Decreased Glibenclamide Uptake in Hepatocytes of Hepatocyte Nuclear Factor-1α-Deficient Mice: A Mechanism for Hypersensitivity to Sulfonylurea Therapy in Patients With Maturity-Onset Diabetes of the Young, Type 3 (MODY3)

Hnf-1α homozygous (Hnf-1α^−/−) mutants, heterozygotes (Hnf-1α^+/-), and wild-type (Hnf-1α^+/+) mice were maintained on a 12-h light/dark cycle with ad libitum access to water and a standard diet in the Laboratory of Animal Research Center, a pathogen-free animal facility at Rockefeller University. Hnf-1α^−/− mice were generated in the laboratory of F. Gonzalez (National Institutes of Health, Bethesda, MD) (4).

[Cyclohexyl-2,3-³H]glibenclamide ([³H]glibenclamide, 52 Ci/mmol) was purchased from New England Nuclear (Boston, MA). Glibenclamide and collagenase type IV were obtained from Sigma (St. Louis, MO). Glibenclamide was prepared as a stock solution in DMSO at a concentration of 20 mmol/l. Other reagents were purchased from Sigma and were of the highest quality available. Acetonitrile, n-hexane, and dichloromethane were from Sigma and of high-performance liquid chromatography (HPLC) grade.

Insulin secretion was studied in vivo using 4- to 6-week-old Hnf-1α^−/− mice and their Hnf-1α^+/+ littermates. Mice were fasted for 10–14 h before glibenclamide (1 mg/kg) was injected into the tail vein. Blood samples were collected by retro-orbital bleeds at −15 and 0 min before glibenclamide injection and at 2, 5, 10, 30, and 60 min after injection in heparinized tubes. The plasma was immediately separated by centrifugation and stored at −20°C for measurement of insulin.

Hnf-1α^−/− mice and their wild-type littermates were used for studying glibenclamide removal from the blood in vivo. For plasma level measurements, glibenclamide at the dose of 1 mg/kg (with 0.2 μCi/g [³H]glibenclamide) was injected into the tail vein of anesthetized mice, and small amounts of blood (40–60 μl) were collected by retro-orbital bleeds at 0, 2, 5, 10, 30, 60, 120, and 180 min. Determination of [³H]glibenclamide concentrations in plasma was performed on 20 μl plasma by addition of 2 ml scintillation fluid. Glibenclamide concentrations were determined using a scintillation counter.

Hepatocytes were prepared from 5- to 7-week-old Hnf-1α^−/− mice and their wild-type littermates according to the collagenase perfusion method of Ferre et al. (17) with the following modifications: 1) mice were anesthetized with Avertin (2.5 g 2,2,2-tribromoethanol in 1.5 ml tert-amylalcohol dissolved in 0.9% NaCl; dose 0.4 ml/25 g) and 2) their portal veins were cannulated and their livers were perfused with buffer A (10 mmol/l HEPES, 143 mmol/l NaCl, 7 mmol/l KCl, 0.2 mmol/l EDTA, pH 7.4), followed by buffer B (50 mmol/l HEPES, 100 mmol/l NaCl, 7 mmol/l KCl, 5 mmol/l CaCl₂, pH 7.6), containing collagenase (0.5 mg/ml), at 37°C. After perfusion, the livers were excised, transferred to Dulbecco’s modified Eagle’s medium (DMEM), and cut into small pieces. The resulting cell suspension was passed through nylon mesh filters, centrifuged at 50g, and resuspended in DMEM, supplemented with 10% fetal bovine serum to inhibit residual collagenase activity. The hepatocytes were plated onto gelatin-coated plates at a density of 10⁵ cells/cm²; cultured in DMEM; supplemented with 10% fetal bovine serum, pyruvate (110 mg/ml), l-glutamine (2 mmol/l), penicillin, and streptomycin; and incubated at 37°C in 5% CO₂ atmosphere. The viability of the cells was tested with 0.4% trypan blue and was 80–90%. Uptake experiments were performed within 24 h after the preparation of hepatocytes.

Extraction and chromatographic separation were performed according to Rydberg et al. (18) using a modified protocol. Briefly, for extraction, 500 mg liver tissue was homogenized in 2 ml PBS, and the cytosolic fraction was prepared by centrifugation at 14,000g for 10 min at 4°C. Serum was diluted with five volumes PBS before extraction. Then, 10% (vol/vol) of 2 mol/l HCl was added to cytosolic fraction or serum, and extraction was performed with six volumes n-hexane-dichloromethane (1:1, vol/vol) for 10 min at room temperature. The organic phase was transferred to a conical tube and evaporated to dryness at 37°C under a gentle stream of air, and the residue was dissolved in 80 μl methanol. Recoveries from liver and gut were 85 ± 5% and from serum were 100 ± 5%. For chromatographic separation of glibenclamide and its metabolites, a BioCad Perfusion Workstation (PerSeptive Biosystems) with a reverse-phase C₁₈ column (5 μm, 4.6 mm × 250 mm; Vydac) was used. The system was run at room temperature, and the mobile phase contained acetonitrile-PBS (28:72, vol/vol). The flow rate was 1 ml/min with pressures ranging from 11 to 17 MPa. Before separation, the system was equilibrated with a mobile phase for 1 h. A total of 20 μl of the sample was injected, and 96 samples (0.2 ml) were collected starting at the time of injection. Radioactivity in all samples was measured by liquid scintillation counting. The recovery of radioactivity was 100 ± 10%.

Protein concentrations were quantified by the bicinchoninic acid protein assay (Sigma) according to the manufacturer’s instructions and using bovine serum albumin as a standard. Insulin concentrations were determined by a double-antibody radioimmunoassay (Linco, St. Louis, MO) using rat insulin as standard.

Results are expressed as means ± SE. Statistical analysis was performed by a two-tailed Student’s t test for unpaired data. Differences were considered to be significant at P < 0.05.

To assess sulfonylurea-induced insulin secretion in vivo, we administered glibenclamide (1 mg/kg body wt) intravenously and compared insulin secretion between wild-type mice, Hnf-1α^+/− mice, and their Hnf-1α^−/− littermates. After glibenclamide injection, Hnf-1α^+/− mice exhibited insulin levels at 2, 5, 10, 30, and 60 min that were indistinguishable from those of wild-type animals. Furthermore, blood glucose levels did not differ in Hnf-1α^+/+ and Hnf-1α^+/− animals after intravenous glibenclamide administration. We then compared serum insulin levels after a glibenclamide challenge of Hnf-1α^−/− and Hnf-1α^+/+ animals. Fasting insulin levels from blood samples collected at −15 and 0 min were comparable between the two groups of mice (0.27 ± 0.06 and 0.36 ± 0.11 ng/ml in Hnf-1α^+/+ animals vs. 0.27 ± 0.03 and 0.24 ± 0.04 ng/ml in Hnf-1α^−/− animals). Insulin levels at 2 min after glibenclamide were not decreased in Hnf-1α^−/− mice (2.48 ± 0.38 ng/ml) compared with Hnf-1α^+/+ mice (2.36 ± 0.32 ng/ml). As shown in Fig. 1, a similar pattern of insulin response was also observed at later time points, without significant differences. The ability of Hnf-1α^−/− mice to markedly increase insulin secretion after a glibenclamide challenge is in sharp contrast to the profound decrease in insulin secretion responses to secretagogues such as glucose, glyceraldehydes, and arginine (19; data not shown). This finding indicates that Hnf-1α-deficient mice may be a useful model to elucidate the molecular basis for the hypersensitivity to sulfonylureas in Hnf-1α-dependent diabetes.

We next evaluated the pharmacokinetics of glibenclamide in the Hnf-1α^−/− mice. [³H]glibenclamide levels were measured in plasma of wild-type and Hnf-1α^−/− mice after a bolus intravenous injection. The rate of disappearance of [³H]glibenclamide from the plasma was markedly reduced in Hnf-1α^−/− mice compared with Hnf-1α^+/+ mice (Fig. 2). The ratio of the area under the curve of Hnf-1α^−/− versus Hnf-1α^+/+ was 4.2. Glibenclamide plasma concentrations remained significantly elevated for 2 h in Hnf-1α^−/− mice compared with Hnf-1α^+/+ mice. Glibenclamide was rapidly cleared (91% after 2 min and 96% after 5 min) in Hnf-1α^+/+ mice after intravenous administration of [³H]glibenclamide. In contrast, plasma glibenclamide levels decreased to only 50% after 2 min and to 16% after 1 h in Hnf-1α^−/− animals. The half-life of glibenclamide in Hnf-1α^−/− and Hnf-1α^+/+ mice was 30 s and 2 min, respectively. These results demonstrate that glibenclamide clearance from the blood is markedly reduced in Hnf-1α^−/− mice. Because the liver is the major site for drug metabolism of glibenclamide, we hypothesized that the reduced clearance of glibenclamide was due to a decrease in hepatic uptake or impaired metabolism of the sulfonylurea.

Several extrinsic serum factors in Hnf-1α^−/− mice could in theory be responsible for the decreased uptake of glibenclamide in mutant Hnf-1α^−/− animals. For instance, Hnf-1α^−/− mice have elevated bile acid serum levels that may inhibit glibenclamide uptake into hepatocytes in vivo (20,21). To rule out the potential involvement of inhibitors of glibenclamide transport, such as cholate, that may interact with glibenclamide clearance in vivo, we analyzed the uptake of [³H]glibenclamide into cultured primary hepatocytes that were isolated from Hnf-1α^+/+ and Hnf-1α^−/− littermate mice. The time course for the net uptake of [³H]glibenclamide into cultured hepatocytes is shown in Fig. 3A. Glibenclamide uptake into primary hepatocytes from Hnf-1α^−/− mice was markedly lower than that of the hepatocytes derived from Hnf-1α^+/+ livers. Kinetics parameters of glibenclamide uptake into cultured hepatocytes were determined using a Lineweaver-Burk plot analysis (data not shown). The calculated V_max was 2.4-fold higher in Hnf-1α^+/+ hepatocytes than in Hnf-1α^−/− hepatocytes (301 vs. 128 pmol · min⁻¹ · mg⁻¹ protein, respectively). These results demonstrate that differences in glibenclamide uptake were due to a primary effect and not secondary to a difference in serum composition between Hnf-1α^+/+ and Hnf-1α^−/− mice. Furthermore, the data indicate that a decrease in hepatic transport and/or a defect in liver glibenclamide metabolism is responsible for the defect in clearance of glibenclamide in mutant Hnf-1α mice.

To investigate if a defect in glibenclamide metabolism contributed to the decrease in hepatic uptake of glibenclamide, we analyzed glibenclamide and its metabolites in liver extracts. Livers from Hnf-1α^+/+ and Hnf-1α^−/− mice were excised 3 h after intravenous injection of [³H]glibenclamide, and compounds were extracted and measured by HPLC analysis (Fig. 4A and B). The glibenclamide peak was identified by spiking with [³H]glibenclamide. The two metabolite peaks (4-trans-hydroxyglibenclamide and 3-cis-hydroxyglibenclamide) have been described by Rydberg et al. (18). The liver content of glibenclamide and its two major hydroxylated metabolites are shown in Table 1. Glibenclamide content was about 20-fold higher in the livers of Hnf-1α^+/+ mice compared with Hnf-1α^−/− mice (158.6 ± 6.9 vs. 8.3 ± 0.7 pmol · l⁻¹ · mg⁻¹ protein). The ratios of 4-trans-hydroxyglibenclamide (M1) and 3-cis-hydroxyglibenclamide (M2) to glibenclamide were increased in Hnf-1α^−/− livers (14.1 and 13.0%, respectively) compared with Hnf-1α^+/+; mice (8.7 and 7.3%, respectively). This indicated that glibenclamide is metabolized normally in hepatocytes of Hnf-1α^−/− mice. Therefore, a defect in glibenclamide transport by hepatocytes is likely to be responsible for the reduced clearance of glibenclamide in the Hnf-1α^−/− mice.

To demonstrate that the radioactivity measured in the plasma was indeed glibenclamide and not its metabolites, a serum extract from Hnf-1α^−/− mice was also analyzed by HPLC 5 min after intravenous injection of [³H]glibenclamide. As shown on Fig. 4C, more than 95% of the counted radioactivity in the plasma of Hnf-1α^−/− mice was glibenclamide. In contrast, HPLC analysis from gut extracts obtained 3 h after intravenous injection of [³H]glibenclamide revealed the presence of predominately two hydroxylated metabolites (Fig. 4D). These data suggest that glibenclamide itself, and not its metabolites, constitutes the majority of the plasma levels in Hnf-1α^−/− mice. Furthermore, the presence of glibenclamide metabolites in the intestine of Hnf-1α^−/− mice indicates that secretion of glibenclamide metabolites into the biliary system is functional. We therefore conclude that hepatic metabolism of glibenclamide as well as biliary secretion is functional in Hnf-1α^−/− mice.

In this study, we examined the role of HNF-1α in glibenclamide metabolism to study possible mechanisms for hypersensitivity to sulfonylureas in HNF-1α-dependent diabetes (MODY3). As a model system for MODY 3, we used mice in which the Hnf-1α gene was inactivated by targeted mutagenesis (4). Heterozygous Hnf-1α mice do not develop diabetes, and islets have normal insulin secretion in response to glucose and other insulin secretagogues (5,19,22). Furthermore, we did not detect differences in insulin secretion or blood glucose levels after an intravenous glibenclamide infusion between wild-type and Hnf-1α^+/− mice, indicating that haplo-insufficient mice are not suitable for studying the pharmacogenetics of glibenclamide in Hnf-1α deficiency. However, Hnf-1α^−/− mice are diabetic, and their pancreatic islets have profound defects in glucose- and arginine-stimulated insulin secretion (19). Pancreatic islets from Hnf-1α^−/− mice also exhibit significantly reduced insulin secretory responses to tolbutamide compared with Hnf-1α^+/− animals in vitro (5,19). However, in vivo, we have observed that insulin responses to glibenclamide are similar in Hnf-1α^−/− and Hnf-1α^+/+ mice. This discrepancy may be explained by the increased glibenclamide concentration to which pancreatic islets of Hnf-1α^−/− animals are exposed. Therefore, this finding suggests that β-cells that are deficient in Hnf-1α are normo- or hyposensitive to the actions of sulfonylureas, thus making it unlikely that a β-cell intrinsic mechanism is responsible for the increased sensitivity to sulfonylureas.

HNF-1α is expressed in the liver and the kidney—two organs that play a crucial role in the detoxification of glibenclamide. Therefore, we tested the hypothesis that hepatic metabolism is affected in Hnf-1α^−/− mice. The elimination of glibenclamide from the body depends on hepatic metabolism and biliary clearance. In the liver, glibenclamide is converted into two major metabolites: 4-trans-hydroxyglibenclamide and 3-cis-hydroxyglibenclamide (18). Approximately 50% of these compounds are excreted in the urine and 50% in the bile. The first step of glibenclamide metabolism involves the transport by an as yet unknown energy-independent facilitative diffusion process (15). The uptake of glibenclamide can be inhibited by other organic anions including the unconjugated bile acid cholate, as well as loop diuretics such as bumetanide (20,23,24). We have previously shown that Hnf-1α^−/− mice have elevated serum bile acid levels due to increased de novo synthesis and decreased uptake by hepatocytes (21). Therefore, the high levels of bile acids present in sera of Hnf-1α^−/− mice could have contributed to the decreased clearance of glibenclamide observed in vivo. To control for these confounding factors, we studied glibenclamide uptake into cultured hepatocytes of Hnf-1α^−/− and Hnf-1α^+/+ mice and showed that V_max was approximately threefold lower in mutant mice. The results demonstrate that the decreased clearance of glibenclamide is caused by a primary defect in hepatic clearance.

Inactivation of glibenclamide in the liver occurs through hydroxylation by cytochrome P450 enzymes followed by glucuronoconjugation and sulfonoconjugation (23). Thus, depending on the limiting step of glibenclamide metabolism, reduced uptake of glibenclamide into hepatocytes may be attributed to an impairment of hepatic uptake or metabolism. To investigate if hydroxylation of glibenclamide was impaired, we measured its metabolites in livers of Hnf-1α^−/− and Hnf-1α^+/+ animals. Using HPLC analysis, we showed that the ratios of glibenclamide and its two hydroxylated metabolites 4-trans-hydroxyglibenclamide and 3-cis-hydroxyglibenclamide were significantly increased in liver extracts of Hnf-1α^−/− mice compared with wild-type littermates. This result indicates that a defect in glibenclamide metabolism is not responsible for decreased glibenclamide uptake in hepatocytes of Hnf-1α^−/− mice. Furthermore, the increase of the relative metabolite concentrations in liver extracts of Hnf-1α^−/− mice may be due to the reduced glibenclamide concentrations in the hepatocytes of Hnf-1α^−/− animals. Together, our data strongly suggest that decreased transport across the hepatocyte plasma membrane appears to be responsible for the increased biological half-life of glibenclamide in mice lacking Hnf-1α.

Petzinger and Fückel (15) have shown that glibenclamide transport occurs through a carrier-mediated system. Several classes of transport systems have been described that mediate hepatic organic anion uptake mechanisms (25). Organic anion transporter proteins play a critical role in this process by transporting a broad spectrum of substrates in a sodium-independent manner. We have recently shown that gene expression of oatp1 (Slc21a1), Lst-1 (Slc21a6), OATPD (SLC21A11), and oat2 are downregulated or absent in livers of Hnf-1α^−/− mice (21). These carriers are candidates for facilitating the transport of glibenclamide across the hepatocyte membrane. We cannot rule out that an intracellular binding protein may be involved in glibenclamide hepatic uptake as well. Meanwhile, identification of the glibenclamide transporter or intracellular binding protein remains to be established.

An example of the genetic defect altering the pharmacological response to sulfonylurea in diabetes was described 20 years ago, and a monogenic control of tolbutamide metabolism in humans was proposed (26). Our results are consistent with the mechanism described in this study. Scott and Pofferbarger (26) observed a ninefold variation in the rate of tolbutamide disappearance from plasma after intravenous administration in humans. This variation was characterized by a trimodal frequency distribution, suggestive of monogenic inheritance. The binding of tolbutamide to serum proteins was not correlated to the tolbutamide clearance. It was therefore proposed that the microsomal hepatic conversion of free tolbutamide to hydroxytolbutamide was the impaired step. However, this process can be viewed as two separate steps: first, removal of free tolbutamide from the blood by hepatocytes, and second, enzymatic conversion by hepatic microsomes. Therefore, genetic variation in genes encoding the Hnf-1α-regulated glibenclamide transporter as well the sulfonylurea-metabolizing P450 enzymes may be responsible for differences in plasma half-lives and susceptibility to these drugs in the population.

Glibenclamide is one of the most widely used orally active sulfonylureas in the treatment of type 2 diabetes. Hypoglycemia is the most important and most often fatal adverse effect of sulfonylureas (23). We propose that hypoglycemia may occur more frequently in subjects with HNF-1α deficiency treated with glibenclamide because of a delayed disappearance of the drug from the blood. Very low doses of short-acting sulfonylureas should therefore be used initially in the treatment of MODY3 patients, and these drugs should be discontinued in subjects with liver disease. Monitoring sulfonylurea plasma concentration in MODY3 patients may also be useful for avoiding hypoglycemia.

This work was supported by National Institutes of Health Grant RO1 DK55033-04 (M.S.) and Medical Scientists Training Program Grant GM07739 (D.Q.S.), the Deutsche Forschungsgemeinschaft (C.W.), and the Human Science Frontier Organization (M.S.).

We would like to thank Dr. F. Gonzalez for providing Hnf-1α mutant mice. We would also like to thank Robert and Harriet Heilbrunn and Allen and Frances Adler for their generous support.