Modification by Arachidonic Acid of Extracellular Adenosine Metabolism and Neuromodulatory Action in the Rat Hippocampus*

Adenosine and arachidonate (AA) fulfil opposite mod-ulatory roles, arachidonate facilitating and adenosine inhibiting cellular responses. To understand if there is an inter-play between these two neuromodulatory systems, we investigated the effect of AA on extracellular adenosine metabolism in hippocampal nerve terminals. AA (30 m M ) facilitated by 67% adenosine evoked release and by 45% ATP evoked release. These effects were not significantly modified upon blockade of lipooxygenase or cyclooxygenase and were attenuated (52–61%) by the protein kinase C inhibitor, chelerythrine (6 m M ). The ecto-5 * -nucleotidase inhibitor, a , b -methylene ADP (100 m M ), caused a larger inhibition (54%) of adenosine re- lease in the presence of AA (30 m M ) compared with control (37% inhibition) indicating that the AA-induced extracellular adenosine accumulation is mostly originated from an increased release and extracellular catabolism of ATP. This AA-induced extracellular adenosine accumulation is further potentiated by an AA-induced decrease (48%) of adenosine transporters capacity. AA (30 m M ) increased by 36–42% the tonic inhibition by endogenous extracellular adenosine of adenosine A 1 receptors in the modulation of acetylcholine release and of CA1 hippocampal synaptic transmission in hippocampal slices. These as a possible loop to facilitation of Spector The fractional release expressed in terms of the percentage of total radioactivity in the tissue at the time of sample collection. The release of tritium evoked by each period of electrical stimulation, i.e. the evoked release (expressed as fractional release), was calculated by integration of the area of the peak upon subtraction of the estimated basal tritium outflow from the total outflow due to electrical stimulation. The evoked release of tritium under these conditions is Ca 2 1 - and tetrodotoxin-sensitive and is mostly constituted by [ 3 H]acetylcholine (17, 18). The effect of drugs on the release of tritium was expressed by alterations of the ratio between the evoked release due to the second stimulation period and the evoked release due to the first stimulation period (S 2 /S 1 ratio). Biochemical Analysis— For the determination of the energy charge, the synaptosomes were incubated without or with AA, PLA 2 , or melittin for 30 min, homogenized by sonication, and adenine nucleotides were extracted and analyzed by HPLC (17, 18). To evaluate cellular integrity, the synaptosomes were placed in the superfusion chambers and super- fused for 30 min with gassed Krebs without or with AA, PLA 2 , or melittin in a close-circuit manner. Cellular disruption was determined by comparing the lactate dehydrogenase activity in the superfusion medium with that found in the synaptosomes upon their solubilization, by sonication, with 2% (v/v) Triton X-100 (2). To study the metabolism of AA in hippocampal synaptosomes, [ 3 H]AA (5 m M , 21.8 m Ci/ml) was incubated for 20 min at 37 °C with hippocampal synaptosomes (0.36– 0.42 mg protein) in a final volume of 200 m l, in the absence or presence of indomethacin (20 m M ) and/or NDGA (50 m M ). The samples were then diluted to a concentration of 80% (v/v) ethanol, vortexed, and kept on

Whenever neurotransmitter release occurs, there is also a release of adenosine as such and of ATP (1), which is extracellularly catabolized into adenosine (2,3). Increased neuronal activity also causes an increase in extracellular arachidonic acid (AA) 1 (4), a free fatty acid that is found sterified in the sn-2 position of membrane phospholipids and can potentially be released by a number of phospholipases, in particular by phospholipases A 2 (5). Adenosine and AA fulfil opposite roles in regulating cell to cell communication in the nervous system. For instance, in the hippocampus, AA facilitates while adenosine A 1 receptor activation inhibits intracellular free calcium accumulation and the release of excitatory neurotransmitters (6 -8); AA facilitates while adenosine inhibits synaptic transmission and plasticity (4,9); AA has been proposed to be a mediator of epileptogenic phenomena (10) whereas adenosine is considered an endogenous anticonvulsant (11); AA potentiates hypoxic damage (12), while adenosine is an endogenous neuroprotector (13). However, despite this parallel between the genesis and opposite actions of adenosine and AA, it is not known whether there is an inter-play between these two neuromodulatory systems.
We now report that AA causes an increase in the extracellular accumulation of adenosine upon stimulation of nerve terminals from the rat hippocampus. This AA-induced increase of extracellular adenosine seems to be partly protein kinase C-mediated and to be due to a combination of an AA-induced facilitation of ATP release and an increased extracellular formation of adenosine from released ATP together with an AAinduced inhibition of extracellular adenosine removal. This AA-induced increase in endogenous extracellular adenosine is likely to be the basis of the greater tonic facilitatory effects of adenosine A 1 receptor antagonists on acetylcholine release and synaptic transmission caused by AA.

H]adenosine and [ 3 H]acetylcholine release as well as on synaptic transmission.
Release of Adenosine from Hippocampal Synaptosomes-Hippocampal synaptosomes from 6-week-old male Wistar rats (140 g) were prepared by differential sucrose-Percoll density gradient centrifugations, as described previously (2). The synaptosomes were resuspended in oxygenated Krebs solution of the following composition (mM): NaCl 124, KCl 3, NaH 2 PO 4 1.25, MgSO 4 1, CaCl 2 2, NaHCO 3 26, glucose 10, pH 7.4, and gassed with a 95% O 2 and 5% CO 2 mixture. This synaptosomal suspension was equilibrated at 37°C for 10 min, loaded with [2,3,8Ј-3 H]adenosine (3 Ci/ml, 0.05 M) for 20 min at 37°C and then layered over Whatman GF/C filters into four parallel 90-l superfusion chambers (adapted from Swinny filter holders, Millipore, Bedford, MA) through the aid of a roller pump (flow rate of 0.6 ml/min, kept constant throughout the experiment). The chamber volume plus dead volume was approximately 0.6 ml. A series of four parallel superfusion chambers was used to enable both control and test conditions to be performed in duplicate from the same batch of synaptosomes. After setting up the synaptosomes, a 40-min washout period was performed before starting sample collection. The effluent was then collected (release period) in 3-min fractions. To quantify tritium release, 500 l of the each of the effluent samples were analyzed by scintillation counting, after addition of 5 ml of scintillation liquid (Scintran T, Wallac, Turku, Finland). In some experiments, [ 3 H]adenosine in the effluent samples was separated. This procedure consisted in the lyophilization of 1.2 ml of the effluent, resuspension in a 70% (v/v) methanol solution to extract adenosine, HPLC separation of adenosine, and collection of the HPLC eluent corresponding to the adenosine peak that was scintillation counted to determine the amount of The synaptosomes were stimulated for 2 min with 20 mM K ϩ (isomolar substitution of Na ϩ by K ϩ in the superfusion Krebs solution) at 21 min after starting sample collection. In some experiments, veratridine stimulation (10 M for 2 min) was used instead of K ϩ stimulation. Tested drugs, applied through the superfusion Krebs solution, were added 9 min before K ϩ stimulation, and removed 13 min after the offset of stimulation. When testing the modifications of the effect of a drug by a modifier, this modifier was applied 15 min before starting sample collection and was present throughout the experiment. At the end of the experiments, the filters were removed from the superfusion chambers and analyzed by scintillation counting for determination of tritium retained by the synaptosomes. Radioactivity was expressed in terms of disintegrations per second per mg of protein (Bq/mg) in each chamber. The fractional release was expressed in terms of the percentage of total radioactivity present in the preparation at the beginning of the collection of each sample. The amount of radioactivity (expressed as fractional release) released by K ϩ stimulation was calculated by integration of the area of the peak upon subtraction of the estimated basal tritium release from the total release of tritium obtained upon K ϩ stimulation. The basal release was assumed to decline linearly from 2 min before onset of stimulation to the 8th min after onset of stimulation.
Release of ATP from Hippocampal Synaptosomes-Hippocampal synaptosomes were resuspended and equilibrated at 37°C in 2 ml of Krebs/HEPES solution of the following composition (mM): NaCl 124, KCl 3, NaH 2 PO 4 1.25, MgSO 4 1, CaCl 2 2, HEPES 26, glucose 10, pH 7.4. Aliquots of 230 l of synaptosomes (0.52-0.61 mg of protein) were placed in a reaction tube to which 50 l of luciferin-luciferase solution (FL-AAM from Sigma dissolved in 5 ml of water, kept at 4°C) and 10 l of Krebs/HEPES containing the test drug were added. The mixture was placed in a luminometer (Wallac 1250 from LKB Turku, Finland) and the electrical signal generated by the photomultiplier recorded. After obtaining a stable baseline, the photomultiplier entry was closed, 10 l of a Krebs/HEPES solution with concentrated KCl (to attain a final concentration of 20 mM) were added, the mixture resuspended, the photomultiplier entry was opened again, and the variation in signal recorded between 15 and 30 s was used to estimate the evoked release of ATP. The calibration curve for ATP was linear between 5 and 1000 pM (see Ref. 1).
Extracellular Catabolism of ATP and AMP in Hippocampal Synaptosomes-Time course kinetic studies of the extracellular catabolism of adenine nucleotides were performed as described previously (2,3). Briefly, the hippocampal synaptosomes were resuspended in 1 ml of Krebs/HEPES solution and aliquots of 100 l were added to incubation vials with 400 l of Krebs/HEPES solution, to which AA (30 M) and/or the protein kinase inhibitor chelerythrine (6 M) were added, and kept at 37°C. After 10 min incubation, the initial substrate, i.e. ATP (30 M) or AMP (10 M), was added. The kinetic protocols consisted of a 5-or 10-min incubation period with sample collection (50 l) at 0, 1, 2.5, 5, 7.5, and 10 min for AMP and at 0, 0.5, 1, 2, 3, and 5 min for ATP. The zero time was defined as the sample collected immediately after (approximately 2-5 s) addition of the initial substrate. Each collected sample was centrifuged (14,000 ϫ g for 10 s) and the supernatant (40 l) was ice-stored for HPLC analysis (2,3). After the 5-or 10-min incubation, the synaptosomes were pelleted by centrifugation (14,000 ϫ g for 10 s). The pelleted synaptosomes were homogenized in 200 l of 2% (v/v) Triton X-100 to determine total lactate dehydrogenase activity and protein content (14). The remaining bathing solution was used to quantify lactate dehydrogenase activity, an index of cellular disruption (2), which was always lower than 3% of total lactate dehydrogenase activity in the synaptosomes.
Transport of Adenosine in Hippocampal Synaptosomes-The synaptosomes were resuspended in 1 ml of Krebs/HEPES solution and equilibrated at 37°C. All adenosine transport assays were conducted at 37°C in a total volume of 300 l, containing ϳ0.15 mg of protein, basically as described by Gu et al. (15), with minor modifications. Transport was initiated by addition of 0.25-2 M [2-3 H]adenosine, which was added at least 10 min after exposure of synaptosomes to tested drugs, and incubations were for 15 s. Transport was terminated by addition of 5 ml of ice-cold inhibitor/stop mixture (100 M dipyridamole and 1 mM adenosine in Krebs/HEPES solution), followed by filtration through Whatman GF/C filters and washing of the reaction tube and filter with 5 ml of ice-cold inhibitor/stop mixture. The filters were then analyzed by scintillation counting for determination of tritium retained by the synaptosomes after addition of 5 ml of scintillation mixture. Adenosine transport was calculated as the difference between total adenosine incorporation into synaptosomes and the nonspecific Electrophysiological Recordings of Hippocampal Synaptic Transmission-One 400-m hippocampal slice, obtained as described previously (3), was transferred to a 1-ml recording chamber for submerged slices and continuously superfused with gassed Krebs solution, kept at 30°C, at a flow rate of 3 ml/min. Drugs were added to this superfusion solution. Electrophysiological recordings of field excitatory post-synaptic potentials (fEPSP) were obtained as described previously (3). Stimulation (rectangular pulses of 0.1 ms applied once every 15 s) was delivered through a bipolar concentric electrode placed on the Schaffer fibers, in the stratum radiatum near the CA3/CA1 border. Orthodromically evoked fEPSPs were recorded through an extracellular microelectrode (4 M NaCl, 2-5 M⍀ resistance) placed in the stratum radiatum of the CA1 area. The intensity of the stimulus was adjusted to evoke a fEPSP of 0.7-1 mV without appreciable population spike contamination. Recordings were obtained with an Axoclamp 2B amplifier coupled to a DigiData 1200 interface (Axon Instruments, Foster City, CA) and averages of 8 consecutive responses were continuously monitored on a personal computer with the LTP 1.01 software (16). Responses were quantified as the initial slope of the averaged fEPSPs. Superfusion of a slice with drugs was started after responses were stable for at least 20 min.
Release of Acetylcholine from Hippocampal Slices-The release of [ 3 H]acetylcholine was performed as described previously (17,18). Briefly, rat hippocampal slices were loaded with [ 3 H]choline (30 Ci/ml, 0.3 M) for 30 min, washed, placed in 100-l Perspex chambers, and superfused with oxygenated Krebs solution containing 10 M hemicholinium-3 at 30°C with a flow rate of 0.6 ml/min. The preparations were stimulated twice (S 1 and S 2 ) at 6 and 36 min with supramaximal monopolar square-wave pulses with a duration of 3 ms and an amplitude of 40 V, delivered with a frequency of 5 Hz for a period of 2 min (600 pulses), through platinum electrodes. Drugs were added to the superfusion medium 15 min before S 2 and remained in the bath up to the end of the experiment. When we evaluated the modification of the effect of a drug by a modifier, this modifier was applied 15 min before sample collection was started and hence was present during S 1 and S 2 .
At the end of the release experiments, 5 ml of scintillation mixture (Scintran T) was added to a 500-l aliquot of each eluent fraction and to 100 l of the homogenized slices (sonicated in 500 l of 3 M perchloric acid and 2% Triton X-100). Radioactivity was expressed in terms of disintegrations per minute per mg of protein in each chamber (dpm/ mg). Protein was quantified according to Spector (14). The fractional release was expressed in terms of the percentage of total radioactivity present in the tissue at the time of sample collection. The release of tritium evoked by each period of electrical stimulation, i.e. the evoked release (expressed as fractional release), was calculated by integration of the area of the peak upon subtraction of the estimated basal tritium outflow from the total outflow due to electrical stimulation. The evoked release of tritium under these conditions is Ca 2ϩ -and tetrodotoxinsensitive and is mostly constituted by [ 3 H]acetylcholine (17,18). The effect of drugs on the release of tritium was expressed by alterations of the ratio between the evoked release due to the second stimulation period and the evoked release due to the first stimulation period (S 2 /S 1 ratio).
Biochemical Analysis-For the determination of the energy charge, the synaptosomes were incubated without or with AA, PLA 2 , or melittin for 30 min, homogenized by sonication, and adenine nucleotides were extracted and analyzed by HPLC (17,18). To evaluate cellular integrity, the synaptosomes were placed in the superfusion chambers and superfused for 30 min with gassed Krebs without or with AA, PLA 2 , or melittin in a close-circuit manner. Cellular disruption was determined by comparing the lactate dehydrogenase activity in the superfusion medium with that found in the synaptosomes upon their solubilization, by sonication, with 2% (v/v) Triton X-100 (2). To study the metabolism of AA in hippocampal synaptosomes, [ 3 H]AA (5 M, 21.8 Ci/ml) was incubated for 20 min at 37°C with hippocampal synaptosomes (0.36 -0.42 mg protein) in a final volume of 200 l, in the absence or presence of indomethacin (20 M) and/or NDGA (50 M). The samples were then diluted to a concentration of 80% (v/v) ethanol, vortexed, and kept on ice for 20 min. The samples were then centrifuged at 1,000 ϫ g for 10 min to pellet the precipitated protein material and the supernatant was evaporated to dryness on a rotary evaporator at 37°C. Extraction of lipid material was performed by resuspension in 300 l of 30% (v/v) methanol, a procedure that yields good recoveries (86 Ϯ 3%, n ϭ 4, using synaptosomes previously boiled for 30 min) of [ 3 H]AA, as previously reported (19). Separation of eicosanoids (100 l injected) in standards and samples was achieved by an adapted reverse-phase HPLC method (see Ref. 19) using a Beckman Gold system equipped with a Lichrospher C-18 (5 m) column, imposing a sequence of isocratic elutions (flow rate:1.25 ml/min) starting with 55% (v/v) methanol in buffered water (prepared by adding 1 ml of glacial acetic acid to 1 liter of water and adjusting the pH to 5.8 with 1 M NH 4 OH) for 22 min followed by 100% methanol for 14 min. The retention times of the tested eicosanoids, identified by UV detection at 220 nm, were: PGE 2 7.77 Ϯ 0.06 min, PGF 2␣ 10.51 Ϯ 0.09 min, 5-HETE and 12-HETE co-eluting at 24.76 Ϯ 0.09 min, and AA 26.08 Ϯ 0.07 min. The HPLC eluent of the chromatographed lipid extracts of synaptosomes incubated with [ 3 H]AA was collected every minute and analyzed by scintillation counting to monitor the disappearance of tritium from [ 3 H]AA and the appearance of tritium in other eicosanoid peaks.
Statistics-The values are presented as mean Ϯ S.E. To test the significance of the effect of an agonist versus control, a paired Student's t test was used. When making comparisons from a different set of experiments with control, one-way analysis of variance (ANOVA) was used, followed by Dunnett's test. p Ͻ 0.05 was considered to represent a significant difference.

RESULTS
Exogenously AA Facilitates the Evoked Outflow of Adenosine-The stimulation of hippocampal synaptosomes with K ϩ (20 mM for 2 min) resulted in an increase of tritium (Fig. 1A, open symbols) and of [ 3 H]adenosine (Fig. 1B, open symbols) in the superfusion effluent, which represents 1.12 Ϯ 0.03% (n ϭ 12) and 0.164 Ϯ 0.007% (n ϭ 12) of total tritium retained by the synaptosomes, respectively. The addition of AA (30 M) to the superfusate 9 min before K ϩ stimulation caused a 28 Ϯ 3% (n ϭ 4) decrease in the basal outflow of tritium ( caused a lower facilitation mainly of the evoked outflow of [ 3 H]adenosine but also of tritium (Fig. 1C). To investigate whether the effect of AA on tritium outflow was influenced by detergent-like effects or oxidative phosphorylation uncoupling caused by AA (e.g. Ref. 21), we tested the effect of AA on the energy charge and on the release of lactate dehydrogenase, an intracellular marker, from superfused rat hippocampal synaptosomes. As shown in Table I, AA (30 M) did not significantly change either the energy charge or the release of lactate dehydrogenase. In contrast, AA (100 M) decreased the energy charge by 8.8 Ϯ 0.1% (n ϭ 4, p Ͻ 0.05) and caused the release of 12 Ϯ 2% (n ϭ 4, p Ͻ 0.05) of total lactate dehydrogenase during the 30-min superfusion period (Table I). Thus, as was previously proposed to occur (18,21), the lower effects of AA (100 M) on the outflow of tritium and [ 3 H]adenosine might probably be due to a modification of the homeostasis and viability of the synaptosomes.
Since AA caused virtually identical effects on tritium outflow and [ 3 H]adenosine outflow, the tritium outflow was used as a measure of [ 3 H]adenosine outflow in the remaining experiments. It was also observed that the effect of AA on the K ϩevoked tritium outflow were consistent from experiment to experiment, whereas there was a large variation in the effects of AA on the basal tritium outflow, which precluded a systematic investigation of the effect of AA on basal tritium outflow.
Incubation of the synaptosomes with BSA (1%) 15 min before starting sample collection did not significantly modify the K ϩevoked outflow of tritium (n ϭ 3). However, the effect of AA (30 M) on the K ϩ -evoked outflow of tritium was almost abolished in the presence of 1% BSA ( Fig. 2A). Superfusion of the synaptosomes 9 min before K ϩ stimulation with another trans-unsaturated fatty acid, linolenic acid (30 M), also facilitated the evoked tritium outflow ( Fig. 2A) by 42 Ϯ 5% (n ϭ 3). In contrast, the saturated free fatty acid arachidic acid (30 M, n ϭ 2) did not significantly modify the K ϩ -evoked tritium outflow ( Fig. 2A).
Several effects exerted by eicosanoids in the hippocampus are mediated by cyclooxygenase or lipoxygenase metabolites of AA (22,23). Indeed, AA was metabolized by hippocampal nerve terminals with the formation mainly of lypoxygenase products but also cyclooxygenase products (see dashed line in Fig. 2B and data in Table II). But the AA-induced facilitation of the evoked release of adenosine does not appear to be due to any of its cyclooxygenase metabolites, since preincubation of the synaptosomes with the cyclooxygenase inhibitor, indomethacin (20 M, n ϭ 3) from 15 min before starting sample collection onwards, did not significantly modify the AA (30 M)-induced facilitation of K ϩ -evoked tritium outflow (Fig. 2C). Indomethacin (20 M) completely prevented the formation of PGE 2 and PGF 2␣ from AA (Table II), indicating that this concentration of indomethacin might be supramaximal to inhibit cyclooxygenase in nerve terminals, as observed by others (24). As shown in  (Table II), indicating that this concentration might be supramaximal to inhibit lypoxygenase in nerve terminals, as ob-  served by others (25). However, the observation that the combined presence of indomethacin (20 M) and NDGA (50 M) only attenuated by 14 Ϯ 3% (n ϭ 3) the AA (30 M)-induced facilitation of K ϩ -evoked tritium outflow (Fig. 2C), suggests that this effect of AA is mostly due to the effect of AA as such rather than to any of its metabolites. As shown in Table II Endogenously Produced AA Facilitates the Evoked Outflow of Adenosine-When phospholipase A 2 (PLA 2 , 2 units/ml), an enzyme that releases fatty acids from the sn-2 position of phospholipids where AA is mostly bound, was superfused 9 min before K ϩ stimulation, it enhanced the K ϩ -evoked tritium outflow by 73 Ϯ 5% (n ϭ 4) (Fig. 3A). Incubation of the synaptosomes with BSA (1%) from 15 min before starting sample collection onward prevented this facilitatory effect of PLA 2 (2 units/ml) on the evoked tritium outflow (Fig. 3A).
We then tested the effect of melittin that, at concentrations below 3 M, activates endogenous PLA 2 in nerve terminals (26) without disrupting nerve terminals (18,27). When melittin (1 M) was superfused 9 min before K ϩ stimulation, it enhanced the evoked tritium outflow by 85 Ϯ 7% (n ϭ 4) (Fig. 3B). We and others have previously shown that PLA 2 inhibitors are able to prevent some effects of melittin (6,7,18,27) and this constitutes an important control to ascribe PLA 2 as a selective target of melittin, since this peptide causes a general membrane perturbation at concentrations higher than 3-10 M (28). Incubation of the synaptosomes with AACOCF 3 (20 M, n ϭ 4), an inhibitor of cytosolic PLA 2 and of Ca 2ϩ -independent PLA 2 activities (29), from 15 min before starting sample collection onward, did not modify evoked tritium outflow but prevented the facilitatory effect of melittin (1 M) on the evoked tritium outflow (Fig. 3B). Incubation of the synaptosomes with BSA (1%) from 15 min before starting sample collection onward, also prevented the facilitatory effect of melittin (1 M, n ϭ 4) on the evoked tritium outflow (Fig. 3B).
AA Preferentially Increases Adenosine Formation from Released Adenine Nucleotides-The extracellular accumulation of adenosine may result from two different sources: either from a release of adenosine as such through the bidirectional nonconcentrative adenosine transporters or the formation of adenosine resulting from the extracellular catabolism of released adenine nucleotides (reviewed in Ref. 30). The discrimination between these two possible metabolic sources is possible by comparing the effect of AOPCP, an inhibitor of ecto-5Ј-nucleo-tidase, the enzyme responsible for the formation of adenosine from adenine nucleotides, with the effect of NBTI and dipyridamole, two inhibitors of the bidirectional non-concentrative adenosine transporters (1) (Fig. 4A). This suggests that the AA-induced facilitation of [ 3 H]adenosine outflow is mostly due to increased formation of adenosine from released adenine nucleotides.
AA Increases the Evoked Release of ATP-To directly test if AA facilitated the release of adenine nucleotides, we investi-

FIG. 3. Endogenous AA also increases the evoked adenosine outflow.
In A is shown the facilitation of the K ϩ (20 mM)-evoked outflow of tritium from rat superfused hippocampal synaptosomes by phospholipase A 2 (PLA 2 , 2 units/ml), which releases AA from the sn-2 position of phospholipids, and the ability of BSA (1%), which quenches AA, to prevent this PLA 2 -induced facilitation. In B is shown the facilitation of the evoked outflow of tritium by melittin (1 M), which stimulates endogenous PLA 2 , and the ability of the PLA 2 inhibitor AA-COCF 3 (20 M) and BSA (1%) to prevent this melittin-induced facilitation. PLA 2 or melittin were added 9 min before K ϩ stimulation whereas the modifiers were added 15 min before starting sample col- gated the effect of AA on the evoked release of ATP from hippocampal nerve terminals. The stimulation of hippocampal synaptosomes with K ϩ (20 mM) resulted in an evoked release of ATP (41 Ϯ 6 pmol/mg protein, n ϭ 4). This evoked release of ATP was Ca 2ϩ -dependent, since omission of Ca 2ϩ in the Krebs solution blocked the K ϩ -induced ATP release (data not shown). AA (3-30 M) caused a concentration-dependent facilitation of the evoked release of ATP (Fig. 4B). However, a higher concentration of AA (100 M) caused a lower facilitation of the evoked release of ATP (Fig. 4B), as occurred for the evoked outflow of [ 3 H]adenosine (Fig. 1C).

AA Modifies the Formation of Adenosine from Adenine Nucleotides by Ecto-nucleotidases-
The results obtained indicate that the main metabolic pathway of extracellular adenosine formation induced by AA is through the catabolism of released ATP. The ecto-nucleotidase pathway, which is responsible for the formation of adenosine from released ATP (34), is not a linear sequence of ecto-enzymes generating extracellular adenosine as a monotonous function of the amounts of ATP released, since the "rate-limiting" enzyme, ecto-5Ј-nucleotidase, is feedforwardly inhibited by the initial substrates of the pathway, ATP and ADP (35). Thus, we investigated the effect of AA (30 M) on the ability of hippocampal synaptosomes to catabolize extracellular ATP and to form adenosine from extracellular AMP. Fig. 6A shows that ATP was extracellularly catabolized with less efficiency in AA (30 M)-treated synaptosomes (10.1 Ϯ 0.5 nmol of ATP catabolized per min per mg protein, n ϭ 4) than in control synaptosomes (12.8 Ϯ 0.6 nmol/min/mg protein, n ϭ 4). In contrast, Fig. 6B shows that the formation of aden- osine from released adenine nucleotides might be more efficient in AA (30 M)-treated synaptosomes since the activity of ecto-5Ј-nucleotidase (the enzyme that forms adenosine from extracellular adenine nucleotides) was greater in AA (30 M)-treated synaptosomes (2.06 Ϯ 0.08 nmol of AMP catabolized per min per mg protein, n ϭ 5) than in control synaptosomes (0.69 Ϯ 0.05 nmol/min/mg of protein, n ϭ 5). This nearly 3-fold increase of ecto-5Ј-nucleotidase activity by AA might be partially mediated via protein kinase C since the activity of ecto-5Ј-nucleotidase was reduced to 1.41 Ϯ 0.06 nmol/min/mg of protein (n ϭ 4) in the simultaneous presence of AA (30 M) and of the protein kinase C inhibitor, chelerythrine (6 M). By itself, chelerythrine (6 M) did not significantly modify the activity of ecto-5Ј-nucleotidase (0.65 Ϯ 0.04 nmol/min/mg of protein, n ϭ 4).
AA Inhibits Adenosine Transport- Fig. 6C suggests that the adenosine that accumulates extracellularly is likely to have a higher half-life in AA (30 M)-treated synaptosomes since there is a decreased efficiency of adenosine transporters upon treatment with AA (30 M). The kinetic parameters indicate that there is a lower capacity of adenosine transporters in AA (30 M)-treated synaptosomes (V max ϭ 0.96 Ϯ 0.09 pmol/mg of protein/15 s, n ϭ 4) than in control synaptosomes (V max ϭ 1.83 Ϯ 0.15 pmol/mg of protein/15 s, n ϭ 4), without significant changes in K m (1.30 Ϯ 0.21 M in AA-treated synaptosomes and 1.17 Ϯ 0.13 M in control synaptosomes, n ϭ 4) (Fig. 6C). This AA-induced decrease of the capacity of adenosine transporters may mostly involve protein kinase C activation since the capacity of adenosine transporters nearly returned to control values in AA (30 M)-treated synaptosomes in the presence of the protein kinase C inhibitor, chelerythrine (6 M) (V max ϭ 1.03 Ϯ 0.09 pmol/mg of protein/15 s, n ϭ 4). By itself, chelerythrine (6 M) did not significantly modify the capacity of adenosine transporters (V max ϭ 0.99 Ϯ 0.07 pmol/mg of protein/15 s, n ϭ 4).

Effect of AA on Adenosine A 1 Receptor Modulation of Hippocampal Synaptic Transmission-In the Schaffer fibers/CA1
pyramid synapses of rat hippocampal slices, extracellular adenosine is present in amounts high enough to tonically inhibit synaptic transmission via A 1 receptor activation (3,36). Since endogenous adenosine only partially attenuates synaptic transmission and provided the concentration of the A 1 receptor antagonist is supramaximal, the variation in the levels of endogenous extracellular adenosine can be evaluated by monitoring the effect of a supramaximal concentration of an A 1 receptor antagonist, so that the greater the facilitatory effect of the To rule out that AA might be mostly affecting the functioning of adenosine A 1 receptors rather than the extracellular levels of adenosine, we tested the effect of the closest chemical analogue of adenosine, 2-chloroadenosine, on hippocampal synaptic transmission in the absence and presence of AA (30 M). In control conditions, 2-chloroadenosine inhibited fEPSP slope with an EC 50 of 373 nM (95% confidence interval: 354 -392 nM, n ϭ 7). In the presence of AA (30 M), the EC 50 of 2-chloroadenosine to inhibit synaptic transmission was increased to 524 nM (95% confidence interval: 499 -549 nM, n ϭ 4; p Ͻ 0.05). To test if this shift to the right of the concentration response to 2-chloroadenosine was due to an enhanced tonic activation of inhibitory A 1 receptors by endogenous adenosine, we tested the effect of AA on the inhibition by 2-chloroadenosine of hippocampal synaptic transmission in the presence of adenosine deaminase. In the presence of adenosine deaminase (2 units/ ml), the EC 50 of 2-chloroadenosine was 337 nM (95% confidence interval: 318 -356 nM, n ϭ 6), and AA (30 M) failed to decrease the potency of 2-chloroadenosine to inhibit synaptic transmission (n ϭ 4).
Effect of AA on Adenosine A 1 Receptor Modulation of Hippocampal Acetylcholine Release-We then attempted to confirm this functional relevance of the AA-induced accumulation of extracellular adenosine concentration in a different experimental paradigm, by directly measuring neurotransmitter re- lease from hippocampal slices. The field electrically evoked release of acetylcholine from hippocampal slices is also tonically inhibited by endogenous adenosine acting through adenosine A 1 receptors (17) and DPCPX is a selective A 1 receptor antagonist that has a supramaximal effect at concentrations above 20 nM in this system (17). In control conditions, DPCPX (50 nM) caused a 18 Ϯ 2% (n ϭ 4) facilitation of the evoked [ 3 H]acetylcholine release, whereas in the presence of AA (30 M), DPCPX (50 nM) caused a 31 Ϯ 5% (n ϭ 4) enhancement of the evoked [ 3 H]acetylcholine release (Fig. 7B). Again, to rule out that the effect of AA could be due to an increased efficiency of A 1 receptors, we tested the effect of the selective adenosine A 1 receptor agonist, N 6 -cyclopentyladenosine, in the absence and presence of AA (30 M). In control conditions, N 6 -cyclopentyladenosine (10 nM to 1 M) inhibited the evoked [ 3 H]acetylcholine release in a concentration-dependent manner with an EC 50 of 53 nM (n ϭ 4). In the presence of AA (30 M, n ϭ 3-4), the EC 50 of N 6 -cyclopentyladenosine was slightly (p Ͼ 0.05) increased to 86 nM, confirming that the effect of AA is due to an increase in the extracellular levels of adenosine rather than to an increase functioning of A 1 receptors. DISCUSSION The present results show that low micromolar concentrations of AA stimulated the evoked adenosine outflow from rat hippocampal synaptosomes, with a maximal effect recorded at 30 M AA. These concentrations of AA are within the range of concentrations reached in the extracellular medium upon neuronal activity (7,41). At higher concentrations of AA (100 M), which are estimated to be reached during pathological conditions such as upon ischemic insults (42), there was a decreased facilitation of evoked adenosine outflow together with a facilitation of basal adenosine outflow. These effects of 100 M AA may be due to disturbance of membrane integrity and/or uncoupling of oxidative phosphorylation (18,21), since 100 M AA, in contrast with the lower concentrations of AA tested, caused a reduction in the energy charge of the synaptosomes and an increased release of lactate dehydrogenase, an intracellular marker (Table I).
The present results also suggest that endogenously produced AA is able to enhance the evoked adenosine outflow from hippocampal nerve terminals, since superfusion of the synaptosomes with PLA 2 , which hydrolyses the sn-2 position of phospholipids where AA is normally sterified, and with melittin, a peptide that activates endogenous PLA 2 in nerve terminals (26), mimicked the effect of exogenously added AA in low micromolar concentrations. However, the observation that the amplitude of the facilitatory effects of both PLA 2 and melittin are greater than that of exogenously added AA suggests that the generation of endogenous AA is more effective than exogenously added AA at raising free AA levels (7,43). The effect of both PLA 2 and melittin was attenuated by albumin, a water soluble protein that binds with high affinity free, but not lipidsterified, fatty acids (44), supporting the requirement of AA production for PLA 2 and melittin to exert their facilitatory effects on adenosine outflow. Since the effect of both PLA 2 and melittin on the evoked adenosine outflow was attenuated by albumin, which binds AA with significantly higher affinity than its metabolites (45), the facilitation of adenosine outflow may be caused by endogenous AA itself rather than by its cyclooxygenase or lipoxygenase metabolites. As observed by other in different brain regions (24,46,47), we now observed that AA metabolism in hippocampal nerve terminals occurs both through the cyclooxygenase and lipoxygenase routes. The observation that inhibition of cyclooxygenase, failed to modify the effect of exogenously added AA and that inhibition of lipoxygenase, by using effective concentrations of NDGA, only inhibited by nearly 15% the effect of exogenously added AA reinforces the idea that the facilitatory effect of AA on the evoked adenosine release mostly results from the action of AA itself rather from that of its metabolites. This lack of interference of arachidonic acid metabolites on the modulation of neurotransmitter release by adenosine in the rat hippocampus and striatum has previously been reported (39).
Excitatory synapses in the hippocampus use glutamate as the main neurotransmitter but also release ATP upon stimulation (1,48), which is also an excitatory neurotransmitter in the hippocampus (49). AA is proposed to behave as a retrograde messenger in excitatory transmission in the hippocampus, being released upon activation of cholinergic and glutamatergic receptors and enhancing the release of glutamate and acetylcholine (4,18). The present observation that AA also enhances the release of ATP, together with the previous observations that P 2 receptor activation triggers AA release (e.g. Ref. 50) and that AA facilitates P 2 receptor activity (51), extends the concept of retrograde facilitation to another excitatory neurotransmitter in the hippocampus. Besides behaving as a neurotransmitter, ATP can also be a source of extracellular adenosine, upon its extracellular catabolism by the ecto-nucleotidase pathway (1, 3). This metabolic source of adenosine is particularly important in nerve terminals (30), where extracellular adenosine mostly acts as a neuromodulator independently of its homeostatic role (30,36). The observations that the ecto-5Ј-nucleotidase inhibitor, ␣,␤-methylene ADP, caused a larger inhibition of the evoked adenosine outflow in the presence than in the absence of AA and that AA enhanced the evoked release of ATP, in the same concentration range as AA enhanced the evoked outflow of adenosine, indicates that the AA-induced extracellular accumulation of adenosine upon stimulation is mostly due to enhanced formation of adenosine from released ATP. Interestingly, we also noted that AA increased the activity of ecto-5Ј-nucleotidase, which is the rate-limiting step of the pathway (35). This might attenuate the physiological impact of the fedforward inhibition of ecto-5Ј-nucleotidase by ATP and/or ADP and will tend to form adenosine in a linear manner, favoring the activation of inhibitory A 1 receptors (see Ref. 21).
The present results also show that AA inhibited adenosine transporters of hippocampal nerve terminals. Extracellular adenosine can also be originated from the release of adenosine as such through the nucleoside transporters (52) and this metabolic source of adenosine is depressed by AA. However, this source of adenosine is more important when metabolic imbalance occurs, rather than during synaptic transmission, where adenosine formation from adenine nucleotides seems to predominate at the synaptic level (reviewed in Ref. 30). Thus, it is anticipated that AA may increase tonic adenosine modulation upon synaptic transmission but might decrease adenosine homeostatic role during stress conditions in the hippocampus.
It was observed that a protein kinase C inhibitor attenuated the AA-induced release of adenosine and ATP. AA is a direct activator of protein kinase C activity (31) and the AA facilitation of the evoked release of glutamate (4) and ␥-aminobutyric acid (27) in the hippocampus is also attenuated by protein kinase C inhibitors. It is interesting to note that activation of protein kinase C inhibits adenosine transporters (53,54) and we now observed that an inhibitor of protein kinase C nearly prevented the AA-induced decrease of the capacity of adenosine transporters. Thus, it is tempting to speculate that protein kinase C activation by AA might be a common pathway for the effect of AA on ATP and adenosine release and extracellular adenosine metabolism. However, AA also produces direct action, apparently not involving intracellular transducing systems, in several membrane targets (e.g. Ref. 55). In this respect, it has been reported that the AA-induced enhancement of membranes ATPase activity in rat brain microvessels might be mediated by lipid peroxide metabolites of AA (56) and we now observed that inhibition of protein kinase C only partially reverts the AA-induced enhancement of ecto-5Ј-nucleotidase activity, although in other systems a strictly protein kinase C-dependent activation of ecto-5Ј-nucleotidase has been reported (e.g. Ref. 57).
The AA-induced increase in evoked extracellular adenosine accumulation was confirmed at the functional level by the greater effect of adenosine A 1 receptor antagonists to facilitate neurotransmitter release and synaptic transmission in hippocampal slices in the presence rather than absence of AA. It is well established that increased levels of the endogenous agonist increase the effect of supramaximal concentrations of exogenously added antagonist and may shift to the right the concentration response curve of exogenously added agonists when the levels of the endogenous agonist are high enough to tonically activate the studied receptor (58). This has been experimentally confirmed in relation to adenosine neuromodulation in the hippocampus, by showing that manipulations that increase the levels of endogenous extracellular adenosine increase the facilitatory effect of supramaximal concentrations of A 1 receptor antagonists and slightly shift to the right the con-centration response curve of A 1 receptor agonists (e.g. Ref. 59). Thus, the presently observed AA-induced increase of the efficiency of supramaximal concentrations of A 1 receptor antagonists together with the lack of enhancement in the potency of A 1 receptor agonists is consistent with the idea that AA increased the evoked extracellular adenosine concentration, rather than to an AA-induced modification of A 1 receptor functioning (60). Since it is known that endogenous adenosine exerts an efficient tonic inhibition of synaptic plasticity phenomena (9), this increased tonic adenosine inhibition in the presence of AA might constitute an auto-control mechanism to increase the threshold of AA-mediated plasticity phenomena in the hippocampus (4).
In summary, the observation that a facilitatory modulator (AA) increases the extracellular concentration and the physiological efficiency of an inhibitory neuromodulatory system (adenosine) represents a direct demonstration of a functional interaction between neuromodulatory systems to control the release of neurotransmitters and synaptic transmission.