Proc. Nati. Acad. Sci. USAVol. 87, pp. 1620-1624, February 1990Medical Sciences

Apparent hydroxyl radical production by peroxynitrite: Implicationsfor endothelial injury from nitric oxide and superoxide

(endothelium-derived relaxing factor/desferrioxamine/ischemia/superoxide dismutase)

JOSEPH S. BECKMAN*t, TANYA W. BECKMAN*, JUN CHEN*, PATRICIA A. MARSHALL*,AND BRUCE A. FREEMAN*tDepartments of *Anesthesiology and *Biochemistry, University of Alabama at Birmingham, Birmingham, AL 35233

Communicated by Irwin Fridovich, December 4, 1989 (received for review October 4, 1989)

ABSTRACT Superoxide dismutase reduces injury in manydisease processes, implicating superoxide anion radical (02 ) asa toxic species in vivo. A critical target of superoxide may benitric oxide (NO-) produced by endothelium, macrophages,neutrophils, and brain synaptosomes. Superoxide and NO- areknown to rapidly react to form the stable peroxynitrite anion(ONOO-). We have shown that peroxynitrite has a pKa of 7.49± 0.06 at 3TC and rapidly decomposes once protonated witha half-life of 1.9 sec at pH 7.4. Peroxynitrite decompositiongenerates a strong oxidant with reactivity similar to hydroxylradical, as assessed by the oxidation of deoxyribose or dimethylsulfoxide. Product yields indicative of hydroxyl radical were5.1 ± 0.1% and 24.3 + 1.0%, respectively, of added perox-ynitrite. Product formation was not affected by the metalchelator diethyltriaminepentaacetic acid, suggesting that ironwas not required to catalyze oxidation. In contrast, desferri-oxamine was a potent, competitive inhibitor of peroxynitrite-initiated oxidation because of a direct reaction between des-ferrioxamine and peroxynitrite rather than by iron chelation.We propose that superoxide dismutase may protect vasculartissue stimulated to produce superoxide and NON under patho-logical conditions by preventing the formation of peroxynitrite.

Vascular injury secondary to ischemia/reperfusion, inflam-mation, xenobiotic metabolism, hyperoxic exposure, andother diseases results in loss of endothelial barrier function,adhesion of platelets, and abnormal vasoregulation. Theability of superoxide dismutase (SOD) to often reduce endo-thelial injury indirectly implicates the participation of super-oxide anion radical (Oj-) with many pathological processes(1). While O-' can be directly toxic (2), it has a limitedreactivity with most biological molecules, raising questionsabout its toxicity per se (3). To account for the apparenttoxicity of O-j in vivo, the secondary production of thefar-more-reactive hydroxyl radical (HO-) is frequently pro-posed to occur by the iron-catalyzed Haber-Weiss reaction:

20- + 2H+ H202 + 02 [1]

O°2+ Fe3+ 02 + Fe2+ [21

Fe2+ + H202-> HOP + OH- + Fe3+. [3]

Although low molecular weight scavengers of HO-, such asmannitol, dimethylthiourea, and dimethyl sulfoxide (DMSO),and the iron chelator desferrioxamine reduce oxidant injury,generation of strong oxidants by the iron-catalyzed Haber-Weiss reaction is not an entirely satisfactory explanation forSOD-inhibitable injury in vivo. Formation of HOP by thispathway requires the interaction of Oj-, H202, and suitablychelated iron-all maintained at low concentrations in vivo

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

1620

because of efficient scavenging systems. The rate constantfor the reduction of Fe3" by O2 is only 1 x 106 M-1 s- (4),while other cellular compounds such as ascorbate can reduceiron and are present in much higher concentrations than °2-(5). Thus, the contribution of °2 to HOP formation by theiron-catalyzed Haber-Weiss reaction may be limited in vivo,suggesting that other reactions may be important in under-standing O-j toxicity to endothelium. We propose that nitricoxide (NO-), a stable free radical, reacts with O-j in manypathological states to yield secondary cytotoxic species.

Recently, endothelium, macrophages, and brain synapto-some preparations have been shown to produce NOR byoxidizing arginine by a calcium-activated NADPH-depen-dent enzyme (6-9). NOR appears to be a major form of theendothelium-derived relaxing factor (EDRF) (10). Vasodila-tory agents such as acetylcholine, ATP, and bradykinininitiate a receptor-mediated influx of Ca2 , triggering theproduction and extracellular release of NO-, which thenactivates soluble heme-containing guanylate cyclases to pro-duce cGMP in vascular smooth muscle and platelets. In-creased cGMP promotes relaxation in vascular smooth mus-cle and inhibits platelet aggregation as well as adhesion ofplatelets to endothelium (11). Macrophages produce NOR aspart of their cytotoxic armamentarium (6).The half-life ofEDRF and NON ranges from 4 to 50 sec (12),

which is approximately doubled by SOD (13, 14). NOR doesnot bind directly to the copper of SOD (15), suggesting thatstabilization involves the scavenging of O- . Because NORcontains an unpaired electron and is paramagnetic, it rapidlyreacts with °-- to form peroxynitrite anion (ONOO-) in highyield (16). In alkaline solutions, ONOO- is stable but has aPKa of 6.6 at 0C (17) and decays rapidly once protonated.°2 + NO-* ONOO- + H+= ONOOH

->HO- + N02-* NN37 + HW. [4]

In the gaseous phase, decomposition of peroxynitrous acidto form HO- and nitrogen dioxide (NO29) is important in theformation of smog and acid rain. Although HOP and NO2 canrecombine to form nitric acid, the rate constant for thisreaction in solution (>3 x 104 M-1-s-; refs. 18 and 19) ismuch slower than most reactions involving HOP. Still, HOPformed by homolytic decomposition of HOONO in aqueoussolutions may not escape beyond the solvent cage beforereacting with NO2', making HOP undetectable. However,peroxynitrite formed by the reaction ofH202 and nitrite atpH2.0 will hydroxylate benzene rings and polymerize methyl

Abbreviations: DMSO, dimethyl sulfoxide; DTPA, diethyltriamine-pentaacetic acid; EDRF, endothelium-derived relaxing factor;MDA, malonyldialdehyde; SOD, superoxide dismutase.tTo whom reprint requests should be addressed at: Department ofAnesthesiology, University of Alabama at Birmingham, Birming-ham, AL 35294.

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

Proc. Natl. Acad. Sci. USA 87 (1990) 1621

methacrylate, reactions characteristic of a free-radical pro-cess (20, 21). By measuring oxygen evolution from HOGoxidation of H202, a 32% yield of HO- could be detectedduring the reaction of H202 and nitrite at pH 5.0 (18).Furthermore, oxygen evolution was inhibited by the HO0scavengers ethanol and benzoate. We have extended theseobservations to more physiological conditions and show thatperoxynitrite initiates many reactions currently used to inferthe action of HO. We propose that the formation of peroxy-nitrite from O2* and NO- is important in recognizing potentialmechanisms of SOD-inhibitable oxidant injury to endothe-lium.

METHODS

Peroxynitrite Synthesis. Peroxynitrite was synthesized in aquenched-flow reactor (22). Solutions of (i) 0.6M NaNO2 and(ii) 0.6 M HCl/0.7 M H202 were pumped at 26 ml/min intoa tee-junction and mixed in a 3-mm diameter by 2.5-cm glasstube. The acid-catalyzed reaction of nitrous acid with H202to form peroxynitrous acid was quenched by pumping 1.5 MNaOH at the same rate into a second tee-junction at the endof the glass tubing. Excess H202 was removed by passageover a 1 x 5 cm column filled with 4 g of granular MnO2. Thesolution was frozen at -200C for as long as a week. Peroxy-nitrite tends to form a yellow top layer due to freeze frac-tionation, which was scraped for further studies. This toplayer typically contained 170-220 mM peroxynitrite as de-termined by absorbance at 302 nm in 1 M NaOH (E302 nm =1670 M-1cm-1; ref. 23). Interference by other absorbingcompounds (e.g., nitrite) was corrected by subtracting thefinal absorbance after adding peroxynitrite to 100 mM po-tassium phosphate (pH 7.4).HO- Assays. The production of HOG from peroxynitrite

decomposition was assayed by the oxidation of DMSO toformaldehyde (24) and deoxyribose to malonyldialdehyde(MDA; ref. 25). All assays were conducted at 370C andincubated for 3 min to allow peroxynitrite to fully decom-pose. Catalase (10 units/ml final concentration; Worthing-ton) was added at the end of each reaction to remove traceH202 left after MnO2 treatment. To assay MDA formed fromdeoxyribose, 1 ml of 2.8% trichloroacetic acid and 1.0 ml of1% thiobarbituric acid in 0.1 M NaOH adjusted to pH 3.5were sequentially added to 1-ml samples, and the mixtureswere then boiled for 10 min. Residual deoxyribose and nitritepresent in the peroxynitrite stock solution caused the ap-pearance ofa shoulder overlapping at 532 nm, the absorbancemaximum oftheMDA thiobarbiturate product. To correct forthis shoulder, absorbances were read at 510, 532, and 560 nm.The shoulder was assumed to decrease linearly over thesewavelengths and was subtracted by standard matrix methods(26). Standards of MDA were prepared daily by the hydrol-ysis of 100 AM 1,1,3,3-tetramethoxypropane.DMSO reacts with HOGto form a methyl radical, ultimately

yielding 0.5 mol each of formaldehyde and methanol underaerobic conditions. Formaldehyde was assayed by reactionwith purpald (Aldrich) as described by Johansson and HakianBorg (27).

Peroxynitrite Decomposition. Peroxynitrite decompositionwas followed by absorbance changes at 302 nm every 0.075s after adding peroxynitrite to a rapidly stirred, temperature-controlled cuvette. The pH of each reaction was measuredafter peroxynitrite addition. The kinetics of peroxynitritedecomposition were strictly first order over at least threehalf-lives as reported (17). The decrease in absorbance at 302nm was used to calculate apparent first-order rate constantsby nonlinear regression (Enzfitter; Elsevier BIOSOFT Pub-lishers).Data Analysis. To estimate the total amount of HO- formed

from peroxynitrite decomposition, a Scatchard-like plot was

used to find the amount of HO trapped in the presence ofinfinite detector concentrations (5). If the rate constant ofHO reacting with a given concentration of detector D to forma product P is kd and the pooled rate constants of HO-undergoing all other reactions are summed to give kr, then themolar concentration of product [P] detected in an assay usinga detector molar concentration [D] may be written as

kd[D]kd[DI + kr

Eq. 5 can be linearized to give the equivalent form of aScatchard plot in which

[P]/[D] = -(kd/kr)[P] + (kd/kr)[HO°]. [6]

In a plot of [P]/[D] versus [P] for a range of detectorconcentrations, the x axis intercept equals the HO- concen-tration that can be trapped at infinite detector concentrations.The inhibition of apparent HO- production, estimated from

the effect of various scavengers on yield, was calculated asdescribed by Winterbourn (28). If Fi is the fraction of inhi-bition from a scavenger i at concentration Si, then simplecompetitive kinetics predict

ks5 Fj[D]kD (1 - Fi)[St] [7]

where ks, and kD are the respective rate constants for reactionof HO- with scavenger and detector. The ratio ks,/kD is alsothe concentration of detector divided by the concentration ofscavenger i required to give 50% inhibition of product yield.If scavenging occurs by simple competition, values of FA[DJ/(1 - F,)[S1] for each scavenger plotted against the known rateconstants of ks, for reaction of HO- with scavenger Si willyield a linear relationship of slope 1/kD (see Fig. 5).

RESULTS

Peroxynitrite addition to 50 mM deoxyribose or DMSO in 50mM potassium phosphate (pH 7.4) at 370C resulted in linear,peroxynitrite concentration-dependent increases of productfor each detector (Fig. 1). By assuming a stoichiometry of0.5mol of product per mol of peroxynitrite added to DMSO, theslopes calculated from different HO* detection systems inFig. 1 corresponded to a yield of 18.8 ± 0.4% formaldehydefor DMSO and 1.53 ± 0.04% MDA for deoxyribose. Productyields from deoxyribose and DMSO were greater at acid pHvalues and were described by the Henderson-Hasselbalchequation, with the apparent pKa being 7.42 ± 0.14 and 7.79± 0.08, respectively (Fig. 2).Peroxynitrite pKa and Half-Life. Keith and Powell (17)

determined the pKa of peroxynitrite to be 6.6 at 0C bymeasuring the rate of peroxynitrite decomposition as a func-tion of pH:

ki= kHA[H+]/([H'1 + Ka), [8]where k1 is the apparent rate constant of peroxynitritedecomposition at a given pH, kHA is the first-order rateconstant for peroxynitrous acid (ONOOH) and Ka is theionization constant for peroxynitrous acid. First-order rateconstants for peroxynitrite decomposition measured over apH range from 4 to 8.5 (Fig. 3) indicated that the pKa was 7.49± 0.06 at 370C. The rate constant kHA for decay of peroxy-nitrous acid was 0.654 ± 0.049 s51, which corresponded to ahalf-life of 1.06 s. At pH 7.4, the half-life of peroxynitriteincreased to 1.9 s.

Medical Sciences: Beckman et al.

[5]

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

1622 Medical Sciences: Beckman et al.

250 500 750Peroxynitrite, AM

FIG. 1. Yield of formaldehyde (m) and MDA (A) from 50 mMDMSO and deoxyribose, respectively, as a function of added per-

oxynitrite at 37TC. The pH of each buffer (50 mM potassiumphosphate) was adjusted to give a final pH of 7.4 after peroxynitriteaddition. Total pH changes were <0.15 pH units. Bars indicatestandard deviations for three replicates per point.

Maximal Yield ofHO. The yield offormaldehyde from 250,uM peroxynitrite increased hyperbolically with greater de-tector concentrations of DMSO at pH 6.0 (Fig. 4). Whenperoxynitrite was allowed to decompose fully by mixing itwith buffer 3 min before DMSO, the trace amounts offormaldehyde produced were the same as controls withDMSO alone. Based upon the estimates from Scatchard-likeplots for each of the HOG detection systems, the maximumyields at infinite detector concentrations relative to theamount of peroxynitrite added at pH 6.0 were 24 ± 1%

formaldehyde from DMSO and 7.0 ± 0.1% MDA fromdeoxyribose.

Competitive Inhibition. Classical HO" scavengers resultedin a concentration-dependent inhibition of yield in all threeHO- assays, with the mean values of Fi[D]/(1 - Fi)[Si] foreach scavenger increasing linearly when plotted against ks,the known rate constants for scavengers reacting with HOE(Fig. 5). The exceptions were when deoxyribose and DMSOwere used as scavengers, both of which deviated stronglyfrom the otherwise linear relationship. The reciprocals of theslopes gave estimates for kD, the reaction rate of peroxyni-trite with the detector, of 1.8 x 109 M-1 s'1 for deoxyriboseand 1.3 x 109 M-1 s-l for DMSO. The kD for deoxyribosereacting with peroxynitrite agrees fairly well with its reactionrate for HOG (1.8 x 109 M-1's-l), while the kD for DMSO isonly 20% of its reaction rate with HO" (7 x 109 M-1 s-'). Therelatively low kD ofDMSO for peroxynitrite may account for

30.

S 20 O20

vll 1001~

5.0 6.0 7.0 8.0 9.0 10.0

pH

FIG. 2. The effect of pH on product yield from 250 tLM peroxy-nitrite at 370C. Potassium phosphate (50 mM) was used over the pHrange of 5.0 to 8.0, while potassium pyrophosphate was used over therange of 7.8 to 9.0. The final pH was determined after peroxynitriteaddition. Solid lines are results of nonlinear regression using theHenderson-Hasselbalch equation. Symbols are as defined in Fig. 1.

FIG. 3. Rate of peroxynitrite decomposition as a function of pH.The pseudo-first-order rate constants at various pH values were

determined by the decrease in A302. The line was fitted by nonlinearregression to Eq. 8, giving a pKa of 7.49 and kHA of 0.654 s-l. Thebuffers were 50 mM sodium acetate (v), 50mM potassium phosphate(*), and 50 mM potassium pyrophosphate (m).

its relatively poor scavenging ability observed in assays usingdeoxyribose as a detector.

In the panels of Fig. 5, the fitted lines did not intercept theorigins. Because the rate constants plotted on the x axis are

literature values for reaction with HO-, the nonzero inter-cepts may be due to a small and relatively constant differencebetween the reaction of each scavenger with peroxynitrite as

compared with HO*.Reaction of Peroxynitrite with Desferrioxamine. To test the

potential contribution of transition metals to peroxynitrite-initiated oxidation, the effects of the metal chelators DTPAand desferrioxamine were examined. In the three HOG assays,DTPA (0.1 mM) had no effect on product yield. In contrast,desferrioxamine was a potent, concentration-dependent in-hibitor of DMSO oxidation by peroxynitrite, but less so forMDA formation from deoxyribose (Fig. 6). Desferrioxaminebound to equimolar ferric iron did not influence yield in thetwo detection systems.

DISCUSSIONPeroxynitrite decomposition generates a strong oxidant ableto initiate many reactions currently used to implicate theaction of HO-. Halfpenny and Robinson (20, 21) previously

35

30

,25

20

15CZ

10

5 tFormnaldehyde, FM0~~~~~~

0 10 20 30 40 50

DMSO, mM

FIG. 4. Yield of formaldehyde as a function of DMSO concen-

tration from 250 gM peroxynitrite. Peroxynitrite (m) was added tovarious concentrations of DMSO at 370C in 50 mM potassiumphosphate (pH 6.0). Data points are means of triplicates. As a

control, DMSO was added 3 min after peroxynitrite had been addedto buffer alone (A). (Inset) Data are transformed to give a Scatchard-like plot (Eq. 6).

100

75

t 50"00

25

0to 0.10

1000

pH

Proc. Natl. Acad. Sci. USA 87 (1990)

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

Proc. Natl. Acad. Sci. USA 87 (1990) 1623

~3 3.

2 .~~~~~~~~2

0 1 2 3 4 5 6 78 0 1 2 3 4 5 6 7 8

ksi, M-1-s-' (x10-9)

FIG. 5. Inhibition by HOW scavengers of product yield from 10mM DMSO (a) or 10 mM deoxyribose (b) after addition of 250 /iMperoxynitrite; ks. is the rate constant for the reaction of HO- withscavenger i. To maximize yields, studies with DMSO and deoxyri-bose as a detector system used 50 mM potassium phosphate (pH 6.0)at 37TC. The scavengers were Tris (A), mannitol (A), ethanol (m),deoxyribose (o), sodium formate (A), benzoic acid (v), salicylic acid(n), DMSO (v), and uric acid (*). Each scavenger was tested overa concentration range of 1-225 mM, and points represent an averageof eight different concentrations tested in duplicate and optimized foreach scavenger to span the range giving 50% inhibition. Rateconstants were those used in ref. 28.

demonstrated the hydroxylation and polymerization of phe-nolic rings, reactions characteristic of HO-, by peroxynitriteat pH 2.0. We have shown herein that peroxynitrite oxidizeddeoxyribose to MDA and DMSO to formaldehyde withsubstantial yields under physiological conditions. Further-more, HOG scavengers were able to reduce the apparent HOGyield in a concentration-dependent fashion in the approxi-mate order of the known rate constants for the scavengersreacting with HOG.

Oxidation of deoxyribose and DMSO by peroxynitritedecreased at higher pH with pKa values close to the pKa of7.49 for peroxynitrite decomposition. The pH dependence iscurious because the HOG assays were end-point assays incu-bated for a minimum of eight half-lives of peroxynitrite at pH9.0. Even at the most alkaline pH tested, sufficient time hadlapsed for >99% of the peroxynitrite anion to become pro-tonated and decompose as peroxynitrous acid. Hence, yieldof oxidized products from deoxyribose or DMSO should nothave been affected by pH. One possible explanation is thatperoxynitrous acid decays by two pathways, the first due todirect internal rearrangement of cis-peroxynitrite (29, 30) togive nitrate without forming strong oxidants and the seconddue to homolytic fission of trans-peroxynitrite to form HOGplus NO2 (Eq. 9).

Internal rearrangement

O==N

. OT-H+ + NO3HO

Homolytic fission

NO, + HO&

0-OH [9]

O=N\N+OH9+ H+

{deactivation}

The transition state for internal rearrangement may have ashorter half-life due to a lower activation energy, making itless likely to be deactivated by the loss of a proton. Alter-natively, the vibrational energy necessary for homolyticfission may be dissipated more readily by proton loss. Thus,internal rearrangement would be favored over homolyticcleavage at high pH, resulting in a lower yield of oxidationproducts.

20 40 60 80 100Desferrioxamine, /uM

FIG. 6. Inhibition of product yield by desferrioxamine of 250 ,uMperoxynitrite with 10 mM DMSO (i) and deoxyribose (v) in 50 mMpotassium phosphate (pH 7.4) and at 37TC.

Desferrioxamine and the Role of Metal Ions. The formationofHO -like oxidants from peroxynitrite did not require metalcatalysis, as shown by the absence ofDTPA inhibition in thethree HOG assays. However, peroxynitrite anion can complexwith iron and copper (31), making metal-catalyzed reactionspossible and in need of further investigation.Much of the evidence for a role of iron in free-

radical-mediated injury in vivo derives from the reduction ofinjury by desferrioxamine. In our experiments, desferriox-amine was a strong inhibitor of peroxynitrite oxidation,which was due to a direct, concentration-dependent reactionbetween desferrioxamine and peroxynitrite rather than bymetal chelation. Desferrioxamine contains three hydroxamicacids (-NOH) adjacent to carbonyl groups, which appears tobe the functional groups involved in peroxynitrite decompo-sition, since these sites are occupied in ferric desferrioxam-ine, which was not inhibitory. The potent inhibition ofperoxynitrite-mediated oxidation by low concentrations ofdesferrioxamine suggests that scavenging of peroxynitritemay be an alternative antioxidant mechanism of desfemox-amine in addition to its role in iron chelation.The present study has focused upon the apparent formation

of HOG from peroxynitrite decomposition, but an equalamount of NO2' is also expected to be formed. NO2' containsan unpaired electron, making it reactive with many freeradicals including intermediates produced by HOG attack ondetector molecules (32). These potential secondary reactionswill reduce apparent HOG yield noncompetitively with respectto the detector, so that Scatchard-like plots can only give aminimal estimate of apparent HOG yield.NO2- is a highly toxic and potent oxidant (Em = 0.9 V),

capable of initiating fatty acid oxidation (33, 34) and nitro-sylation of aromatic amino acids (35). The possibility ofnitrogen-based products being produced in vivo during en-dothelial injury presents a testable conjecture of our hypoth-esis. In support of this possibility, a nitrogen-centered radicalhas been detected during reperfusion of an ischemic isolatedrat heart preparation (36).

Implications for Endothelial Oxidant Ihqury. Peroxynitritemay have considerably greater toxicity than HOG generatedextracellularly; it is formed by a diffusion-limited reactionbetween 0-- and NO-, it has a 1.9-s half-life at pH 7.4 thatpermits diffusion over several cell diameters, it decomposesto generate a potent oxidant similar to HOG in reactivity, andit may cross cell membranes through anion channels as hasbeen demonstrated for Oj- (37).Many pathophysiological processes including reperfusion

of ischemic tissue, acute inflammation, and sepsis mightinitiate events that stimulate NO- production. For example,ischemic endothelium will accumulate Ca2+, but will be

Medical Sciences: Beckman et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0

1624 Medical Sciences: Beckman et al.

unable to support NO- synthesis without oxygen (38). Reper-fusion may stimulate rapid NO- synthesis by providing theoxygen needed to produce NO-, since the intracellular mes-senger, Ca2+, and other substrates-arginine and NADPH-would already be made available by the ischemic insult.

Rates of NO- production have not been measured underpathological conditions and will be an important test of thehypotheses advanced here. However, it is possible to calcu-late that substantial amounts of NO- may be produced insmall vessels by assuming rates of production equivalent tothat of rabbit aorta. Rabbit aortic endothelium can produce 40pmol of NO- min1-cm-2 of luminal surface when stimulatedby ATP (7). In a 50-,um-diameter vessel, the surface area ofendothelium enclosing a 1-liter volume of blood is 2 x 105cm2. Then, the intraluminal rate of NO- production couldtheoretically reach 8 gM min-. Ca2+ entry into endotheliumis also known to increase °-2 production (39). Since the rateof peroxynitrite formation depends upon the product of °-2-and NO- concentrations, it will increase 100-fold for every10-fold increase in °2- and NO- concentration. Thus, rela-tively small increases in rates of°-2 and NO- production maygreatly increase rates of peroxynitrite formation to poten-tially cytotoxic levels.One of the most puzzling effects of SOD is to understand

how it can protect ischemic tissue in experimental modelswhen injected into the circulation just prior to reperfusion.Injury to the endothelium is a major consequence of ische-mia/reperfusion injury, causing edema formation due to theloss of barrier function and favoring platelet adhesion toendothelium. The protective action of SOD may in part bedue to preventing the decomposition of NO- by scavenging°-2, which would help maintain normal vasodilation andinhibit thrombosis. We propose that SOD also protectsendothelium in vivo by preventing the formation of perox-ynitrite, which is toxic due to its decomposition to formpotent, cytotoxic oxidants. On the other hand, the simulta-neous generation of °-2 and NO- by macrophages andactivated neutrophils (6, 40) presents an additional mecha-nism mediating the cytotoxic action of these cell types.

We thank Drs. Irwin Fridovich and James N. Siedow (DukeUniversity, Durham, NC) for their advice, and Kelly Blair and JohnMorgan for their technical assistance. This work was supported bygrants NS 24338 (to J.S.B.) and NS 24275 (to B.A.F.) from theNational Institutes of Health and grants from the Alabama Affiliateof the American Heart Association (to J.S.B.).

1. Freeman, B. A., Jackson, R. M., Matalon, S. & Harding,S. M. (1988) in Endothelial Cells, ed. Ryan, U. S. (CRC, BocaRaton, FL), pp. 13-32.

2. Fridovich, I. (1986) Arch. Biochem. Biophys. 247, 1-11.3. Sawyer, D. T. & Valentine, J. S. (1981) Acc. Chem. Res. 14,

393-400.4. Winterbourn, C. C. (1979) Biochem. J. 182, 625-628.5. Babbs, C. F. & Griffin, D. W. (1989) Free Radicals Biol. Med.

6, 493-503.6. Marietta, M. A., Yoon, P. S., Iyengar, R., Leaf, C. D. &

Wishnok, J. S. (1988) Biochemistry 27, 8706-8711.7. Schmidt, H. H. H. W., Nau, H., Wittfoht, W., Gerlach, J.,

Prescher, K.-E., Klein, M. M., Niromand, F. & Bohme, E.(1988) Eur. J. Pharmacol. 154, 213-216.

8. Palmer, R. M. J., Ashton, D. S. & Moncada, S. (1988) Nature(London) 333, 664-666.

9. Knowles, R. G., Palacios, M., Palmer, R. M. J. & Moncada, S.(1989) Proc. Natl. Acad. Sci. USA 86, 5159-5162.

10. Moncada, S., Herman, A. G. & Vanhoutte, P. M. (1987)Trends Pharmacol. Sci. 8, 365-368.

11. Busse, R., Luckhoff, A. & Bassenge, E. (1987) Naunyn-Schmiedebergs Arch. Pharmacol. 336, 566-571.

12. Furchgott, R. F. & Vanhoutte, P. M. (1989) FASEB J. 3,2007-2018.

13. Rubanyi, G. & Vanhoutte, P. M. (1986) Am. J. Physiol. 250,H815-H821.

14. Gryglewski, R. J., Palmer, R. M. J. & Moncada, S. (1986)Nature (London) 320, 454-456.

15. Gorren, A. C. F., deBoer, E. & Wever, R. (1987) Biochim.Biophys. Acta 916, 38-47.

16. Blough, N. V. & Zafiriou, 0. C. (1985) Inorg. Chem. 24,3504-3505.

17. Keith, W. G. & Powell, P. E. (1969) J. Chem. Soc. A., 453.18. Mahoney, L. R. (1970) J. Am. Chem. Soc. 92, 5262-5263.19. Gratzel, M., Henglein, A., Lilie, J. & Beck, G. (1969) Ber.

Bunsenges. Phys. Chem. 73, 646-653.20. Halfpenny, E. & Robinson, P. L. (1952) J. Chem. Soc., 928-

938.21. Halfpenny, E. & Robinson, P. L. (1952) J. Chem. Soc., 939-

946.22. Reed, J. W., Ho, H. H. & Jolly, W. L. (1974) J. Am. Chem.

Soc. 96, 1248-1249.23. Hughes, M. N. & Nicklin, H. G. (1968) J. Chem. Soc. A,

450-456.24. Klein, S. M., Cohen, G. & Cederbaum, A. I. (1981) Biochem-

istry 20, 6006-6012.25. Halliwell, B. & Gutteridge, J. M. C. (1981) FEBS Lett. 128,

347-352.26. Salaris, S. C. & Babbs, C. F. (1988) J. Pharmacol. Methods 20,

335-345.27. Johansson, L. H. & Hakian Borg, L. A. (1988) Anal. Biochem.

174, 331-336.28. Winterbourn, C. C. (1987) Free Radicals Biol. Med. 3, 33-39.29. Burkholder, J. B., Hammer, P. D. & Howard, C. J. (1987) J.

Phys. Chem. 91, 2136-2144.30. Bayliss, N. S. & Bucat, R. B. (1975) Aust. J. Chem. 28,

1865-1878.31. Hughes, M. N. & Nicklin, H. G. (1970) J. Chem. Soc. A,

925-928.32. Elliot, A. J. & Simsons, A. S. (1984) Can. J. Chem. 62,

1831-1835.33. Pryor, W. A. & Lightsey, J. W. (1981) Science 214, 435-437.34. Patel, J. M. & Block, E. R. (1986) Am. Rev. Respir. Dis. 134,

1196-1202.35. Pritz, W. A., Monig, H., Butler, J. & Land, E. J. (1985) Arch.

Biochem. Biophys. 243, 125-134.36. Zweier, J. L., Flaherty, J. T. & Weisfeldt, M. L. (1987) Proc.

Natl. Acad. Sci. USA 84, 1404-1407.37. Lynch, R. E. & Fridovich, I. (1978) J. Biol. Chem. 253,

4697-4699.38. De Mey, J. G. & Vanhoutte, P. M. (1983) J. Physiol. (London)

335, 65-73.39. Matsubara, T. & Morris, Z. (1986) J. Cell. Physiol. 127,

207-210.40. McCall, T. B., Boughton-Smith, N. K., Palmer, R. M. J.,

Whittle, B. J. R. & Moncada, S. (1989) Biochem. J. 261,293-296.

Proc. Natl. Acad. Sci. USA 87 (1990)

Dow

nloa

ded

by g

uest

on

Aug

ust 4

, 202

0