Heart and Circulatory Physiology

Enhanced mitochondrial superoxide in hyperglycemic endothelial cells: direct measurements and formation of hydrogen peroxide and peroxynitrite

Celia Quijano, Laura Castro, Gonzalo Peluffo, Valeria Valez, Rafael Radi


Hyperglycemic challenge to bovine aortic endothelial cells (BAECs) increases oxidant formation and cell damage that are abolished by MnSOD overexpression, implying mitochondrial superoxide (O2•−) as a central mediator. However, mitochondrial O2•− and its steady-state concentrations have not been measured directly yet. Therefore, we aimed to detect and quantify O2•− through different techniques, along with the oxidants derived from it. Mitochondrial aconitase, a sensitive target of O2•−, was inactivated 60% in BAECs incubated in 30 mM glucose (hyperglycemic condition) with respect to cells incubated in 5 mM glucose (normoglycemic condition). Under hyperglycemic conditions, increased oxidation of the mitochondrially targeted hydroethidine derivative (MitoSOX) to hydroxyethidium, the product of the reaction with O2•−, could be specifically detected. An 8.8-fold increase in mitochondrial O2•− steady-state concentration (to 250 pM) and formation rate (to 6 μM/s) was estimated. Superoxide formation increased the intracellular concentration of both hydrogen peroxide, measured as 3-amino-2,4,5-triazole-mediated inactivation of catalase, and nitric oxide-derived oxidants (i.e., peroxynitrite), evidenced by immunochemical detection of 3-nitrotyrosine. Oxidant formation was further evaluated by chloromethyl dichlorodihydrofluorescein (CM-H2DCF) oxidation. Exposure to hyperglycemic conditions triggered the oxidation of CM-H2DCF and was significantly reduced by pharmacological agents that lower the mitochondrial membrane potential, inhibit electron transport (i.e., myxothiazol), and scavenge mitochondrial oxidants (i.e., MitoQ). In BAECs devoid of mitochondria (rho0 cells), hyperglycemic conditions did not increase CM-H2DCF oxidation. Mitochondrial O2•− formation in hyperglycemic conditions was associated with increased glucose metabolization in the Krebs cycle and hyperpolarization of the mitochondrial membrane.

  • mitochondria
  • hyperglycemia
  • aconitase
  • MitoSOX

in endothelial cells, superoxide (O2•−) is involved in the physiological regulation of vascular tone (39, 68) and underlies the pathogenesis of cardiovascular diseases, diabetes complications, ischemia-reperfusion, and sepsis (34). Superoxide is formed by the mitochondrial electron transport chain (62), NAD(P)H oxidases (6, 67), uncoupled nitric oxide synthase (NOS) (57), xanthine oxidase (65), and cytochrome P-450 (34). The relative importance of these sources varies with the physiological or pathological stimuli and the tissue. Superoxide is the precursor hydrogen peroxide (H2O2), peroxynitrite, and other strong oxidizing species that, when produced at low levels, modulate gene expression, signal transduction cascades, and enzyme activity but, if their levels are high and/or sustained, produce oxidative damage (34, 41, 68).

In particular, the oxidants formed during hyperglycemic challenge to endothelial cells appear to be involved in the pathogenesis of diabetic complications (3). Brownlee et al. (4, 40) postulated that O2•− produced in the mitochondria forms oxidants that trigger the formation of advanced glycation end products (AGEs), increase glucose flux through the polyol and hexosamine pathways, and activate protein kinase C (PKC) and nuclear factor-κB (NF-κB). This proposal of the role of mitochondrial O2•− is based on the observation that MnSOD expression suppresses oxidant formation1 and normalizes these routes of hyperglycemic damage (40). However, direct measurements of mitochondrial O2•− formation and establishment of its steady-state concentrations during hyperglycemic challenge to endothelial cells are lacking. Indeed, O2•− detection in cells is not an easy task, since many of the existing probes either are not specific or are prone to pitfalls, such as promoting artifactual O2•− production even in systems were the radical is not being formed (59). Aconitase inactivation is a sensitive tool for the detection and quantitation of O2•− formation in the mitochondria (21). Although aconitase also can be inactivated by nitric oxide-derived species (7, 28), a recent study from our group (61) showed that in cells producing low levels of nitric oxide (·NO), aconitase inactivation remains sensitive to variations in O2•− concentration. Besides, recent studies have identified and characterized a fluorescent, distinguishable product from the reaction between hydroethidine and O2•−, hydroxyethidium (OH-Etd+), providing the means to selectively detect the formation of O2•−. Hydroethidine does not form O2•− artifactually; besides, its novel mitochondrially targeted derivative (Mito-HE; trade name: MitoSOX) allows the detection of O2•− in this subcellular compartment (43, 50, 70).

Cell damage and/or dysfunction arising from mitochondrial O2•− probably requires the formation of secondary oxidants capable of diffusing to other cellular compartments. Few reports have identified which secondary oxidants are being formed and involved in the degenerative processes of the hyperglycemic endothelial cell. MnSOD overexpression decreases oxidation of the fluorescent probe chloromethyl dichlorodihydrofluorescein (CM-H2DCF) (40), suggesting that peroxynitrite, the reaction product between O2•− and ·NO, rather than H2O2, product of O2•− dismutation, is mainly responsible for oxidant formation. Moreover, in hyperglycemic endothelial cells, tyrosine nitration increases (16, 20) and peroxynitrite is reported to nitrate prostacyclin synthase and oxidize endothelial NOS, inactivating both enzymes (73, 74). Nonetheless, H2O2 levels also could rise in hyperglycemic conditions affecting the expression of NOS and NAD(P)H oxidase (17), perturbing vascular tone and endothelium permeability (9, 10, 14, 56).

In summary, although many studies support a relevant role of oxidants in hyperglycemic damage to endothelial cells, the nature, localization, and levels of the different species formed in these conditions is not well established. In this study, we measured mitochondrial O2•− through aconitase inactivation and Mito-HE oxidation and assessed the formation of secondary O2•−-derived oxidant species (i.e., peroxynitrite and H2O2). We also confirmed the role of mitochondria in oxidant formation in hyperglycemic conditions by performing experiments with BAECs devoid of mitochondria (rho0 cells). The alteration of mitochondrial homeostasis was further confirmed by measuring mitochondrial membrane potential and Krebs cycle activity.



Culture medium (M199) and fetal bovine serum (FBS) were obtained from GIBCO (Invitrogen, Grand Island NY). Iron-supplemented bovine calf serum was obtained from Hyclone (Logan UT). Phenol red-free M199 was obtained from Sigma (St. Louis, MO). Monochlorobimane, 1,1′-dioctadecyl-3,3,3′,3′-tetramethyl-indocarbocyanine perchlorate acetylated low-density lipoprotein (DiI-Ac-LDL), 5(6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate (CM-H2DCFDA), 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolocarbocyanine iodide (JC-1), mito-hydroethidine (MitoSOX red), MitoFluor green, and 4,6-diamidino-2-phenylindole (DAPI) nucleic acid stain were obtained from Molecular Probes-Invitrogen (Eugene, OR). DETA NONOate (NOC-18) was obtained from Alexis (San Diego CA). Xanthine oxidase and glucose oxidase were obtained from Calbiochem (La Jolla, CA). Potassium cyanide was obtained from Mallinckrodt. Ethidium bromide was obtained from Merck (Darmstadt, Germany). Vanadium(III) chloride was obtained from Aldrich (Milwaukee, WI). d-[6-14C]glucose was obtained from Amersham Biosciences (Little Chalfont, UK). All other reagents were purchased from Sigma. Decyltriphenylphosophonium bromide (TPP) and MitoQ, a mixture of mitoquinol [10-(6′-ubiquinolyl)decyltriphenylphosphonium] and mitoquinone [10-(6′-ubiquinonyl)decyltriphenylphosphonium], were kind gifts from Dr. Michael Murphy (Medical Research Council Dunn Human Nutrition Unit, Cambridge, UK) (30). Human recombinant MnSOD was a kind gift from Dr. Daniel Hernandez-Saavedra (Instituto Mexicano del Seguro Social, Mexico City, Mexico) and Dr. Joe McCord.

Cell culture and preparation of rho0 cells.

Bovine aortic endothelial cells (BAECs) were obtained as described previously (52). Briefly, bovine thoracic aortas were acquired from a local slaughterhouse (Frigorífico Carlos Schneck, Montevideo, Uruguay), and cells were obtained by scraping the luminal surface of the aorta. When colonies of BAECs were formed, they were isolated using 8-mm-diameter cloning rings (Sigma). The purity of the culture was assessed by immunocytochemistry using an anti-human von Willebrand factor polyclonal antibody (Sigma) and an acetylated low-density lipoprotein fluorescently labeled (DiI-Ac-LDL). The cells were propagated by subculturing in a 1:4 ratio in M199 with 5% FBS, 5% iron-supplemented calf serum, 10 μM thymidine, 100 U/ml penicillin G, and 100 μg/ml streptomycin sulfate. Medium was routinely changed after 3 days, 1 wk after subculture, and once a week thereafter for no longer than a month. Experiments were conducted using cells between passages 4 and 14. One week after subculture, cells were switched to M199 (containing 5.5 mM glucose) with 0.4% FBS or to M199 with 0.4% FBS plus 24.5 mM glucose for 3 days; medium was changed daily, and experiments were performed on day 10 after subculture.

To obtain mitochondrial DNA-depleted pseudo-rho0 cells, BAECs were subcultured and maintained for 10 days as described previously but with the addition to the medium of ethidium bromide (250 ng/ml), uridine (50 μg/ml), and sodium pyruvate (110 μg/ml) (71); after subculture, the medium was changed every 2 days. The generation of rho0 cells was confirmed by measuring mitochondrial membrane potential (ΔΨm) using JC-1 (as described below) and assessing protein levels of complex IV subunit I by Western blot, resolving cellular proteins (80 μg) in a 10% SDS-PAGE and using a monoclonal anti-complex IV subunit 1 antibody (2 μg/ml) from Invitrogen.

Aconitase activity.

Cells were grown in 100-mm-diameter culture dishes and exposed to different conditions in M199. After treatment, cells were rinsed with phosphate-buffered saline (PBS: 140 mM NaCl, 4 mM KCl, and 10 mM Na2HPO4, pH 7.4), and scraped into 5 ml of ice-cold PBS. Cells were then centrifuged at 1,000 g for 5 min, resuspended in lysis buffer (50 mM Tris·HCl, 20 μM sodium fluorocitrate, and 0.6 mM MnCl2), sonicated, and centrifuged at 14,000 g for 1 min. The supernatant was used to measure enzyme activity. To measure aconitase activity in subcellular fractions, cells were grown in 140-mm-diameter culture dishes and incubated in Krebs buffer (1 mM CaCl2, 4.8 mM KCl, 1.2 mM MgSO4, 118 mM NaCl, 1.2 mM KH2PO4, and 25 mM HEPES) in different conditions. Cells were then washed and scraped into 7 ml of ice-cold fractionation buffer (FB: 0.25 M sucrose, 10 mM Tris·HCl, 0.1 mM EDTA, 2 mM sodium citrate, and 1 mM sodium succinate). Cells from two dishes were pooled together, centrifuged at 1,000 g for 5 min, resuspended in 1 ml of FB, and homogenized at 4°C with a Potter-Elvenhem homogenizer by 10 strokes at 800 rpm. Cell lysates were centrifuged at 1,500 g for 10 min, and the supernatants, containing the cytosolic and mitochondrial fractions, were centrifuged at 13,000 g for 10 min. The mitochondrial pellet was resuspended in 200 μl of FB and centrifuged at 13,000 g for 10 min, and the pellet was resuspended in 100 μl of FB and sonicated for 1 s at maximal power four times, alternating with incubation in an ethanol-ice bath. Aconitase activity in the cell lysate and the different fractions was assayed immediately, using the coupled assay with porcine heart isocitrate dehydrogenase, as described previously (21). Lactate dehydrogenase (LDH; cytosolic marker) and MnSOD (mitochondrial matrix marker) were assayed to evaluate the purity of the fractions. LDH activity was measured following NADH formation at 340 nm (24), and MnSOD activity was determined by measuring the inhibition of cytochrome c reduction by xanthine oxidase/xanthine in the presence of 3 mM potassium cyanide (12, 19). The specific activity of LDH in the mitochondrial fraction was <5% of that in the cytosolic fraction, and no MnSOD activity could be detected in the cytosolic fraction.

Mito-HE oxidation.

Cells were incubated with 1 μM Mito-HE (MitoSOX red) for 10 min, rinsed, incubated for 30 min in phenol red-free M199 under different experimental conditions, and visualized using a Nikon Eclipse TE 200 epifluorescence microscope at a fixed exposure time. Fluorescence arising from the hydroxyethidium derivative (Mito-OH-Etd+) was visualized with a filter cube fitted especially for this purpose with an excitation filter (wavelength/band pass) of 390/22 nm, an emission filter of 605/55 nm, and a beam splitter at 425 nm. The cells also were observed using an excitation filter (wavelength/band pass) of 540/25 nm, emission filter of 605/55 nm, and a beam splitter at 565 nm to assess global oxidation of Mito-HE to both Mito-OH-Etd+ and Mito Etd+. To colocalize Mito-OH-Etd+ and mitochondria, cells were loaded with MitoFluor green (0.25 μM) for 10 min, rinsed, and visualized with a filter cube with an excitation filter (wavelength/band pass) of 480/30 nm, emission filter of 535/40 nm, and beam splitter at 505 nm. All filters were obtained from Chroma (Rockingam, VT).

Aminotriazole-mediated catalase inactivation.

Cells grown in 60-mm-diameter culture dishes were incubated, in different experimental conditions, in phenol red-free M199 and in the presence of 10 mM 3-amino-1,2,4-triazole. At different time intervals, cells were rinsed with PBS, scraped into 1 ml of 50 mM potassium phosphate, 0.1 mM EDTA, and 0.1% Triton X-100, pH 7, and sonicated 5 s at maximal power. Catalase activity was measured spectrophotometrically following the disappearance of H2O2 at 240 nm (ε240 = 43.6 M−1·cm−1) as described previously (42).

Protein tyrosine nitration.

Western blots were performed with cell lysates, and subcellular fractions were obtained as described for the determination of aconitase activity. Proteins (20 μg) were resolved on 12% SDS-PAGE, followed by Western blotting using a polyclonal anti-3-nitrotyrosine antibody (1:2,000) produced in our laboratory (2) and developed using SuperSignal West Femto maximum sensitivity substrate (Pierce, Rockford, IL). All staining was abolished by pretreatment of the blot with 100 mM sodium hydrosulfite (dithionite) in 100 mM sodium borate, pH 9, which reduces 3-nitrotyrosine to 3-aminotyrosine (13).

For immunocytochemistry studies, fixed cells were blocked, and incubated with the polyclonal anti-3-nitrotyrosine antibody (1:50) and, thereafter, with a goat anti-rabbit IgG antibody labeled with Alexa Fluor 594 (1:400). Nuclei were visualized with DAPI (10 μM). Stained cells were visualized using a Nikon Eclipse TE 200 epifluorescence microscope at a fixed exposure time, using a filter cube with an excitation filter (wavelength/band pass) of 540/25 nm, an emission filter of 605/55 nm, and a beam splitter at 565 nm.

CM-H2DCFDA oxidation.

BAECs grown in 24-well plates were incubated with CM-H2DCFDA (5 μM) for 30 min, rinsed, and incubated in phenol red-free M199 under different experimental conditions. At different time points, fluorescence (excitation filter 480 nm, emission filter 520 nm) of the oxidized probe was measured in a plate reader (Fluostar Galaxy, BMG Labtechnologies, Offenburg, Germany).

Glucose consumption and Krebs cycle activity.

For glucose consumption and Krebs cycle activity assays, cells were grown on 140-mm-diameter culture dishes. Glucose concentration in the incubation medium was measured after 24 h of incubation by a coupled assay with glucose oxidase (0.1 U/ml), horseradish peroxidase (0.6 U/ml), and 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (1 mM). Glucose concentration in the samples was determined by measuring the rate of increase in the absorbance at 420 nm and comparing it with a calibration curve. Krebs cycle activity was determined as the amount of 14CO2 produced by cells incubated with d-[6-14C]glucose (55 mCi/mmol). Freshly harvested BAECs were suspended in 1 ml of Krebs buffer in the presence of either 5.5 mM glucose + 0.25 μCi d-[6-14C]glucose or 30 mM glucose + 1.36 μCi d-[6-14C]glucose. A filter paper soaked in 2 M KOH was placed in the cap, and the flasks were sealed. After 3 h of incubation, 0.1 ml of 6 N H2SO4 was added to the flask and additional overnight incubation under agitation was performed. Filters were placed into scintillation vials containing liquid scintillation cocktail (OptiPhase Supermix; Wallac, Turku, Finland), and radioactivity was measured in a scintillation counter (Wallac Trilux).

Mitochondrial membrane potential.

Mitochondrial membrane potential (ΔΨm) was measured using the J-aggregate-forming lipophilic cation JC-1. Cells were grown on 24-well plates and incubated in M199 with 0.4% FBS in the different conditions and for different time lengths. JC-1 (0.5 μM) was added to the wells, cells were incubated at 37°C for 1 h and rinsed with Krebs buffer, and fluorescence emission at 520 and 590 nm was measured after excitation at 485 nm in a plate reader (Fluostar Galaxy; BMG Labtechnologies). The JC-1 aggregate/monomer ratio was used as an index of ΔΨm. Cells were visualized using a Nikon Eclipse TE 200 epifluorescence microscope, equipped with a digital camera, at a fixed exposure time. JC-1 aggregate fluorescence was observed using a filter cube with an excitation filter (wavelength/band pass) of 540/25 nm, an emission filter of 605/55 nm, and a beam splitter at 565 nm, whereas the monomer was observed using a filter cube with an excitation filter (wavelength/band pass) of 480/30 nm, an emission filter of 535/40 nm, and a beam splitter at 505 nm.

Nitrite and nitrate concentrations.

Cells were grown in 60-mm-diameter culture dishes and incubated in different conditions for 24 h in phenol red-free M199. Nitrite and nitrate concentrations were determined using the Griess method, reducing nitrate with vanadium(III) chloride (38).

MnSOD activity and expression.

MnSOD activity was assessed in cells and mitochondrial fractions by measuring the inhibition of cytochrome c reduction by xanthine oxidase/xanthine in the presence of 3 mM potassium cyanide (19). MnSOD expression was assessed by Western blot. Cellular proteins (40 μg) were resolved in a 13% SDS-PAGE, and Western blots were performed using a rabbit polyclonal anti-human MnSOD antibody, kindly provided by Dr. Ling-Yi Chang (National Jewish Medical and Research Center, Denver, CO).

Glutathione levels.

BAECs grown in 24-well plates were incubated for 3 h in different conditions in phenol red-free M199. At the end of the incubation period, monochlorobimane (40 μM) (54) was added to the incubation medium. Fluorescence was recorded before (blank) and 3 h after the probe was added, in a Fluostar Galaxy plate reader (excitation filter 355 nm, emission filter 460 nm). Glutathione (GSH) content in BAECs was determined by HPLC after derivatization with monobromobimane, compared with GSH standards (33).

Biochemical analyses.

Stock solutions of NOC-18 were prepared in Krebs buffer or phenol red-free M199 immediately before use, and .NO production rates were determined spectrophotometrically following the oxidation of oxy- to methemoglobin (ε577 = 11 mM−1·cm−1) (66). The rate of H2O2 production by glucose oxidase was calculated by measuring oxygen consumption in a water-jacketed Clark-type electrode (YSI model 5300) calibrated with air-saturated distilled water (oxygen concentration 217 μM at 37°C). Protein concentration was determined using the Bradford method.

Image processing.

All image processing was performed with the program ImageJ 1.36 B (Wayne Rasband, National institutes of Health, http://rsb.info.nih.gov/ij/).

Data analysis.

Experiments were performed a minimum of three times with similar results being obtained. Results are expressed as means (SD) or a representative example. For statistical analysis, independent t-tests were performed using the program OriginPro 6.1, with α = 0.05, P < 0.05 for statistical significance.


Aconitase inactivation due to mitochondrial superoxide formation during hyperglycemic challenge to BAECs.

Oxidants formed during hyperglycemic challenge to endothelial cells are thought to arise from mitochondrial O2•−, because MnSOD overexpression inhibits their formation (40). Because O2•− selectively reacts with aconitase, leading to the disruption of the Fe-S cluster (21), we assessed O2•− formation by measuring aconitase inactivation in BAECs.

Confluent cultures of BAECs were incubated in either 5.5 mM glucose (resembling normoglycemic conditions, control) or 30 mM glucose (resembling short-term hyperglycemic condition) for 3 h, and a decrease in total aconitase activity to 78 ± 5% of control values was observed. Longer incubation periods (3 days) did not significantly alter aconitase inactivation (69 ± 7% activity with respect to control values). Superoxide membrane permeability is limited (27, 35, 58); therefore, its reactivity is mostly confined to the compartment where it is being formed. Consequently, we explored the subcellular distribution of aconitase activity in BAECs exposed to hyperglycemic conditions for 3 h, observing that aconitase inactivation (to 40% of control values) occurred mainly in the mitochondrial compartment (Fig. 1A). As a positive control, we exposed the cells to 10 μM 2,3-dimethoxy-1,4-naphtoquinone (DMNQ; a redox cycling agent that produces O2•− in different cell compartments) for 2 h, which reduced both cytosolic and mitochondrial aconitase activities to 2 and 70% of control levels, respectively (not shown).

Fig. 1.

Hyperglycemic challenge to endothelial cells results in aconitase inactivation by increased superoxide formation. A: bovine aortic endothelial cells (BAECs) were incubated for 3 h in 5.5 mM glucose (normoglycemic, N) or 30 mM glucose (hyperglycemic, H) in the absence or presence of NOC-18 (N+NOC, H+NOC). Aconitase activity was measured in mitochondrial and cytosolic fractions. Inset: primary traces of aconitase activity were obtained by a coupled assay measuring NADH formation spectrophotometrically at 340 nm. Cytosolic (triangles, ∼0.18 mg protein) and mitochondrial (squares, ∼0.044 mg protein) fractions in normoglycemic (open symbols) and hyperglycemic conditions (filled symbols) are shown. The results are a representative example of 4 independent experiments. B: BAECs were incubated for 3 h in 5.5 mM glucose (N), 30 mM glucose (H), or 5.5 mM glucose + 10 μM 2,3-dimethoxy-5-methyl-1,4-benzoquinone (Q) or for 24 h in 200 μM buthionine sulfoximine (BSO). Intracellular GSH levels were assessed by adding monochlorobimane (40 μM) and measuring the fluorescence after 3 h of incubation. GSH levels are expressed as a percentage of control (normoglycemic conditions, containing 650 ± 20 pM/106 cells). Values are means (SD); n = 4. *Significantly different from value in normoglycemic conditions. C: MnSOD expression in BAECs incubated for 3 h in 5 mM glucose (N) or 30 mM glucose (H) was assessed by Western blot with an anti-MnSOD antibody (top). Human recombinant MnSOD was used as standard (St). MnSOD specific activity (SA; in units/mg) of the cell lysates and the pure enzyme are shown at bottom.

Aconitase activity depends on the reactions leading to the disruption of the Fe-S cluster (e.g., the reaction with O2•−) and to its reassembling by reincorporation of Fe2+ (24). GSH influences the rate of aconitase reactivation (22); thus we measured its concentration using monochlorobimane. GSH intracellular levels were not modified by the exposure to hyperglycemic conditions (Fig. 1B), suggesting that aconitase inactivation was not due to a decrease in the reactivation rate, but rather to a higher inactivation rate, by O2•−. As a control, we incubated the cells with a redox cycling quinone (2,3-dimethoxy-5-methyl-1,4-benzoquinone) reported to produce large levels of O2•− (29, 51), which reduced the GSH levels to 25% of control levels (Fig. 1B). To confirm that monochlorobimane was reacting mainly with GSH (54) and not with other protein or nonprotein thiols, we treated the cells with 200 μM buthionine-[S,R]-sulfoximine (a specific inhibitor of the enzyme γ-glutamyl-cysteine synthetase) (26) for 24 h, before the exposure to monochlorobimane. This treatment reduced the fluorescence by >50% (Fig. 1B), indicating that the probe could sense variations in GSH levels. Total GSH content of BAECs was determined by HPLC after derivatization with monobromobimane (33) to be 650 ± 20 pM/106 cells (not shown).

MnSOD activity and expression also were determined in cells exposed to normoglycemic and hyperglycemic conditions, and no difference was observed between the different conditions (0.8 ± 0.2 U/mg and 0.8 ± 0.2 U/mg cell protein, respectively) (Fig. 1C), implying that the increase in O2•− steady-state concentrations was due to an increase in the formation of the radical.

When the incubations were performed in the presence of a ·NO donor (1 mM NOC-18, half-life ∼1,000 min), which produced a ·NO flux of 1 μM/min (Fig. 1A), inactivation of aconitase was observed in both the cytosolic and mitochondrial compartments. A similar degree of ·NO-dependent aconitase inactivation was observed in the cytosol of normoglycemic and hyperglycemic cells. However, in mitochondria, ·NO further amplified the differences in inactivation between normoglycemic and hyperglycemic cells.

Mito-HE oxidation by mitochondrial superoxide during hyperglycemic challenge to BAECs.

Superoxide formation in mitochondria was also assessed using Mito-HE (MitoSOX), a hydroethidine derivative targeted to the mitochondria (50). Inside a cell, the hydroethidine moiety can be oxidized to ethidium (Mito-Etd+) by several oxidants and also can react with O2•−, specifically producing the hydroxyethidium derivative (Mito-OH-Etd+) (50, 70). Mito-OH-Etd+ formation was assessed with an epifluorescence microscope using 390-nm excitation and 605-nm emission filters, since OH-Etd+ presents an excitation maximum at 396 nm that is not present in the excitation spectrum of Etd+ (50). As shown in Fig. 2B, the fluorescence of the cells under hyperglycemic conditions was more intense than that of their normoglycemic counterparts (Fig. 2A). Antimycin A (1 μM) was used as a positive control and produced an increase in fluorescence similar to that observed in hyperglycemic conditions (Fig. 2C), but its distribution was more diffuse. Mito-OH-Etd+ fluorescence in hyperglycemic cells colocalized well with mitochondria, as visualized with MitoFluor green, a mitochondrial staining dye (Fig. 2, D–F). When both Mito-OH-Etd+ and Mito-Etd+ were excited, using 540-nm excitation and 605-nm emission filters, cells in hyperglycemic conditions had a small increase in fluorescence compared with their normoglycemic counterparts, and much higher fluorescence was observed in cells treated with 1 μM antimycin A (see Supplemental Fig. 1). (Supplemental data for this article is available online at the American Journal of Physiology-Heart and Circulatory Physiology website.)

Fig. 2.

Hyperglycemic conditions increase Mito-hydroethidine (MitoSOX) oxidation in mitochondria. A–C: fluorescence microscopy (magnification ×400) images of BAECs loaded with MitoSOX (1 μM) and incubated with 5.5 mM glucose (A), 30 mM glucose (B), or 5.5 mM glucose + 1 μM antimycin A (C). Images show the fluorescence exciting at 390 nm, mainly from the hydroxyethidium derivative (Mito-OH-Etd+). D–F: fluorescence microscopy (magnification ×400) images of BAECs loaded with MitoSOX (1 μM) and MitoFluor green (0.25 μM) and incubated in 30 mM glucose for 1 h. Images show the fluorescence arising from Mito-OH-Etd+ (D) or MitoFluor green (E) and the overlay of both signals (F). Contrast and brightness of the images were enhanced using Image J; exactly the same parameters were used in A–C.

Intracellular H2O2 concentration during hyperglycemic challenge to BAECs.

To establish whether the increase in mitochondrial O2•− formation led to an increase in the levels of H2O2 in cells, we measured catalase inactivation in the presence of 3-amino-2,4,5-triazole in cells exposed to hyperglycemic conditions. Compound I, formed by the two-electron oxidation of the heme by H2O2, reacts specifically and irreversibly with aminotriazole, inactivating the enzyme. Therefore, the rate of catalase inactivation in cells incubated with aminotriazole is related to H2O2 concentration in the vicinity of the enzyme (25, 49). In fact, rapid aminotriazole-dependent catalase inactivation was observed when the cells were exposed to intracellular (20 μM DMNQ; Fig. 3A) or extracellular H2O2 sources (glucose oxidase, producing a 1 μM/min H2O2 flux; not shown). In the first 2 h of incubation, catalase activity decreased exponentially to 22 and 18% of the control value, respectively.

Fig. 3.

Hyperglycemic conditions increase aminotriazole-mediated catalase inactivation in BAECs. A: BAECs were incubated with 10 mM aminotriazole in the absence (N) or presence of 20 μM 2,3-dimethoxy-1,4-naphtoquinone (DMNQ). B: BAECs were incubated with (circles) or without (squares) 10 mM aminotriazole in the presence of 5.5 mM glucose (open symbols) or 30 mM glucose (filled symbols). Catalase specific activity (SA) was measured after different incubation periods and expressed as a percentage of control values (SA = 8 ± 2 U/mg). Values are means (SD); n ≥ 4. *Significantly different from value in normoglycemic conditions. Insets: logarithmic plots used to derive the rate constants of exponential decay in catalase activity (kAT).

Significant aminotriazole-dependent catalase inactivation by endogenously produced H2O2 was observed only after 3 h of incubation, and although long incubations were performed (32 h), total inactivation was not achieved (Fig. 3B). Still, in hyperglycemic conditions, more aminotriazole-mediated inactivation of catalase was observed (Fig. 3B), evidence of increased H2O2 formation. If excess glucose was replaced by 3-deoxyglucose, a nonmetabolizable analog, catalase inactivation was similar to that of normoglycemic conditions (Table 1). Catalase activity in the presence of aminotriazole decayed exponentially with pseudo first-order rate constants of 0.046 ± 0.02 and 0.0.062 ± 0.01 h−1 for normoglycemic and hyperglycemic conditions, respectively (Fig. 3B, inset). When 1 mM NOC-18 was present during the incubation, less inactivation of catalase in both normoglycemic and hyperglycemic conditions was observed, suggesting that O2•− was diverted from dismutation to peroxynitrite formation (Table 1).

View this table:
Table 1.

Aminotriazole-mediated catalase inactivation in BAECs

To control that this method was sensitive to variations in H2O2 production at the mitochondrial level, we incubated the cells with aminotriazole in the presence of the mitochondrial uncoupler carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP; 0.5 μM), myxothiazol (0.05 μM), an inhibitor of the electron transport chain known to decrease O2•− formation (62), or antimycin A (1.2 μM) for 24 h. Whereas myxothiazol and FCCP inhibited catalase inactivation, antimycin A increased it (Table 1), in total agreement with the effect that these compounds produce in H2O2 production at the mitochondrial level.

Nitrite and nitrate formation by BAECs in hyperglycemic conditions.

We assessed NO formation, measuring nitrite and nitrate levels in the culture medium, after 24 h of incubation in the presence of 5.5 or 30 mM glucose. BAECs in hyperglycemic conditions produced levels of nitrate and nitrite similar to those in normoglycemic conditions (NO3: 1.0 ± 0.4 and 0.9 ± 0.2 nmol/mg cell protein, respectively; NO2: 1.2 ± 0.8 and 1.1 ± 0.4 nmol/mg cell protein, respectively; not shown).

Protein tyrosine nitration in BAECs in hyperglycemic conditions.

As BAECs produce ·NO, the increase in O2•− production observed under hyperglycemic conditions could lead to peroxynitrite formation and protein tyrosine nitration. Immunocytochemistry using an anti-3-nitrotyrosine antibody showed increased staining after 24 h in hyperglycemic conditions. 3-Nitrotyrosine appeared to be evenly distributed in the entire cell, and a speckled pattern could be appreciated (Fig. 4A). Western blots of mitochondrial fractions using an anti-3-nitrotyrosine antibody showed that various proteins were nitrated in hyperglycemic conditions and confirmed that 3-nitrotyrosine was increased in mitochondrial fractions (Fig. 4B).

Fig. 4.

Hyperglycemic conditions increase total and mitochondrial protein tyrosine nitration in BAECs. A: immunocytochemistry of protein 3-nitrotyrosine in BAECs incubated in 5.5 mM glucose (normoglycemia) and 30 mM glucose (hyperglycemia) for 24 h (magnification, ×600). 3-Nitrotyrosine staining is shown in red, and the nucleus is stained in blue. Contrast and brightness of the images were enhanced using Image J; exactly the same parameters were used in both images. B: Western blots with an anti-3-nitrotyrosine antibody of cellular and mitochondrial proteins of BAECs incubated in 5.5 mM glucose (N) or 30 mM glucose (H). The blots are representative examples of 3 independent experiments.

Mitochondrial oxidant formation during hyperglycemic challenge to BAECs.

Both H2O2 and peroxynitrite are precursors of strong oxidant species [e.g., hydroxyl radical, oxoferryl species, nitrogen dioxide, carbonate radical (48)] that can oxidize the fluorescent probe CM-H2DCF (53, 69). BAECs exposed to hyperglycemic conditions for 3 h (Fig. 5A) oxidized the fluorescent probe CM-H2DCF with higher rates than cells in normoglycemic conditions, evidencing an increase in oxidant formation. When excess glucose was substituted by 3-deoxyglucose, the rate of CM-H2DCF oxidation resembled that of the control (Fig. 5A). The increase in the production of oxidants was abolished by the addition of both FCCP (0.5 μM) and myxothiazol (0.5 μM), in agreement with previous reports (40) and supporting our observations on mitochondrial increase in the formation of O2•−, H2O2, and peroxynitrite (Fig. 5B). In addition, the incubation with MitoQ (0.1 μM), an antioxidant quinone selectively targeted to the mitochondria (30), also decreased the oxidation of CM-H2DCF in hyperglycemic conditions (Fig. 5C), whereas the incubation with 0.1 μM TPP (the triphenylphosphonium moiety responsible for targeting MitoQ to the mitochondria) did not. Incubation with concentrations of TPP ≥0.5 μM also decreased CM-H2DCF oxidation, probably due to a decrease in ΔΨm. To assess oxidant formation in long-term exposure to hyperglycemic conditions, cells were incubated in 30 mM glucose for 3 days. No significant difference in oxidant formation was observed between short- and long-term exposure (not shown).

Fig. 5.

Hyperglycemic challenge to BAECs increases mitochondria-derived oxidants. A: BAECs loaded with chloromethyl dichlorodihydrofluorescein (CM-H2DCF; 5 μM) were incubated in 5.5 mM glucose (N), 30 mM glucose (H), or 5.5 mM glucose + 24.5 mM deoxyglucose (D). B: BAECs loaded with CM-H2DCF were incubated in 5.5 mM glucose (N) or 30 mM glucose (H) alone or plus 0.5 μM carbonyl cyanide p-trifluoromethoxyphenylhydrazone (H + FCCP) or 0.5 μM myxothiazol (H + Myx). C: BAECs were loaded with 0.1 μM MitoQ or 0.1 μM decyltriphenylphosophonium bromide (TPP) and CM-H2DCF (5 μM), washed, and incubated in 30 mM glucose. Fluorescence measurements were performed at different time points. The results are a representative example of 3 independent experiments.

To further evaluate mitochondrial involvement in oxidant formation in hyperglycemic conditions, we produced mitochondria-depleted BAECs (rho0 cells) by growing the cells in the presence of ethidium bromide, a compound that intercalates between the base pairs of the DNA, leading to mutations in mitochondrial DNA (31). Rho0 cells had negligible levels of mitochondrial complex IV subunit 1 (Fig. 6A), and although control cells showed multiple mitochondria with JC-1 aggregates with red fluorescence, evidencing the existence of mitochondria with electrochemical gradient, very few could be observed in rho0 cells (Fig. 6B). Upon exposure to hyperglycemic conditions, no increase in CM-H2DCF oxidation was observed in rho0 cells (Fig. 6C).

Fig. 6.

Mitochondria-depleted rho0 cells do not respond with increased oxidant formation to hyperglycemic challenge. A: cell lysates of BAECs and rho0 cells (80 μg) were analyzed by Western blot with an antibody anti-complex IV subunit 1. Liver mitochondria (Mit; 20 μg) were loaded for control. B: BAECs and rho0 cells were loaded with JC-1 and aggregate (top images) and monomer distribution (bottom images) was observed. Contrast and brightness of the top images were enhanced using Image J; exactly the same parameters were used in both images. C: BAECs (circles) and rho0 cells (triangles) were loaded with CM-H2DCF (5 μM) and incubated in 5.5 mM glucose (open symbols) or 30 mM glucose (filled symbols). Fluorescence measurements were performed at different time points. Values are means (SD); n = 4. *Significantly different from value in normoglycemic conditions.

Hyperglycemic conditions increase glucose consumption, Krebs cycle activity, and ΔΨm in BAECs.

To explore the mechanisms involved in mitochondrial oxidant formation during hyperglycemic challenge to BAECs, we studied glucose metabolization. Cells exposed to 30 mM glucose consumed more glucose than their counterparts under normoglycemic conditions (0.65 ± 0.2 vs. 0.38 ± 0.1 μmol/mg protein, respectively; not shown). Incubation of BAECs in the presence of d-[6-14C]glucose showed that in hyperglycemic conditions, two times more [6-14C]CO2 was produced, evidencing an increase in glucose metabolization in the Krebs cycle (Fig. 7A). BAECs in hyperglycemic conditions showed a gradual increase in ΔΨm, as determined by the ratio of aggregate-to-monomer fluorescence of the cationic lipophilic probe JC-1 (Fig. 7B).

Fig. 7.

BAECs under hyperglycemic conditions metabolize more glucose and have hyperpolarized mitochondria. A: BAECs were incubated with either 5.5 mM glucose + 0.25 μCi d-[6-14C]glucose (N) or 30 mM glucose + 1.36 μCi d-[6-14C]glucose (H), and [6-14C]CO2 formation was assessed. Values are means (SD); n = 6. B: BAECs were incubated in 5.5 mM glucose (control) and 30 mM glucose (H) for different time periods. Mitochondrial membrane potential (ΔΨm) was measured with JC-1 (0.5 μM). The ratio of fluorescence emission at 590 and 520 nm was used as an index of ΔΨm, and results are expressed as the percent increase with respect to the control (normoglycemic) condition. Values are means (SD); n ≥ 3. *Significantly different from value in normoglycemic conditions.


Radical and oxidant species, namely, O2•−, H2O2, ·NO, and peroxynitrite, are constantly formed by endothelial cells. These molecules participate in endothelial cell signaling but also are related to damage and dysfunction in pathological settings (34, 68). In particular, it has been postulated that during hyperglycemic challenge to endothelial cells, radicals arising from mitochondrial O2•− oxidize DNA, activating poly(ADP-ribose) polymerase (PARP), which in turn inactivates glyceraldehyde-3-phosphate dehydrogenase (GADPH), leading to AGEs formation, increased glucose flux through the polyol and hexosamine pathways, and PKC and NF-κB activation (4, 15, 20, 40).

In this study, we have directly measured O2•− formation in mitochondria, along with species derived from it. Using aconitase inactivation for the detection of O2•− (24), we observed that exposure to hyperglycemic conditions enhanced aconitase inactivation in BAECs mainly in mitochondria. Considering that under normoglycemic conditions ∼14% aconitase is inactive, as reported for other mammalian cells (24), whereas under hyperglycemic conditions this value rises to 59% (Fig. 1A) without any change in GSH levels (Fig. 1B), we calculated an 8.8-fold increase in intramitochondrial O2•− steady-state concentration ([O2•−]ss) with respect to normoglycemic conditions. Furthermore, because the steady-state concentration of active aconitase reflects the rates of inactivation by O2•− (kinact = 8.1 × 106 M−1·s−1) (28) and reactivation (kreact = 0.0014 s−1),2 a mitochondrial [O2•−]ss of 250 pM could be estimated for hyperglycemic conditions.

Since MnSOD is not inactivated under these conditions (Fig. 1C), the increase in [O2•−]ss reflects an increase in O2•− formation rate. Having determined a MnSOD specific activity in BAECs (0.8 ± 0.2 U/mg cell protein) and with that of purified MnSOD (4,000 ± 1,000 U/mg) (12, 36, 37, 64) along with the molecular mass of the enzyme (86 kDa for the bovine enzyme) (36) and a mitochondrial volume of 0.2 μl/mg cell protein (calculated from data in Ref. 44), a concentration of MnSOD inside the mitochondria of 12 ± 3 μM could be estimated, in agreement with previous reports (45). Since the main reaction of O2•− consumption in our cell system of low ·NO output is the MnSOD-catalyzed dismutation (k = 2 × 109 M−1·s−1) (5), we anticipate a rate of O2•− formation in the mitochondria in hyperglycemic conditions of 6 μM/s.

Mitochondrial O2•− formation was also assessed using MitoSOX, a hydroethidine derivative targeted to the mitochondria (Mito-HE). In hyperglycemic conditions, increased epifluorescence was observed with excitation at 390 nm, suggesting that Mito-OH-Etd+, product of the reaction with O2•−, was being formed (Fig. 2, A and B). Mito-Etd+ is probably being formed as well in hyperglycemic conditions (see Supplemental Fig. 1), and its contribution to the fluorescence could not be discarded since it also fluoresces, although much less (50), when excited at this wavelength. HPLC analysis of the oxidation products is required to provide conclusive evidence of the relative amounts of each product.3 Colocalization studies with MitoFluor green showed that the fluorescent oxidation products were present mainly in mitochondria (Fig. 2, D–F) in normoglycemic and hyperglycemic cells, and since the cells were loaded with Mito-HE before the exposure to hyperglycemic conditions, the probe was probably oxidized inside the compartment. Thus our studies confirm that mitochondria are key sites of O2•− formation during hyperglycemic challenge to aortic endothelial cells. In cells exposed to antimycin A, however, the fluorescence was more diffuse and probably not restricted to mitochondria. As antimicyn A produces depolarization of the mitochondrial membrane (55), it probably promotes the release of Mito-OH-Etd+ to the cytosol.

Several oxidant molecules can arise from O2•−, SOD-catalyzed dismutation produces H2O2, and O2•− fast reaction with ·NO forms peroxynitrite. Our results on aminotriazole-dependent catalase inactivation showed that in hyperglycemic conditions, H2O2 formation increased and impacted on the steady-state concentration of H2O2 ([H2O2]ss) in the cell (Fig. 3B). With the pseudo first-order rate constants obtained for the decay in catalase activity in the presence of aminotriazole (kAT) and the rate constant for the reaction between catalase and H2O2 to form compound I (k1 = 1.7 × 107 M−1·s−1) (8), [H2O2]ss in the peroxisome was estimated (49), rendering values of 0.76 ± 0.2 and 1.02 ± 0.2 pM for cells in normoglycemic and hyperglycemic conditions, respectively, in agreement with the observed increase in [O2•−]ss. However, the H2O2 concentrations and differences between conditions were smaller than those of O2•−, and whereas differences in [O2•−] could be observed after 3 h of incubation, those in [H2O2] only became apparent after 24 h under hyperglycemic conditions. These discrepancies are probably due, in the first place, to antioxidant systems such as GSH peroxidase, which already has been reported to protect catalase from aminotriazole inactivation (25). Antunes et al. (1) have observed that H2O2 consumption by antioxidant enzymes and membrane permeability result in the establishment of a concentration gradient between cytosol and peroxisomes given by [H2O2]cytosol/[H2O2]peroxisome = 3, and a similar gradient is probably established between the mitochondria and the cytosol. Second, since mitochondrial volume is ∼7% of the cytosolic volume (44), oxidants being formed in this organelle will suffer dilution when escaping to the cytosol, undergoing a steep decrease in concentration.

Tyrosine nitration, in cells under hyperglycemic conditions, evidenced peroxynitrite formation in agreement with related reports (20, 73). Although increased nitration was observed in the mitochondria, it also could be observed throughout the cell, suggesting that peroxynitrite escapes out of the mitochondria to other cellular compartments (Fig. 4). BAECs incubated in normoglycemic and hyperglycemic conditions presented similar levels of the ·NO-derived metabolites nitrite and nitrate, in agreement with Srinivasan et al. (56), although others have reported differently (11, 14). Nevertheless, as proposed recently (46), an increase in the formation of O2•− is sufficient to account for the observed increase in peroxynitrite formation and tyrosine nitration (Fig. 4).

Peroxynitrite can give rise to radical species such as hydroxyl radical, nitrogen dioxide, and carbonate radical, as well as highly oxidant oxo-metal intermediates (48), that are probably responsible for CM-H2DCF oxidation. Whereas in BAECs hyperglycemic conditions led to CM-H2DCF oxidation (Fig. 5), in rho0 cells hyperglycemia did not increase CM-H2DCF oxidation (Fig. 6C), supporting the hypothesis that the primary oxidants were being produced in the mitochondria. Auspiciously, our work showed that the MitoQ could scavenge the oxidants formed under hyperglycemic conditions (Fig. 5B), suggesting that targeting antioxidants to mitochondria might be a successful pharmacological strategy against the development of diabetes complications.

When endothelial cells were incubated with the ·NO donor aconitase, inactivation was observed in the cytosol and mitochondria of both normoglycemic and hyperglycemic cells (Fig. 1A), in agreement with related observations (23). In mitochondria, aconitase inactivation was higher in hyperglycemic than in normoglycemic conditions and was probably largely mediated by peroxynitrite, arising from the reaction between O2•− and ·NO and its derived radicals (i.e., carbonate radical) (61). Actually, incubation with a ·NO donor lowered the intracellular [H2O2] (Table 1), evidencing O2•− diversion to peroxynitrite formation.

In hyperglycemic conditions, BAECs consumed and metabolized more glucose in the Krebs cycle (Fig. 7A), in agreement with a previous report (40), despite aconitase inactivation. Although this might seem paradoxical, the reaction catalyzed by aconitase is not a rate-limiting step of the Krebs cycle; thus its inactivation might not alter the net flux through the pathway. We also observed hyperpolarization of the mitochondrial membrane (Fig. 7B), in agreement with that observed in neurons exposed to high glucose (63). As reported by Korshunov et al. (32), a small increase in ΔΨm, slightly above that of state 3, can result in a strong increase in oxidant formation by mitochondria. Therefore, our results suggest that increased flux of equivalents through the electron transport chain might increase ΔΨm and the reduction of the complexes, leading to electron diversion from reduction by cytochrome oxidase to O2•− formation.

Mitochondria are now acknowledged as relevant signaling organelles, and mitochondrial oxidants are considered to play an important role as signaling molecules, for instance, involved in AMPK activation in endothelial cells (47, 72). Mounting evidence suggests that spatial location is a relevant issue in oxidant signaling in a cell, and in physiological conditions, oxidants appear to be formed in restricted microdomains and subcellular compartments; therefore, the appearance of an oxidant in a different compartment can trigger a specific defense response (60). The spatial distribution of the different oxidant species is defined not only by the site of production but also by their net charge, lipophilicity, and reactivity. Our results suggest that in hyperglycemic conditions, O2•−, formed by the electron transport chain, is retained in the mitochondria, since no aconitase inactivation was observed in the cytosol. However, its reaction products, H2O2 and peroxynitrite, can diffuse out of the compartment, impacting on the cell redox status, and are probably responsible for the reactions leading to hyperglycemic cell damage.


This work was supported by Programa de Desarrollo Tecnológico (Uruguay), Fondo Clemente Estable (Uruguay), and Comisión Sectorial de Investigación Científica (Uruguay) (to C. Quijano and L. Castro) and by the International Center for Genetic Engineering and Biotechnology and Howard Hughes Medical Institute (to R. Radi). G. Peluffo was partially supported by Programa de Investigaciones Biomédicas (Uruguay). V. Valez was partially supported by Fondo Clemente Estable (Uruguay). R. Radi is a Howard Hughes International Research Scholar.


We thank Dr. Lucía Piacenza, Dr. Adriana Cassina, and Dr. Florencia Irigoin (Universidad de la República, Montevideo, Uruguay) for collaboration in Krebs cycle activity determinations, mitochondrial purification, and glutathione measurements, respectively, and Phillip H. Chumley (University of Alabama at Birmingham, Birmingham, AL) and Dr. Bruce A. Freeman (University of Pittsburgh, Pittsburgh, PA) for assistance in the establishment of the BAEC culture. We thank the Frigorífico Carlos Schneck (Montevideo, Uruguay) for providing the bovine aortas to establish the BAEC culture.


  • 1 Oxidant formation in this report was assessed by the oxidation of the fluorescent probe CM-H2DCF. This probe is oxidized by strong oxidants such as hydroxyl radical (OH), carbonate radical CO2•−, nitrogen dioxide (·NO2), and oxidants derived from H2O2 reaction with heme proteins, but not by O2•− (53, 69).

  • 2 Taking into account that GSH levels in BAECs were not affected by the exposure to hyperglycemic conditions and assuming that iron levels are not modified either, we can presume that the rate constant for aconitase reactivation will be similar to that reported for other mammalian cells (24).

  • 3 Although hydroxyethidium and ethidium formation by cells has been assessed using HPLC-based techniques (18, 70), the extraction and quantification of the mitochondrion-targeted derivatives Mito-OH-Etd+ and Mito-Etd+ produced in cells has not yet been performed using this method.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract