Ouabain Suppresses Glucose-Induced Mitochondrial ATP Production and Insulin Release by Generating Reactive Oxygen Species in Pancreatic Islets

Male Wistar rats weighing 180–230 g were fed standard lab food ad libitum with free access to water in an air-conditioned room with a 12-h light, 12-h dark cycle. Islets of Langerhans were isolated by collagenase digestion. Insulin release from intact islets was monitored using either batch incubation or a perifusion system as previously described (19). For batch incubation experiments, islets were preincubated at 37°C for 30 min in Krebs-Ringer bicarbonate buffer (KRBB) supplemented with 2.8 mmol/l glucose and 0.2% BSA (fraction V). Groups of five islets were then batch-incubated for 60 min in 0.7 ml KRBB with test materials. In some experiments, NaCl and NaHCO₃ were replaced by choline chloride and choline bicarbonate, respectively, and 5 μmol/l atropine sulfate was added to prevent cholinergic effects. At the end of the incubation period, islets were pelleted by centrifugation, and aliquots of the buffer were sampled. For perifusion experiments, groups of 20 islets were placed in parallel chambers (400 μl each) of a perifusion apparatus and perifused with the same medium at a rate of 0.7 ml/min at 37°C. The medium was continuously gassed with 95% O₂ and 5% CO₂. Islets were perifused for 30 min to establish a stable insulin secretory rate at the basal level of glucose. Ouabain and the stimulating level of glucose were added to the medium according to each experimental protocol. The samples were collected at the times indicated in the figures. The amount of immunoreactive insulin was determined by radioimmunoassay, using rat insulin as a standard. Experiments using the same protocol were repeated three times to ensure reproducibility.

The oxygen consumption in isolated islets was measured by the method of Hutton and Malaisse (20) with slight modification. O₂ consumption was determined by the fluctuating levels of pO₂, which was monitored using a dual-channel oxygen monitoring system with a Clark-type micro-electrode (YSI model 5300; Instech Laboratories, Horsham, PA). The perifusate, KRBB medium equilibrated with 95% O₂/5% CO₂ (vol/vol) atmosphere, flowed through the chamber (70 μl volume) containing 1,000 islets, separated by a circular spacer of cellulose acetate, at a rate of 0.25 ml/min at 37°C. Islets were perifused for at least 30 min to attain a steady state in the presence of 2.8 mmol/l glucose, and the experiment was then performed. Ouabain (1 mmol/l) was introduced into the medium 15 min before 16.7 mmol/l glucose. The relationship between pO₂ and O₂ content of the medium was determined from the O₂ solubility data, with appropriate corrections for atmospheric pressure and H₂O vapor pressure.

Isolated islets were cultured for 14–18 h with RPMI 1640 medium (containing 10% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 11.1 mmol/l glucose) at 37°C in humidified air containing 5% CO₂. The ATP content in islets was determined by luminometric method as previously described (21). In brief, after groups of 10 islets were preincubated at 2.8 mmol/l glucose for 30 min, they were batch-incubated for 60 min in 0.8 ml KRBB with 16.7 or 2.8 mmol/l glucose in the presence or absence of 1 mmol/l ouabain. Ouabain was applied to the medium 15 min before 16.7 mmol/l glucose. Ca²⁺-free media were prepared with Ca²⁺-free KRBB plus 1 mmol/l EGTA. In some experiments, NaCl and NaHCO₃ were replaced by choline chloride and choline bicarbonate, respectively, and 5 μmol/l atropine sulfate was added. The reaction was stopped by immediate addition and mixing of trichloroacetic acid (TCA), to a final concentration of 5%. The islets were sonicated at 4°C and centrifuged (2,000g, 3 min), and a fraction (0.7 ml) of the supernatant was mixed with 1 ml water-saturated diethylether. The ether phase containing TCA was discarded. The step was repeated four times, and a fraction of the extracts (0.1 ml) was diluted with 0.1 ml of 20 mmol/l HEPES (pH 7.4 with NaOH). The ATP concentration in the solutions was measured by adding 0.1 ml luciferin-luciferase solution (Turner Designs, Sunnyvale, CA) in a bioluminometer (model 20e; Turner Designs). To draw a standard curve, blanks and ATP standards were run through the entire procedure, including the extraction steps.

Because the 16.7 mmol/l glucose-induced ATP increment is small because of the large stable ATP pool in fresh islets, as previously reported (22), we have presented data from cultured islets, in which the increment of ATP content by glucose is easier to observe because of the small stable ATP pool. Experiments using the same protocol were repeated three times to ensure reproducibility.

Mitochondrial membrane potential (ΔΨm) was measured by 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolcarbocyanine iodide (JC-1) fluorescence as previously reported (23,24). JC-1 exhibits potential-dependent accumulation in mitochondria by a fluorescence emission shift from green (∼525 nm) to red (∼595 nm). Isolated islets were incubated for 15 min in KRBB containing 2.8 mmol/l glucose and 0.2% BSA at room temperature in the dark with 10 μg/ml JC-1. The islets were washed in PBS and incubated with 0.25% trypsin and 1 mmol/l EDTA solution (Life Technologies, Grand Island, NY) for 3 min at 37°C, diluted by 20 ml cold PBS, and dispersed by pipetting. The dispersed islet cells were resuspended in 400 μl Ca²⁺-free medium and applied to glass cuvettes. After preincubation for 20 min at 37°C, the fluorescence was determined using a spectrofluorophotometer (RF 5000; Shimadzu, Kyoto, Japan) with excitation wavelength at 490 nm and emission wavelength at 590 nm and with stirring medium containing dispersed cells in cuvettes at 37°C. At time zero, basal fluorescence was determined, when glucose and ouabain, at final concentrations of 16.7 mmol/l and 1 mmol/l, respectively, were added. Cuvettes were incubated in humidified air containing 5% CO₂ at 37°C, and determinations were performed at the times indicated in the figures. Fluorescence was corrected by subtracting parallel blanks in which islet cells were not loaded with JC-1 and by DNA content, which was measured as previously described (21). Experiments using the same protocol were repeated three times to ensure reproducibility.

ROS production in islet cells under Ca²⁺-free conditions was measured by 5- (and 6-) chloromethyl-2′,7′ -dichlorofluorescein (CM-DCF) fluorescence as previously reported (15,16,24). The dispersed islet cells prepared by cold PBS and trypsin EDTA were incubated in KRBB containing 10 μmol/l CM-DCFH diacetate, a reduced CM-DCF conjugated with an acetate group, and 2.8 mmol/l glucose for 20 min at 37°C, and then washed in Ca²⁺-free medium three times. During loading, the acetate groups on CM-DCFH diacetate are removed by intracellular esterase, trapping the probe inside the cells. Production of ROS could be measured by changes in fluorescence, since oxidation of CM-DCFH produced fluorescent product CM-DCF in the cells. At time zero, basal fluorescence was determined using a spectrofluorophotometer (RF 5000) with excitation wavelength at 505 nm and emission wavelength at 540 nm and with stirring medium containing dispersed cells in cuvettes at 37°C. After 60-min incubation in 400 μl Ca²⁺-free KRBB containing test materials and 16.7 mmol/l glucose, ROS production was determined. Fluorescence was corrected by subtracting parallel blanks in which islet cells were not loaded with CM-DCFH diacetate and by DNA content. Experiments using the same protocol were repeated three times to ensure reproducibility.

The mitochondrial suspension from freshly isolated islets was prepared by repeated centrifugation, as previously reported (25,26), with slight modification. First, isolated islets were homogenized in medium A consisting of (mmol/l) 50 HEPES, 100 KCl, 1.8 ATP, 1 EGTA, and 2 MgCl₂ and 0.5 mg/ml BSA (electrophoretically homogeneous) (pH 7.00 at 37°C with KOH). After precipitation of cell debris and nuclei by centrifugation, the supernatant was more rapidly centrifuged (10,000g) to obtain a pellet containing the mitochondrial fraction. Afterward, the precipitation diluted by 200 μl of solution A was centrifuged again, and finally rinsed three times in medium B consisting of (mmol/l) 20 HEPES, 3 KH₂PO₄, 1 EGTA, 12 NaCl, 0.3 MgCl₂, 148 KCl, and 4 carnitine and 0.5 mg/ml BSA (electrophoretically homogeneous) (pH 7.10 with KOH). In some experiments, the ratio of NaCl and KCl was changed as indicated in the figures. The mitochondrial fraction in 500 μl medium B was kept on ice until use. The reaction was started by adding 5 μl mitochondrial suspension to 495 μl prewarmed medium B (37°C) supplemented with 0.5 mmol/l succinate, 50 μmol/l ADP, and 1 μmol/l diadenosine pentaphosphate (DAPP). The reaction was stopped by the addition of 0.5 μmol/l antimycin A. The samples were cooled to room temperature, and the ATP concentration in the solutions was measured by adding luciferin-luciferase solution to each sample with a bioluminometer.

To draw a standard curve, blank and ATP standards were added to parallel samples containing the complete incubation mixture except the mitochondrial suspension.

Thapsigargin, myxothiazol, carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP), ADP, DAPP, antimycin A, and α-tocopherol were purchased from Sigma (St. Louis, MO); succinate was purchased from Aldrich (Steimheim, Germany); ATP was purchased from Kohjin (Tokyo, Japan); and JC-1 and CM-DCFH diacetates were purchased from Molecular Probes (Eugene, OR). All other agents, including ouabain, were obtained from Nacalai Tesque (Kyoto, Japan).

Results are expressed as means ± SE. Statistical significance was evaluated by unpaired Student’s t test or paired t test. P < 0.05 was considered significant.

Ouabain inhibited 16.7 mmol/l glucose–stimulated insulin release from 60-min–incubated islets in a concentration-dependent manner (Table 1).

When the glucose concentration was raised from 2.8 to 16.7 mmol/l, 1 mmol/l ouabain had a dual effect on glucose-induced insulin release (Fig. 1A): 1 mmol/l ouabain significantly increased insulin release in the presence of 2.8 mmol/l glucose and enhanced the secretion more intensively after application of 16.7 mmol/l glucose (Fig. 1B); however, the compound inhibited insulin secretion 22 min after administration of 16.7 mmol/l glucose (Fig. 1A). The inhibition was reversed within 35 min after withdrawal of ouabain from the medium, whereas continuous exposure to ouabain sustained the inhibition of insulin secretion (Fig. 1A). The suppressive effect of 1 mmol/l ouabain on 16.7 mmol/l glucose–induced insulin secretion was greater than the stimulatory effect during 60-min incubation. Total release during 60-min incubation was calculated from the data shown in Fig. 2A (control 129.6 ± 7.9 vs. ouabain 86.9 ± 2.5 ng · 20 islets⁻¹ · 60 min⁻¹; P < 0.01).

To determine the influence of the cationic changes produced by ouabain, we examined insulin release under conditions of replacement of extracellular Na⁺ by choline, in which ouabain fails to raise intracellular Na⁺ concentration (27). In the absence of Na⁺, insulin release from both 10-min- and 60-min-incubated islets in the presence of 16.7 mmol/l glucose was inhibited significantly compared with that in the presence of Na⁺. However, ouabain still produced an inhibitory effect on insulin release from 60-min-incubated islets, although it had no effect on insulin release from 10-min-incubated islets in the absence of ambient Na⁺ (Table 2).

The elevation of O₂ consumption remained for 120 min with 16.7 mmol/l glucose present. ΔO₂ consumption at 60 min was 2.82 ± 0.61 pmol · islet⁻¹ · min⁻¹ in control samples. On the other hand, O₂ consumption in the presence of 1 mmol/l ouabain remained at a higher level until 30 min and was subsequently reduced (ΔO₂ consumption at 60 min -1.65 ± 0.73 pmol · islet⁻¹ · min⁻¹; P < 0.05 vs. control). However, the decrease was restored with a time delay of about 30 min after withdrawal of the compound (Fig. 2).

In medium containing physiological concentrations of Ca²⁺ and Na⁺, ATP content in islets incubated with 16.7 mmol/l glucose was significantly greater than in islets with 2.8 mmol/l glucose. However, in the presence of 1 mmol/l ouabain, 16.7 mmol/l glucose failed to increase ATP content. In the presence of the basal level of glucose, ATP content was not affected by ouabain (Table 3). In islets, increasing [Ca²⁺]_i increases ATP consumption (28), which leads to a decrease of intra-islet ATP content. Ouabain depolarizes the plasma membrane and increases Ca²⁺ influx via voltage-dependent Ca²⁺ channels. To eliminate the effect of increased [Ca²⁺]_i induced by ouabain, the ATP content under ambient Ca²⁺-free conditions was examined. Although 16.7 mmol/l glucose increased the ATP content in the absence of Ca²⁺, the increase also was abolished by ouabain (Table 3). This inhibitory effect of ouabain occurred in a concentration-dependent manner. Ouabain at 10⁻⁵ mol/l decreased ATP content significantly and at 10⁻³ mol/l maximally (Table 4). When extracellular Na⁺ was replaced by choline, ouabain still inhibited the ATP increase induced by 16.7 mmol/l glucose (Table 3). Ouabain also decreased ATP content in fresh (noncultured) islets in the presence of 16.7 mmol/l glucose (data not shown).

To evaluate the effect of ouabain on ΔΨm, JC-1 fluorescence was measured in the presence of 16.7 mmol/l glucose without Ca²⁺. After addition of 16.7 mmol/l glucose to the medium, the fluorescence increased gradually, indicating hyperpolarization of ΔΨm, whereas the basal level of fluorescence continued in the presence of 2.8 mmol/l glucose during the measurement. Ouabain significantly inhibited glucose-induced hyperpolarization of ΔΨm 30 min after administration. JC-1 fluorescence decreased to less than basal value after the addition of 1 μmol/l FCCP (Fig. 3).

The generation of ROS was examined in the presence of 16.7 mmol/l glucose in the absence of Ca²⁺. The fluorescence showed concentration-dependent increments when islet cells were incubated for 60 min with various concentrations of hydrogen peroxide (0 to ∼2 mmol/l) (Fig. 4A).

Myxothiazol, which inhibits the respiratory chain at site III (29), inhibited ROS production significantly (71.3 ± 2%; P < 0.01 vs. control). FCCP, which increases proton leak of the mitochondrial membrane, also suppressed it (85 ± 3%; P < 0.01 vs. control). Thapsigargin, an inhibitor of Ca²⁺-ATPase in endoplasmic reticulum, did not enhance ROS production (99.9 ± 1.4%). However, ROS production was significantly higher in islet cells exposed to 1 mmol/l ouabain (114 ± 3.7%; P < 0.05); the increment is approximately equivalent to that by 50 μmol/l hydrogen peroxide. In the presence of myxothiazol, there was no ouabain-induced increase in ROS production (Fig. 4B).

To evaluate the effect of ROS on ouabain-induced inhibition of ATP production, 100 μmol/l α-tocopherol, an antioxidant, was added to the medium 30 min before ouabain administration. α-Tocopherol significantly inhibited the ouabain-induced increase in ROS generation in the presence of 16.7 mmol/l glucose without Ca²⁺ (ouabain 122.8 ± 3% vs. ouabain + α-tocopherol 98.8 ± 4%; P < 0.05) (Fig. 5A). Although ouabain decreased ATP content significantly (control 9.3 ± 0.8 vs. ouabain 5.8 ± 0.9 pmol/islet; P < 0.05), ouabain did not affect ATP content in the presence of α-tocopherol (α-tocopherol 8.9 ± 0.9 vs. ouabain + α-tocopherol 8.2 ± 0.8 pmol/islet; NS) (Fig. 5B). Early enhancement of insulin release by ouabain was still observed in the presence of α-tocopherol. However, the antioxidant prevented the delayed inhibitory effect of ouabain (Fig. 5D). The total insulin release during 60-min incubation calculated from the data shown in Fig. 5D was not significantly different in control and ouabain-treated islets (α-tocopherol 146.2 ± 10.0 vs. α -tocopherol + ouabain 128.5 ± 4.8 ng · 20 islets⁻¹ · 60 min⁻¹; NS).

We examined the direct effect of ouabain and the Na⁺ concentration on ATP production from isolated mitochondria of islets. ATP production in the presence of succinate and ADP was linear within 60 min (data not shown). However, ATP production from mitochondria in the absence of succinate and in the presence of succinate and antimycin A did not show any increase within 60 min. Ouabain had no significant effect on ATP production from mitochondria compared with control values, but exposure to 50 μmol/l hydrogen peroxide inhibited it. In addition, the increasing Na⁺ concentration in medium B enhanced ATP production from mitochondria (Fig. 6).

In the present study, we demonstrate that ouabain suppresses the ATP content increase by glucose in pancreatic islets regardless of intracellular cationic changes by reducing mitochondrial ATP production rather than by increasing ATP consumption. We also show that the inhibition by ouabain of ATP production in islets is caused by increased ROS production in the mitochondria, as was found recently in cardiac myocytes (14–16).

Because 1 mmol/l ouabain, at which concentration Na⁺/K⁺-ATPase activity is maximally suppressed, is commonly used to investigate the effect of the compound on insulin secretion (7,10,12, 30), we used that concentration in all experiments in the present study. In addition, 1 mmol/l ouabain does not have irreversible, cytotoxic effects, as shown by the complete recovery of insulin secretion after withdrawal of the compound.

Inhibition of Na⁺/K⁺-ATPase activity is known to depolarize the plasma membrane abruptly (2), resulting in Ca²⁺ influx and increased [Ca²⁺]_i (10–12). The augmentation of insulin secretion in the basal state and in the early phase during glucose stimulation in the presence of ouabain, accordingly, should be attributable to the rise in [Ca²⁺]_i, consistent with previous reports (8–10,13). Under Na⁺-deprived conditions, early enhancement by ouabain was not observed. Ouabain is reported not to alter intracellular Na⁺ concentration ([Na⁺]_i) under Na⁺-deprived conditions (27) in which Na⁺ influx and Na⁺/K⁺ exchange do not occur. It should not cause membrane depolarization under those conditions because the depolarizing effect depends on blockade of the electrogenic effect of Na⁺/K⁺-ATPase due to Na⁺/K⁺ exchange. Ca²⁺ influx through voltage-dependent Ca²⁺ channels, therefore, should not be triggered. These results support attribution of the early enhancement of insulin release to the [Ca²⁺]_i increase by ouabain.

Because ouabain generates a higher level of [Ca²⁺]_i and enhances insulin release, it might raise ATP consumption by increasing exocytosis of insulin granules (31–33) and by enhancing Ca²⁺ uptake into endoplasmic reticulum through Ca²⁺-ATPase (34,35), thus reducing the ATP content in the islets. Furthermore, a preceding high [Ca²⁺]_i level has been reported to desensitize to increased intramitochondrial Ca²⁺ concentration, which impairs the insulin secretory response (36). Therefore, we performed experiments under Ca²⁺-depleted conditions to eliminate ATP consumption and desensitization due to the depolarizing effect of the compound. The intracellular ATP content was reduced even in such conditions. The effect of ouabain was maximal at ∼10⁻³ mol/l and half-maximal (IC₅₀) at ∼10⁻⁵ mol/l, which is higher than the concentration-dependent effects of the compound on Na⁺/K⁺-ATPase inhibition, for which its IC₅₀ is ∼10⁻⁷ mol/l (37).

Mitochondrial ATP production is driven by the proton-motive force that includes the mitochondrial membrane potential generated by the electron transport chain, and the rate of ATP synthesis in mitochondria is closely correlated with ΔΨm (38). Because ouabain reduced the hyperpolarizing effect of glucose on ΔΨm in a Ca²⁺-depeleted condition, the compound decreases mitochondrial ATP production independently of changes in [Ca²⁺]_i. Moreover, we have reported previously that ouabain also inhibited the ATP increment induced by α-ketoisocaproate, which is metabolized in mitochondria (4). Accordingly, the ouabain-induced inhibition of ATP content is a reflection of the reduction of ATP production, especially that in mitochondria.

Ouabain reduced not only the increment in ATP content and the hyperpolarization of ΔΨm by glucose, but also the increment in O₂ consumption by glucose. Because increased O₂ consumption occurs in uncoupling (20), ouabain-induced suppression of mitochondrial ATP production is not mediated by uncoupling, and the suppression should derive from direct or indirect effects on the respiratory chain. Because ouabain did not affect ATP production from isolated mitochondria directly, the effect should be indirect via intracellular signal transduction.

Ouabain induces an increment in [Na⁺]_i and a reduction in intracellular K⁺ concentration in β-cells in medium containing physiological concentrations of monovalent cations (27,30). Extracellular manipulation of monovalent cation concentrations without using ouabain has been reported to affect O₂ consumption, glucose utilization, and glucose oxidation, suggesting that intracellular concentrations of monovalent cations may affect mitochondrial metabolism (39). However, these results were not necessarily due to intracellular cationic alternation without change in Na⁺/K⁺-ATPase activity, because the manipulations themselves can affect Na⁺/K⁺-ATPase activity (2,37). Ouabain suppressed the insulin release and the increment in ATP content induced by glucose in islets even under the Na⁺-deprived condition in which ouabain failed to raise [Na⁺]_i (27). This suggests that ouabain does not necessarily require altered [Na⁺]_i to suppress mitochondrial metabolism, a theory supported by the observation that ATP production from isolated mitochondria is not suppressed by increasing the Na⁺ concentration in medium but rather is enhanced, as previously reported (25).

We found that hydrogen peroxide, the most abundant ROS inhibited mitochondrial ATP production from isolated mitochondria. Meachler et al. (17) previously demonstrated that transient exposure to hydrogen peroxide suppresses hyperpolarization of ΔΨm, the increment in insulin secretion, and the increase in ATP content induced by glucose in pancreatic β-cells. Moreover, Xie and colleagues (14–16) have shown recently that ouabain initiates signal cascades independently of changes in [Ca²⁺]_i and [Na⁺]_i and that it enhances Ras-dependent ROS generation in cardiac myocytes. Ouabain-induced ROS production also was observed in A7r5 cells and HeLa cells (16). Recently, ROS have been regarded as intracellular second messengers that act to regulate activation of downstream effectors of Ras and Rac (40,41). The interaction of Na⁺/K⁺-ATPase with ouabain initiates the activation of multiple interrelated signal pathways beginning with activation of Src kinase through protein kinase C, followed by Src-induced transactivation of epidermal growth factor, recruitment and activation of Ras, and activation of Ras-dependent pathways that communicate with mitochondria to increase ROS generation in cardiac myocytes (42,43).

In pancreatic islets, as reported in other tissues (29,44), metabolic inhibitors such as myxothiazol (an inhibitor of the site III complex of mitochondrial electron transport) and FCCP (an uncoupler) decrease ROS production. Moreover, ouabain induces ROS production in islets, which also occurs independently of [Ca²⁺]_i. Ouabain-induced ROS production was found to be completely abolished in the presence of myxothiazol, suggesting that its generation takes place mainly in mitochondria, due to increased synthesis or increased release. We noticed in this study that inhibition of Ca²⁺-ATPase at the endoplasmic reticulum, another ATP consumer, by thapsigargin (35) increased ATP content in the presence of glucose (5) but did not enhance ROS generation. These data indicate that ouabain-induced ROS generation decreases mitochondrial ATP production.

We examined the effect of ouabain on ATP content in the presence of α-tocopherol, a ROS scavenger. Because α-tocopherol restored the ouabain-induced suppression of ATP content and the late-phase inhibition of glucose-induced insulin release while early enhancement of insulin release was not affected, ROS clearly plays a more important role in the late suppression of insulin release by the compound.

Ouabain is known to be present endogenously in rat and human plasma as a kind of adrenal cortex hormone (45) and as a paracrine hormone secreted from the central nervous system (46). In type 2 diabetic patients, the plasma endogenous ouabain concentration is reported to be higher (47), while the Na⁺/K⁺-ATPase activity in erythrocytes is significantly lower (48), than in nondiabetic subjects. Whereas even at 10⁻⁵ mol/l, the effect of the compound on ATP consumption and insulin release was pronounced in the present study, that concentration of ouabain is higher than in human plasma (10⁻¹⁰ to 10⁻⁹ mol/l) (47), so the physiological and pathophysiological role of ouabain in Na⁺/K⁺-ATPase inhibition in insulin secretion remains to be determined.

In conclusion, we show that ouabain, a Na⁺/K⁺-ATPase inhibitor, suppresses mitochondrial ATP production by generating ROS via signal transduction, not involving cationic change. This mechanism may underlie the inhibitory effect of the compound on glucose-induced insulin release. To determine if the increment in ROS generation is Ras-dependent, as it is in cardiac myocytes, further investigation is required.

This study was supported in part by Grants-in-Aids for Creative Basic Research (10NP0201) and for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan, a grant from “Research for the Future” Program of the Japan Society for the Promotion of Science (JSPS-RFTF 97 I 00201), and Health Sciences Research Grants for Research on Human Genome, Tissue Engineering and Food Biotechnology from the Ministry of Health, Labor and Welfare. E.M. is a research fellow of the Japan Society for Promotion of Science.

The authors thank S. Nawata for technical assistance.