Redox Control of Exocytosis: Regulatory Role of NADPH, Thioredoxin, and Glutaredoxin

Rat β-cells were purified from male Wistar rats after collagenase digestion of the pancreata, handpicking of the islets of Langerhans, dissociation of islets into single cells by trypsin in calcium-free medium, and autofluorescence-activated cell sorting (7,21). Cell purity and viability were ≥90%. β-Cells were reaggregated briefly and cultured for 16 h in serum-free Ham’s F10 medium containing 1% BSA and 10 mmol/l glucose before experimentation. MIN6 cells (obtained from Drs. H. Ishihara and Y. Oka, Sendai, Japan), which are known to be glucose responsive in terms of metabolism and insulin release (22), were cultured in Dulbecco’s modified Eagle’s medium containing 25 mmol/l glucose and 15% heat-inactivated FCS. Cells were only used below passage number 33. INS-1E cells (passage 65) were cultured in RPMI 1640 medium supplemented with 10% FCS, 100 μg/ml penicillin, 100 μg/ml streptomycin, and 50 μmol/l 2-mercaptoethanol. PC12 cells were cultured in RPMI 1640 medium supplemented with 10 mmol/l HEPES, 1 mmol/l Na-pyruvate, 5% horse serum, 10% FCS, 100 IU/ml penicillin, and 0.1 mg/ml streptomycin. For RNA analysis, expression of GRX1 and TRX1 was studied in freshly isolated islet β-cells and non–β-cells and snap-frozen liver, muscle, fat, brain, pituitary, lung, and kidney from male Wistar rats. For Western blots, islets and control tissues were taken from female Sprague Dawley rats.

Pyridine nucleotides were extracted from batches of 1 × 10⁵ rat β-cells and 2 × 10⁵ MIN6 cells and quantified by enzymatic cycling (11). After a 30-min incubation in 0.5 ml of Earle’s HEPES buffer (pH 7.35) containing 0.5% BSA and the indicated glucose concentration, cells were briefly centrifuged, and the supernatant fluid was aspirated. Pellets were frozen in liquid N₂ and homogenized by sonication for 1 min on ice in 50 μl 40 mmol/l NaOH/5 mmol/l cysteine solution. Sonicates were microfuged (30 s at 4°C), after which the supernatant fluid was distributed as follows: 10 μl + 5 μl 0.2M HCl (assay for NADP⁺), 10 μl + 5 μl 40 mmol/l NaOH/5 mmol/l cysteine (NADPH), 10 μl + 5 μl 0.04M HCl (assay for NAD⁺), and 15 μl undiluted extract (NADH analysis). After heating the samples for 10 min at 60°C, the enzymatic cycling reaction (1 h, 37°C) was started by adding 100 μl cycling reagent mixture per test sample. Enzymes were purchased from Boehringer Biochemicals, Mannheim, Germany. To lower the blank values of the NAD cycle, the malate and alcohol dehydrogenase stocks were treated by charcoal adsorption before use. The cycling reaction was stopped by heating for 3 min in a boiling water bath and 10 min centrifugation at 3,000 rpm. The indicator reaction was carried out fluorimetrically (excitation 340 nm/emission 460 nm) on a Hitachi F-2000 fluorimeter (Hitachi, Tokyo, Japan) using 90 μl of the cycling reaction product and adding 900 μl indicator reagent, measuring the amount of 6-phosphogluconolactone (NADP-cycle) or malate (NAD-cycle) formed during the cycling step. The appropriate pyridine nucleotide standard series were extracted in the same media as the unknown samples and run in parallel reaction tubes through the cycling and detection steps.

Insulin release was measured in the supernatant buffer of rat β-cells and MIN6 cells after 30-min static incubations in the same conditions as used for the pyridine nucleotide measurements. Cellular insulin content was measured after extraction of the cell pellet in 0.25% BSA in 2M acetic acid. Media and cellular insulin content was quantified via radioimmunoassay (21). Basal insulin release, determined at 1 mmol/l glucose concentration was subtracted from all release values to obtain glucose-induced release data.

Isolated β-cells were prepared and whole-cell capacitance measurements were performed using EPC7/9 patch-clamp amplifiers with the software suite Pulse + X-Chart Extension (version 8.6 or later; HEKA, Lambrecht-Pfalz, Germany), as previously described (23,24). Primary islet cells or PC12 cells were seeded in Nunc plastic petri dishes and were used for experiments the following day. All chemicals were of analytical grade (Sigma). The extracellular solution contained 138 mmol/l NaCl, 5.6 mmol/l KCl, 2.6 mmol/l CaCl₂, 1.2 mmol/l MgCl₂, 5 mmol/l HEPES, and 5 mmol/l glucose (pH 7.4 with NaOH). In Figs. 3A–F and 7, 20 mmol/l tetra-ethyl-ammonium substituted for equimolar amounts of NaCl to block outward K⁺ currents that otherwise obscured the (smaller) inward Ca²⁺ currents. The intracellular solution in Figs. 3A–F and 7 contained 125 mmol/l Cs-glutamate, 10 mmol/l CsCl, 10 mmol/l NaCl, 1 mmol/l MgCl₂, 5 mmol/l HEPES, 3 mmol/l Mg-ATP, 0.1 mmol/l cAMP, and 0.05 mmol/l EGTA, with NADPH added as indicated (pH 7.2 with CsOH). Cs⁺ salts were used to block outward K⁺ currents. In Fig. 3G and H and Fig. 4, the control intracellular solution consisted of 125 mmol/l K-glutamate, 10 mmol/l KCl, 10 mmol/l NaCl, 1 mmol/l MgCl₂, 5 mmol/l HEPES, 3 mmol/l Mg-ATP (except in Fig. 4A and B), 0.1 mmol/l cAMP, 10 mmol/l EGTA, and 9 mmol/l CaCl₂, with pyridine nucleotides added as indicated in text or figures (pH 7.2 with KOH). Recombinant rat GRX and TRX (IMCO, Stockholm, Sweden) were directly dissolved in the patch electrode solution to their final concentrations.

The mRNA expression profile of rat purified β-cells as well as from control tissues (liver, muscle, fat, brain, lung, pituitary, and kidney) were analyzed using rat 230A microarrays from Affymetrix (Santa Clara, CA). Total RNA was used to prepare biotinylated cRNA according to the standard Affymetrix protocol. Briefly, total RNA was reverse transcribed using the SuperScript Choice System (Invitrogen, Carlsbad, CA) via oligo-dT primers containing a T7 RNA polymerase promotor site. Complementary DNA was in vitro transcribed and labeled using the RNA transcript labeling kit (Enzo, Farmingdale, NY). The concentration of labeled cRNA was measured using the NanoDrop ND-1000 spectrophotometer. Labeled cRNA was fragmented in a fragmentation buffer during 35 min at 94°C. The quality of labeled and fragmented cRNA was analyzed using the Bioanalyzer 2100 (Agilent, Waldbronn, Germany). Fragmented cRNA was hybridized to the array during 16 h at 45°C. The arrays were washed and stained in a fluidics station (Affymetrix) and scanned using the Affymetrix 3000 GeneScanner. Raw data were analyzed using the Affymetrix GCOS software. Signal intensities were scaled using the global scaling method, taking 150 as target intensity value. Scaling factors were less than three times different from each other in all used arrays. Signal intensities for GRX and TRX were compared between individual arrays from β-cells (n = 3) on the one hand and the eight control tissues (islet non–β-cells, liver, muscle, fat, brain, pituitary, lung, and kidney; n = 3 each), allowing GCOS analysis of 72 pairwise comparisons between β-cell samples and the tissue samples. Mean GRX1-to-actin and TRX1-to-actin signal ratios in β-cells and other tissues were also compared using two-tailed unpaired Student’s t tests, taking P < 0.01 as threshold for significance.

To confirm the microarray data by quantitative RT-PCR, cDNA was prepared from total RNA using random hexamer primers and moloney murine leukemia virus reverse transcriptase (RevertAid H Minus First Strand cDNA synthesis kit; Fermentas, St. Leon-Rot, Germany), following the manufacturer’s protocol. Primers and dual-labeled probes (Table 1) were designed using OligoAnalyzer 3.0 software (available from http://biotools.idtdna.com/analyzer/). β-Actin primers and dual-labeled probe were synthesized by Proligo (Paris, France) and those for GRX1 and TRX1 by Eurogentec (Seraing, Belgium). Real-time PCR was performed on a Rotor-Gene 3000 instrument (Corbett Research, Mortlake, Australia), using ABsolute QPCR Mix (ABgene, Epsom, U.K.), according to the manufacturer’s instructions. Cycle threshold values were determined by Rotor-Gene 6.0.16 software. All samples were amplified in duplicate reactions, and every experiment was repeated using cDNA samples from independent tissue/cell preparations. In each run, two no-template controls were used for the transcripts under investigation to test for contamination of reagents. Real-time PCR cycling conditions were the following for GRX1: initial denaturation/polymerase activation step at 95°C for 15 min, followed by 45 cycles of denaturation at 95°C for 20 s and annealing/elongation at 57°C for 60 s; for TRX1: the same, but annealing/elongation step at 60°C. For each transcript, serial dilutions of a control template were used to generate standard curves (cycle threshold versus cDNA concentration). Reaction efficiencies (E) were calculated from the slopes of the standard curves, according to the following equation: E = 10(−1/slope). All transcripts were amplified with high efficiencies (E ranging from 1.83 to 2.00), and linear regression of all standard curves showed a high linear correlation r² > 0.990). The relative expression of target mRNA levels were calculated as a ratio relative to the β-actin reference mRNA (25).

Tissue samples (lung, kidney, liver, brain, skeletal muscle [m. soleus], and fat [intraperitoneal]) were collected from female Sprague Dawley rats. Islets of Langerhans were isolated from rats by collagenase digestion of pancreas followed by handpicking. Tissues and islets were quickly frozen in liquid nitrogen and stored in −80°C until homogenization. Tissue samples, islets, and INS-1E cells (obtained from Dr. C.B. Wollheim, University of Geneva, Geneva, Switzerland) were homogenized in 50 mmol/l TES buffer, pH 7.4, containing 250 mmol/l sucrose, 1 mmol/l EDTA, 0.1 μmol/l EGTA, and complete protease inhibitor cocktail (Roche, Stockholm, Sweden) on ice using an ULTRA-TURRAX (IKA-WERK; Janke & Kunkel, Staufen, Germany) or by repeated aspiration through a needle (Ø0.4 μm). After homogenization, tissue debris was removed by centrifugation (20,000g, 15 min, 4°C), and the supernatants were collected. Protein content of the extract was measured using a kit (DC protein assay; Bio-Rad, Hercules, CA). Forty micrograms of total protein content were electrophoresed on 12% SDS-PAGE, and the separated proteins were transferred onto a polyvinylidine fluoride membrane (Amersham Pharmacia Biotech, Uppsala, Sweden). The membrane was blocked with 5.0% nonfat dry milk in Tris-buffered saline with Tween (pH 7.4; 0.15 mol/l NaCl, 10 mmol/l Tris-HCl, and 0.1% Tween 20) for 1 h at room temperature or overnight at 4°C. After blocking, the membrane was incubated with polyclonal rabbit anti-GRX (1:250) and anti-TRX antibody (1:1,000) (a gift from Dr. Barcena, University of Cordoba, Cordoba, Spain [26]) and monoclonal mouse anti-β actin (1:15,000) in Tris-buffered saline with Tween for 1 h at room temperature. After washing with Tris-buffered saline with Tween, the membrane was incubated with peroxidase-linked anti-rabbit IgG (dilution 1:5,000; Amersham Bioschience) or peroxidase-linked anti-mouse IgG (dilution 1:20,000; Jackson ImmunoResearch,West Groove, PA) for 1 h at room temperature. Detection was done by an enhanced chemiluminescence kit (Amersham). Quantification of protein band densities was calculated by Scion Image software (Scion, Frederick, MD).

Dispersed islet cells were attached to poly-d-lysine–coated glass coverslips, fixed in 4% paraformaldehyde (Merck, Darmstadt, Germany) for 1 h at 4°C, postfixed in methanol (Merck) for 30 s, and treated with 50 mmol/l ammoniumchloride (Merck) to block free aldehyde groups. Immunofluorescent double staining was performed for insulin with TRX or GRX, respectively, using a guinea pig anti-insulin antibody (a gift of Dr. Van Schravendijk, Diabetes Research Center, Brussels, Belgium) diluted 1:500, and the TRX/GRX antibodies were diluted 1:10 for 60 min at room temperature. Binding was detected with Cy3-labeled anti-rabbit Ig and fluorescein isothiocyanate–labeled anti-guinea pig Ig secondary antibodies (Jackson Immunoresearch Laboratories). Immunofluorescent double staining for SNAP25-TRX and SNAP25-GRX was performed using 1:100 diluted anti–SNAP-25 (Synaptic Systems, Göttingen, Germany) for 60 min at room temperature on membrane patches of dispersed islet cells (27). Controls included the omission of the primary antibody. Preparations were counterstained with 4′,6-diamidino-2-phenylindole, and digital images images were obtained with a Zeiss Axioplan II microscope equipped with a Sensys cooled CCD camera and Quips imaging software (Vysis, Downers Grove, IL).

Intact pancreatic islets were fixed in 4% paraformaldehyde (Merck, Stockholm, Sweden) with 0.5% glutaraldehyde for 1 h and were then imbedded in Lowicryl and cut into ultrathin sections (60–80 nm) using an LKB MK III Ultratome. GRX immunoreactivity was visualized using a CM 10 electron microscope (Philips, Eindhoven, the Netherlands) after incubation with a rabbit-raised primary antibody (1:60) followed by incubation with a donkey anti-rabbit secondary antibody conjugated with 12-nm gold particles. The subcellular gold particle density was analyzed using the software package from Scion.

The biochemical analysis consisted of cellular extractions and determination of pyridine nucleotide levels after 30-min incubations at different glucose concentrations, using a sensitive enzymatic cycling method. Overall rates of glucose oxidation accelerated about fourfold in both fluorescence-activated cell sorter–purified rat β-cells and in the glucose-responsive murine β-cell line MIN6 when the extracellular substrate level was raised from 1 to 10 mmol/l (data not shown). In both primary β-cells and MIN6 cells, glucose consistently increased the cellular NADPH content (Fig. 1A). Assuming a water space of ∼0.6 pl per rat β-cell (28) and a uniform nucleotide distribution in the cellular water space, the average NADPH concentration in primary rat β-cells was noted to increase from 42 μmol/l in 1 mmol/l glucose to ∼69 μmol/l at 10 mmol/l glucose, which is close to previous biochemical measurements of the NAPDH content in rat islets (13). The steepest increase occurred between the physiological interval of 5–10 mmol/l glucose (Fig. 1A), while a further rise up to 80 μmol/l was observed in the presence of 20 mmol/l glucose. The sum of NADPH and NADP⁺ was unchanged when glucose levels were increased, indicating that a shift of redox state from NADP⁺ to NADPH occurs, as was expected (Fig. 1A). Elevation of glucose caused a significant increase in NADPH and a decrease in NADP⁺ in both primary β-cells (Fig. 1A) and INS1 cells (Fig. 1B). Consequently, the glucose-induced rise in the NADPH-to-NADP⁺ ratio (Fig. 2A) was more prominent (almost fourfold over the whole concentration range tested) than that of NADPH content (twofold increase between 1 and 20 mmol/l glucose). In rat β-cells, the ratios averaged 0.40 ± 0.05 vs. 0.93 ± 0.20 at 1 and 10 mmol/l glucose, respectively (n = 5, P = 0.03). Insulin release was measured during static incubations from the same cell types during the same time period as the analysis of pyridine nucleotide content (Fig. 2B). The cellular NADPH-to-NADP⁺ ratio correlated strongly to the measured rates of insulin release (R² = 0.87, P = 0.0001; Fig. 2C). Interestingly, the dose-response curves of the glucose-dependent rise in the NADPH-to-NADP⁺ ratio and insulin release were hyperbolic in MIN6 cells and sigmoidal in rat β-cells (Fig. 2A and B).

To assess if changes in the cytosolic NADPH concentration directly influence exocytosis, we next used the standard whole-cell configuration of the patch-clamp technique combined with capacitance measurements of exocytosis (23). As illustrated in Fig. 3, addition of 100 μmol/l NADPH stimulated the increase in membrane capacitance elicited by trains of ten voltage-clamp depolarizations by 84 and 102% over that observed using the control patch electrode solution in both mouse and rat β-cells, respectively. The dose and time dependency of this effect is compatible with a rapid and physiological effect on insulin release for the following reasons. First, significant acceleration of exocytosis was detected well within 1 min of injection of the pyridine nucleotide, indicating a prompt effect on the insulin release machinery (24). Second, the concentration-response curve (Fig. 3G and H) indicates that 100 μmol/l NADPH is a saturating concentration, and half-maximal activation is attained at 45 μmol/l (i.e., close to the mean concentration estimated in intact β-cells during incubations with physiological glucose concentrations). In the absence of ATP, Ca²⁺-elicited exocytosis was suppressed and NADPH failed to stimulate release rates (Fig. 4A and B). Our interpretation of this data is that the action of the pyridine nucleotide is distinct from, and requires, ATP-dependent priming of the dense core granules (23,24). Since glucose not only increases NADPH but also decreases NADP⁺ (Fig. 1A), thereby changing the NADPH-to-NADP⁺ ratio to a greater extent than that of the absolute NADPH concentration, we next tested if coinjection of NADP⁺ could also affect exocytosis. Figure 4C and D shows that the effect of the saturating concentration of NADPH (100 μmol/l) was completely abolished by a threefold molar excess of NADP⁺, while injection of the oxidized nucleotide alone did not influence exocytosis. Since the NADPH-to-NADP⁺ ratio shifts in intact rat and mouse β-cells from ∼0.4 in the absence of glucose to 1.5 in the presence of 20 mmol/l glucose (Fig. 2A), NADP⁺ appears to be an antagonist of the effect of NADPH so that glucose has two actions on the putative pyridine nucleotide–sensitive target(s): 1) generation of an activator and 2) removal of an inactivator. This situation would be analogous to the glucose-induced shift in the adenine nucleotides ADP and ATP (29), which have inverse effects on K_ATP channels (30).

To further clarify the roles of pyridine nucleotides as putative signaling molecules for exocytosis, glucose-induced changes in NADH were measured and effects of NADH on exocytosis were quantified. While glucose raised the content of NADH in rat β-cells (Fig. 1C) and MIN6 cells (Fig. 1D), this effect was small over the tested range of glucose concentrations. Furthermore, there was no detectable decrease in NAD⁺ in MIN6 cells, and the measured NADH-to-NAD⁺ ratios in these cells were between 3 and 10 times lower than in rat β-cells, the latter being close to that previously reported in rat islets ex vivo (13). Consequently, in the pooled dataset of β-cells and MIN6 cells, the cellular NADH-to-NAD⁺ ratio was not correlated to the rate of insulin release (Fig. 2D), but a positive correlation might be present when data of primary β-cells are considered separately from those of MIN6 cells. Moreover, the glucose depencency of the NADH-to-NAD⁺ ratio was linear rather than hyperbolic or sigmoidal (Fig. 2A). At the electrophysiological level, the addition of 300 μmol/l NADH in the patch-clamp analysis did not mimic the effects of NADPH. As can been seen in Fig. 4E and F, this condition did not stimulate exocytosis; in fact, a modest but consistent inhibition was observed. As such, these data not only exclude a nonspecific reaction but makes it also unlikely that the effects on insulin exocytosis are causally related to the glycolytic production of NADH (18,19). To address the issue of cellular specificity, our experimental design was repeated in catecholamine-releasing PC12 cells, a well-characterized model of regulated exocytosis (Fig. 4G and H). Similarly, in this model, the intracellular addition of NADPH elevated Ca²⁺-elicited membrane capacitance by 50%, suggesting that this pyridine nucleotide plays a novel and thus far unrecognized role in the potentiation of Ca²⁺-regulated exocytosis.

To explore the mechanism of the NADPH-mediated actions on the exocytotic machinery, we assessed the possibility that particular redox acceptor proteins (20) might mediate the effect of NADPH on exocytosis. In line with this hypothesis, we examined the expression level of the GRX1 and TRX1 genes at both the mRNA (Fig. 5A) and protein (Fig. 5B) levels. Using Affymetrix microarrays, we observed that expression signals of GRX1 mRNA were significantly higher in pancreatic β-cells than in flow-sorted islet non–β-cells as well as seven tested extrapancreatic tissues (Fig. 5A; P < 0.01 for all comparisons). Moreover, all 72 pairwise comparisons of the 3 β-cell samples, with 24 samples from other cells and tissue (Affymetrix GCOS program), gave “increased in β-cells” calls, the mean signal log ratio of these comparisons being 1.7 and the lowest value 0.9. On the contrary, TRX1 mRNA signals in β-cells were significantly lower than those in liver, fat, lung, and kidney (P < 0.001); Fig. 5B). These data were corroborated by quantitative RT-PCR analaysis, with a few minor differences like a higher GRX1 mRNA signal in liver and a lower TRX1 signal in adipose tissue. At the protein level, we observed high expression of GRX in both pancreatic islets and in INS1 cells as well as in brain (Fig. 5B) and intermediate signals in adipose tissue. Interestingly, the TRX1 protein level was highest in islets and INS1 cells and lowest in liver and muscle, while the inverse situation was observed for mRNA levels.

To assess whether islet TRX and GRX can be localized in β-cells, we performed light microscopic analysis of dissociated islet rat cells. Figure 6A shows that TRX and GRX immunostaining of β-cells using polyclonal TRX and GRX antibodies (26) exhibits a punctuate distribution pattern. Higher-resolution immunostaining of the inner face of the plasma membrane suggests the two redox proteins are present in a nonuniform clustered distribution that is disctinct from SNAP25 in lipid rafts (Fig. 6B). At the electron microscopic level, it was further documented that GRX preferentially localizes in particular subcellular domains such as the nucleus and secretory granules (Fig. 6C and D).

To assess the potential functional relevance of GRX and TRX for insulin secretion, we next studied the effect of an altered GRX-to-TRX ratio upon NADPH-induced exocytosis by coadministration of GRX or TRX together with NADPH (100 μmol/l) in the intracellular pipette solution. As is shown in Fig. 7A and B, a supplement of GRX (1 μmol/l) further augmented NADPH- and depolarization-elicited exocytosis by 54% (P < 0.05); on the contrary, addition of extra TRX (1 μmol/l) completely counteracted the stimulatory action of NADPH and reduced the exocytotic response by 66% (P < 0.01). It was verified that the stimulation of insulin release by GRX was specific for NADPH. Indeed, when NADH (100 μmol/l) was included in the intracellular solution instead of NADPH, GRX failed to enhance depolarization-evoked exocytosis (172 ± 28 fF, n = 6 and 210 ± 44, n = 13, with and without GRX, respectively, means ± SE). Moreover, in the presence of a permissive NADPH concentration (100 μmol/l) but in the presence of a threefold molar excess of NADP⁺, GRX failed to enhance depolarization-evoked exocytosis (average capacitance increase 405 ± 35 and 408 ± 32 fF without and with GRX, respectively, n = 5).

The activity of GRX is under the control of the glutathione system (31) to mediate several important NADPH-dependent cellular processes, such as DNA synthesis (32), protein folding and transcriptional regulation (33), signal transduction (31), as well as defense against oxidative stress and apoptosis (34,35). Therefore, it can be anticipated that the reduced form of glutathione, acting as an intermediate between NADPH and reduced GRX, would also potentiate Ca²⁺-dependent exocytosis. As is shown in Fig. 7C and D, intracellular addition of glutathione (1 mmol/l) stimulated exocytosis by 46% (P < 0.05).

The present report has studied redox regulation of insulin release following the hypothesis that the reducing power of the cells, in combination with the expression of the redox proteins GRX1 and TRX1, is critical for the amplification of the glucose-induced Ca²⁺-mediated signaling pathway. Our biochemical analysis of the four pyridine nucleotides confirms earlier studies (12,13) showing that both NADPH/NADP⁺ and NADH/NAP⁺ acutely increase in β-cells when glucose in the extracellular medium rises. To our knowledge, however, this study is the first to demonstrate via electrophysiological recordings that an experimentally imposed rise in the NADPH-to-NADP⁺ ratio in itself does cause exocytosis, amplifying the Ca²⁺- and ATP-mediated response. The specificity of this phenomenon is underlined by showing that 1) alterations in the NADH-to-NAD⁺ ratio do not have such an effect and 2) that the effectiveness of the NADPH-to-NADP⁺ ratio upon exocytosis can be further modulated by GRX1 and TRX1, the two best-known acceptor proteins for NADPH electrons (20).

GRX1 has previously been proposed to play a role in exocytosis on basis of its immunohistochemical localization in synaptic terminals in the neurohypophysis (26), and an increased NADPH-to-NADP⁺ ratio has been linked to the intense wave of exocytosis of cortical granules oocyte fertilization (36). Redox-regulated posttranslational modification of exocytosis-regulating t-SNARE proteins has been proposed to result in an increased rate of maturation, or priming, of secretory vesicles in yeast (37). A similar accelerated formation of mature insulin granules by NADPH/GRX may provide an explanation for the potentiating affect of NADPH on exocytosis evoked by Ca²⁺ and ATP in neuroendocrine cells, and we propose in this report that regulated exocytosis is linked to cellular metabolism via the cellular redox state. The sequence of events seems to be nutrient-induced reduction of NADP⁺ and flow of electrons from NADPH to glutathione, GRX1, and TRX1. This pathway would be particularly important for the pancreatic β-cell, which has to match the extracellular glucose concentration to the rate of insulin release. Indeed, this idea is consistent with previous reports that cataplerosis of citrate (15) as well as pyruvate cycling (17) in β-cells is linked to insulin secretion. Furthermore, β-cells with a high insulin secretory competence upregulate mRNA expression encoding NADPH-generating pathways such as glucose-6-phosphate dehydrogenase (16) and ATP-citrate lyase (16).

Our results (mRNA/protein expression studies and electrophysiological analysis of exocytosis) suggest that both the expression level of GRX/TRX in the β-cell as well as the NADPH/NADP⁺ redox status are important factors for the regulation of exocytosis. It is conceivable that such redox regulatory system is further regulated by accessory proteins such as the thioredoxin regulatory protein (VDUP1), whose expression is glucose dependent (38,39). Our data are of particular interest in the context of human type 2 diabetes, in which there is dysregulation of insulin release by glucose. In addition to initiating electrical activity by ATP-dependent closure of the β-cell K_ATP channels, glucose metabolism also exerts a potentiating action on the insulin release apparatus via a K_ATP channel–independent pathway (5). The latter pathway is impaired in animal models of type 2 diabetes (40) and appears to be defective in human type 2 diabetic patients (41). The identity of the alternative metabolic signal that amplifies insulin secretion has remained an unsettled issue, particularly with respect to the question of whether glutamate (formed from the citric acid cycle intermediate α-ketoglutarate) fulfills this function (42,43). Without being mutually exclusive to the glutamate hypothesis (42), our data suggest that the glucose-mediated generation of NADPH and the reduction of GRX and TRX are newly identified components of redox regulation of exocytosis.

This work was supported by grants from the Swedish Research Council (no. 12234), the European Foundation for the Study of Diabetes, and the Novo Nordisk and Crafoord Foundations to E.R. and from the Fonds voor Wetenschappelijk Onderzoek-Vlaanderen (FWO-Vlaanderen) (G.0130.99), the Juvenile Diabetes Research Foundation (1-2002-801), and the Katholieke Universiteit-Leuven (GOA/2004/11) to F.S. The position of R.I. was partly funded by Lund University, and those of S.D. (research assistant) and F.S. (research professor) were supported by the FWO-Vlaanderen. MIN6 cells were kindly donated by H. Ishihara and Y. Oka (Tohoku University Graduate School of Medicine, Sendai, Japan). INS-1E cells were a gift from Dr. C.B. Wollheim (University of Geneva, Geneva, Switzerland). Affinity-purified polyclonal rabbit anti-rat GRX and TRX antisera were kindly provided by Dr. J.A. Barcena (Department of Biochemistry and Molecular Biology, University of Cordoba, Cordoba, Spain; see [26]).

All experiments were approved by the ethical committees at Lund University, Lund, Sweden; the Katholieke Universiteit, Leuven, Belgium; and the Vrije Universiteit, Brussels, Belguim. We thank G. Stangé for rat β-cell purifications, V. Berger and G. Schoonjans for help with the insulin assays, L. Van Lommel for help with the microarray RNA analysis, L. Eliasson for help with the electron microscopy, S. Hinke for manuscript review, and H. Mulder for valuable discussions.