The Role of Phosphoinositide 3-Kinase/Akt Signaling in Low-Dose Mercury–Induced Mouse Pancreatic β-Cell Dysfunction In Vitro and In Vivo

We purchased 18–25 g male ICR mice from the Animal Center of the College of Medicine, National Taiwan University (Taipei, Taiwan). The Animal Research Committee of College of Medicine, National Taiwan University, conducted the study in accordance with the guidelines for the care and use of laboratory animals. Mice were housed in a room at a constant temperature of 22 ± 2°C with 12-h light-dark cycles. Mice were randomly assigned to pretreatment groups, weighed, and administered HgCl₂, MeHg, or vehicle (oral application by gavage). These groups were given 0, 50, and 500 μg · kg⁻¹ · day⁻¹ HgCl₂ or 20 μg · kg⁻¹ · day⁻¹ MeHg (7 days/week) in the presence or absence of NAC (150 mg/kg) for 2–4 weeks. Each group contained 16 mice. All mice were observed daily for clinical signs. Following collection of blood samples, the animals were exsanguinated, while still under anesthesia, and necropsied. A necropsy examination was performed on all animals from each group. All pathological findings were recorded. A histopathological examination was performed on organs from groups treated with 0 and 500 μg/kg HgCl₂ and 20 μg/kg MeHg.

Blood samples were collected from the peripheral vessels of mice under ketamine anesthesia (85 mg/kg i.p.). Whole blood was centrifuged at 3,000g for 10 min, and plasma was obtained and assayed immediately using the lipid peroxidation assay kit (Calbiochem). An absorbance of 586 nm was measured using an enzyme-linked immunosorbent assay (ELISA) microplate reader.

Blood samples were collected after an overnight fast. Blood glucose levels were determined using the SureStep blood glucose meter (Lifescan, Milpitas, CA). For oral glucose tolerance testing, d-glucose (1 g/kg) was administered by stomach tube after an overnight fast. Blood samples were collected before and 30, 60, 90, 120, and 180 min after delivery of the glucose load.

Islets of Langerhans were isolated by collagenase digestion from the mouse pancreas as previously described (27,28). After separation on a Ficoll gradient, the islets were further purified by handpicking to eliminate any remaining exocrine tissue. Whole islets were maintained in culture medium consisting of RPMI-1640 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin/amphotericin B at 37°C in an atmosphere of 95% air/5% CO₂ before experimentation.

Pancreatic β-cell–derived HIT-T15 cells were cultured in a humidified chamber with a 95% air/5% CO₂ mixture at 37°C and maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum and containing 11.1 mmol/l glucose as described (29). In some experiments, transfection in HIT-T15 cells was performed using the Lipofectin reagent (BD Clontech) according to the recommendations of the manufacturer with 2 μg total DNA (or vector) per sample. The efficiency of transfection (∼80%) was determined by using an equal amount of a plasmid encoding the green fluorescent protein under the cytomegalovirus promoter. Plasmids containing the DN-p85 (δ-p85) or DN-Akt (Akt K179A) (30,31) were kindly provided by Dr. R.H. Chen and Dr. M.L. Kuo (Institutes of Molecular Medicine and Toxicology, National Taiwan University, Taiwan).

HIT-T15 insulin secretion experiments were performed in Krebs Ringer buffer (KRB), and islet secretion studies were performed in islet culture media for a 1-h experimental incubation with a 95% air/5% CO₂ mixture at 37°C, as previously described (32). All experimental agents were mixed together in experimental solutions (KRB or media) and then added to the cells at the start of the experimental incubation. To measure the amount of insulin secreted, aliquots of samples were collected from the plasma or experimental solutions at indicated time points and subjected to insulin antiserum immunoassay according to the manufacturer’s instruction (Mercodia, Uppsala, Sweden).

Cells were washed with fresh media from dishes and cultured in 96-well plates (2 × 10⁵/well), then stimulated with HgCl₂ and MeHg. After 24 h, the medium was replaced with fresh medium containing 30 μl (2 mg/ml) 3-(4,5-dimethyl thiazol-2-yl-)-2,5-diphenyl tetrazolium bromide. After incubation for 4 h, the medium was removed and 1 ml dimethyl sulfoxide added to dissolve blue formazan crystal. An ELISA reader was used for fluorescence detection; absorption was 570 nm.

Intracellular ROS generation was monitored by flow cytometry using peroxide-sensitive fluorescent probe (2′,7′-dichlorofluorescin diacetate [DCFH-DA; Molecular Probes]). In brief, HIT-T15 cells or islets were coincubated with 20 μmol/l DCFH-DA for 15 min at 37°C. In the presence of a proper oxidant, DCFH was oxidized into the highly fluorescent 2′,7′-dichlorofluorescein (DCF). After incubation with the dye, cells and islets were resuspended in ice-cold PBS and placed on ice in darkness for flow cytometry analysis. Using trypsin, the dispersed islets and intracellular peroxide levels were measured with the FACScan flow cytometry device (Becton Dickinson, Franklin Lakes, NJ). More than 100 islets from at least three separate islet isolations were studied for each group. The results were expressed as the mean fluorescence intensity. In some experiments, after 15 min of incubation of DCFH-DA, the islets were washed twice and images captured by a Leica DMIL inverted microscope equipped with a charged-coupled device camera.

Cellular ATP concentration was determined by the ATP bioluminescent assay kit (Sigma), which measures the intracellular ATP level using the ATP-dependent luciferin-luciferase reaction.

Fifty micrograms of protein of each cell lysate were subjected to electrophoresis on 10% SDS-polyacrylamide gels. The samples were then electroblotted on polyvinylidene difluoride membranes. After blocking, blots were incubated with anti-Akt and anti–phospho-Akt (New England BioLabs) antibodies in PBS and 0.1% Tween 20 for 1 h followed by two washes (15 min each) in PBS and 0.1% Tween 20. The membranes were then incubated with horseradish peroxidase–conjugated secondary antibodies for 30 min. Enhanced chemiluminescence reagents (Amersham) were used to depict the protein bands on membranes.

PI3K activity was assayed as described previously (33). Cell extracts were incubated with 2 μg anti-p85 antibody overnight at 4°C. The immunocomplex was precipitated with protein A sepharose for 1 h at 4°C and washed three times with lysis buffer, twice with LiCl buffer, and twice with TNE buffer (10 mmol/l Tris, 100 mmol/l NaCl, and 1 mmol/l EDTA), and then the immunoprecipitated enzymes were subjected to a kinase assay by competitive ELISA, according to the manufacturer’s instructions (Echelon Biosciences, Salt Lake City, UT).

The levels of mercury in blood were determined by cold vapor atomic absorption spectrophotometer combined with the flow-injection analysis system (34), and analysis was performed with a PerkinElmer 5100PC AAS equipped with a PerkinElmer FIAS-200 flow-injection analysis system and AA WinLab software. The detection limit for mercury was ∼0.1 ppb (μg/l).

The values in the text are given as means ± SE. The significance of difference was evaluated by the paired Student’s t test. When more than one group was compared with one control, significance was evaluated according to one-way ANOVA. Statistical difference was indicated by P < 0.05.

To understand whether mercury-induced oxidative stress could affect the function of pancreatic β-cells, we investigated the in vitro effects of low-concentration HgCl₂ and MeHg (0.2–1 μmol/l) on the β-cell–derived HIT-T15 cells. Exposure of cells to mercury for 24 h did not affect cell viability (data not shown). Measurement of DCF fluorescence as an indicator of ROS formation, after exposure of HIT-T15 cells to 0.5 and 1 μmol/l mercury for 1 h, significantly increased ROS levels, which were inhibited by the antioxidant NAC (1 mmol/l) (Fig. 1A). Furthermore, measurement of insulin secretion from HIT-T15 cells after exposure to mercury under various concentrations of glucose (5.5–30 mmol/l) showed that HgCl₂ or MeHg (0.5 and 1 μmol/l, respectively) significantly inhibited insulin secretion (Fig. 1B). Cells treated with NAC effectively prevented the inhibition of insulin secretion in HIT-T15 cells exposed to mercury (Fig. 2A). The PI3K inhibitor LY294002 (5 μmol/l) also effectively prevented the mercury-induced inhibition of insulin secretion in HIT-T15 cells (Fig. 2B). In a 4-h incubation, 5 μmol/l LY294002 alone did not affect the insulin secretion; however, it was capable of increasing the insulin secretion in HIT-T15 cells at concentration of 10 μmol/l (131.6 ± 1.2% of control, n = 5).

To further evaluate the involvement of PI3K/Akt signaling in the responses triggered by mercury-induced oxidative stress, the PI3K and Akt activities in HIT-T15 cells were examined by determining the PI3K activity and phosphorylated status of Akt. Immunoprecipitates with the anti-p85 antibody revealed a substantial increase in PI3K activity 5 min after the exposure of mercury (1 μmol/l) (Fig. 3A, a). The significant activation of Akt was also shown in mercury-treated cells (Fig. 3A, b). LY294002 markedly reduced the PI3K activity and phosphorylation of Akt induced by mercury in HIT-T15 cells (data not shown). NAC effectively inhibited the mercury-induced Akt phosphorylation (Fig. 3A, b) but not the increased PI3K activity (Fig. 3A, a).

The next aim of the investigation was to ascertain whether inhibition of PI3K/Akt activities might affect the mercury-induced inhibition of insulin secretion in HIT-T15 cells. To address this issue, cells were transfected with a dominant-negative form of the regulatory subunit (p85) of PI3K (DN-p85) or a dominant-negative form of Akt (DN-Akt) (35). As shown in Fig. 3B, a, DN-p85 or DN-Akt transfection markedly decreased the phosphorylation of Akt. The levels of Akt protein expression in DN-Akt–transfected cells were increased compared with control. Blocking the PI3K and Akt activities with DN-p85 and DN-Akt greatly prevented the 1 μmol/l mercury–triggered insulin secretion inhibition (Fig. 3B, b). The control pcDNA3 transfection did not affect the mercury-induced responses. However, blocking the PI3K activity with DN-p85 did not affect ROS generation by mercury (Fig. 3B, c). Therefore, these results imply that both PI3K and ROS independently regulated Akt signaling–related mercury-induced insulin secretion inhibition.

Exposure of isolated mouse islets to 1 μmol/l mercury for 1 h increased DCF-sensitive ROS formation, which was prevented by 1 mmol/l NAC (Fig. 4). The role of PI3K/Akt signaling was further investigated. A significant phosphorylation of Akt in isolated mouse islets was shown 1 h after the treatment of mercury (1 μmol/l) (Fig. 4B, a). NAC and LY294002 (5 μmol/l) markedly reversed the increase of Akt phosphorylation by mercury (Fig. 4B, a). Furthermore, mercury (1 μmol/l) was capable of inhibiting the insulin secretion in mouse islets, which could also be reversed by NAC and LY294002 (Fig. 4B, b).

The next aim of the investigation was to ascertain whether mercury-induced oxidative stress might affect the function of pancreatic β-cells in vivo and the involvement of PI3K/Akt signaling. A 2- or 4-week exposure to low-dose HgCl₂ (50 or 500 μg · kg⁻¹ · day⁻¹) or MeHg (20 μg · kg⁻¹ · day⁻¹) (oral application by gavage) to mice caused a significant decrease in plasma insulin levels (Fig. 5A). Moreover, a significant phosphorylation of Akt was shown in islets isolated from mice exposed to mercury for 2 weeks (Fig. 5B). Mercury-exposed mice also showed an elevation in blood glucose (Fig. 6A) and glucose intolerance (Fig. 6B). Moreover, the result of plasma lipid peroxidation assay after exposure to HgCl₂ or MeHg to mice showed that plasma malondialdehyde levels were significantly increased at 2–4 weeks (Fig. 7). These mercury-triggered responses in mice could also be reversed by the treatment of NAC (150 mg/kg) (Figs. 5–7).

Mice exposed to mercury over a 2- to 4-week period had elevated blood mercury levels ranging from ∼4.5 to 60 μg/l (Fig. 8A). There was a similar blood mercury range for treatment with 500 μg/kg HgCl₂ and 20 μg/kg MeHg. The elevated blood mercury and decreased plasma insulin and elevated blood glucose levels were gradually reversed after mercury withdrawal (Fig. 8A, blood mercury; Fig. 8B, plasma insulin; blood glucose [mg/dl]: HgCl₂ 4 weeks 144.2 ± 8.4, withdrawal 2 weeks 112.4 ± 5.8, withdrawal 4 weeks 89.7 ± 6.7, n = 8; MeHg 4 weeks 122.1 ± 7.7, withdrawal 2 weeks 93.6 ± 5.4, withdrawal 4 weeks 82.9 ± 4.2, n = 8; age-matched controls 81.6 ± 4.9, n = 8).

There were no clinical signs and grossly detectable lesions (including brain, liver, kidney, lung, heart, adrenal, gonad, and intestine) related to the administration of mercury for 2–4 weeks in mice, and there was no effect on whole-body weight (data not shown).

Oxidative stress has been demonstrated to be produced under diabetic conditions and was likely involved in the progression of pancreatic β-cell dysfunction found in diabetes (3,36). It has been demonstrated that mitochondrial ROS suppressed glucose-induced insulin secretion from β-cells (2). Despite several studies showing that mercury could induce oxidative stress with subsequent oxidative damage in several organs or systems (16–19,37), the precise action and mechanism of mercury-induced oxidative stress on the pancreatic β-cell function is unknown. To investigate the effects of mercury-induced oxidative stress on the function of pancreatic β-cells, we first investigated the in vitro effects of HgCl₂ and MeHg on the β-cell–derived HIT-T15 cells and isolated mouse islets. Our results showed that submicromolar-concentration exposure to mercury was capable of triggering ROS production and decreasing insulin secretion in HIT-T15 cells and isolated islets, which could be prevented completely by treatment of cells with antioxidant NAC, indicating that ROS was involved in mercury-induced β-cell insulin secretion inhibition. On the other hand, changes in glucose levels to insulin secretion are related to intracellular ATP levels. Following glucose entry into pancreatic β-cells, products of glucose metabolism enter the mitochondrial respiratory chain, which uses them to generate ATP. The increase in ATP inhibits ATP-sensitive potassium channels, which in turn stimulate insulin secretion (38). Here we found that increasing doses of mercury did not change the intracellular ATP levels in HIT-T15 cells after exposure to 0.2–1 μmol/l HgCl₂ or MeHg at 1–4 h (data not shown), and hence, it may not be a primary toxic mechanism in dysfunction of pancreatic β-cells induced by low-concentration mercury. Next, we investigated the in vivo effects of mercury-induced oxidative stress on the insulin secretion and blood glucose regulation in mice. The results showed that a 2- or 4-week oral exposure to low-dose mercury caused an increase in plasma lipid peroxidation levels, a decrease in plasma insulin levels, and an elevation in blood glucose and glucose intolerance. NAC could prevent these mercury-induced responses. These findings indicate that oxidative stress is involved in the toxic mechanism in mercury-induced hypoinsulinemia and hyperglycemia. In addition, it has been found that endogenous nitric oxide (NO) production could be stimulated by glucose and implicated this glucose-stimulated NO production in physiological glucose-stimulated insulin secretion (32). Whether dysfunction or deregulation of the NO system is involved in mercury-induced β-cell damage may need to be clarified in the future.

Some studies (6,7) have suggested that as insulin resistance developed in peripheral tissues, a potential result of impaired PI3K activation in β-cells promoted a linked increase in insulin secretion to maintain glucose homeostasis. Eto et al. (39) have also indicated that PI3K suppressed glucose-stimulated insulin secretion by affecting postcytosolic Ca²⁺ elevation signals. It has been shown that redox stress regulated cell proliferation and apoptosis of human hepatoma through PI3K/Akt pathway (40). Angiotensin II–induced Akt phosphorylation has been demonstrated to be mediated by ROS, which contributed to angiotensin II–induced hypertrophy in vascular smooth muscle cells (10). However, the role of PI3K/Akt signaling pathway in the oxidative stress–induced pancreatic β-cell dysfunction is still unclear. The present work showed that low-dose mercury could enhance the phosphorylation of Akt, which was reversed by NAC. Moreover, the inhibition of PI3K/Akt activity with the PI3K inhibitor LY294002, or by expressing the dominant-negative p85 or Akt, effectively attenuated the inhibition of insulin secretion induced by mercury. LY294002 was capable of increasing the insulin secretion in HIT-T15 cells at a concentration of 10 μmol/l. However, unexpectedly, we found that NAC could not inhibit mercury-increased PI3K activity, and the inhibition of PI3K activity by expressing the dominant-negative p85 also could not suppress mercury-induced ROS production in β-cells. Therefore, these results implicate that both PI3K and ROS independently regulate Akt signaling–related, mercury-induced insulin secretion inhibition and suggest that low-dose mercury–induced oxidative stress and PI3K activation cause pancreatic β-cell dysfunction through an Akt signaling pathway.

A recent study of mercury exposure in an urban pediatric population has shown that mean urinary mercury was 1.08 ± 1.82 μg/l (range 0.2–11.7) and found that 5% of subjects had unsuspected elevated urinary mercury levels (>5 μg/l) (41). The Food and Drug Administration advises women to limit the consumption of fish with up to 0.5 ppm (μg/ml) mercury to 14 oz/week or with up to 1.0 ppm mercury to 7 oz/week (42). Recently, the Wisconsin Department of Natural Resources lowered its mercury in fish action level by a factor of 10× to 0.05 ppm based on subtle alterations in neurodevelopment in children born to mothers who eat large amounts of fish (43). A study (44) of blood mercury levels in workers exposed to the preparation of mercury cadmium telluride layers has shown that the mercury value was estimated at 1.60 ± 0.20 μg Hg/l in control and at 10.72 ± 1.34 μg Hg/l in phase I and 8.08 ± 1.15 μg Hg/l in phase II; an individual who met with a mercury accident showed 226 μg Hg/l of blood, which decreased to 25 μg Hg/l after 3 months. It has been reported that sensitive nonhuman mammals showed significant adverse effects of mercury when dietary levels were 1,100 μg/kg body wt (45). The study of Nakagawa has shown that the total mercury concentrations in the hair of ordinary diseased people, including diabetes, were from 2.08 to 36.5 ppm; those values were considerably higher than that of healthy people of the same age-groups (46). In the present study, we used low doses of inorganic mercury (50 and 500 μg/kg) and organic methylmercury (20 μg/kg) to expose mice. The blood mercury content ranges are ∼3.9–5.8 and 38.4–72.2 μg/l in 50 and 500 μg · kg⁻¹ · day⁻¹ HgCl₂–treated mice, respectively, or 40.2–64.6 μg/l in 20 μg · kg⁻¹ · day⁻¹ MeHg–treated mice, which were exposed over a 2- to 4-week period (0.83–1.62 μg/l in age-matched controls). Moreover, no evidence of mercury-induced target organ toxicity was seen in either the clinical pathology parameters or histomorphologic evaluations. Thus, the increased blood mercury levels in the present study approach the human-exposed levels.

An increased incidence of diabetes has been found to exist in patients with documented Minamata disease (blood mercury >200 μg/l [20]) in Japan (26). Takeuchi and Eto (47) reported that disturbance of pancreatic islet cells were found in autopsy cases of Minamata disease. Nevertheless, a health-surveillance study in the population living in a methylmercury-polluted area ∼30 years after the outbreak of Minamata disease by Futatsuka et al. (48) showed that the prevalence of liver disease, renal disease, and diabetes was not higher in this methylmercury-polluted area compared with other areas in Japan, contrary to what was expected based on standard mortality ratios and pathological findings. However, the results of Sakamoto et al. (49) and Harada et al. (50) showed that no significant difference in the mercury concentrations in the erythrocytes and scalp hair was observed between the cases with Minamata disease and the controls about several decades after the outbreak of Minamata disease. Around 1960, marine products were heavily polluted by methylmercury in the Minamata area. When the cause of Minamata disease had already proved to be methylmercury-contaminated marine products, the residents on the local coast were hesitant to eat those marine products, so the patient’s mercury levels were gradually reduced (50). These findings may explain why there are no positive correlations between those diseases and methylmercury exposure ∼30 years after the outbreak of Minamata disease, as indicated by the study of Futatsuka et al. (48). These findings also imply that some diseases induced by methylmercury exposure may be reversible if the exposure was terminated. Our present results support this possibility; we found that the elevated blood mercury, decreased plasma insulin, and elevated blood glucose levels in mercury-treated mice were gradually reversed after mercury exposure was terminated. Thus, the present study indicates that low-dose mercury is capable of altering blood insulin and glucose homeostasis, although individuals may tend to ignore this risk of long-term exposure to low-dose mercury.

In conclusion, as indicated in Fig. 8C, our results strongly suggest an essential and specific role for the Akt signaling in low-dose mercury–induced, oxidative stress–caused pancreatic β-cell dysfunction. Moreover, low-dose mercury–induced oxidative stress causes pancreatic islet Akt activation and induces hypoinsulinemia and hyperglycemia in mice in the absence of overt mercury toxicity. These observations also give further evidence to confirm the possibility that mercury is an environmental risk factor for diabetes.

This study was supported by grants from the National Science Council of Taiwan (NSC93-2314-B-002-178); the Department of Health, Executive Yuan of Taiwan (DOH95-TD-B-111-TM003); and the National Taiwan University Hospital (93-A018).