Role of Free Radicals in the LimbTeratogenicity of L-NAME(NG-Nitro-L-Arginine Methyl Ester): A NewMechanistic Model of Vascular DisruptionA.G. FANTEL,* L.D. STAMPS, T.T. TRAN, B. MACKLER, R.E. PERSON, AND N. NEKAHIDepartment of Pediatrics, University of Washington, Seattle, Washington 98195-6320

ABSTRACT In continuing studies of limb ef-fects resulting from fetal exposure to NG-nitro-L-argininemethyl ester (L-NAME), we examined the early timecourse of vascular changes and the effectiveness offetal intraamniotic injection. Vascular engorgement andhemorrhage occurred within 4 hr of L-NAME treatmenton gestational day (gd) 17, and direct injection ap-peared to be as effective as maternal intraperitonealinjection in inducing limb hemorrhage. Further studiesexamined protein nitration and electron transport inhibi-tion in tissues of exposed fetuses. L-NAME causedsignificant increases in nitrotyrosine (NT) formation inlimb but not in heart or brain, and reduced electrontransport rates in limb. Three agents, a-phenyl-N-t-butylnitrone (PBN), a radical trap and inhibitor of induc-ible nitric oxide synthase (iNOS), allopurinol, an inhibitorof xanthine oxidase, and aminoguanidine, a relativelyspecific inhibitor of iNOS, significantly moderated limbhemorrhage and protein nitration in distal limb. Theseresults suggest that L-NAME works directly on the fetallimb vasculature and indicate a cytotoxic role forperoxynitrite, a potent oxidant and nitrating agent thatis the reaction product of nitric oxide and superoxideanion radical. We propose that L-NAME and othervasoactive toxicants disrupt the fetal limb in a sequen-tial process. Initially, nitric oxide (NO) is depleted,causing hemorrhage and edema in the limb. Withinhours, iNOS is induced, resulting in cytotoxic tissueconcentrations of NO and reactive nitrogen speciesthat induce apoptosis and/or necrosis in the limb. Wesuggest that L-NAME exposure may serve as a modelof vascular disruptive limb malformations. Teratology60:151–160, 1999. r 1999 Wiley-Liss, Inc.

The capacity of NG-nitro-L-arginine methyl ester (L-NAME) and related nitric oxide synthase inhibitors tocause limb reduction malformations is well-docu-mented (Diket et al., ’94; Pierce et al., ’95; Salas et al.,’95; Fantel et al., ’97). Defects, which occur whenexposure takes place relatively late in gestation, maybe similar to those induced by other cardioactive vasoac-tive and cardioactive agents, including calcium channelblockers (Danielsson et al., ’89, ’90), cocaine (Websterand Brown-Woodman, ’90), phenytoin (Danielsson et

al., ’92), epinephrine (Jost, ’53a,b), and uterine vascularclamping (Leist and Grauwiler, ’74). Similar limb mal-formations, attributed to vascular disruption, in hu-mans have been reported to occur at an estimatedfrequency of 0.22 per 1,000 newborns (L. Holmes,personal communication). Several exposures have beenassociated with these defects, including cocaine (Shein-baum and Badell, ’92), phenytoin, and chorionic villussampling (Olney et al., ’94, ’95; Mastroiacovo and Botto,’94), although each association remains controversial.Significantly, each of these exposures has the capacityto induce transient reductions in uteroplacental perfu-sion, and it has been proposed that one consequence ofsuch exposure is the formation of free radicals in fetaltissues (Fantel et al., ’92; Fantel, ’96).

Limb sensitivity to L-NAME first appears on gesta-tional day 16 (gd-16) when a few minor digital defectscan be induced, and peaks on gd-17–18 (Fantel et al.,’97). This contrasts with the onset of sensitivity to otherhemodynamic limb teratogens, which occurs as early asgd-14. As with the other agents, the earliest limb effectsof L-NAME consist of vascular engorgement, fluidleakage, and hemorrhage into mesenchyme. Studies ofuterine vascular clamping have shown these changesas early as 60 min after unclamping (Leist and Grau-wiler, ’74). Importantly, the incidence of defects causedby L-NAME is higher than that caused by other vasoac-tive limb teratogens. In contrast to these other agents,all exposed fetuses may have dysmorphic limbs at doseslethal neither to gravida nor fetuses (Fantel et al., ’97).Furthermore, the limb hemorrhages associated withL-NAME exposure are more severe and frequentlyinvolve complete limbs.

The mechanism of L-NAME limb teratogenicity isunknown, but several proposals have been advanced.Diket et al. (’94), unable to detect changes in maternal

Grant sponsor: National Institute of Environmental Health Sciences;Grant numbers: ES-06361, P30 ES-07033.

*Correspondence to: Alan Fantel, Department of Pediatrics 356320,University of Washington, Seattle, WA 98195-6320.E-mail: agf@u.washington.edu

Received 20 January 1999; Accepted 7 May 1999

TERATOLOGY 60:151–160 (1999)

r 1999 WILEY-LISS, INC.

blood pressure in gravid rats exposed to L-NAME,proposed that L-NAME acts directly on the fetal vascu-lature. Fantel et al. (’97) suggested that L-NAMEcauses uteroplacental vasoconstriction and that reac-tive oxygen species (ROS) form during reoxygenation ofhypoxic fetal tissues. Greenberg et al. (’97) performed acomparative study of L-NAME, amiloride, arginine-deficient diet, and other nitric oxide synthase inhibi-tors, and concluded that arginine deficiency, ratherthan nitric oxide inhibition, was the basis of limbreductions in L-NAME-exposed fetuses.

Deficiencies in each of these studies led us to recon-sider the basis of L-NAME limb teratogenicity. Thestudy of Diket et al. (’94) employed chronic dosing andambiguous gestational dating that may have obscuredmaternal pressor changes and ignored possible alter-ations in uterine blood flows. Fantel et al. (’97) failed toexamine maternal hemodynamic changes, and bothDiket et al. (’94) and Greenberg et al. (’97) failed toconsider parallels to the many cardioactive and vasoac-tive agents that induce similar limb malformationpatterns during late gestation. These agents actpromptly, making it unlikely that a fetal argininedeficiency state could develop and cause teratogenicchanges. Also, the absence of a dose-response studymodel in the paper by Greenberg et al. (’97) reducesconfidence in conclusions about relative teratogenicity.Also, Gregg et al. (’98) reported the presence of limbreduction defects in mice lacking a functioning gene forthe endothelial isoform of nitric oxide synthase, support-ing the role of nitric oxide depletion in limb defects.

The studies described here examined both immediateand direct fetal effects of L-NAME in order to helpassess the possible roles of altered maternal and fetalhemodynamics. They also considered the contributionof free radicals, and more specifically, of reactive nitro-gen species to pathogenesis, enabling formulation of anovel, two-stage teratogenic model that may serve moregenerally for the pathogenesis of vascular disruptivemalformations and for the teratogenicity of vasoactiveagents.

MATERIALS AND METHODS

Animal treatment

Time-mated, primigravid rats were obtained from acommercial vendor (B & K Laboratories, Fremont, CA).Pathogen-free animals were housed in our animalfacilities, where light cycled on a 12-hr basis and foodand water were provided ad libitum. On gd-17, gravidasreceived single intraperitoneal injections of L-NAME(25 or 50 mg/kg, i.p.; Sigma Chemical Co., St. Louis,MO) or saline vehicle. Thirty minutes prior to treat-ment with L-NAME (25 or 50 mg/kg) or vehicle, somegravidas received single intraperitoneal injections ofa-phenyl-N-t-butylnitrone (PBN, 50 mg/kg in saline,i.p.; Sigma) or aminoguanidine (AG, 50 mg/kg, insaline, i.p.; Sigma). Other gravidas received three injec-tions of allopurinol (ALLO, 100 mg/kg in saline, i.p.;Sigma) 18 hr prior to, 30 min prior to, and 30 min after

administration of L-NAME (50 mg/kg). For biochemicalstudies, fetuses were dissected on the morning of gd-18and hearts, brains, or limbs from each experimentalgroup were pooled and frozen at 280°C. For teratogenicevaluation, gravidas were sacrificed on the morning ofgd-18 or gd-20, while some litters were examinedearlier. Representative fetuses were photographed.

For direct exposure experiments, gravidas were anes-thetized with halothane and N2O on gd-17, and a smallincision was made in the abdominal wall. One horn ofthe uterus was gently pulled through the incision, and50 µl of an L-NAME solution (2 mg/ml in 0.9% saline) orsaline vehicle were injected directly into the amnioticsac. We estimate that the complete implantation siteweighed approximately 2 g, giving an approximate doseof 25 µg/g, corresponding to 25 mg/kg maternal doses,assuming even distribution of L-NAME between gravidaand fetus. Following injections, the uterus was re-turned to the abdomen and the incision was closed withstainless steel wound clips. Fetuses were examined 24hr later under a stereoscopic dissecting microscope, andrepresentative examples were photographed.

Limb dysmorphogenicity

Gravidas were sacrificed and laparotomized underdeep halothane/N2O anesthesia at 4, 24, or 72 hr aftertreatment. Uteri were removed, individual implanta-tion sites were opened, and external examination offetuses was performed under a stereoscopic dissectingmicroscope. Representative specimens were photo-graphed. In order to compare severity of limb hemor-rhage, the following scale was developed: 0, normal; 1,minimal hemorrhage in distal digit(s); 2, moderatehemorrhage in foot and/or hand; 3, hemorrhage, edemain entire limb(s); and 4, extensive hemorrhage andedema with necrotic limbs separating from body and/orhemorrhage extending into adjacent flank.

Fetal transfer of L-NAME

L-NAME (50 mg/kg, i.p.) was administered on gd-17.Twenty minutes later, the gravida was killed, andblood, a muscle sample, and fetuses were removed foranalysis of L-NAME concentrations by a modification ofthe method of Tabrizi-Fard and Fung (’96). Tissue(approximately 100 µg) was sonicated on ice, andprotein was precipitated using 6 µl 70% perchloric acid.After centrifugation, supernatant was removed andadjusted to pH 3 with NaOH. Processed samples wereinjected onto a C8 HPLC column and eluted with amobile phase consisting of an aqueous solution of 18.5mM heptanesulfonic acid, 10% methanol (pH 2.7) at 1.5ml/min. Detection was performed at 280 nm, andprofiles were compared to those from authentic stan-dard.

Nitrotyrosine formation

Methods used for extraction of nitrotyrosine (NT) andtyrosine (TYR) were based on modifications of themethod of Skinner et al. (’97). Fifty milligrams of tissue

152 A.G. FANTEL ET AL.

(heart, brain, or distal limb) were sonicated twice at 120W for 20 sec in 300 µl of 50 mM potassium phosphatebuffer (pH 7.4) on ice. Supernatant was then digested inproteinase K (1 U/10 mg tissue) at 55°C for 18 hr. Afterthe digest was cooled on ice, 3 ml of ice-cold acetonitrilewere added, and the mixture was vortexed to facilitatetransfer of NT and TYR to the organic phase. After 10min on ice, 1 ml of organic phase was transferred to amicrofuge tube and dried under vacuum. Dried sampleswere stored at 4°C until quantitation.

Quantitation of NT and TYR was performed bymodifications of the methods of Kamisaki et al. (’96)and Shigenaga et al. (’97). Dried material, suspended in50 µl of water, provided sufficient material for two 20-µlaliquots. Authentic NT and TYR standard was added toone aliquot, while 20 µl of water were added to thesecond. Both spiked and unspiked aliquots underwentprecolumn derivatization as described below. Five min-utes prior to injection on the column, 8 µl of 0.1 Msodium borate (pH 8.7) and 8 µl of 4-fluoro-7-nitrobenzo-2-oxa-1,2-diazole (NBD-F) in ethanol (5 mg/ml) weremixed with each sample and incubated for 1 min at60°C. The derivatization reaction was stopped by theaddition of 12 µl of 0.1 M HCl, and samples wereinjected onto the column. Derivatization was performedimmediately prior to injection, since signal rapidlydecays. Concentrations of NBD-F up to 50 mg/ml wereused to derivatize the sample, with no increase insignal strength.

Nitrotyrosine and tyrosine were separated on tworeverse phase C18 columns (8 3 10 compression wa-ters), connected in series by a short length of stainlesssteel tubing. Isocratic elution was performed with amobile phase containing 0.1 M sodium phosphate (pH7.2) and methanol (45:55) at a flow rate of 1 ml/min.Detection of derivatized TYR and NT was performedwith a fluorescence detector (lex 5 470; lem 5 530).Results are reported as nitrotyrosine/tyrosine peakheights. These methods enabled detection of standarddown to 10 nM, which is the lower limit of our detectioncurve and which is sufficiently sensitive for 20-µlinjections of extracted fetal tissues, yielding NT in the40–50-nM range.

Inhibition of electron transport (NADH oxidase)

Experiments were performed only at the highermaternal dose of L-NAME because of the large tissuevolumes required for these assays. Because of morpho-logical similarities between groups, these experimentswere not repeated with fetuses pretreated with PBN.Gravidas received intraperitoneal injections of L-NAME (50 mg/kg) on gd-17. After 4 hr, they weresacrificed under deep halothane/N2O anesthesia andfetuses were removed. Hearts, limbs, and brains weredissected free and pooled individually. They were ultra-sonically disrupted in 1–2 ml of a solution of sucrose,Tris, and EDTA (0.25 M, 10 mM, and 1 mM, respec-tively) and centrifuged at 600g for 10 min. The pelletwas discarded, and the supernatant was used as the

source of broken mitochondria. An aliquot was reservedfor the determination of protein content by the methodof Lowry et al. (’51).

Supernatant, suspended in 30 mM potassium phos-phate buffer, was added to the oxygen polarograph,which records the disappearance of oxygen. After asteady-state rate of O2 loss was attained, NADH wasadded to make a final concentration of 1.0 mM, and thenew rate of O2 disappearance was recorded. The firstrate was subtracted from the former, and O2 consump-tion was expressed as µatoms of O2 consumed/min/mgprotein.

Statistical analyses

Discontinuous traits were analyzed by the x2 test,while continuous variables were subject to paired t-testor to analysis of variance (ANOVA) followed by post hocTukey’s test. Significance was nominally set at 0.05.

RESULTS

Limb hemorrhage

Exposure to L-NAME caused prompt engorgementand subsequent hemorrhage from the fetal limb vascu-lature, irrespective of whether injection was deliveredto the maternal peritoneum or the fetal amniotic sac.Injection of saline was inconsequential. Fetuses receiv-ing L-NAME by direct injection had a significantlyhigher incidence of limb hemorrhage at 24 hr (P , 1026

by x2). Figure 1 presents representative fetuses exam-ined 4 hr after maternal exposure to L-NAME (50mg/kg) with and without PBN pretreatment. Dilationof blood vessels and pooling of blood in limbs can beseen in both, but are more severe and extensive in thefetus that was not pretreated. Figure 2 presents a fetusexamined 4 hr after intraamniotic injection of L-NAME(approximately 25 µg/g). Its appearance is comparableto those exposed via maternal intraperitoneal injection.Figure 3 presents pups on gd-20 after exposure toL-NAME in the presence and absence of PBN. Limbhemorrhage was detected in all groups exposed toL-NAME with or without PBN-pretreatment. As shown,PBN appeared to protect against the hemorrhagiceffects of the low L-NAME dose, but was ineffectiveagainst the higher dose when examined at 72 hr.Higher PBN doses were also ineffective, and maternaltoxicity was observed in gravidas exposed to the combi-nation of the high L-NAME dose and PBN (100 mg/kg).Defect incidence was dose-dependent on gd-20, basedeither on the number of fetuses or litters, as wereported previously (Fantel et al., ’97). Defect severitywas also dose-dependent, and PBN pretreatment signifi-cantly moderated both the incidence and severity oflimb hemorrhage at the low L-NAME dose. By contrast,it was without significant effect at the high L-NAMEdose. These data are presented in Table 1.

Both AG and ALLO effectively and significantlyreduced the severity and incidence of limb hemorrhagecaused by the high L-NAME dose, in contrast to PBN,

FREE RADICALS IN L-NAME LIMB REDUCTION 153

which was only effective against the low dose. Thesedata are presented in Table 2.

L-NAME distribution

Twenty minutes after maternal injection of 50 mg/kgL-NAME, the concentration in maternal blood was 61µg/ml, while concentrations in maternal muscle andfetus were 13 and 34 µg/g, respectively, indicating rapiddistribution to the fetus.

Nitrotyrosine formation

The formation of NT in fetal tissues was examined 24hr after maternal treatment. Results, expressed as theratio of NT to TYR, are presented graphically in Figure4. Labels designate the injected doses of PBN andL-NAME (PBN/L-NAME), respectively, in mg/kg. Be-cause PBN exposure alone had no obvious or significanteffects on NT formation, data from these controls wereomitted from Figure 4. Within individual tissue groups,post hoc analyses compared the results of each experi-ment with its own control rather than with the averageof controls of all experiments.

Nitrotyrosine percent was significantly higher incontrol brain than in other control tissues (Fig. 4).Within tissue groups, the change in NT percent wassignificantly increased only in the limbs of fetusesexposed to the low L-NAME dose. Pretreatment withPBN significantly decreased NT percent only in thelimbs of fetuses exposed to the low L-NAME dose, andthere was no significant difference between controlsand fetuses pretreated with PBN and subsequentlyexposed to L-NAME. Although the high L-NAME doseappeared to increase percent NT and PBN appeared toincrease it further (Fig. 4b), the differences were notstatistically significant.

In order to help understand the PBN results, weexamined the formation of nitrotyrosine in L-NAME-

exposed fetuses pretreated with either AG or ALLO. Inthese later experiments, improved separation of NTresulted in higher NT/TYR ratios. Nitrotyrosine ratioswere significantly lower in both groups of pretreatedfetuses than they were in those exposed only to the highL-NAME dose, as shown in Figure 5. Therefore, bothAG and ALLO effectively moderated protein nitrationassociated with the high L-NAME dose, while PBN wasonly effective with the low L-NAME dose.

Electron transport activity

As noted, low NADH oxidase activities limited thesestudies to 50 mg/kg L-NAME alone. Activity was high-est in homogenates prepared from fetal brain, andlowest in limb. Following maternal treatment withL-NAME, electron transport activity was significantlyreduced in limb but unchanged in heart or brain (Fig. 6).

DISCUSSION

As previously demonstrated, L-NAME is a potentlimb toxicant when rats are exposed during late fetallife. At doses of 25 and 50 mg/kg, all litters containedpups with hemorrhagic limbs. Superficial hemorrhagewas also observed elsewhere, especially in the head,

Fig. 1. Fetuses 4 hr after exposure to L-NAME (50 mg/kg, i.p.) ongd-17. Fetus at left received no pretreatment, while fetus at right waspretreated with PBN (50 mg/kg, i.p.) 30 min prior to L-NAME. Allinjection solutions were made up in saline.

Fig. 2. Fetus 24 hr after exposure to L-NAME (25 µg/g, approximate)by direct injection into the amnion on gd-17. Injection volume con-sisted of 50 µl of saline. Inset: Hemorrhage in distal portion of digit oflower extremity.

154 A.G. FANTEL ET AL.

back, and phallus shortly after treatment or at term,but no associated terata have been observed or re-ported. All fetuses were affected at the higher dose, 72%were affected at the lower dose, and the severity scorewas responsive to dose. Because limb responses toexposure via the fetal intraamniotic or maternal intra-peritoneal route were comparable, we conclude thatL-NAME most likely acts directly on the fetus, and thatmaternal hemodynamic changes are of limited patho-genic importance. Unequivocal acceptance of this con-clusion will require demonstration that intraamnioticL-NAME does not alter maternal hemodynamics, ei-ther systemically or at the uteroplacental interface, orthat such changes, if they occur, are inconsequential.

Studies of direct, intraamniotic injection of L-NAMEargue against a major role for systemic maternalchanges, since the total amount of L-NAME injecteddivided by the maternal body weight yields an approxi-mate dose of 0.2 mg/kg, well below the nonteratogenicdose of 10 mg/kg (Fantel et al., ’97), even assumingcomplete distribution.

The inability of Diket et al. (’94) to detect pressorchanges in gravidas (albeit chronically exposed by theoral route) supports their contention that fetal vascularalterations resulting from the depletion of NO are moreimportant for the development of hemorrhage than arematernal vascular effects. This conclusion is consistentwith reports of limb deficiency in endothelial nitric

Fig. 3. Fetuses on gd-20 after maternal injection of L-NAME on gd-17, with and without PBNpretreatment. All injection solutions were made up in saline. Arrows indicate hemorrhagic areas. 1,Control fetus; 2, L-NAME (25 mg/kg, i.p.); 3, L-NAME (50 mg/kg, i.p.); 4, PBN (50 mg/kg, i.p.); 5, PBN (50mg/kg, i.p.) 1 L-NAME (25 mg/kg, i.p.); 6, PBN (50 mg/kg) 1 L-NAME (50 mg/kg, i.p.).

FREE RADICALS IN L-NAME LIMB REDUCTION 155

oxide synthase (eNOS)-deficient mice (Gregg et al., ’98)and our studies of direct injection of L-NAME. Fetallimb vasculature first becomes sensitive to L-NAMEaround gd-16, while limb sensitivity to other exposuressuch as vascular clamping (Leist and Grauwiler, ’74)and cocaine (Webster and Brown-Woodman, ’90) firstoccurs as early as gd-14. We speculate that this differ-ence may be related to the developmental timing ofcationic amino acid transporters that carry both argi-nine and L-NAME across cell membranes. Proof willrequire analysis of fetal transfer of maternally adminis-tered L-NAME prior to gd-16. Although it is clear thatL-NAME distributes to fetuses in utero on gd-17, ourstudies fail to support a role for fetal arginine deficiencyin the limb hemorrhage induced by L-NAME (Green-berg et al., ’97), since it is unlikely that a deficiencystate would develop and exert its effect within severalhours of exposure.

Spatial patterns of protein nitration correlated withpatterns of tissue disruption. Limbs showed increasesin nitration at both effective L-NAME doses, despite thehigher control values in heart and brain, althoughparadoxically the change was insignificant at the higherL-NAME dose. Maternal pretreatment with PBN signifi-cantly reduced the incidence and severity of limbhemorrhage caused by the low L-NAME dose, but waswithout measurable effect at the high dose. PBN hasbeen reported to reduce both the vascular disruptiveeffects of cocaine and the formation of lipid peroxides in

Fig. 4. Peak ratio, nitrotyrosine to tyrosine (NT/TYR) in tissues offetuses 24 hr after exposure to L-NAME on gd-17, with and withoutPBN pretreatment (50 mg/kg, i.p.). Each bar represents the mean ofthree experiments, using one litter per treatment group in eachexperiment. TYR and NT were extracted and separated by HPLC, asdescribed in Materials and Methods. Fluorescent detection was de-tected on samples derivatized with NBD-F (see text). *Significantlydifferent from 0/0 control (P , 0.05). **Significantly different fromL-NAME-only (P , 0.05). a: L-NAME, 25 mg/kg, i.p. b: L-NAME, 50mg/kg, i.p.

TABLE 2. Amelioration of L-NAME-induced limbhemorrhage by allopurinol and aminoguanidine

Treatment*Total

examinedAbnormal

(%)

Averageseverityscore**

L-NAME 34 34 (100) 3.2 6 0.2L-NAME 1 allopurinol 30 10 (33)*** 1.2 6 0.3***L-NAME 1

aminoguanidine 28 9 (32)*** 0.6 6 0.2***

*L-NAME, 50 mg/kg, i.p.**Mean 6 standard error.***Significantly different from L-NAME only.

TABLE 1. Combined effects of L-NAME and PBN on limb hemorrhage

PBN(mg/kg)

L-NAME(mg/kg) No.

Litterno.

Litters withabnormal fetuses

Average severity*(mean 6 SE)

Percentabnormal**

0 0 82 7 0 0 6 0 050 0 85 7 0 0 6 0 00 25 69 6 6 1.1 6 0.1 72

50 25 90 8 6 0.6 6 0.1*** 49***0 50 62 6 6 2.5 6 0.1 100

50 50 69 6 6 2.3 6 0.1 99

*Average severity score significantly higher in all groups exposed to L-NAME compared tocontrol without L-NAME by ANOVA and post hoc Tukey’s test.**Percent abnormal significantly higher in all groups exposed to L-NAME, compared tocontrol without L-NAME.***Significantly different from L-NAME (25 mg/kg).

156 A.G. FANTEL ET AL.

organogenesis-stage mouse embryos (Zimmerman etal., ’94).

Understanding the basis of the ameliorative activityof PBN is complicated by two features. First, PBN is aspin trap antioxidant that could reduce concentrationsand consequently toxicity of reactive oxygen species. Asdiscussed below, one oxygen radical, superoxide anion,reacts with NO to form peroxynitrite, a potent oxidizingagent most likely involved in the oxidation and nitra-tion of proteins and other macromolecules. Second,PBN is an inhibitor of inducible nitric oxide synthase(iNOS), the presumptive source of the cytotoxic quanti-ties of NO that form peroxynitrite. The correspondencebetween the spatial patterns of tissue breakdown andprotein nitration and the capacity of PBN to moderateboth features at the low L-NAME dose support a rolefor peroxynitrite in the limb reduction malformationsresulting from L-NAME exposure. This conclusion re-ceived additional support from the spatial pattern ofelectron transport inhibition in L-NAME-treated fe-tuses, since NO and peroxynitrite effectively reduceelectron flow (Drapier and Hibbs, ’88; Cassina andRadi, ’96).

Studies employing ALLO and AG were undertaken inan effort to determine the properties of PBN that arerelevant to the moderation of L-NAME-induced limbtoxicity. Allopurinol inhibits xanthine oxidase, reducingsuperoxide anion radicals from that source, while AG isa relatively specific inhibitor of iNOS. Pretreatmentwith either agent effectively ameliorated the limb ef-fects of the high L-NAME dose, and both reduced theformation of nitrotyrosine. The ALLO findings indicatethat superoxide plays a role in limb toxicity, and takenalone might suggest a role for oxidative stress or

macromolecular oxidation. On the other hand, the AGresults strongly implicate NO formation catalyzed byiNOS. A reasonable explanation of the combined find-ings is that both relatively stable radicals, O2

2 and NO,form during ischemia, or more likely during reperfu-sion, and react to produce highly reactive and cytotoxicperoxynitrite.

Based on 1) prompt engorgement and hemorrhagefrom fetal limb vasculature following L-NAME expo-sure, 2) increased nitrotyrosine formation in limbs afterL-NAME treatment, 3) reductions in electron transportrates in fetuses exposed to L-NAME, and 4) moderationof L-NAME-induced limb hemorrhage and nitrotyro-sine formation by PBN, AG, and ALLO pretreatment,we propose the following sequential mechanism of limbteratogenicity: L-NAME inhibits endothelial nitric ox-ide synthase (eNOS), the activity of which is critical tothe maintenance of vascular homeostasis (Palmer etal., ’87; Moncada and Higgs, ’91) in limb. As NOconcentrations in endothelium decline, vascular perme-ability increases in postcapillary venules, resulting inhemorrhage and edema into surrounding mesenchyme(Baldwin et al., ’98; Grisham et al., ’98). Subsequently,NO concentrations increase significantly as ischemiainduces iNOS via the hypoxia-responsive element inthe iNOS gene (Melillo et al., ’95) or as a result ofcytokine activation. Elevated NO concentrations mayalso result from nonenzymatic reduction of nitriteunder conditions of low tissue pH and increased reduc-tion pressure (Zweier et al., ’95). In any case, NO reactswith superoxide to form peroxynitrite. This sequence,

Fig. 5. Peak ratio, nitrotyrosine to tyrosine (NT/TYR) in tissues offetuses 24 hr after exposure to L-NAME (50 mg/kg), with and withoutAG or ALLO pretreatment. Three litters were used for L-NAME onlyand L-NAME 1 ALLO, and two were used for L-NAME 1 AG. TYRand NT were extracted and separated by HPLC, as described inMaterials and Methods. Fluorescent detection was detected on samplesderivatized with NBD-F (see text). *Significantly different from L-NAME only (P , 0.05).

Fig. 6. NADH oxidase activities in fetal tissues 4 hr after exposure toL-NAME (50 mg/kg, i.p.). Four litters were used for limb and brain,and three were used for heart. Tissues were sonicated and centrifuged(600g), and the supernatant was used as the source of brokenmitochondria. Rates of oxygen utilization were determined in theoxygen polarograph in the presence and absence of NADH as theelectron donor. The difference between the two rates representedNADH oxidase activity. *Significantly lower than control limb(P , 0.05).

FREE RADICALS IN L-NAME LIMB REDUCTION 157

Fig. 7. Proposed scheme for limb effects of L-NAME. Other vasoactive toxicants may produce similareffects by increasing tissue levels of superoxide anion. Also shown are points at which modifiers ofteratogenicity could act.

158 A.G. FANTEL ET AL.

along with points at which the experimental inhibitorscould intervene, is presented in Figure 7.

The temporal pattern of induction of vascular engorge-ment and its moderation by PBN, AG, and ALLOsuggest that iNOS is rapidly induced in fetal limb, witha corresponding increase in NO. It has been reportedthat NO formation peaks approximately 12 hr afteriNOS induction (Miyajima and Kotake, ’97), a transcrip-tional process. The basis of precocious fetal formation ofNO by iNOS remains to be determined, but it isconceivable that this process represents an adaptiveresponse to potential uteroplacental ischemia. NO re-versibly inhibits electron transport by binding cyto-chrome a3 in mitochondrial complex IV, stacking elec-trons on the chain and promoting electron leakage toavailable molecular oxygen, forming superoxide (Cas-sina and Radi, ’96). Superoxide then avidly reacts withNO to form peroxynitrite (OONO2). Peroxynitrite and/orits products, including peroxynitrous acid, hydroxylradical, NO2, and N2O3 attack macromolecules, depletethiols, and irreversibly inhibit electron transport, initi-ating apoptosis (Yabuki et al., ’97; Richter, ’97) and/ornecrosis.

The rate constant for the reaction between O22 and

NO that forms peroxynitrite (OONO2) is 6.9 3 109 M21

sec21 (Huie and Padmaja, ’93). Under normal condi-tions, NO concentrations in vasculature exceed those ofsuperoxide by 3–4 orders of magnitude (Grisham et al.,’98). This concentration differential ensures sufficientNO to perform physiological functions mediated throughthe activation of guanylate cyclase. These include themaintenance of vascular tone and integrity. At the sametime, O2

2 concentrations are insufficient to permitformation of toxicologically significant quantities ofmore highly reduced and consequently more reactiveoxidants such as hydroxyl radical.

It is well-recognized that a ‘‘burst’’ of ROS formationtakes place during the reperfusion of ischemic tissues.Although the subcellular site(s) of formation of theseROS is controversial, most believe that superoxide isformed by xanthine oxidase converted from xanthinedehydrogenase during ischemia (Granger et al., ’81;Roy and McCord, ’83) and the mitochondrial electrontransport chain (Kehrer and Park, ’90; Park et al., ’91).Because reaction with this increased O2

2 depletes NOas does L-NAME, we propose that L-NAME administra-tion may represent a useful model of vascular disrup-tive limb malformation, whether of exogenous or endog-enous origin. In adult tissues, a large portion of thesuperoxide that depletes NO forms in circulating neu-trophils and macrophages. Additional reductions in NOconcentrations occur as eNOS is transcriptionally andposttranscriptionally downregulated by hypoxia (Mc-Quillan et al., ’94), and reductions in arginine transportfurther depress eNOS activity (Block et al., ’95).

While is clear that the limbs are most sensitive to theteratogenicity of L-NAME and other vasoactive agentsduring fetal life, the basis of spatial sensitivity remainsto be determined. One possibility is that limb vascula-

ture is highly and/or precociously dependent on NO. Onthe other hand, we have shown that superoxide dismu-tase activities are significantly lower in limb than inother fetal tissues (excepting brain; Mackler et al., ’98),and that limb has a significantly greater capacity toaccumulate superoxide anion radical (Fantel et al., ’95).It is possible, therefore, that the sensitivity of limbsresults from heightened capacity to form O2

2, whichcan deplete NO required to maintain vascular integrity.

These studies indicate that L-NAME causes promptvascular changes in fetal limb. They confirm the dose-responsive nature of L-NAME limb teratogenicity andspatially correlate L-NAME-induced limb hemorrhagewith protein nitration and reductions in electron trans-port activity. These data, combined with the capacity ofPBN, AG, and ALLO to reduce the incidence andseverity of limb hemorrhage as well as protein nitrationin limb, suggest that 1) L-NAME acts directly on thefetal vasculature by depleting NO, causing vascularengorgement and leakage; 2) following initial NO deple-tion, NO concentrations increase, most likely as aconsequence of induction of iNOS during ischemia/reperfusion; 3) secondarily increased NO concentra-tions result in peroxynitrite formation; and 4) peroxyni-trite and its products play a role the final malformationsinduced by L-NAME. If this scheme is correct, L-NAMEexposure may serve as an experimental model of utero-placental ischemia/reperfusion and of limb reductionassociated with vascular disruption.

LITERATURE CITEDBaldwin AL, Thurston G, Al Naemi H. 1998. Inhibition of nitric oxide

synthesis increases venular permeability and alters endothelialactin cytoskeleton. Am J Physiol 274:1776–1784.

Block EK, Herrera H, Couch M. 1995. Hypoxia inhibits L-arginineuptake by pulmonary artery endothelial cells.Am J Physiol 269:574–580.

Cassina A, Radi R. 1996. Differential inhibitory action of nitric oxideand peroxynitrite on mitochondrial electron transport. Arch Bio-chem Biophys 328:309–316.

Danielsson BRG, Reiland S, Rundqvist E, Danielson M. 1989. Digitaldefects induced by vasodilating agents: relationship to uteroplacen-tal blood flow. Teratology 40:351–358.

Danielsson BRG, Danielson M, Reiland S, Rundqvist E, Dencker L,Regard CG. 1990. Histological and in vitro studies supportingdecreased uteroplacental blood flow as explanation for digitaldefects after administration of vasodilators. Teratology 41:185–193.

Danielsson BRG, Danielson M, Rundqvist E, Reiland S. 1992. Identi-cal phalangeal defects induced by phenytoin and nifedipine suggestfetal hypoxia and vascular disruption behind phenytoin teratogenic-ity. Teratology 45:247–258.

Diket AL, Pierce MR, Munshi UK, Voelker CA, Eloby-Childress S,Greenberg SS, Zhang X-J, Clark DA, Miller MJS. 1994. Nitric oxideinhibition causes intrauterine growth retardation and hind-limbdisruptions in rats. Am J Obstet Gynecol 171:1243–1250.

Drapier JC, Hibbs JB Jr. 1988. Differentiation of murine macrophagesto express nonspecific cytotoxicity for tumor cells results in L-arginine-dependent inhibition of mitochondrial iron-sulfur enzymesin the macrophage effector cells. J Immunol 140:2829–2838.

Fantel AG. 1996. Reactive oxygen species in developmental toxicity:review and hypothesis. Teratology 53:196–217.

Fantel AG, Barber CV, Carda MB, Tumbic RW, Mackler B. 1992.Studies of the role of ischemia/reperfusion and superoxide anionradical production in the teratogenicity of cocaine. Teratology46:293–300.

FREE RADICALS IN L-NAME LIMB REDUCTION 159

Fantel AG, Person RE, Tumbic RW, Nguyen T-D, Mackler BM. 1995.Studies of mitochondria in oxidative embryotoxicity. Teratology52:190–195.

Fantel AG, Nekahi N, Shepard TH, Cornel LM, Unis AS, Lemire RG.1997. The teratogenicity of NG-nitro-L-arginine methyl ester (L-NAME), a nitric oxide synthase inhibitor, in rats. Reprod Toxicol11:709–717.

Granger DN, Rutili G, McCord JM. 1981. Superoxide radicals in felineintestinal ischemia. Gastroenterology 81:22–29.

Greenberg SS, Lancaster JR, Xie J, Sarphie TG, Zhao X, Hua L,Freeman T, Kapusta DR, Giles TD, Powers DR. 1997. Effects of NOsynthase inhibitors, arginine-deficient diet, and amiloride in preg-nant rats. Am J Physiol 273:1031–1045.

Gregg AR, Schauer A, Shi O, Liu Z, Lee CGL, O’Brien WE. 1998. Limbreduction defects in endothelial nitric oxide synthase-deficient mice.Am J Physiol 275:2319–2324.

Grisham MB, Granger DN, Lefer DJ. 1998. Modulation of leukocyte-endothelial interactions by reactive metabolites of oxygen andnitrogen: relevance to ischemic heart disease. Free Radic Biol Med25:404–433.

Huie RE, Padmaja S. 1993. The reaction of NO with superoxide. FreeRadic Res Commun 18:195–199.

Jost A. 1953a. Degenerescence des extremities du foetus de ratprovoquee par l’adrenaline. C R Soc (Paris) 236:1510–1512.

Jost A. 1953b. La degenerescence des extremites du foetus de rat sousdes actions hormonales (acroblapsie experimentale) et la theorie desbulles myelencephaliques de Bonnevie. Arch Fr Pediatr 10:865–870.

Kamisaki Y, Wada K, Nakamoto K, Kishimoto Y, Kitano M, Itoh T.1996. Sensitive determination of nitrotyrosine in human plasma byisocratic high-performance liquid chromatography. J Chromatogr BBiomed Appl 685:343–347.

Kehrer JP, Park Y. 1990. Oxidative stress during hypoxia in isolated-perfused rat heart. Biol React Intermed 4:299–304.

Leist KH, Grauwiler J. 1974. Fetal pathology following uterine-vesselclamping on day 14 of gestation. Teratology 10:55–68.

Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. 1951. Proteinmeasurement with the Folin-phenol reagent. J Biol Chem 193:265–275.

Mackler B, Person RE, Nguyen T-D, Fantel AG. 1998. Studies of thecellular distribution of superoxide dismutases in adult and fetal rattissues. Free Radic Res 28:125–129.

Mastroiacovo P, Botto LD. 1994. Chorionic villus sampling andtransverse limb deficiencies: maternal age is not a confounder. Am JMed Genet 53:182–186.

McQuillan LP, Leung GK, Marsden PA, Kostyk SK, Kourembanas S.1994. Hypoxia inhibits expression of eNOS via transcriptional andposttranscriptional mechanisms. Am J Physiol 267:1921–1927.

Melillo G, Musso T, Sica A, Taylor LS, Cox GW, Varesio L. 1995. Ahypoxia-responsive element mediates a novel pathway of activationof the inducible nitric oxide synthase promoter. J Exp Med 182:683–1693.

Miyajima T, Kotake Y. 1997. Optimal time and dosage of phenylN-tert-butyl nitrone (PBN) for the inhibition of nitric oxide synthaseinduction in mice. Free Radic Biol Med 22:463–470.

Moncada S, Higgs EA. 1991. Endogenous nitric oxide: physiology,pathology and clinical relevance. Eur J Clin Invest 21:361–374.

Olney RS, Khoury MJ, Botto LD, Mastroiacovo P. 1994. Limb defectsand gestational age at chorionic villus sampling. Lancet 344:476.

Olney RS, Khoury MJ, Alo CJ, Costa P, Edmonds LD, Flood TJ, HarrisJA, Howe HL, Moore CA, Olsen CL, Panny SR, Shaw GM. 1995.Increased risk for transverse digital deficiency after chorionic villussampling: results of the United States Multistate-Control Study,1998–1992. Teratology 51:20–29.

Palmer RMJ, Ferrige AG, Moncada S. 1987. Nitric oxide releaseaccounts for the biological activity of endothelium-derived relaxingfactor. Nature 327:524–526.

Park Y, Invest L, Kanekai S, Kerhrer JP. 1991. Oxidative changes inhypoxic rat heart tissue. Am J Physiol 260:1395–1405.

Pierce RL, Pierce MR, Liu H, Kadowitz PJ, Miller MJS. 1995. Limbreduction defects after prenatal inhibition of nitric oxide synthase inrats. Pediatr Res 38:905–911.

Richter C. 1997. Reactive oxygen and nitrogen species regulatemitochondrial Ca21 homeostasis and respiration. Biosci Rep 17:53–66.

Roy RS, McCord JM. 1983. Superoxide and ischemia: conversion ofxanthine dehydrogenase to xanthine oxidase. In: Greenwald RA,Cohen G, editors. Oxy radicals and their scavenger systems. Volume2, cellular and medical aspects. New York: Elsevier Press. p 143–153.

Salas SP, Altermatt F, Campos M, Giacaman A, Rosso P. 1995. Effectsof long-term nitric oxide synthesis inhibition on plasma volumeexpansion and fetal growth in the pregnant rat. Hypertension26:1019–1023.

Sheinbaum KA, Badell A. 1992. Physiatric management of twoneonates with limb deficiencies and prenatal cocaine exposure. ArchPhys Med Rehabil 73:385–388.

Shigenaga MK, Lee HH, Blount BC, Shigeno SCET, Yip H, Ames BN.1997. Inflammation and Nox-induced nitration: assay for 3-nitrotyro-sine by HPLC with electrochemical detection. Proc Natl Acad SciUSA 94:3211–3216.

Skinner KA, Crow JP, Skinner HB, Chandler RT, Thompson JA, ParksDA. 1997. Free and protein-associated nitrotyrosine formationfollowing rat liver preservation and transplantation. Arch BiochemBiophys 342:282–288.

Tabrizi-Fard MA, Fung H-L. 1996. Reversed-phase high-performanceliquid chromatography method for the analysis of nitro-arginine inrat plasma and urine. J Chromatogr B 679:7–12.

Webster WS, Brown-Woodman PDC. 1990. Cocaine as a cause ofcongenital malformations of vascular origin: experimental evidencein the rat. Teratology 41:689–697.

Yabuki M, Inai Y, Yoshioka T, Hamazaki K, Yasuda T, Inoue M, UtsumiK. 1997. Oxygen-dependent fragmentation of cellular DNA by nitricoxide. Free Radic Res 26:245–255.

Zimmerman EF, Potturi RB, Resnick E, Fisher JE. 1994. Role ofoxygen free radicals in cocaine-induced vascular disruption in mice.Teratology 49:192–201.

Zweier JL, Wang P, Samouilov A, Kuppusamy P. 1995. Enzyme-independent formation of nitric oxide in biological tissues. Nat Med1:804–809.

160 A.G. FANTEL ET AL.