Neuron,

Vol. 8, 3-11, January, 1992, Copyright

0 1992 by Cell Press

Nitric Oxide, a Novel Neuronal Messenger David S. Bredt and Solomon H. Snyder Departments of Neuroscience, Pharmacology and Molecular Sciences, Psychiatry and Behavioral Sciences Johns Hopkins Medical Institutions Baltimore, Maryland 21205

Nitric oxide (NO) is a free radical gas, clearly distinct from nitrous oxide, N20, which is used as an anesthetic and is chemically stable. N O has long been known to occur in bacteria, but until about 4 years ago, there was no evidence for its biological functions in vertebrates. Evidence since that time has established a major role for N O as a messenger molecule in at least three systems: white blood cells, where N O mediates tumoricidal and bactericidal effects; blood vessels, where it represents endothelium-derived relaxing factor activity; and, most recently, as a neuronal constituent with functions very much like those of a neurotransmitter. Identification and Brain

of N O in Endothelium,

Macrophages,

The first evidence that N O is formed endogenously came from studies of dietary nitrates as a source of carcinogenic nitrosamines. Tannenbaum and associates (Green et al., 1981a, 1981b; Wagner et al., 1983) noted that urinary nitrate excretion is maintained in humans and rats fed diets low in nitrate content. A subject with infectious diarrhea excreted very high levels of nitrates, suggesting a role for inflammatory processes as a source. Mice selectively deficient in macrophages evince low urinary nitrate excretion (Stuehr and Marletta, 1985), and isolated cultures of macrophages form large amounts of nitrates when stimulated byendotoxin (lyengar et al., 1987). Arginine was shown to be a source of the nitrates, as its removal from incubation media prevents the formation of nitrites and nitrates (Hibbs et al., 1987a). Hibbs and collaborators (1987b) noted that the bactericidal and tumoricidal actions of macrophages in culture are abolished when arginine is removed from the medium. Two independent lines of investigations led to identification of a role for N O in blood vessels. Furchgott and Zawadzki (1980) observed that relaxation of blood vessels by acetylcholine is abolished when the endothelial layer is removed, but can be restored by reapplying endothelial cells to the smooth muscle layer in a fashion indicating that a diffusible molecule from the endothelium mediates the muscle relaxation. The endothelium-derived relaxing factor was notoriously unstable and difficult to isolate despite efforts by several laboratories. Insight came from a parallel line of investigation showing that the smooth muscle-re-

Review

laxing effects of nitroglycerine and other organic nitrate vasodilators involve an active metabolite, NO, whose properties are very much like those of the endothelium-derived relaxing factor (Murad et al., 1978; Feelisch and Noack, 1987). Moncada and colleagues (Palmer et al., 1987, 1988) directly demonstrated that chemically detectable N O synthesized from arginine is released in sufficient amounts from endothelial cells to account for all endotheliumderived relaxing factor activity. In the brain, Garthwaite and collaborators (1988) noted that dissociated cultures of neonatal cerebellar cells release a factor whose actions on blood vessels are reminiscent of NO. NO-forming activity was directly demonstrated from arginine in brain extracts (Radomski et al., 1990). The regulation of N O synthase (NOS), the enzyme that generates NO, by neurotransmitters in the brain was first examined in the cerebellum, on the basis of what is known of how N O dilates blood vessels. N O brings about smooth muscle relaxation by binding to iron in the heme that is part of soluble guanylyl cyclase, stimulating the formation of cGMP. cGMP in turn stimulates a protein kinase, which phosphorylates the light chain of myosin, eliciting relaxation. The cerebellum contains the highest levels of cGMP in the brain, with its formation stimulated by glutamate acting via N-methyl-o-aspartate (NMDA) receptors (Ferrendelli et al., 1974). By monitoring the conversion of arginine to N O or to citrulline, we demonstrated in cerebellar slices that NOS activity, which is formed stoichiometrically with NO, is enhanced 300% in response to NMDA receptor stimulation (Bredt and Snyder, 1989). The concentration-response relationships for NOS activation are the same as those for the stimulation of cGMP levels. The enhanced NOS activity is responsible for the increased levels of cCMP, since we (Bredt and Snyder, 1989) and others (Garthwaite et al., 1989) observed that NC-monomethylarginine (L-NMMA), an inhibitor of NOS, completely prevents the stimulation of cGMP formation, with a concentration-response relationship which is the same as that for NOS inhibition. The effect is specific, as it is reversed by arginine and not by other amino acids. Nitric Oxide Synthase Major insights into N O disposition have come from characterization of NOS. With crude or partially purified preparations, it was apparent that the macrophage enzyme differs substantially from the brain and endothelial enzymes, both of which are quite similar if not identical. All forms of NOS use arginine as their substrate, form citrulline stoichiometricallywith NO, and require nicotinamideadenine dinucleotide phosphate (NADPH) as an electron donor. However, the brain and endothelial enzymes are stimulated by cal-

Neuron

4

cium even in crude preparations, whereas the macrophage enzyme does not require calcium. Additional insight into N O formation and its regulation has come from purification of the brain enzyme (Bredt and Snyder, 1990) and its molecular cloning (Bredt et al., 1991c). initial attempts to purify the enzyme, the majority of whose activity is soluble, revealed that its catalytic activity was lost even with single purification steps (Bredt and Snyder, 1990). However, enzyme activity could be reconstituted by recombining different fractions from DEAE columns, indicating that this purification procedure dissociates a crucial cofactor. In view of the calcium requirement for enzyme activity, calmodulin was examined as a possible cofactor and turned out to be essential for enzyme activity. Half-maximal stimulation bycalmodulin, in the presence of calcium, was observed at 10 nM concentration. This finding explained what had been a perplexing question, namely, that in cerebellar slices glutamatergic agonists could stimulate NOS activity 3-fold within seconds, as early as one could monitor activity (Bredt and Snyder, 1989). The likely explanation for this rapid effect is that NMDA receptors open calcium ion channels, with the calcium binding to calmodulin to activate NOS. Based on the requirement of NOS for NADPH, a 2’,5’-ADP affinity chromatography column provided extensive purification in a single step (Bredt and Snyder, 1990). The purified brain enzyme has a monomer molecular weight of 160 kd. NOS appears to be a highly regulated enzyme. Purified NOS is phosphorylated by cAMP-dependent protein kinase, protein kinase C (PKC), and calciumlcalmodulin-dependent protein kinase II (Bredt et al., 1991a). With all three enzymes, phosphorylation is stoichiometric and occurs on serine, but the three enzymes phosphorylate distinct peptides. To determine the effects of phosphorylation on NOS enzymatic activity, we utilized kidney cells stably transfected with NOS cDNA. Activation of PKC in these cells with phorbol ester leads to a rapid phosphorylation of NO5 accompanied by a greater than SO% decrease in NOS activity (Bredt et al., 1991a). Downregulation of NOS activity by PKC phosphorylation represents a mechanism for “cross-talk” between these two major messenger systems. NOS has tightly bound flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN), with 1 molecule of each per molecule of NOS (Bredt et al., 1991c). Presumably, electrons are transferred successively between NADPH and the two flavins as part of NOS catalytic activity. NOS also contains a tightly bound molecule of tetrahydrobiopterin and 1 moleof iron per NOS monomer (Mayer et al., 1991). The primary structure of brain NOS, revealed by molecular cloning (Bredt et al., 1991c), indicates that the protein has an a helical, calmodulin-binding consensus sequence and a CAMP-dependent protein kinase phosphorylation sequence (Figure 1). Sequences for phosphorylation by PKC or calciumlcalmodulin protein kinase are not sufficiently selective to be de-

NH2

P P

CaM

N+4DPH IFMN!

IFAD{

L&

H

Figure 1. Schematic Model the Cofactor

Recognition

Showingthe Sites within

~0s

CPR

Spatial Relationships NOS and CPR

of

Predicted sites for calmodulin (CaM) binding and protein phospho+ation within the NO5 sequence and the transmembrane domain (TM@ in the CPR sequence are noted (adapted with permission from Nature).

finitively identified in NOS. Well-defined sites for binding of NADPH, FMN, and FAD are apparent. Of all sequenced mammalian proteins, NOS displays close homology only with cytochrome P-450 reductase (CPR). The two proteins are most similar in the C-terminal half of NOS, which shows 58% homologywith CPR over 641 amino acids. NOS also has substantial homology with bacterial sulfite reductase. NOS, CPR, and sulfite reductase are unique in possessing binding sites for NADPH, FMN, and FAD in the same polypeptide. In the two reductases, the flavins constitute a transport chain with electrons shuttling between the isoalloxazine rings, a reaction that is likely to take place also in NOS. The similarity of CPR to the C-terminal half of NOS suggests that in phylogeny, CPR may have provided theelectron transport necessaryfor NOS activityeven though it was a distinct protein. When the N-terminal and C-terminal halves of NOS are expressed together, the mixture provides NOS catalyticactivity(D. S. Bred: and S. H. Snyder, unpublished data). Thus, the C-terminal half can transfer electrons”in trans”to the N-terminal domain of NOS, suggesting that CPR might have a similar property. Molecular cloning of macrophage NOS reveals substantial homology with brain NOS, especially in the areas that resemble CPR in the C-terminal portion of the molecule (C. Lowenstein, D. S. Bredt, and S. H. Snyder, unpublished data). Purification of a neutrophil NOS indicates differences from both macrophage and brain enzymes (Yui et al., 1991). The neutrophil enzyme, like the macrophage enzyme, does not require calmodulin, but like the brain enzyme, does require calcium. Yet another form of NOS, which is dependent upon addition of catalase, has been identified (Manitz and Schrader, 1991, Biol. Nitric Oxide, abstract). By contrast, brain and macrophage NOS can form N O without the addition of catalase. NOS Localization Because of its instability, N O cannot readily be localized in tissues. With theavailabilityof pure brain NOS, antisera were obtained, permitting immunohistochem-

Review: Nitric Oxide 5

ical localization of the enzyme (Bredt et al., 1990), and molecular cloning has enabled NOS mRNA to be localized by in situ hybridization (Bredt et al., 1991b). The antiserum employed reacts with brain and endothelial NOS, but does not recognize macrophage NOS, so that the distribution of the macrophage enzyme is not known.Throughoutthe body, NOS immunoreactivity occurs exclusively in neurons or in the endothelium of blood vessels. Within the brain, no glia are stained. NOS activity with properties of the brain enzyme has been reported in cultures of astrocytes (Murphy et al., 1990), which seems at variance with the absence of glial staining for NOS in the brain. Conceivably, astrocyte proliferation following brain injury might stimulate NOS expression. Specificity of the immunohistochemistry is evident from the close parallel between levels of immunoreactivity and of NOS catalytic activity in different brain regions. The highest densities of NOS occur in neurons in the cerebellum and olfactory bulb, with the accessory olfactory bulb being even more heavily labeled (Figure 2). In the cerebellum NOS mRNA occurs most prominently in the granule cell layer, whereas immunoreactive NOS is concentrated in the molecular as well as the granule layer, indicating that NOS in the molecular layer probably occurs.largely in processes of granulecells. NOS in thecerebellum is also highlyconcentrated in basket cells. Purkinje cells of the cerebellum are devoid of NOS. This pattern of localizations provides acellular basis for the regulation of NOS and cGMP by glutamate. cGMP is quite prominent in Purkinje cells. Presumably, stimulation of NMDA receptors on granule cells or basket cells triggers the formation of NO, which diffuses to Purkinje cells to activate guanylyl cyclase. Glia may also be involved, as cGMP has been visualized by immunohistochemistry in astrocytes of the cerebellar granule cell layer as well as in Bergmann glia, whose processes radiate through the’molecular layer (de Vente et al., 1989). In immature rat cerebellum, NMDA elicits cGMP accumulation in cerebellar astrocytes and Bergmann glia, as does nitroprusside, which generates NO. Stimulation of cGMP levels by NMDA, but not by nitroprussjde, is abolished by inhibitors of NOS (Bredt and Snyder, 1989; Garthwaite et al., 1989). However, recent evidence indicates that concentrations of NMDA necessary to stimulate glial accumulation of cCMP are much higher than those that enhance cGMP formation physiologically in the cerebellum, presumably in Purkinjecells (Garthwaite, 1991). Thus, a physiologic role for glia as targets of N O is uncertain. ’ In the cerebral cortex and hippocampus NOS occurs in scattered, isolated cells that are medium to large aspiny neurons. Granule cells of the dentate gyrus possess abundant NOS, but no pyramidal cells of the hippocampal layers are stained. Similarly, in the corpus striatum NOS appears in scattered, medium to largeaspinyneurons in both cell bodiesand neuropil.

While most areas display NOS-containing cells, prominent NOS in the islands of Callejae is confined to fibers. These localizations of NOS may have bearing upon how N O exerts its physiologic effects in the brain. In the cerebellum cGMP appears to be involved in the mediation by N O of glutamate actions on NMDA receptors. Does cGMP mediate all actions of N O in the brain? If so, one would expect a similar localization of NOSandcGMPorguianylyl cyclase.Various immunohistochemical studies give somewhat different patterns for localizations of guanylyl cyclase and cGMP (Nakane et al., 1983; Chan-Palay and Palay, 1979). However, in no instance is there a close relationship of these markers with the distribution of NOS. Thus, other molecular targets for N O presumably exist. Because of its highly reactive properties, N O could interact with many chemical groups as well as with iron in enzymes other than guanylyl cyclase. The intense staining for NOS in only 2% of cerebral cortical, striatal, and hippocampal neurons does not fit with the distribution of any known neurotransmitter. In some areas of the brain there is colocalization of NOS with particular neurotransmitters. For instance, in the corpus striatum all NOS-staining neurons also stain for somatostatin and neuropeptide Y, but in other areas, such as the pedunculopontine nucleus of the brain stem, NOS neurons lack somatostatin and neuropeptide Y. By contrast, in the pedunculopontine nucleus all NOS neurons stain also for choline acetyltransferase, whereas NOS neurons of the striatum lack the enzyme. NOS neurons colocalize with NADPH diaphorase (NDP) (T. M. Dawson et al., 1991; Bredt et al., 1991b). NDP staining reflectsa blue precipitateobtained from Nitro Blue Tetrazolium only in the presence of NADPH (Thomas and Pearse, 1964). Throughout the brain and, with limited exceptions, throughout the peripheral nervous system, NDP and NOS localizations are identical. This is particularly striking in areas of the brain such as the cerebral cortex, where only I%-2% of neuronal cells, scatteredin a seemingly random array, are positive for NOS and are invariably positive for NDP. The coincidence of NOS and NDP is not species specific to rat, but is also demonstrable in monkey brain (Bredt et al., 1991b). That NOS catalytic activityfullyaccountsfor all NDP catalytic staining has been established in experiments utilizing cloned and expressed NOS (T. M. Dawson et al., 1991). NOS cDNA was transfected into human kidney 293 cells, which prior to transfection stain for neither NOS nor NDP. Following transfection, cells stain for both, with the ratio of NOS and NDP staining being identical to the ratios of staining in neurons in the brain and periphery. In principle one might also be able to purify NDP catalytic activity to determine whether it copurifies with NOS. However, since NDP activity can be provided by diverse oxidative-reductive activities utilizing NADPH, numerous other enzymatic activities could give rise to presumed NDP activ-

Figure 2. Localization

of NOS in Brain and Peripheral

Tissues

Histologic localizations of NOS protein, NADPH diaphorase activity, and mRNA in brain. Adjacent sagittal brain sections were processed for NOS immunohistochemistry (A), NADPH diaphorase histochemistry (B), and NOS in situ hybridization (C). All three methods show densest staining in the accessory olfactory bulb (AOB), pedunculopontine tegmental nucleus (PPN), and cerebellum (CB), with lesser staining in the dentate gyrus of the hippocampus (DC), main olfactory bulb (OB), superior and inferior colliculus (C), and supraoptic nucleus (SO). Intensely staining isolated cells are apparent scattered throughout the cerebral cortex (CX), caudate putamen (CP), and basal forebrain. Some regions enriched in NOS protein and diaphorase staining are devoid of NOS mRNA, suggesting that in these regions the NOS protein has been transported in nerve fibers distant from its site of synthesis. These regions include the molecular layer of the cerebellum, the islands of Callejae (ICj), and the neurophil of the caudate putamen and cerebral cortex. (D) Section through rat adrenal gland shows dark labeling of ganglioncells(CC)in themedulla(M),with nostainingofadrenal cortex (CX). (E) Staining in the pituitary for NOS is enriched in the posterior lobe(P), with negligible immunoreactivity in the intermediate (I) and anterior (A) lobes. (F) Longitudinal section through the rat duodenum reveals intense white immunofluorescence in the neurons of the myenteric plexus (MP) and their associated axons running in parallel with the inner circular muscle layer (.IC). S, submucosa. (G) Cross section of aorta with strong white immunofluorescence of the endothelial cells, whose nuclei can be seen bulging

Review: Nitric Oxide 7

Arginine

o* 1. Ca-/CaM

Arginine-OH

02 NADPH Citrulline + C~**,CdYl

NO’

t

NOS-flavin -

-

NOS-flavinH2 UADPH NBT

-

t

NBT fomazan

“NADPH Dmohorase”

Figure 3. Biosynthesis

of Nitric

Oxide

NOS initially receives electrons from NADPH in a calcium/calmodulinand arginine-independent step that presumably involves reduction of one or both of the associated flavins. The reduced NOS will rapidly reoxidize in the presence of an ap propriate electron acceptor, such as Nitro Blue Tetrazolium (NBT), giving rise to the NADPH diaphorase activity. In the presenceofcalcium,calmodulin,and molecular oxygen, the reduced NOS will hydroxylate the guanidino group of arginine in a P-4!+ like monooxygenase reaction (Stuehr et al., 1991). The tightlyassociated hydroxyarginine is then further oxidized by NOS to citrulline and NO.

ity. Indeed, multiple protein fractions provide NDP catalytic activity that is distinct from the NDP activity intrinsic to purified NOS (Figure 3) (Hope et al., 1991). One striking feature of NDP neurons is their selective resistance to destruction in clinical neurodegenerative conditions such as Huntington’s disease (Ferrante et al., 1985), as well as following ischemic (Uemura et al., 1990) or neurotoxin-induced (Beal et al., 1986; Koh et al., 1986) destruction of neural tissue. In Huntington’s disease up to 95% of striatal neurons degenerate, while virtually all NDP neurons survive. Treatment of primary cerebral cortical cultures with NMDA destroys 90% or more of neurons, but NDP neurons survive. NOS activity, which accounts for the NDP staining, may be responsible for this resistance. Calcium overload is one potential mechanism of neurotoxicity, and N O can reduce intracellular levels of calcium by mechanisms unrelated to cGMP (Garg and Hassid, 1991). The diaphorase activity of NOS may be relevant, since induction in neuronal cultures of a similar enzyme, DT diaphorase, protects neurons from oxidative toxicity (Murphy et al., 1991). Outside the brain NOS occurs in a number of striking locations, and NDP colocalizes at all sites (T. M. Dawson et al., 1991). In the pituitary gland, fibers and terminals in the posterior lobe stain intenselywhereas the intermediate and anterior lobes have very low lev-

in the lumen (arrow). L, lumen; E, endothelial cell layer; M, tunica media. (H) Oblique section through thecoronaryarterywith perinuclear labeling of a sheet of endothelial cells. MC, myocardium. (I) Dark immunoperoxidase staining of a distal cerebral vessel shows labeling restricted to endothelial cells (E). A, adventitia. (I) Anterior communicating cerebral artery shows labeling of the endothelium as well as the dense innervation about the adventitia. (A, B, and C) Adapted from Neuron; (D) adapted from Proc. Nat/. Acad. Sci. USA; (E, C, H, I, and J) adapted from Nature.

els of NOS. There is also abundant NOS in neurons of the supraoptic and paraventricular nuclei of the hypothalamus that project to the posterior pituitary. In the adrenal gland NOS occurs in discrete ganglion cells and fibers in the medulla, making close contact with the chromaffin cells that secrete catecholamines (Bredt et al., 1990; T. M. Dawson et al., 1991). The adrenal cortex is the”exception that proves the rule,” in that substantial NDP staining occurs in the absence of NOS. Oxidative enzymes that are associated with steroid synthesis and that utilize NADPH presumably account for NDP staining there. In both the adrenal medulla and the posterior pituitary, secretion is accompanied by massive increases in blood flow (Breslow et al., 1987). Indeed, it has been virtually impossible to dissociate these two, and thus it was thought possible that selective enhancement of blood flow provoked secretion. Splanchnic nerve stimulation augments both blood flow and catecholamine secretion from the adrenal medulla. Nitroarginine, a potent inhibitor of NOS, blocks the augmentation in adrenal medullary blood flow produced by splanchnic nerve stimulation, but is without effect on catecholamine secretion (Breslow et al., 1992). In the retina, a plexus of nerve fibers in the choroid stains for NOS, as does a limited population of amacrine cells in the inner nuclear layer and occasional cells in the ganglion cell layer (Bredt et al., 1990; T. M. Dawson et al., 1991). The NOS neurons in the blood vessels of the choroid and limbus appear to derive from the sphenopalatine (pterygopalatine in the rat) ganglia and thus likely also contain vasoactive intestinal polypeptide (VIP) (R. Yamamoto, D. S. Bredt, S. H. Snyder, and R. A. Stone, unpublished data). Neuronal cells and processes of the myenteric plexus throughout the gastrointestinal tract stain for NOS (Bredt et al., 1990). Detailed analysis reveals that NOS is exclusively associated with inhibitory motor neurons whose axons project posteriorly in the anal direction and interact selectively with the inner circular muscle layer (Furness et al., 1991, Aust. Physiol. Pharmacol. Sot., abstract). In these neurons NOS is selectively colocalized with VIP. Besides neuronal sites, NOS is concentrated in endothelial layers of larger blood vessels, such as the aorta, coronary arteries, and blood vessels in the brain (Bredt et al., 1990). Large cerebral blood vessels also display NOS in nerve fibers within the adventitia, a pattern not evident in smaller distal cerebral or peripheral vessels. Lesions of the VIP-containing input fromthesphenopalatinegangliamarkedlyreducethis staining, indicating a similar source for NOS in vasodilator nerves in blood vessels of the eye and brain (Nozaki et al., 1991, Stroke Meeting, abstract). Neuronal

Functions of NOS

Specific neuronal functions of NOS have been best clarified in regions where N O affects smooth muscle tone. The high concentration of NOS in the myenteric

a

plexus throughout the gastrointestinal pathway suggests that NO might mediate functions of these nerves. One major activity of the myenteric plexus is to mediate nonadrenergic-nonchofinergic(NANC) relaxation of the intestines, an important component of peristalsis. Studies over many years established that classic biogenic amine neurotransmitters cannot account for NANC relaxation; purines (Burnstock et al., 1991) and peptides (Goyal et al., 1980) appeared to be alternative candidates. Potent and selective inhibitors of NOS have provided strong evidence that NO mediates NANC relaxation. In numerous parts of the gastrointestinal system, including the lower esophageal sphincter (Tttrup et al., 1991), the stomach (Desai et al., 1991), the large and small intestines (Bult et al., 1990), and the anococcygeus muscle (Gillespie et al., 1989; Ramagopal and Leighton, 1989; Gibson et al., 1990), NOS inhibitors such as L-NMMAand nitroarginine potently block NANC relaxation. These effects are reversed by arginine, confirming their selectivity. Nitroprusside, which generates NO, mimics the effects of NANC neuronal stimulation. Similar experiments have established a role for NO in relaxation of the corpora caveronosae of the penis (Ignarro et al., 1990) and in the neurally mediated relaxation of cerebral blood vessels (Toda and Okamura, 1990). In all these systems, NO fulfills many criteria of a neurotransmitter. Thus, its synthetic enzyme is present in the relevant neurons, it can mimic the effects of physiologic nerve stimulation, and blockade of its formation prevents the effects of nerve stimulation. In the intestine the colocalization of NOS and VIP in myenteric plexus neurons suggests some type of interaction. Li and Rand (1990) obtained evidence for contributions to NANC relaxation by both, as antibodies to VIP partially reduce NANC relaxation and L-NMMA abolishes the portion resistant to VIP antibodies.Cotransmission by NOandVIPmayalsooccur in neurons in the eye and cerebral arteries where VIP and NOS are colocalized. NO may also play a role in synaptic plasticity such as long-term potentiation (LTP). Bohme et al. (1991) examined LTP in the CA1 region of the hippocampus in slice preparations. In the slice, 0.1 uM nitroarginine completely blocks LTP, an effect reversed by L-arginine, while nitroprusside produces a long-lasting enhancement in synaptic efficacy that is not additive with tetanus-induced LTP. O’Dell et al. (1991) have also observed blockade of CA1 LTP by both nitroarginine, which blocks LTP when injected into the postsynaptic cell, and by hemoglobin, which is not taken up by cells and presumably binds NO that diffuses from the postsynaptic cell into the extracellular space. O’Dell et al. (1991) also found that NO enhances spontaneous transmitter release from cultured hippocampal pyramidal cells, suggesting that NO may serve as a retrograde factor in LTP. Interestingly, immunohistochemical studies indicate that CA1 pyramidal cells lack NOS, although they stain for NADPH diaphorase, suggesting that they might contain a different isoform (Bredt

et al., 1991b; D. S. Bredt and S. H. Snyder, unpublished data). Long-term depression (LTD) in the cerebellum is another model of synaptic plasticity in which conjunctive stimulation of climbing and parallel fibers evokes LTD of parallel fiber-Purkinje cell transmission (Ito, 1989). This system has been implicated as the cellular mechanism for cerebellar motor learning. Shibuki and Okada (1991) demonstrated release of NO followingstimulationofclimbingfibersand blockadeofLTD by L-NMMA and hemoglobin. Moreover, nitroprusside or cGMP was able to substitute for stimulation of climbing fibers in evoking LTD. In cultured Purkinje nejlrons LTD may not require NO signaling, as it is unaffected by nitroarginine or hemoglobin (Linden and Connor, 1991). Besides its proposed role in protecting NOS neurons from neurotoxicity, NO may itself induce neurotoxicity. Neurotoxicity associated with cerebral ischemia is thought to involve glutamatergic stimulation via NMDA receptors. In support of this, NMDA antagonists block neuronal destruction in various ischemic stroke models (Choi, 1991). Though NMDA stimulates NO formation in NOS neurons, they are selectively resistant to NMDA neurotoxicity. A way out of this dilemma would be to propose that normally the neurons making NO release it to elevate cGMP in adjacent neurons without toxicity. However, in the presence of large, toxic levels of glutamate, NOS neurons would behave like macrophages, releasing large amounts of NO to kill nearby neurons. This possibility has been explored in primary cerebral cortical neuronal cultures. When NMDA is added to cultures briefly, 80% of the neurons die within 24 hi(V. L. Dawson et al., 1991). In this system nitroarginine and L-NMMA prevent neurotoxicity elicited by NMDA and related excitatory amino acids (V. L. Dawson et al., 1991). This effect is competitively reversed by L-arginine. Moreover, depletion of arginine in the culture medium, by adding arginase or utilizing arginine-free growth medium, prevents NMDA toxicity. Nitroprusside produces dose-dependent cell death that parallels cCMP formation. Moreover, hemoglobin, which complexes NO, prevents the neurotoxic effects of both NMDA and nitroprusside. Taken together, these findings establish that NO mediates the glutamate neurotoxicity in cultures. Conceivably, this NOcouldariseeitherfrom neuronsorfrom microglia, which are the equivalent of macrophages in the brain. These possibilities were distinguished by taking advantage of findings that, although they resist NMDA toxicity, NOS neurons are more susceptible than other neurons to kainate or quisqualate toxicity (Koh et al., 1986). Low doses of quisqualate, which selectively destroy NOS neurons, reduce the toxicity of subsequently administered NMDA(T. M. Dawson and S. H. Snyder, unpublished data). Thus, the NOS neurons are presumably the major source of the NO that mediates NMDA toxicity. Recent evidence directly implicates NO in neuronal

Review: Nitric Oxide 9

damage associated with vascular strokes. In mice with ligations of the middlecerebral artery, intraperitoneal injections of nitroarginine substantially reduce the volume of infarcted brain tissue (Nowicki et al., 1991). Neuronal protection occurs even when nitroarginine is administered after arterial ligation, suggesting a potential therapeutic application in stroke patients. These protective effects are exerted at strikingly low doses of nitroarginine, with 1 mg/kg given 5 min and 3, 6, 24, and 36 hr after artery ligation providing 70% protection. By contrast, maximal protection by the NMDA antagonist MK-801 is only 55%. Because of their charged guanidino groups, arginine derivatives might not be expected to penetrate into the brain. However, in rat 5 mglkg nitroarginine administered intraperitoneally daily for 4 days irreversibly inhibits brain N O activity (Dwyer et al., 1991). Conclusions Clearly, N O is an interesting messenger molecule. In the nervous system it fulfills most of the criteria of a neurotransmitter. However, it is certainly an aberrant transmitter. It is not stored in synaptic vesicles, but instead seems to be formed on demand. N O is not released by exocytosis, nor does it act upon receptor proteins on neuronal membranes. Instead, it diffuses out of one neuron and into another neuron, with its receptor being iron at the active site of guanylyl cyclase or some other molecular target. It is conceivable that N O is only the first of a class of unstable, small molecular messengers. The close similarity of NOS and CPR is provocative and perplexing. One common element is that CPR is the electron donor for the cytochrome P-450 enzymes, which form free radicals and related molecules. Might some such molecules have synaptic functions? To explore this possibility, we have mapped CPR and demonstrated localizations in selected neuronal populations, complementary in part to those of NOS (M. Fotuhi, T. M. Dawson, and S. H. Snyder, unpublished data). Earlier, other workers showed selective concentrations of CPR in catecholamine neurons in the brain (Haglund et al., 1984). Certain cytochrome P-450 enzymes demonstrate selective neuronal localizations in the brain (Warner et al., 1988). Another “atypical” neuronal messenger might be carbon monoxide (Marks et al., 1991). As in the N O system, carbon monoxide can be formed by two types of heme oxygenase, type 1, which occurs primarily in peripheral tissues and is inducible, and type 2, which is not inducible and occurs in high concentrations in the brain (Trakshe1 et al., 1988). By in situ hybridization, we have demonstrated selective localizations of mRNA for the type 2 heme oxygenase in discrete neuronal populations in the hippocampus and cerebellum (D. Hirsch, A. Verma, C. Glatt, and S. H. Snyder, unpublished data). Whether or not N O is unique or merely the first of a new class of atypical neuromodulators, it is clear

that the ubiquitous distribution of N O in neuronal populations throughout the body, where its physiologic functions have already been demonstrated, portends a new way of thinking about neuronal communication.

This work was supported by USPHS grants DA-00266, MH-18501, Contract DA271-g&7408, and Research Scientist Award DA-00074 (to S.H.S.), training grant GM-07309 (to D.S.B.), a grant from the International Life Sciences Institute, and a gift from Bristol-Myers Squibb. References Beal, M. F., Kowall, N. W., Ellison, D. W., Mazurek, M. F., Swartz, K. J., and Martin, J. B. (1986). Replication of the neurochemical characteristics of Huntington’s disease by quinolinic acid. Nature 327, 168-171. Bohme, C. A., Bon, C., Stutzmann, J.-M., Doble, A., and Blanchard, J.-C. (1991). Possible involvement of nitric oxide in long-term potentiation. Eur. J. Pharmacol. 799, 379-381. Bredt, D. S., and Snyder, 5. H. (1989). Nitric oxide mediates glutamate-linked enhancement of cCMP levels in the cerebellum. Proc. Natl. Acad. Sci. USA 86, 9030-9033. Bredt, D. S., and Snyder, S. H. (1990). Isolation of nitric oxide synthetase, a calmodulin-requiring enzyme. Proc. Natl. Acad. Sci. USA 87, 682-685. Bredt, D. S., Hwang, P. M., and Snyder, S. H. (1990). Localization of nitric oxide synthase indicating a neural role for nitric oxide. Nature 347, 768-770. Bredt, D. S., Ferris, C. D., and Snyder, S. H. (1991a). Nitric oxide synthase regulatory sites: phosphorylation by cyclic AMP dependent protein kinase, protein kinase C, and calcium/calmodulin protein kinase; identification of flavin and calmodulin binding sites. J. Biol. Chem., submitted. Bredt, D. S., Glatt, C. E., Hwang, P. M., Fotuhi, M., Dawson, T. M., and Snyder, S. H. (1991b). Nitric oxide synthase protein and mRNA are discretely localized in neuronal populations of the mammalian CNS together with NADPH diaphorase. Neuron 7, 615-624. Bredt, D. S., Hwang, P. H., Glatt, C., Lowenstein, C., Reed, R. R., and Snyder, S. H. (1991c). Cloned and expressed nitric oxide synthase structurally resembles cytochrome P-450 reductase. Nature 352, 714-718. Breslow, M. J., Jordan, D. A., Thellman, S. T., and Traystman, R. J. (1987). Neural control of adrenal medullary and cortical blood flow during hemorrhage. Am. J. Physiol. 252, H521-H528. Breslow, M. J., Tobin, J. R., Bredt, D. S., Ferris, C. D., Snyder, S. H., and Traystman, R. J. (1992). Role of nitric oxide in adrenal medullary vasodilation during catecholamine secretion. Eur. J. Pharmacol., in press. Bult, H., Boeckxstaens, G. E., Pelckmans, P. A., Jordaens, F. H., Van Maercke, Y. M., and Herman, A. C. (1990). Nitric oxide as an inhibitory non-adrenergic non-cholinergic neurotransmitter. Nature 345, 346-347. Burnstock, C., Satchell, D. C., and Smythe, A. (1991). .A comparison of the excitatory and inhibitory effects of non-adrenergic, non-cholinergic nerve stimulation and exogenously applied ATP on a variety of smooth muscle preparations from different vertebrate species. Br. J. Pharmacol. 46, 234-242. Char-r-Palay, V., and Palay, S. L. (1979). lmmunocytochemical localization of cyclic CMP: light and electron microscope evidence for involvement of neuroglia. Proc. Natl. Acad. Sci. US.A 76,14851488.

Choi, D. W. (1991). Glutamate neurotoxicity nervous system. Neuron 1, 623-634.

and diseases

of the

Dawson,T. M., Bredt, D. S., Fotuhi, M., Hwang, P. M.,and Snyder, S. H. (1991). Nitric oxide synthase and neuronal NADPH diaphorase are identical in brain and peripheral tissues. Proc. Natl. Acad. Sci. USA 88, 7797-7801. Dawson, V. L., Dawson, T. M., London, E. D., Bredt, D. S., and Snyder, S. H. (1991). Nitric oxide mediates glutamate neurotoxicity in primary cortical culture. Proc. Natl. Acad. Sci. USA88,63686371. de Vente, J., Bol, J. G., and Steinbusch, H. W. (1989). Localization of cGMP in the cerebellum of the adult rat: an immunohistochemical study. Brain Res. 504, 332-337. Desai, K. M., Sessa, W. C., and Vane, J. R. (1991). Involvement of nitric oxide in the reflex relaxation of the stomach to accommodate food or fluid. Nature 357, 477-479. Dwyer, M. A., Bredt, D. S., and Snyder, S. H. (1991). Nitric oxide synthase: irrevesible inhibition by L-N%itroarginine in brain in vivo and in vitro. Biochem. Biophys. Res. Commun. 776, 11361141. Feelisch, M., and Noack, E. A. (1987). Correlation between nitric oxide formation during degradation of organic nitrates and activation of guanylate cyclase. Eur. J. Pharmacol. 739, 19-30. Ferrante, R. j., Kowall, N. W., Beal, M. F., Richardson, E. P., jr., Bird, E. D., and Martin, j. B. (1985). Selective sparing of a class of striatal neurons in Huntington’s disease. Science 230, 561-563. Ferrendelli, J.A., Chang, M. M.,and Kinscherf, D.A. (1974). Elevation of cyclic CMP levels in central nervous system by excitatory and inhibitory amino acids. J. Neurochem. 22, 535-540. Furchgott, R. F., and Zawadzki, J. V. (1980). The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature 288, 373-376. Garg,lJ.C.,and Hassid,A. (1991). Nitricoxidedecreasescytosolic free calcium in BALBlc 3T3 fibroblasts by a cyclic GMP-independent mechanism. J. Biol. Chem. 266, 9-12. Garthwaite, Conference

J. (1991). Proceedings of the Second on Nitric Oxide, in press.

International

Garthwaite, J., Charles, S. L., and Chess-Williams, R. (1988). Endothelium-derived relaxing factor release on activation of NMDA receptors suggests roleas intercellular messenger in the brain. Nature 336, 385-388. Garthwaite, J., Garthwaite, G., Palmer, R. M. J., and Moncada, S. (1989). NMDA receptor activation induces nitric oxide synthesis from arginine in rat brain slices. Eur. J. Pharmacol. 772,413-416. Gibson, A., Mirzazadeh, S., Hobbs, A. J., and Moore, P. K. (1990). L-NC-monomethyl arginine and L-NC-nitro arginine inhibit nonadrenergic, non-cholinergic relaxation of the mouse anococcygeus muscle. Br. J. Pharmacol. 99, 602-606. Gillespie, J. S., Liu, X., and Martin, W. (1989). The effects of L-arginine and NC-monomethyl L-arginine on the response of the rat anococcygeus muscle to NANC nerve stimulation. Br. J. Pharmacol. 98, 1080-1082. Goyal, R. K., Rattan, S., and Said, S. I. (1980). VIP as a possible neurotransmitter of non-cholinergic non-adrenergic inhibitory neurones. Nature 288, 378-380.

Hibbs, J. B., Jr.,Taintor, R. R., and Vavrin, 2. (1987a). Macrophage cytotoxicity: role for L-arginine deiminase and imino nitrogen oxidation to nitrite. Science 235, 473-476. Hibbs, J. B., Jr., Vavrin, Z., and Taintor, R. R. (198715). L-Arginine is required for expression of the activated macrophage effector mechanism causing selective metabolic inhibition in target cells. J. Immunol. 738, 550-565. Hope, B.T., Michael, G. J., Knigge, K. M., and Vincent, S. R. (1991). Neuronal NADPH diaphorase is a nitric oxide synthase. Proc. Natl. Acad. Sci. USA 88, 2811-2814. Ignarro, L. J., Bush, P. A., Buga, G. M., Wood, K. S., Fukuto, J. M., and Rajfer, J. (1990). Nitric oxide and cyclic GMP formation upon electrical field stimulation cause relaxation of corpus cavernosum smooth muscle. Biochem. Biophys. Res. Commun. 770,843850. Ito, M. (1989). Long-term 85-102.

depression.

Annu.

Rev. Neurosci.

12,

lyengar, R., Stuehr, D. J., and Marletta, M. A. (1987). Macrophage synthesis of nitrite, nitrate, and N-nitrosamines: precursors and role of the respiratory burst. Proc. Natl. Acad. Sci. USA 84,63696373. Koh, J.-Y., Peters, S., and Choi, D. W. (1986). Neurons containing NADPH-diaphorase are selectively resistant to quinolinate toxicity. Science 234, 73-76. Li,C.G.,and Rand,M.J.(1990). Nitricoxideandvasoactiveintestinal polypeptide mediate non-adrenergic, non-cholinergic inhibitory transmission to smooth muscle of the rat gastric fundus. Eur. J. Pharmacoi. 797, 303-309. Linden, D. J., and Connor, J. A. (1991). Long-term depression of glutamate currents in cultured cerebellar Purkinje neurons does not require nitric oxide signalling. Eur. J. Neurosci., in press. Marks, C. S., Brien, J. F., Nakatsu, K., and McLaughlin, B. E. (1991). Does carbon monoxide have a physiological function? Trends Pharmacol. Sci. 12, 185-188. Mayer, B., John, M., Heinzel, B., Werner, E. R., Wachter, H., Schultz, G., and Bohme, E. (1991). Brain nitric oxide synthase is a biopterinand flavin-containing multi-functional oxidoreductase. Fed. Eur. Biol. Sot. 288, 187-191. Murad, F., Mittai, C. K.,Arnold, W. P., Katsuki, S., and Kimura, H. (1978). Cuanylate cyclase: activation by azide, nitro compounds, nitric oxide and hydroxyl radical and inhibition by hemoglobin and myoglobin. Adv. Cyclic Nucleotide Res. 9, 145-158. Murphy, S., Minor, R. L., Jr., Welk, G., and Harrison, D. C. (1990). Evidence for an astrocyte-derived vasorelaxing factor with properties similar to nitric oxide. J. Neurochem. 55, 349-351. Murphy, T. I-!., DeLong, M. J., and Coyle, J. T. (1991). Enhanced NAD(P)H: quinone reductase activity prevents glutamate toxicity produced by oxidative stress. J. Neurochem. 56, 990-995. Nakane, M., Ichikawa, M., and Deguchi, T. (1983). Light and electron microscopic demonstration of guanylate cyclase in rat brain. Brain Res. 273, 9-15. Nowicki,J. P., Duval, D., Poignet, H., and Scatton, B. (1991). Nitric oxide mediates neuronal death after focal cerebral ischemia in the mouse. Eur. J. Pharmacol., in press.

Green, L. C., Ruiz-de-Luzuriaga, K., Wagner, D. A., Rand, W., Istfan, N., Young, V. R., and Tannenbaum, S. R. (1981a). Nitrate biosynthesis in man. Proc. Natl. Acad. Sci. USA 78, 7764-7768.

O’Dell,T. j., Hawkins, R. D., Kandel, E. R., and Arancia, 0. (1991). Tests of the roles of two diffusible substances in LTP: evidence for nitric oxide as a possible early retrograde messenger. Proc. Natl. Acad. Sci. USA, in press.

Green, L. C.,Tannenbaum, synthesis in the germfree 58.

Palmer, R. M. J., Ferrige, A. G., and Moncada, S. (1987). Nitric oxide release accounts for the biological activity of endothelium-derived relaxing factor. Nature 327, 524-526.

S. R.,and Goldman, P. (1981 b). Nitrate and conventional rat. Science 272,56-

Haglund, L., Kohler, C., Haaparanta, T., Goldstein, M., and Gustafsson, J.-A. (1984). Presence of NADPH-cytochrome 450 reductase in central catecholaminergic neurones. Nature 307,259-262.

Palmer, R. M. J., Ashton, D. S., and Moncada, endothelial cells synthesize nitric oxide from 333, 664-666.

S. (1988). Vascular L-arginine. Nature

Review: Nitric Oxide 11

Radomski, M. W., Palmer, R. M. j., and Moncada, S. (1990). An L-arginine/nitric oxide pathway present in human platelets regulates aggregation. Proc. Natl. Acad. Sci. USA 87, 5193-5197. Ramagopal, M. V., and Leighton, H. J. (1989). Effects of NGmonomethyl-L-arginine on field stimulation-induced decreases in cytosolic Ca *+ levels and relaxation in the rat anococcygeus muscle. Eur. J. Pharmacol. 774, 297-299. Shibuki, K., and Okada, D. (1991). Endogenous lease required for long-term synaptic depression lum. Nature 349, 326-328. Stuehr, D. synthesis: response Acad. Sci.

nitric oxide rein the cerebel-

J., and Marletta, M. A. (1985). Mammalian nitrate biomouse macrophages produce nitrite and nitrate in to Escherichia co/i lipopolysaccharide. Proc. Natl. USA 82, 7738-7742.

Stuehr, D. J., Kwon, N. S., Nathan, C. F.,and Griffith, 0. W. (1991). Nw-hydroxy-L-arginine is an intermediate in the biosynthesis of nitric oxide from L-arginine. J. Biol. Chem. ,266, 6259-6263. Thomas, E., and Pearse, A. G. E. (1964). The solitary Histochemical demonstration of damage-resistant with a TPN-diaphorase reaction. Acta Neuropathol.

active cells. nerve cells 3, 238-249.

Toda, N., and Okamura, T. (1990). Possible role of nitric oxide in transmitting information from vasodilator nerve to cerebroarterial muscle. Biochem. Biophys. Res. Commun. 770, 31X3-313. Trakshel, C. M., Kutty, R. K., and Maines, M. D. (1988). Resolution of the rat brain heme oxygenase activity: absence of a detectable amount of the inducible form (HO-I). Arch. Biochem. Biophys. 260, 732-739. Tttrup, A., Svane, D., and Forman, ing NANC inhibition in opossum Am. J. Physiol. 260, C385-G389.

A. (1991). Nitric oxide mediatlower esophageal sphincter.

Uemura, Y., Kowall, N. W., and Beal, M. F. (1990). Selective sparing of NADPH-diaphorase-somatostatin-neuropeptide Y neurons in ischemic gerbil striatum. Ann. Neurol. 27, 620-625. Wagner, D.A.,Young,V. R., andTannenbaum, 5. R. (1983). Mammalian nitrate biosynthesis: incorporation of 15NH3 into nitrate is enhanced by endotoxin treatment. Proc. Natl. Acad. Sci. USA 80, 4518-4521. Warner, M., Kohler, C., Hansson, T., and Gustafsson, J.-A. (1988). Regional distribution of cytochrome P-450 in the rat brain: spectral quantitation and contribution of P-450b,e and P450c,d. J. Neurochem. 50, 1057-1065. Yui, Y., Hattori, R., Kosuga, K., Eizawa, H., Hiki, K., and Kawai, C. (1991). Calmodulin-independent nitric oxide synthase from rat polymorphonuclear neutrophils. J. Biol. Chem. 266,12544-12547.

Nitric oxide, a novel neuronal messenger.

Neuron, Vol. 8, 3-11, January, 1992, Copyright 0 1992 by Cell Press Nitric Oxide, a Novel Neuronal Messenger David S. Bredt and Solomon H. Snyder D...
2MB Sizes 0 Downloads 0 Views