Activation by P2X7 Agonists of Two Phospholipases A2 (PLA2) in Ductal Cells of Rat Submandibular Gland

Isolated ductal cells of rat submandibular gland phospholipid pools were labeled with [3H]arachidonic acid (AA). The tracer was incorporated preferentially to phosphatidylcholine (46% of the lipidic fraction). Extracellular ATP induced the release of [3H]AA to the extracellular medium in a time- and dose-dependent manner (EC50 = 220 μm). Among other agents tested, only 2′,3′-O-(4-benzoylbenzoyl)adenosine 5′-triphosphate (Bz-ATP) was able to mimic the effect of ATP (EC50 = 15 μm), without activation of phospholipase C. The purinergic antagonists oxidized ATP, suramin, and Coomassie Blue partly inhibited the response to 1 mm ATP and 100 μmBz-ATP; the response was also blocked by the addition of Mg2+ or Ni2+. Expression of P2X7receptor mRNA in these cells was confirmed by reverse transcription-polymerase chain reaction. In the presence of extracellular calcium, the phospholipase A2 inhibitor 2-(p-amylcinnamoyl)amino-4-chlorobenzoic acid (a nonspecific inhibitor), arachidonyl trifluoromethylketone (AACOCF3, an inhibitor of the calcium-dependent cytosolic PLA2(cPLA2)), and bromoenol lactone (an inhibitor of the calcium-independent PLA2 (iPLA2)) inhibited the release of [3H]AA induced by ATP and Bz-ATP. In the absence of extracellular calcium, the release of [3H]AA in response to the purinergic agonists was still observed; this response was not affected by AACOCF3 and completely blocked by bromoenol lactone. ATP and Bz-ATP stimulated a calcium-independent secretion of kallikrein, which could be blocked by BEL but which was enhanced by AACOCF3. It is concluded that the P2X7 receptor in ductal cells is coupled to kallikrein secretion through a calcium-dependent cPLA2 and a calcium-independent iPLA2.

concentration of calcium ([Ca 2ϩ ] i ) both in rat submandibular gland (RSMG) acini and ducts. This result suggested that ATP might regulate both phases of secretion. In acini, ATP increased the intracellular concentration of Na ϩ (19), activated the Na ϩ /H ϩ exchanger, and opened a chloride channel (20). It also inhibited the response to agonists activating the L-␣-phosphatidylinositol 4,5-bisphosphate-selective phospholipase C (21). Two purinergic responses were reported in pure RSMG ducts. At low concentrations, ATP activated a P2Y 1 receptor, while at high concentrations, it stimulated a P2X receptor (18,22). The ionotropic receptor triggered the secretion of kallikrein. The molecular characterization of the ionotropic receptor present on RSMG ductal cells has not been achieved yet. Both P2X 4 and P2X 7 transcripts have been observed in rat submandibular glands (23), and, according to Buell et al. (6), P2X 4 are expressed in RSMG acini but not in RSMG ducts. It can thus be speculated that the P2X receptors expressed in ducts are of the P2X 7 type.
The functional consequences of the activation of this receptor have not been determined. The coupling of some P2Y receptors to phospholipases C and phospholipases A 2 has been clearly established (24 -26). Reports on the regulation of the activity of phospholipases by P2X receptors are more scarce (27)(28)(29). The purpose of this work was to study the contribution of ionotropic purinergic receptors to the release of arachidonic acid by RSMG ductal cells. The phospholipases A 2 involved in this response and their role in the secretion in response to purines were also explored.
Isolation of RSMG Ducts-Male Sprague-Dawley rats (200 g) fed ad libitum and given free access to water were used. The rats were killed by exposure to diethylether, and the glands were immediately dissected and finely minced. The suspension of ductal cells was prepared as described previously (18) with some modifications. Briefly, the minced glands of one rat were digested in the presence of 2.6 mg of collagenase P (1.52 units/mg) for 20 min at 37°C under constant shaking (90 cycles/min) in 10 ml of HEPES-buffered saline (HBS) containing (mM): 24.5 HEPES, 96 NaCl, 6 KCl, 2.5 NaH 2 PO 4 , 11.5 glucose, 5 pyruvate, 5 glutamate, 5 fumarate, 1% (v/v) glutamine-free amino acid mixture (Life Technologies, Inc.), and 0.125% (w/v) bovine serum albumin (BSA). The pH was adjusted to 7.4 with NaOH. Ten minutes after the beginning of the digestion and at the end of the digestion, the cells were aspirated several times with 10-, 5-, and 2-ml glass pipettes. The suspension was then washed three times with an isotonic saline solution, resuspended in 10 ml of collagenase-free HBS and incubated for 10 min in the presence of 0.06 mg of DNase I (420 units/mg). The crude suspension was pipetted again, filtered through a nylon mesh and washed five times in a saline solution. Cells were resuspended in 4 ml of HBS and distributed in two tubes containing 6 ml of an isotonic Percoll solution (40%). The tubes were centrifuged at 4,000 ϫ g at 4°C for 10 min. At the end of the centrifugation, one population of cells had sedimented while the other had remained on the top of the Percoll. The microscopic examination revealed that the upper band was a ductal suspension while acini had sedimented through the Percoll. The protein content of the ductal suspension was 3.6 Ϯ 0.6 mg/pair of glands (n ϭ 10) measured with a Bio-Rad protein assay kit using ␥-globulin as standard. The assay of kallikrein (a ductal marker) confirmed that the upper layer was enriched 4.4-fold in kallikrein activity with respect to the crude cellular suspension. Ducts were aspirated, washed, and resuspended in HBS buffer for further experiments. H]AA) were added, and the suspension was incubated at 25°C with gentle shaking. At various times, 50-l aliquots were transferred to Eppendorf tubes and centrifuged at 16,000 ϫ g for 1 min. The pellet was washed with 0.9% NaCl, the lipids were extracted, and the incorporation of [ 3 H]AA was estimated. Labeling was very rapid, and steady state was reached after approximately 60 min. At that time, 80 Ϯ 3% of the total cellular precursor was incorporated into the lipidic fraction. Most of this radioactivity was distributed among various phospholipids; phosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol ϩ phosphatidyl serine incorporated, respectively, 46 Ϯ 1%, 18.8 Ϯ 0.3%, and 5 Ϯ 1% of the total lipidic fraction.
To study the metabolism of phosphatidylcholine and phosphatidyl inositol, ductal cells from 1 or 3 rats, respectively, were incubated for 90 min as described above in 1 ml of HBS containing 0.5 mM CaCl 2 and 4 Ci/ml [methyl- 14 Assay for [ 3 H]Arachidonic Acid Release-At the end of the isotopic labeling, ductal cells were washed and incubated for 1 h in 6 ml of calcium-free HBS in the absence of the tracer. After this incubation, the ducts were washed again and resuspended in 6 ml of calcium-free HBS. Assays were performed at 37°C under constant shaking and started by adding 100 l of cell suspension to 400 l of HBS containing CaCl 2 (final concentration 0.5 mM) and the agonists. The reaction was stopped by centrifugation for 30 s at 10,000 ϫ g. Four hundred l of the supernatants were transferred to vials and mixed with 4 ml of scintillation mixture. Radioactivity of the samples was measured in a scintillation spectrometer (model 2000 CA, Tri-Carb, Packard). In order to estimate the radioactivity already present in the medium at the start of each assay, several samples of the cellular suspension were taken during the experiments, mixed with 400 l of HBS, and directly centrifuged, and the radioactivity present in the supernatant was counted. These results were used as blank values and used to estimate the release of [ 3 H]AA during the incubation at 37°C. Under these assay conditions, 1 mM ATP or 100 M Bz-ATP promoted the release up to 8 Ϯ 1% (n ϭ 10) and 14 Ϯ 3% (n ϭ 10), respectively, of the incorporated isotope after 20 min of stimulation. In order to characterize the nature of the radioactive compound released by the ductal cells, the supernatant was extracted with chloroform:methanol:HCl (200:100:1). This extract was concentrated under vacuum, and the lipids were analyzed on TLC plates using two different elution systems: chloroform:methanol: acetic acid:water (90:8:1:0.8) for the separation of AA and its oxidation metabolites and petroleum ether:diethylether:acetic acid (60:45:1) for the separation of AA and diacylglycerol. Unlabeled AA was used as standard (R F ϭ 0.72 and 0.43 for the above described elution systems, respectively) and was visualized using iodine vapor. The bands were scraped from the plates, and the radioactivity was measured by liquid scintillation. The [ 3 H]AA released represented about 80% of the total counts found in the supernatant.
Extraction of Cellular Lipids and Their Separation by TLC-Total cellular lipids were extracted as described previously (30) with minor modifications. Briefly, pellets from ductal cells were resuspended in 0.9% NaCl and transferred to ice-cold glass tubes containing 500 l of chloroform:methanol:HCl (200:100:1). After 30 min in an ice bath, a final concentration of 0.1 M HCl was added and the tubes were vortexed and let at room temperature. The tubes were then centrifuged at 200 ϫ g for 15 min; 200 l of the organic phase were vacuum-dried in an automatic SpeedVac concentrator (Savant AS290), and resuspended in 40 l chloroform. Aliquots (30 l) were spotted on 20 ϫ 20-cm Silica Gel 60 F 254 plates. An ascending chromatography was performed using chloroform:methanol:acetic acid:water (75:45:12:3) as the eluant for the analysis of phospholipids. Spots were identified by comigration with authentic standards, which were visualized by exposure to iodine. The silica gel containing radioactivity was scraped into scintillation vials and quantified by liquid scintillation.
RNA Extraction and Reverse Transcription-Polymerase Chain Reaction-Total RNA was extracted from gland tissue and purified cell fractions using a genomic-free RNA extraction kit (GStract RNA isolation kit II, Maxim Biotech, Inc.) based on the Chomczynski and Sacchi method (31). First strand cDNA was synthesized from 1.5 g of RNA using random primers and avian myeloblastosis virus reverse transcriptase in accordance with the supplier's instructions. Specific primers for the detection of P2X 7 transcripts were designed using the Genetics Computer Group program GCG Primer. Upstream and downstream primers were based on unique sequences within exon 8 and exon 12, respectively, stretching bases 958 -980 (P2X 7 .up) and 1323-1341 (P2X 7 .down) of the P2X 7 sequence (GenBank accession no. X95882). The amplified product was thus predicted to be 384 base pairs in length. These regions correspond to the intracellular and carboxylterminal extracellular loops, respectively, of the corresponding amino acid sequence. PCR was performed as described previously (32) with some modifications. A hot-start touchdown PCR protocol with thinwalled tubes was employed involving denaturation at 95°C and extension at 72°C. The annealing stages began at 65°C, dropped by 1°C every two cycles until 56°C, at which all subsequent annealings were performed. Each step lasted for 50 s, and a total of 45 cycles were performed. The reaction mixture contained 1.5 mM magnesium. Amplified products were analyzed by electrophoresis in 1.8% agarose gels and viewed with ethidium bromide staining. A X174 HaeIII digest was used as a size standard. In order to confirm that the amplicons observed corresponded to RNA and not contaminating genomic DNA, we performed PCR under identical conditions with rat genomic DNA. The amplicon in this case was Ͼ Ͼ1400 base pairs, indicating the presence of the four predicted introns in the gDNA between exons 8 and 12 (data not shown). Thus, the PCR products obtained from the submandibular gland and from the acinar and ductal cells were derived from expressed mRNA and not genomic DNA.
Calcium Measurements-[Ca 2ϩ ] i was determined as described previously (18). Briefly, ducts from one rat were resuspended in 3 ml of HBS. Aliquots (1 ml) were incubated at 25°C with 1 ml of HBS buffer in the presence of 0.5% (w/v) BSA, 0.25 mM CaCl 2 and 2 M Fura-2/AM. After 45 min, 1 ml of the suspension was removed, washed with isotonic NaCl, and resuspended in 2 ml of magnesium-free HBS without BSA or amino acids. The cells were constantly stirred in the cuvette, and the excitation wavelength was switched every second from 340 nm to 380 nm (slit width 4 nm). The light emitted at 505 nm was recorded (slit width 8 nm). The voltage of the photomultiplier was 700 -750 V. At the start of each assay, the signal observed after excitation at 345 nm was arbitrarily set at 50% maximal scale by adjusting this voltage. At the end of the assay, the traces were calibrated with the successive addition of 0.5 mM digitonin and 40 mM EGTA (pH 8.5 with Tris). Autofluorescence was measured at both wavelengths in cells that were not loaded with the Ca 2ϩ indicator and was subtracted from all the data before calculation of the ratios. The calcium concentration was estimated by the ratio method (33), using a K d value of 263 nM.
Activity of the Polyphosphoinositide-specific Phospholipase C-The ductal cells obtained from 3 rats were labeled with myo-[2-3 H]inositol free of magnesium as described previously (21). At the end of the incubation, the cells were washed twice with 10 ml of HBS in the absence of labeled inositol and finally resuspended in 4.5 ml of HBS in the presence of 10 mM LiCl. They were preincubated in this medium for 10 min at 37°C. Aliquots (100 l) were incubated in a final 500 l of HBS medium in the presence of the tested agents for 5 min at 37°C. At the end of the incubation, the samples were centrifuged at 10,000 ϫ g for 30 s. The supernatant was discarded, and 500 l of 10% (w/v) ice-cold trichloroacetic acid were added to the pellet. The samples were centrifuged for 2 min at 10,000 ϫ g, and the supernatant was transferred to glass tubes. The pellets were washed again with 500 l of 10% trichloroacetic acid, centrifuged and the supernatants were pooled. The trichloroacetic acid was extracted with diethylether saturated with water. The extract was neutralized with 1 M KOH and diluted with 8 ml of 10 mM HEPES, 2 mM EDTA (pH 7.4). The inositol phosphates were isolated using a Dowex AG1-X8 column (34). The glycerophosphoinositol was eluted with 12 ml of 30 mM ammonium formate. The inositol derivatives (inositol mono-, bis-, and trisphosphates) were eluted simultaneously from the Dowex column using 12 ml of 0.7 M ammonium formate and 0.1 M formic acid. The radioactivity eluted off the column was quantified by liquid scintillation counting. Each determination was performed in triplicate.
Kallikrein Secretion-The ductal cells isolated from 1 rat were resuspended in 4 ml of magnesium-free HBS. The stimulation of the cells with the agonist was performed for 10 min as described for the [ 3 H]AA release assay. At the end of the incubation, the cells were centrifuged for 30 s at 10,000 ϫ g. Four hundred l of the supernatant were transferred to Eppendorf tubes and kept on ice. The kallikrein assay was performed as described previously (18). The assay was run in a 1-ml plastic cuvette using 300 l of Tris buffer (0.5 M, pH 8.2), 300 l of substrate solution (2 mg/ml N␣-benzoyl-DL-arginine p-nitroanilide), and 300 l of the sample. The absorbance was measured at 405 nm during 3 min in a Uvikon 943 spectrophotometer. The results were plotted as absorbance versus time, and the activity of the enzyme was calculated using the slope of the linear part of the curve. An aliquot of the ductal suspension was homogenized by sonication (2 pulses of 5 s at 8 m with a Soniprep 150) and assayed for kallikrein content. This result gave an estimate of the total content of kallikrein in the cell suspension. Results were expressed as a percentage of the total kallikrein content released during the incubation at 37°C. These results were corrected with regard to the kallikrein present in the medium prior to the incubation of the cells with the agonists.
Statistical Analysis-Unless otherwise indicated, values in figures are given as means Ϯ standard error (S.E.) of n experiments performed in triplicate with n different cell preparations. Statistical significance of the results was determined using a Student's test and represented by *** (p Յ 0.001), ** (0.001 Յ p Յ 0.01), * (0.01 Յ p Յ 0.05), and NS (not significant) (0.05 Յ p).  (Table I). Among the nonpurinergic secretagogues tested, only 100 M carbachol was able to induce a small effect (118 Ϯ 8%). The stimulation of the cells for 20 min with calcium ionophores (100 nM ionomycin or 1 M A23187) or the depletion of the intracellular calcium stores in cells incubated for 20 min with 1 M thapsigargin had no significant effect on the release of [ 3 H]AA from the cells (Table I). In separate experiments, it was shown that all these agents (epinephrine, carbachol, ionomycin, and thapsigargin) increased the [Ca 2ϩ ] i in the ductal suspension (data not shown).

Release of [ 3 H]AA from RSMG Ductal Cells in Response to Purinergic Agonists-RSMG ductal cells were prelabeled with
The release of [ 3 H]AA in response to ATP and Bz-ATP was dose dependent (Fig. 2). ATP significantly increased the release of [ 3 H]AA at 100 M, its half-maximal and maximal effects being observed at 220 M and 1 mM. The RSMG ductal cells were more sensitive to Bz-ATP, which significantly increased the release of  (Table II). Among these antagonists, oxidized ATP (oATP) is an inhibitor of ecto-ATPases and an irreversible inhibitor of P 2Z receptor in human lymphocytes (29) and in mouse macrophages (35). The cells were preincubated with oATP for 60 min before exposure to the purinergic agonist. By itself, oATP did not affect the basal release of [ 3 H]AA. At a 100 M concentration, oATP inhibited by 52 Ϯ 6% and 46 Ϯ 6% the [ 3 H]AA release in response to 1 mM ATP and 100 M Bz-ATP, respectively. At 500 M oATP, the ATP and Bz-ATP purinergic responses were inhibited by 77 Ϯ 7% and 76 Ϯ 9% (Table II). The second antagonist tested was suramin. This inhibitor of ecto-ATPases is also an inhibitor of purinergic receptors, which does not discriminate between the P2X and the P2Y subtypes and an inhibitor of P 2Z receptors (29,35). At a 500 M concentration, suramin inhibited by 50% the response to 100 M Bz-ATP or 1 mM ATP. Coomassie Blue was the last inhibitor tested. It has been reported that Coomassie Blue blocks the response of submandibular acini to Bz-ATP (21).

Effect of Divalent Cations on the Release of [ 3 H]AA in Response to Purinergic
Agonists-It is generally recognized that the active form of ATP on P2X 7 (P 2Z ) receptors is the free tetraionic form (ATP 4Ϫ ). The complexation of ATP by divalent cations might thus decrease the concentration of the active form. However, divalent cations might also block the nonspecific cation channels coupled to the ionotropic receptors. Magnesium (18) and nickel (36) were used in these studies. The concentrations of the different forms of ATP in the presence of either Ni 2ϩ or 5 mM Mg 2ϩ was estimated using the Solgaswater program (37) and the equilibrium constants from different well documented sources (38,39). According to our calculations (Table III) (Fig. 3). The half-maximal inhibitory concentrations were 0.18 and 0.24 mM with respect to 100 M Bz-ATP and 1 mM ATP. The response to the purinergic agonists was totally abolished in the presence of concentrations of Ni 2ϩ higher than 0.75 mM. The inhibition exerted by 0.75 mM Ni 2ϩ on the [ 3 H]AA release in Bz-ATP-stimulated cells was reversed by the addition of 1.5 mM DTPA (a chelator of metal ions) to the medium (Fig. 4, left panel). By itself, DTPA slightly affected (0.05 Ͻ p) the maximal release of [ 3 H]AA in the absence of Ni 2ϩ . This inhibition might be secondary to the chelation of external calcium (see above). In order to maintain the availability of Ca 2ϩ in the presence of 1.5 mM DTPA, these experiments were thus performed in the presence of 1 mM calcium, instead of 0.5 mM. That is the reason why the maximal effect on the [ 3 H]AA release is higher in these experiments than the results shown in Fig. 3. Considering that the addition of nickel in the medium did not significantly decrease the [ATP 4Ϫ ], nickel could exert its inhibition on the nonspecific cation channels coupled to the P2X 7 (P 2Z ) receptors.

TABLE II Effect of P2 receptor inhibitors and antagonists on the release of [ 3 H]arachidonic acid in rat submandibular ductal cells stimulated with ATP or Bz-ATP
This hypothesis was tested in parallel experiments, where the effect of nickel on the [Ca 2ϩ ] i in response to purines was tested. RSMG ductal cells were loaded with Fura-2 and resuspended in HBS medium in the presence of 1 mM CaCl 2 . When the cells were incubated in control conditions, the addition of 100 M Bz-ATP to the medium sharply increased the [Ca 2ϩ ] i from 202 Ϯ 41 to 646 Ϯ 94 nM (n ϭ 3, p Ͻ 0.01) and the [Ca 2ϩ ] i continued to rise slowly thereafter (Fig. 4, right panel). The addition of 1.5 mM DTPA at 10 min caused a slight drop on the calcium level (from 1264 Ϯ 42 to 1067 Ϯ 112 nM, 0.05 Ͻ p). This decrease might be due to the binding of some calcium to DTPA and the subsequent decrease of the free calcium concentration in the medium. When the RSMG ductal cells were resuspended in the presence of 0.75 mM Ni 2ϩ , the initial response to 100 M Bz-ATP was completely inhibited (Fig. 4, right panel). The [Ca 2ϩ ] i increased slowly reaching 367 Ϯ 95 nM at 10 min. It can also be noted that the intracellular Fura-2 was not quenched by nickel, confirming that at the opposite to calcium and manganese (18), this divalent cation cannot enter the RSMG ductal cells via the nonspecific cation channels coupled to the P2X 7 receptors. The chelator added 10 min after Bz-ATP almost totally reversed, for the first 10 s, the inhibition exerted by Ni 2ϩ on the response to Bz-ATP. Five minutes after the addition of DTPA, the [Ca 2ϩ ] i increased to 550 nM.
Expression of P2X 7 Receptor mRNA in RSMG Cells-The P 2Z receptor has been recently cloned in rat brain by Surprenant et al. (8) and termed P2X 7 according to the current classification. The expression of the P2X 7 -receptor mRNA in ducts and acini was demonstrated by RT-PCR using rat whole brain as a positive control. As shown in Fig. 5     Two other more specific inhibitors were tested next. AACOCF 3 is a potent, slow, tight-binding, reversible inhibitor of cPLA 2 and it proved to be highly specific for the 85-kDa cytosolic calcium-dependent PLA 2 (cPLA 2 ) (42). In our conditions, incubation of the cells with AACOCF 3 for 2 min prior to the addition of the purinergic agonists produced a concentration-dependent decrease in the stimulated [ 3 H]AA release (Fig. 6, left panel), whereas it had no effect on the basal [ 3 H]AA release. The inhibition averaged 15 Ϯ 3% at a 300 nM AACOCF 3 concentrations. Half-maximal inhibition could be observed at 2 M and at a maximal 100 M concentration, AACOCF 3 inhibited by 84 Ϯ 4% the response to 1 mM ATP (data not shown) and by 73 Ϯ 6% the response to 100 M Bz-ATP (Fig. 6, left panel). A full inhibition could never be observed. These results suggested that a cPLA 2 was at least partly responsible for the release of [ 3 H]AA in response to purinergic agonists but that some other PLA 2 might also be involved. We examined the possibility that a calcium-independent PLA 2 was present in RSMG ducts. The release of [ 3 H]AA was measured in the absence of calcium but in the presence of 0.5 mM EGTA. In these conditions, Bz-ATP could not increase the [Ca 2ϩ ] i (18, data not shown). The release of [ 3 H]AA in response to Bz-ATP was inhibited by 66 Ϯ 3% but the stimulatory effect of Bz-ATP on the release of [ 3 H]AA remained highly significant when compared with control (p Ͻ 0.001). The addition of various concentrations of AACOCF 3 to the medium did not affect this calcium-insensitive release of [ 3 H]AA in response to Bz-ATP (Fig. 6, left panel). This result confirmed that AACOCF 3 inhibited the cPLA 2 , which is a calcium-dependent PLA 2 . The presence of a calcium-insensitive (iPLA 2 ) in RSMG ducts was confirmed with BEL, an inhibitor of the iPLA 2 (43). In the presence of extracellular calcium, BEL slightly inhibited the basal release of [ 3 H]AA by 18% and 11% at 3 and 10 M, respectively, whereas it had no significant effect at 30 and 100 M. BEL dose-dependently inhibited the release of [ 3 H]AA in response to either ATP (IC 50 1 M; data not shown) or Bz-ATP (IC 50 2 M; Fig. 6, right panel). At a maximal 100 M concentration of BEL, the response to the two purinergic agonists was inhibited by 71 Ϯ 3%. In the absence of extracellular calcium, 30 M BEL fully inhibited the residual response to Bz-ATP (Fig. 6, right panel).
BEL and AACOCF 3 were also used to test the role of the iPLA 2 and the cPLA 2 in the remodeling of cellular phospholipids. RSMG ductal cells were preincubated for 5 min either in basal conditions or in the presence of 100 M BEL or AACOCF 3 and then incubated in the same conditions but in the presence of [ 3 H]AA. After 60 min, the radioactivity present in the phospholipid fraction was estimated. The presence of BEL and AACOCF 3 decreased by, respectively, 90 Ϯ 1% and 51 Ϯ 3% the incorporation of [ 3 H]AA in the phospholipids (data not shown). These data suggest that iPLA 2 (which is not specific for arachidonyl-containing phospholipids) has a constitutive activity which participates in the selective incorporation of arachidonic acid in phospholipids.

Lack of Correlation between the Activation of a Phospholipase C and the Release of [ 3 H]AA-As reported previously, ATP-and
Bz-ATP-stimulated [ 3 H]AA release could reflect phospholipase C activation by the purinergic receptors and subsequent stimulation of a calcium-or protein kinase C-dependent phospholipase A 2 (26,30). In order to determine the relationship between the [ 3 H]AA release and the phospholipase C activation, the cells were labeled with myo- [2-3 H]inositol. The stimulatory effect of Bz-ATP on [ 3 H]AA release from ducts was not secondary to the activation of phospholipase C, since 100 M Bz-ATP, which was the best agonist on [ 3 H]AA release, did not significantly increase the liberation of inositol phosphates; the radioactivity present in the fraction containing IP 1 ϩIP 2 ϩIP 3 in-creased from 200 Ϯ 47 dpm in the absence of Bz-ATP to 263 Ϯ 73 dpm in its presence (NS, 0.05 Ͻ p). At the opposite, 100 M carbachol, which did not significantly increase the release of [ 3 H]AA (Table I), increased the radioactivity present in the inositol phosphate fraction to 1,366 Ϯ 152 dpm (p Ͻ 0.001 when compared with control). These results confirmed that the increase in the [Ca 2ϩ ] i through the phospholipase C activation was not responsible for the Bz-ATP-stimulated [ 3 H]AA release.
Role of the Two Phospholipases A 2 in the Physiology of the Ductal Cells-It has been reported previously that the activation of ionotropic receptors increases the release of kallikrein by RSMG ductal cells (18). The role of the cPLA 2 and iPLA 2 in this response was tested next (Fig. 7). The ducts were preincubated with 100 M AACOCF 3 or BEL in a medium containing 0.5 mM Ca 2ϩ , and the release of kallikrein in the medium was measured after incubation either in basal conditions or in the presence of 100 M Bz-ATP. The two inhibitors had divergent effects on the secretion of kallikrein both in basal conditions and in the presence of the purinergic agonist. AACOCF 3 increased the basal release of kallikrein from 18 Ϯ 2% to 50 Ϯ 7% (n ϭ 3). It also enhanced the secretory response to Bz-ATP. Bz-ATP increased the secretion of kallikrein to 42 Ϯ 1% in the absence of AACOCF 3 and to 75 Ϯ 4% in its presence. Taking into account that AACOCF 3 is a structural analogue of AA, we tested whether exogenous AA could also increase kallikrein secretion. At 1, 10, and 100 M, AA had no effect on the secretion of kallikrein. BEL was a strong inhibitor of kallikrein secretion, which decreased to 3 Ϯ 1% in basal conditions and to 6 Ϯ 2% in the presence of Bz-ATP (Fig. 7). The inhibitory effect of BEL was dose-dependent (Fig. 8). Half-maximal inhibitory concentrations were 10 M both on basal and on Bz-ATP-stimulated kallikrein release. Considering that BEL was originally designed to inhibit the serine protease ␣-chymotrypsin (44), BEL could inhibit the activity of kallikrein rather that inhibit its secretion. To test this hypothesis, RSMG ductal cells were homogenized and the esterase activity of the homogenate was measured in the absence or in the presence of increasing concentrations of BEL. At a 100 M concentration, the inhibition exerted by BEL on the activity of kallikrein never exceeded 15% (data not shown). These results confirmed that BEL inhibited the secretion of kallikrein rather than the activity of the enzyme. The involvement of an iPLA 2 in the secretory response to Bz-ATP was further confirmed by measuring the release of kallikrein in the absence of extracellular calcium. In this condition, the basal release of kallikrein was not significantly affected and the secretory response to Bz-ATP was decreased from 46 Ϯ 2% in the presence of 0.5 mM calcium to 36 Ϯ 4% in the presence of 0.5 mM EGTA. The response was still highly significant when compared with basal release (p Ͻ 0.01, n ϭ 4). DISCUSSION The present results constitute the first demonstration by RT-PCR of the expression of P2X 7 in ductal cells of RSMG. This receptor is coupled with the activation of two distinct phospho- H]AA-labeled ductal cells were washed and resuspended in magnesium-free HBS medium containing 0.125% BSA. Cells were stimulated with 100 M Bz-ATP in the presence of 0.5 mM CaCl 2 in a final volume of 500 l. The reaction was carried out at 37°C and stopped by centrifugation at indicated times. The lipids extracted from the pellets were separated by TLC as described under "Experimental Procedures" and the radioactivity present in the bands of PC, PE, and PSϩPI was determined. Data represent mean Ϯ S.E. of three independent experiments performed in triplicate. The percentage of 3 H content with respect to unstimulated controls and statistical significance are also shown.  lipases A 2 . Indeed, ATP and Bz-ATP increased the release of [ 3 H]AA from prelabeled RSMG ductal cells. Based on the response to various agonists, on the sensitivity to antagonists like Coomassie Blue, or on the effect of divalent cations on the response to ATP, two types of purinergic receptors on RSMG ductal cells have been described (18). At low concentrations, ATP stimulated a G-protein-coupled P2Y 1 receptor that was very sensitive to 2-MeSATP. The occupancy of this receptor increased the intracellular concentration of IP 3 , and an intracellular pool of calcium was mobilized. The removal of calcium from the extracellular medium or the addition of magnesium did not suppress this response to ATP. This receptor is present in freshly isolated RSMG ductal cells, but its coupling with an effector disappears during cell cultures (22). At high concentrations, ATP and its analog Bz-ATP both induced a massive increase of the [Ca 2ϩ ] i , which was fully dependent on the pres-ence of calcium in the extracellular medium (18). This response was blocked by Coomassie Blue or by the decrease of the concentration of ATP 4Ϫ secondary to the addition of magnesium to the medium. The secretion of kallikrein by ductal cells was increased in response to P 2Z agonists (18). As shown in this work, ATP and Bz-ATP, two agonists of the P 2Z receptors, increased the release of [ 3 H]AA from RSMG ductal cells. This response was not reproduced by agonists of the P2Y receptors. The release of [ 3 H]AA was not secondary to the activation of a polyPI-specific phospholipase C since an agonist like carbachol, which activated this enzyme, had no significant effect on the release of [ 3 H]AA. It was inhibited by oATP, suramin, or Coomassie Blue, antagonists of the P 2Z receptors. The presence of magnesium in the medium also suppressed the release of [ 3 H]AA in response to ATP and Bz-ATP, further confirming that P 2Z receptors were involved in this response. Nickel also inhibited the response to the purines. We could calculate that at concentrations of nickel as high as 1 mM, the concentration of ATP 4Ϫ in the medium should be sufficient to fully stimulate the P 2Z receptors. RSMG ductal cells were impermeant to nickel since the ion did not quench the fluorescence of intracellular Fura-2 and since the inhibition by nickel was instantaneously reversed by the addition of DTPA, a non-permeant chelator of the ion. The inhibition by nickel could thus only be explained by a direct blockade by this cation of the nonspecific cation channel coupled to P 2Z receptors. From these results, it could be concluded that the activation of the nonspecific cation channels coupled to P 2Z receptors increased the release of [ 3 H]AA from RSMG ductal cells.
The [ 3 H]AA released in the medium in response to purinergic agonists originated from the cellular phospholipids, mainly PC. Indeed, the decrement in the content of [ 3 H]AA of this class of phospholipids was in agreement with the increment of the [ 3 H]AA released into the extracellular medium. Several lipases could be involved in the release of [ 3 H]AA from prelabeled phospholipids. It has been reported previously that P 2Z agonists can activate a phospholipase D in BAC1.2F5 macrophages (27) or in lymphocytes (29). The phosphatidic acid generated by this enzyme could be hydrolyzed by a phospholipase A 2 or could generate diglycerides after hydrolysis by a phosphatidate-phosphohydrolase. A diglyceride lipase would then liberate [ 3 H]AA from these diglycerides. Such a pathway has been described in peritoneal mast cells (45) and accounts for the release of [ 3 H]AA in response to muscarinic agonists in RSMG acinar cells (46). Preincubation of [ 3 H]AA-labeled RSMG ductal cells with propranolol (an inhibitor of the phosphatidate phosphohydrolase, 47) did not affect the subsequent ATP-or Bz-ATP-release of [ 3 H]AA (data not shown). The release of [ 3 H]AA was inhibited by inhibitors of PLA 2 . These results suggest that a PLA 2 and not a diglyceride lipase is responsible for the release of [ 3 H]AA. At the present time, we are unable to exclude the activation of a phospholipase D previous to the activation of PLA 2 . It should, however, be mentioned that: 1) the purines did not increase the release of free choline from phospholipids labeled with [ 3 H]choline; and 2) the activation of PLD by P 2Z agonists is dependent on bivalent cation influx (29), but the release of [ 3 H]AA in response to ATP could be observed in the absence of divalent cation in the medium and in the presence of EGTA, a chelator of these cations.
Two major groups of intracellular PLA 2 have been described. The calcium-dependent cytosolic PLA 2 (cPLA 2 ) has a molecular mass of 85 kDa. This enzyme is rather specific for 2-arachidonyl phospholipids. The NH 2 terminus of this enzyme is a calcium-binding domain analogous to the C2 domain of protein kinase C. This enzyme cannot be activated in the absence of an increase of [Ca 2ϩ ] i . After interacting with the C2 domain of the enzyme, calcium promotes its translocation from the cytosol to intracellular membranes, bringing the calcium-independent catalytic site to the membrane substrate (48). Various calciumindependent phospholipases A 2 have been described (49). Ankyrin domains have been described in these iPLA 2 , which could explain their self-association to form catalytically competent species and their interaction with proteins that may regulate their catalytic properties (50). Considering that no inhibitor is fully specific of the iPLA 2 , the release of [ 3 H]AA in the absence of extracellular calcium is the best evidence for the involvement of this enzyme in the hydrolysis of cellular phospholipids (49). In our hands, the removal of extracellular calcium inhibited but did not suppress the release of [ 3 H]AA in response to Bz-ATP. This result suggested that P2X 7 -receptors could activate both calcium-dependent and calcium-independent phospholipases A 2 . This was further confirmed by the use of AACOCF 3 (an inhibitor of cPLA 2 ) and BEL (an inhibitor of iPLA 2 ). Neither of these inhibitors is fully specific; it has been claimed that AACOCF 3 could inhibit iPLA 2 (49,51) and that BEL could inhibit the phosphatidate phosphohydrolase (52). In the presence of extracellular calcium, none of these two inhibitors could completely block the release of [ 3 H]AA. The inhibition exerted by AACOCF 3 was not observed in the absence of extracellular calcium, confirming that this inhibitor blocked a calcium-sensitive PLA 2 . The presence of a cPLA 2 in RSMG ductal cells was confirmed by Western blots using commercially available rabbit polyclonal antibodies (data not shown). The poor quality of the gels (due to the high background) precluded any study of the phosphorylated state of the enzyme by the electrophoretic mobility shift method. We were also unable to demonstrate the translocation of the enzyme from the cytosol to a membrane compartment in the presence of Bz-ATP (data not shown). The inhibition exerted by BEL on the release of [ 3 H]AA was still observed in the absence of extracellular calcium; the stimulation of [ 3 H]AA release by Bz-ATP was completely blocked by a combination of calcium removal and BEL. This confirmed that P2X 7 agonists activated an iPLA 2 . The mechanisms involved in the activation of iPLA 2 remain poorly understood. The enzyme has several ankyrin domains, which are involved in the interaction with other proteins like phosphofructokinase or in the interaction between several monomers of iPLA 2 (50). It has indeed been reported that the active form of the enzyme has an oligomeric structure. More recently, it was shown that the splicing of the introns can generate isoforms that have no catalytic activity but can interact with the iPLA 2 and inhibit its activity (53). It has been recently reported that in macrophages, P 2Z agonists activate a transcription factor, NF-B (54). Like iPLA 2 , this protein has several ankyrin domains (55), and splice variants of a precursor behave as antagonists (54). The activation of iPLA 2 and NF-B by P2X 7 agonists might involve similar mechanisms like activation of the ubiquitin-proteasome pathway (56).
Considering that P2X 7 receptors activate both a cPLA 2 and an iPLA 2 and promote the secretion of kallikrein, the correlation between these responses remained to be established. BEL was a strong inhibitor of both basal and Bz-ATP-stimulated release of kallikrein. Half-maximal concentrations of BEL on [ 3 H]AA release and on kallikrein secretion were similar, suggesting that the two responses were coupled. Furthermore, Ni 2ϩ was able to block in a dose-dependent manner the Bz-ATP-stimulated kallikrein secretion (data not shown). These results suggest that iPLA 2 was responsible for the release of kallikrein in response to purinergic agonists. The contribution of this phospholipase A 2 to amylase secretion from rat parotid acini (57) and to insulin secretion by pancreatic ␤ cells (58) has also been reported. This is consistent with the fact that calcium is not required for the fusion of secretory granules with the plasma membrane (59). Additionally, the presence of an iPLA 2 on the membrane of rat parotid secretory granules has been reported (60). The stimulation by AACOCF 3 of kallikrein secretion by RSMG ductal cells (this report) or of amylase by rat parotid acini (57) is more difficult to interpret. This response was observed at concentrations higher than 10 M. These concentrations have no lytic effect (measured by the release of lactate dehydrogenase or the release of Fura-2, data not shown). Considering that 1) the activation of the two phospholipases A 2 and the hydrolysis of PC did not lead to an accumulation of intracellular lysophospholipids, 2) lysophospholipids are known fusogens, 3) cPLA 2 is a strong lysophospholipase (it was recently shown that cPLA 2 could hydrolyze both sn1 and sn2 isomers of palmitoylglycero-3-phosphocholine; Ref. 61), 4) AACOCF 3 is an inhibitor of the lysophospholipase activity of cPLA 2 (61), 5) the major role of iPLA 2 would be the remodeling of phospholipids in membranes (49), and 6) macrophages expressing P 2Z receptors spontaneously fuse (62), we would like to propose the following model. The iPLA 2 would be activated by P2X 7 agonists. This enzyme, which has a weak specificity for arachidonyl-containing phospholipids, would release free fatty acids and sn1-acyl lysophospholipids (probably sn1-acyl lysophosphorylcholine). The lysophospholipids would favor the fusion between the membrane of the secretory granule and the apical plasma membrane. These lysophospholipids could be metabolized in two ways. 1) They might be reacylated to phospholipids (the Lands cycle) by an acylase specific for arachidonic acid. This would explain why a preincubation of the RSMG ductal cells with BEL abolished the incorporation of [ 3 H]AA in the cellular phospholipids. 2) The lysophospholipids might be hydrolyzed by the lysophospholipase activity of cPLA 2 . According to this second hypothesis, the secretory response to AACOCF 3 could be secondary to the accumulation of lysophospholipids generated by the constitutive activity of iPLA 2 and the inhibition of cPLA 2 by AACOCF 3 . However, other hypotheses cannot be excluded. It has been shown that AACOCF 3 can be reduced to AACOCH 3 , which, by itself, increased the stimulated level of AA in cell-based assay and the production of 12-hydroxyeicosatetraenoic acid by ionophorestimulated platelets (63). The increased level of AA in response to these metabolites could trigger exocytosis. Another hypothesis would be that AACOCF 3 behaves as an analog of AA and stimulates the secretory process, although in our hands the presence of AA did not affect the basal secretion of kallikrein.
In conclusion, P2X 7 agonists increase the release of [ 3 H]AA from RSMG ductal cells by activating two phospholipases A 2 , a cPLA 2 and an iPLA 2 . The activation of the iPLA 2 is responsible for the secretion of kallikrein by ductal cells in response to P2X 7 activation.