Vo1.173, No. 3,1990
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS Pages 932-939
December 31,1990
FREE RADICALGENERATIONBY EARLYGLYCATION PRODUCTS: A MECHANISM FOR ACCELERATED ATHEROGENESIS IN DIABETES CATHLEEN J.MULLARKEY, DIANE EDELSTEIN AND MICHAEL BROWNLEE
Division of Endocrinology, Department of Medicine and Diabetes Research Center, Albert Einstein College of Medicine, Bronx, NY 10461 Received November i, 1990
ABSTRACT Non-enzymaticglycation of reactive amino groups in model proteins increased the rate of free radical production at physiologic pH by nearly f i f t y fold over non-glycated protein. Superoxide generation was confirmed by electron paramagnetic resonance measurements with the spin-trap phenyl-t-butyl-nitrone. Both Schiff base and Amadori glycation products were found to generate free radicals in a ratio of 1:1.5. Free radicals generated by glycated protein increased peroxidation of membranes of linoleic/arachidonic acid vesicles nearly 2-fold over control, suggesting that the increased glycation of proteins in diabetes may accelerate vascular wall lipid oxidative modification. ©199oAcadem: Press, Inc.
Diabetes is associated with a substantially increased prevalence of atherosclerotic disease and cardiovascular mortality, even in the absence of hypertension, smoking, and hyperlipidemia (I). This two to four-fold increase in cardiovascular risk also occurs in hypercholesterolemic diabetic patients, resulting in a leftward shift of the risk vs. cholesterol level curve. Although hyperglycemia and hyperinsulinemia have both been implicated as pathogenetic factors in this process (2-4), diabetic populations with cholesterol
levels
significantly lower than Caucasians have proportionately lower risk of cardiovascular disease, despite apparently similar degrees of hyperglycemia and hyperinsulinemia (5,6). These observations suggested to us that the atherogenic effect of diabetes is mediated through changes in lipoprotein-arterial wall interactions. Much recent work on altered lipoprotein-arterial wall interactions has focused on the role of oxidatively modified LDL (7). Unlike native LDL, oxidatively modified LDL can recruit circulating monocytes, and then through uptake via the scavenger-receptor pathway, transform them into foam cells (8,9). Endothelial and smooth muscle cell membrane injury by oxidatively modified LDL has been postulated to play a role in atherosclerotic plaque development (7,10), and oxidatively modified LDL products have been specifically identified in aortic lesions, but not in normal adjacent tissue (11,12). Treatment of WHHLrabbits with the antioxidant drug probucol prevents the progression of atherosclerosis 0006-291x/90 $1.5o Copyright © 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
932
Vol. 173, No. 3, 1990
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
in this animal model of familial hypercholesterolemia, suggesting that oxidative modification of arterial wall lipids is an important pathogenic mechanism in vivo (13,14). A number of
i n i t i a t i n g steps have been proposed for the oxidative
modification of arterial wall lipids, all of them involving the participation of cells and divalent metal ions (7).
Basedon the known chemistry of Schiff base
and Amadori products formed during the nucleophilic addition of glucose to protein amino groups (nonenzymatic glycation), we hypothesized that such early glycation products on proteins deposited in the arterial wall could themselves generate free radicals capable of oxidizing lipids (15,16).
Since the level of
these products reflects ambient glucose concentration, such a mechanism could account for
increased oxidation of vascular wall
lipids
and accelerated
atherogenesis in hyperglycemic diabetic patients.
MATERIALSAND METHODS Chemicals. Ribonuclease A Type XIIA (chromatographically pure, from bovine pancreas), nitroblue tetrazolium (grade I l l ) , cytochrome c Type VI (from horse heart), sodium borohydride (98% pure), sodium cyanoborohydride (90-95% pure), superoxide dismutase (from bovine erythrocytes), catalase (from bovine liver, thymol free), xanthine (crystalline--99-100% pure), xanthine oxidase (grade I, from buttermilk suspension in 2.3M ammonium sulfate solution containing 0.02% sodium salicylate), and diethylenetriamepentacetic acid (DETAPAC) were obtained from Sigma. Sodium phosphate (monobasic), anhydrous D-glucose, and THAMwere obtained from Fischer Scientific. Phenyl-T-butyl nitrone (PBN)(98% pure) and aminoguanidine HCI were obtained from Aldrich Chemical Co. Vitrogen 100 was obtained from Collagen Corporation. C-14 radioactive glucose was obtained from Dupont-New England Nuclear. SephadexPD-IO columns were obtained from Pharmacia. Preparation of 61ycated Proteins. All glycation reactions were conducted at 37°C in 0.2 M sodium phosphate pH 7.8 containing Pen VK (60 mg/liter), gentamicin (40 mg/liter), fungazone (250 ug/liter), polymyxin B (10 mg/liter), EDTA 5mM, and PMSF i mM, as previously described (17). RNase (10 mg/ml) was incubated for varying times in buffer containing O-500mM glucose. After incubation, the samples were passed through PDIO sephadex columns to separate glycated protein from free glucose. Collagen gels (Vitrogen) prepared by addition of 8 ml of chilled vitrogen 100 collagen to i ml of 0.1NaOH and I ml of IOX phosphate buffered saline solution, then added to 35 mmpolystyrene wells, each so that each well contained the equivalent of 3 mg of bovine Type I collagen. Gelation Was accomplished by incubation at 37oC for 20 minutes followed by drying in a laminar flow hood. Gels were glycated by adding 4 ml. of glucose solution (50 mM to 500mM) and incubating for 7 days. After incubation, the wells were washed with 0.03 M sodium carbonate buffer (pH 10.4). Determination of Nitroblue Tetrazolium and Cytochrome c Reduction Rates.
The rate of nitroblue tetrazolium and cytochrome c reduction by glycated proteins, non-glycated proteins, and 10 mMglucose alone were determined using the methods of Beauchamp (18), and Fridovich (19), respectively. Reaction mixtures contained either 100 uM EDTA or 500 uM DETAPAC in addition to the standard reactants, in order to inhibit divalent metal-catalyzed free radical generation (20). Samples for NBT assay were equilibrated with air during the f i r s t 10 minutes, and the rate of reduction was determined as the change in As6o over the 10-20 minute interval. Samples for the Cytochrome C reduction assay were read at Asso. For collagen gels, the NBT assay was carried out by adding 933
Vol. 173, No. 3, 1990
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
NBT reaction mixture to each well in a volume of 1.5 ml. This reaction mixture was then transferred to i ml cuvettes and read at O, 10, 20, and 30 min. Inhibition of NBT and Cytochrome c Reduction by Superoxide Dismutase and Catalase. The contribution of superoxide and hydroxyl radicals to the reduction of NBT and cytochrome c by glycated protein was assessed by performing these assays in the presence of either superoxide dismutase (50 ug/ml) or catalase (500 ug/ml) (21). Electron Paramagnetic ResonanceMeasurements. Roomtemperature EPR spectra were obtained on a Varian E-112 spectrometer operating at X-band. The f i e l d modulation was at an amplitude of IG at 100 kHz frequency. The spectrometer is operated with a TEl02 rectangular resonant cavity into which a quartz capillary containing the sample is placed. The incident microwave power was 15-17 mW. The spin trap Phenyl-t-butyl-nitrone (PBN) was used (22). Since i t is sparingly soluble in aqueous solutions, i t was added to the samples d i r e c t l y as a solid in order to maximize the concentration. Xanthine oxidase (0.42 units/ml) was added to 200 mMphosphate buffer saturated with xanthine and PBN and incubated for 30 minutes to generate superoxide for a standard spectrum. Samples containing glycated RNAase (70 mg/ml), phosphate RNAase, glycated RNAase plus superoxide dismutase (5000 U/ml), and 10 mMglucose alone were incubated in phosphate buffer pH 7.8 containing i mM DETAPAC, and then transferred to quartz capillary tubes for spectral analysis. Reduction of Schiff Base and Amadori Products. Reduction of Schiff base adducts alone by NaBH3CN and of both Schiff base adducts and Amadori products by NaBH4 were carried out by adding a 200 molar excess of sodium cyanoborohydride or sodium borohydride to the incubated samples (23). Sodium cyanoborohydridetreated samples were exposed to air in a fume hood for two hours, refrigerated overnight and then passed through a sephadexPDIOcolumn to separate free glucose and sodium cyanoborohydride from RNAase. Sodiumborohydride-treated sampleswere exposed to air under a fume hood for 15 minutes, refrigerated overnight and then passed through sephadex PDIO columns. Quantitation of Schiff Base and Amadori Products. Samples containing 20 mg protein were divided equally. One portion was reduced with 400 mMNaBH3CNat pH 5.0 for 2 hrs at RT, and then the pH was adjusted to 8. Both portions were then reduced with NaB~H4(200 mM, S.A. 8 mCi/mmol) for 15 minutes at RT and 60 minutes at 4°C (24). Sampleswere dialyzed exhaustively and Amadori product cpm/mmole and Amadori product + Schiff base cpm/mmole were then determined. Assessment of Lipid Peroxidation by Glycated Protein. Stable preparations of linoleic/arachidonic acid vesicles were generated as described by Gebicki and Hicks (25). Linoleic acid (40 mg) and arachadonic acid (0.4 mg) in 0.1M Tris HCI pH 8.0 were vortexed and diluted to 20 ml. Glycated and non-glycated RNase (5 mg) were added to 2.5 ml of liposomes (total volume 10 ml) and incubated in the presence of 500 micromolar DETAPAC for 72 hrs at 37o C. Peroxidation was measured as an increase in conjugated diene absorption at 234nm (Em=28,000) and as an increase in thiobarbituric acid reactive substances (26). RESULTS
Effect of Protein Glycation on Nitroblue Tetrazolium and Cytochrome c Reduction Rates.
In all experiments conducted with glycated RNase, a ten-fold
increase in the rate of reduction occurred after only 1 day of protein incubation with glucose.
Incubation of RNase with glucose for one week increased the rate
of reduction an additional 5-fold. little
Incubations beyond this time resulted in
i f any further increase in reduction rate (Figure I ) . 934
In contrast, no
Vol. 173, No. 3, 1990
rr
BIOCHEMICAL AND BIOPHYSICAL RESEARCHCOMMUNICATIONS
7°I.
3.5-
Glucose(10 mM) BufferedRNAase GlycatedRNAase
6.0
rr
3.0-
8
~C z2
,~" 20t ~ x 4.0 ~
r~
x
.__.~ o ~= 2'0 1.2E
#_~ 2.0
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS Pages 932-939
December 31,1990
FREE RADICALGENERATIONBY EARLYGLYCATION PRODUCTS: A MECHANISM FOR ACCELERATED ATHEROGENESIS IN DIABETES CATHLEEN J.MULLARKEY, DIANE EDELSTEIN AND MICHAEL BROWNLEE
Division of Endocrinology, Department of Medicine and Diabetes Research Center, Albert Einstein College of Medicine, Bronx, NY 10461 Received November i, 1990
ABSTRACT Non-enzymaticglycation of reactive amino groups in model proteins increased the rate of free radical production at physiologic pH by nearly f i f t y fold over non-glycated protein. Superoxide generation was confirmed by electron paramagnetic resonance measurements with the spin-trap phenyl-t-butyl-nitrone. Both Schiff base and Amadori glycation products were found to generate free radicals in a ratio of 1:1.5. Free radicals generated by glycated protein increased peroxidation of membranes of linoleic/arachidonic acid vesicles nearly 2-fold over control, suggesting that the increased glycation of proteins in diabetes may accelerate vascular wall lipid oxidative modification. ©199oAcadem: Press, Inc.
Diabetes is associated with a substantially increased prevalence of atherosclerotic disease and cardiovascular mortality, even in the absence of hypertension, smoking, and hyperlipidemia (I). This two to four-fold increase in cardiovascular risk also occurs in hypercholesterolemic diabetic patients, resulting in a leftward shift of the risk vs. cholesterol level curve. Although hyperglycemia and hyperinsulinemia have both been implicated as pathogenetic factors in this process (2-4), diabetic populations with cholesterol
levels
significantly lower than Caucasians have proportionately lower risk of cardiovascular disease, despite apparently similar degrees of hyperglycemia and hyperinsulinemia (5,6). These observations suggested to us that the atherogenic effect of diabetes is mediated through changes in lipoprotein-arterial wall interactions. Much recent work on altered lipoprotein-arterial wall interactions has focused on the role of oxidatively modified LDL (7). Unlike native LDL, oxidatively modified LDL can recruit circulating monocytes, and then through uptake via the scavenger-receptor pathway, transform them into foam cells (8,9). Endothelial and smooth muscle cell membrane injury by oxidatively modified LDL has been postulated to play a role in atherosclerotic plaque development (7,10), and oxidatively modified LDL products have been specifically identified in aortic lesions, but not in normal adjacent tissue (11,12). Treatment of WHHLrabbits with the antioxidant drug probucol prevents the progression of atherosclerosis 0006-291x/90 $1.5o Copyright © 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
932
Vol. 173, No. 3, 1990
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
in this animal model of familial hypercholesterolemia, suggesting that oxidative modification of arterial wall lipids is an important pathogenic mechanism in vivo (13,14). A number of
i n i t i a t i n g steps have been proposed for the oxidative
modification of arterial wall lipids, all of them involving the participation of cells and divalent metal ions (7).
Basedon the known chemistry of Schiff base
and Amadori products formed during the nucleophilic addition of glucose to protein amino groups (nonenzymatic glycation), we hypothesized that such early glycation products on proteins deposited in the arterial wall could themselves generate free radicals capable of oxidizing lipids (15,16).
Since the level of
these products reflects ambient glucose concentration, such a mechanism could account for
increased oxidation of vascular wall
lipids
and accelerated
atherogenesis in hyperglycemic diabetic patients.
MATERIALSAND METHODS Chemicals. Ribonuclease A Type XIIA (chromatographically pure, from bovine pancreas), nitroblue tetrazolium (grade I l l ) , cytochrome c Type VI (from horse heart), sodium borohydride (98% pure), sodium cyanoborohydride (90-95% pure), superoxide dismutase (from bovine erythrocytes), catalase (from bovine liver, thymol free), xanthine (crystalline--99-100% pure), xanthine oxidase (grade I, from buttermilk suspension in 2.3M ammonium sulfate solution containing 0.02% sodium salicylate), and diethylenetriamepentacetic acid (DETAPAC) were obtained from Sigma. Sodium phosphate (monobasic), anhydrous D-glucose, and THAMwere obtained from Fischer Scientific. Phenyl-T-butyl nitrone (PBN)(98% pure) and aminoguanidine HCI were obtained from Aldrich Chemical Co. Vitrogen 100 was obtained from Collagen Corporation. C-14 radioactive glucose was obtained from Dupont-New England Nuclear. SephadexPD-IO columns were obtained from Pharmacia. Preparation of 61ycated Proteins. All glycation reactions were conducted at 37°C in 0.2 M sodium phosphate pH 7.8 containing Pen VK (60 mg/liter), gentamicin (40 mg/liter), fungazone (250 ug/liter), polymyxin B (10 mg/liter), EDTA 5mM, and PMSF i mM, as previously described (17). RNase (10 mg/ml) was incubated for varying times in buffer containing O-500mM glucose. After incubation, the samples were passed through PDIO sephadex columns to separate glycated protein from free glucose. Collagen gels (Vitrogen) prepared by addition of 8 ml of chilled vitrogen 100 collagen to i ml of 0.1NaOH and I ml of IOX phosphate buffered saline solution, then added to 35 mmpolystyrene wells, each so that each well contained the equivalent of 3 mg of bovine Type I collagen. Gelation Was accomplished by incubation at 37oC for 20 minutes followed by drying in a laminar flow hood. Gels were glycated by adding 4 ml. of glucose solution (50 mM to 500mM) and incubating for 7 days. After incubation, the wells were washed with 0.03 M sodium carbonate buffer (pH 10.4). Determination of Nitroblue Tetrazolium and Cytochrome c Reduction Rates.
The rate of nitroblue tetrazolium and cytochrome c reduction by glycated proteins, non-glycated proteins, and 10 mMglucose alone were determined using the methods of Beauchamp (18), and Fridovich (19), respectively. Reaction mixtures contained either 100 uM EDTA or 500 uM DETAPAC in addition to the standard reactants, in order to inhibit divalent metal-catalyzed free radical generation (20). Samples for NBT assay were equilibrated with air during the f i r s t 10 minutes, and the rate of reduction was determined as the change in As6o over the 10-20 minute interval. Samples for the Cytochrome C reduction assay were read at Asso. For collagen gels, the NBT assay was carried out by adding 933
Vol. 173, No. 3, 1990
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
NBT reaction mixture to each well in a volume of 1.5 ml. This reaction mixture was then transferred to i ml cuvettes and read at O, 10, 20, and 30 min. Inhibition of NBT and Cytochrome c Reduction by Superoxide Dismutase and Catalase. The contribution of superoxide and hydroxyl radicals to the reduction of NBT and cytochrome c by glycated protein was assessed by performing these assays in the presence of either superoxide dismutase (50 ug/ml) or catalase (500 ug/ml) (21). Electron Paramagnetic ResonanceMeasurements. Roomtemperature EPR spectra were obtained on a Varian E-112 spectrometer operating at X-band. The f i e l d modulation was at an amplitude of IG at 100 kHz frequency. The spectrometer is operated with a TEl02 rectangular resonant cavity into which a quartz capillary containing the sample is placed. The incident microwave power was 15-17 mW. The spin trap Phenyl-t-butyl-nitrone (PBN) was used (22). Since i t is sparingly soluble in aqueous solutions, i t was added to the samples d i r e c t l y as a solid in order to maximize the concentration. Xanthine oxidase (0.42 units/ml) was added to 200 mMphosphate buffer saturated with xanthine and PBN and incubated for 30 minutes to generate superoxide for a standard spectrum. Samples containing glycated RNAase (70 mg/ml), phosphate RNAase, glycated RNAase plus superoxide dismutase (5000 U/ml), and 10 mMglucose alone were incubated in phosphate buffer pH 7.8 containing i mM DETAPAC, and then transferred to quartz capillary tubes for spectral analysis. Reduction of Schiff Base and Amadori Products. Reduction of Schiff base adducts alone by NaBH3CN and of both Schiff base adducts and Amadori products by NaBH4 were carried out by adding a 200 molar excess of sodium cyanoborohydride or sodium borohydride to the incubated samples (23). Sodium cyanoborohydridetreated samples were exposed to air in a fume hood for two hours, refrigerated overnight and then passed through a sephadexPDIOcolumn to separate free glucose and sodium cyanoborohydride from RNAase. Sodiumborohydride-treated sampleswere exposed to air under a fume hood for 15 minutes, refrigerated overnight and then passed through sephadex PDIO columns. Quantitation of Schiff Base and Amadori Products. Samples containing 20 mg protein were divided equally. One portion was reduced with 400 mMNaBH3CNat pH 5.0 for 2 hrs at RT, and then the pH was adjusted to 8. Both portions were then reduced with NaB~H4(200 mM, S.A. 8 mCi/mmol) for 15 minutes at RT and 60 minutes at 4°C (24). Sampleswere dialyzed exhaustively and Amadori product cpm/mmole and Amadori product + Schiff base cpm/mmole were then determined. Assessment of Lipid Peroxidation by Glycated Protein. Stable preparations of linoleic/arachidonic acid vesicles were generated as described by Gebicki and Hicks (25). Linoleic acid (40 mg) and arachadonic acid (0.4 mg) in 0.1M Tris HCI pH 8.0 were vortexed and diluted to 20 ml. Glycated and non-glycated RNase (5 mg) were added to 2.5 ml of liposomes (total volume 10 ml) and incubated in the presence of 500 micromolar DETAPAC for 72 hrs at 37o C. Peroxidation was measured as an increase in conjugated diene absorption at 234nm (Em=28,000) and as an increase in thiobarbituric acid reactive substances (26). RESULTS
Effect of Protein Glycation on Nitroblue Tetrazolium and Cytochrome c Reduction Rates.
In all experiments conducted with glycated RNase, a ten-fold
increase in the rate of reduction occurred after only 1 day of protein incubation with glucose.
Incubation of RNase with glucose for one week increased the rate
of reduction an additional 5-fold. little
Incubations beyond this time resulted in
i f any further increase in reduction rate (Figure I ) . 934
In contrast, no
Vol. 173, No. 3, 1990
rr
BIOCHEMICAL AND BIOPHYSICAL RESEARCHCOMMUNICATIONS
7°I.
3.5-
Glucose(10 mM) BufferedRNAase GlycatedRNAase
6.0
rr
3.0-
8
~C z2
,~" 20t ~ x 4.0 ~
r~
x
.__.~ o ~= 2'0 1.2E
#_~ 2.0
