Arachidonic Acid

Daniele Piomelli

Professor of Pharmacology; 360 Med Surge II; University of California, Irvine; Irvine, CA 92697-4625.

When a neurotransmitter binds to its receptor on the membrane of a target neuron, it triggers the formation of second messengers, responsible for translating receptor occupation into cellular responses. For example, the binding of dopamine to D1-type receptors stimulates the activity of adenylyl cyclase, which catalyzes the conversion of ATP into cyclic AMP. This second messenger, in turn, binds to and activates a specific protein kinase, protein kinase A, which puts inorganic phosphate on select intracellular proteins. Phosphorylation modifies the biological activity of these proteins and constitutes the basis for many physiological effects of dopamine in the central nervous system (CNS) (see Cholinergic Transduction and Signal Transduction Pathways for Catecholamine Receptors).

This model of transmembrane signaling assumes that the range of action of a second messenger is confined to the intracellular environment. In agreement with this view, most "classical" signaling systems—cyclic AMP, cyclic GMP, Ca2+, inositol trisphosphate, and diacylglycerol—produce their effects by binding to protein receptors located within the cell, whether they be protein kinases, protein phosphatases, Ca2+-binding proteins, or ion channels. Such a model is not likely to account, however, for all known transduction pathways. Examples of more complex scenarios include the arachidonic acid cascade, examined in the present chapter, and nitric oxide, outlined in Nitric Oxide and Related Substance as Neural Messengers).

A schematic picture of the ways in which arachidonic acid and its metabolites may act in regulating neuronal activity is shown in Fig. 1. Arachidonic acid is released from phospholipids in cells stimulated by many first messengers, including neurotransmitters, neuromodulators, and neurohormones. The free fatty acid has, as such, a short lifespan, during which it may interact with and affect the activity of ion channels and protein kinases within the cell. Alternatively, it may be transformed to a family of metabolites—the eicosanoids—which may also produce important effects on intracellular targets. In both cases, the arachidonic acid cascade affects neuronal excitability by fulfilling the primary criteria defining a second messenger system—that is, receptor-dependent formation and intracellular site of action.

Where the eicosanoids differ from "classical" second messengers is in their ability to cross the cell membrane, diffuse through the extracellular space, and interact with high-affinity receptors located on neighboring neurons (Fig. 1). Eicosanoid receptors have been characterized in the brain and have been shown to be linked to second messengers, such as cyclic AMP, very much like the receptors recognized by dopamine, noradrenaline, and so on. Therefore, thanks to the ability to branch at the same time within and without a cell, the arachidonic acid cascade may give rise both to intracellular second messengers and to local mediators, bridging the gap between transmembrane and transcellular communication. This two-pronged role may be important in integrating the responses of postsynaptic neurons with the activity of presynaptic terminals and of other contacting cells.

In resting cells, arachidonic acid is stored within the cell membrane, esterified to glycerol in phospholipids (Fig. 2). A receptor-dependent event, requiring a transducing G protein, initiates phospholipid hydrolysis and releases the fatty acid into the intracellular medium. Three enzymes may mediate this deacylation reaction: phospholipase A2 (PLA2), phospholipase C (PLC), and phospholipase D (PLD), whose different sites of attack on the phospholipid backbone are shown in Fig. 2 (inset). PLA2 catalyzes the hydrolysis of phospholipids at the sn (stereospecific numbering)-2 position. Therefore, this enzyme can release arachidonate in a single-step reaction. By contrast, PLC and PLD do not release free arachidonic acid directly. Rather, they generate lipid products containing arachidonate (diacylglycerol and phosphatidic acid, respectively), which can be released subsequently by diacylglycerol- and monoacylglycerol-lipases (Fig. 2).

Once released, free arachidonate has three possible fates: reincorporation into phospholipids, diffusion outside the cell, and metabolism. Metabolism is carried out by three distinct enzyme pathways expressed in neural cells: cyclooxygenase, lipoxygenases, and cytochrome P450. Several products of these pathways act within neurons to modulate the activities of ion channels, protein kinases, ion pumps, and neurotransmitter uptake systems. The newly formed eicosanoids may also exit the cell of origin and act at a distance, by binding to G-protein-coupled receptors present on nearby neurons or glial cells. Finally, the actions of the eicosanoids may be terminated by diffusion, uptake into phospholipids, or enzymatic degradation.

Neurons can take up preformed arachidonic acid, but they cannot synthesize it ex novo, as other cells do, by elongation and desaturation of dietary linoleic acid. Yet, neuronal lipids are highly enriched in arachidonate, raising the question as to how does the fatty acid get there. The liver is a major source, via the circulation, but two types of cells in the CNS appear also to play an important role: cerebral endothelium and astrocytes. These cells accumulate circulating linoleate, use it to synthesize arachidonic acid, and secrete the latter into the interstitial medium, making it available to neurons (45, 46).

Neurons take up free arachidonic acid and store it rapidly by esterifying it to membrane phospholipids (10, 26). As a result, only trace levels of free arachidonate may be found in resting cells. Such tight control, justified both by the signaling role of this lipid and by its potential toxicity, is exerted by two concerted enzymatic activities, arachidonoyl-coenzyme A (CoA) synthetase and arachidonoyl-CoA:lysophospholipid transferase (note that a lysophospholipid lacks one of the two phospholipid acyl chains).

Arachidonoyl-CoA synthetase catalyzes the ATP- and Mg2+-dependent formation of arachidonoyl-CoA, using fatty acid and reduced CoA as substrates (35, 48, 83, 88). Next, the activated fatty acid is incorporated into lysophospholipid by arachidonoyl-CoA:lysophospholipid transferase (11!popup(ch59ref11)). After ultracentrifugation of brain extracts, both enzymes are found in the particulate fraction, and indirect evidence suggests that they may be organized in a multienzyme complex on the intracellular aspect of the neuronal membrane (77).

Several neuromodulators stimulate the deacylation of phospholipids, causing release of free arachidonate. These include excitatory amino acids (such as glutamate), biogenic amines (such as serotonin and histamine), and peptides (such as bradykinin) (1, 14, 17, 30, 36, 57, 58, 59, 60). Even though the final effect of these various substances on arachidonate turnover is similar, they may use different mechanisms to achieve it. As we have seen above, at least three distinct phospholipases are thought to generate free arachidonic acid, either directly or indirectly: PLA2, PLC, and PLD. Recent studies have shown that all of them may be activated by neurotransmitters.

Julius Axelrod and his colleagues at the National Institutes of Health have used primary cultures of hippocampal neurons to study the effect of serotonin on arachidonic acid release (17). They labeled neuronal phospholipids by prolonged incubation with [3H]arachidonic acid, and then exposed the neurons to serotonin or to drugs acting at select serotonin (5-HT) receptors. They discovered that stimulating the 5-HT2 receptor, a subtype known to be linked to transducing G proteins, resulted in the accumulation of unesterified radioactive fatty acid. Which phospholipase activity mediated this effect? To answer this question, Axelrod and his colleagues examined the ability of serotonin to stimulate the formation of lysophosphatidylcholine, which (as shown in Fig. 2) is produced selectively by PLA2 activity, but not by PLC or PLD. Using a radiolabeled precursor, they found that the quantity of radioactive lysophospholipid in the membrane was increased by serotonin, strongly arguing for a participation of PLA2 in the response (17). These results, and those obtained in several other laboratories using different experimental preparations (30, 64), support the idea that PLA2 may play a widespread role in receptor-dependent release of arachidonic acid. Despite these progresses, important information on the mechanism of activation of PLA2 in neurons is still lacking. For example, most researchers believe that a G protein ensures the coupling of receptors with PLA2. This convinction is based on the ability of pertussis toxin (a Bordetella toxin which inactivates two families of G proteins, Gi and Go) to prevent receptor-stimulated arachidonate release, as well as on the ability of nonhydrolyzable GTP analogues to evoke it (7). The precise identity of the G protein(s) involved remains, however, unknown, because the existing pharmacological tools do not allow us to discriminate among the various members of the Gi and Go families. Likewise, recent findings indicate that multiple PLA2s may be expressed in neurons and in other cells (8, 12, 82, 92). Do these different isoforms couple selectively to different receptors? Or rather, do they serve distinct functions? And if so, which functions? Answering these questions will require the development of new classes of PLA2 inhibitors, more specific and more potent than those available at present. We have seen above that—in addition to PLA2—arachidonic acid release may also proceed from the sequential activation of PLC, diacylglycerol-lipase and monoacylglycerol-lipase. The reactions carried out by these enzymes, which were discovered in the laboratory of Philip Majerus (5), are shown in Fig. 2: PLC cleaves the polar heads of phospholipids, thereby forming diacylglycerol, which is then hydrolyzed to glycerol and free fatty acids by diacylglycerol-and monoacylglycerol-lipases (16). Recently, Pierre Morell and colleagues, at the University of North Carolina, were able to show that, in primary cultures of sensory neurons, bradykinin may evoke arachidonic acid release by activating selectively this enzyme pathway. Neurons obtained from the spinal cord of embryonic rats were labeled by incubation with various radioactive lipid precursors and were exposed to bradykinin. Application of the neuroactive peptide raised the levels of unesterified arachidonate, but had no effect on lysophospholipids, arguing against an involvement of PLA2. By contrast, appearance of the free fatty acid was preceded by a transient increase in diacylglycerol content, likely caused by PLC activation, which took place within a few seconds of exposure to bradykinin. In addition, arachidonate release could be prevented by an inhibitor of diacylglycerol lipase, the compound RG 80267 (1).

In contrast with PLA2 and PLC, participation of PLD in receptor-dependent arachidonic acid release has not been demonstrated yet. However, one of the products of its activity, phosphatidic acid (the other is a phospholipid head-group, such as choline or inositol), is dephosphorylated to diacylglycerol, which, as we have seen, enters the diacylglycerol-lipase pathway yielding free arachidonate (Fig. 2) (15). In addition, the ability of some neurotransmitters to stimulate PLD activity adds further support to the possibility that this lipase may participate in receptor-mediated arachidonate release (15).

Nonhydrolyzable GTP analogues, such as GTP-g-S, have been very useful to determine the role of G proteins in transmembrane transduction. As a rule, their ability to produce a certain response is taken as good evidence for the presence of a G-protein-mediated coupling mechanism. By using GTP analogues, Carol Jelsema and Julius Axelrod have provided the first evidence of an inhibitory control by G proteins over the activity of PLA2. While studying signaling events in retinal photoreceptors, they observed that flashing light on dark-adapted rod outer segments (ROS) enhanced PLA2 activity. However, when the ROS were exposed to light after incubation with GTP-g-S, this increase was significantly smaller. They concluded that an unidentified G protein, which could be activated by the GTP analogue, exerted an inhibitory action on the activity of retinal PLA2 when this enzyme was stimulated by light (27, 28).

Do neurotransmitter receptors link to inhibition of arachidonate release? Experiments carried out on a heterologous expression system, in the laboratory of Jean-Charles Schwartz in Paris, suggest this possibility (78). Chinese hamster ovary (CHO) cells were transfected with a plasmid vector directing expression of histamine H2-type receptor, which is known to be positively coupled to adenylyl cyclase via a Gs protein. CHO cells were no exception to this rule, and the transfected receptor was found to be very effective in evoking cyclic AMP formation when stimulated with an H2 agonist. Unexpectedly, in addition to this response, H2-receptor occupation was also found to reduce the release of arachidonic acid evoked by raising intracellular Ca2+ levels (a stimulus for PLA2). The mechanism underlying this effect has been only partially uncovered, but the evidence collected allows us to draw a few conclusions. First, inhibition of arachidonate release was independent of the rises in cAMP produced by stimulating the H2 receptor, because membrane-permeant cAMP analogues did not mimic the response. Second, inhibition was not secondary to a reduction in Ca2+ entry, because H2-receptor stimulation had no effect on either basal or stimulated Ca2+ levels. The results, therefore, support the possibility that transfected H2 receptors in CHO cells are directly coupled to inhibition of PLA2 activity (78). It remains to be determined whether a similar response occurs in neurons or in other cells expressing this receptor constitutively.

Several structurally different neurotransmitter receptors—including D2-dopaminergic and a2-adrenergic—share the ability to reduce adenylyl cyclase activity and to lower cAMP levels in cells, through the intermediate of an "inhibitory" G protein (Gi). When transfected in CHO cells, receptors of this group produce, in addition, what appears to be a "silent" facilitation of arachidonic acid release. Namely, receptor activation has no effect, per se, on arachidonate release, but, if release is triggered by a second agent—for example, by stimulation of a different receptor or by a Ca2+ ionophore—it will greatly potentiate it (18, 63). This novel form of regulation involves, like adenylyl cyclase inhibition, a Gi protein, as shown by the ability of pertussis toxin to inhibit the response, and of GTP-g-S to mimic it (63).

The three pathways of arachidonic acid metabolism discovered in most animal tissues—lipoxygenases, cyclooxygenase, and cytochrome P450—have been also described in brain (Fig. 3).

Lipoxygenases are a family of enzymes which catalyze the oxygenation of arachidonic acid, each lipoxygenase forming a distinct hydroperoxy-eicosatetraenoic acid (HPETE) (90). HPETEs may undergo a series of metabolic transformations—what is referred to as a lipoxygenase pathway. Here, we will focus our attention on the two lipoxygenases whose presence in the CNS has been best characterized: 12- and 5-lipoxygenase (Fig. 3). 12-Lipoxygenase converts arachidonic acid into 12(S)-HPETE (containing a -OOH group on the chiral carbon 12), which may be further metabolized into four distinct products: an alcohol [12(S)-hydroxy-eicosatetraenoic acid, 12(S)-HETE], a ketone (12-keto-eicosatetraenoic acid, 12-KETE), or two epoxy alcohols (hepoxilin A3 and B3) (54, 55, 60, 61).

The sequence of reactions initiated by 5-lipoxygenase is more complex. To become active, 5-lipoxygenase requires three cofactors: Ca2+, ATP, and an integral membrane protein called five lipoxygenase-activating protein (FLAP). Inactive 5-lipoxygenase binds Ca2+ and ATP and translocates onto the membrane, where it anchors to FLAP. Membrane translocation activates 5-lipoxygenase, which carries out two sequential reactions: First, it converts arachidonic acid into 5(S)-HPETE; second, it dehydrates 5(S)-HPETE to yield the epoxide, leukotriene A4 (LTA4). The newly formed leukotriene leaves the active site of 5-lipoxygenase but, being itself quite short-lived, is rapidly metabolized to form, by hydrolysis, LTB4 (via an LTA4-hydrolase) or, by addition of glutathione, LTC4 (via a glutathione-S-transferase) (68).

Brain 12-lipoxygenase was purified, and a complementary DNA encoding it was cloned (51, 86). Immunohistochemical studies revealed the occurrence of this enzyme in neurons (particularly in hippocampus, striatum, and olivary nucleus), as well as in glial and in cerebral endothelial cells (50). In agreement with these findings, 12-lipoxygenase metabolites are among the most abundant eicosanoids produced by nervous tissue, as first shown by Lidia Sautebin and co-workers, at the University of Milan (69).

5-Lipoxygenase activity in the CNS was demonstrated by Samuelsson and his colleagues at the Karolinska Institut in Stockholm and was confirmed by further studies (32, 39, 68, 75)). Even though several of the products formed have been identified (notably, LTC4 and LTB4), little is known on the distribution in the CNS of 5-lipoxygenase, FLAP, and glutathione-S-transferase. The laboratory of Takao Shimizu in Tokyo has shown that LTA4-hydrolase is expressed in virtually all regions of the brain, suggesting that—in addition to converting LTA4 into LTB4—this enzyme may serve more general functions, possibly unrelated to arachidonic acid metabolism (75).

Cyclooxygenase (prostaglandin G/H synthase) catalyzes the stepwise conversion of arachidonic acid into two short-lived intermediates, prostaglandin G (PGG) and prostaglandin (PGH). The latter is metabolized to PGs, prostacyclin (PGI2), and thromboxane A2 (TXA2) by the activity of specific enzymes: prostaglandin isomerases for the various PGs, prostacyclin synthase for PGI2, and thromboxane synthase for TXA2 (Fig. 3).

Since the pioneering work of Bengt Samuelsson (at the Karolinska Institut) and Leonard Wolfe (at the Hospital for Sick Children in Toronto), three most prominent cyclooxygenase products have been identified in nervous tissue (PGF2a, PGD2, and PGE2), and the enzymes involved in their biosynthesis have been purified and characterized (22, 31, 52, 79, 89). Immunohistochemical studies, carried out primarily by Osamu Hayaishi and his colleagues (80) have established the presence of these enzymes in both neurons and glia. These studies have been supported by experiments demonstrating that primary cultures enriched in either neurons or glial cells have the ability to synthesize prostaglandins (72, 81)).

Cytochrome P450, the microsomal enzyme complex participating in drug metabolism, may also act on endogenous arachidonic acid, catalyzing its conversion into epoxy-eicosatrienoic acids (EETs). The epoxide ring of these EETs may be cleaved by the action of epoxide hydrolases, to yield the corresponding vicinal diols. In addition, cytochrome P450 has been shown to produce a family of HETEs by hydroxylation (monooxygenation) of arachidonic acid (Fig. 3) (41).

Even though mammalian brain tissue contains very low levels of cytochrome P450, several isoforms of this enzyme were detected in both neural and glial cells by immunohistochemistry, and biosynthesis of arachidonate metabolites via the cytochrome P450 pathway has been reported (2, 29, 84).

A novel arachidonic acid derivative was recently isolated from brain and was identified as the ethanolamide of arachidonic acid (Fig. 4). This compound was shown to (a) inhibit the specific binding of a radiolabeled agonist to the cannabinoid receptor and (b) produce inhibition of the twitch response in mouse vas deferens, a typical response to cannabinoids. These properties have led to the suggestion that arachidonoylethanolamide (dubbed "anandamide" after the sanskrit word for bliss, "ananda") may act as the endogenous ligand for brain cannabinoid receptors (13). The pathways leading to the biosynthesis and the degradation of anandamide in the CNS are not known.

The arachidonic acid cascade is arguably the most elaborate signaling system neurobiologists have to deal with. Not only can it generate multiple messenger molecules (at least 16, according to a conservative estimate limited to the brain), but these molecules may act both within and without the neuron, bringing into play intracellular as well as extracellular targets. To help find our way in this complex network, it may be helpful, rather than listing all known neuronal actions of the eicosanoids, to discuss in greater depth a few examples representative of the roles of these lipids as either intracellular second messengers or transcellular mediators.

The potential role of arachidonic acid in mediating K+ channel modulation and presynaptic inhibition of neurotransmitter release—our first example of a second messenger role for the eicosanoids—was first suggested by experiments carried out in the laboratory of James Schwartz at Columbia University, using the simple nervous system of the marine mollusk, Aplysia californica. Aplysia has large, easily identifiable and well-characterized neurons, which can be dissected out individually and maintained in culture for several days. When these neurons were stimulated with the neurotransmitter, histamine, they released both 12- and 5-lipoxygenase products (58). Histamine is known to exert inhibitory actions on identified Aplysia neurons, causing membrane hyperpolarization and reducing neurotransmitter release at specific synapses. This clue led to the idea that arachidonic acid metabolites may be involved in inhibitory responses, and it prompted the study of a neurotransmitter with better-defined electrophysiological effects. FMRF-amide, a neuroactive tetrapeptide, hyperpolarizes Aplysia sensory neurons and inhibits neurotransmitter release at sensory–motor synapses, by increasing the activity of a subclass of K+ channels termed S-K+ channels. To determine whether arachidonic acid metabolites participate in the effects of FMRF-amide, a series of biochemical, pharmacological, and electrophysiological experiments were carried out. First, application of FMRF-amide to sensory cells resulted in the formation of 12- and 5-lipoxygenase metabolites. Next, drugs that inhibit PLA2 and lipoxygenase activities prevented the electrophysiological actions of FMRF-amide on sensory neurons, whereas cyclooxygenase inhibitors had no effect. Finally, applications of arachidonic acid or 12-HPETE mimicked the effects of FMRF-amide on both S-K+ channels and transmitter release. By contrast, 12-HETE, 5-HPETE and 5-HETE had no effect (59).

In subsequent studies, it was shown that the actions of 12-HPETE on S-K+ channel activity required metabolism of the hydroperoxyacid via an enzymatic activity sensitive to cytochrome P450 inhibitors (4). The results support the possibility that a 12-lipoxygenase metabolite, possibly a hepoxilin, acts as a second messenger in mediating the effects of FMRF-amide on K+ channels and neurotransmitter release (Fig. 5).

FMRF-amide is a modulatory neurotransmitter, and its effects on K+ channels occur within seconds of its application and last as long as the application lasts. Repeated administrations of FMRF-amide may result, however, in long-term changes in neuronal excitability. Samuel Schacher and his colleagues at Columbia University found that prolonged exposure of Aplysia sensory neurons to FMRF-amide (2 hr), produced a depression in synaptic transmission that lasted for several days and was accompanied by a significant reduction in the number of varicosities on sensory neuron dendrites. As with the short-term modulation discussed above, this form of long-term depression may be mediated by arachidonic acid. In agreement, the number of varicosities was significantly reduced 24 hr after a 2-hr application of arachidonic acid to sensory neurons. This effect is likely to involve gene expression and protein synthesis, because it could be prevented by protein synthesis inhibitors (43, 44).

Actions of the eicosanoids similar to those described in Aplysia have also been found in mammalian brain. For example, 12-lipoxygenase products were shown to inhibit glutamate release from hippocampal mossy fiber nerve endings (19), whereas 5-lipoxygenase metabolites were found to increase the activity of muscarine-inactivated M-K+ channels in rat hippocampal CA1 neurons (71). K+ channel modulation by lipoxygenase products has also been reported in a number of non-neural cells, including heart myocytes (33, 34).

Phosphorylation of specific proteins in the presynaptic nerve terminal may participate, together with ion channel modulation, in regulating neurotransmitter release. The state of phosphorylation of the synaptic-vesicle-associated protein, synapsin I, is thought to regulate the availability of synaptic vesicles for exocytosis. In its dephosphorylated state, synapsin I may cross-link synaptic vesicles to the surrounding cytoskeletal lattice. According to this model, when synapsin I is phosphorylated on its "tail"-region by Ca2+/calmodulin-dependent protein kinase II, its interaction both with synaptic vesicles and with cytoskeletal elements is reduced, resulting in dissociation of the vesicles from the cytoskeleton. This would, in turn, increase the number of vesicles available for exocytosis. Therefore, reducing the state of phosphorylation of synapsin I may be a way to reduce synaptic strength independent of, and possibly parallel to, ion channel modulation (see Electrophysiology).

Arachidonic acid and its metabolites may regulate neurotransmitter release partly through such a phosphorylation-dependent mechanism. In agreement, experiments carried out in Paul Greengard's laboratory at the Rockefeller University showed that lipoxygenase-derived eicosanoids are potent and selective inhibitors of purified Ca2+/calmodulin-dependent protein kinase II. 12-HPETE inhibited activity of this protein kinase with a half-maximal effect at a concentration of 0.7 mM. By contrast, the eicosanoid has no effect on the activities of protein kinase C, cAMP-dependent protein kinase, Ca2+/calmodulin-dependent protein kinase I and III, or the Ca2+/ calmodulin-activated phosphatase, calcineurin (62).

The effects of the eicosanoids on K+ channels and on Ca2+/calmodulin-dependent protein kinase II may be integrated in a model, shown in Fig. 5, for the role played by these lipids in presynaptic inhibition. Free arachidonic acid, produced as a result of receptor activation, is metabolized by 12-lipoxygenase to form 12-HPETE. The hydroperoxyacid may, on the one hand, modulate the activity of K+ channels and, on the other, inhibit Ca2+/calmodulin-dependent protein kinase II, reducing Ca2+-evoked protein phosphorylation. These two parallel effects might be synergistic in decreasing synaptic strength.

In addition to these actions mediated by the eicosanoids, arachidonic acid and other fatty acids may regulate neuronal excitability directly, by mechanisms that do not involve metabolism or intervention of other second messenger pathways. For example, fatty acids may modify the activity of a variety of ion channels, possibly by interacting with hydrophobic binding sites within the channel protein (like local anesthetic or antiarrhythmic drugs). The interested reader is referred to a recent review on this topic (53). In this section, I will examine instead an additional potential role of unsaturated fatty acids—that of second messengers in the receptor-dependent stimulation of protein kinase C (PKC) activity.

PKC, which was originally described as a Ca2+- and phospholipid-dependent protein kinase activated by diacylglycerol, is now recognized to consist of a family of at several related isoenzymes with different properties and distribution in the brain. Nishizuka and his colleagues (47, 73) have recently shown that low concentrations of unsaturated fatty acids (1–10 mM), which are likely to be attained in stimulated cells, activate with high selectivity a single PKC isoform, type I, in the absence of Ca2+ and phospholipid.

What is the physiological significance of this stimulation, and how does it relate to the well-characterized ability of diacylglycerol to activate PKC? To address these questions, David Linden and collaborators (in John Connor's laboratory at the Roche Institute) have carried out a series of electrophysiological experiments on primary cultures of rat cerebellar neurons (38). In rat cerebellum, type I and type II PKC are segregated: Type I is expressed in Purkinje cells, whereas type II is expressed in granule cells. The application of phorbol esters, which causes a nonselective stimulation of all PKC isoforms, reduced voltage-gated K+ currents equally in both neurons. By contrast, the administration of unsaturated fatty acids affected K+ currents selectively in Purkinje cells, not in granule cells. The findings suggest that, in neurons expressing type I PKC, activation of this protein kinase isoform may result from the receptor-dependent stimulation of PLA2 activity and from the generation of free arachidonic acid (as well as other fatty acids) (38).

The examples discussed thus far illustrate one possible role played by the eicosanoids—that of intracellular second messengers. We will examine now the alter ego of the arachidonic acid cascade: its ability to act as a transcellular (or trans-synaptic) signaling system.

Diffusible signals may exert important neurophysiological functions. Neurons in the CNS are organized as interconnected groups of functionally related cells (e.g., in sensory systems). A diffusible factor released from a neuron into the interstitial fluid, and able to interact with membrane receptors on adjacent cells, would be ideally used to "synchronize" the activity of an ensemble of interconnected neural cells. Furthermore, during development and in certain forms of learning, postsynaptic cells may secrete regulatory factors which diffuse back to the presynaptic component, determining its survival as an active terminal, the amplitude of its sprouting, and its efficacy in secreting neurotransmitters—a phenomenon known as retrograde regulation. The participation of arachidonic acid metabolites in retrograde signaling and in other forms of local modulation of neuronal activity has been proposed.

Long-term potentiation of synaptic transmission (LTP) is a mammalian model of synaptic plasticity and information storage. LTP is believed to consist of two phases: induction and maintenance. Induction is initiated by the postsynaptic entry of Ca2+, which occurs through glutamate N-methyl-D-aspartate (NMDA)-type receptor channels. Maintenance appears to be produced at least partly by presynaptic mechanisms. To bridge postsynaptic induction with presynaptic maintenance, the existence of a diffusible retrograde messenger was proposed (6).

Arachidonic acid was suggested as a potential candidate for this role (59). In agreement, stimulation of glutamate receptors evokes arachidonic acid release from a variety of neural cell preparations (14, 36). In addition, nonselective PLA2 inhibitors (such as p-bromophenacylbromide) prevent induction of LTP, while application of arachidonic acid (or other unsaturated fatty acids) to hippocampal slices causes a slow-onset enhancement of synaptic transmission that resembles LTP (37, 87). The mechanism of action of arachidonic acid in enhancing neurotransmission remains to be established, and several potential targets have been proposed. The fatty acid may increase glutamate release from hippocampal nerve terminals, block glutamate uptake, or potentiate NMDA receptor current (41, 93). Alternatively, it may act by enabling presynaptic glutamate receptors to produce enhanced glutamate release (25).

Orna Harish and Mu-Ming Poo at Columbia University have provided evidence suggesting that a 5-lipoxygenase metabolite of arachidonic acid may act as a retrograde messenger at the developing neuromuscular synapse (21) (Fig. 6). Using primary cultures of innervated muscle cells from Xenopus, they found that injections of GTP-g-S into the myocyte caused an increase in the frequency of spontaneous synaptic currents (SSCs), an indication that acetylcholine release from presynaptic terminals was enhanced. They concluded that a G-protein-driven signal was released from the muscle cell, crossed the synaptic cleft, and acted on the presynaptic neuron to modulate transmitter secretion. To determine the nature of this diffusible signal, they injected drugs that activate cAMP-dependent protein kinase or PKC, but found no effect. However, when they loaded arachidonic acid into the myocyte, a significant increase in SSC frequency occurred, an effect which could be prevented by the selective 5-lipoxygenase inhibitor, AA861. In agreement with an involvement of 5-lipoxygenase-mediated metabolism, the postsynaptic application of 5-HPETE, but not of 12-HPETE, resulted in an increase in the spontaneous synaptic events (21).

After the discovery of the PGs in the CNS, much attention has been given to the roles these eicosanoids may play in modulating neurotransmission, by interacting with presynaptic or with postsynaptic PG receptors. The existence of such receptors is well-demonstrated, and, in the brain, high-affinity binding sites have been described for both PGE2 and PGD2 (65, 66, 74, 85, 91). Peripheral PGE2 receptors are subdivided into three subtypes—EP1, EP2 and EP3—characterized by distinct pharmacological properties and intracellular signaling systems (9). EP1 receptors are coupled to phosphoinositide-specific PLC activity, while EP2 and EP3 receptors stimulate and inhibit, respectively, adenylyl cyclase activity. Recently, a cDNA encoding an EP3-type PGE2 receptor was isolated and characterized, and expression of its mRNA in brain tissue was demonstrated by Northern blot. Sequence analysis of EP3-type receptor cDNA revealed the presence in this molecule of seven putative transmembrane domains, characteristic of G-protein-coupled receptors (76).

Presynaptic PG receptors have been often, but not always, linked to inhibition of neurotransmitter release. For example, PGE2 (as well as its analogue, PGE1) inhibits noradrenaline release in a variety of nervous tissue preparations (20, 23, 24, 67). This inhibitory role is by no means universal, however. For example, in dorsal root ganglion neurons in culture, PGE2 was shown to increase Ca2+ conductance and to stimulate release of substance P (49). Such an effect may be related to the sensitizing and hyperalgesic properties of this PG, and it might mediate the hyperalgesia produced, in the spinal cord, by the stimulation of glutamate and substance P receptors (40).

The receptor-dependent effects produced by lipoxygenase products in brain have been poorly characterized thus far, even though the existence of a high-affinity binding site for the leukotriene, LTC4, was reported (70). Both LTC4 and LTB4 were shown to evoke the rapid release of luteinizing hormone (LH), when applied at picomolar concentrations in primary cultures of anterior pituitary cells (32). Because gonadotropin-releasing hormone was shown to stimulate leukotriene biosynthesis, it is possible that the leukotrienes play a role in LH secretion.

Cerebellar Purkinje neurons display a remarkable response to the iontophoretic administration of LTC4. The leukotriene was found to cause a slowly developing increase in the firing rate of these cells, which could last for up to 1.5 hr after application. The lack of effect of LTB4, along with the ability of a leukotriene-receptor antagonist, the compound FPL 55712, to prevent this response, indicates the selective participation of LTC4 receptors (56). The physiological significance of this intriguing response remains unknown.

published 2000