Nitric Oxide: a Spatial Second Messenger H Schulman

PERSPECTIVE Nitric oxide: a spatial second messenger H Schulman

Department of Neurobiology, Stanford University School of Medicine, Stanford, CA 94305-5401, USA

Fast synaptic transmission by excitatory neurotransmitters endows neural tissue with remark- able computational power but is limited in its ability to adapt with experience. Information in the nervous system can also be distributed beyond the synapse by biogenic amines or neuropeptides released at non-synaptic sites to elicit volume transmission and by the spatial second messenger nitric oxide which can diffuse from a synapse to a volume of surrounding neural tissue. Volume transmission and spatial signaling complement classic point-to-point transmission and may coordinate local activity among groups of neurons and glia. An interest- ing dialogue exists between these systems. Nitric oxide produced in response to excitation of a neuron by glutamate stimulates exocytosis and release of glutamate and dopamine from nearby neurons. Dopamine in turn can alter the strength of glutamatergic synapses and affect synaptic plasticity. Keywords: nitric oxide; volume transmission; long-term potentiation (LTP); calcium; exocytosis; synaptic vesicles; dopamine

The view of the nervous system as hard wired, with course, and broad physiological actions may coordi- information relayed quickly from synapse to synapse nate local activity among groups of neurons and comp- via chemical messengers, inspired representation of lement classic point-to-point transmission. the brain as a powerful computer. While this metaphor Historically, an early concept of the brain as a holis- is still useful, realistic models of brain function must tic organ shifted to a focus on individual neurons cap- accommodate a wet brain in which some informational able of fast synaptic transmission. The view of brain molecules are not restricted to their synapse of origin function has changed again in recent years, in part due but diffuse within a limited volume of neural tissue. to the acknowledgement that neuromodulators have This is exemplified by biogenic amines such as dopa- anatomically widespread effects on the functional state mine and serotonin found in neurons with widespread and perhaps even the wiring of neuronal networks. The and diffuse projections and released en passsant from debate between Ramon y Cajal and Camillo Golgi that non-synaptic varicosities, giving rise to what has been ended a century ago with the formulation of the neuron termed volume transmission.1 Such neuromodulators doctrine established the anatomical basis but not the diffuse through extracellular fluid and act on receptors functional basis for the working of the nervous system.7 at multiple sites rather than being restricted to the syn- Golgi believed that a holistic function could be achie- aptic cleft. An additional class of informational mol- ved by a protoplasmic reticulum in which branches of ecule, acting over a volume of neural tissue that cell bodies were continuous with each other. Cajal includes many synapses, is nitric oxide (NO).2 I use the demonstrated that neurons were individual units with term NO in a general sense, without specifying whether bounded structures that don’t share their cytoplasm the nitrogen monoxide is the free radical NO· or NO+. and which must therefore transmit information via NO is small and sufficiently hydrophobic to readily contacts that we now call synpases. This neuron doc- diffuse through cytoplasm, membrane, and extracellu- trine led to the demonstration of point-to-point fast lar fluid alike. It can therefore act intracellularly or chemical transmission that is the tenet of modern neu- intercellularly on other neurons, glia, and vasculature robiology. It was only later appreciated that the compu- within a volume delimited by its half-life and dif- tational ability of the brain required functional units fusion. More recent studies suggest that NO can stimu- composed of relatively large networks of neurons that late release of the excitatory neurotransmitter gluta- communicate primarily via fast chemical transmission. mate, the neuromodulatory biogenic amines, and The concept of functional units was extended by the neuropeptides, providing for a cascade of release notion of volume transmission by neuromodulatory events that further extends its reach.3–6 Such neurotransmitters and neuropeptides1 and more recently to transmission with its slow onset, relatively long time NO.2,8 Is NO a neurotransmitter, neuromodulator, or second messenger? Although NO is often described as an Correspondence: H Schulman, Department of Neurobiology, Stanford University School of Medicine, Stanford, CA 94305- atypical neurotransmitter, it seems most appropriate to 5401, USA. E-mail: schulmanȰcmgm.stanford.edu place NO in the category of a second messenger. Neuro- Received and accepted 30 August 1996 transmitters and neuromodulators are synthesized and Nitric oxide: a spatial second messenger H Schulman 297 stored in advance of need and released by exocytosis. been implicated in synaptic plasticity underlying By contrast, NO, like most second messengers, is syn- memory, synaptogenesis, the link between activity and thesized from an abundant but inert precursor (from l- blood flow, and in toxicity associated with ischemia arginine). NO does differ from other second messengers and oxidative stress. By 1992 NO was heralded by 2+ like IP3 or Ca in the ability to act intercellularly as Science as the molecule of the year. well as intracellularly and may therefore be regarded I will use LTP as an example to show how NO can as a spatial second messenger.2 NO action is limited by function as a spatial messenger through both its direct its reactivity, particularly with superoxide anion, and actions as well as secondarily via release of other sig- it can be scavenged by the heme moiety in circulating naling molecules. LTP may be a cellular counterpart hemoglobin. Depending on the redox state, NO (in the of associative learning because synaptic transmission form of NO+) can avoid such inactivation by reacting between neurons that are temporally coactive during with cysteine residues on proteins or with smaller induction of LTP are associated by the resulting thiol-containing compounds to form S-nitrosothiols.9 increase in synaptic strength. In the hippocampus, for The NO in nitrosothiols has a much longer half-life example, a potentiation of neurotransmission can be than NO·. Nitrosothiols can serve as effective inter- measured for hours or days if activity in presynaptic cellular carriers of NO since they can subsequently inputs from CA3 to the neurons in CA1 release gluta- release NO to act on its targets. Synthesis of NO is mate at the same time that the postsynaptic CA1 neu- tightly regulated. Some forms of NO synthase (NOS) rons are depolarized. For many, the synapse specificity are constitutively active but present in low amounts of this potentiation makes it difficult to conceive of a until a stimulus induces an increase in transcription of role for NO, and indeed many still do not accept NO its mRNA and synthesis of the protein. The activity of as a mediator of LTP on both an experimental and another form of NOS, termed the neuronal isoform, is theoretical basis. For others, however, a spatial mess- completely dependent on a rise in Ca2+ for its activity. enger such as NO was born of necessity.8 A retrograde The enzyme is present but inactive and produces NO messenger, ie one that goes from postsynaptic CA1 to on demand when a cell stimulus produces a sufficient presynaptic CA3 neurons, was needed to explain how increase in the Ca2+/calmodulin complex to activate the presynaptic terminals could tell that Ca2+ was elev- the enzyme. ated in postsynaptic spines of the CA1 neurons. The The biological action of NO was first recognized as NO hypothesis argued that postsynaptic activity and an endothelial-derived relaxing factor (EDRF), a pri- the rise in Ca2+ would generate NO which would dif- mary determinant of blood pressure.10 Acetylcholine fuse in a volume of tissue and change synaptic efficacy stimulates synthesis of NO by the endothelium from presynaptically or postsynaptically.12 Subsequent which it diffuses to nearby smooth muscle leading to pharmacological data convincingly demonstrated that vasodilatation. Similar action on smooth muscle selective inhibition of NOS with arginine analogues enables it to regulate the nonadrenergic-noncholinergic blocked LTP and that addition of the NOS substrate, l- control of peristalsis by the gut. Peripheral actions of arginine, overcame the block.13 Such an action for NO NO also include the killing of tumor cells and bacteria outside the synapse where it is synthesized is sup- by macrophages. The circuitry of the brain seemed too ported by the finding that hemoglobin, which scav- precise and elegant to utilize such a ‘promiscuous’ enges extracellular NO, attenuated LTP. Furthermore, agent. Granted, NO could serve to couple neuronal synapses in which the presynaptic neurons are even activity with blood flow through an effect on the weakly active became potentiated if the activity was microcirculation, but could it possibly have the neces- coincident with exposure to bath-applied NO.14 NO sary finesse to participate in information flow in the would not act indiscriminately on all nearby synapses. brain? Regardless of the spread from its postsynaptic origin, The realization that NO was an important player in only the presynaptic terminals that are active and the CNS began with the finding that activation of exposed to a sufficient concentration of NO would be NMDA receptors and resultant Ca2+ entry trigger NO potentiated. synthesis which then stimulates cGMP production in The finding that the basic unit of synaptic change in neurons.11 These authors realized that NO could func- LTP may involve more than one synapse further impli- tion as an intercellular messenger and affect presynap- cates a spatial messenger. The induction of LTP at one tic terminals and glial cells, but there was no formal synapse was found to be communicated to synapses on appreciation of how NO actions would be sufficiently nearby cells which also became potentiated.15,16 restricted spatially and temporally to function in Potentiation was distributed to nearby terminals that neurotransmission. The NO hypothesis presented by were coactive but not when NO synthesis was blocked Gally, Edelman and colleagues offered a way out of the in the donor neuron.16 NO therefore acts as a spatial dilemma.12 Neurons could use NO because of its ability messenger to synchronize the spread of potentiation to diffuse rather than despite it. They proposed that and may extend the time window for coactivity from activity-dependent synthesis and diffusion of NO milliseconds to hundreds of milliseconds or seconds. could be important for long-term potentiation (LTP) in Proponents of NO in LTP will find little solace from hippocampal CA1 neurons, control of cerebral blood the first set of NOS knockouts, however. Targeted gene flow to areas of active neurons, and the refinement of disruption of neither the neuronal (activatable) nor the synaptic contacts in late development. NO has now epithelial (inducible) forms of NOS isoforms block Nitric oxide: a spatial second messenger H Schulman 298 LTP.17,18 The effects of the double knockout are anxi- these recombinant proteins with NO in fact increases ously awaited. formation of the VAMP 2/SNAP-25/syntaxin 1a hetero- NO may also communicate the fact of ongoing trimer and decreases binding of n-sec-1 to syntaxin activity by regulating vesicular release in a volume of 1a.23 The biochemical modification is likely to be S- neural tissue surrounding the active synapses. nitrosylation mediated by NO+. Since neuronal NO is Enhanced release of neurotransmitters (eg dopamine produced primarily in an activity-dependent manner, and norepinephrine) and neuropeptides (eg corticotro- fusion of a variety of vesicles including those storing pin releasing factor) in response to NO has been docu- biogenic amines, glutamate, neurotrophic factors, and mented.3,5,6,19 In some studies application of NO is other neuropeptides could be synchronized with alone sufficient to elicit release while in other studies activity. NO is implicated because inhibitors of NOS are found One scenario for activity-dependent coordination at to block release. For example, stimulation of norepi- multiple sites via NO is shown schematically for the nephrine and glutamate release from cortical synapto- CA1 region of hippocampus in Figure 1. When ter- somes by addition of NMDA is blocked by inhibitors minal A releases glutamate on target synaptic spine B of NOS and by scavenging of NO.20 Exogenous appli- and the net input to this neuron depolarizes it suf- cation of NO did not obviate the need for NMDA, sug- ficiently, there is a rise in Ca2+ in spine B and LTP is gesting that an additional signal induced by NMDA induced. The amount of excitatory synaptic activity receptor activation is required for release. In striatal and postsynaptic depolarization in a given area would slices application of NO clearly increased dopamine thereby be temporally correlated with NO synthesis. release without involvement of NMDA receptors.3 During synaptic activity and for a second or so after- Evoked release occurs without NO via the partici- wards NO would diffuse within a local volume of neu- pation of numerous proteins in docking of synaptic ral tissue to elicit both homosynaptic and heterosynap- vesicles at release sites and in Ca2+-dependent fusion tic effects. NO will amplify the initial signal (glutamate and release.21,22 However, in some neurons the role of release from terminal A) by stimulating further release Ca2+ may be to stimulate synthesis of NO which then of glutamate from this terminal (homosynaptic) which somehow triggers release. In PC12 cells, a pheochromo- would lead to more NO synthesis by spine B in a feed- cytoma-derived cell line, a significant fraction of the forward manner. NO would also stimulate release of effect of membrane depolarization on release of acetyl- glutamate from proximal terminal C (heterosynaptic) choline and dopamine was unexpectedly found to be but not distal terminal D. NO-stimulated exocytosis blocked by inhibitors of NO.4 Ca2+ may therefore stimu- from terminal C may occur regardless of whether it is late release both directly, by conventional exocytosis, coactive with terminal A, but the threshold for LTP at and via a pathway that utilizes NO. its synapse may not be reached and potentiation may An examination of exocytosis from hippocampal not occur unless it is coactive, thus limiting the spread synaptosomes suggests that NO acts on the conven- of potentiation.16 This feedforward amplification may tional release machinery.5 The study monitored explain part of the requirement for NO in LTP, ie to exocytosis from synaptic vesicles stained with a fluor- facilitate a threshold level of glutamate release at ter- escent dye as well as release of labeled glutamate. With minals A and C to induce LTP. When a high local con- both methods, exocytosis stimulated by NO and by centration of NO is achieved, NO may also initiate a membrane depolarization was robust and occurred in long-term modification of processes that directly a similar time course. The surprising finding was that added NO stimulated release without a rise in Ca2+.It is quite possible that release of dopamine from PC12 cells and striatal slices can occur by a similar route. Since the half-life of NO is far longer than typical neuronal Ca2+ transients, this mechanism offers a route by which local synaptic activity produces NO that affects release of glutamate from nearby synapses and dopamine from other nearby release sites even after Ca2+ levels have returned to baseline. The NO-mediated pathway requires the same synaptic molecules shown to be essential for conventional Ca2+-dependent exocytosis.23 Selective proteolytic cleavage of three critical synaptic proteins VAMP, syntaxin, and SNAP-25 by botulinum neurotoxin F, C, and A, respectively, similarly affected exocytosis stimulated by either membrane depolarization or NO. Bio- chemical and genetic studies have defined certain pro- Figure 1 Synaptic activity produces a rise in NO (shaded tein–protein interactions which are necessary (eg a sphere) that is limited in time and space which can effect VAMP/SNAP-25/syntaxin complex) and others whose release of glutamate and other neurotransmitters and neuro- interactions must be decreased (eg n-sec-1/syntaxin 1a) peptides from both the active synapse (A) and from nearby in order to promote release.21,22 Direct treatment of active or inactive synapses (C) or varicosities (E). Nitric oxide: a spatial second messenger H Schulman 299 underlie either a presynaptic or postsynaptic References expression of LTP. For example, it may act presynaptically to increase probability of evoked release or post- 1 Fuxe K, Agnati L. Volume Transmission in the Brain: Novel Mech- synaptically to stimulate fusion of a reserve pool of anisms for Neural Transmission. Raven Press: New York, 1991, 602 pp. vesicles containing AMPA receptors and thereby 2 Edelman GM, Gally JA. Nitric oxide: linking space and time in the increase postsynaptic sensitivity to glutamate. brain. Proc Natl Acad Sci USA 1992; 89: 11651–11652. Some NO may also be able to directly stimulate 3 Hanbauer I, Wink D, Osawa Y, Edelman GM, Gally JA. Role of release of neuromodulators, such as dopamine, from nitric oxide in NMDA-evoked release of [3H]-dopamine from striatal slices. NeuroReport 1992; 3: 409–412. nearby sites, eg from diffuse projections of dopami- 4 Hirsch DB, Steiner JP, Dawson TM, Mammen A, Hayek E, Snyder nergic neurons originating from midbrain and ventral SH. Neurotransmitter release regulated by nitric oxide in PC-12 tegmental areas. This would provide for an activity- cells and brain synaptosomes. Curr Biol 1993; 3: 749–754. dependent coupling between activation of NMDA 5 Meffert MK, Premack BA, Schulman H. Nitric oxide stimulates Ca2+-independent synaptic vesicle release. Neuron 1994; 12: receptors on spine B and release of dopamine (and per- 1235–1244. haps neuropeptides) at heterologous sites such as var- 6 Raber J, Koob GF, Bloom FE. Interleukin-2 (IL-2) induces cortico- icosity E on a nearby dopaminergic neuron. Dopamine tropin-releasing factor (CRF) release from the amygdala and may, in turn, be essential for the persistent or late involves a nitric oxide-mediated signaling: comparison with the phase of LTP. It is known that dopamine antagonists hypothalamic response. J Pharmacol Exp Ther 1995; 272: 815–824. 7 Shepherd GM. Foundations of the Neuron Doctrine. Oxford Uni- block expression of the persistent phase of LTP and versity Press: New York, 1991, 338 pp. that application of dopamine elicits a persistent 8 Schuman EM, Madison DV. Nitric oxide and synaptic function. potentiation.24 Finally, if there is sufficient local Annu Rev Neurosci 1994; 17: 153–183. activity, NO and bioactive substances released by NO 9 Stamler JS. Redox signaling: nitrosylation and related target interactions of nitric oxide. Cell 1994; 78: 931–936. would couple neuronal activity with oxygen supply by 10 Furchgott RF, Zawadski JV. The obligatory role of the endothelial increasing local blood flow and with uptake and cells in the relaxation of arterial smooth muscle by acetylcholine. metabolism of neurotransmitters by signaling to nearby Nature 1980; 288: 373–376. glial cells (not shown). 11 Garthwaite J, Charles SL, Chess-Williams R. Endothelium-derived Even if the functional effect of NO beyond the syn- relaxing factor release on activation of NMDA receptors suggests role as intercellular messenger in the brain. Nature 1988; 336: apse is limited to neighboring coactive synapses it is 385–388. difficult to accept that what is good for the gut is good 12 Gally JA, Montague PR, Reeke GN, Edelman GM. The NO hypoth- for the brain. However, nature is generally parsimoni- esis: possible effects of a short-lived, rapidly diffusible signal in ous and applies good solutions for biological problems the development and function of the nervous system. Proc Natl Acad Sci USA 1990; 87: 3547–3551. in different contexts. It is unreasonable to expect that 13 Schuman EM, Madison DV. A requirement for the intercellular NO action could somehow be restricted to point-to- messenger nitric oxide in long-term potentiation. Science 1991; point information transfer when used in brain while 254: 1503–1506. serving as a spatial messenger in the periphery. The 14 Zhuo M, Small SA, Kandel ER, Hawkins RD. Nitric oxide and car- bon monoxide produce activity-dependent long-term synaptic function of the cortex and hippocampus may, in fact, enhancement in hippocampus. Science 1993; 260: 1946–1950. require that the basic unit of plasticity not be restricted 15 Bonhoeffer T, Staiger V, Aertsen A. Synaptic plasticity in rat hippo- to changes at one synapse at a time. Whatever loss campal slice cultures: local ‘Hebbian’ conjunction of pre- and post- would be incurred in the capacity of information stor- synaptic stimulation leads to distributed synaptic enhancement. age, due to a distribution of synaptic changes beyond Proc Natl Acad Sci USA 1989; 86: 8113–8117. 16 Schuman EM, Madison DV. Locally distributed synaptic potenti- the initial site, could be made up by an enrichment of ation in the hippocampus. Science 1994; 263: 532–536. signaling that now incorporates neuromodulators and 17 Huang PL, Dawson TM, Bredt DS, Snyder SH, Fishman MC. Tar- which coordinates changes at a limited set of synapses. geted disruption of the neuronal nitric oxide synthase gene. Cell The effect of cognitive functions such as memory on 1993; 75: 1273–1286. 18 O’Dell TJ, Huang PL, Dawson TM, Dinerman JL, Snyder SH, Kandel emotional states and reward systems may rely on spa- ER, Fishman MC. Endothelial NOS and the blockade of LTP by tial signaling and volume transmission. The dialogue NOS inhibitors in mice lacking neuronal NOS. Science 1994; 265: can go in both directions, of course. Fast synaptic 542–546. transmission may mediate or modulate release of bio- 19 Zhu XZ, Luo LG. Effect of nitroprusside (nitric oxide) on endogen- genic amines as described above while the biogenic ous dopamine release from rat striatal slices. J Neurochem 1992; 59: 932–935. amines are known to regulate fast synaptic trans- 20 Montague PR, Gancayco CD, Winn MJ, Marchase RB, Friedlander mission by modulating the excitability of neurons. I MJ. Role of NO production in NMDA receptor-mediated neuro- somehow doubt that either Cajal or Golgi would be dis- transmitter release in cerebral cortex. Science 1994; 263: 973–977. pleased to learn that some information flow may occur 21 Calakos N, Scheller RH. Synaptic vesicle biogenesis, docking, and fusion: a molecular description. Physiol Rev 1995; 76: 1–29. among groups of synapses that are neither in direct 22 Sudhof TC. The synaptic vesicle cycle: a cascade of protein–protein synaptic contact nor are part of the same anatomical interactions. Nature 1995; 375: 645–653. reticulum. A more accurate description of what I have 23 Meffert MK, Calakos NC, Scheller RH, Schulman H. Nitric oxide detailed above will likely come from a better molecular modulates synaptic vesicle docking/fusion reactions. Neuron 1996; 16: 1229–1236. understanding of NO action and an appreciation of the 24 Huang YY, Kandel ER. D1/D5 receptor agonists induce a protein dialogue between specific neurotransmitter and neuro- synthesis-dependent late potentiation in the CA1 region of the hip- modulator systems. pocampus. Proc Natl Acad Sci USA 1995; 92: 2446–2450. Nitric oxide: a spatial second messenger H Schulman 300