A method for the detection of superoxide in biological systems

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 215, No. 2, May, pp. 367-3’78, 1982

A Method for the Detection GERALD

M. ROSEN,l

Departments

of

of Superoxide

ELI FINKELSTEIN,

AND

in Biological ELMER

Pharmna logy and Surgery, Duke University Lhrham, North Carolina 27710 Received

December

Systems

J. RAUCKMAN

Medical

Center,

9, 1981

The ability to detect superoxide in biological milieu is filled with a number of difficult problems. For example, the ferricytochrome c assay method cannot be used in the presence of NADPH-cytochrome P-450 reductase since cytochrome c is preferentially reduced by this enzyme. We have found that the superoxide-dependent oxidation of one particular hydroxylamine, 2-ethyl-1-hydroxy-2,5,5-trimethyl-3-oxazolidine, to its corresponding nitroxide, 2-ethyl-2,5,5-trimethyl-3-oxazolidinoxyl, can be used to quantitate superoxide production by hepatic microsomes and purified enzymes. We determined that this assay method is free from most of the problems inherent in other methods for the identification of superoxide.

In 1969, McCord and Fridovich reported the isolation of an enzyme from erythrocytes which was found to inhibit the superoxide-dependent reduction of cytochrome c by xanthine oxidase (1). In addition to establishing the physiological importance of superoxide, this study provided a convenient assay for the specific detection of this free radical in biological milieu. The ease of this assay was an important factor behind the many studies which were to follow (2-8). Subsequent to these investigations, other methods have been developed to detect superoxide and are summarized in Table I. In general, the closer one approximates in tivo conditions, the more difficult it becomes to reliably quantitate superoxide production. The methodologies currently employed to measure superoxide generation in biological systems involve the reaction of superoxide with an indicator to form a stable product. This reaction can involve oxidation, reduction, or binding of superoxide to the indicator. Because a variety of different reaction pathways are involved, the sensitivity, efficiency, and specificity of 1 Author dressed.

to whom

corespondence

should

detecting superoxide varies greatly. For example, the rate constant for the spin trapping of superoxide by 5,5,-dimethyl-lpyrroline-N-oxide (DMP0)2 is 10 M-’ s-l, while the rate constants for the reaction of superoxide with cytochrome c and tetranitromethane at pH 7.8 are about 6 X lo5 and 2 X 10’ M-l s-l (Table I). Since superoxide can act as either a one-electron oxidant or reductant (5), many of the methods described in Table I are prone to artifacts. For example, the oxidation of epinephrine to adrenochrome is a six-electron process and involves the formation of quinone and semiquinone intermediates (14, 22). Superoxide oxidizes epinephrine to a semiquinone which then can reduce molecular oxygen to give more superoxide. Thus, in this reaction, superoxide is both an initiator and a propagator of the free radical reaction. Thus, any enzyme which x Abbreviations used: DMPO, 5,5-dimethyl-l-pyrroline-N-oxide; OXANOH, 2-ethyl-1-hydroxy-2,5,5trimethyl-3-oxazolidine; OXANO, 2-ethyl-2,5,5-trimethyl-3-oxazolidinoxyl; DETAPAC, diethylenetriaminepentaacetic acid, SOD, superoxide dismutase; TEMPO, 2,2,6,6-tetrametbylpiperidinoxyl; TEMPOH, l-hydroxyl-2,2,6,6-tetramethylpiperidine; PMA, pborbol myristate acetate.

be ad-

36 7

0003-9861/82/060367-12$02.00/O Copyright All rights

0 1982 by Academic Press. Inc. of reproduction in any form reserved.

368

ROSEN,

FINKELSTEIN,

AND

TABLE METHODS

Indicator Reductions Ferricytochrome Tetranitromethane Nitroblue tetrazolium

Product

c

Ferrocytochrome

I

OF QUANTITATING

Method

c

Nitroform anion FOrmaZOll

Optical,

Em = 20 mK’

cm-’

& = 14.8 mM-’ cm-l Esm = 30 mK’ cm-’

EM = 2.0 mK’

E,

= 24.5 mK’

Optical, EPR EPR

Other TMPO DMPO Lactoperoxidase

TMPO-OOH DMPO-OOH Compound III

EPR (spin trapping) EPR (spin trapping) Optical, 598 nM

of 0; with indicator

Rat constant* (K’ s-1)

Optical, Optical, Optical,

Adrenoehrome Semiquinane OXANO

are for the reaction

SUPEROXIDE

of detection

Oxidations Epinephrine Tiron OXANOH

a Rate constants

RAUCKMAN

6 X l@ 3.4 x 106 2 x 108

cd

cm-’

Reference

(9,

6)

(10) (9, 11)

(12) (13)

5.6 x 10‘

5 x 10s 6.7 X l@

7

04) (15)

(16) (1-o

10

(17-19) (14, 20. 21)

at pH 7.8.

can reduce the quinone to the semiquinone will, in effect, result in the overestimation of the quantity of superoxide generated. Such an enzyme is NADPH-cytochrome c reductase (23, 24). For this reason, superoxide production reported for this enzyme (4750 nmol/mg/min) using the adrenochrome assay method is erroneous (25). As in the case of epinephrine, NADPHcytochrome c reductase reduces nitroblue tetrazolium to a free radical intermediate which in the presence of oxygen generates superoxide (26, 27). Similarly, different problems exist with other methods described in Table I. Tiron is not a practical indicator of superoxide because the semiquinone radical formed by the oxidation of tiron by superoxide has a half-life of only 10 s (15). Spin trapping superoxide as mentioned previously is most inefficient with a rate constant no greater than 10 M-’ s-l (Table I). The cytochrome c reduction method is inhibited by low-molecularweight reductants such as ascorbate and thiols (6), by enzyme systems which rapidly reduces cytochrome c such as NADPHcytochrome c reductase (9), and by enzymes which rapidly oxidize cytochrome c such as cytochrome oxidase (6). Derivatized cytochrome c analogs are not without limitations (6, 28). Ideally, an assay method for the measurement of superoxide should involve only a one-electron

change in the indicator, should result in the formation of a stable product, and should be relatively free from nonspecific reactions. We have previously demonstrated that superoxide generated by purified FADcontaining monooxygenase can oxidize dialkylhydroxylamines to the corresponding nitroxides (29). The resultant nitroxides were stable indefinitely and easily detectable by electron paramagnetic resonance (EPR) spectrometry. The stoichiometry of the reaction was one nitroxide formed per superoxide, indicating a simple one-electron oxidation of the hydroxylamine as shown. RzN-OH

+ 03 + H+ -

RsN’O

+ H202.

Prior studies, however, by other groups had shown that hydroxylamine autoxidation can itself result in superoxide production (30, 31). In the current study, we demonstrate that the autoxidation of hydroxylamines is due to transition metal impurities, and that this autoxidation can be prevented by chelating agents. Rate constants for the reaction of superoxide with various hydroxylamines were determined and indicate that hydroxylamine oxidation is a method of suitable efficiency for the detection of superoxide. The superoxide-

SUPEROXIDE

DETECTION

BY HYDROXYLAMINE

dependent oxidation of one particular hydroxylamine, 2-ethyl-1-hydroxy-2,5,5trimethyl-3-oxazolidine (OXANOH) to its corresponding nitroxide, 2-ethyl-2,5,5-trimethyl-3-oxazolidinoxyl (OXANO) was found to be free from nonspecific reactions and was used to quantitate superoxide production by hepatic microsomes and purified NADPH-cytochrome c reductase (Table I). Finally, we demonstrate that these hydroxylamines easily diffuse into cells and as such this methodology is useful in the detection of superoxide in cell preparations. MATERIALS

AND

METHODS

General comments. Cytochrome c (Type III), diethylenetriaminepentaacetic acid (DETAPAC), xanthine, bovine erythrocyte superoxide dismutase (SOD), and phorbol myristate acetate were purchased from Sigma Chemical Company, St. Louis, Missouri. Chelex100 ion-exchange resin was obtained from Bio-Rad. All other chemicals were of reagent grade. All buffers were passed through a Chelex-100 ion-exchange column according to the method of Poyer and McCay (32) to remove trace metal impurities. Xanthine oxidase was a gift from Dr. Irwin Fridovich, Department of Biochemistry, Duke University. Purified pig hepatic FAD-containing monooxygenase was a gift from Dr. Daniel M. Ziegler, Department of Chemistry, University of Texas, Austin. The structures of the nitroxides and hydroxylamines used are shown in Fig. 1. The nitroxides were prepared by methods previously described (33). TEMPO (2,2,6,6-tetramethylpiperidinoxyl) and OXAN0 were reduced to their corresponding hydroxylamines, TEMPOH (l-hydroxyl-2,2,6,6-tetramethylpiperidine) and OXANOH, respectively, by bubbling a 10 mM solution of the nitroxide with hydrogen in the presence of the catalyst platinum for 45 min (34). The hydroxylammines should be prepared fresh just prior to use since they are unstable when stored under reducing conditions. However, the corresponding nitroxides are very stable lasting for weeks in a water solution and years in a desiccator. The rate of hydroxylamine oxidation (to produce the corresponding nitroxide) was monitored by measuring the appearance of the electron paramagnetic resonance (EPR) signal. The field strength was set at the midfield peak and scanned with time using a Varian Associates Model E-9 spectrometer. NADPH-cytochrome c reductase was purified from rat hepatic microsomes according to the procedure of Yasukochi and Masters (35), with minor modifications. Emulgen 913 (KAO-Atlas, Japan) at 30% higher concentration was used instead of Renex 690,

369

OXIDATION

9

-Q -G?H

TEMPO

TEMPOH

0

6 OXANO

FIG.

1. Structures

DMPO

6H OXANOH

of the spin

probes.

and the reductase was purified through the affinitycolumn step and no further. Microsomes were prepared from 125-g male Sprague-Dawley rats as follows. Livers were homogenized in 50 mM chelexed potassium phosphate buffer at pH 7.4, containing 0.1 mM DETAPAC and 0.25 M sucrose. The homogenate was centrifuged at 9OOOg for 15 min to sediment mitochondria, nuclei, and cell debris. The supernatant from this 15-min spin was then centrifuged at 100,OOOg for 60 min to sediment the microsomal fraction. The supernatant (cytosol) was saved for experimentation. The microsomes were washed by resuspending the pellet in 0.15 M KCl, 0.1 mM DETAPAC at pH 7.4 and centrifuging this mixture at 100,OOOg for 40 min. After washing the microsomes several times, the microsomal fraction was resuspended in the KC1 solution to be used throughout the experiments described herein. Mitochondria were isolated by resuspending the nuclear-mitochondria-debris fraction in the phosphate-DETAPAC-0.25 M sucrose buffer. This solution was layered over an equal volume of 0.34 M sucrose (containing 50 mM potassium phosphate, chelexed, pH 7.4, and 0.1 mM DETAPAC) and centrifuged at 7OOg for 10 min. The supernatant solution was centrifuged at 50009 for 10 min. The pellet was washed with 0.25 M sucrose buffer, resuspended and centrifuged at 50009 for 10 min. Mitochondria were resuspended in the KC1 solution. Nuclei were prepared as described previously (36). The Coomassie blue dye binding method of Bradford (37) was used for protein assay, with bovine gamma globulin as a standard. The dye was purchased from Bio-Rad. Hepatocytes were prepared from 200-g male Sprague-Dawley rats by the method of Seglen (38). Type IV collagenase was obtained from Sigma Chemical Company. For the microsomal experiments, the DMPO-OH assay for superoxide used 0.18 M DMPO, 5 X 10m4 M cysteine, and 1 pg/ml SOD as a control. The OXANOH assay method (for microsomes, cytosol, nuclei, and mitochondria) used 5 X 10e4 M or 1 mM OXANOH, and 5 fig/ml SOD as a control. The TEMPO reduction assayed employed 2 X lo-’ M TEMPO, 2 X lo-” M cysteine, and 10 pg/ml SOD as a control. Partition coefficients were determined using equal

370

ROSEN,

FINKELSTEIN, TABLE

AND

RAUCKMAN

II

EFFECTSOFMETALSANDCHELATORSONAUTOXIDATIONOF

OXANOH OXANOH

None

0.73 0.050 12.4

lo-’ M DETAPAC 1o-6 M CL?+

lo-6 M cU*+ 1O-6 M Cu*+, lo-’ 1O-6 M &a+, lo-’ 1Om6 M 10m6 M 10m6 M 1W6 M 10m6 lo-’ 10m6 1O-6 10e6 10e6 10e6 10m6

M M M M

Cu’+, Cu’+, Cu’+, CL?+,

Fe3+ Fe3+ Fe3+, Fe3+, M Fea+, M Fe3+, M Fe3+, M Fez+,

oxidation

(pM.min-')

Addition

a

M DETAPAC M EDTA M diethyldithiocarbamate

lo-’ 10F4 M cysteine lo-* M orthophenanthroline 1W4 M CN-

0.075 0.19 0.21 0.73 0.78 0.081b

10e4 lo-’ lo-’ lo-” 10e6 10e6

1.28 1.12 0.050 0.13 0.14 0.17 0.085 0.022

M M M M M M

DETAPAC cysteine orthophenanthroline EDTA CL?+, 5 X lo-* M DETAPAC Cu a+ , 5 x 1o-4 M DETAPAC,

5 x 1om4 M cysteine

The general conditions were 50 mM chelexed potassium phosphate buffer, pH 7.8, containing 5 X 10e4 Ferric chloride and cupric acetate were the metal salts used. “The reaction had gone to completion before a rate could be measured. bThe rate of OXANOH oxidation by Cur+ and CN- was biphasic with an extremely rapid initial rate followed by a very slow rate. The slow rate is listed above. Note.

M OXANOH.

volumes of 10 mM phosphate buffer, pH 7.4, and noctanol. The concentration of DMPO in the buffer was determined by its uv absorbance (17), while concentrations of OXANO and TEMPO were measured by their EPR signals. OXANOH and TEMPOH concentrations were determined by ferricyandie oxidation to OXANO and TEMPO, respectively, then measuring their EPR signal accordingly. RESULTS

Copper Dependence on Hydroxylamine Autoxidation The spontaneous oxidation of OXANOH to the corresponding nitroxide, OXANO, was monitored by EPR spectrometry. As shown in Table II, this oxidation was metal-ion dependent. The autoxidation was faster in nonchelexed than in chelexed buffer. Furthermore, chelating agents decreased the rate of autoxidation to negligble levels. Addition of trace quantities of Cu2+, but not Fe3+, greatly stimulated the

rate of OXANOH oxidation (Table II). Chelating agents varied in their abilities to prevent copper ion-catalyzed oxidation. For example, at a level of 100 PM Cu2+, the relative inhibitory properties of various chelators were found to be: DETAPAC > EDTA > diethyldithiocarbamate > cysteine > orthophenanthroline. Cyanide was unusual in that it stimulated the rate of OXANOH oxidation which was stoichiometric with the amount of Cu2+ present. This is most likely due to the reduction of Cu2+ to Cu’+ by OXANOH followed by the formation of the stable CUE complex (39). In the absence of a chelator, Cu2+ was found to adhere tenaciously to the quartz EPR cell, and could not be removed by prolonged flushing with distilled water. This resulted in artifactually high rates for OXANOH oxidation, unless the adventitious Cu2+was removed. This was accomplished by rinsing the EPR cell with 1 mM

SUPEROXIDE

DETECTION

BY HYDROXYLAMINE

371

OXIDATION

DETAPAC, then flushing the cell with distilled water after each experiment. The data suggest that as little as 1 nM free copper ions can mediate a significant rate of OXANOH oxidation. The use of the chelator, DETAPAC, prevents this artifact. Rate Constant for Reaction with OXANOH

of Superoxide

In order to determine the efficiency of OXANOH oxidation as a method for detecting superoxide, the rate constant for reaction of superoxide with OXANOH at pH 7.8 was determined by competitive kinetics. In this method, a rate constant is determined by using a substance with a known rate constant to inhibit the reaction of OXANOH with superoxide. A xanthine-xanthine oxidase superoxidegenerating system was employed and superoxide dismutase (SOD) was used as a competitive inhibitor. The concentration of OXANOH was held constant, while the concentration of SOD was varied. SOD was standardized against cytochrome c according to the method of McCord and Fridovich (1). The data were analyzed using (18) (WV) - 1 =

~soD(SOD)/~~XANOH(OXAN~H).

In this equation, V and v represent the rates of OXANOH oxidation in the absence and presence of SOD, respectively. The second-order rate constants ksoD and hoxANon are for the reaction of superoxide with SOD and OXANOH, respectively. The data presented in Fig. 2 were consistent with this equation; the plot of (V/v)-1 versus SOD was found to be linear. The ratio of the rate constants for the reaction of superoxide with OXANOH or cytochrome c (koXANOH/k~hromec) is 1.12 X lo-‘. Assuming that the rate of reaction of superoxide with cytochrome c is 6 X lo5 M-’ s-l at this pH (9), koXANOH becomes 6.7 X lo3 M-l s-l. Similarly, the rate constant for reaction of superoxide with TEMPOH was found to be 1.7 X lo3 M-’ s-l. Control experiments indicated that SOD and OXANOH did not inhibit each other. Like-

SOD

(units)

FIG. 2. Inhibition of OXANOH oxidation by SOD. V is the rate of OXANOH oxidation in the absence of SOD, while o is the rate of OXANOH oxidation in the presence of SOD. Conditions were 50 pM chelexed potassium phosphate buffer, pH 7.8, containing 500 FM OXANOH, 400 PM DETAPAC, and xanthine oxidase such that the rate of superoxide production was 3 rM/min, in a total volume of 0.5 ml. Units of SOD are defined by the assay method of McCord and Fridovich (1).

wise, catalase did not diminish the rate of oxidation of OXANOH in the presence of the xanthine-xanthine oxidase system. This is consistent with our earlier observations (29). Ll$fusibility of OXANO Membranes

across Plasma

In order for a hydroxylamine to serve as a probe for intracellular superoxide it should have a suitable lipid/water partition coefficient. The partition coefficient should be large enough so that the probe can diffuse across plasma membranes and enter the cell, yet not be so high that the probe is concentrated in the lipid phase and thus unavailable for reaction with superoxide. In Table III, the octanol/water partition coefficients of OXANOH, OXANO, TEMPOH, TEMPO, and DMPO are listed. The nitroxides have higher octanol/water partition coefficients than do the corresponding hydroxylamines, which in turn have much higher partition coefficients than the spin trap DMPO. The nitroxides, TEMPO and OXANO, had lipid/water partition coefficients which were lo- and

ROSEN,FINKELSTEIN.AND

372 TABLE III

OCTANOL/~ATER PARTITION COEFFICIENTS OF VARIOUS PROBES Probe DMPO OXANO OXANOH

TEMPO TEMPOH

Partition

coefficient 0.093 10.1

3.7 64.6 6.64

Partition coefficients were determined by measuring the partitioning of these substances between equal volumes of 10 mM phosphate buffer, pH 7.4, and n-octanol. Note.

3-fold greater than their corresponding hydroxylamines. An experiment with whole blood suggested that OXANO can penetrate erythrocytes. OXANO was chosen instead of OXANOH because OXANO is directly detectable by EPR spectrometry. The addition of 1 mM OXANO to whole blood resulted in a slight pertubation of its aqueous EPR signal when compared to the spectrum observed in phosphate buffer. The low-field and mid-field peaks were not affected, whereas the high-field peak was broadened and decreased in height by approximately 10%. A splitting of the highfield peak due to partitioning of OXANO into lipid was not detectable. In plasma, the high-field peak was decreased by 5% as compared to an aqueous control. Thus, as judged by its EPR spectrum, most of the OXANO appeared to be free in solution. This fulfills one of the aforementioned criteria for a diffusible probe to detect superoxide, namely, that a large concentration of the probe should be available for reaction with superoxide. By adding OXANO to whole blood, allowing it to incubate at room temperature, and then centrifuging the red blood cells down at 4°C we could determine the partitioning of OXANO between plasma and red blood cells as a function of time. When first added to whole blood, the ratio of OXANO concentration in plasma to that of whole blood was 1.71 f 0.08 (mean -t SD of three independent experiments). Since

RAUCKMAN

the hematocrit of the sample was 46, it can be determined that most of the OXANO was initially located extracellularly. After 20 min, however, the ratio was 1.05 f 0.09, indicating that the plasma and red blood cell compartments had equilibrated and that OXANO can easily cross plasma membranes. Effects of Thiols on OXANOH

Oxidation

It has been reported that thiols directly reduce nitroxides (40), however, we have found no evidence of this either using TEMPO or OXANO. Various thiols, including cysteine, glutathione, and dithiothreitol did not reduce either TEMPO or OXANO, as long as a metal ion chelating agent (e.g., DETAPAC) was present. This was determined by monitoring the intensity of the nitroxide EPR signal in the presence of excess thiol. It was determined that the simultaneous presence of superoxide and thiol led to the rapid reduction of TEMPO to the corresponding hydroxylamine, TEMPOH but not OXANO. This superoxide-thioldependent reduction of TEMPO is the subject of a future paper, nevertheless, this observation clearly demonstrates that the OXANOH oxidation method for detecting superoxide can be carried out in the presence of thiols. Superoxide Production by Putified Microsomal Enzymes Superoxide production by purified NADPH-cytochrome P-450 reductase and FAD-containing monooxygenase was determined using the OXANOH oxidation assay method (Table IV). Unlike the other procedures for the detection of superoxide discussed in the introduction of this paper, NADPH-cytochrome P-450 reductase does not inhibit the oxidation of OXANOH to OXANO. OXANOH oxidation by both FAD-containing monooxygenase and NADPH-cytochrome P-450 reductase are NADPH dependent and are completely inhibited by SOD. Control experiments determined that both purified enzymes cannot reduce OXANO to its corresponding

SUPEROXIDE

DETECTION

BY HYDROXYLAMINE

hydroxylamine, OXANOH. The rate of superoxide production of FAD-containing monooxygenase, as measured by the OXANOH oxidation method was 0.39 mol superoxide/m01 FAD/min. This is in excellent agreement with a value of 0.36 mol superoxide/mol FAD/min, which was obtained using the cytochrome c assay method. The rates of superoxide production by both NADPH-cytochrome P-450 reductase and FAD-containing monooxygenase were approximately 2-fold higher at pH 8.3 than those measured at pH 7.4. On a perprotein basis, NADPH-cytochrome P-450 reductase generated superoxide at a lofold faster rate than did FAD-containing monooxygenase. Superoxide Production Microsmes

by Hepatic

Superoxide production by rat liver microsomes and NADPH was measured using two different methods, namely, OXANOH oxidation and DMPO spin trapping in the presence of cysteine. The spin-trapping method, described previously (17), was used with the following modification. The superoxide adduct of DMPO, DMPOOOH, is unstable, but can be converted into the relatively stable product, DMPOOH by thiols (18). For this reason, the spin-trapping assays were carried out in the presence of cysteine. In order to correct for the effects of residual SOD, and different efficiencies of detecting superoxide, xanthine oxidase was added as an internal standard. Xanthine oxidase of known activity was added to the microsomal system, and the increase in the measured rate of superoxide production was compared to the actual rate of superoxide generated by the xanthine-xanthine oxidase system. A control experiment indicated that microsomes do not affect the rate of superoxide production by the xanthine-xanthine oxidase system. In this manner, we could determine the relative efficiency of microsomal superoxide detection by each of the different methods. SOD was added as a control to determine the specificity of each of

373

OXIDATION TABLE

IV

SUPEROXIDE PRODUCTIONBY

Enzyme FADM, pig Reductase, FADM, pig Reductase, Reductase, Reductase,

liver pig liver liver pig liver rat liver beef adrenal

PURIFIED

ENZYMES

PH

Superoxide production” (nmol 0; mg-’ min-‘)

8.3 8.3 7.4 7.4 7.4 7.4

6.12 64.4 3.38 32.3 17.7 20.0

a Rates of superoxide production are based on the assumption that the molecular weight per flavin of FAD-containing monooxygenase (FADM) is 65,000, and that the molecular weight per flavin of the NADPH cytochrome P-450 reductase (reductase) is 39,000. Buffers used were 0.1 M chelexed potassium phosphate, pH 7.4, containing 0.1 mM DETAPAC, or 50 mM chelexed potassium phosphate, pH 8.3,50 mM tricine, containing 0.1 mM DETAPAC. Additions were 5 X 10e4 M OXANOH, and 150 rg/mi NADPH.

the procedures. The actual rate of superoxide generation by hepatic microsomes was determined using the following expression: Actual rate of microsomal superoxide production equals measured rate of superoxide generation divided by the efficiency of superoxide detection. Using OXANOH as an example, the efficiency of superoxide detection was determined by dividing the increase in the rate of microsomal OXANOH oxidation caused by adding xanthine-xanthine oxidase to the microsomes and NADPH by the rate of OXANOH oxidation caused by xanthinexanthine oxidase alone. This is analogous to the use of an internal standard in scintillation counting and helps correct for nonspecific interferences. Table V compares the rate of microsomal superoxide production, as determined by several different methods. The method with the best specificity was DMPO-OH formation, as 83% was inhibited by SOD. The level of DMPO-OH production under these conditions is much greater than that in the absence of cysteine. In the absence of this thiol, most of the DMPO-OH observed appears to be due

ROSEN, FINKELSTEIN,

374

AND RAUCKMAN

TABLE

V

COMPARISON OF DIFFERENT METHODS OF MEASURING MICROSOMAL SUPEROXIDE AT pH 7.4”

Method DMPO-OH formation OXANOH oxidation OXANOH oxidation Cytochrome c reduction

Percentage inhibition by control 83 18 96 0

Percentage efficiency of 0; detection

Rate of 0, productionb (nM/mg min)

+SOD +SOD -NADPH

76 81 81

2.34 0.73 3.90

+SOD

-

Type of control

a The values refer to the corrected rates of superoxide produced by microsomes and NADPH, as determined by different methods. Conditions are listed under Materials and Methods. The efficiency of 0, detection was determined by (for example) dividing the increase in the rate of OXANOH oxidation caused by adding xanthine and xanthine oxidase to the microsomal system, by the rate of OXANOH oxidation caused by xanthine and xanthine oxidase alone. The addition of xanthine alone had no effect on microsomal superoxide production. b The rate of superoxide production was calculated by subtracting the control rate from the experimental rate.

directly to hydroxyl radical trapping, as its formation is inhibited by catalase and by hydroxyl radical scavengers such as ethanol. Similar results were reported by Lai and Piette (41). In the presence of cysteine, most of the DMPO-OH production becomes insensitive to catalase and hydroxyl radical scavengers, and becomes sensitive to SOD. In chemical studies, we find that only 3% of DMPO-OOH is converted to DMPO-OH via hydroxyl radical trapping of DMPO in the absence of cysteine (42), while in the presence of cysteine, this conversion becomes quantitative. The net result is that DMPO-OH formation can be used as a measure of superoxide production, if thiols are present. There was no discernible inhibition of microsomal cytochrome c by SOD. This is due to the high background level of cytochrome c reduction initiated by the enzyme NADPH-cytochrome P-450 reductase. The observation that microsomal OXANOH oxidation is only 18% inhibitable by SOD is troublesome since we find that it is almost completely inhibited in homogeneous preparations (i.e., purified soluble enzymes). The possibility that OXANOH is a substrate for a NADPH-

requiring microsomal enzyme was excluded by determining that OXANOH (1 mM) does not stimulate microsomal NADPH oxidation. Microsomes were also found not to oxidize OXANOH at an appreciable rate in the absence of NADPH. An alternative explanation for the low inhibition of microsomal oxidation of OXANOH by SOD is that SOD is excluded from the site of formation and reaction of superoxide with OXANOH in these heterogeneous preparations. This was tested by sonicating microsomes in the presence of SOD. In these sonicated preparations SOD inhibited 27% of the OXANOH oxidation under experimental conditions otherwise identical to those where OXANOH oxidation was only 18% inhibited by SOD. Sonication was found not to alter the rate of OXANOH oxidation. These data, combined with the high lipid solubility of OXANOH, suggests that the limited SOD inhibition of microsomal OXANOH is caused by compartmentalization, OXANOH can diffuse to sites in the microsomes where superoxide is generated, whereas SOD cannot. None of the methods listed in Table V for the detection of superoxide were as efficient in the presence of microsomes and

SUPEROXIDE

DETECTION

BY HYDROXYLAMINE

NADPH as they were with xanthine-xanthine oxidase alone. It was also determined that both DMPO-OH (spin-trapping method) and OXANOH oxidation were sensitive to the concentration of microsomes used in the experiment (i.e., efficiency of superoxide detection decreased with increasing protein concentration). The question as to whether or not microsomal superoxide is an artifact caused by enzymatic damage during microsomal preparation was investigated. It is known, for example, that xanthine dehydrogenase only produces superoxide during its purification (43). If this phenomenon were also true of microsomal enzymes, then the inclusion of sulfhydryl or protease inhibitors during the preparation of this fraction should diminish microsomal superoxide production. However, we found that neither glutathione (1 mM) nor the specific lysosomal protease inhibitor leupeptin (50 pM) depressed microsomal superoxide production. Since lysosomal proteases have an acid pH optimum, and since homogenizing tissue in KC1 as opposed to sucrose results in greater organelle damage, the effects of different homogenization mixtures and pH values were determined. Liver was homogenized either in 0.25 M sucrose, 50 mM potassium phosphate, 0.1 mM DETAPAC; or 1.15% KCl, 0.1 M potassium phosphate, 0.1 mM DETAPAC, at pH values of 6.8, 7.4, and 7.8. Microsomes were prepared as described under Materials and Methods. No significant difference in superoxide production was noted. However, there was greater mitochondrial contamination in the KCl-prepared microsomes, as judged by the activity of cytochrome oxidase. Thus, superoxide production by microsomes is not due to enzyme damage caused by tissue homogenization, and may be significant in viva. Superoxide Production by Other Subcellular Fractions There may be enzymes in other cellular fractions that generate superoxide. We examined this possibility by preparing mitochondria, nuclei, and cytosol from rat liver and determining the rate of

OXIDATION

375

OXANOH oxidation with these various hepatic preparations. OXANOH was incubated with hepatic mitochondria and a-ketoglutaric acid (0.1 mM). We were unable to observe any oxidation of OXANOH. This is not surprising since mitochondria are known to contain SOD (Mn species) (44). The cytosol which contains Cu-Zn SOD was inhibited with diethyldithiocarbamate (100 PM). Under these conditions we did not observe any oxidation of OXANOH to its corresponding nitroxide, OXANO. In the case of hepatic nuclei, we had previously demonstrated that hamster hepatic nuclei contain SOD (Cu-Zn) which is easily removed by extensively washing the nuclei (45). Under these conditions, the rate of OXANOH oxidation was found to be 0.70 f 0.02 nmol/min/mg protein. This rate of superoxide production correlates well with published observations (45) and is only a mere fraction of microsomal superoxide production (Table V). OXANOH

Oxidation

by Whole Cells

The diffusibility of OXANO, and the lack of interference by thiols, suggested that OXANOH oxidation may be a useful method of detecting superoxide production by whole cells. We tried this method using human neutrophils, and on isolated rat hepatocytes, and compared the results obtained with different superoxide assays. In all of the experiments conducted using neutrophils, 50 pM DETAPAC was included in the Hanks solution. The added DETAPAC does not diminish the available calcium ions since the Hanks solution contains about 1 mM calcium ions. However, the DETAPAC improves the linearity of the cytochrome c assay for superoxide, and greatly reduces the background rate of OXANOH oxidation, presumably by chelating trace quantities of copper present in the Hanks solution. An analogy may be drawn to the use of Ca’+-EDTA or Cazf-DETAPAC mixtures in treating heavy metal poisoning. The binding constants for heavy metals to these chelators are greater than 7 orders of magnitude higher than the binding constants of these chelators to calcium. In this fashion, the

376

ROSEN,

3

6

9 TIME

12 (mln)

FINKELSTEIN,

15

16

FIG. 3. OXANOH oxidation by stimulated neutrophils. The conditions were 50 pM DETAPAC, 4 X lo6 neutrophils, and 1 mM OXANOH at 25°C. The bottom curve is the resting rate of OXANOH oxidation; the middle curve has 20 pM PMA, and 5 pg SOD present; and the top curve has only 20 @% PMA present.

heavy metals are removed without removing body calcium ions (46). Unstimulated neutrophils did not reduce OXANO, nor did they oxidize OXANOH. Neutrophils which were stimulated with phorbol myristate acetate (PMA) exhibited a lag phase and then oxidized OXANOH at a constant rate. SOD almost completely inhibited OXANOH oxidation (Fig. 3). PMA stimulated neutrophils also exhibited a lag phase before reducing cytochrome c. The rates were not as linear as the OXANOH assay, however, addition of 50 pM NaCN increased the linearity of the cytochrome c assay, indicating that at least part of the problem was due to cytochrome oxidase (6) (data not shown). Unfortunately, the cytochrome c assay consistently yielded a higher rate for superoxide production by neutrophils than did the OXANOH method (Table VI). The cytochrome c method was also found to be dependent on the cytochrome c concentration, that is the rate increased as the cytochrome c concentration was increased. This is in contrast to the reduction of cytochrome c by xanthine-xanthine oxidase, which is insensitive to changes in cytochrome c concentration at levels of 5 PM. On the other hand, the OXANOH oxidation rate was unaffected by probe concentration.

AND

RAUCKMAN

The data presented in Table VII illustrate the results obtained with the OXANOH oxidation assay in isolated rat hepatocytes. The interpretation of these data is difficult because of the high levels of SOD present in hepatocytes. Hepatocytes exhibit a high basal rate of OXANOH oxidation which was inhibited by NaCN, but not by either DETAPAC or cysteine. We initially thought that this inhibition was due to cyanide’s interference in the mitochondrial electron transport system which might account for OXANOH oxidation. However, isolated mitochondria do not oxidize OXANOH. Unfortunately, the reason for the basal rate of OXANOH oxidation and inhibition by cyanide is unclear. Nevertheless, in the presence of 2 mM NaCN, paraquat initiated a dose-related increase in OXANOH oxidation. Cyanide was required in order to measure the stimulatory effect of paraquat, presumably because cyanide inhibits intracellular SOD. Thus, superoxide, generated from the reduction of oxygen by paraquat free radical, oxidizes OXANOH to the corresponding nitroxide, OXANO. TABLE COMPARISON MEASURING PMN

Method Cytochrome 27.5 68.8 68.8

VI

OF DIFFERENT METHODS SUPEROXIDE PRODUCTION

OF AT 25°C

Number of PMNs (X10-6/ml)

Rate of superoxide production (pM min-‘)

1.25 1.25 2.50

3.7 5.7 11.0

1.25 1.25

1.5 1.5

c (pM)

OXANOH (M) 5 x 1o-4 1 x 10-s

Note. Conditions were Hanks balanced salt solution, pH 7.4, containing 2 mg/ml glucose and 5 X 10m5 M DETAPAC. Each experiment was initiated by the addition of 20 pM phorbol myristate acetate. The results shown are the mean of triplicate determinations. In all cases, SOD completely inhibited cytochrome c reduction, and caused better than a 95% inhibition of OXANOH oxidation. The rates shown above are during the linear period of the assays. PMN is polymorphonuclear neutrophils.

SUPEROXIDE TABLE OXANOH

BY HYDROXYLAMINE

VII

OXIDATION BY HEPATOCYTES OXANOH oxidation (nmol OXANOH oxidized/lo6 cells. min)

Addition None 2 mbi 2 mM 2 mM 2 mM 2 mhi 2 mM 2 mM 2 mM

DETECTION

0.22

paraquat DETAPAC CNCN-, 0.2 mM paraquat CN-, 1.0 mM paraquat CN-, 2.0 mM paraquat CN-, 4.0 mM paraquat CN-, 10 mM paraquat

0.24 0.25 0.02

0.07 0.16 0.22 0.31 0.34

Note. Conditions were tyrodes solution, pH 7.4, containing 5 X 10m4M OXANOH, lo-’ M DETAPAC, and 0.93 X lo6 hepatocytes/ml, in a total volume of 0.5 ml at 25’C. Ceil viability was about 80% as determined by trypan blue exclusion. The order of addition of reagents was paraquat, cyanide, then OXANOH.

DISCUSSION

The OXANOH oxidation assay method for the detection of superoxide is the only oxidative method which involves a oneelectron change in the indicator molecule, and in which a stable product is produced. Our data indicate that it is a useful method for detecting superoxide. Its advantages over the cytochrome c assay procedure are that it is less subject to nonspecific interferences. For example, the OXANOH oxidation method is not inhibited by either thiols or NADPH-cytochrome P-450 reductase. Since the OXANOH oxidation procedure is an EPR assay method, it is not subject to turbidity-a problem with assay methods that use spectrophotometric techniques. However, the OXANOH method for the detection of superoxide is not without its problems and limitations. For example, this method is less efficient at detecting superoxide than is cytochrome c. The OXANOH assay procedure yields results comparable to those observed with cytochrome c in homogeneous soluble enzyme systems such as xanthine oxidase or purified FAD-containing monooxygenase. Different results are obtained in more

OXIDATION

377

complex biological systems. The reasons for these discrepancies are unknown, but may involve differences in compartmentalization. Since OXANOH is a weak amine, it will accumulate in acidic organelles like lyposomes, whereas the weak acid superoxide would be extruded from weak acidic compartments. Hence, in membranes with pH gradients, the OXANOH oxidation method would underestimate superoxide concentrations. ACKNOWLEDGMENT This research is supported in part from a grant from the National Institutes of Health, GM 25188, and a grant from the Department of Defense, DAAG 29-86-K-9675. REFERENCES 1. MCCORD, J. M., AND FRIDOVICH, I. (1969) J. Biol. Chewz. 244,6049-6055. FRIDOVICH, I. (1972) Act. Chem Res. 5, 321-326. FRIDOVICH, I. (1974) Advan EnzymoL 41,35-97. FRIDOVICH, I. (1975) Annu Rev. Biochem 44,147159. FRIDOVICH, I. (1976) in Free Radicals in Biology (Pryor, W. A., ed.), Vol. 1, pp. 239-277, Academic Press, New York. 6. MCCORD, J. M., CRAPO, J. D., AND FRIDOVICH, I. (1977) in Superoxide and Superoxide Dismutases (Michelson, A. M., McCord, J. M., and Fridovich, I., eds.), pp. 11-18, Academic Press, New York/London. 7. FRIDOVICH, I. (1978) Science 201, 875-880. 8. FRIDOVICH, I. (1979) in CIBA Foundation Symposium No. 65: Oxygen Free Radicals and Tissue Damage, pp. 77-93, Elsevier, Amsterdam. 9. MCCORD, J. M., AND FRIDOVICH, I. (1968) J. Biol. Chem.

243, 5753-5760.

10. KOPPENOL, W. H., VAN BUUREN, K. H. J., BUTLER, J., AND BRAAMS, R. (1976) Biochim Biophys. Acta 449, 157-168. 11. FORMAN, H. J., AND FRIDOVICH, I. (1973) Arch. Biochem.

Biophys.

158,396-400.

12. MILLER, R. W. (1970) Can& J. Chem 48,935-939. 13. BIELSKI, B. H. J., SHIUE, G. G., AND BAJUK, S. (1980) J. Phys. Chem 84, 830-833. 14. BORS, W., SARAN, M., LENGFELDER, W., MICHEL, C., FUCHS, C., AND FRENZEL, C. (1978) Photo them

Photobiol;

28, 629-638.

15. GREENSTOCK, C. L., AND MILLER, R. W. (1975) B&him Biophys. Acta 396, 11-16. 16. FINKELSTEIN, E. (1980) Ph.D. dissertation, Duke University, Durham, N.C. 17. FINKELSTEIN, E., ROSEN, G. M., RAUCKMAN, E. J.,

378

18. 19. 20.

21. 22. 23.

24.

25. 26. 27.

28. 29. 30. 31.

ROSEN,

FINKELSTEIN,

1 16,676AND PAXTON, J. (1979) MOL &W?WCO 685. FINKELSTEIN, E., ROSEN, G. M., AND RAUCKMAN, E. J. (1980) J. Amer. Chem Sot 102,4994-4999. FINKELSTEIN, E., ROSEN, G. M., AND RAUCKMAN, E. J. (1980) Arch. Bidem. Biophys. 200,1-16. KUTHAN, H., TSUJI, H., GRAF, H., ULLRICH, V., WERRINGLOER, J., AND ESTABROOK, R. W. (1978) FEBS L&t. 91,343~345. NAKAMURA, S., AND YAMAZAKI, I. (1969) Biochim. Biophys. Acta 189, 29-37. MISRA, H. P., AND FRIDOVICH, I. (19’72) J. Biol. Chem. 247, 3170-3175. MASON, R. P. (1979) in Reviews in Biochemical Toxicology (Hodgson, E., Bend, J. R., and Phiipot, R. N., eds.), pp. 151-200, Elsevier, Amsterdam. AUST, S. D., ROERIG, D. L., AND PEDERSON, T. C. (1972) Biochem. Biophvs. Res. Ccnnmun 47, 1133-1137. CASSEL, R. (1979), Ph.D. dissertation, Duke University, Durham, NC. AUCLAIR, C., TORRES, M., AND HAKIM, J. (1978) FEBS L.&t. 89, 26-28. AUCLAIR, C., AND BOIVIN, P. (1980) in International Conference on Oxygen and Oxy-Radicais in Chemistry and Biology, Austin, Texas, p 10. (Abstract) AZZI, A., MONTECUCCO, C., AND RICHTER, C. (1975) B&hem. Biophys. Res. Commun 65,597-603. RAUCKMAN, E. J., ROSEN, G. M., AND KITCHELL, B. B. (1979) Mol. Pharnuwo 1 15,131-137. KONO, Y. (1978) Arch Biochem. Biophys. 186,189197. ELSTNER, E. F., AND HEUPEL, A. (1976) And B&hem 70, 616-620.

AND

RAUCKMAN

32. POYER, J. L., AND MCCAY, Chem 246, 263-269. 33. ROSEN, G. M., RAUCKMAN, K. W. (1977) Tori& I&.

P. B. (1971)

J. BioL

E. J., AND 1, 71-74.

HANCK,

34. ROZANTSEV, E. G. (1970) Free Nitroxyl Radicals, pp. 93-94, Plenum, New York. 35. YASUKOCHI, Y., AND MASTERS, B. S. S. (1976) J. Biol Chem 251, 5337-5344. 36. PATPON, S. E., ROSEN, G. M., RAUCKMAN, E. J., GRAHAM, D. G., SMALL, B., AND ZIEGLER, D. M. (1980) Mol. Phurmaco L 18, 151-156. 37. BRADFORD, 254.

M. M. (1976)

Anal

Biochem.

72.248-

38. SEGLEN, P. 0. (1972) Exp. Cell Res. 74, 450-454. 39. MARTELL, C., personal communications. 40. MORRISETT, J. B., AND DROW, H. R. (1969) J. Biol. Chem 244, 5083-5086. 41. LAI, C. S., AND PIE’ITE, L. H. (1978) Arch. Btixdem Biophys. 190, 27-38. 42. FINKELSTEIN, E., ROSEN, G. M., AND RAUCKMAN, E. J. (1982) Mol. PhurnwxoL, in press. 43. DELLA CORTE, E., AND STRIPE, F. (1972) B&hem J. 126,739-745. 44. MCCORD, J., BOYLE, J. A., DAY, E. D., RIZZOLO, L. J., AND SALIN, M. L. (1977) in Superoxide and Superoxide Dismutases (Michelson, A. M., McCord, J. M., Fridovich, I., eds.), pp. 129-138, Academic Press, New York/London. 45. PATTON, S. E., ROSEN, G. M., AND RAUCKMAN, E. J. (1980) Md Phxwmu.col 18, 588-593. 46. GOLDSTEIN, A., ARONOW, L., AND KALMAN, S. M. (1974) in Principles of Drug Action: The Basis of Pharmacology, pp. 12,116,402, Wiley, New York.