Dopamine D2 Receptors Potentiate Arachidonate Release via Activation of Cytosolic, Arachidonate‐Specific Phospholipase A2

Abstract: Several Gi‐linked neurotransmitter receptors, including dopamine D2 receptors, act synergistically with Ca2+‐mobilizing stimuli to potentiate release of arachidonic acid (AA) from membrane phospholipids. In brain, AA and its metabolites are thought to act as intracellular second messengers, suggesting that receptor‐dependent potentiation of AA release may participate in neuronal transmembrane signaling. To study the molecular mechanisms underlying this modulatory response, we have now used Chinese hamster ovary cells transfected with rat D2‐receptor cDNA, CHO(D2). Two antisense oligodeoxynucleotides corresponding to distinct cDNA sequences of cytosolic, AA‐specific phospholipase A2 (cPLA2) were synthesized and added to cultures of CHO(D2) cells. Incubation with antisense oligodeoxynucleotides inhibited D2 receptor‐dependent release of AA but had no effect on D2‐receptor binding or D2 inhibition of cyclic AMP accumulation. In addition, pharmacological experiments showed that D2 receptor‐dependent AA release was prevented by nonselective phospholipase inhibitors (such as mepacrine) but not by inhibitors of membrane‐bound, non‐AA‐specific PLA2 (such as p‐bromophenacyl bromide). cPLA2 is expressed in brain tissue. The results, showing that cPLA2 participates in receptor‐dependent potentiation of AA release in CHO(D2) cells, suggest that this phospholipase may serve a similar signaling function in brain.

Ca'-mobilizing stimuli to potentiate release of arachidonic acid (AA) from membrane phospholipids . In brain, AA and its metabolites are thought to act as intracellular second messengers, suggesting that receptor-dependent potentiation of AA release may participate in neuronal transmembrane signaling . To study the molecular mechanisms underlying this modulatory response, we have now used Chinese hamster ovary cells transfected with rat D2 -receptor cDNA, CHO(D2 ) . Two antisense oligodeoxynucleotides corresponding to distinct cDNA sequences of cytosolic, AA-specific phospholipase A2 (cPLA2 ) were synthesized and added to cultures of CHO(D2 ) cells. Incubation with antisense oligodeoxynucleotides inhibited D2 receptor-dependent release of AA but had no effect on D2 -receptor binding or D2 inhibition of cyclic AMP accumulation . In addition, pharmacological experiments showed that D2 receptor-dependent AA release was prevented by nonselective phospholipase inhibitors (such as mepacrine) but not by inhibitors of membrane-bound, non-AA-specific PLA2 (such as pbromophenacyl bromide) . cPLA2 is expressed in brain tissue . The results, showing that cPLA 2 participates in receptor-dependent potentiation of AA release in CHO(D2 ) cells, suggest that this phospholipase may serve a similar signaling function in brain. Key Words: Chinese hamster ovary cells-Transfected cells-Second messengers-G protein-linked receptors . J. Neurochem. 64, 2765Neurochem. 64, -2772Neurochem. 64, (1995 . Several neurotransmitters evoke the receptor-dependent hydrolysis of membrane phospholipids and the release of free arachidonic acid (AA) from neurons and astrocytes . In many cases, this reaction is thought to occur through the activation of phospholipase A2 (PLA2), a family of structurally heterogeneous lipases that catalyze the hydrolytic cleavage of glyeerophospholipid at the srt-2 position (where AA is most often esterified) yielding free fatty acid and lysophospholipid (for review, see Piomelli, 1993) .

Daniel Vial and Daniele Piomelli 2765
Based on their molecular structure, subcellular distribution, and phospholipid selectivity, members of the PLA2 family may be divided into two groups, high molecular weight, cytosolic PLA 2 and low molecular weight, membrane-bound (secretory) PLA 2 (Glaser et al,, 1993 ;Mayer and Marshall, 1993) . A cytosolic PLA2 with an apparent molecular mass of 100-110 kDa (by sodium dodecyl sulfate-polyacrylamide gel electrophoresis) that selectively hydrolyzes AA-containing phospholipids has been purified from several sources, and a lull-length cDNA encoding it has been isolated and sequenced (Clark et al ., 1990(Clark et al ., , 1991 . In vitro, the activity of this PLA2, termed cPLA2 , is stimulated by free Cat ' at concentrations (0. l -1 NM) that are likely to be reached when Ca -1 -mobilizing receptors are stimulated in intact cells, su esting that this activity may participate in receptor-~endent AA release (Clark et al., 1991 ) . In suppor his possibility, it was shown that when Chinese hat ovary (CHO) cells overexpressing transfected ( , were stimulated with extracellular ATP, an a at purinergic P2 receptors, release of AA was enhanced if compared with wild-type CHO cell et al., 1992) . In contrast with cPLA2 , low molect sight forms of PLA 2 (14-18 kDa) are either sect Ir membranebound, hydrolyze phospholipids w lesser degree of selectivity, and are activated in by free Cat' atmillimolarconcentrations (Waite,7) .Their pos sible participation in receptor-stimuli i release of AA and of other fatty acids has also been s igested (Pernas et al ., 1991 ) .
A group of G; protein-coupled neurotransmitter re-ceptors, including DZ -dopaminergic, a2-adrenergic, and m2and m4-muscarmic receptors, act synergistically with Ca t+ -mobilizing stimuli to produce AA release. For example, in transfected CHO cells labeled by incubation with [ 3H]AA, stimulation of D2 receptors potentiates the release of [`H ] AA evoked by application of ATP or Ca" ionophores but has no effect on basal ['H]AA release (Felder et a1 ., 1991 ;Piomelli et al ., 1991) . Like adenylyl cyclase inhibition, this response is likely to involve a protein of the G;/G family, as suggested by its sensitivity to pertussis toxin, and by the ability of GTP-y-S, a nonhydrolyzable GTP analogue to mimic it (Piomelli et al ., 1991 ;Di Marzo et al ., 1993) .
Receptor-dependent potentiations of AA release similar to those seen in transfected CHO cells have been demonstrated in neurons and astrocytes (Marin et al ., 1991 ;Schinelli et al ., 1994) . Despite its potentially important role in signal transduction, however, the molecular mechanism underlying receptor-dependent potentiation of AA release is still poorly understood . In the present study, we used CHO cells transfected with rat D 2 -receptor cDNA, CHO(DZ ), to examine the possible participation of AA-specific, high molecular weight cPLA Z in this facilitatory response . We report that antisense deoxyoligonucleotides (ODNs) directed against specific sequences of cPLA Z inhibit DZ receptor-dependent potentiation of AA release, without affecting inhibition of adenylyl cyclase activity or other second messenger pathways . The results, showing that cPLAZ may mediate a signal transduction pathway activated by D2 receptors, support a role for this phospholipase in brain signaling.

Receptor binding
CHO(DZ ) cells were incubated in DMEM containing 0.25% trypsin (GIBCO) for 5 min at room temperature, and the reaction was stopped by adding 10 ml of supplemented DMEM . Detached, intact cells were collected by centrifugation, resuspended in DMEM, and binding measurements were performed using '25 1-sulpride (Amersham) as described (Martres et al ., 1984). Curves were analyzed by nonlinear regression .
[3H]AA release After washing residual ODNs, cells (24-well plates) were labeled by incubation with I'H ] AA (Amersham; 200-220 Ci/mmol, 0.25 pCi/ml) in DMEM (1 ml) containing 0.2% fatty acid-free bovine serum albumin (BSA) for 2 h at 37°C . To eliminate unincorporated radioactivity, cells were washed with I ml of DMEM plus BSA, before incubating them for 30 min at 37°C in 1 ml of DMEM containing final concentrations of the appropriate drugs. [ 3H]AA release was determined by liquid scintillation counting of samples (0 .5 ml) of the incubation medium .
Cyclic AMP accumulation ODN-containing medium was removed, and cells (96well plates) were incubated for 10 min in 0 .15 ml of DMEM containing isobutylmethylxanthine (0 .1 mM) plus forskolin (10 pM) and final concentrations of drugs. After extraction in 0.1 M HCl (0 .1 ml), sonication, and neutralization of the tissue extracts, cyclic AMP concentrations were determined using a radioimmunoassay kit (Amersham), following the manufacturer's instructions .
['H ] Choline release CHO(D2 ) cells (24-well plates) were labeled overnight by incubation with ['H]choline (Amersham, 1 pCi/ml), in the presence of ODNs, washed, and incubated for 1 h at 37°C in DMEM containing choline (1 mM) . After 30-min exposure to drugs, radioactivity was determined in samples (0.5 ml) of incubation medium. ODN treatment had no effect on ['H]choline incorporation into CHO(D2 ) cells, as determined by measuring total radioactivity in cell extracts (data not shown) .
[ 3 H]AA uptake Cells, labeled by incubation with ['H]AA as described above, were washed with Cat}/Mg 2+ -free phosphate-buffered saline (GIBCO) and sonicated briefly in 0 .25 ml of the same medium. Radioactivity was determined in samples (25 pl) of cell extracts .

An antisense ODN directed against cPLA2 prevents ATP-stimulated [3 H ] AA release
In CHO cells, the stimulation of constitutive P2 -type purinergic receptors with extracellular ATP produces three prominent responses, activation of phosphoinositide-specific phospholipase C (PLC), elevation of intracellular Ca 2+ levels, and release of free AA from phospholipids (Gupta et al ., 1990) . Evidence indicates that P2 receptor-dependent AA release is mediated through the activation of cPLA2 . First, CHO(D2 ) cells express constitutively an mRNA encoding for cPLA 2 , as revealed by northern blot and RT/PCR analyses . Samples of poly (A) + RNA from CHO(D 2 ) cells were subjected to agarose gel electrophoresis, transferred onto nitrocellulose, and incubated with a 32p_ labeled cDNA probe for human cPLA2 (see Experimental Procedures) . The probe hybridized with a transcript of the expected size (3 .4 kb) (Fig . IA) . RT/PCR analysis, performed on samples of total RNA from CHO(D2 ) cells, revealed the presence of a 0 .5-kb product whose identity was confirmed by an additional PCR (Fig .  1B) . Second, when cPLA2 expression was enhanced in CHO cells by transfection with a cDNA encoding for human cPLAz , extracellular ATP was much more potent in evoking AA release than in nontransfected  (1) one or (2) two consecutive PCR with nested oligonucleotide primers (see Experimental Procedures for further details) . Results are from one experiment typical of two (northern blot) or three (PCR) . 2767 cells, or in cells transferred with a secretory PLA 2 (Lin et al., 1992) .
Antisense ODNs are effective in preventing expression of select target proteins, when they are either injected intracellularly or added to the medium of cells in culture (Sorscher et al., 1991 ;Wang et al ., 1992 ;Holopainen and Wojcik, 1993;Wahlestedt et al., 1993) . To determine whether antisense ODNs may inhibit expression of cPLA2 , CHO(D 2 ) cells were incubated for 24 h in culture medium containing one of three distinct 21-mer ODNs, termed antisense cPLA 2 -1 (complementary to a 5'-untranslated sequence in human and rat cPLAz cDNA but including the ATG initiation codon), sense cPLA2 -1 (homologous to the same cDNA sequence), and missense cPLA 2-1 (random nucleotide sequence) (50 yM, see Experimental Procedures) . To limit degradation of the ODNs, the cells were maintained in serum-free medium during treatment and labeled with a short (2 h) incubation in the presence of ['H]AA . Although these labeling conditions resulted in an incorporation of ['H ] AA into CHO (D2 ) lipids that was only 10% of that obtained with overnight labeling, the application of ATP (100 pM) was found to produce a significant release of ['H]AA from cells that had been exposed to either sense cPLA2 -1 or missense cPLA 2-1 (Fig . 2) . In the absence of any addition, quinpirole plus ATP produced on average a 400% increase of basal ['H]AA release (Piomelli et al ., 1991) . In contrast, the effect of ATP on ['H]AA release was completely abolished in cells treated with antisense cPLA 2-1 (Fig . 2) .
The ability of antisense cPLA 2-1 to prevent ATPinduced release of ['H]AA may have resulted from a nonspecific action of this ODN leading to defective signaling at the P 2 receptor . To address this possibility, we examined the effect of ATP on intracellular Ca 2+ levels in CHO(D 2 ) cells that had been exposed to antisense, sense, or missense cPLA2 -1 (24 h, 50 pM) . No difference was seen in the Ca t ' response to ATP (100 ICM) after these treatments, suggesting that antisense cPLA 2 -1 does not interfere either with the binding of ATP to the P2 receptor or with receptor coupling to phosphoinosifide-specific PLC and intracellular Ca 2 ' rises ( Table 1) .
The activities of arachidonoyl-CoA synthetase and arachidonoyl-CoA lysophospholipid transferase are necessary for the esterification of AA into phospholipid (Waku, 1992) . An undesired effect of antisense cPLA2 -1 on these enzyme activities may affect phospholipid labeling with  Table  1, no significant difference was seen in labeling among cells exposed to antisense, sense, or missense cPLA2-1 .
Antisense ODNs against cPLA2 prevent D 2 receptor-dependent [ 3H]AA release In CHO(D Z) cells, DZ-receptor agonists are potent in enhancing [3H]AA release, when such release is evoked by stimulating Cat'-mobilizing receptors (such as PZ receptors) or by applying a Cat' ionophore (Felder et al ., 1991 ;Piomelli et al ., 1991 ;Di Marzo et al., 1993) . Although these results had suggested the participation of a PLAZ activity in this response, the lack of selective PLA 2 inhibitors has hindered the characterization of this activity.
To examine a possible involvement of cPLA 2, we first studied the effect of antisense cPLA2-1 on DZreceptor potentiation of ATP-induced [ 3 H]AA release . As shown in Fig . 2, incubation with antisense cPLA2-1 markedly inhibited the release of 13 H ] AA produced by the application of ATP plus quinpirole (1 p,M) . In contrast, treatment with sense or missense ODNs had no effect. A similar inhibition of the response to quinpirole was obtained when [ 3 H]AA release was evoked by applying the Ca`' ionophore, A23187 (Fig. 3) .
The inhibitory effect of antisense cPLA 2-1 did not result from a nonspecific action on D,-receptor function. In support of this conclusion we found, first, that DZ -receptor binding was not affected by incubation with the antisense ODN . In two separate experiments, 121 I-sulpride binding to intact CHO(DZ ) cells was displaced by quinpirole with a K, of 0.21 -0.02 nM in cells exposed to antisense cPLA 2-1, and 0.18 -!-0.05 nM in cells exposed to sense cPLA,-I . Likewise, no significant difference in B ,x was seen with the two ODNs (antisense, 1,606 ± 274 sites/cell ; sense, 1,804  Fig . 1 legend . -305 sites/cell) . In addition, the ability of quinpirole to reduce forskolin-induced accumulation of cyclic AMP was not modified by treatment with any of the ODNs used (Table 1 ) . It is interesting that although antisense cPLA2 -1 inhibited the response to quinpirole, it did not completely abolish the release of ['H]AA induced by A23187 alone . Like other Ca 2+ ionophores, A23187 activates in a nonspecific manner several Ca" -dependent phospholipases whose activities may catalyze J'H J AA deacylation, including PLC and phospholipase D (PLD) (Piomelli, 1993) .
Next, we examined the effect of an additional antisense ODN, named antisense cPLA,-2, complementary to a translated sequence of cPLA 2 cDNA, which we used as a primer for RT/PCR analysis (see Experimental Procedures) . The results, depicted in Fig . 4, show that treatment with antisense cPLA2 -2 (24 h, 50 pM) decreased the release of ['H ] AA produced by quinpirole plus A23187, without affecting the response to A23187 alone . Sense and missense cPLA 2 -2 had no effect (Fig . 3) . Pharmacological blockade of cPLA2 Several alkylating agents that inhibit secretory PLA 2 by modifying covalently the enzyme at its active site have little or no effect on cPLA 2 activity. For example, the histidine reagent, p-bromophenacyl bromide, which is potent in inhibiting the activity of secretory PLA 2 , is ineffective on purified cPLA, (Mayer and Marshall, 1993) . In contrast, mepacrine, which interferes with the availability of phospholipid substrate, acts on cPLA 2 with an efficacy comparable with that observed with secretory lipases (Chang et al ., 1987;Xin and Mattera, 1992) .
We examined the effects of a series of chemically unrelated PLA2 inhibitors on the release of ['H ] AA from CHO(D,) cells, stimulated either with A23187 or with a combination of A23187 and quinpirole (Fig .  5) . ['H ] AA release was prevented most effectively by mepacrine (50 pM), and only partially reduced by aristolochic acid, dimethyleicosadienoic acid, and octadecylbenzoacrylic acid . p-Bromophenacyl bromide augmented rather than inhibited [ 3 H ] AA release from CHO(D 2 ) cells (Fig . 4), possibly by interfering with reuptake of ['H ] AA into cell lipids .

DISCUSSION
The experiments described above suggest that, in CHO(D,) cells, the facilitation of Ca -+ -evoked ['H]AA release produced by stimulating dopamine D2 receptors results from activation of AA-specific, high molecular weight cPLA2 . Three observations support this conclusion . First, cPLA 2 is constitutively expressed in CHO(D,) cells, as indicated by northern blot and RT/PCR detection of the iuRNA coding for this protein . Next, incubation of CHO(D2 ) cells with two distinct antisense ODNs directed against cPLA 2 markedly reduced D2 receptor-dependent potentiation of ['H ] AA release . Finally, the response was prevented by mepacrine, a nonselective PLA 2 blocker, but not by p-bromophenacyl bromide, a drug that inhibits membrane-bound PLA2 selectively over cPLA 2 .
As the conclusion of the present study relies mainly upon results obtained in knockout experiments with antisense ODNs, it will be important to examine, at the outset, the efficacy and selectivity of these agents . Next, we will discuss the potential significance of the results for signal transduction in the CNS .
Efficacy and selectivity of antisense ODNs against cPLA2 Synthetic ODNs in antisense orientation, i .e ., designed to hybridize to complementary sequences of mRNA, are widely used as inhibitors of translation in intact cells . Their inhibitory effects are thought to result either from translational block, by interference with ribosomal processing, or from induced degradation of target mRNA, by stimulation of ribonuclease H activity, which cleaves the RNA moiety of RNA/ DNA hybrids (for review, see Colman, 1990 ;Hélène and Toulmé, 1990) .
In cells in culture, two routes of ODN administration have been used successfully, i .e ., intracellular injection and addition to the culturing medium . Intracellular injection offers the advantage of a quantitative delivery of the intact ODN to its site of action, limiting the losses associated with incomplete membrane permeation and cellular degradation (Kleuss et al ., 1991) . Unfortunately, this technique is not suitable for many biochemical determinations in which the quantity of cells is a limiting factor . In such cases, long-term incubation of cells in a medium containing relatively high ODN concentrations may represent a feasible alternative approach, provided that appropriate controls for possible nonspecific effects of the ODN are performed (Sorscher et al ., 1991 ;Wang et al., 1992;Holopainen and Wojcik, 1993) .
In the present study, CHO(D,) cells were incubated for 24 h in serum-free medium containing different ODNs at concentrations up to 50 1,tM . Under these conditions, the release of [ 3 H ] AA induced by stimulating D2 receptors in the presence of ATP was found to be inhibited . To determine the selectivity of this effect, the following control experiments were performed : (1) the ODN antisense cPLA 2-1, directed against a 5' untranslated sequence of human and rat cPLA 2 cDNA sequence (Clark et al., 1991), prevented ATP-induced [ 3 H]AA release, mediated by cPLA 2 (Lin et al ., 1992), but had no effect on ATP-induced elevation of intracellular Ca 2+ levels (Gupta et al., 1990) ; (2) neither a sense-oriented ODN directed against the same cPLA 2 sequence nor a missense ODN (random nucleotide sequence) had such effect ; (3) treatment with antisense cPLA2-1 did not affect D 2 receptor binding, inhibition of forskolin-stimulated cyclic AMP accumulation, [ 3 H] AA incorporation into lipid or Ca 2+ ionophore-induced [ 3H]choline release (a measure of PLC and PLD activities) ; (4) an additional ODN, antisense cPLA 2-2, designed to hybridize to nucleotides 741-763 of cPLA2 , was also effective in reducing D= receptor-dependent [ 3 H]AA release, whereas its J. Neurochem ., Vol. 64, No. 6, 1995 D. VIAL AND D. PIOMELLI sense homologue and an additional missense ODN were not . Together, the results suggest strongly that the antisense ODNs used in this study exert their actions on [ 3H] AA release through a selective inhibition of cPLA2 expression .
Possible signaling functions of cPLA2 in the brain Several forms of receptor-dependent regulation of AA release have been described (see, for review, Piomelli, 1993) . Some neurotransmitter receptors are positively coupled to AA release . For example, glutamate acting at N-methyl-D-aspartate receptors evokes AA release from a variety of neuronal cell types (Dumuis et al ., 1988;Lazarewicz et al ., 1988 ;Sanfeliu et al ., 1990) . Other receptors may be linked to the inhibition of AA release . For example, in transfected CHO cells, activation of rat H Z receptors prevents the release of AA induced by the Ca'+ ionophore A23187 (Traiffort et al ., 1992) . A third group of receptors facilitates the release of AA evoked by stimuli (neurotransmitters or drugs) that elevate intracellular Ca 2+ levels . This permissive effect has been demonstrated in CHO cells transfected with G ;-coupled receptors, such as D Z -dopaminergic, a2 -adrenergic, m2and mamuscarinic, and 5-HT, A serotonin receptors (Felder et al ., 1991 ;Piomelli et al., 1991 ;Raymond et al ., 1992) . A G;/G,, protein may participate in transducing this effect, as indicated by the ability of pertussis toxin to inhibit the response and of a nonhydrolyzable GTP analogue, GTP-y-S, to mimic it (Piomelli et al ., 1991 ;Di Marzo et al ., 1993) . Although the precise molecular mechanism underlying the potentiating effect of DZ receptors on AA release remains unknown, two hypotheses are compatible with the available experimental evidence . G ;-coupled receptors, such as the D 2 receptor, may activate a "permissive" aG ;-protein that would enhance, but not directly evoke, cPLA 2 activity. Alternatively, G ;-coupled receptors may release "free" ß-y subunits, which have been reported to stimulate PLA, activity and AA release (Jelsema and Axelrod, 1987) .
Permissive actions of G;-coupled receptors on AA release have been shown to occur in neural cells . In primary cultures of striatal neurons, the release of AA evoked by applications of ATP or A23187 may be enhanced by addition of D2 -receptor agonists . In contrast, these drugs have no effect when applied alone to the neurons (Schinelli et al ., 1994) . Furthermore, in striatal astrocytes, AA release may be evoked by the combined application of somatostatin, a neuropeptide, and methoxamine, an a,-adrenergic agonist, drugs that have no effect when applied alone (Marin et al ., 1991) .
Despite its potentially important role in intracellular signaling, the molecular mechanism underlying receptor-dependent facilitation of AA release remains unknown. Central to the understanding of such mechanism is the unequivocal identification of the phospholipase activity involved . This identification is made difficult, however, by the existence of multiple path-ways of AA release and by the lack of selective enzyme inhibitors for each of these pathways . At least three esterase activities participate in deacylating AA-containing phospholipids, PLA,, PLC, and PLD (Piomelli, 1993) . In addition, most tissues, including the brain, express a variety of PLA, activities distinguished by their different subcellular distribution and regulatory properties (Woelk et al ., 1974 ;Gray and Strickland, 1982a,b ;Yoshihara and Watanabe, 1990 ;Hirashima et al ., 1992) .
The results reported in this study, obtained in a heterologous expression system, indicate that a single PLA, isoform, AA-selective cPI,A,, may mediate, on the one hand, receptor-stimulated AA release (Lin et al ., 1992), and participate, on the other, in receptordependent facilitation of such release . Expression of cPLA, in brain tissue (Yoshihara and Watanabe, 1990 ;Hirashima et al ., 1992) supports a similar role in neural cells .