Expression and Trans-synaptic Regulation of P2x4 and P2z Receptors for Extracellular ATP in Parotid Acinar Cells

Trans-synaptic regulation of muscarinic, peptidergic, and purinergic responses after denervation has been reported previously in rat parotid acinar cells (McMillian, M. K., Soltoff, S. P., Cantley, L. C., Rudel, R., and Talamo, B. R. (1993) Br. J. Pharmacol. 108, 453–461). Characteristics of the ATP-mediated responses and the effects of parasympathetic denervation were further analyzed through assay of Ca2+ influx, using fluorescence ratio imaging methods, and by analysis of P2x receptor expression. ATP activates both a high affinity and a low affinity response with properties corresponding to the recently described P2x4 and the P2z (P2x7)-type purinoceptors, respectively. Reverse transcription-polymerase chain reaction analysis reveals mRNA for P2x4 as well as P2x7 subtypes but not P2x1, P2x2, P2x3, P2x5, or P2x6. P2x4 protein also is detected by Western blotting. Distribution of the two types of ATP receptor responses on individual cells was stochastic, with both high and low affinity responses on some cells, and only a single type of response on others. Sensitivity to P2x4-type activation also varied even among cells responsive to low concentrations of ATP. Parasympathetic denervation greatly enhanced responses, tripling the proportion of acinar cells with a P2x4-type response and increasing the fraction of highly sensitive cells by 7-fold. Moreover, P2x4 mRNA is significantly increased following parasympathetic denervation. These data indicate that sensitivity to ATP is modulated by neurotransmission at parasympathetic synapses, at least in part through increased expression of P2x4mRNA, and suggest that similar regulation may occur at other sites in the nervous system where P2x4 receptors are widely expressed.

Relatively little is known about the physiological regulation of P 2x receptors. We have shown that removal of the autonomic innervation alters ATP responses in parotid acinar cells, which therefore provide a good model for investigating trans-synaptic regulation of receptor-mediated signaling. We previously described two ATP responses in rat parotid acinar cells, a P 2z/x7 type and a high affinity ATP receptor with distinctly different P 2x -like pharmacology (1).
Expression of the P 2x -type proteins in heterologous systems shows that each can form functional, homomeric, ligand-gated ion channels with a significant permeability to Ca 2ϩ ions as well as other inorganic cations, including Na ϩ and K ϩ . Properties of P 2x7 receptors differ substantially from the other subtypes of P 2x receptors and are similar to the P 2z receptor response that has been extensively described in macrophages and related cells. P 2z receptors not only function as cation-permeable channels but also are capable of mediating the flux of larger organic molecules of sizes ranging up to 900 Da without regard to the charge on the molecule (17,18). Although P 2z receptors are present on various cells of the immune lineage (17, 19 -24), only parotid acinar cells are directly innervated and regulated by neuronal activity.
Sympathetic and parasympathetic nerves control the secretion and composition of parotid saliva (25). Physiologically mediated changes in tonic patterns of nerve activity elicited by surgical denervation or by dietary manipulation alter the rat parotid gland, modifying acinar cell proliferation, cell size, and sensitivity of parotid secretion to neurotransmitters (26 -30). Parasympathetic denervation enhances secretory responses elicited in vivo by activation of calcium-mobilizing neurotransmitter receptors linked to phospholipase C (26 -28). Our previous denervation studies also established that trans-synaptic regulation of sensitivity to these neurotransmitters can be demonstrated in vitro in dissociated cell suspensions and showed for the first time that sensitivity to ATP is increased very dramatically (1,31). This suggests that P 2x purinoceptors are modulated by changes in synaptic activity and that ATP plays a role in the physiologically important regulation of food intake and metabolism.
Here, we further characterize and identify the high and low affinity ATP receptors in parotid acinar cells and determine whether the receptors are co-expressed or independently distributed. P 2x receptor types were classified in individual cells by pharmacology of the Ca 2ϩ i response to ATP, as analyzed by fluorescence ratio imaging of cells loaded with Fura-2. The properties of the responses are consistent with those observed for homomeric P 2x4 receptors and P 2x7 receptors in expression sys-tems. RT-PCR 1 amplification provided evidence for expression of the cognate mRNAs. Imaging experiments established that 1) the high and low affinity responses are independently expressed across acinar cells and likely mediated by P 2x4 and P 2x7 receptors, respectively; 2) denervation produces an increase in sensitivity to ATP (decrease in threshold) in individual cells; and 3) denervation leads to an increase in the number (proportion) of cells with high affinity ATP responses. Both responses to denervation could be explained, at least in part, by quantitative RT-PCR data showing that P 2x4 receptor mRNA increases following denervation. These data suggest that elevated P 2x4 receptor protein contributes to the observed increases in the sensitivity and total magnitude of the glandular response.

EXPERIMENTAL PROCEDURES
Dissociated Acinar Cell Preparation-Dissociated parotid acinar cells were prepared by trypsin and collagenase treatment as described previously (32). The cell pellet was suspended in oxygenated HEPES/ Ringer buffer of the following composition (in mM) 120 NaCl, 5 KCl, 2.2 MgCl 2 , 1 CaCl 2 , 20 HEPES, 5 ␤-hydroxybutyrate, 10 glucose, and 0.1% bovine serum albumin, pH 7.4. This preparation is composed of single cells and small clumps containing up to five cells; acinar cells comprise 85% or more of the dissociated cell preparation.
Parasympathetic Denervation-Unilateral deafferentation of one parotid gland of Sprague-Dawley rats (50 -100 g) was carried out surgically by avulsion of the right auriculotemporal nerve, which carries post-ganglionic parasympathetic nerve fibers. Tetracycline HCl (Polyotic, for veterinary use; 500 mg/liter drinking water; American Cyanamid, Wayne, NJ) was provided in drinking water ad libitum for 4 days following surgery. The contralateral gland served as control. Dissociated acinar cells were prepared 2-3 weeks after denervation.
Fura-2 Loading-Ca 2ϩ i was assayed in single cells using the fluorescent calcium indicator dye Fura-2 (Molecular Probes). Cells were loaded with Fura-2 acetoxymethyl ester (1.5 M) for 60 min at room temperature as described previously (32). Washed cells were suspended in Mg 2ϩ -free HEPES/Ringer buffer, and 150 l of Fura-2 loaded cells were plated on an acid-washed glass coverslip coated with concanavalin A (1 mg/ml) 5-10 min before mounting in the perfusion chamber for Ca 2ϩ i measurements. Ca 2ϩ i was estimated by the ratio method, using a K of 224 nM for Ca 2ϩ binding to Fura-2 (33). Maximum and minimum fluorescence values were obtained by the addition of ionomycin and EGTA, respectively, to the perfusion chamber at the end of experiments. In most cases, calcium responses are indicated as the ratio of the fluorescence values at 340 and 380 nm excitation. Rapid exchange of solutions was achieved by superfusion (4 ml/min) of cells in an open chamber.
Fluorescence Ratio Imaging-Cells were imaged using a Nikon Diaphot inverted fluorescence microscope equipped with a Xenon light source, Fura-2 barrier and emission filter sets, and a Cohu CCD camera. A computerized filter wheel (Sutter Instruments) equipped with an electronic shutter regulated excitation at 340 or 380 Ϯ 10 nm. Data were collected on a 486 computer equipped with an Itex 100 framegrabbing board. Software was developed to collect data simultaneously from as many as 50 cells, by placing circles of about 80 pixels in area over intracellular regions of fluorescent cells imaged on a 512 ϫ 512 video monitor. Fluorescent cells were visualized using a Nikon fluor 40ϫ oil immersion objective and excitation light at 380 nm. Cells with the morphology of polarized acinar cells were selected for analysis, including both fully dissociated individual cells as well as polarized cells organized in clusters of three to five cells around a lumen. Strings of duct cells were not analyzed. Ratio-pair data from all selected areas were usually collected every 5-7 s during the experiment. Data from a single frame were saved with no averaging. Background fluorescence was determined by placing a data collection circle in an area without cells; this value was subtracted for each data point before calculating the ratio of 340:380 measurements. Fluorescence images of cells with superimposed data collection circles as well as bright-field (Nomarski) images were saved and printed for each experimental run. Data were analyzed off-line.
Reagents were of analytical grade and were obtained from Sigma unless otherwise noted.
Reverse Transcription-Polymerase Chain Reaction-Total RNA was prepared by a single step method using Ultraspec, a commercially available isolation system (Biotecx, Houston, TX). The parotid glands were denervated as described above, removed 2-3 weeks later, and immediately immersed in liquid nitrogen. Control and denervated samples were crushed in liquid nitrogen and then homogenized in 4 ml of Ultraspec. Purified total RNA was quantified by absorbance at 260 nm, and 10 g was run on a formaldehyde gel to confirm the integrity of the RNA as indicated by the preservation of the 28 and 18 S rRNA. First strand cDNA was synthesized from 10 g total RNA using Moloney murine leukaemia virus reverse transcriptase (Life Technologies, Inc.) with random hexamers as primers (Promega). Rat brain and rat dorsal root ganglia cDNAs were prepared from total RNA obtained in the same manner. Genomic DNA was prepared from rat liver by a standard procedure (34). Briefly, the liver was excised, minced, frozen in liquid nitrogen, and crushed to a fine powder. The powdered tissue was suspended in digestion buffer containing 100 mM NaCl, 10 mM Tris-Cl, 25 mM EDTA, 0.8% sodium dodecyl sulfate, 0.1 mg/ml proteinase K, pH 8.0, and then incubated with shaking at 50°C for 18 h. An equal volume of phenol/chloroform/isoamyl alcohol was added, and the nucleic acids were extracted. The DNA was precipitated from the upper layer with 1.07 M ammonium acetate and 57% ethanol and resuspended in Tris/ EDTA buffer (pH 8).
The P 2x -receptor and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNAs were amplified with specific primers using Taq DNA polymerase (Life Technologies, Inc.) according to the manufacturer's instructions. The 50-l reaction mixture contained 10 l of DNA template, 1 ϫ Taq buffer, 800 M dNTPs, 1.5 mM magnesium, 50 pmol of each primer, and 2.5 units of Taq polymerase. The products were visualized by electrophoresis on a 2% agarose gel by ethidium bromide staining. The primer sequences and expected sizes of the products are shown in Table I. The identity of the products was confirmed by sequencing the amplified cDNA (Tufts DNA Sequencing Facility, Boston).
Quantification of mRNA Levels by Competitive PCR-The basis for competitive PCR is the inclusion of a known quantity of a synthetic internal standard that is amplified by the same primers as the cDNA of interest. Specific competitive templates were constructed to determine the amounts of GAPDH and P 2x4 -receptor cDNAs in the RT-DNA samples. For each experiment, a standard curve is obtained by adding known quantities of the internal standard to the PCR mixture containing the cDNA of interest. The level of amplification of each template depends upon its relative abundance in the mixture. The product from 1 The abbreviations used are: RT-PCR, reverse transcription-polymerase chain reaction; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; bp, base pair(s); GDP␤S, guanosine 5Ј-(␤-thio)diphosphate; PPADS, pyridoxal phosphate-6-azophenyl-2Ј,4Ј-disulfonic acid; H 2 DIDS, dihydro-4,4Ј-diisothiocyano 2,2Ј stilbenedisulfonate. the internal standard is shorter than the product from the cDNA, and they can be separated by gel electrophoresis (see Fig. 7a). The ratio of the intensities of the two bands is derived by densitometric measurements of the ethidium bromide-stained gel. Ratios from data of Fig. 7a are plotted in Fig. 7b. At a ratio of 1.0, the intensity of the two bands is the same, and the quantity of mRNA in the original sample may be read on the x axis. The competitive template for P 2x4 was constructed using a plasmid (pcDNA3, Invitrogen) containing the P 2x4 -receptor cDNA (courtesy of G. Buell,Glaxo (Ref. 12)). Two HaeII restriction endonuclease sites within the sequence amplified by PCR with primer set P 2x4 (up/lo) were employed to remove 200 bp from the cDNA. The fragments were then ligated back into the empty vector to create the template, P 2x4 -⌬HaeII. A similar template was generated from a plasmid containing the human GAPDH cDNA (pHcGAP; ATCC, Rockville, MD) by restriction digest with XbaI and BstXI to remove 95 bp. The resulting product was then blunted with T4 DNA polymerase (New England Biolabs) and ligated to form GAPDH ⌬X/B. The nucleotide sequence recognized by the primers is the same for both the human and rat GAPDH cDNA. Identical primers were used for competitive templates and the cDNA samples, and products differed only slightly in size, so both were amplified with the same efficiency. In some experiments, annealing between the heterologous strands of the competitor and cDNA products was observed. The third band generated in this manner was not included in the analysis because it contains equal quantities of both products and did not affect the quantification.
Heterologous Expression and Immunoblotting of P 2x4 -Receptor Protein-Immunoblotting of P 2x4 -receptor protein was carried out using samples from rat parotid gland and, as a control, human embryonic kidney 293 cells (293 cells), which heterologously expressed the receptor. A plasmid containing the full-length cDNA sequence for the rat P 2x4 receptor (p464) was transfected into the 293 cells using the calcium phosphate precipitation method of Okayama and Chen, as modified by Pritchett et al. (35). Protein samples from parotid glands were prepared by freezing and pulverizing the tissue in liquid nitrogen followed by direct transfer into SDS sample buffer containing 62.5 mM Tris-HCl, bromphenol blue, 0.1 mg/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 2 mM EDTA, pH 6.8. The sample was then homogenized for 30 s using a Brinkman polytron and boiled for 5 min. Extracts from 293 cells were prepared by scraping the cells from 100-mm tissue culture dishes in SDS sample buffer and homogenizing the sample as described above. Total protein was separated by 10 or 12% SDS-polyacrylamide gel electrophoresis, and the P 2x4 receptor was detected by immunoblotting with an antibody (final concentration, 2 g/ml; kindly donated by Gary Buell, Glaxo, Geneva) against the COOH-terminal of the rat protein. Crude antiserum and affinity-purified antibody were used in these experiments. Some blots were also reacted with antibody preabsorbed for 2 h at 20°C with the antigenic peptide (20 g/ml). Protein concentrations were determined by the Coomassie protein reagent assay (Pierce). P 2x4 -receptor-like immunoreactivity was visualized by the enhanced chemiluminescence detection method (Pierce).

RESULTS
The basis for increased sensitivity of denervated parotid acinar cells to ATP was investigated by characterizing Ca 2ϩ i responses in individual cells and by molecular analysis.
The properties of the Ca 2ϩ i responses have been partially determined in previous studies and can be evaluated through Fura-2 analysis. Although muscarinic, ␣-adrenergic, and substance P receptors in parotid acinar cells elevate Ca 2ϩ i through G protein-coupled activation of phospholipase C and subsequent production of inositol 1,4,5, trisphosphate (32,36,37), our results show that the ATP responses do not seem to be of the metabotropic type. Both ATP responses required extracellular Ca 2ϩ . ATP is not effective in stimulating inositol phosphate formation in this preparation (however, see also Ref. 38), and whole cell patch clamp recordings demonstrate that activation of whole cell currents by ATP does not involve a GDP␤Ssensitive step (36). P 2z responses in parotid cells resemble those in other cell types, requiring relatively high concentrations of ATP (100 M or greater) to stimulate influx of both Na ϩ and Ca 2ϩ (39). 3Ј-O-(4-benzoyl)benzoyl-ATP is the most potent and effective agonist, and the response is inhibited by high concentrations of divalent cations such as Mg 2ϩ , Brilliant Blue G, reactive blue 2, and H 2 DIDS (36,40). The second type of response to ATP is of higher affinity (EC 50 less than 10 M), insensitive to nucleotides UTP, GTP, adenosine, and ␣,␤-methylene ATP and weakly responsive to 2-methylthio-ATP. Further, it is neither activated by 3Ј-O-(4-benzoyl)benzoyl-ATP nor inhibited by high concentrations of divalent cations, Brilliant Blue G, reactive blue 2, and H 2 DIDS (1,36,40). An additional distinguishing feature of the two responses to ATP is that they are differentially modulated by protein kinase inhibitors (39).

Ca 2ϩ
i is elevated in parotid acinar cells by at least four different neurotransmitters, including agonists of the metabotropic G protein-coupled receptors and P 2 purinergic receptors for extracellular ATP. To determine whether the muscarinic and P 2 receptors are differentially distributed across acinar cells, Fura-2 loaded cells were perfused sequentially with maximally effective concentrations of the agonists carbachol (30 or 100 M) and ATP (30 or 300 M). Between agonist doses, cells were washed with buffer until Ca 2ϩ i values returned to basal levels. Almost all cells responded to maximal concentrations of carbachol, and cells that did not respond to carbachol rarely showed responses to ATP. Only the cells responding to carbachol were analyzed. Among the carbachol-responsive cells, 93% (75/81) also responded to 300 M ATP (Fig. 1), a concentration that would activate both high and low affinity receptors. This indicated that the majority of cells had P 2x receptors. Addition of a maximal dose of carbachol rapidly elevated Ca 2ϩ i to a peak value that then declined to a slightly lower steady-state level followed by recovery to a basal value when agonist was washed out.

Two Distinct ATP Responses (Receptors) in Individual Parotid Acinar Cells
Heterogeneity Shown by Dose-dependent Responses-Previous studies using cell suspensions demonstrated that the response to ATP is biphasic and that there are two pharmacologically distinct responses. However, under those assay conditions, the signal is a composite of responses from a large population of cells, and it is difficult to determine whether the two responses arise from subsets of cells with different receptors or whether the two receptors are homogeneously distributed across all responding cells. In the present studies, the distribution of the two distinct receptor types was characterized on individual cells. Cells were assayed in buffer without added Mg 2ϩ to optimize the P 2z response. Perfusion with both low and high concentrations of ATP identified some cells that displayed only the low affinity or only the high affinity response as defined below, but many cells showed both responses (Fig. 1). In 73% of the cells with responses to 300 M ATP (164/225 cells), a small, rapid elevation in Ca 2ϩ i was detected at low concentrations of ATP (1-30 M, high affinity response), and Ca 2ϩ i usually reversed to basal level when agonist was washed out (Fig. 1). At higher concentrations of ATP (300 M), Ca 2ϩ i increased more slowly and reached higher levels (low affinity response). After exposure to 300 M ATP, the elevated Ca 2ϩ i did not always reverse to basal level after washing out the ATP (Fig. 1).

Effects of Mg 2ϩ and High Concentrations of Divalent Cations on ATP Responses in Single Cells-In parotid acinar cell suspensions, the large increase in Ca 2ϩ
i in response to a high concentration of ATP (300 M) is primarily mediated by P 2z receptors. Thus, Mg 2ϩ or high concentrations of Ca 2ϩ would be expected to reverse or block this response in cells with low affinity receptors (see for example, Ref. 41). To test this, single cells were first exposed to 300 M ATP, and Mg 2ϩ was then added to the perfusion chamber (final concentration ϭ 1 mM). The effect of 300 M ATP was reversed by Mg 2ϩ (n ϭ 22). When 1 mM Mg 2ϩ was present together with ATP in solution, the low affinity response to 300 M ATP was blocked (Fig. 3b). The effect of a high concentration of divalent cation also could be demonstrated by stimulating cells with various concentrations of ATP, first at low concentrations of CaCl 2 (3 mM) and then at high concentrations of CaCl 2 (10 mM) (Fig. 2). When cells suspended in Mg 2ϩ -free KRH containing 3 mM CaCl 2 were exposed to 30 and 60 M ATP, Ca 2ϩ i rose rapidly and then recovered to basal when ATP was removed by washout. Under identical conditions, exposure of the same cells to 300 M ATP led to a larger and less rapid increase in Ca 2ϩ i . Raising the extracellular Ca 2ϩ concentration from 3 to 10 mM (Mg 2ϩ -free conditions, Fig. 2) did not significantly alter the response of the cells to 30 and 60 M ATP, but the large response to 300 M ATP was reduced to a value comparable with that elicited by 30 M ATP (n ϭ 10 cells). This depression by high Ca 2ϩ was not nonspecific, as the response to carbachol at the end of the experiment was normal. Mg 2ϩ also did not alter the response to low concentrations of ATP, indicating that divalent cations do not interfere with the high affinity response/receptor, consistent with our previous findings in cell suspensions. Inhibition of the low affinity response by the addition of 1 mM Mg 2ϩ (at 1 mM Ca 2ϩ ) or an increase of the calcium concentration from 3 to 10 mM might be due to alterations in the amounts of different species of ATP (such as ATP 4-, Mg 2ϩ ATP, Ca 2ϩ ATP), which vary in agonist efficacy. However, the effect of divalent cations on the low affinity response does not directly correlate with ATP 4concentration, and the response can still be detected at 3 mM Ca 2ϩ but not in 1 mM Ca 2ϩ plus 1 mM Mg 2ϩ , even though the ATP 4concentration is calculated to be the same (for 300 M ATP total, ATP 4ϭ 15 M in 1 mM Ca 2ϩ plus 1 mM Mg 2ϩ versus 15.3 M ATP 4in 3 mM Ca 2ϩ ) (42). These data are consistent with a recent report that divalent cations inhibit P 2z receptors principally by a direct effect on the receptor protein (41).
ADP Blocks Only the High Affinity ATP Responses-We have shown previously that the low affinity calcium response to Ͼ300 M ATP in a suspension of parotid acinar cells is activated only weakly if at all by ADP but that ADP blocked the more sensitive (high affinity) ATP response (1). In the present studies, we examined the effects of ADP on individual cells shown to have a high affinity response. Perfusion with 500 M ADP alone had variable effects on Ca 2ϩ i , with little or no effect on Ca 2ϩ i in most cells examined. However, in some cells a small response similar to the high affinity ATP (30 M ATP) response was observed. In these cells responding to 30 M ATP (n ϭ 18), addition of ADP reversed the Ca 2ϩ i response mediated by the high affinity ATP receptors even in the continued presence of ATP (Fig. 3a). However, ADP neither blocked (Fig. 3b) nor reversed (data not shown) the low affinity response mediated by high concentrations (300 M) of ATP. Under conditions where only the high affinity receptor is activated (in the presence of Mg 2ϩ ), no response to ATP was observed when 500 M ADP also was present (Fig. 3b). The response to 300 M ATP subsequently recovered after ADP was removed by washout (Fig. 3b). These results indicate that ADP is a weak partial agonist at the high affinity ATP receptor, inhibiting further responses to 30 M ATP but with no inhibitory effect on the low affinity receptor.
Brilliant Blue G Blocks Only the Low Affinity ATP Responses-Brilliant Blue G blocks the P 2z response as well as binding of [␣-32 P]ATP in rat parotid acinar cell suspensions (1,40). Consistent with these results, addition of Brilliant Blue G (1 M final concentration) reversed the low affinity response (in 33/34 cells), even in the continued presence of ATP (Fig. 3c), but had no effect on the high affinity ATP response (detected in Mg 2ϩ -containing solution; data not shown). The effects of Brilliant Blue G were irreversible (Fig. 3c).

Heterogeneous Distribution of the Two Distinct ATP Responses among Individual Cells
The effects of Mg 2ϩ , high concentrations of Ca 2ϩ , Brilliant Blue G, and ADP on single cell responses to ATP clearly dem-FIG. 2. High concentrations of cations selectively block the low affinity ATP 4؊ response. Cells were perfused with the indicated concentrations of ATP in Mg 2ϩ -free HEPES/Ringer medium containing either 3 mM CaCl 2 (broken bar) or 10 mM CaCl 2 (solid bar). Dotted lines across the trace indicate where perfusion with buffer continued but data collection was interrupted for 100 -246 s. Note that two distinct responses to ATP can be distinguished and that only the low affinity response is selectively blocked by 10 mM CaCl 2 in both cells shown in the trace. Carbachol was used at the end of the run as a control. The trace shown is representative of 10 cells from three different cell preparations.
onstrate the presence of two distinct responses, likely to correspond to P 2x4 and P 2x7 receptors (Table II). Evaluation of these characteristics across individual cells showed that the P 2x4and P 2z -type ATP responses were differentially distributed (Table III). Among the cells that responded to ATP, 47% (106/225) showed both the P 2x4 -and P 2z -type responses; 27% (61/225) had only the P 2x4 -type response, and 26% (58/225) showed only the P 2z -type response. This distribution was not an artifact of experimental design, as the diversity of response was observed on cells in the same experimental run, with some cells having both responses while other cells within the same acinar cluster had only one type or the other.

Molecular Characterization of P 2x Receptor Expression by RT-PCR and Western Blotting
Identification of the High and Low Affinity Receptors as the P 2x4 and P 2x7 Subtypes, Respectively-The low affinity/P 2z response to ATP in rat parotid acinar cells has the properties of heterologously expressed P 2x7 receptors. The physiological properties of the high affinity response are those of P 2x receptors, but previous data did not allow us to clearly identify which of the known cDNA clones is likely to mediate the response. To determine which P 2x receptor mRNA subtypes are expressed, we carried out RT-PCR on total RNA from rat parotid gland using oligonucleotide primers which discriminate between the seven identified P 2x receptor clones (Fig. 4). The major transcripts amplified from parotid gland cDNA were P 2x4 and P 2x7 . Fig. 4a shows the results obtained using the P 2x4 (up/lo) and P 2x7 (trunc) primers (see Table I); as predicted, the sizes of these transcripts were 618 and 578 bp, respectively, and their identity was confirmed by sequencing (very low levels of PCR products were detected with P 2x1 and P 2x3 primers, but identification was not supported by sequencing results). No product was amplified using primers specific for P 2x2 , P 2x5 , and P 2x6 , consistent with reports that P 2x receptor subtypes 1, 2, 3, 5, and 6 are not present in the rat salivary gland (15). We verified the functionality of primers that did not amplify a product from rat parotid cDNA by using rat genomic DNA (Fig. 4b) or cDNA from rat brain and dorsal root ganglia (Fig. 4c) as templates. With genomic DNA as a template for P 2x1 , P 2x2 , and P 2x4 primers, the products were all larger than those obtained using cDNA, indicating that the primers amplify across intron/exon boundaries in the P 2x receptor genes. This established that we do not have any genomic DNA contamination in our cDNA preparations. In the case of the P 2x2 receptor gene, whose sequence is known, the size of the product obtained using genomic DNA as a template (2218 bp) corresponds to that predicted (43) (Fig. 4b). Controls for P 2x3 , P 2x5 , and P 2x6 recep-Mg 2ϩ -free HEPES/Ringer buffer ([Mg] o ϭ 0; broken line). Cells were then stimulated in the absence of Mg 2ϩ with 500 M ADP followed by 300 M ATP solution also containing 500 M ADP as indicated. Note that ADP did not block the large response mediated by 300 M ATP in the absence of Mg 2ϩ . c, Brilliant Blue G (BBG) reverses the low affinity ATP response. After a control stimulation with 100 M carbachol, cells were perfused with 300 M ATP (Mg 2ϩ -free). After the peak elevation in Ca 2ϩ i was observed, perfusion was stopped by turning off the suction (under these conditions, cells continue to be exposed to 300 M ATP solution, which is present in the chamber), and Brilliant Blue G (1 M final concentration) was added to the chamber. Note that Brilliant Blue G reversed the low affinity response in the continued presence of ATP. Brilliant Blue G was washed out by perfusion with Mg 2ϩ -free HEPES/ Ringer buffer, and cells were stimulated again with 300 M ATP (Mg 2ϩfree) as indicated. No further response to ATP was seen. A second stimulation with carbachol was effective, but gave a slightly smaller response compared with the Ca 2ϩ i elevation mediated by the first stimulation. There seemed to be no high affinity ATP response in this cell. Trace shown is representative of 34 cells from three different cell preparations. i response in this cell, and no further response to 300 M ATP (in the presence of Mg 2ϩ , 300 M ATP activates only the high affinity receptor) was observed. Cells were then perfused with HEPES/ Ringer buffer (containing 1 mM MgCl 2 ) to remove nucleotides, and cells were perfused with 300 M ATP (containing 1 mM MgCl 2 ) subsequently. Note that the response to ATP recovered in the absence of ADP. After the peak response to ATP in Mg 2ϩ -containing solution was observed, both Mg 2ϩ and ATP were washed out by perfusing the cells with tors, using cDNA templates from brain and dorsal root ganglia, yielded products of the predicted size, that is 856, 613, and 1081 bp, respectively. The identity of these products was confirmed by sequencing. As further confirmation, other primer pairs targeted to different regions of the P 2x4 and P 2x7 receptors (see Table I) also amplified products of the expected sizes from rat parotid cDNA (data not shown). The products amplified by the P 2x7 (COOH-terminal) and P 2x7 (3Ј-untranslated region) primers include a sequence that encodes the necessary amino acids for the large pore-forming function of the P 2x7 /P 2z receptor (17). Each of these sets of primers yielded a single product, indicating that the P 2x7 -receptor subunit in rat parotid gland is probably identical to the protein in other tissues. These data support the conclusion that the low affinity response to ATP is mediated by the P 2x7 receptor and that the high affinity response is probably due to P 2x4 receptors. P 2x4 -Receptor Protein Is Expressed in Rat Parotid Gland-We have further examined the expression of P 2x4 receptors in the rat parotid gland by immunoblotting of proteins separated on denaturing SDS gels. Western blotting was carried out with a specific antibody raised against 16 amino acids in the COOH terminus of the receptor protein. The affinity-purified antibody recognizes a single protein in extracts from 293 cells transfected with a plasmid containing the P 2x4 -receptor cDNA (Fig.  5a). The antibody does not react with protein from 293 cells transfected with empty vector. Binding of the antibody is blocked by pre-incubation with the peptide to which the antibody was raised. The estimated size of the immunoreactive protein is substantially larger than the size predicted from the amino acid sequence (approximately 43 kDa), possibly due to post-translational glycosylation. Having confirmed that the antibody is specific for P 2x4 receptors in a heterologous expression system, we tested for the presence of this receptor in total protein extracts from rat parotid gland. Immunoblotting with crude antiserum detects a protein in parotid samples that is the same size as the protein recognized in 293 cells expressing the P 2x4 receptor, and binding is blocked by pre-incubation with the antigenic peptide (Fig. 5b). This protein was very susceptible to degradation, and it was necessary to flash-freeze and crush the tissue in liquid N 2 prior to adding sample buffer to obtain an intact sample.

Sensitivity
Receptor Sensitivity to ATP in Individual Cells-The threshold concentration of ATP for activation of individual cells that had P 2x4 -type responses (defined by response to 30 M ATP) was determined by perfusing the cells with concentrations of ATP ranging from 0.3 to 30 M. Ca 2ϩ i responses of 20 nM or more above basal were scored as positive. Cells responding to 0.3-1.0 M ATP were classified as supersensitive. Sensitivity of control cells was variable (Fig. 6). Activation by 0.3 M ATP was infrequent (3% of cells) in control preparations. At higher concentrations of ATP (1 and 3 M), the proportion of responsive cells increased to 31 and to 72%, respectively (the number of cells responding to 30 M ATP was set at 100%).
Parasympathetic Denervation Supersensitivity to ATP-Properties of the enhanced response to ATP were examined in individual cells after parasympathetic denervation.
2-3 weeks following parasympathetic denervation, the wet weight of parotid glands was reduced to 56.6 Ϯ 2.5% (n ϭ 7) of that of the contralateral glands. The fraction of cells with P 2x4 -type responses (to 30 M ATP) increased dramatically in denervated glands. For reasons that are not clear, in these experiments and all subsequent experiments, fewer control responses in individual cells Cells that responded to ATP (n ϭ 225) were further analyzed, and responses to ATP were classified as high affinity or low affinity based on the effects of Mg 2ϩ , high Ca 2ϩ , ADP, and brilliant blue G. Cells showing either low or high affinity responses were indicated as "ϩ", and cells that did not show high or low affinity responses were indicated as "Ϫ". Percentages in parentheses indicate the fraction of total cells that had each of the response distributions noted. Cells that did not respond to ATP were not included in this  4. PCR of P 2x -receptor DNA. Agarose gels stained with ethidium bromide show the products amplified by PCR using specific primers for the seven identified P 2x -receptor sequences (see Table I). a, the predominant transcripts in rat parotid gland cDNA are P 2x7 (X7, using P 2x7 (trunc) primers) and P 2x4 (X4, using P 2x4 (up/lo) primers) receptors. A faint band is detected with primers for P 2x1 receptor (X1), but its identity was not confirmed by sequencing. No product is detected with primers specific for P 2x2 receptors (X2). Amplification of GAPDH cDNA (G) was included as a control. b, the primers used in a were validated using rat genomic DNA. c, very low levels of PCR product were seen with P 2x6 , P 2x3 , and P 2x5 receptor primers in rat parotid gland. These primers amplified a product of the expected size from brain (B) or dorsal root ganglion (D) cDNA. cells had high affinity receptors compared with the original series of experiments evaluating the pharmacology of the low and high affinity ATP receptors. We found that among control cells that responded well to carbachol, only 20% also responded to 30 M ATP (46/233 cells). This does not seem to be a function of the age of the animal nor to be related to the method of preparing dissociated cells and was observed for cells prepared from unoperated animals as well as from control glands in operated animals. P 2x4 -type responses to 30 M ATP were much more frequent in cells from denervated glands and were observed in 67% of the cells that responded to carbachol (222/330, n ϭ 10 cell preparations).
Sensitivity to ATP also increased dramatically in cell preparations from parasympathetically denervated glands as shown by a drop in the threshold dose of ATP needed for activation. As noted above, in control preparations only 3% of cells with P 2x4 responses were activated by 0.3 M ATP (cells responding to 30 M ATP set at 100%), whereas 21% of cells in denervated preparations were responsive. At 1.0 M ATP, the proportion of responding cells increased from 31% in control cells to 54% in denervated cells. On the other hand, higher concentrations of ATP (3 M) triggered a response in nearly the same proportion of P 2x4 -responsive cells in control (72%) and denervated (80%) preparations. These data are expressed in terms of the fraction of cells that are activated at different threshold concentrations of ATP in Fig. 6.
Regulation of P 2x4 Receptor mRNA Levels following Parasympathetic Denervation-With supporting evidence that the high affinity response to ATP is mediated by the P 2x4 receptors, we tested the hypothesis that up-regulation of P 2x4 receptor expression underlies the supersensitivity of acinar cells to ATP following parasympathetic denervation. Competitive, quantitative PCR assays measured the relative levels of P 2x4 -receptor mRNA in control and denervated rat parotid glands as shown in Fig. 7. Rat parotid RNA preparations were obtained 2 weeks after parasympathectomy from control and denervated glands of eight animals. Fig. 7, a and b, show data obtained in a typical competitive PCR experiment using samples from a single animal. There was substantial variation in the yield of RNA from the glands, probably because of atrophy of the tissue and increased amounts of connective tissue in the denervated gland. We controlled for these changes by normalizing to the levels of mRNA for GAPDH determined by competitive PCR. The relative amount of P 2x4 -receptor mRNA, corrected for GAPDH mRNA content, was higher in the denervated gland than in the control tissue in every set of samples (n ϭ 8). There was some variation in these values, but on average, 2.8 times more P 2x4receptor mRNA was present in the denervated tissue (Fig. 7c) compared with the control side (n ϭ 7). In one animal, the level of P 2x4 receptor mRNA in the control gland was below detectable limits even though the levels of GAPDH mRNA indicated that the yield of cDNA in this preparation was not lower than average. P 2x4 receptor mRNA was up-regulated in the denervated gland from this animal, which is consistent with the data obtained from the other animals. The data from this one animal are not included in Fig. 7c.
Our results show that parasympathetic denervation causes up-regulation of P 2x4 receptor mRNA and support the hypothesis that the increased sensitivity of rat parotid acinar cells to ATP can be attributed, at least in part, to increased expression of P 2x4 receptor. DISCUSSION Electrolyte secretion in parotid acinar cells is mediated by mobilization of calcium from intracellular stores via receptors that are linked to phospholipase C. In contrast, extracellular ATP also increases Ca 2ϩ i but does so primarily by activating influx of extracellular Ca 2ϩ , independently of phospholipase C activation and without the participation of heterotrimeric guanine-nucleotide binding proteins (32,36). Examination of the The cells were transfected with the empty vector, pcDNA3 (Control), or the plasmid p464, which contains the cDNA for the P 2x4 receptor (P 2x4 ). Blots were probed with the affinity-purified antibody to the P 2x4 receptor (left) or with antibody preabsorbed with the antigenic peptide (right). b, immunoblot of protein from a rat parotid gland. The blots were probed with crude antiserum to the P 2x4 receptor (left) or with preabsorbed antiserum (right). The mobilities of the molecular weight standards are shown. FIG. 7. Competitive RT-PCR of rat parotid gland P 2x4 -receptor mRNA. a, ethidium bromide-stained agarose gels showing the results from control (left) and parasympathetic-denervated (right) parotid glands. In each lane, the top bands are the products from the target cDNA, and the distribution of responsiveness to neurotransmitters on individual cells indicates that almost all cells with a muscarinic response also respond to maximally effective concentrations of ATP. However, closer examination reveals heterogeneity in purinergic receptor subtype, magnitude of the response, and sensitivity to particular agonists. Regulation of these factors determines the physiological set point and capacity of the organ to respond to a stimulus.
Previous observations showing two different responses to ATP in suspensions of parotid acinar cells could not establish whether there were two separate populations of cells, one with high and another with low affinity receptors, or whether both receptors co-existed on the same cells (1,36,40). The present studies on individual acinar cells clearly demonstrate a stochastic distribution of two distinct purinergic responses and show that the response pathways are independent. Properties of the low affinity response are characteristic of the P 2z response and correspond to properties of the P 2x7 receptor. Taken together with RT-PCR and in situ hybridization evidence for P 2x7 mRNA, this suggests a functional role for P 2x7 in the parotid gland.
A high affinity response corresponds in its properties to the cloned P 2x4 receptor (11,12,14). P 2x4 receptors are widely expressed in the rat central nervous system, in peripheral ganglia, and in serous and mucous acini of rat salivary gland as well as respiratory and laryngeal epithelia (12). Patch clamp recordings of P 2x4 receptors transiently expressed in 293 cells are similar to antagonist-insensitive currents recorded in dissociated rat submandibular gland acinar cells, indicating that expression of a single type of subunit is adequate to convey properties of the receptor seen in vivo (12). The parotid high affinity ATP receptor described here has similar properties; it desensitizes very little, responds to ATP concentrations between 0.3 and 30 M, and is not activated by ␣,␤-methylene ATP, UTP, GTP, or adenosine. Analysis by RT-PCR confirms in situ hybridization reports that P 2x4 is expressed in rat parotid gland and that P 2x 1, 2, 3, 5, and 6 receptor subtypes are not present (15). Further, a protein immunoreactive with P 2x4 antibodies is observed in Western blots of parotid gland extracts, corresponding to the band seen in HEK293 cells heterologously expressing P 2x4 . However, properties of the P 2x4 receptor of parotid cells may differ somewhat from those reported in the 293 cell expression system and in submandibular cells, as parotid cells are not well activated by 2-methylthio-ATP. Further evaluation of the pharmacology using the antagonists PPADS and suramin, which distinguish P 2x4 and P 2x6 receptors from the others, could not be carried out in this analysis because these drugs absorb light in the wavelength range of the calcium-sensitive dyes.
Not all parotid acinar cells showed P 2x4 responses, similar to the limited distribution of P 2x responses reported in a submandibular acinar cell preparation (12). Although almost all parotid acinar cells have the P 2z response, as noted above, some have only the P 2x4 or only the P 2z response, supporting the idea that these receptors are distinct and independent of one another. Co-expression of two different ATP receptors may confer different temporal responses as well as the possibility of acti-vating different second messenger pathways.
Sensitivity to ATP was variable in parotid cells with P 2x4 responses. Few of the P 2x4 -responsive control cells were supersensitive to ATP, but the fraction of cells responding to ATP increased as the concentration of ATP was raised, consistent with the dose-dependent recruitment of cells with different sensitivities. Heterogeneity is also seen in pancreatic acinar cells, which show differential sensitivity of Ca 2ϩ i response to cholecystokinin (44) and to stimulation of amylase secretion by acetylcholine (45). Coupling of acinar cells through gap junctions has been reported in mouse salivary gland acinar cells (46), but in the present investigation, cell coupling did not seem to contribute to sensitivity. Ca 2ϩ i measurements were recorded from individual parotid acinar cells present as singlets, as part of doublets or triplets, or within acinar formations consisting of five to six cells. However, we did not observe a spread of Ca 2ϩ i from cell to cell in acini in the dissociated preparations from either contralateral unoperated parotid glands or denervated glands (data not shown), nor did we note any systematic difference in the sensitivity of cells within acini versus isolated single cells. Further, we observed acinar cells with different sensitivities within the same acinus, indicating that they were not coupled.
Homeostatic regulation of physiological events traditionally is thought to provide accommodation to changing conditions and to compensate for altered activity or demands on a particular system. Nerve activity influences both sensitivity and functional responses of parotid acinar cells. Removal of the parasympathetic innervation, the primary regulator of fluid, and electrolyte secretion (25,(47)(48)(49) results in adaptive "supersensitivity" or enhanced sensitivity of the gland to muscarinic agonists in vivo (26,50). Diet also tunes the animal's response to food ingestion as well as its metabolism and modifies salivary gland physiology (51,52). The development of supersensitivity to ATP following denervation suggests that parasympathetic activation patterns and signals modulate secretory activity through a pathway that is functionally mediated by ATP. The source of the ATP is not known, but it is likely to be the parasympathetic nerve terminals. ATP is stored at high concentrations and released along with acetylcholine from synaptic nerve endings (53) and with noradrenaline from sympathetic nerve terminals (54). However, there may be other non-synaptic sources of ATP. It has been suggested that ATP is transported through the cystic fibrosis transmembrane conductance regulator, for example, although this is still controversial (55). There is compelling evidence for the release of nucleotides from many tissues and cell types through non-lytic processes (56 -63). Wounded or damaged cells also may release high concentrations of ATP that function in pathological situations, and ATP has been implicated in programmed cell death in lymphocytes (7).
Increased P 2x4 signaling after denervation might be attributed to an increase in receptor number, to direct receptor modification, or through interaction with other proteins. Our data suggest that more cells respond to ATP because more cells express P 2x4 mRNA and receptor protein. Increased sensitivity may be due to higher receptor density per cell. Advances in development of new specific agonist or antagonist compounds lower bands are the products from the competitor DNA template, 418 bp for P 2x4 HaeII and 500 bp for GAPDH ⌬X/B. The approximate numbers of copies of internal standard that were added to each PCR reaction are indicated. b, quantification of P 2x4 -receptor (using P 2x4 (up/lo) primers) (left) and GAPDH (right) mRNA levels by densitometry of the bands in a. Lines shown are fitted by linear regression. The relative levels of GAPDH, a housekeeping gene, are not expected to change during denervation. The ratio of the band intensity for the competitive internal standard relative to the intensity of the band for the endogenous cDNA template was determined. At the point where the ratio is 1.0, the concentration of the endogenous template equals the concentration of added internal standard. Filled symbols are controls, and open symbols are denervated samples. c, averaged data for the effect of parasympathetic denervation on the relative levels of P 2x4 -receptor mRNA in rat parotid gland (n ϭ 7, see text). The levels of P 2x4 -receptor and GAPDH mRNA were determined as shown in a and b for eight animals. P 2x4 -receptor mRNA was normalized to GAPDH mRNA content to adjust for changes in the mass of the parotid glands and possible difference in the recovery of the mRNA from the control and denervated tissue. Solid bar is control and patterned bar is denervated data. Data report the mean Ϯ S.E. that can be used to assay receptor binding will be required to evaluate the role of receptor number or changes in binding affinity for ATP. Additionally, other factors may also contribute to the increase in functional receptors. The possibility that pre-existing receptors are activated following denervation cannot be ruled out. Recently, we reported that purinergic receptor responses in parotid acinar cell suspensions are greatly potentiated by treatment with protein kinase inhibitors, leading to the hypothesis that the response is modulated by protein kinases and that the dephosphorylated state of the receptor may be the most active form (39). A component of the sensitivity increase triggered by parasympathetic denervation could be mediated by alteration of the phosphorylation/dephosphorylation state of either the receptor or of another modulatory molecule that regulates receptor function.
The physiological role of extracellular ATP and the P 2x receptors in parotid acinar cells is still unclear. Synaptic modulation via depolarization and calcium entry may modify parotid secretion. At other synapses, ATP acts as a modulatory cotransmitter, and ATP has been reported to inhibit muscarinic and tachykinin responses through the P 2z receptor in submandibular or parotid cells (38,64,65), although relatively high concentrations of ATP are required for activation in the presence of divalent cations.
In conclusion, our results clearly indicate the presence and heterogeneous distribution of both P 2x and P 2z purinoceptor responses to extracellular ATP in parotid acinar cells. These data correlate with the robust expression of P 2x4 and P 2x7 mRNA in parotid glands, suggesting that ATP-gated channels containing P 2x4 and P 2x7 subunits mediate physiological responses. Parasympathetic denervation increases the number of cells with P 2x4 ATP responses, the level of P 2x4 mRNA, and the proportion of supersensitive cells, suggesting an important physiological role for these receptors in parotid cells. Transsynaptic regulation is probably mediated by increased expression of receptors, but modulation by other mechanisms may also contribute. Cloning of the P 2x7 receptor will now permit further studies of how this channel is regulated and how cellspecific interactions can generate differences in pore-forming abilities of the P 2z receptor observed in parotid and other cell types (66 -68). These studies provide the basis for developing molecular and protein probes for a better understanding of signal transduction mechanisms in the different ligand-gated channel and G protein-coupled receptors of this family and provide the foundation for structural and functional studies. Examination of regulation by neuronal signaling will give further insight into the physiological role and mechanisms of modulation of receptors in this important and widely distributed receptor family.