Defective Pancreatic β-Cell Glycolytic Signaling in Hepatocyte Nuclear Factor-1α-deficient Mice*

Mutations in the hepatocyte nuclear factor-1α (HNF-1α) gene cause maturity onset diabetes of the young type 3, a form of type 2 diabetes mellitus. In mice lacking the HNF-1α gene, insulin secretion and intracellular calcium ([Ca2+] i ) responses were impaired following stimulation with nutrient secretagogues such as glucose and glyceraldehyde but normal with non-nutrient stimuli such as potassium chloride. Patch clamp recordings revealed ATP-sensitive K+currents (KATP) in β-cells that were insensitive to suppression by glucose but normally sensitive to ATP. Exposure to mitochondrial substrates suppressed KATP, elevated [Ca2+] i , and corrected the insulin secretion defect. NAD(P)H responses to glucose were substantially reduced, and inhibitors of glycolytic NADH generation reproduced the mutant phenotype in normal islets. Flux of glucose through glycolysis in islets from mutant mice was reduced, as a result of which ATP generation in response to glucose was impaired. We conclude that hepatocyte nuclear factor-1α diabetes results from defective β-cell glycolytic signaling, which is potentially correctable using substrates that bypass the defect.

Maturity-onset diabetes of the young (MODY) 1 (1), an autosomal dominant form of non-insulin-dependent diabetes mellitus in which affected subjects develop hyperglycemia generally before the age of 25 years (1, 2), may be due to mutations in the transcription factor hepatocyte nuclear factor-1␣ (HNF-1␣) (3). These mutations appear to cause diabetes by inducing alterations in ␤-cell function. Insulin secretion is abnormally low in individuals with HNF-1␣ diabetes, non-diabetic subjects with mutations in HNF-1␣ demonstrate reduced insulin secretory responses to high glucose, and insulin levels may be reduced even under basal conditions (4,5).
The mechanisms whereby mutations in the gene for HNF-1␣ lead to abnormalities in insulin secretion have not been determined. We have demonstrated previously that mice lacking both alleles of the HNF-1␣ gene (HNF-1␣ (Ϫ/Ϫ)) have substantially reduced insulin secretory responses to glucose compared with wild type mice (HNF-1␣ (ϩ/ϩ)) whereas heterozygous HNF-1␣ (HNF-1␣ (ϩ/Ϫ)) are normal (6). In this study, experiments have been performed in wild type mice, and in heterozygous and homozygous HNF-1␣-deficient mice (7), to define the nature of the defects present in ␤-cell signaling pathways that are responsible for reduced insulin secretory responses to glucose in HNF-1␣ diabetes.

EXPERIMENTAL PROCEDURES
Pancreatic Islet and ␤-Cell Isolation-Pancreatic islets were isolated from 6 -12-week-old HNF-1␣ (Ϫ/Ϫ) mice and their age-matched heterozygote (ϩ/Ϫ) and wild type (ϩ/ϩ) controls by collagenase digestion and differential centrifugation through Ficoll gradients using a modification of procedures previously described (8). Islets were either cultured overnight prior to experiments performed to measure insulin secretion or plated on glass coverslips and cultured for 48 -96 h for measurements of intracellular calcium ([Ca 2ϩ ] i ). Single ␤-cells used for membrane current recordings were obtained from islets dispersed by trituration (200 strokes through a 200-l pipette tip) in Ca 2ϩ -and Mg 2ϩ -free Hank's solution at room temperature. Cells were plated on glass coverslips and maintained in culture for 2-6 days. All cultures were performed using RPMI 1640 medium supplemented with fetal bovine serum, 11.6 mM glucose, 100 microunits/ml penicillin, and 100 g/ml streptomycin as described previously (9). Medium was replaced every third day.
Insulin Secretion from Isolated Perifused Pancreatic Islets-Following a base-line sampling period during which insulin concentrations were measured at 1-min intervals, the following secretagogues were added sequentially to the perifusate (a modified Krebs-Ringer buffer (KRB) containing (in mM) 119 NaCl, 4.7 KCl, 2 CaCl 2 , 1.2 MgSO 4 , 1.2 KH 2 PO 4 , 25 NaHCO 3 , 2 glucose) for a period of 10 -15 min: D-glyceraldehyde (24 mM), methyl pyruvate (24 mM), tolbutamide (100 M), and KCl (20 mM). Insulin concentrations were measured in the effluent perifusate every third minute and every minute for 3 min immediately after a change in perifusion conditions. DNA was measured in aliquots of islets from the same mice and the insulin secretory responses expressed as picomoles/liter/g of DNA. The average insulin concentration in the effluent perifusate during administration of each secretagogue was calculated for each individual experiment, and the means for each secretagogue were compared between HNF-1␣ (ϩ/Ϫ) and (Ϫ/Ϫ) islets. Since glyceraldehyde administration was associated with a very pronounced and brief first phase of insulin secretion and a much lower second phase, the mean peak insulin secretory response (the single point of highest response) was compared between groups.
Assay Methods-Insulin concentrations were measured by a double-antibody radioimmunoassay using a rat insulin standard. The intraassay coefficient of variation for this technique is 7%. All samples were assayed in duplicate. DNA content of pancreatic islets was quantified in aliquots of 20 islets from three to four (ϩ/Ϫ) and (Ϫ/Ϫ) mice using a fluorimetric assay as described previously (10). Intracellular Ca 2ϩ Concentration in Pancreatic Islets- [Ca] i was estimated using fluorescence imaging techniques previously used by our laboratory (11).
Membrane Current Recordings-The bathing solution contained (in mM): 137 NaCl, 4.7 KCl, 2 CaCl 2 , 1.2 Mg SO 4 , 1.2 KH 2 PO 4 , 10 HEPES (pH 7.45), 2-10 glucose, and the ␤-cell was dialyzed with an internal solution containing (in mM): 120 KCl, 5 NaCl, 10 EGTA, 0 -5 MgATP, 10 HEPES (pH 7.2). The resistance of the patch pipette ranged between 1.0 and 2.5 megohms. Care was taken during the break-in procedure to minimize the negative pressure utilized to gain access, as this increased the lifetime of the recordings. No series resistance compensation was used. Experiments were performed at room temperature. Current was normalized by reference to the cell capacitance. ␤-Cells were identified by a combination of morphometric criteria and the presence of nutrientsuppressible ATP-sensitive K ϩ currents. Average values are expressed as mean Ϯ S.E. NAD(P)H Fluorescence Measurements-Mutant and control islets were maintained in culture as described above for 48 -96 h to allow adherence to coverslips. Using a fluorescence imaging system, cellular autofluorescence was excited by ultraviolet light at 360 nm and the emitted fluorescence intensity was measured at 460 nm. Experiments were performed at 37°C in islets that were perifused with KRB. The relative increase in fluorescence intensity was obtained by dividing the fluorescence intensity following stimulation with secretagogues by the mean fluorescence intensity measured during the base-line resting condition (KRB ϩ 2 mM glucose).
Measurements of Islet Glycolytic Flux-Islet glycolytic flux was determined by measuring the formation of 3 H 2 O from D-[5-3 H]glucose (12). Duplicate batches of 10 islets from HNF-1␣ (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) mice were incubated in 100 l of KRB containing either 2 or 14 mM glucose, 5 mM HEPES, 0.2% w/v bovine serum albumin, and 1 Ci of The incubations were carried out in 1-ml Eppendorf tubes inside 20-ml glass scintillation vials containing 1 ml of distilled water and fitted with airtight rubber seals. Following a 30-min preincubation, in KRB containing 2 mM glucose, incubations were performed for 120 min at 37°C with continuous shaking. Islet metabolism was stopped by the addition of 0.1 ml of 0.1 M HCl, injected into the inner tube through the rubber seal. The sealed scintillation vials were subsequently left at room temperature overnight to allow equilibration of the 3 H 2 O produced by the islets with the water in the outer vials. Liquid scintillation fluid was then added to the outer tube and mixed, and the radioactivity was quantified using a scintillation analyzer (Tricarb 2200CA, Packard Instrument Co., Downer's Grove, IL). The recovery of 3 H from known amounts of 3 H 2 O was 40 Ϯ 5% under these experimental conditions. The rate of glycolytic flux was corrected for recovery and expressed as picomoles of glucose/islet/2 h.
Magnesium Green Fluorescence Measurements-Magnesium Green (MgG) fluorescence was measured in islets as an indirect measure of intracellular ATP. ATP has a greater affinity than ADP for Mg 2ϩ (13,14). Therefore, as intracellular ATP increases, the cytosolic ionized Mg 2ϩ concentration falls, reflected by a reduction in fluorescence intensity in cells loaded with MgG. Control and mutant islets were isolated and maintained in culture on coverslips, as described above, for 48 -96 h. Islets were loaded with MgG acetoxymethylester (5 M, Molecular Probes, Eugene, OR) for 30 min at 37°C in a humidified incubator. Islets were perifused as described previously (11) using oxygenated KRB containing variable concentrations of glucose and ketoisocaproic acid (KIC), and fluorescence intensity was measured using a fluorescence imaging system as described above. Fluorescence was excited by ultraviolet light at 495 nM, and emission was measured at 535 nM. The relative decrease in fluorescence intensity was obtained by dividing the maximum decrease in fluorescence intensity following stimulation with secretagogues by the mean fluorescence intensity measured during the base-line equilibration period (KRB ϩ 2 mM glucose).
Measurement of Islet ATP Content-ATP content of pancreatic islets was measured by a quantitative bioluminescence assay using a modification of a previously described protocol (15). Experiments were performed on islets from HNF-1 ␣ (ϩ/ϩ), and (Ϫ/Ϫ) mice that had been incubated overnight in RPMI containing 11.6 mM glucose. On day 2, groups of 15 islets were hand-picked and pre-incubated in KRB containing 2 mM glucose for 45 min. Since the HNF-1␣ (Ϫ/Ϫ) islets were smaller than the (ϩ/ϩ) islets, the smaller control islets were picked for these experiments. Following the pre-incubation, islets were trans-ferred to separate tubes containing 0.5 ml of KRB with either 2 or 14 mM glucose and incubated for 5 min. After incubation, 0.5 ml of ice-cold 10% trichloroacetic acid was added and the tubes were vortex-mixed and kept on ice for 15 min. Following centrifugation, a fraction of the supernatant was removed and mixed with 1.5 ml of diethyl ether to remove the trichloroacetic acid. After vortex mixing, the ether phase was discarded and the procedure was repeated three times. The final extracts were diluted 1:1 with an ATP assay buffer and the ATP concentration of the supernatant was assayed using an ATP bioluminescence kit (FL-AA, Sigma) as per the manufacturer's instructions. ATP was measured in duplicate 0.1-ml aliquots from each sample, mixed with 0.1 ml of the ATP assay mix containing luciferase and luciferin. ATP is consumed and light emitted when luciferase catalyzes the oxidation of D-luciferin. The emitted light was measured using a luminometer (Monolight 1500, Analytical Luminescence Laboratories, San Diego, CA).
Statistical Analysis-Results are expressed as mean Ϯ S.E. The statistical significance of differences between animal groups was evaluated using the Wilcoxon rank-sum test or one-way analysis of variance where appropriate. Differences were considered to be significant at p Ͻ 0.05.

RESULTS
We have previously demonstrated that insulin secretion, in response to an increase in glucose concentration from 2 to 26 mM, in islets from HNF-1␣ (Ϫ/Ϫ) animals is substantially (approximately 85%) lower than in islets from both HNF-1␣ (ϩ/ϩ) or (ϩ/Ϫ) animals (6). In order to further define the insulin secretory defect in the HNF-l␣ (Ϫ/Ϫ) mouse, isolated pancreatic islets were stimulated with secretagogues that provoke insulin release by distinct mechanisms. Increases in extracellular glucose concentration normally initiate insulin secretion from pancreatic ␤-cells by inducing closure of ATP-sensitive K ϩ channels (KATP) via an increase in the cytosolic ATP/ADP ratio (16,17). This leads to membrane depolarization and influx of Ca 2ϩ through voltage-dependent Ca 2ϩ channels. The elevation in [Ca 2ϩ ] i is a necessary prerequisite for insulin granule exocytosis. Sulfonylurea drugs like tolbutamide stimulate insulin secretion by directly blocking KATP (18). As was the case with glucose, tolbutamide produced a significantly smaller secretory response in HNF-1␣ (Ϫ/Ϫ) compared with (ϩ/Ϫ) is- Stimulation of islets from HNF-1␣ (Ϫ/Ϫ) mice with glyceraldehyde, a metabolic intermediate that enters glycolysis distal to glucose 6-phosphate, produced only 30% of the peak secretory response seen in (ϩ/Ϫ) islets when applied to the (Ϫ/Ϫ) islets (0.24 Ϯ 0.06 versus 0.73 Ϯ 0.18 pmol/liter/ng of DNA; p Ͻ 0.05) (Fig. 1A). In addition, the area under the curve for [Ca 2ϩ ] i in (Ϫ/Ϫ) islets in response to glyceraldehyde was 28% of the (ϩ/Ϫ) value (Table I). These data suggest that specific steps in the early signal transduction process stimulated by glucose were impaired in the HNF-1␣ (Ϫ/Ϫ) islets. The observation that KCl induced equivalent increases in [Ca 2ϩ ] i in HNF-1␣ (Ϫ/Ϫ) and (ϩ/Ϫ) islets (Table I) suggests that later steps responsible for the elevation in [Ca 2ϩ ] i were unaffected. We have previously shown that 14 mM glucose failed to elevate [Ca 2ϩ ] i in the (Ϫ/Ϫ) islets (6).
In order to further evaluate alterations in proximal steps in the glucose-stimulated stimulus-secretion coupling pathway, we measured changes in KATP currents (I KATP ) in single pancreatic ␤-cells utilizing standard whole-cell voltage clamp techniques. Exposure of single ␤-cells to stimulatory concentrations of glucose ordinarily results in a rapid suppression of I KATP . HNF-1␣ (Ϫ/Ϫ) ␤-cells displayed large amplitude I KATP that were fully activated on break-in or progressively increased during the 5-10 min dialysis period ( Fig. 2A). By contrast, in A-C, depolarizing pulses of 160 ms in duration were delivered at 10-mV increments from a holding potential of Ϫ60 to ϩ40 mV. Shown are currents recorded immediately following a break in (left panel), and the 600 s of (right panel), intracellular dialysis. Recordings were made in the presence of 10 mM glucose (A and B) and 2 mM glucose (C). D, time course of KATP current activation (measured by pulsing from Ϫ60 to ϩ40mV) expressed as current density subsequent to break-in in the HNF-1␣ (Ϫ/Ϫ) and HNF-1␣ (ϩ/Ϫ) ␤-cells. The inset shows the biphasic alteration in KATP current in the (ϩ/Ϫ) ␤-cells on a magnified current density scale. HNF-1␣ (ϩ/Ϫ) and (ϩ/ϩ) ␤-cells the amplitude of I KATP initially was small or completely absent, then increased and subsequently decreased rapidly during dialysis (Fig. 2, B and C).
To determine whether alterations of KATP were responsible for the reduced glucose responsiveness in the HNF-1(Ϫ/Ϫ) ␤-cells, different concentrations of ATP or NADH were applied via the patch pipette to block I KATP . In the (Ϫ/Ϫ) ␤-cells, I KATP had normal sensitivity to ATP and NADH: the IC 50 for ATP was approximately 75 M (Fig. 3, A and B), while both ATP and NADH caused complete current suppression at 500 M (n ϭ 6), in accordance with published data (17), and confirmed in the case of ATP in (ϩ/Ϫ) ␤-cells bathed in 2 mM glucose. Furthermore, both HNF-1␣ (Ϫ/Ϫ) and (ϩ/Ϫ) ␤-cells were equivalently suppressed by 1 mM tolbutamide (Fig. 3C) (18). These findings suggested that the glucose insensitivity of HNF-1␣ (Ϫ/Ϫ) ␤-cells might be related to an inability to produce ATP following the metabolism of glucose. This hypothesis was supported by the finding that ␤-cells from (ϩ/Ϫ) animals bathed in 10 mM glucose, together with the ATP synthesis uncouplers sodium azide (Fig. 3D) or FCCP (data not shown) (19,20) produced I KATP of a similar magnitude to those seen in HNF-1␣ (Ϫ/Ϫ) ␤-cells bathed in 10 mM glucose alone.
To further define the defect in ATP production, we studied the effect of a number of specific substrates and inhibitors of ATP synthesis. The methyl ester of pyruvate, a potent insulin secretagogue that generates large amounts of mitochondrial ATP (21,22), was used to specifically stimulate mitochondrial ATP production. Exposure of the HNF-1␣ (Ϫ/Ϫ) ␤-cells to 20 -50 mM methyl pyruvate induced an immediate (Ͻ20 s) suppression of I KATP (Fig. 4), which was reversible following addition of sodium azide, demonstrating that the effect is mediated through a stimulation of mitochondrial ATP production. Leucine, another stimulator of mitochondrial ATP synthesis (23), elicited similar effects on I KATP in the homozygous mutant ␤-cells that were reversed by sodium azide (data not shown). These results suggest that the ability of mitochondria to generate ATP from leucine or methyl pyruvate is unaffected by the loss of HNF-1␣ expression. Consistent with the effectiveness of mitochondrial substrates in suppressing I KATP , methyl pyruvate produced equivalently robust elevation in [Ca 2ϩ ] i in the mutant and control islets (Table I). Furthermore, the mean insulin secretory responses to methyl pyruvate were not significantly different in HNF-1␣ (ϩ/Ϫ) and (Ϫ/Ϫ) islets (0.64 Ϯ 0.04 versus 0.44 Ϯ 0.12 pmol/liter/ng of DNA; p Ͼ 0.05) (Fig. 1B). Iodoacetic acid (IAA) inhibits glycolysis at the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) preventing NADH and glycolytic ATP generation (24). When perifused prior to glucose, 1 mM IAA markedly attenuated the insulin secretory response to glucose in control islets ((ϩ/Ϫ) or (ϩ/ϩ)), rendering them as unresponsive to glucose as the HNF-1␣ (Ϫ/Ϫ) islets (Fig. 5).
Glycolysis generates ATP directly and indirectly via production of NADH. To determine if the impairment of ATP production is secondary to failure to reduce NAD ϩ to NADH, NAD(P)H generation by islets in response to glucose was estimated using fluorescence imaging. Fluorescence intensity at 460 nm increased by 72% and 71% over basal levels in response to 14 mM glucose in the HNF-1␣ (ϩ/ϩ) (Fig. 6, A and D) and (ϩ/Ϫ) islets (data not shown) but only by 23% in the (Ϫ/Ϫ) islets (p Ͻ 0.05; Fig. 6, B and D). IAA also abolished NAD(P)H responses of (ϩ/ϩ) islets to glucose (Fig. 6C) to levels similar to those seen in (Ϫ/Ϫ) islets with glucose alone (Fig. 6B). Methyl pyruvate (24 mM) induced a 2.1-fold increase in NAD(P)H in HNF-1␣ (Ϫ/Ϫ) islets, which tended to be greater than the responses in (ϩ/ϩ) islets (1.7-fold increase) (Fig. 7, A, B, and D). That the large increases in NAD(P)H in both (Ϫ/Ϫ) and (ϩ/ϩ) islets induced by methyl pyruvate occurred by stimulation of Krebs cycle metabolism was supported by the lack of effect of iodoacetate on methyl pyruvate-induced increases in NAD(P)H (Fig. 7, C and D).
These results predict a reduction in ATP production in response to glucose in the (Ϫ/Ϫ) islets. In order to determine if this occurred, changes in Magnesium Green fluorescence were measured in islets from HNF-1 ␣ (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) mice in response to stimulation by 14 mM glucose and 10 mM KIC (a stimulator of mitochondrial metabolism) in the presence of 2 mM glucose (Fig. 9). The expected small reduction in MgG fluorescence was consistently seen in islets from (ϩ/ϩ) and (ϩ/Ϫ) mice. The average maximal decrease in fluorescence intensity in the (ϩ/ϩ) islets in response to a step increase from 2 to 14 mM glucose was 13.5 Ϯ 1.7% and in the (ϩ/Ϫ) mice was 10.7 Ϯ 2.6%. Although these differences within each of the groups represented statistically significant differences from base-line, the magnitude of the change in fluorescence was not different between the (ϩ/ϩ) and (ϩ/Ϫ) islets. In contrast, glucose did not induce a significant change (2.4 Ϯ 1.0%) in fluorescence intensity in the (Ϫ/Ϫ) islets. The average change in fluorescence was significantly less than in control islets (p Ͻ 0.03 compared with (ϩ/Ϫ) and p Ͻ 0.005 compared with (ϩ/ϩ) islets; Fig. 9C). In contrast, there was no difference in the response to 10 mM KIC in the three groups. The average maximal decrease in fluorescence intensity in response to KIC was 13.6 Ϯ 1.8, 15.5 Ϯ 1.6, and 14.5 Ϯ 2.1 in the (ϩ/ϩ), (ϩ/Ϫ), and (Ϫ/Ϫ) islets, respectively (p Ͼ 0.05; Fig. 9D).

DISCUSSION
This study was undertaken to further characterize the defect in ␤-cell function in HNF-1␣ (Ϫ/Ϫ) mice. We show that there is a defect in ATP production in response to glucose in (Ϫ/Ϫ) ␤-cells related to failure to generate NADH from glycolytic substrates.
We have demonstrated that glucose (6) and glyceraldehydeinduced insulin secretion is attenuated in HNF-1␣ (Ϫ/Ϫ) islets while the secretagogue effects of methyl pyruvate and KClinduced depolarization are normal. Although insulin secretion following application of tolbutamide was reduced in islets from homozygous mutant mice, it is likely that this is secondary to a defect in ATP production. This hypothesis accords with the observation that the sensitivity of KATP to sulfonylureas is low in cells with impaired ATP production (25,26).
To determine whether there was a specific abnormality in the KATP channel or its regulation, direct measurements using electrophysiological techniques were undertaken using the secretagogues whose mode of action is summarized in Fig. 11. ␤-Cell membrane depolarization induced by high glucose is initiated by closure of the KATP channel via an increase in the ATP/ADP ratio (16,17). Using standard whole-cell voltage clamp configuration recordings, we demonstrated that the ability of glucose to induce closure of KATP channels in the HNF-1␣ (Ϫ/Ϫ) mouse ␤-cells was significantly reduced. Since the KATP current was blocked by the addition of ATP and NADH to the cell, the failure to generate ATP appears to account for the reduced glucose responsiveness of the KATP channel in HNF-1␣ (Ϫ/Ϫ) mouse ␤-cells. This conclusion is supported by the finding that treatment of control islets with uncouplers of oxidative phosphorylation, FCCP and sodium azide, generated large KATP currents, similar to those seen in the (Ϫ/Ϫ) cells.
To determine whether ATP formation in HNF-1␣ (Ϫ/Ϫ) islets  [5-in response to glucose stimulation is reduced, two approaches were utilized. The first technique involved the measurement of Magnesium Green fluorescence in islets. This method of measuring ATP takes advantage of the fact that ATP has a greater affinity for Mg 2ϩ than ADP so that the intracytoplasmic [Mg 2ϩ ] decreases when ATP is formed (13,14). Using this technique, we have demonstrated a significantly smaller decrease in MgG fluorescence intensity in (Ϫ/Ϫ) islets compared with control islets in response to glucose stimulation, thus providing indirect evidence to support the hypothesis that there is a defect in ATP generation in response to glucose in HNF-1␣ (Ϫ/Ϫ) islets. Second, we measured the ATP content of islets from HNF-1␣ (ϩ/ϩ) and (Ϫ/Ϫ) islets incubated at 2 and 14 mM glucose using a quantitative bioluminescence assay. Whereas the ATP content of HNF-1␣ (ϩ/ϩ) islets increased in response to glucose stimulation, no increase was found in the (Ϫ/Ϫ) islets. Taken together, these data support the hypothesis that there is a defect in ATP production by HNF-1␣ (Ϫ/Ϫ) islets in response to glucose stimulation.
Glucose-induced inhibition of I KATP has previously been shown to depend on glycolysis, distal to the generation of glyceraldehyde 3-phosphate (27), a conclusion consistent with the observation that glyceraldehyde is an effective secretagogue in normal islets. In particular, the production of NADH by GAPDH and its consequent conversion into ATP by mitochondria is critical to normal glucose signaling in ␤-cells (27,28). A defect in glucose-induced insulin secretion in HNF-1␣ (Ϫ/Ϫ) islets caused by the inability of glucose metabolism to produce ATP could be the consequence of reduced NAD(P)H responses to glucose in these islets. In agreement with these predictions, whereas glucose induced a robust elevation in NAD(P)H in (ϩ/Ϫ) and (ϩ/ϩ) islets, the rise in NAD(P)H in the HNF-1␣ (Ϫ/Ϫ) islets was markedly attenuated.
Failure of glucose-induced ATP generation could result from a defect occurring at several points in the pathway of glucose metabolism, which occurs in two phases: glycolysis in the cytosol and oxidation of the resulting NADH in the mitochondria. Methyl pyruvate was used to determine at which site the defect occurs in HNF-1␣ (Ϫ/Ϫ) islets. Although pyruvate alone does not stimulate insulin secretion (29), probably due to its restricted transport into the mitochondrion, methyl pyruvate freely permeates the mitochondrial membrane and thereby generates a large amount of ATP that closes the KATP channel in normal ␤-cells (28). In fact, methyl pyruvate at various concentrations suppressed I KATP in HNF-1␣ (Ϫ/Ϫ) islets. In addition, leucine, an amino acid that is metabolized in mitochondria and produces ATP, was also capable of suppressing ATP content was measured in islets from HNF-1␣ (ϩ/ϩ) and (Ϫ/Ϫ) following incubation at 2 (open bars) and 14 (hatched bars) mM glucose, using a quantitative bioluminescence assay. The ATP content of HNF-1␣ (ϩ/ϩ) islets was 30% higher in islets cultured at 14 mM glucose compared with islets cultured at 2 mM glucose (n ϭ 11 mice, p Ͻ 0.05). In contrast, the ATP content of HNF-1 ␣ (Ϫ/Ϫ) islets was 10% lower in islets incubated at 14 mM glucose than in islets incubated at 2 mM glucose (n ϭ 5 mice, p Ͼ 0.05).
FIG. 11. Metabolic signaling pathways in the ␤-cell. The figure shows the fate of representative secretagogues: glucose and glyceraldehyde enter glycolysis, methyl pyruvate (me-pyruvate) generates ATP after entering the tricarboxylic acid (TCA) cycle in the mitochondrion and tolbutamide depolarizes the ␤-cell membrane after binding to the sulfonylurea receptor. An increase in the ATP/ADP ratio leads to depolarization of the ␤-cell membrane by closing the KATP channel. Also shown are the sites of action of two inhibitors of energy production; IAA inhibits glycolysis at the level of GAPDH, and FCCP inhibits the terminal steps of oxidative phosphorylation (Ox-Phos). Other abbreviations used include glyceraldehyde 3-phosphate (GAH-P) and 1,3bisphosphoglycerate (P-GP).
KATP currents in the homozygous mutant ␤-cell. Furthermore, the reduction in MgG fluorescence intensity in (Ϫ/Ϫ) islets stimulated by 10 mM KIC was equivalent to that in control islets suggesting that mitochondrial ATP generation is intact in the mutant islets. Taken together, these data indicate that under appropriate conditions the mitochondria in the (Ϫ/Ϫ) ␤-cell are capable of synthesizing ATP and suggest that the defect in HNF-1␣ (Ϫ/Ϫ) ␤-cells is either in generating or transporting NADH into mitochondria.
Since decreased production of ATP could be secondary to reduced formation of NADH, NAD(P)H fluorescence was measured in islets following stimulation with glucose and methyl pyruvate. The increase in fluorescence after glucose stimulation was significantly lower in the (Ϫ/Ϫ) islets than in control islets, whereas the responses to methyl pyruvate are similar. These findings confirm that mitochondrial NADH generation is not reduced in islets from HNF-1␣ (Ϫ/Ϫ) mice, again suggesting that cytoplasmic NADH generation or its transfer into the mitochondrion is abnormal. The demonstration of reduced glucose utilization in HNF-1 (Ϫ/Ϫ) islets in response to glucose stimulation also supports the hypothesis that the impairment in ATP generation in mutant islets results from a defect in cytosolic glucose metabolism. The 3 H of the fifth carbon of the labeled glucose yields a tritiated water molecule during the reaction catalyzed by enolase (30,31), implying that the defect in glycolytic signaling is proximal to the enolase reaction. Although the precise nature of the signaling defect remains elusive, it is possible that a reduction in the expression of one or more of the enzymes of the glycolytic pathway proximal to the enolase step accounts for the defects observed in glucose-stimulated signal transduction.
The HNF family of transcription factors has been implicated in the regulation of key glycolytic enzymes in the liver (32,33), and expression of multiple HNF subtypes including HNF-1␣ has been reported in pancreatic islets and insulinoma cells (34,35). The ability of mitochondrial substrates such as methyl pyruvate to 1) increase NAD(P)H, 2) block KATP, 3) elevate intracellular [Ca 2ϩ ] i , and 4) stimulate insulin secretion in the HNF-1␣ (Ϫ/Ϫ) islets holds out the hope that similar agents could have therapeutic benefit in the treatment of HNF-1␣ diabetes or perhaps even other forms of type 2 diabetes.