Vasoactive Intestinal Peptide (VIP) and Pituitary Adenylate Cyclase-activating Polypeptide (PACAP) Potentiate the Glutamate-evoked Release of Arachidonic Acid from Mouse Cortical Neurons EVIDENCE FOR A cAMP-INDEPENDENT MECHANISM*

Glutamatergic neurotransmission is associated with release of arachidonic acid (AA) from membrane phospholipids of both neurons and astrocytes. Since free AA has been shown to enhance glutamate-mediated synaptic transmission, it can be postulated that glutamate release and AA formation constitute a positive feed-back mechanism for sustained excitatory neurotransmission. In the present study, we examined whether the gluta-mate-evoked release of AA could be modulated by pep- tides. Using mouse cortical neurons in primary cultures, we show that the release of AA evoked by glutamate is potentiated by vasoactive intestinal peptide and pituitary adenylate cyclase-activating polypeptide (PACAP). This effect is mediated through the activation of PACAP I receptors. However, several arguments show that this potentiating mechanism does not involve the cAMP/PKA pathway. 1) Increasing intracellular cAMP by either cholera toxin, forskolin, or 8-Br-cAMP treatments does not affect the glutamate-evoked release of AA; 2) poten- tiation of the glutamate response by PACAP is not pre-vented by the PKA inhibitor 8-Br- R p -cAMPS. Also, an involvement of the phospholipase C protein kinase C pathways is unlikely since inhibitors of both phospho- lipase C ( i.e. U-73122) and protein kinase C ( i.e. Ro 31-8220)

Glutamatergic neurotransmission is associated with release of arachidonic acid (AA) from membrane phospholipids of both neurons and astrocytes. Since free AA has been shown to enhance glutamate-mediated synaptic transmission, it can be postulated that glutamate release and AA formation constitute a positive feed-back mechanism for sustained excitatory neurotransmission.
In the present study, we examined whether the glutamate-evoked release of AA could be modulated by peptides. Using mouse cortical neurons in primary cultures, we show that the release of AA evoked by glutamate is potentiated by vasoactive intestinal peptide and pituitary adenylate cyclase-activating polypeptide (PACAP). This effect is mediated through the activation of PACAP I receptors. However, several arguments show that this potentiating mechanism does not involve the cAMP/PKA pathway. 1) Increasing intracellular cAMP by either cholera toxin, forskolin, or 8-Br-cAMP treatments does not affect the glutamate-evoked release of AA; 2) potentiation of the glutamate response by PACAP is not prevented by the PKA inhibitor 8-Br-R p -cAMPS. Also, an involvement of the phospholipase C protein kinase C pathways is unlikely since inhibitors of both phospholipase C (i.e. U-73122) and protein kinase C (i.e. Ro 31-8220) do not affect the potentiation of the glutamate response by PACAP. These observations indicate an effect mediated by PACAP I receptors, which does not involve the second messenger pathways classically associated with activation of this type of receptors. Furthermore, results indicate that this potentiating mechanism mediated by PACAP I receptor acts at a level downstream of the glutamate receptor-mediated calcium influx.
Glutamate is the main excitatory neurotransmitter in the cerebral cortex. As glutamate is released into the synaptic cleft, a specific reuptake system present on both neuronal and glial cells adjusts the extracellular concentrations of this neurotransmitter between 0.2 M and 200 M (1). Activation of glutamatergic receptors results in both rapid changes of ionic currents and second messenger formation. Among the second messenger pathways stimulated by glutamate, the release of AA 1 has been shown in brain slices (2) and primary cultures of neurons or astrocytes (3)(4)(5). The glutamate-evoked release of AA could be of particular relevance to the physiological regulation of the excitatory signaling, since it has been suggested that AA itself could amplify glutamatergic neurotransmission. Indeed, AA enhances glutamate release from presynaptic terminals (6), inhibits glutamate reuptake (7), and enhances NMDA currents (8). Interestingly, certain components of long term potentiation appear to involve AA formation (9).
AA is an unsaturated fatty acid predominantly located in the sn-2 position of membrane phospholipids from where it can be released by various types of phospholipases, mainly by phospholipase A 2 (PLA 2 ) or by sn-2 diacylglycerol lipase (for review, see Ref. 10). In neurons, the glutamate-evoked release of AA appears to involve the activation of PLA 2 (3,4,11). The release of AA is tightly modulated by different mechanisms; in particular, PLA 2 activity is regulated by phosphorylation. Thus, both PLA 2 activity and the subsequent release of AA are enhanced by the cAMP/cAMP-dependent protein kinases (cAMP/PKA) pathway (12,13).
In the present study, we addressed the question of whether activation of the cAMP/PKA pathway in neurons would modulate the glutamate-evoked release of AA. For this purpose, we investigated the effect of VIP and PACAP on the glutamate response, since both peptides are potent and efficient stimulators of cAMP formation in cortical neurons (14 -17). Both VIP and PACAP interact with the G s protein-coupled PACAP I and PACAP II receptors. PACAP I receptors are activated by PACAP in the nanomolar range and by VIP in the micromolar range, whereas PACAP II receptors are activated in the nanomolar range by both peptides (18,19).
Here we report that VIP and PACAP, by activating PACAP I receptors, potentiate the glutamate-evoked release of AA from primary cultures of cerebral cortical neurons. However, the cascade of events that leads to this potentiation is not mediated by G s proteins, nor by the cAMP/PKA or the phospholipase C (PLC)/PKC pathway. Therefore, our data describe a potentiating effect mediated by the activation of PACAP I receptors, which does not involve second messenger pathways classically associated with the activation of these types of receptors.

Methods
Cell Culture-Cortical neurons devoid of glial cells were prepared as described previously (11). Briefly, the cerebral cortex of 16-day-old Swiss albino mice embryos was dissected by removing the olfactory bulbs, striatum, hippocampus, and meninges. Cells were dissociated mechanically and were plated (800,000 cells/ml) on 12-well Falcon culture dishes (1 ml/well) previously coated with 15 g/ml polyornithine. For intracellular Ca 2ϩ ([Ca 2ϩ ] i ) imaging, cells were plated on glass coverslips also coated with 15 g/ml polyornithine, followed by 2 g/ml laminin. In all cases, coating was carried out in serum-free medium. The culture medium was composed of DMEM supplemented with 2 mM glutamine, 7.5 mM NaHCO 3 , 5 mM HEPES buffer (pH 7.0), 100 g/ml streptomycin, 60 g/ml penicillin, 25 g/ml insulin, 100 g/ml transferrin, 60 M putrescine, 20 nM progesterone, and 30 nM sodium selenite. Cells were cultured at 37°C in a humidified atmosphere of 95% air and 5% CO 2 for 5-7 days in vitro, a time point at which they reach maximum viability and exclusively express neuronal markers (11).

Measurement of [ 3 H]Arachidonic
Acid Release-Measurement of [ 3 H]AA release was carried out as described previously (11). Briefly, cortical neurons cultured in 12-well culture dishes were incubated for 16 -20 h in the presence of [ 3 H]AA (1 Ci/well) by adding 100 l/well of DMEM containing [ 3 H]AA and 0.01% BSA as a lipid carrier. After washing three times at 37°C with a Locke-HEPES buffer (L-H buffer (in mM): NaCl, 145; KCl, 5.5; CaCl 2 , 1.1; MgCl 2 , 1.1; NaHCO 3 , 3.6; glucose, 5.5; HEPES, 20; pH, 7.4) containing 0.2% BSA, cells were preincubated for 15 min in L-H buffer, supplemented with adenosine deaminase (1 IU/ml). This latter procedure was used to degrade endogenous adenosine, which has been previously shown to affect glutamatemediated signal transduction (20). Unless otherwise stated, inhibitors or antagonists were also added during this preincubation period. Neurons were then exposed for 15 min at 37°C to the various agents in a similar L-H buffer, lacking Mg 2ϩ (a condition necessary to yield full activity of NMDA receptors; Refs. 21 and 22) and in the presence of adenosine deaminase (1 IU/ml). At the end of the stimulation, the incubation medium was collected, centrifuged at 200 ϫ g for 5 min, and an aliquot of the supernatant was assayed for 3 H by liquid scintillation counting. To avoid misestimation due to the presence of BSA, the protein content was determined in sister cultures by the method of Bradford (23).
Thin Layer Chromatography Analysis of [ 3 H]Arachidonic Acid Release-Characterization by thin layer chromatography (TLC) of the 3 H-labeled product(s) released from cortical neurons was carried out as described previously (11). This procedure allows complete separation of AA from its metabolites including prostaglandins and most lipoxygenase metabolites, as well as from other classes of lipids such as phospholipids and mono-and diacylglycerols.
Determination of cAMP Levels-cAMP levels were determined by radioimmunoassay under conditions similar to those used for the [ 3 H]AA release measurement. Briefly, cortical neurons cultured in 12-well dishes were washed three times with L-H buffer at 37°C and subsequently preincubated for 15 min in L-H buffer supplemented with IBMX (1 mM) and adenosine deaminase (1 IU/ml). Neurons were then exposed for 15 min at 37°C to the agents in a L-H buffer lacking Mg 2ϩ , supplemented with IBMX (1 mM), adenosine deaminase (1 IU/ml), and glutamate (100 M). Glutamate was added in these experiments because cortical neurons possess metabotropic glutamate receptors, which are negatively coupled to the stimulation of adenylyl cyclase (24). Indeed, as cortical neurons were incubated in presence of 100 M glutamate, the accumulation of cAMP evoked by 1 M VIP and 100 nM PACAP was partially inhibited (data not shown). Therefore, we set out to determine whether agents that are known to stimulate cAMP formation were still efficient in the presence of 100 M glutamate (see "Results"). The reaction was stopped by replacement of the incubation buffer with ice-cold L-H buffer and subsequent sonication. The suspension was boiled for 10 min and centrifuged for 2 min at 9980 ϫ g, and an aliquot of the supernatant was taken to assess cAMP levels (using 125 I-cAMP as a tracer; Ref. 25 (11). Briefly, cells were plated in the presence of myo-[2-3 H]inositol (4 Ci/ml) (in 12-well culture dishes). After 5-7 days in vitro, neurons were washed three times with L-H buffer (1 ml/well) at 37°C, then preincubated for 15 min in L-H buffer supplemented with lithium (10 mM) and adenosine deaminase (1 IU/ml), and finally exposed for 15 min to the agents in a L-H buffer lacking Mg 2ϩ , supplemented with lithium (10 mM) and adenosine deaminase (1 IU/ml). Isolation of [ 3 H]IPs by ion-exchange chromatography (Dowex AG 1-X8) was performed as described previously (see Ref. 11).

Determination of Intracellular [Ca 2ϩ ]-Determination of [Ca 2ϩ
] i was carried out as described previously (11). Briefly, cortical neurons grown on glass coverslips were studied with dual emission microfluorimetry. Cells were loaded with 12 M INDO-1 AM and then exposed to various agents (in L-H buffer) using a multichannel cell superfusion device. All responses were measured in the absence of Mg 2ϩ and in the presence of D-serine (100 M). D-Serine was used instead of glycine because it yields full activation of NMDA receptor on perfused neurons and does not activate glycine-strychnine-sensitive chloride channels (26). The concentration of [Ca 2ϩ ] i (nM) was calculated according to the equation described by Grynkiewicz et al. (27). Under those conditions, 97% of neurons tested responded to 100 M glutamate.
Statistical Analysis-Results are expressed as mean Ϯ S.E. of n independent determinations. Data were statistically analyzed using InStat, GraphPad Software, San Diego, CA.  (Fig. 1). EC 50 values for these potentiating effects were 1 M for VIP and 0.7 nM for PACAP (Fig. 2), suggesting that these peptides might exert their effect by activating PACAP I receptors. In agreement with this hypothesis, the release of [ 3 H]AA evoked by the concomitant application of VIP (1 M), PACAP (10 nM), and glutamate was similar to the response obtained with PACAP (10 nM) and glutamate (n ϭ 9; data not shown). Therefore, in the following experiments, potentiation of the glutamate-evoked release of [ 3 H]AA was investigated using 10 nM PACAP.

VIP and PACAP Potentiate the Glutamate-evoked Release of [ 3 H]Arachidonic
The effect of secretin was also studied, because this peptide shares a high degree of sequence homology with VIP and PACAP (18,19). However, secretin (1 M) had no effect on either basal [ 3 H]AA release or on the glutamate response ( Fig. 1).
Thin layer chromatography (TLC) was used to verify that glutamate alone or in the presence of PACAP evoked the release of [ 3 H]AA exclusively. Indeed, in three separate experiments, the increase in 3 H-labeled product(s) evoked by gluta-mate either alone or in the presence of PACAP was exclusively recovered in the AA fraction. No significant radioactivity was recovered in the fraction containing AA metabolites.

Activation of the Adenylyl Cyclase-Protein Kinase A Pathway Does Not Potentiate the Glutamate-evoked Release of [ 3 H]Arac-
hidonic Acid-PACAP I receptors are coupled to G s proteins and have been shown to activate adenylyl cyclase (18). Indeed, PACAP (in the presence of glutamate) induced a strong accumulation of cAMP in cortical neurons (Fig. 3A). A similar accumulation of cAMP was obtained by either direct activation of the G s proteins with cholera toxin (CTx) or by direct stimulation of adenylyl cyclase with forskolin (Fig. 3A). However, the glutamate-evoked release of [ 3 H]AA was not affected in neurons similarly treated with either CTx or forskolin (Fig. 3B), indicating that intracellular increases in cAMP are not sufficient to potentiate the glutamate response.
The involvement of PKA is also unlikely. A specific cell membrane permeant inhibitor of PKA activity, i.e. 8-Br-R p -cAMPS (100 M) did not affect the release of [ 3 H]AA evoked by either glutamate alone or by the co-application of glutamate plus PACAP (Fig. 4) (Table I). To assess the possible involvement of the PLC pathway in AA release, we used a potent inhibitor of several types of PLC, i.e. U73122 (29,30). When added to cortical neurons, U73122 (3 M) inhibited the PACAP plus glutamate-induced formation of [ 3 H]IPs by 84% (Table I), indicating that this inhibitor was active on the PLC activity present in cortical neurons. However, PLC inhibition by U73122 (3 M) did not affect the [ 3 H]AA release evoked by the co-application of glutamate and PACAP (10 nM) (Fig. 4), indicating that PLC activation is not necessary to potentiate the glutamate response.
The involvement of PKC is also unlikely. Indeed, we tested a potent inhibitor of PKC activity, i.e. Ro 31-8220, which has a specific effect between 1 and 10 M on intact cells (31). Ro 31-8220 (10 M) did not affect the release of [ 3 H]AA evoked by either glutamate alone or by the co-application of glutamate plus PACAP (10 nM) (Fig. 4). Furthermore, activation of PKC by application of the phorbol ester PMA (0.1 M) did not affect the glutamate-evoked release of [ 3 H]AA (1.03 Ϯ 0.16-fold of the glutamate response, n ϭ 6).

Potentiation of the Glutamate-evoked Release of [ 3 H]Arachidonic Acid by PACAP Occurs when Both NMDA and AMPA/ Kainate Receptors Are Activated-
The glutamate-evoked release of [ 3 H]AA in primary cultures of cortical neurons is mediated by two receptor subtypes: AMPA/kainate and NMDA (11). Activation of AMPA/kainate receptors with either kainate (100 M) alone or glutamate plus MK801 (1 M to block NMDA receptors) evoked the release of [ 3 H]AA; both of these reponses were similarly potentiated by PACAP (Fig. 5). Activation of NMDA receptors with either NMDA (100 M) alone or glutamate plus DNQX (10 M to block AMPA/kainate receptors) evoked the release of [ 3 H]AA; both reponses were also potentiated by PACAP (Fig. 5).
Finally, although metabotropic receptors are not involved in the glutamate-evoked release of AA from cortical neurons in primary cultures (11), we considered the possibility that PACAP could reveal such a coupling. This hypothesis can, however, be ruled out since (1S,3R)-1-aminocyclopentane-1,3dicarboxylic acid (1 mM) in the presence of PACAP did not evoke a significant release of [ 3 H]AA (n ϭ 9; data not shown). Taken together, these results suggest that the potentiating mechanism mediated by the activation of PACAP I receptors could be operative at a level further downstream of the glutamate induced Ca 2ϩ influx. To strengthen this hypothesis, we examined whether PACAP would affect the release of [ 3 H]AA evoked by a receptor-independent pathway, such as a large   increase in [Ca 2ϩ ] i evoked by the Ca 2ϩ ionophore ionomycin. Indeed, the release of [ 3 H]AA induced by ionomycin (10 M) was potentiated by PACAP (Fig. 5). DISCUSSION In the present study, we show that VIP and PACAP potentiate, in a concentration-dependent manner, the glutamateevoked release of AA from primary cultures of cortical neurons. Considering the EC 50 values of this potentiating effect (1 M for VIP and 0.7 nM for PACAP; see Fig. 2), it can be postulated that both peptides potentiate the glutamate response by interacting with PACAP I receptors (18,19). Further strengthening this view is the result showing that secretin, which interacts with PACAP II, but not with PACAP I receptors (18,19), had no effect on the glutamate-evoked AA release.

PACAP Potentiates the Glutamate-evoked Release of [ 3 H] Arachidonic Acid at a Level
It is well established that PACAP I receptors are coupled to G s proteins (18). However, activation of G s protein per se was not sufficient in potentiating the glutamate response, since the release of AA evoked by glutamate was not affected when the cells were treated with CTx (a toxin that activates G s protein by ADP-ribosylation). It should be emphasized that cells responded to CTx, since a strong accumulation of cAMP was measured (see Fig. 3). This result suggests that the PACAP I receptors involved in this particular potentiating effect are not coupled to the G s protein family. The family of G i/o proteins can also be ruled out, since the potentiation of the glutamateevoked release of AA by PACAP is not affected by treating the neurons with 1 g/ml pertussis toxin for 24 h (data not shown).
In line with the non-involvement of G s proteins, several arguments allow to rule out the cAMP-PKA pathway. Indeed, the glutamate-evoked release of AA was not affected by (i) treatments that induce an accumulation of cAMP to the same extent than that measured in the presence of PACAP (i.e. forskolin and CTx; Fig. 3) or (ii) application of cell membrane permeant cAMP (i.e. 8-Br-cAMP). Furthermore, the potentiating effect of PACAP was (i) not inhibited by a specific PKA inhibitor (i.e. 8-Br-R p -cAMPS; Fig. 4) and (ii) not mimicked by a specific PKA activator (i.e. 8-Br-S p -cAMPS).
In order to rule out the involvement of PLC in the glutamateevoked release of AA, we used the specific inhibitor U73122. At a concentration shown here to dramatically inhibit PLC activity, this compound did not affect the release of AA evoked by glutamate alone or in the presence of PACAP (Fig. 4). This result confirms that PLC is not involved in the glutamateevoked release of AA (11), nor in its potentiation by PACAP. Although a role of PKC was not expected, mainly in view of the non-involvement of PLC as discussed above, this possibility could be ruled out since: (i) the specific inhibitor of PKC Ro 31-8220 did not affect the AA release evoked by glutamate alone or in the presence of PACAP (Fig. 4) and (ii) the activation of the PKC by application of the phorbol ester PMA did not affect the glutamate-evoked release of AA (see "Results").
Recently, it has been shown that a PACAP-like neuropeptide present in Drosophila stimulates a distinct protein kinase pathway, which is independent of PKA or PKC, i.e. the mitogen-activated protein kinase (MAP kinase) cascade (35). Since MAP kinase directly phosphorylates cPLA 2 (36), an activation of this pathway could be considered for the potentiating effect of PACAP on the glutamate-evoked release of AA. However, the use of available inhibitors, such as PD098059 (37,38), did not allow to prove the involvement of the MAP kinase pathway (data not shown). Furthermore, whether cPLA 2 is the enzyme involved in the glutamate-evoked release of AA from cortical neurons remains to be demonstrated. Therefore, at this point, we can only speculate on a possible involvement of the MAP kinase pathway in the potentiating effect of PACAP.
The Ca 2ϩ influx mediated by glutamate receptor activation is a necessary step in the glutamate-evoked release of AA (3,32,33). Confirming these results, ionomycin induced a release of AA from cortical neurons in primary culture (Fig. 5). This is consistent with the fact that PLA 2 activities are enhanced by Ca 2ϩ specifically interacting with a Ca 2ϩ -binding domain on cPLA 2 (39). Furthermore, Ca 2ϩ -regulated PLA 2 activities are present in primary cultures of cortical neurons (11). Thus, one possible step in the signal transduction cascade associated with the activation of PACAP I receptors would be the enhancement of the glutamate-induced Ca 2ϩ influx. In primary cultures of cortical neurons, the glutamate-evoked release of AA is mediated by the activation of AMPA/kainate as well as NMDA receptor subtypes (11). Activation of glutamate receptors enhances [Ca 2ϩ ] i by several mechanisms: (i) a depolarizationinduced gating of VSCC channels; (ii) in several neocortical neurons, which express specific AMPA receptors subunits, a Ca 2ϩ influx that occurs directly through the ionotropic receptor channel (40), or (iii) through the NMDA receptor channel (41). In the present study, an involvement of VSCC could be ruled out, since potent blockers of these channels, such as nifedipine and lanthanum, did not reduce the AA release evoked by glutamate plus PACAP. Furthermore, since the kainate as well as the NMDA-evoked release of AA were similarly potentiated by PACAP (Fig. 5) a mechanism acting on a specific receptor subtype can also be excluded, suggesting a level of interaction that is downstream of the receptor-induced Ca 2ϩ influx. Further strengthening this view is the fact that PACAP did not significantly affect the glutamate-induced enhancement in [Ca 2ϩ ] i (Table II). Even more striking is the observation that the release of AA evoked by ionomycin was also potentiated by PACAP (Fig. 5).
In summary, results reported here suggest that the substrate, whose activity is modulated following the activation of PACAP I receptors, is involved at a level downstream of the Ca 2ϩ influx. Whether this substrate is PLA 2 itself remains to be demonstrated. This study also shows that activation of PACAP I receptors alone (by VIP or PACAP) is not sufficient to promote the release of AA, whereas it clearly potentiated a Ca 2ϩ -dependent release of AA evoked by either glutamate or ionomycin.
Is VIP a cell-specific neuromodulator of the glutamate signaling? The morphological characteristics and physiological properties of VIP containing neurons in the neocortex, the region from which the cultures used in this study were prepared, have been characterized in detail (for review, see Ref. 42). In particular, VIP is almost exclusively localized to a homogeneous population of bipolar, radially oriented neurons (43). The dendrites of these neurons extend from the pial surface to layer V-VI arborizing to a maximum of 100 -120 m in the horizontal plane. The axons of VIP-containing neurons are ] i were measured in INDO-1-loaded cortical neurons exposed to agents for a period of 2 min in an incubation medium lacking Mg 2ϩ . Data represent [Ca 2ϩ ] i calculated as previously described (11), corresponding to n individual neurons analysed for each condition and chosen at random from 5 independent coverslips. No significant difference was found when glutamate and glutamate plus PACAP responses were compared (two-tailed unpaired Student's t test; p Ͼ 0.05). radially oriented and ramify within the radial domain of the dendritic arborization (43). These morphological characteristics imply that the actions of VIP released from VIP-containing neurons are confined within cortical columnar ensembles (for review, see Ref. 44). Since glutamate is released from activated afferent pathways to a given cortical area, VIP-containing bipolar neurons are ideally positioned to receive laminarly specified excitatory inputs from afferent pathways. Frequent asymmetrical (excitatory) synapses have been identified on VIPcontaining neurons (45); furthermore, glutamate (in addition to other depolarizing stimuli) has been shown to evoke VIP release (46). Thus, the activation of excitatory afferents can result in synergistic interactions between VIP and glutamate on potential common target cells within discrete cortical domains. From the data presented in this study, and from previous observations showing that VIP actually inhibits the glutamateevoked release of AA from astrocytes in primary cultures (5), it follows that the concomitant release of VIP and glutamate within discrete radially oriented cortical domains will selectively enhance the release of AA within neuronal rather than glial populations of activated cortical columns. This cell typespecific modulation of the glutamate signaling by VIP could result in a transient sharpening of the information flow to target neurons resulting in the enhancement of the signal-tonoise ratio of AA-mediated cellular processes. Indeed, AA has been shown to regulate the efficacy of glutamatergic neurotransmission by several mechanisms. AA increases (i) NMDA currents (possibly by interacting with the ion channel protein itself) (8,47) and (ii) the glutamate concentration in the synaptic cleft (possibly by enhancing synaptic release of glutamate and by inhibiting its reuptake) (48,7). In line with these observations are the results obtained in the hippocampus showing that certain components of long term potentiation appear to require AA formation (9,49). Therefore, data reported here suggest that VIP and PACAP could participate in the amplification of glutamate signaling, as already shown for the glutamate-induced c-fos expression (16).