Mechanisms of Nitric Oxide-Mediated Neurotoxicity in Primary Brain Cultures

The Journal of Neuroscience. June 1993, 13(6): 2651-2661

Mechanisms of Nitric Oxide-mediated Neurotoxicity in Primary Brain Cultures

Valina L. Dawson,’ Ted M. Dawson ,2.3 Duane A. Bartley,2 George R. Uhl,1.2.3 and Solomon H. SnydeC4 ‘National Institute on Drug Abuse, Addiction Research Center, Laboratory of Molecular Neurobiology, Baltimore, Maryland 21224 and Departments of *Neuroscience, 3Neurology, 4Pharmacology and Molecular Sciences, and Psychiatry, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205

In addition to mediating several physiological functions, ni- of macrophagecytotoxicity (Hibbs et al., 1987; Marietta, 1989; tric oxide (NO) has been implicated in the cytotoxicities ob- Ignarro, 1990), nitric oxide (NO) appearsto be a neuronal mes- sewed following activation of macrophages or excess stim- sengerin the brain and PNS, satisfying many criteria of a neu- ulation of neurons by glutamate. We extend our previous rotransmitter (Garthwaite, 199 1; Snyder and Bredt, 1991; Daw- observations of glutamate-stimulated, NO-mediated neuro- son et al., 1992; Snyder, 1992). NO synthase(NOS) is localized toxicity in primary cultures of rat fetal cortical, striatal, and to the samediscrete neuronal populations in the brain that stain hippocampal neurons. Neurotoxicity elicited by either NMDA for NADPH-diaphorase (T. M. Dawson et al., 1991; Hope et or sodium nitroprusside (SNP) exhibits a similar concentra- al., 199l), a histochemical marker for neuronsthat resistcertain tion-effect relationship and time course. The concentration- forms of neurotoxicity (Beal ct al., 1986; Koh et al., 1986; Koh effect curve of NMDA-induced neurotoxicity is shifted to the and Choi, 1988). NOS catalytic activity accountsfor NADPH- right in the presence of nitro-L-arginine and farther to the diaphorasestaining, as transfection of human kidney 293 cells right in arginine-free media. The rank order of potency of with NOS cDNA provides NADPH-diaphorase and NOS stain- several NO synthase (NOS) inhibitors in preventing neuro- ing in the sameproportions as in neurons(T. M. Dawsonet al., toxicity is the same as the rank order of these compounds 1991). in inhibiting NOS, and this inhibition is stereospecific. NMDA Glutamate neurotoxicity elicited via NMDA subtypesofglu- neurotoxicity is also prevented by flavoprotein Inhibitors and tamate receptorsappears to mediate much of the neurotoxicity calmodulin inhibitors, fitting with the roles of flavoproteins in focal ischemia, as NMDA antagonistsblock such neurotox- and calmodulin as NOS regulators. 8-Bromo-cGMP and icity (Choi, 1988, 1990). Glutamate ncurotoxicity may also play guanylyl cyclase inhibitors do not affect neurotoxicity, while a role in neurodegenerative diseasessuch as Alzheimer’s and superoxide dismutase attenuates neurotoxicity. NOS neu- Huntington’s discascs(Choi, 1988; Meldrum and Garthwaite, rons appear to be the source of neurotoxic NO in culture, as 1990). At NMDA receptors, glutamate triggers the opening of lesions of these neurons with 20 CAMquisqualate diminish cation-permeablechannels. The entry of calcium through these subsequent NMDA neurotoxicity. Moreover, NMDA neuro- channels into cells stimulates NOS activity (Bredt and Snyder, toxicity develops over time in culture coincident with the 1989) by binding to calmodulin, which is a cofactor for NOS expression of NOS. lmmunohistochemical localization of NOS (Bredt and Snyder, 1990). in cultures and intact brain demonstrates widespread dis- Recently we demonstrated that NMDA neurotoxicity is at- tribution of the cell processes suggesting that NOS neurons tenuated in primary cerebral cortical cultures by the coappli- contact the majority of cortical neurons and so could mediate cation of NOS inhibitors or removal of the precursor of NO, widespread neurotoxicity. L-arginine (V. L. Dawson et al., 1991; T. M. Dawson et al., [Key words: NADPH-diaphorase, glutamate, NMDA, nitric 1993). These protective effectsare reversedby addition ofexcess oxide synthase, excitotoxicity, neurodegeneration] L-argininc. In addition, sodium nitroprusside (SNP), which gen- erates NO, mimics NMDA-induced neurotoxicity. Both SNP Besidesits roles as endothelial-derived relaxing factor (Furch- and NMDA neurotoxicities are also blocked by hemoglobin, gott, 1990; Ignarro, 1990; Moncada et al., 1991) and a mediator which binds NO. Together, these results implicate NO as a potential mediatorofNMDA neurotoxicity. In the presentstudy we examine the detailed mechanismsregulating NO mediation Received Oct. 5, 1992; revised Jan. 4, 1993; accepted Jan. I I, 1993. of glutamate neurotoxicity. We thank Dr. Charles E. Spivak for performing the electrophysiologic experi- men& This work was supported by USPHS Grants MH- 1850 1 and DA-00266, Materials and Methods Contract DA 271-90-7408, and Research Scientist Award DA-00074 to S.H.S., and NlGMS Pharmacology Research Associate Training Program and an Intra- Cell cultures mural Research Training Award (IRTA) from the NIH to V.L.D. T.M.D. is a Primary cell cultures were prepared from fetal rats (gestation day 14 for Pfizer postdoctoral fellow and is supported by grants from the American Academy of Neurology, the French Foundation for Alzheimer’s Research, the DANA Foun- cortex andcaudate-putamen cultures, gestation day I7 for hippocampal dation, and USPHS CIDA NS 01578-01. We gratefully acknowledge support of cultures). The various brain regions were dissected under a microscope, the W. M. Keck Foundation. incubated for 20 min in 0.027% trypsin/saline solution (5% phosphate- Correspondence should he addressed to Solomon H. Snyder, Departments of buffered saline, 40 mM sucrose, 30 mM glucose, IO mM HEPES, pH Neuroscienceand Pharmacology and Molecular Sciences,725 North Wolfe Street, 7.4). and transferred to modified Eagle’s medium (MEM), 10% horse Baltimore, MD 2 1205. serum,10% fetalbovine serum, 2 mM glutamine. Cells were dissociated Copyright 0 I993 Society for Neuroscience 0270-6474193113265 I-I I $05.00/O by trituration, counted, and plated in I5 mm multiwell (Nunc) plates 2652 Dawson et al. - Mechanisms of Nitric Oxide Neurotoxicity

coated with polyomithine at a density of u x 10’ cells per well. Four NOS immunohistochemistry. days after plating, the cells were treated with IO pg/ml of 5-Buoro-2’- deoxyuridine for 3 d to inhibit proliferation of non-neuronal cells. Cells Cells were washed three times with CSS and fixed for 30 min at 4°C in were maintained in MEM, 5% horse serum, 2 mM glutamine in 8% a 4% PF, 0.1 M PB. The cells were then washed with TBS. The cells CO,, humidified, 37°C atmosphere. The medium was changed twice a were then pcrmeabilized with 0.2% TX-100 in TBS for 5 min, followed week. Mature neurons (more than 21 d in culture) were used in all by blocking with 4% normal goat serum (NGS), 0.1% TX-100 in TBS experiments except for the experiments examining the ontogeny and for I hr. This was followed by-incubating the cells with affinity-purified development of NMDA neurotoxicity in relation to NMDA currents anti-NOS antibodies overnight at 4°C (Bredt et al.. 1990. 1991: T. M. and expression of NOS. In the mature cultures the percentage ofneurons Dawson et al., I99 I). The c&s were then rinsed three times in TBS for is approximately 70-90% of the total number of ceils as assessed by IO min each. After rinsing, the cells were incubated with biotin-con- neuron-specific enolase (NSE) and glial fibrillary acidic protein (GFAP) jugated secondary antibody (goat anti-rabbit; Vector Laboratories) for immunocytochemical staining of neurons and astrocytes, respectively 1 hr at room temperature in 1.5% NGS, TBS, 0.1% TX-100. After an (V. L. Dawson, T. M. Dawson, D. A. Bartley, G. R. Uhl, and S. H. additional three washes in TBS, the cells were incubated with an avidin- Snyder, unpublished observations). biotin-peroxidase complex (1:50; Vector Elite, Vector Laboratories) in TBS for 45 min at room temperature. The cells were again rinsed three times for IO min each in TBS and were developed with a substrate Cytotoxicity solution consisting of 0.01% H,O, and 0.5 mg/ml diaminobenzidine in Cells were exposed to test solutions as previously described (V. L. Daw- TBS. Cells were then rinsed in TBS containing 0.02% sodium azide. son et al., 1991). The cells were washed three times with a Tris-buffered All NOS-positive cells in each well were counted utilizing an inverted control salt solution (CSS) containing I20 rnM NaCI, 5.4 mM KCI, 1.8 microscope mM CaCI,, 25 mM Tris-HCI, and 15 mM glucose, pH 7.4 at room lmmunohistochemistry for NSE (Incstar) and GFAP (Incstar) was temperature. Except for exposures to kainate, all other drug solutions performed as described above with substitution of NSE antiserum or were applied to the cells for 5 min and then washed away with CSS GFAP antiserum for anti-NOS antibody. replaced with MEM containing 21 mM glucose and the cells were put NOS immunohistochemistry of rat brain was performed as described back in the incubator. Exposures to kainate were performed in MEM, (Bredt et al., 1990, 1991; T. M. Dawson et al., 1991). Briefly, male 21 mM glucose for 24 hr in the incubator. Twenty to twenty-four hours Sprague-Dawley rats (I 50-250 pm) were perfused with phosphate-buf- after exposure to drug solutions the cells were exposed to 0.4% trypan fered saline (PBS; 50 mM PB, 0.9% NaCI, pH 7.4, 4°C) followed by blue’in CSS to stain the residue of nonviable cells. Two to four pho- perfusion with 4% PF in 0. I M PB, pH 7.4,4”C. The brains were removed toprints at I 0-20x were made of each well. Viable versus nonviable and postfixed for 2 hr in 4% PF in 0.1 M PB, pH 7.4, 4°C. This was cells were counted, with approximately 500-l 500 cells counted per well. followed by ctyoprotection in 20% glycerol in 0.1 M PB, pH 7.4, 4°C At least two separate experiments utilizing four separate wells were (v/v) overnight. Thick (40 rm) sections were cut with a sliding micro- performed so that a minimum of 4000-l 2,000 neurons were counted tome (Microm) and sections were stained with the anti-NOS antibody for each data point. Ten percent of the photomicrographs were counted as described for the cortical cultures. by an additional observer blinded to the arrangement of photomicrographs, study design, and treatment protocol. An inter-rater reliability Measurement of NOS catalytic activity. of greater than 90% was consistently observed for the cell counting. In some of the experiments photomicrographs were made before and after Mature cortical neurons were treated with 500 FM NMDA for 5 mitt treatment using a transparent grid at the bottom of each culture plate. utilizing identical conditions as in the cytotoxicity assay. At 2,4, 8, and Viable and nonviable neurons in identical fields were counted by an 24 hr after NMDA administration, the cells were scraped from the observer blinded to study design and treatment protocol. Under the culture wells and homogenized in 50 mM Tris-HCI buffer containing I conditions used for assessment of neurotoxicity, no appreciable loss of mM EDTA and I mM EGTA, pH 7.4. Cortical cultures were also treated neurons after the various treatment conditions was identified when with 3.5 rig/ml lipopolysaccharide (LPS) and 5 U/ml r-interferon for comparing “before” and “after” photomicrographs. Furthermore, there 24 hr and then homogenized in the above buffer. NOS catalytic activity was no significant difference between the results obtained with either was assessed by measuring the conversion of ‘H-arginine to ‘H-citrulline method of assessing neurotoxicity. as previously described (Bredt and Snyder, 1989, 1990) in the presence and abscncc of calcium as well as NADPH. After incubation for 15 min at room temperature, the assays were quenched with two 2 ml aliquots Electrophysiology of 20 mM HEPES, pH 5.5, 2 mM EDTA applied to I ml columns of Cortical neurons in sister cultures were voltage clamped by whole-cell Dowex AGSOWX-8 (Na . form). ‘Hcitrulline was quantified by liquid patch clamp as previously described (Hamill et al., I98 I). Cells were scintillation counting ofthe 4 ml flow through. All values were compared bathed in an extracellular solution containing 137 mM NaCl, 2.5 mM to control vehicle-treated cells. KCI, 2 mM CaCl,, IO mM HEPES, and IO mM glucose, pH 7.3. Micro- pipettes were pulled from 1.5 mm fiber-fill capillary tubing and filled with an intracellular solution containina 140 mM KCI. 2 mM MaCI,. I1 Materials mM EGTA, I mM CaCI,, and IO mM HEPES, pH 7.2: Micropipette tip NMDA and MK801 were purchased from Research Biochemicals In- internal diameters were approximately I rrn with resistances of ap- corporated. A23 I87 was obtained from Calbiochem. Diphenylenio- proximately 7 MR. The target cell was microsuperfused (Demirgoren et donium (DPI) was obtained from Kodak. DPI and A23187 were dis- al., 1991) with extracellular solution containing 0.3 PM tetrodotoxin solved in dimethyl sulfoxide (DMSO) and diluted to a final concentration with or without 200 PM NMDA, IO PM glycine. Recordings were made of 0.1% DMSO. All other compounds were dissolved in CSS. LPS was in the whole-cell voltage-clamp mode (Hamill et al., I98 1) using a List obtained from GIBCO, and recombinant murine y-interferon, from EPC-7 amplifier (Adams and List Associates, Ltd., Great Neck, NY). Upstate Biotechnology, Inc. ‘H-arginine was obtained from New En- Signals were recorded simultaneously on an FM tape recorder and a gland Nuclear/DuPont. L-Iminoethylomithine (NIO) was a generous Gould strip chart recorder. gift from Dr. Salvador Moncada, The Wellcome Foundation, Ltd.; SIN-I (3-morpholino-sydnonimine-hydrochloride) and C88-3934 were gifts Diaphorasestaining of Cassella AG. Unless otherwise noted, all other chemicals were purchased from Sigma (St. Louis, MO). Reduced hemoglobin was prepared Cells were washed three times with CSS and fixed for 30 min at 4°C in as described by Martin et al. (I 985). Solutions of SNP, SIN- I, and C88- a 4% paraformaldehyde (PF), 0.1 M phosphate buffer (PB). The PF 3934 were prepared in the dark immediately before application to the solution was washed away with Tris-buffered saline (TBS) 50 mM Tris- cultures to minimize photolysis. Cell culture media and arginine-free HCI, I .5% NaCl, pH 7.4. The reaction solution containing I mM NADPH, media were obtained from GIBCO. Arginine can be depleted from 0.2 mM nitroblue tetrazolium, 0.2% Triton X-100 (TX-loo), 1.2 mM cultures by growing them in arginine-free media for 24 hr in the presence sodium azide, 0.1 M Tris-HCI, pH 7.2, was applied to the fixed cell of glutamine, which inhibits argininosuccinate synthetase (Sessa et al., cultures for 1 hr at 37°C. The reaction was terminated by washing away 1990). Cortical neurons remain viable in arginine-free media for up to the reaction solution with TBS. Ah diaphorase-positive cells in each 72 hr (Dawson, Dawson, Bartley, Uhl, and Snyder, unpublished obser- well were counted utilizing an inverted microscope. vations). Tha Jcurnalof Nauroaciaace, June 1993. U(6) 3663

Results Effects of NOS inhibitors on NMDA neurotoxicity. To evaluate the relationship of NO generation to NMDA neurotoxicity, we examined concentration-response relationships and the time course ofcell death for both SNP and NMDA in cortical cultures (Fig. 1). The shapes of the concentration-response CUNCS for 80 NMDA and SNP are essentially the same. Neurotoxicity in- creases from about 10% to more than 80% cell death with both SNP and NMDA over about 2 orders of magnitude, suggesting 60 the absence of cooperativity. The time course of toxicity for NMDA and SNP is comparable with minimal effects over the 40 first 4 hr proceeding to almost maximal neurotoxicity after 12 hr. This lag phase, following a brief 5 min application of NMDA, described previously by others (Choi, 1987; Choi et al., 1988; 20 Koh et al., 1990), is relatively unique to this form of neurotoxicity; for example, it is not evident with kainate toxicity. Thus, the close similarity of SNP and NMDA neurotoxic effects both 0 in concentration-response relationships and time course is con- 10 sistent with NO mediating NMDA toxicity. However, since NMDA (pM) SNP breaks down to NO and Fe(CN,)‘- it is possible that SNP kills neurons by a process unrelated to NO, as SNP has non- spe$fic actions that include blockade of NMDA receptors (East B et ai., 199 1; Kiedrowski et al., 1991; Manzoni et al., 1992a). It 70 has been proposed that the nonspecific action is due to the Fe(CN,)‘- moiety of SNP, as K.,[Fe(CN),] had similar effects 60 (East et al., 1991; Kiedrowski et al., 199 1, 1992; Manzoni et al., 1992a). To determine whether SNP toxicity could involve so Fe(CN,)2-, we applied KJFe(CN),] at concentrations up to 3 mM under identical exposure conditions and found no signifi- 40 cant neuronal injury. Interestingly, concentrations as low as 100 PM K,[Fe(CN),] are completely neuroprotective against NMDA 30 toxicity (data not shown). Further evidence that NO is neurotoxic is that SIN- 1, another agent that releases NO in aqueous 20 solutions, also elicits neuronal injury, while C88-3934, a struc- turally related compound that does not release NO, does not 10 elicit neuronal toxicity (data not shown). Other NO releasers elicit neurotoxicity in cortical cultures (Lustig et al., 1992b), 0 fitting with our observations. 12 4 12 24 Confirming our previous observations (V. L. Dawson et al., 1991), the NOS inhibitor NC-nitro+arginine (N-Arg) blocks Time (hours) NMDA cell death (Fig. 2). The concentration-response rela- Figure I. Dosedependence and time courseof NMDA andSNP neu- tionship for NMDA is shifted to the right in the presence of 100 rotoxicity. In primarycortical cultures SNP elicitscell death in a dose- PM N-Arg. This protective effect is only overcome by an NMDA dependentfashion that parallelsNMDAdose dependence(A). Fe(CNJ-, concentration of 7 mM. Arginine-free medium also attenuates producedwhen SNP releasesNO, doesnot elicit toxicity whenapplied to cellsin concentrationsequivalent to SNP (datanot shbwn).The-time NMDA neurotoxicity, producing a rightward shift of the NMDA courseofcell deathfollowing a 5 min anDlicationofSNP mimicsNMDA concentration-response curve. Arginine-free medium has a cell death(B). Eachdata point representsthe meanf SEM of at least greater effect on NMDA neurotoxicity than coapplication of 100 two separateexperiments in whichat leastfour wellswere treated with PM N-Arg, a maximally effective concentration in culture. At 7 the variousagents per experiment; 500-I 500neurons were counted per mM NMDA, neither N-Arg nor arginine-free medium prevented well. toxicity. Additionally, the NMDA antagonist MK80 1, which blocks the neurotoxicity of 1 mM NMDA, fails to block the Various arginine derivatives differ in their potenciesas NOS neurotoxicity elicited by 7 mM NMDA. Thus, the toxicity ob- inhibitors as well as their selectivity for brain as compared to servedat concentrationsof NMDA at or exceeding7 mM NMDA macrophageNOS (Moncada et al., 199 1). Accordingly, we ex- appearsnot to involve NMDA receptors or NO. amined concentration-responserelationships for a seriesof NOS Previously we reported that a single high concentration of inhibitors (Fig. 3). N-Arg is most potent with 50% protective L-arginine can reverse the protective effects of N-Arg (V. L. effectsat 20 PM. N-methyl-arginine (NMA), a weaker inhibitory Dawson et al., 1991). In the present study we have explored a ofbrain NOS than N-Arg, is about one-eighthas potent as N-Arg detailed concentration-responserelationship for L-arginine (Fig. in protecting againsttoxicity. NIO, which has a greater afinity 2, inset). Half-maximal reversal of the protective effects of ar- for macrophage NOS and has intermediate potency between ginine-free medium isevident at 25 PM L-arginine, with maximal N-Arg and NMA in inhibiting brain NOS (McCall et al., 199I), effects near 100 PM. also displays intermediate potency in blocking NMDA toxicity. 2654 Dawscm et al. l Mechanism of Nitric Oxide Naurotoxicity

Figure 2. NMDA neurotoxicity is inhibited by N-Arg or arginine-free media (AF-MEM). The dose-response of cortical cultures to increasing concentrations of NMDA is shifted to the right by N-Arg and further to the right in arginine-free media with a more than 20-fold increase in the 50% lethal concentration of NMDA. Adding graded concentrations of L-arginine to 300 j.rcM NMDA in arginine-free media reveals a requirement for L-arginine with an EC,, L 25 PM (inset). All data points represent the mean + SEM of at least two separate experiments in which at 10-4 10-S 10-2 least four wells were treated with var- 10-S ious agents per experiment; 500-I 500 neurons were counted per well. The concentration of NMDA is molar. Concentration of NMDA

The methyl ester of N-Arg (NAME) also has an intermediate of cortical neurons, an effect not observed with NOS inhibitors potency in inhibiting NOS activity and NMDA neurotoxicity, that are arginine derivatives. with an EC,, of 3 10 PM. The effectsof NMA are stereospecific, We also compared effectsof NOS inhibitors on neurotoxicity as 500 PM N-methyl-D-arginine provides no protection against in cultures from different brain regions. In these studies we NMDA neurotoxicity. evaluated ncurotoxicity elicited by glutamate, quisqualate,and In addition to arginine analogs,NOS can also be inhibited by kainate in addition to NMDA (Table 1). Glutamate and its three DPI, which binds the flavin moiety critical to NOS activity analogs produce substantial neurotoxicity in all three brain (Stuehr et al., 1991). DPI inhibits NMDA toxicity with an EC,, regions. In all the brain regions N-Arg producesthe most pro- of 30 nM (Fig. 4). DPI also blocks NADPH-diaphorase staining tcction againstglutamate and NMDA, with somewhatless pro-

loo- c ii 8 80- zi f 80.

2 40- i --e NAME d p 20- +3- NIO

8 lo- o- 1 1 10 100 1000 1 ” .-.--3 CONC (pU) NOS Inhibitors 0 1onh4 5onM 1oonM 5oonM 1 CM Figure 3. Concentration-response of NOS inhibitors in attenuating Cone DPI 300 PM NMDA toxicity. All data points represent the mean k SEM of Figure 4. DPI attenuates NMDA neurotoxicity and NADPH-diapho- at least two separate experiments in which at least four wells were treated rase staining. All data points represent the mean + SEM (n 2 8) in with various agents per experiment and 500-l 500 neurons were counted which approximately 4000-I 2,000 neurons were counted for each data per well. point. The Journal of Neuroscknoe. June 1993. 13(6) 2655

Table 1. Inhibition of neurotoxicity by N-Arg and reversal by L-arginine

Cortex Caudate-putamen Hippocampus Treatment (% killed f SEM) (O/o killed + SEM) (% killed f SEM) 500 phi glutamate 49.2 -c 3.7 74.5 k 5.1 58.0 + 3.1 + 100 PM N-Arg 14.5 2 3.1+ 25.9 + 2.9* 2.7 f 1.8* +N-Arg + 1 mM L-Arg 50.1 f 1.8 74.4 f 5.3 52.8 f 1.9 300 FM NMDA 62.0 2 2.7 63.5 k 3.0 62.8 + 3.2 + 100 HIM N-Arg 20.3 f 3.3’ 15.5 f 2.4+ 18.7 k 1.3* +N-Arg + 1 mM L-Arg 64.6 + 5.0 60.8 + 2.6 56.7 f 2.7 500 PM quisqualate 60.7 k 4.1 64.9 + 3.5 60.3 f 2.0 + 500 &tM N-Arg 41.5 + 4.1: 25.6 + 3.4* 49.2 f 4.5 +N-Arg + 5 mM L-Arg 63.0 ? 4.5 67.1 + 4.0 56.2 f 2.2 100 PM kainate 83.9 k 4.0 85.6 + 2.5 83.2 + 3.9 +500 /AM N-Arg 83.0 2 4.0 67.6 f 5.5* 73.9 rt 4.9

Data are the means + SEM (n = 6-26). Each data point represents a minimum of 4000-12,000 neurons counted. In some experiments up to 40,ooO neurons were counted. Toxicity was assessed by trypan blue exclusion as described in Materials and Methods. Significant overall values were obtained using a one-way between-groups ANOVA. Specific comoarisons on all possible oainvise combinations were made with the Student’s I test for independent means; l , p < 0.05: tection against quisqualateand negligible protection in the cc- N-Arg. Reduced hemoglobin, which binds extracellular NO, rebral cortex and hippocampus againstkainate. In the caudate markedly prevents A23187 toxicity, similar to its protection puthmen, N-Arg produces modest but statistically significant against NMDA toxicity (V. L. Dawson et al., 1991). A23187 protection against kainate. In all brain regions excessarginine toxicity is also prevented in arginine-free medium. reversesthe effects of N-Arg. Calcium activation of NOS involves calmodulin. To evaluate The role of calcium, cGMP, and superoxideanion in NMDA the role of calmodulin in NMDA toxicity, we examined the neurotoxicity. Large increasesin intracellular calcium following effects of calmidazolium, a calmodulin antagonist (Table 2). NMDA receptor activation have been implicated as a primary Calmidazolium blocks NMDA neurotoxicity to a similar extent mediator of NMDA neurotoxicity (Choi, 1987, 1988; Choi et as N-Arg. W7, which binds calmodulin, also blocks NMDA al., 1988; Meldrum and Garthwaite, 1990). To ascertain wheth- toxicity. er cell death mediated by increased intracellular calcium in- The concentration-responsecurve for SNP toxicity in cortical volves NO, we examined effects of agents influencing NO dis- cultures parallels its effects in stimulating cGMP concentrations position following calcium influx (Fig. 5). The calcium ionophore in thesecultures (V. L. Dawson et al., 1991). Additionally, the A23187 (30 PM, applied for 5 min) kills about 70% of cortical same conditions that modulate NMDA toxicity alter cGMP cells, as assessed by trypan blue exclusion 20-24 hr later, similar formation in a similar manner (Dawson et al., 1991a). To as- to the effects of NMDA (Choi, 1987). This ncurotoxicity is certain whether cGMP mediates NMDA- and NO-mediated blocked by N-Arg to a similar extent as the inhibition of NMDA neurotoxicity, we examined the effects of methylene blue, an toxicity. Moreover, L-arginine rcvcrscs the protective effects of inhibitor of guanylyl cyclase (Table 3). Methylene blue fails to block both NMDA- and SNP-induced toxicity. Moreover, 8-bromo-cGMP, a derivative ofcGMP that penetratesinto cells, has no effect upon neurotoxicity elicited by NMDA or SNP. Exposure to cultures with 8-bromo-cGMP alone doesnot elicit neurotoxicity. These observations confirm recent findings of Greenberg and collaborators (Lustig et al., 1992a),who explored a variety of guanylyl cyclase inhibitors and cGMP derivatives. NO can combine with the superoxideanion to form peroxynitrite, which decomposesinto the hydroxyl free radical (OH’) and the nitrogen dioxide free radical (NO;), both of which are highly reactive and potentially toxic (Beckman et al., 1990; Radi et al., 1991 a,b). To evaluate the role of the superoxide anion in NMDA and NO neurotoxicity, we examined the effects of su-

A23ln + N-kg + N-kg *Mb AFUEM + L-kg Table 2. Calmodulin inhibitors attenuate NMDA neurotoxicity Figure 5. Neurotoxicity elicited by the calcium ionophore A23187. A23187 is affected by the same agents that influence NMDA neuro- Treatment % cell death + SEM toxicity. The NOS inhibitor N-Arg (100 FM) inhibits A23 I87 (30 jiM) 300 /.LM NMDA 63.8 +_ 4.2 neurotoxicity, which can be overcome with excess (I mM) substrate, + IO I.~Mcalmidazolium 20.9 + 5.0* L-arginine (L-A&. Toxicity is also attenuated in cultures exposed lo 30 PM A23 187 in arginine-free media (AF-MEM) or in the presence of 500 +200 &LM WJ 23.3 f 6.6; PM reduced hemoglobin (Hgb). Data are means 2 SEM (n 2 8); ap- Data are means + SEM (n 2 8). Each data point represents approximately 4000- proximately 4000-12,000 neurons were counted for each data point. 12,OCQ counted neurons. Toxicity was assessed by trypan blue exclusion as de- Toxicity was assessed by trypan blue exclusion. Significance was deter- scribed in Materials and Methods. Significance was determined by the Student’s mined by Student’s t test for independent means; *, p < 0.01. t test for independent means; *, p < 0.0 I. 2651) Dawaon et al. * Mechanisms of Nitric Oxide Neurotoxicity

Table 3. Superoxide dismotase but not guanylyl cyclase modulates NMDA and SNP neurotoxicity

%,cell death Treatment (+ SEM) 300 FM NMDA 60.6 + 5.4 + 100PM methyleneblue 53.9 f 2.2 + I mM 8-bromo-&MP 69.3 r 6.4 +lOOUSOD 32.1 + 4.0* 300 FM SNP 51.3 f 3.7 + 100PM methyleneblue 48.8 + 4.9 + I rnM I-bromo-cGMP 48.4 r 4.4 + 100 U SOD 18.0 + 3.7* NMDA QUIS PT ~231 a7 QUIS PT Data are the means + SEM (n 2 8). Each data point represents 4000-12,000 counted neurons. Toxicity was assessed by trypan blue exclusion as described in Figure 7. Selectivelykilling NOS neuronsprior to NMDA or A23 187 Materials and Methods. Significance was determined by Student’s I test for in- exposure reduces toxicity. Treating primary cortical neurons with 20 dependent means; l , p < 0.05. PMquisqualate, 24 hr prior lo treatingthese neurons with 300PM NMDA or 30 PM A23 187(QCJJS Py), reducesNMDA neurotoxicity 7 1% and A23 187 neurotoxicity 94%. Neurons in sister cultures treated with 20 peroxide dismutase (SOD), an enzyme that scavengesthe su- PM quisqualate still produce currents on application of 200 jt~ NMDA peroxide anion (Table 3). Addition of SOD to the exposure (truce ins@. Data are means+ SEM (n 2 8). Significancewas deter- solution reduces the toxicity elicited by NMDA, SNP, and mined by Student’s f test for independent means; *, p < 0.01. A23187. Evidencethat NOS neuronsare the sourceofN0 that mediates neurotoxicity. NO can be formed by NOS in macrophagcsand positive neuronsversus the total neuronal population following endothelial cellsas well asneurons expressing NOS. In the brain, exposure to graded concentrations of NMDA or quisqualate microglia function as macrophages.To determine whether NOS (Koh et al., 1986; Koh and Choi, 1988). NOS neuronsarc also neuronsin the cultures are the sourceof the neurotoxic NO, we resistantto the toxic effectsof exogenouslyapplied NO (Fig. 6). took advantage of the differential sensitivity of NOS neurons We examined the neurotoxicity of 300 FM NMDA in cultures and other neuronsto various toxins (Fig. 6). While about 60% treated 24 hr earlier with 20 PM quisqualate, a concentration of the total neuronal population of cortical cultures are killed that destroys 85-95% of NOS neurons(Fig. 7). In thesecultures by 300 PM NMDA or 300 PM SNP, only about 25% and lo%, NMDA toxicity is profoundly attenuated. In sistercultures pre- respectively, of NOS neurons die with the two treatments. In treated in the same manner with 20 PM quisqualate, 200 PM contrast, 20 PM quisqualate kills less than 20% of the total NMDA still elicits calcium currents, indicating the persistence neuronal population, but kills about 85-95% of NOS neurons. of functional NMDA receptors. Toxicity elicited by the calcium These resultsresemble those of Choi and co-workers, who con- ionophorc A23 187 is also attenuated by quisqualate pretreat- trasted the degreeof cell death elicited in NADPH-diaphorase- ment (Fig. 7). In addition, a critical number of NOS neurons (approximately 100 NOS-positive cells per 15 mm well) were required to observean NO component to NMDA neurotoxicity (data not shown). These experiments strongly implicate NOS neurons as the sourceof NO that mediates NMDA toxicity. To examine the role of NOS neurons in NMDA toxicity by another paradigm, we evaluated NMDA toxicity at different days in culture while monitoring the percentageof cells that stain positive for NADPH-diaphorase (Fig. 8). NMDA toxicity is not evident for the first 16 d in culture, and then develops gradually, reachingfull effectsat 20 d in culture. NADPH-diaph- erase staining of the cells displays a virtually identical time course. Despite the absenceof NMDA toxicity in early stages of the culture, NMDA-mediated calcium currents are evident at day 7 and day 14, though they are somewhatsmaller than at - day 2 I. Although young cultures are resistantto NMDA toxicity, tom tow ms they are sensitive to 300 PM SNP, which produces the same rN- - nauroM Murons n”roM IlWlClU extent of cell killing at 7 and 14 d as at 2 1 d (data not shown). 2o~oul6 304 pM NMQA 3IOflSNP Inducible macrophageNOS activity is calcium independent, while neuronal NOS requires added Ca2+(Marlctta, 1989). We Figure 6. Selectivesensitivity of NOS neuronslo quisqualate.NOS neuronsin primary corticalcultures resist NMDA and SNP neurotox- asscsscda potential role for microglial, inducible NOS by ex- icity but are moresensitive than other neuronsto quisqualate.Low- amining whether there was any calcium-independent NOS cat- dose,20 FM quisqualate(QZJ1.Y) produces substantial loss of NOSneu- alytic activity 2, 4, 8, and 24 hr after a 5 min application of ronswithout much decreasein the total cell population.The total cell NMDA. Despite the presenceof abundant calcium-dependent countsinclude NOS neurons. NMDA (300 PM) and SNP (300 PM) elicit morethan 60%cell death in the total neuronal population but only 25% NOS catalytic activity, no appreciablecalcium-independent NOS and 13% loss of NOS neurons,respectively. Data are means+ SEM (n catalytic activity was detected (data not shown). In addition, 2 8). application of LPS and y-interferon, potent inducers of mac- The Journal of Neuroscknce, June 1993, 13(6) 2667

NWDA WDA NYDA apA I 9 151 D@Y7

120-

100.

60.

n NYDA Figure 8. The developmentof NMDA % _ g NADPWD = 60 neurotoxicity correspondswith the ap- pearanceof NOS neurons.After dis- 8 sociatedcortical cells were plated (day ap 0), cellswere periodicallyassessed for 40. NMDA neurotoxicity and NADPH- diaphorase stain. Neurotoxicity and NADPH-diaphorasestaining were first 20. observedon day I7 in culture,with dc- velopment of full expression by day 20. NMDA-induced currents were ob- -I- servedas early as day 7 in culture Ncu- “I rotoxicity to 300 PM SNP was observed 4 7 10 14 16 17 16 at day 7 and did not vary over time. Day in Culture Data arc means* SEM (n 2 9).

rophageNOS, fails to induce a calcium-independentNOS. Thus, ncuronal injury in whole rat brain cultures (Demerle-Pallardy neuronal NOS seemsto be the primary source of ncurotoxic et al., 1991). Whether the use of whole brain cultures or other NO in primary cortical cultures. factors accounted for the failure of NOS inhibitors to block BecauseNOS occurs in only l-2% of cerebral cortical cells, neurotoxicity is unclear. we might ask how NO is able to kill the majority of cells in Besidesthe cerebral cortex, we observed blockade of toxicity cultures. To examine this question, we conducted both dark- by NOS inhibitors in caudate-putamen and hippocampal cul- and bright-field microscopy of sections of intact adult rat ce- tures. N-Arg blocks glutamate and NMDA toxicity to a greater rebral cortex (Fig. 9) and cortical cultures (Fig. 10 and data not extent than quisqualatetoxicity. At the concentrations of quis- shown). While the number of NOS cells in a given field may be qualatc and the mode of application that WChave employed, it few, processesof the cells ramify extensively to cover virtually is likely that quisqualate may act in part via NMDA receptors all the surfaceof the culture (Fig. 10). Since the extensive ram- (Koh ct al., 1990) which may account for the portion of quis- ification of NOS processesin culture is difficult to illustrate. we qualate toxicity prevented by N-Arg. N-Arg does not block show the extensive amount of NOS processesin a tissuesection kainate toxicity in the cerebral and hippocampalcultures, which of intact adult rat cerebral cortex (Fig. 9). Dead cells following fits with observations of Coyle and associates(Puttfarcken et NMDA treatment occur in patches. These patchesarc largely al., 1992) that kainate toxicity involves free radicals and not concentrated about processesand cell bodies of NOS neurons NO. N-Arg does prevent a portion of kainate neurotoxicity in (Fig. 10). the striatal cultures. Perhapskainate-sensitive calcium channels (Murphy et al., 1987) causeexcessive release of NO from NOS Discussion neuronsand account in part for the neuronal injury associated In the present study we confirm and extend our previous ob- with kainate in the caudate-putamen. Interestingly, NOS inhib- servations that NO is responsible,at least in part, for NMDA- itors do not prevent glutamate neurotoxicity in cerebellar gran- type glutamate neurotoxicity. The relative potencies of several ule cells (Puttfarcken et al., 1992). All cerebellar granule cells NOS inhibitors in blocking neurotoxicity parallel their potencies are enriched in NOS (Bredt et al., 1991). Their resistance to as NOS inhibitors. The ability of NOS inhibitors to protect NMDA toxicity fits with our observation that NOS neuronsare againstNMDA neurotoxicity in brain culturesas well as in brain uniquely resistant to NMDA and SNP toxicity. On the other slices has now been confirmed by several groups (Izumi et al., hand, R. J. Miller and colleagues (personal communication) 1992; Lustig et al., 1992b; Moncada et al., 1992; Wallis et al., have shown that in cultures ofcerebellar enriched Purkinjc cells, 1992). NO hasalso been implicated in neuronal injury in intact kainate-induced killing of cerebellar Purkinje cells, which lack animals, as N-Arg (1 mg/kg, i.p., rcpcatcd doses)blocks neu- NOS, is attenuated by NOS inhibitors. rotoxicity elicited by ligation of the middle cerebral artery more Application of SNP, which spontaneouslyreleases NO, elicits cffcctivcly than the noncompetitive NMDA antagonist MK80 I cell death in a concentration- and time-dependent fashion that (Nowicki et al., 1991). Additionally, N-Arg is completely neu- parallels NMDA toxicity, observations also made by Chen et roprotective against focal stroke in the 7-d-old rat (Trifilctti, al. (I 99 1) in striatal cultures. SNP elicits neuronal injury in 1992). In one report the NOS inhibitor N-Arg did not block intact animals when injected directly into the hippocampus 2658 Dawson et al. l Mechanisms of Nitric Oxide Neurotoxicity

Figure 9. NOS neurons and processes in cerebral cortex. A, Bright-field photomicrograph illustrating the distribution and density of NOS neurons (arrowheads) in the cerebral cortex from intact adult rat brain. B, Dark-field photomicrograph of A, illustrating the dense plexus of NOS fibers throughout the cortex. Scale bar, 100 pm. The Journal of Neuroscience, June 1993. f3(6) 2659

CORTEX

Figure 10. NMDA kills neurons within the vicinity of NOS neurons and/or processes. A, Low-power Hoffman modulation photomicrography of NOS neurons (arrows) in cortical cultures. B, NMDA lolls neurons in the vicinity of NOS neurons. Dead neurons appear black in these photomicrographs. C, High-power view of cortical cultures stained with anti-NOS antibody. D, High-power view of NMDA neurotoxicity. Scale bar, 100 pm.

(Loiacono and Beart, 1992). KJ(Fe(CN),], which elicits the same the initiation of “delayed neurotoxicity” associatedwith calci- nonspecific effects as the Fe(CN,)*- moiety of SNP, does not um overload (Choi, 1988, 1990; Randall and Thayer, 1992). cause neuronal injury. Several other NO releasersalso cause Our studiesindicate that activation of NOS leadsto theselethal neuronal injury both in primary cultures (Chen et al., 1991; processes,as neurotoxicity directly elicited by a calcium ion- Lustig et al., 1992b) and in intact animals (Smith et al., 1991). ophore A23187 is mediated via NO, being blocked by N-Arg, In addition, 40 nM authentic NO can cause cell death in hip- hemoglobin, and arginine-free medium. Moreover, NMDA tox- pocampal cultures (O’Dell et al., 1991). Thus, whether NO is icity is blocked by calmodulin antagonists,indicating that the from exogenoussources, such as SNP or other NO releasers,or toxicity elicited by calcium involves calmodulin and, presum- is endogenouslygenerated, it may function as a neurotoxin un- ably, the calmodulin-induced activation of NOS. In contrast to der conditions of excessiveNO production or release. the crucial role for calcium and calmodulin, cGMP does not Calcium has been implicated as a major mediator of gluta- appear to mediate the toxicity directly. In our experiments as mate neurotoxicity (Choi, 1988; Meldrum and Garthwaite, well as those of Greenberg and colleagues(Lustig et al., 1992a), 1990). However, someevidence indicates that intracellular cal- the guanylyl cyclase inhibitor methylene blue does not block cium levels do not directly correlate with such toxicity (Michaels neurotoxicity, and 8-bromo-cGMP doesnot influence toxicity. and Rothman, 1990; Dubinsky and Rothman, 1991). For in- Garthwaite and Garthwaite (1988) obtained evidence that cGMP stance, brief (5 min) glutamate exposure elicits a transient ele- may protect againstneurotoxicity in cerebellar slices. vation in intracellular calcium that recovers to the basal level Evidence for participation of superoxide anions in neurotoxin the majority of neuronsin culture (Randall and Thayer, 1992). icity elicited by NMDA, the calcium ionophore, as well as the Despite normal intracellular calcium levels after the initial glu- direct generation of NO from SNP is evident from the protective tamate exposure,irreversible processesare startedthat no longer effects of SOD. The importance of superoxideanions in NMDA require a sustainedelevation of intracellular calcium (Randall neurotoxicity is emphasized by findings that cortical cultures and Thayer, 1992). These early irreversible processeslead to from transgenicmice overexpressingSOD are resistantto NMDA 2660 Dawson et al. - Mechanisms of Nitric Oxide Neurotoxieity neurotoxicity (Chan et al., 1990) and these same animals are Beckman JS (1991) The doubled-edge role of nitric oxide in brain relatively resistant to focal &hernia (Kinouchi et al., 1991). function and superoxide-mediated injury. J Dev Physiol 15:53-59. Beckman JS, Beckman TW, Chen J, Marshall PA, Freeman BA (1990) SOD protection might relate to superoxides directly generated Apparent hydroxyl radical production by peroxynitrite: implications by NOS, as purified NOS directly forms superoxide anions (Pou for cndothclial injury from nitric oxide and superoxide. Proc Nat1 et al., 1992). The combination of the superoxide anion and NO Acad Sci USA 87: 1620-l 624. gives rise to peroxynitrite, which in turn decomposes into OH* Bredt DS, Snyder SH (I 989) Nitric oxide mediates glutamate-linked enhancement of cGMP levels in the cerebellum. Proc Nat1 Acad Sci and NO; free radicals that are highly reactive and might be the 1JSA 86:9030-9033. final common mediators of NO toxicity (Beckman et al., 1990; Bredt DS, Snyder SH (1990) Isolation of nitric oxide synthetase, a Beckman, 199 1; Radi et al., 199 la,b; Dawson et al., 1992, 1993). calmodulin-requiring enzyme. Proc Nat1 Acad Sci USA 87:682-685. Extracellular SOD would not be expected to penetrate cells, Bredt DS, Hwang PM, Snyder SH (1990) Localization of nitric oxide suggesting that the superoxide anion may be released upon glu- synthase indicating a neural role for nitric oxide. Nature 347:768- 770. tamate exposure. However, Bennett and colleagues (Saez et al., Bredt DS, Glatt CE, Hwang PM, Fotuhi M, Dawson TM, Snyder SH 1987) have shown that exogenously administered ‘251-labclcd (1991) Nitric oxide synthase protein and mRNA arc discretely lo- SOD can attain high intracellular levels under depolarizing con- calized in neuronal populations of the mammalian CNS together with ditions. Both NO (Lei et al., 1992; Manzoni et al., 1992b) and NADPH diaphorase. Neuron 7:6 15-624. the superoxide anion (Aizcnman et al., 1989, 1990) block both Chan PH, Chu L, Chen SF, Carlson EJ, Epstein CJ (1990) Reduced neurotoxicity in transgenic mice overexpressing human copper-zinc- NMDA receptor currents and the associated increase in intra- superoxide dismutase. Stroke 2 1:80-82. cellular calcium. Thus, both potential toxic agents might exert Chen J, Chang B, Williams M, Murad F (199 1) Sodium nitroprusside feedback inhibition on NMDA receptors under physiologic con- degenerates cultured rat striatal neurons. Neuroreport 2: 12 l-1 23. ditions. With excessive release of both NO and the superoxide Choi DW (1987) Ionic dependence ofglutamate neurotoxicity. J Neu- rosci 7~369-379. anion, normal regulatory mechanisms might be overwhelmed, Choi DW (I 988) Glutamate ncurotoxicity and diseases of the nervous leading to cell death. As suggested by Lei et al. (1992), NO is system. Neuron 1:623-634. necessary but not sufficient for neuronal injury and is toxic only Choi DW (1990) Cerebral hypoxia: some new approaches and un- in the presence of other factors, such as the superoxide anion. answered questioins. J Neurosci 10:2493-250 1. Why NOS neurons are selectively resistant to various forms Choi DW, Koh J, Peters S (1988) Pharmacology of glutamate neurotoxicity in cortical cell culture: attenuation by NMDA antagonists. of neurotoxicity is not known. SOD could, in principle, protect J Neurosci 8: 185-196. these cells against such toxicity. Conceivably NOS neurons are Dawson TM, Bredt DS, Fotuhi M, Hwang PM, Snyder SH (1991) enriched in SOD, which would account in part for their resis- Nitric oxide synthase and neuronal NADPH diaphorase are identical tancc to toxicity. In the cerebral cortex and in the caudate- in brain and peripheral tissues. Proc Nat1 Acad Sci USA 88:7797- 7801. putamen all somatostatin neurons are positive for NOS (T. M. Dawson TM, Dawson VL, Snyder SH (1992) A novel neuronal mes- Dawson et al., 1991). Inagaki et al. (1991) have observed high senger molecule in brain: the free radical, nitric oxide. Ann Neurol levels of SOD staining in somatostatin neurons of the caudate- 32:297-3 11. putamen. In the caudate-putamen large ACh-containing intcr- Dawson TM, Dawson VL, Snyder SH (1993) Nitric oxide as a me- neurons as well as NADPH-diaphorase neurons are resistant to diator of neurotoxicity. Nat1 lnst Drug Abuse Res Monogr Ser, in press. degeneration in Huntington’s disease (Ferrante et al., 1985), and Dawson VL, Dawson TM, London ED, Bredt DS, Snyder SH (1991) though the ACh-containing intemeurons do not stain for NOS Nitric oxide mediates glutamate neurotoxicity in primary cortical (T. M. Dawson et al., 199 l), these cells do display high densities culture. Proc Nat1 Acad Sci USA 88:6368-6371. of SOD staining (Inagaki et al., 199 1). Demerle-Pallardy C, Lonchampt MO, Chabricr PE, Braquet P (1991) Using several experimental approaches, we have established Absence of implication of L-arginineinitric oxide pathway on neuronal cell injury induced by L-glutamate or hypoxia. Biochem Biophys that the NO that mediates neurotoxicity derives from NOS Res Commun I8 1:456-464. neurons. Thus, destruction of 85-95% of NOS neurons but less Dcmirgoren S, Majewska MD, Spivak CE, London ED (I 99 I) Re- than lO-20°h ofthe total neuronal population in cortical cultures ceptor binding and clectrophysiological effects of dehydroepiandros- attenuates the ability of NMDA to elicit toxicity. Moreover, tcronc sulfate,an antagonistof the GABA, receptor.Neuroscience NMDA toxicity in cortical cultures does not appear until day 45:127-135. 17 and peaks at day 20, coincident with the development of Dubinsky JM, Rothman SM (199 I) Intracellular calcium concentrations during “chemical hypoxia” and exitotoxic neuronal injury. J NOS-positive cells. Finally, although they comprise only ap- Neurosci I 1:2545-2551. proximately l-2% of the total neuronal population, NOS neu- East SJ. Batchclor AM, Garthwaite J (1991) Selective blockade of rons have extensive processes and ramify to influence the ma- N-methyl-D-aspartate receptor function by the nitric oxide donor, jority of neurons in culture as well as in intact animals. Thus, nitroprusside. Eur J Pharmacol 209: 119-I 2 1. under conditions of excessive NO production, this small pop- Fcrrante RJ, Kowall NW, Beal MF, Richardson EP Jr, Bird ED, Martin ulation of neurons could kill a large population of surrounding JB (1985) Selective sparing of a class of striatal neurons in Hun- neurons. tington’s disease. Science 230:561-563. Furchgott RF (1990) The 1989 Ulf von Euler lecture. Studies on endothelium-dependent vasodilation and the endothelium-derived relaxing factor. Acta Physiol Stand 139:257-270. References Garthwaite G, Garthwaite J (1988) Cyclic GMP and cell death in rat Aizenman E, Lipton SA, Loring RH (1989) Selective modulation of cerebellar slices. Neuroscience 26:321-326. NMDA responses by reduction and oxidation. Neuron 2: 1257-I 263. Garthwaite J (1991) Glutamate, nitric oxide and cell-cell signalling Aizenman E, Hartnett KA, Reynolds IJ (I 990) Oxygen free radicals in the nervous system. Trends Ncurosci 14:60-67. reeulate NMDA receDtor function via a redox modulatory site. Neu- Hamill OP, Marty A, Neher E, Sakmann B, Sigworth FJ (198 I) Im- roi 5:841-846. _ proved patch-clamp techniques for high-resolution current recording Beal MF, Kowall NW, Ellison DW, Mazurek MF, Swartz KJ, Martin from cells and cell-free membrane patches. Pfluegers Arch 391:85- JB (1986) Replication of the neurochemical characteristics of Hun- 100. tington’s disease by quinolinic acid. Nature 32 1: 168-l 7 I. Hibbs JB Jr, Taintor RR, Vavrin Z (I 987) Macrophage cytotoxicity: The Journal of Neuroscience, June 1993. 13(6) 2661

role for t-arginine deaminase and imino nitrogen oxidation to nitrite. Meldrum B, Garthwaite J (I 990) Excitatory amino acid neurotoxicity Science 235:473dJ6. and ncurodegenerative disease. Trends Pharmacol Sci ll:379-387. Hope BT, Michael GJ, Knigge KM, Vincent SR (1991) Neuronal Michaels RL, Rothman SM (1990) Glutamate neurotoxicity in vifro: NADPH diaphorase is a nitric oxide synthase. Proc Nat1 Acad Sci antagonist pharmacology and intracellular calcium concentrations. J IJSA 88:281 I-2814. Neurosci 10:283-292. Ignarro LJ (1990) Biosynthesis and metabolism of endothclium-de- Moncada C, Lekieffre D, Arvin B, Meldrum B (1992) Effect of NO rived nitric oxide. Annu Rev Pharmacol Toxicol 30:535-560. synthase inhibition on NMDA- and ischaemia-induced hippocampal lnagaki S, Suzuki K, Taniguchi N, Takagi H (1991) Localization of lesions. Neuroreport 3:53C-532. Mn-superoxide dismutase (Mn-SOD) in cholinergic and somatostat- Moncada S, Palmer RMJ, Higgs EA (I 99 I) Nitric oxide: phvsioloav, in-containing neurons in the rat neostriatum. Brain Rcs 549: 174-l 77. pathophysiology and pharmacology: Pharmacol Rev 43: iO9-142.--. Izumi Y, Benz AM, Clifford DB, Zorumski CF (I 992) Nitric oxide Murohv-. SN. Thaver SA. Miller RJ (1987) The effects of excitatorv inhibitors attenuate N-methyl-D-aspartate excitotoxicity in rat hip- amino acids on;ntracellular calcium in single mouse striatal neurons pocampal slices. Neurosci Lett I35:227-230. in vitro. J Neurosci 7:4 1454 158. Kiedrowski L, Manev H, Wroblewski JT (1991) Inhibition of gluta- Nowicki JP, Duval D, Poignet H, Scatton B (1991) Nitric oxide me- mate-induced cell death by sodium nitroprusside is not mediated by diates neuronal death after focal cerebral ischemia in the mouse. Eur nitric oxide. Neuropharmacology 30: 124-l-l 243. J Pharmacol 204:339-340. Kiedrowski L. Costa E. Wroblewski JT (1992) Sodium nitromusside O’Dell TJ, Hawkins RD, Kandel ER, Arancio 0 (1991) Tests of the inhibits N-methyl-D-aspartatevoked calcium influx via a t&tic ox- roles of two diffusible substances in long-term potentiation: evidence ide- and cGMP-independent mechanism. Mol Pharmacol 41:779- for nitric oxide as a possible early retrograde messenger. Proc Natl 784. Acad Sci USA 88: I 1285-l 1289. Kinouchi H, Epstein CJ, Mizui T, Carlson E, Chen SF, Chan PH (I 99 I) Pou S, Pou WS, Bredt DS, Snyder SH, Rosen GM (1992) Generation Attenuation of focal cerebral ischemic injury in transgenic mice over- of supcroxide by purified brain nitric oxide synthasc. J Biol Chem expressing CuZn superoxide dismutase. Proc Natl Acad Sci USA 88: 267:24173-24176. 11158-11162. Puttfarcken PS, Lyons WE, Coyle JT (1992) Dissociation of nitric Koh J-Y, Choi DW (1988) Vulnerability of cultured cortical neurons oxide generation and kainate-mediated neuronal degeneration in pri- to damage by exocitotoxins: differentialsusceptibility of neurons con- mary cultures of rat cerebellar granule cells. Neuropharmacology 3 I: tainina NADPH-diaohorase. J Neurosci 8:2 153-2 163. 565-575. Koh J-f, Peters S, Choi DW (1986) Neurons containing NADPH- Radi R, Beckman JS, Bush KM, Freeman BA (1991a) Peroxynitrite- diaphorase are selectively resistant to quinolinate toxicity. Science induced membrane lipid peroxidation: the cytotoxic potential of su- 234~73-76. peroxide and nitric oxide. Arch Biochem Biophys 288:481487. Koh J-Y, Goldberg MP, Hartley DM, Choi DW (1990) Non-NMDA Radi R, Beckman JS, Bush KM, Freeman BA (199lb) Peroxynitritc receptor-mediated neurotoxicity in cortical culture. J Neurosci IO: oxidation of sulthydryls. The cytotoxic potential of superoxide and 693-705. nitric oxide. J Biol Chem 266:4244-4250. Lei SZ, Pan Z-H, Aggarwal SK, Chen H-SV, Hartman J, Sucher NJ, Randall RD, Thayer SA (1992) Glutamate-induced calcium transient Lipton SA (1992) Effect of nitric oxide production on the redox triggers delayed calcium overload and neurotoxicity in rat hippocam- modulatory site of the NMDA receptor-channel complex. Neuron pal neurons. J Neurosci 12: 1882-l 895. 8:1087-1099. Saez JC, Kessler JA, Bennett MVL, Spray DC (1987) Superoxide Loiacono RE, Beart PM (1992) Hippocampal lesions induced by mi- dismutase protects cultured neurons against death by starvation. Proc croinjection of the nitric oxide donor nitroprusside. Eur J Pharmacol Natl Acad Sci USA 84:3056-3059. 216:331-333. Sessa WC, Hecker M, Mitchell JA, Vane JR (1990) The metabolism Lustig HS, von Brauchitsch KL, Chan J, Greenberg DA (I 992a) Cyclic of t-arginine and its significance for the biosynthesis of endothelium- GMP modulators and excitotoxic injury in cerebral cortical cultures. derived relaxing factor: L-glutamine inhibits the generation of t.-ar- Brain Res 577~343-346. ginine by cultured endothelial cells. Proc Natl Acad Sci USA 87:8607- Lustig HS, von Brauchitsch KL, Chan J, Greenberg DA (1992b) Eth- 861 I. anol and excitotoxicity in cultured cortical neurons: differential sen- Smith RP, Louis CA, Kruszyna R, Kruszyna H (1991) Acute neu- sitivity of N-methyl-D-aspartate and sodium nitroprusside toxicity. J rotoxicity of sodium axide and nitric oxide. Fundam Appl Toxicol Neurochem 59:2193-2200. 17:120-127. Manzoni 0, Prezeau L, Desagher S, Sahuquet A, Sladeczek F, Bockaert Snyder SH (I 992) Nitric oxide: first in a new class of neurotransmit- J, Fagni L (1992a) Sodium nitroprusside blocks NMDA receptors ters? Science 257:494-496. via formation of ferrocyanide ions. Neuroreport 3:77-80. Snyder SH, Bredt DS (1991) Nitric oxide as a neuronal messenger. Manzoni 0, Prezeau L, Marin P, Deshager S, Bockaert J, Fagni L Trends Pharmacol Sci 12: 125-l 28. (I 992b) Nitric oxide-induced blockade of NMDA receptors. Neuron Stuchr DJ, Fasehun OA, Kwon NS, Gross SS, Gonzalez JA, Levi R, 81653-662. Nathan CF (1991) Inhibition of macrophage and endothelial cell Marletta MA (1989) Nitric oxide: biosynthesis and biological signif- nitric oxide synthase by diphenyleneiodonium and its analogs. FASEB icance. Trends Biochem Sci 14:488492. J 5:98-103. Martin W, Villani GM, Jothianandan D, Furchgott RF (1985) Sclcc- Trifiletti RR (1992) Neuroprotective effects of N”-nitro-L-arginine in tive blockade of endotheliumdependent and glyceryl trinitrate-in- focal stroke in the J-day-old rat. Eur J Pharmacol 2 18: 197-198. duced relaxation by hemoglobin and by mcthylene blue in the rabbit Wallis RA, Panizzon K, Wasterlain CC (1992) Inhibition of nitric aorta. J Pharmacol Exp Ther 232:708-J 16. oxide synthase protects against hypoxic neuronal injury. Neuroreport McCall TB, Feel&h M, Palmer RMJ, Moncada S (I 99 I) Identification 3:645-648. ofN-iminoethyl+omithine as an irreversible inhibitor ofnitric oxide synthase in phagocytic cells. Br J Pharmacol 102:234-238.