Long exposure to high glucose concentration impairs the responsive expression of gamma-glutamylcysteine synthetase by interleukin-1beta and tumor necrosis factor-alpha in mouse endothelial cells.

To elucidate the pathological metabolism of glutathione synthesis in diabetic endothelial cells, we studied the expression of gamma-glutamylcysteine synthetase (gamma-GCS) using a mouse vascular endothelial cell line. Exposing normoglycemic endothelial cells to tumor necrosis factor-alpha (TNF-alpha) or interleukin-1beta (IL-1beta) increased the activity and the mRNA expression of gamma-GCS. The addition of inhibitors for nuclear factor kappaB (NF-kappaB) to the cells caused a loss of the gamma-GCS mRNA expression in response to TNF-alpha. A shift of the concentration of glucose in the medium from 5.5 to 28 mM glucose and a following incubation for 7 days decreased the expression of gamma-GCS mRNA. These cells showed no apparent responses of gamma-GCS mRNA or the activity of NF-kappaB to TNF-alpha or IL-beta. Increase in the GSH concentration of the cells treated with 28 mM glucose restored the expression of gamma-GCS mRNA and its response to TNF-alpha or IL-beta, suggesting that redox regulation is involved in the expression of gamma-GCS. In summary, the expression of gamma-GCS is regulated by TNF-alpha or IL-1beta in endothelial cells mediated by NF-kappaB stimulation, and impairment of the regulation of gamma-GCS in hyperglycemic cells may be a cause of medical complications that develop in diabetes mellitus.

Endothelial cells play important roles in selective transport, anticoagulation, lipid metabolism, vascular tension, vascularization, and immunological regulation. The secretion of some cytokines or growth factors from endothelial cells and their binding to specific receptors mediate these functions (1). In diabetes mellitus, endothelial cell damage is believed to be a significant contributing factor in the development of medical complications. Oxidative stress is a factor involved in cellular injury (2). Increases in the production of oxygen radical species or decreases in the scavenging activity against oxidative stress may play crucial roles in the development of pathological conditions (3).
Glutathione (␥-glutamylcysteinyl glycine, GSH) participates in many biological processes such as the metabolism of amino acids containing sulfur, biosynthesis of leukotrienes and DNA, and in the cellular defense system against oxidative stress by reducing the disulfide linkage of proteins and other cellular molecules or by scavenging free radicals and reactive oxygen intermediates (4). The physiological role of GSH as an antioxidant has been described and substantiated in numerous disorders reflecting increased oxidation as a result of an abnormal GSH metabolism. GSH is synthesized in most mammalian cells by the activity of two ATP-requiring GSH-synthesizing enzymes, ␥-glutamylcysteine synthetase (␥-GCS) 1 and glutathione synthetase. ␥-GCS catalyzes the rate-limiting step of GSH synthesis (5). This enzyme has been purified and the structure of the cDNA from the rat kidney has been determined (6). GSH synthesis is also thought to be an important factor in cellular defense against stress such as radiation and drug resistance (7). We showed evidence that the concentration of GSH and the activity of ␥-GCS are decreased in endothelial cells from experimental diabetic rabbits (8) and that the expression of ␥-GCS mRNA is sensitive to stress such as heat shock or chemical insults (9). However, the pathological significance of the expression of ␥-GCS in the development of diabetic complications in endothelial cells is not known.
To understand the regulatory mechanisms for GSH synthesis under diabetic conditions, we studied changes in the responsiveness of ␥-GCS to IL-1␤ and TNF-␣ in mouse endothelial cells cultured with a high concentration of glucose. We also investigated how the activation of NF-B and ␥-GCS expression by TNF-␣ are linked.
Cell Culture-The mouse hemangioendothelioma (MHE) cell line established from the thyroid, expressing factor VIII and vimetidine (19), was used as vascular endothelial cells. MHE cells can incorporate acetylated low-density lipoprotein and stain positively for factor VIII antigen. The cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum at 37°C in 5% CO 2 under 100% humidity. The cells were harvested by centrifugation at 4°C. The cell extract was prepared by lysis with 4 volumes of 10 mM NaH 2 PO 4 / Na 2 HPO 4 , pH 7.4, containing 0.5 mM EDTA, 0.1 mM 2-mercaptoethanol, and 0.5 mM phenylmethylsulfonyl fluoride, followed by sonication for 2 min. The cellular debris was removed by centrifugation at 4°C for 60 min at 5,000 ϫ g. The levels of ␥-GCS and its related enzymes in the supernatant were estimated.
Enzyme Assay-The activity of ␥-GCS was estimated using L-␣aminobutyrate, L-glutamic acid, and [ 32 P]ATP as substrates as described (20). The activities of glutathione peroxidase, glutathione Stransferase and Cu,Zn-superoxide dismutase (Cu,Zn-SOD) were estimated photometrically as described by Beutler (21). One unit of enzyme activity was expressed as 1 mol of substrate changed per min.
Estimation of ATP and Glutathione-The concentrations of intracellular ATP, GSH, and GSSG were estimated enzymatically as described by Beutler (21).
Northern Blots-The cloned cDNA was isolated as described by Goldwin et al. (7). A ␥-GCS probe (267 base pairs corresponding to nucleotides 54 -320 of rat kidney ␥-GCS) was generated from rat liver mRNA by the polymerase chain reaction. A human Cu,Zn-SOD cDNA fragment was generously provided by Dr. Taniguchi. For control hybridization on the amount of RNA, the complementary oligonucleotide probe for 28 S ribosomal RNA was synthesized, and the sequence is as follows: 5Ј-AACGATCAGAGTAGTGGTATTTCACC-3Ј (22). These probes were radiolabeled with 32 P using a random primer and were used to screen a rat liver cDNA library in gt11. Isolation of cytoplasmic RNA and Northern blotting were essentially as described by Sambrook et al. (23). Cytoplasmic RNAs isolated from MHE cells were subjected to electrophoresis in 1% agarose gels containing 0.6 M formaldehyde, subsequently transferred to nylon membranes, and then hybridized with 32 P-labeled nick-translated probes. Autoradiographed membranes were analyzed using a Fujix Bio-Analyzer BAS-2000 (Fuji Photo Film, Tokyo, Japan). After stripping, the membranes were rehybridized with 32 Plabeled 28 S rRNA probe, and the intensity of the bands at 5.0 kilobases was measured. The relative radioactivity was expressed as a ratio of PSL corrected by the intensity of 28 S rRNA.
Electrophoretic Mobility Shift Assay-The electrophoretic mobility shift assay for NF-B was performed as described by Sen and Baltimore (24) with a slight modification. Briefly, nuclear extracts were incubated with an NF-B-specific 32 P-oligonucleotide. The binding reaction proceeded in a 20-l volume containing 10 g of extract, 4 l of a binding buffer (10 mM Tris, pH 7.5, 40 mM NaCI, 1 mM EDTA, 1 mM 2-mercaptoethanol, 4% glycerol), 2 g of poly(dI-dC) as a nonspecific competitor DNA, 2 g of bovine serum albumin and labeled oligonucleotide (3,000 -6,000 cpm). After a 30-min binding reaction at room temperature, samples were loaded on a 6% nondenaturing polyacrylamide gel and subjected to electrophoresis in 50 mM Tris, 45 mM borate, and 0.5 mM EDTA, pH 8.0. As specificity control, a 100-fold excess of unlabeled probe was applied. The sequence of the binding site for the NF-B probe was 5Ј-GGGATTTCC-3Ј. The DNA binding activity of the extracts was quantified by estimating the amount of the 32 P-labeled NF-B⅐DNA complex excised from the dried gels and was expressed as PSL. To understand the effect of NF-B on ␥-GCS mRNA expression by TNF-␣, antisense DNAs for p50 and p65 cDNA were used. The sequences of the antisense codons for the NF-B probe were 5Ј-AGTCAAAGCAGTGT-TCAAAT-3Ј (20 base pairs corresponding to nucleotides 41-60 of p-50 and 5Ј-TCATCTTCCCGGCAGAGCCA-3Ј (20 base pairs corresponding to nucleotides 22-41 of p65) as described (25,26). Authentic antisense DNA was dissolved in phosphate-buffered saline (9 volumes of isotonic saline and 1 volume of 0.1 M Na 2 HPO 4 /NaH 2 PO 4 , pH 7.4) and added to the incubation medium with a final concentration of 8 M 24 h before TNF-␣ was added as described by Wu and Wu (27). Similarly, protease inhibitors (TLCK and TPCK) that inactivate IB proteolysis were used as described (13).
Western Blots-Lysate (5 g) from cell extracts was separated by SDS-polyacrylamide gel electrophoresis in a 10% gel (9), transferred to a nitrocellulose membrane, and immunologically stained using rabbit anti-p50 and anti-p65 polyclonal IgG, and the bound antibodies were made visible with an alkaline phosphatase-coupled second antibody using a Proto Blot kit. The protein concentration was determined according to Redinbaugh and Turley (28), with bovine serum albumin as the standard.
Nuclear Run-on Transcriptional Assay-The nucleic RNA from cells (1 ϫ 10 8 /ml) were prepared as described (29). For the transcription assay, the following solution was added to 100 l of nucleic RNA suspension in 50 mM Tris-HCl, pH 8.0, containing 40% glycerol, 5 mM  32 P was isolated as described (29). As a standard, the transcriptional rates of Cu,Zn-SOD was shown, since this enzyme has been reported to be irresponsive to TNF-␣ (30). The radioactivity was expressed as a ratio of PSL.
DNA Damage-DNA damage was determined by estimating the formation of the 8-OHdG using high pressure liquid chromatography as described by Kasai et al. (32). Nucleosomes were prepared from MHE cells based on the difference of gravity using a sucrose gradient (33). They were incubated with 25 M tert-butylhydroperoxide for 4 h and formation of 8-OHdG was estimated.
Statistical Analysis-The data are given as the mean Ϯ S.D. Differences were calculated with Student's two-tailed t test.

Concentration of Glutathione and the Activity of Its Related
Enzymes-Incubation of MHE cells with increasing concentrations of glucose resulted in a corresponding decrease in the concentration of GSH after 7 days. The concentration of GSH in 11, 22, and 28 mM glucose (Table I) was 78, 59, and 56% of that with 5.5 mM glucose, respectively. As shown in Table I, the activity of ␥-GCS in cells incubated with 28 mM glucose was 51% of the activity in cells incubated with 5.5 mM glucose.

Effect of Glucose on the Expression of ␥-Glutamylcysteine Synthetase
Decrease in the activity of Cu,Zn-SOD by 69% (p Ͻ 0.01) and glutathione peroxidase by 87% (p Ͻ 0.01) was also observed in cells incubated with 28 mM glucose. No apparent change was noted in the activity of glutathione S-transferase or the concentration of ATP.
Effects of Cytokines on the Concentration of Glutathione-Incubating MHE cells with IL-1␤ or TNF-␣ resulted in a marked increase in the concentration of GSH which reached a maximum after 3 h of 547 and 474%, respectively, followed by a gradual decline (Fig. 1). The following experiments to determine the effects of the cytokines were performed using a 3-h incubation. No stimulatory effect of IL-1␤ and TNF-␣ on the concentration of GSH was observed after the cells had been incubated with 28 mM glucose for 7 days (9.1 Ϯ 2.7 nmol/10 6 cells with IL-1␤ and 9.0 Ϯ 1.9 nmol/10 6 cells with TNF-␣) ( Fig.  2A). The impaired responses to cytokines were dependent on the concentration of glucose; increases in the concentration of GSH in response to TNF-␣ were about 300% with 11 mM glucose, 147% with 22 mM glucose, and 138% with 28 mM glucose with respect to the control. There was no shift of the peak with these high concentrations of glucose (data not shown).
IL-1␤ and TNF-␣ also had stimulatory effects on the level of activity of ␥-GCS in cells grown in 5.5 mM glucose (172 and 192% increase compared with the control, respectively) but no effect on the level of activity of ␥-GCS in MHE cells incubated with 28 mM glucose for 7 days (Fig. 2B). A TNF-␣ receptor binding assay showed no apparent change in the characteristics of the receptor between the two cell groups (data not shown).
This suggested that impairment of the response to TNF-␣ cells exposed to high glucose concentration is not due to changes in the receptor levels.
Northern Blots-Effects of cytokines and glucose concentration on the expression of ␥-GCS mRNA were estimated by Northern blotting. IL-1␤ and TNF-␣ stimulated the expression of ␥-GCS mRNA (Fig. 3). TNF-␣ and IL-1␤ increased the expression of ␥-GCS mRNA by 225 and 220%, respectively. Incubating MHE cells treated with 28 mM glucose for 7 days resulted in a decrease in the level of ␥-GCS mRNA by 30% (lane 4), and the cytokine addition did not affect the expression of ␥-GCS mRNA under these conditions. The results indicate that IL-1␤ and TNF-␣ stimulate the expression of ␥-GCS mRNA followed by concomitant increases in the ␥-GCS activity and GSH concentration and that exposing MHE cells to a high concentration of glucose negates the stimulatory effect of these cytokines.
Activity of NF-B-The effect of TNF-␣ on the expression of NF-B was estimated using a gel-shift assay. The concentration of NF-B was increased 5-fold when MHE cells with 5.5 mM glucose were incubated with TNF-␣ (Fig. 4, lanes 2 and 3, 4350 PSL versus 860 PSL), whereas NF-B concentration did not increase in the cells incubated with 28 mM glucose (lane 5). These results indicated that the high concentration of glucose inhibited cytokine-dependent NF-B activation in MHE cells.
To determine whether the TNF-␣-dependent NF-B activation is linked to the expression of ␥-GCS mRNA, MHE cells in the presence of 5.5 mM glucose were exposed to TLCK and TPCK, which inhibit the activation of NF-B and its binding to DNA, at 37°C for 30 min before the addition of TNF-␣ (13). TLCK inhibited the effect of TNF-␣ on the activation of NF-B (Fig. 5A, lane 4) as well as the expression of ␥-GCS mRNA (Fig.  6, lane 4). TPCK increased the expression of ␥-GCS (Fig. 5A,  lane 5), but it blocked additional stimulation of the expression of ␥-GCS mRNA by TNF-␣ (Fig. 6, lanes 5 and 6). These results suggest that the inhibition of NF-B activation inhibits the TNF-␣-dependent expression of ␥-GCS mRNA.
To obtain additional evidence for this suggestion, we incubated MHE cells with the antisense codons for p50 and p65. Adding the antisense codons 24 h before TNF-␣ inhibited TNF-␣-dependent increase in NF-B concentration (Fig. 5A, lane 6) and the expression of ␥-GCS mRNA stimulated by TNF-␣ (Fig.  6, lanes 7 and 8). An immunological analysis of cell extracts from MHE cells indicated that the concentration of both p50 and p65 proteins increased by TNF-␣ and that the authentic antisense codons inhibited the synthesis of these proteins (Fig.  5B).
Nuclear Run-on Assay-We performed nuclear run-on transcriptional assays to determine whether the decreased expression of the ␥-GCS mRNA in the presence of high glucose concentration was due to altered transcriptional activity of this gene. Shift of the concentration of glucose to 28 mM decreased the transcriptional rates of ␥-GCS by 58% (Fig. 7, lanes 1 and  3). Its transcriptional rates in response to TNF-␣ were 6.0-fold higher in MHE cells incubated with 5.5 mM glucose (lanes 1 and 2) and only 1.1-fold in the presence of 28 mM glucose (lanes 3 and 4). The transcriptional rates of Cu,Zn-SOD in 28 mM glucose slightly changed by 90% compared with that in 5.5 mM glucose, and TNF-␣ had no apparent effect on the transcription of Cu,Zn-SOD in both glucose concentrations (lanes 2 and 4).
Modification of Glutathione Levels-The effect of the changes in the concentrations of GSH and GSSG on the expression of ␥-GCS mRNA was studied. As shown in Table II, when the cells incubated with 5.5 mM glucose were exposed to 1 mM GSH ester 48 h before analysis, the concentration of GSH and the expression of ␥-GCS mRNA increased by 163 and 141%, respectively. GSH ester also increased the GSH/GSSG ratio by 170%. When the cells incubated with 28 mM glucose, in addition to the decrease in the concentration of GSH, the concentration of GSSG increased by 146%, and the GSH/GSSG ratio decreased by 33%. GSH ester added to the cells incubated with 28 mM glucose increased the concentration of GSH by 266% and the expression of ␥-GCS mRNA by 283%, decreased the concentration of GSSG by 78%, and increased the GSH/GSSG ratio by 343%, respectively. Furthermore, it restored the weakened response of ␥-GCS mRNA to TNF-␣ as well as to IL-1B. These data suggest that the expression of ␥-GCS mRNA correlates with the concentration of GSH more than with the concentration of GSSG or the GSH/GSSG ratio.
DNA Damage-To establish if DNA from cells incubated with a high concentration of glucose is sensitive to oxidative stress, DNA damage by oxidative stress was studied by estimating the formation of 8-OHdG (Fig. 8). Nucleosomes prepared from MHE cells incubated with 5.5 mM glucose for 7 days showed no apparent increase in the levels of 8-OHdG after adding 25 M tert-butylhydroperoxide for 4 h. In contrast, nucleosomes from cells incubated with 28 mM glucose showed an increase in the level of 8-OHdG induced by 25 M tert-butylhydroperoxide (5.8 Ϯ 0.4 pmol of 8-OHdG/g of DNA with 28 mM glucose versus 0.6 Ϯ 0.1 pmol of 8-OHdG/g of DNA with 5.5 mM glucose). This DNA damage due to 25 M tert-butylhydroperoxide in MHE cells with 28 mM glucose was restored to normal levels when cells were incubated beforehand with 1 mM GSH ester for 48 h.

DISCUSSION
To maintain cellular functions, relatively high concentrations of GSH are maintained through synthesis, redox cycle supported by GSH reductase and GSH peroxidase, and the active transport of GSSG or GSH S-conjugates. In particular, ␥-GCS regulates the overall turnover of the intracellular GSH (4). The activity of ␥-GCS is regulated by the nonallosteric inhibition by GSH (5). Moreover, the levels of GSH are regulated by the expression of ␥-GCS. An increase in GSH concentration caused by elevated transcriptional levels of ␥-GCS has been reported (7,34). GSH synthesis in response to various stresses occurs rapidly (9). The rapid response of the expression of ␥-GCS mRNA to heat shock or oxidative stress is followed by corresponding elevations of the concentration of GSH, demonstrating the important function of GSH and ␥-GCS in cellular emergencies. We have reported that depletion of GSH causes embryogenesis in rat embryo culture (35). The decreased expression of ␥-GCS mRNA in rat embryos cultured with a high concentration of glucose was reversed when the embryos were exposed to GSH ester, suggesting that oxidative conditions alter the expression of ␥-GCS mRNA.
Endothelial cells face the circulating blood. An increase in cytotoxicity caused by H 2 O 2 was found in endothelial cells cultured in high concentrations of glucose (36), and a weakened defense system against oxidative stress is thought to contribute to the development of diabetic complications. In this study, the exposure of endothelial cells to a high concentration of glucose for 7 days decreased the expression of ␥-GCS mRNA (Fig. 3), followed by a concomitant decrease in the enzyme protein, the enzyme activity, and the concentration of GSH (Table I, Fig. 2), in good agreement with a previous report (8). IL-1␤ and TNF-␣ have ubiquitous biological activities and play diverse biological roles in host systems. They induce chronic inflammatory changes and bring about increases in a variety of defense mechanisms, particularly immunologic and hematologic responses (2). Endothelial cells synthesize IL-1␤ and TNF-␣ and have specific receptors for them. Kalebic et al. have reported (37) that expression of the human immunodeficiency virus in chronically infected monocyte cells is induced by TNF-␣ and suppressed by GSH, suggesting a relationship between cellular GSH and the viral infection mediated by TNF-␣. A protective effect of IL-1␤ against oxidative stress has been reported by Tsan et al. (38). Administering IL-1␤ to rats increased the activities of pulmonary Mn-SOD, and Cu,Zn-SOD. However, the effects of these cytokines on the defense systems in endothelial cells have not been investigated in detail.
This results here represent the first report that TNF-␣ stimulates the expression of ␥-GCS, resulting in an increase in the levels of GSH. The concentration of GSH in endothelial cells increased within 3 h after the addition of TNF-␣ or IL-1␤ and gradually returned to the initial level after 48 h (Fig. 1). This increase in the concentration of GSH was correlated with an increase in the expression of ␥-GCS mRNA and a corresponding increase in the activity of this enzyme (Figs. 2 and 3). Marcho et al. (39) have reported a decrease in the concentration of GSH and an increase in the susceptibility to oxygen toxicity in bovine pulmonary artery endothelial cells treated with TNF-␣ for 18 h. Similar observations were reported by Ishii et al. (40) who found that TNF-␣ administration for 6 h mediates a decrease in the concentration of GSH and an increase in the sensitivity of pulmonary vascular endothelial cells to H 2 O 2 . The reason for the difference in the response between pulmonary endothelial cells reported by Ishii et al. (40) and the results here is unclear.
NF-B is inducible both by IL-1␤ and TNF-␣ (41,42). Simultaneous activation of NF-B and the expression of ␥-GCS by TNF-␣ suggest that the effect of TNF-␣ on the ␥-GCS mRNA is mediated by NF-B (Fig. 4). Proteolytic steps of IB are important in the activation of NF-B (13). The protease inhibitors, TLCK and TPCK, inhibited the TNF-␣-dependent activation of NF-B and the expression of ␥-GCS mRNA (Figs. 5 and 6). The expression of ␥-GCS mRNA was similarly inhibited when NF-B was blocked by its antisense codons (Fig. 6). These results suggest that NF-B initiates expression of ␥-GCS. The result of the nuclear run-on assay for the expression of ␥-GCS stimulated by TNF-␣ indicates that the transcriptional levels of ␥-GCS are regulated by TNF-␣ (Fig. 7), whereas those of Cu,Zn-SOD are not in good agreement with the data reported by Visner et al. (30).
The endothelial cells exposed to high concentration of glucose for 7 days showed a decrease in the expression of ␥-GCS mRNA and its response to TNF-␣ and IL-1␤ (Figs. 2-4). NF-B is regulated by oxidoreductive control mechanisms (43), and the oxidoreductive condition of NF-B regulates its DNA binding activity (44). Galter et al. (45) have reported the inhibitory effect of GSSG on the DNA binding activity of NF-B activation.
A striking observation in this study is that elevation of the concentration of GSH by the addition of GSH ester to the cells incubated with 28 mM glucose resulted in an increase in the expression of ␥-GCS. Furthermore, it restored the weakened response of the expression of ␥-GCS mRNA to TNF-␣ and IL-1␤ (Table II). The effect of modification of GSH metabolism on the increase in the expression of ␥-GCS was apparently paralleled by an increase in the concentrations of GSH. These observations suggest that GSH concentration is important in the regulation of the expression of ␥-GCS and the impaired expression of ␥-GCS in the cells incubated with a high glucose concentration causes changes in the redox regulation system. However, the precise mechanisms by which this responsiveness of ␥-GCS is impaired in the cells remain to be clarified.
Since endothelial cells, which express insulin-independent glucose transporter-1, are exposed to intracellular high glucose concentration levels in diabetes mellitus, we propose that a weak response of GSH synthesis to cytokines together with low levels of GSH, reduces cellular antioxidant activity, and is a potential cause of endothelial cell damage and the development of diabetic complications. This is supported by the findings in this study that the DNA damage is increased by oxidative stress in the endothelial cells exposed to high concentrations of glucose (Fig. 8).
In conclusion, long exposure of endothelial cells to high glucose levels impairs GSH synthesis stimulated by cytokines, which apparently weakens defense systems against oxidative stress in diabetes mellitus.