Purinoceptors Evoke Different Electrophysiological Responses in Pancreatic Ducts

In epithelia, extracellular nucleotides are often associated with regulation of ion transporters, especially Cl− channels. In this study, we investigated which purinoceptors are present in native pancreatic ducts and how they regulate ion transport. We applied whole-cell patch-clamp recordings, intracellular Ca2+ and pH measurements, and reverse transcription-polymerase chain reaction (RT-PCR) analysis. The data show two types of purinoceptors and cellular responses. UTP and ATP produced large Ca2+ transients, a decrease in intracellular pH, 8–10-mV depolarization of the membrane voltage, and a decrease in the whole-cell conductance. The membrane effects were due to closure of K+ channels, as confirmed by dependence on extracellular K+. UTP/ATP effects could be associated with P2Y2 purinoceptors, and RT-PCR revealed mRNAs for P2Y2 and P2Y4 receptors. On the other hand, 2′,3′-O-4-benzoylbenzoyl-ATP induced Ca2+influx and ∼20-mV depolarization of the membrane voltage with a concomitant increase in the whole-cell conductance. These effects were dependent on extracellular Na+, not Cl−, indicating opening of cation channels associated with P2X7purinoceptors. RT-PCR showed mRNAs for P2X7 and P2X4 receptors. In microperfused ducts, luminal (but not basolateral) ATP caused large depolarizations of membrane voltages recorded with microelectrodes, consistent with luminal localization of P2X7 receptors. Thus, P2Y2 (and possibly P2Y4) purinoceptors inhibit K+ channels and may not support secretion in native ducts. P2X7 (and possibly P2X4) receptors are associated with cation channels and may contribute to regulation of secretion.

Extracellular nucleotides are regulators of a wide range of cellular functions in various cells, including regulators of epithelial transport (1). They act through specific P2 receptors, which, according to their molecular structure and signal transduction, have been divided into two distinct receptor families: G protein-coupled receptors (P2Y receptors) and ligand-gated ion channels (P2X receptors) (1). In epithelia cultured from upper respiratory airways, intestine, and exocrine glands, including pancreas, UTP and ATP stimulate Cl Ϫ transport through P2Y receptors, which were described pharmacologically as P2U receptors and, in many cases, may correspond to the cloned P2Y 2 receptors (2)(3)(4)(5)(6)(7)(8)(9)(10)(11). However, similar action of pyrimidines on Cl Ϫ transport could also be exerted through P2Y 4 or P2Y 6 (12,13). Due to their effects on Cl Ϫ transport, nucleotides were proposed as therapeutic agents for treatment of cystic fibrosis, as they could bypass the defective function of the cystic fibrosis transmembrane regulator Cl Ϫ conductance and restore secretion by activation of a Ca 2ϩ -dependent Cl Ϫ conductance (8). Interestingly, recent studies on exocrine glands indicate that the P2Y receptors and associated signal transduction pathways are not static, but dynamic. For example, the P2Y 2 receptor expression increases dramatically in tissue culture of salivary gland cells (14), and the P2Y 1 receptor activity changes during the gland development (15). Thus, it is important to consider which P2 receptors are present in freshly prepared cells from exocrine glands. Although there is evidence for some P2Y-type receptors (4,11,16,17), more commonly accounted purinoceptors are those linked directly to cation influx (18 -23). Occupation of these purinoceptors leads to stimulation of Cl Ϫ transport, volume changes, and secretion of enzymes and peptides (19,21,24), but in some cases, interference with the Ca 2ϩ signaling utilized by other receptor pathways has been reported (25,26). These receptors have been described pharmacologically as P2Z receptors, but it is unclear if these correspond to the pore-forming P2X 7 receptors originally found in immunoreactive cells (1). Interestingly, in situ hybridization analysis shows P2X 4 receptors on salivary gland acini, but not on ducts (27). Thus, from pharmacological studies, it seems that exocrine glands can express receptors belonging to P2X and P2Y families, but their identity, localization, and function are unclear.
Pancreatic ducts modify enzyme-and Cl Ϫ -rich secretion originating from acini by adding a bicarbonate-rich fluid. Bicarbonate secretion is achieved by coupling of ion transport through the luminal cystic fibrosis transmembrane regulator Cl Ϫ channels and the luminal Cl Ϫ /HCO 3 Ϫ exchanger (28 -30). The driving force for the Cl Ϫ exit is kept by the K ϩ exit via the basolateral K ϩ channels (29,(31)(32)(33). The main regulators of pancreatic bicarbonate secretion are secretin, vasoactive intestinal peptide, acetylcholine, and noradrenaline (30,34,35). In earlier studies, we showed that isolated ducts also respond to ATP by releasing stored Ca 2ϩ and initiating Ca 2ϩ influx through channels different from those used by acetylcholine (36 -38). Furthermore, we demonstrated that ATP inhibits secretin-evoked changes in the membrane voltage (37). We used Ca 2ϩ signals and pharmacological profiles to describe two types of receptors: one responding to UTP and ATP and belonging to the P2Y superfamily (P2Y 2 -like receptor) and another responding to BzATP 1 and ATP and belonging to the P2X superfamily (P2X 7 -like receptor) (36). Although epithelial P2 receptors are associated with Ca 2ϩ signaling in ducts, it is not clear how this is further translated into ion transport and thus secretion in native pancreatic ducts. Therefore, the aim of this study was to use the patch-clamp methods and intracellular Ca 2ϩ measurements to elucidate how purinergic receptors regulate ion transport in pancreatic ducts. Furthermore, we aimed at identifying these purinoceptors using molecular biological methods, which were applied to this epithelium for the first time. For this purpose, we used pancreatic duct fragments prepared from collagenase digests of whole rat pancreas. The advantage of this preparation is that we can obtain pure duct cells free of the surrounding connective tissue, blood vessels, and nerves, all likely purinoceptor contaminants. These ducts retain receptors, Ca 2ϩ signals, and electrophysiological responses comparable to the microdissected ducts (29,30,32,34,39). The disadvantage of the preparation is that ducts are too fragile for microperfusion; and thus, the polarity of the responses cannot be studied.
In this study, we show that UTP/ATP inhibit K ϩ channels and thus most likely cannot support secretion. On the other hand, BzATP stimulates cation conductance and may therefore regulate secretory processes stimulated by other secretagogues. RT-PCR analysis revealed that ducts contain mRNAs for P2Y 2 and P2X 7 receptors, likely candidates for the observed effects, as well as mRNAs for P2Y 4 and P2X 4 receptors.

EXPERIMENTAL PROCEDURES
Materials-All standard chemicals including collagenase V, trypsin inhibitor, UTP, ATP, and BzATP were obtained from Sigma (Copenhagen, Denmark). Fura-2/AM and BCECF/AM were obtained from Molecular Probes, Inc. (Leiden, The Netherlands). Tissue culture media and primers for PCR were provided by Life Technologies, Inc.
Preparation of Ducts-Female Wistar rats (100 -200 g), kept on a standard laboratory diet, were killed by cervical dislocation. The pancreas was excised and subjected to collagenase digestion as described previously (39). Briefly, the pancreas was cut into small pieces and incubated with collagenase V (1.3 mg/ml) in Dulbecco's modified Eagle medium 1000/Ham's F-12 mixture containing trypsin inhibitor (0.17 mg/ml). Pancreatic ducts compose Ͻ5% of the total pancreatic tissue, and single intercalated and small intralobular duct fragments (outer diameter of 10 -40 m and length of 100 -200 m) were identified with the aid of a dissection microscope. Any remaining loose connective tissue and small blood vessels were stripped with sharpened forceps. Subsequently, ducts were transferred into a chamber placed on an inverted microscope (Axiovert 100TV (Carl Zeiss, Inc.) or DM IRBE (Leica Inc.)) and inspected with 40ϫ or 63ϫ objectives for "cleanliness." Fig. 1 shows an example of small intralobular ducts obtained using these methods. These "clean" ducts were used for patch-clamp experiments, for molecular biology, and for most fluorescence measurements. For physiological experiments, suction pipettes held ducts on a side such that both the luminal and basolateral sides of the epithelium were accessible to the bathing solutions. For one series of experiments, small intralobular and extralobular ducts (outer diameter of 20 -50 m) were microdissected from fresh rat pancreas. These microdissected ducts were held between concentric pipettes at each end, and the lumen was perfused according to methods described earlier (32). Such ducts were used for microelectrode recordings.
During all physiological experiments, the chamber was continuously perfused at a rate of 10 -15 ml/min to avoid mechanical disturbances, which can cause release of ATP. 2 Control solutions contained 145 mM Na ϩ , 3.6 mM K ϩ , 1.5 mM Ca 2ϩ , 1 mM Mg 2ϩ , 125 mM Cl Ϫ , 25 mM HCO 3 Ϫ , 2 mM phosphate, and 5 mM glucose. The pH was equilibrated to 7.4 with 5% CO 2 in O 2 . The temperature was kept constant at 37°C during all experiments. In some experiments, K ϩ , Na ϩ , or Cl Ϫ concentrations were modified as follows: K ϩ was increased from 3.6 to 20, 25, or 50 mM by equimolar substitution for Na ϩ ; Na ϩ was decreased from 145 to 5 mM and substituted with N-methyl-D-glucamine; and Cl Ϫ was decreased from 125 to 5 mM and substituted with gluconate. The pancreatic ducts were stimulated with ATP, UTP, and BzATP in concentrations of 0.1 mM. In a few experiments, concentrations of agonists were varied from 10 Ϫ8 to 10 Ϫ3 M, except for BzATP, which was not used at concentrations higher than 0.1 mM due to the cost.
Electrophysiological Measurements-The cell responses to nucleotides were monitored in whole-cell nystatin patch-clamp recordings adopted for pancreatic ducts (39). Patch pipettes had resistances of 3-5 megaohms, and the initial seal was at least 1 gigaohm. The success rate of obtaining seals was Ͻ10%, and stable cell membrane voltage recordings were obtained only in a fraction of these. The cell membrane under the pipette was permeabilized to make direct electrical access to duct cells. Nystatin (0.02-0.1 mM) was dissolved in a pipette solution adjusted to pH 7.2 that had the following composition: 11 mM Na ϩ , 125 mM K ϩ , 32 mM Cl Ϫ , 96 mM gluconate, 1 mM Mg 2ϩ , 6 mM phosphate, and 5 mM glucose; and Ca 2ϩ was adjusted to 0.1 M with EGTA. A flowing 1 M KCl electrode was used as a reference. The membrane voltage (V m ) was continuously monitored during experiments in a current-clamp mode (zero current clamp). Periodically, the whole-cell current (I) was measured in a voltage-clamp mode, where the voltage was first clamped to the spontaneous cell voltage and then in increasing and decreasing 10-mV steps of 500-ms duration. In some experiments, V m was clamped at the spontaneous cell voltage, and continuous whole-cell currents were measured. Where possible, the total conductance (G t ) was corrected for the series resistance to obtain the cell membrane conductance. The series resistance was estimated by the compensation circuit of the patch-clamp amplifier (EPC 9, Heka, Lambrecht, Germany), and in some cases where this was not possible, series resistance and cell membrane conductance estimations were made subsequently by fitting an exponential function to the initial part of the current trace (40). Since duct cells can be variably coupled to each other, G t changes with the agonists are relative rather than absolute. In several experiments, ducts were microdissected and microperfused according to methods described earlier (32). Single duct cells were impaled with 100 -200megaohm Ling-Gerard microelectrodes, and V m was recorded with a WPI electrometer.
Measurements of Intracellular Ca 2ϩ and Intracellular pH-The fura-2 method was used to estimate [Ca 2ϩ ] i in pancreatic ducts. Briefly, as we have described earlier (37), ducts were loaded with 0.5-1.0 M fura-2/AM, and the fluorescence intensity of 10 -20 cells was measured with a photomultipler at 510 nm after excitation at 340, 360, and 380 nm. The fluorescence ratio (340/380 nm) was an estimate of [Ca 2ϩ ] i . Signals were calibrated in situ with ionomycin (1.0 -5.0 M). For pH i measurements, ducts were loaded with BCECF/AM as described earlier (41). Emission intensity of duct cells was measured at 510 nm after excitation at 490 and 440 nm. The fluorescence ratio (490/440 nm) was an estimate of pH i . Signals were calibrated in situ with the protonophore carbonyl cyanide p-chlorophenylhydrazone or nigericin (1 or 10 M).
Statistics-Physiological data are presented as original recordings, summaries, and means Ϯ S.E. In most experiments, control and test measurements were made within one duct cell or one duct, and n refers to measurements on different ducts. The paired Student's t test was applied, and p Ͻ 0.05 was accepted as significant. The F-test was used to test for the difference between two regression coefficients obtained for the current-voltage relations.
RNA Isolation, cDNA Synthesis, and Polymerase Chain Reaction Amplification-Total RNA was isolated from small pancreatic ducts (prepared from collagenase digests of rat pancreas) ( Fig. 1) using total RNA isolation reagent (Advanced Biotechnologies Ltd.) according to the manufacturer's instructions. First strand cDNA was synthesized utilizing an anchored oligo(dT) primer using Reverse-iT (Advanced Biotech-2 C. E. Neumann and I. Novak, unpublished observations. FIG. 1. Pancreatic ducts. Shown is an example of two pancreatic ducts freed from the surrounding connective tissue by collagenase digestion and microdissection. Both ducts are intralobular; the duct in the right panel is closest to acinar cells. Bars indicate 20 m. Differential contrast images were obtained with a 63ϫ water objective in a Leica confocal laser scanning microscope. nologies Ltd.). PCR conditions were identical for all primers (Table I) Temperature cycling proceeded as follows: 1 cycle at 94°C for 5 min and 35 cycles at 94°C for 30 s, 55°C for 60 s, and 72°C for 90 s, followed by 72°C for 10 min. PCR products were then subjected to gel electrophoresis on a 1% agarose gel containing ethidium bromide. Possible contamination of blood vessels and acinar cells was ruled out by RT-PCR amplification of thrombin and amylase using two specific primers (Table I). Thrombin is present in endothelial cells, but not duct cells; and similarly, amylase is present in acinar cells, but not duct cells. As a positive control for duct cells, carbonic anhydrase II primers were used. As tissue controls, we used blood vessels obtained from collagenase digests of pancreas as well as the whole pancreas and testis. Fig. 2A shows the effect of extracellular ATP and BzATP (0.1 mM) on [Ca 2ϩ ] i in pancreatic ducts. ATP stimulation, similar to that of acetylcholine, evokes a characteristic biphasic Ca 2ϩ response consisting of a Ca 2ϩ peak and a Ca 2ϩ plateau. The peak response is largely due to release of Ca 2ϩ from intracellular stores, whereas the plateau reflects largely influx of Ca 2ϩ from extracellular medium (36,38). UTP evoked a similar response, but a more pronounced Ca 2ϩ influx component. On the other hand, stimulation with BzATP led to a monophasic Ca 2ϩ response, which was almost fully dependent on extracellular Ca 2ϩ and thus due to Ca 2ϩ influx. On the basis of these and other observations, we postulated that there are at least two types of receptors belonging to both P2Y and P2X families (36). Fig. 2B summarizes the maximal Ca 2ϩ responses for the three purinergic analogues used in this study. The EC 50 values for UTP, ATP, and BzATP were estimated at 5, 4, and ϳ20 M, respectively. ATP, UTP, and BzATP were the best agonists regarding the intracellular Ca 2ϩ signals compared with ITP, 2-methylthio-ATP, ␤,␥-methylene-ATP, and adenosine, as shown earlier (36). Since native pancreatic ducts are not suitable for extensive electrophysiological experiments (see "Experimental Procedures"), we focused on using ATP, UTP, and BzATP as pharmacological markers for further patch-clamp experiments aimed at elucidating how purinoceptors couple to ion transport.

Effect of Nucleotides on Cell Calcium-
Effect of UTP and ATP on Cell Membrane Voltage and Conductance-In contrast to the Ca 2ϩ responses evoked by UTP and ATP, the V m responses measured in whole-cell patchclamp experiments were relatively small, although there was some variability between cells. On average, UTP and ATP evoked ϳ10-mV changes in V m , and in several cells, the effects were very small, as shown in an original recording in Fig. 3A. In contrast, BzATP depolarized V m reversibly by ϳ20 mV in the same cell. Fig. 3B shows a whole-cell current measurement in another duct cell, where again the ATP and UTP responses were small, whereas BzATP stimulated a sustained inward current of ϳ100 pA at a holding potential of Ϫ60 mV. In this experiment, carried out in a current-clamp mode, ATP depolarized V m by 9 mV, and UTP by 4 mV (not shown). Similar  paired experiments were performed on 14 ducts, and Fig. 4A shows the summary of the V m data for the three agonists. The summary shows that UTP and ATP caused small but significant V m depolarizations of 8 Ϯ 2 and 10 Ϯ 2 mV, respectively. BzATP depolarized V m by 22 Ϯ 3 mV in the same experiments, and this effect was significantly larger compared with the UTP and ATP responses (p Ͻ 0.005).
In the next series of experiments, we investigated the electrophysiological events underlying the V m depolarization caused by UTP and ATP. To illustrate how V m changes correlate with the cell conductance changes, Fig. 5A shows an experiment in which the V m responses to UTP and ATP were some of the largest obtained. The corresponding current-voltage relations in Fig. 5B show that during the UTP and ATP stimulation, V m shifted away from the resting voltage; but notably, G t decreased. A summary of the whole-cell conductance data for 14 experiments is given in Fig. 4B. UTP and ATP caused small but significant reductions in G t of 17 and 20%, whereas BzATP increased G t by 18%. In 10 of these experiments, where it was possible to correct G t for series resistance, changes in the cell membrane conductance evoked by UTP, ATP, and BzATP were qualitatively and quantitatively similar.
The depolarization of V m and the reduction in whole-cell conductance, triggered by UTP and ATP, could be due to a closure of K ϩ channels. This theory was tested in currentclamp experiments in which the bath K ϩ concentration was changed from 4 to 20 mM (see "Experimental Procedures"). In unstimulated ducts, this K ϩ concentration change resulted in ϳ20-mV depolarization of V m , as shown in the original recordings in Fig. 6A. During the ATP stimulation, however, the same K ϩ concentration step had a smaller effect, depolarizing V m by ϳ10 mV. Fig. 6B summarizes seven paired experiments with similar protocols in unstimulated and UTP/ATP-stimulated ducts. Thus, in the stimulated ducts, the K ϩ concentration step depolarized V m by 13 Ϯ 3 mV. In the same but unstimulated ducts, the K ϩ step depolarized V m significantly more by 19 Ϯ 1 mV (p Ͻ 0.05). In additional experiments, duct cells were voltage-clamped in bath solution containing 4, 25, or 50 mM K ϩ in the presence or absence of UTP or ATP. An example of such a paired experiment on one duct cell is shown in Fig. 6 (C and  D). At 4 mM K ϩ , UTP decreased G t from 10.8 to 5.8 nanosiemens (p Ͻ 0.001) and shifted the reversal potential by 16 mV away from the K ϩ equilibrium potential (Fig. 6C). At 25 and 50 mM K ϩ (only 50 mM K ϩ is shown in Fig. 6D), the control G t was higher compared with 4 mM K ϩ , as expected for K ϩ currents. UTP decreased G t from 25.3 to 16.0 nanosiemens (p Ͻ 0.001), but the reversal potential was not affected as the cell voltage lies close to the equilibrium potentials for K ϩ , Cl Ϫ , and nonspecific cation conductance. Similar paired observations were obtained for an additional four cells.
Effect of ATP on Intracellular pH-Taken together, UTP/ ATP depolarize V m due to closure of K ϩ channels. One candidate for the K ϩ channel, the renal K ϩ channel (see "Discussion"), is down-regulated by intracellular Ca 2ϩ and acidic pH i . Therefore, in a separate set of experiments, we monitored pH i in duct cells loaded with BCECF in response to ATP. Fig. 7 shows the reversible effect of ATP (0.1 mM) on the pH i of a pancreatic duct. For comparison, this figure also shows the pH i of another duct stimulated with a classical duct secretagogue, secretin (1 nM). In summary, ATP decreased pH i from 7.51 Ϯ 0.08 to 7.40 Ϯ 0.10 in eight ducts (p Ͻ 0.005). In contrast, secretin had no effect on pH i in similar experiments, i.e. pH i was 7.49 Ϯ 0.05 in resting ducts and 7.50 Ϯ 0.05 in stimulated ducts (n ϭ 16).
Effect of BzATP on Cell Membrane Voltage and Conductance-In another series of experiments, we investigated the ionic basis for BzATP-evoked depolarization in V m and increased whole-cell conductance. Fig. 8 shows examples of two protocols carried out in four experiments. Fig. 8A is the original recording of whole-cell currents in a duct stimulated with BzATP in a control 145 mM Na ϩ solution and in a test 5 mM Na ϩ solution. Fig. 8B shows the results of a voltage-clamp protocol carried out on another cell stimulated with BzATP in control and low Na ϩ solutions. The current-voltage relation indicates that BzATP increases the whole-cell conductance, but only in the presence of control Na ϩ concentrations. Hence, these experiments show that BzATP opens cation channels that allow Na ϩ and also Ca 2ϩ influx, as shown earlier (36). In another set of experiments (n ϭ 4), we tested whether BzATP stimulation affects the Cl Ϫ conductance. Fig. 9 presents an example of one of four experiments in which the bath Cl Ϫ concentration was decreased from 125 to 5 mM. This maneuver seemed to have no effect on BzATP-evoked V m changes, which indicates that the Cl Ϫ conductance was unaffected. For comparison, in ducts stimulated with secretin or isoproterenol, a similar Cl Ϫ concentration step causes 15-20-mV depolariza-tion, which is due to an increase in the cystic fibrosis transmembrane regulator Cl Ϫ conductance (30,34).
Microelectrode Recordings on Perfused Ducts-In one series of experiments, the V m of single duct cells was monitored with microelectrodes on microperfused ducts obtained by microdissection, and one such experiment is shown in Fig. 10A. ATP application from the basolateral side led to a small depolarization of 2 Ϯ 1 mV (n ϭ 8), sometimes preceded by a few millivolts of hyperpolarization (also see Ref. 37). These V m changes are very similar to UTP/ATP responses observed using the wholecell patch-clamp recordings in duct fragments (Figs. 3-6). In contrast, luminal application of ATP resulted in a significantly larger and reversible depolarization of 17 Ϯ 3 mV (n ϭ 10). These V m changes were similar to the BzATP responses observed in duct fragments (Figs. 3 and 4).
P2 Purinoceptor Subtype Expression in Pancreatic Duct Cells Analyzed by RT-PCR-Since the pharmacological data presented above indicated the existence of both P2X and P2Y receptors, we used RT-PCR to investigate which subtypes of the purinergic receptor were expressed in ducts. As shown in Fig. 11A, by use of primers for the P2X family (P2X 1 to P2X 7 ), we could detect transcripts of both P2X 4 and P2X 7 . There were two bands for P2X 7 ; the upper band was seen only when total FIG. 6. ATP and UTP inhibit K ؉ conductance. A, effect of a K ϩ concentration step (from 4 to 20 (K20) mM) on the V m of a single duct cell unstimulated and stimulated with 0.1 mM ATP. B, summary of data shown as means Ϯ S.E. for seven such experiments in which duct cells were stimulated with ATP or UTP and the V m response to the 20 mM K ϩ step was compared with that in the same unstimulated cells. Asterisks indicate that the test values for the K ϩ step were significantly different from the control resting V m (**, p Ͻ 0.005; ***, p Ͻ 0.0005), and the response to 20 mM K ϩ was significantly smaller in stimulated versus unstimulated duct cells (p Ͻ 0.05). C and D, I/V curves for one cell exposed to 4 (K4) or 50 (K50) mM K ϩ solution, respectively, in the presence or absence of UTP (0.1 mM). The whole-cell currents were dominated by the K ϩ conductance, which was inhibited by UTP.
RNA from pancreatic ducts was used, and not when RT-PCR was performed on total RNA from other tissues. Whether the upper band is due to mispriming or to a yet unknown P2X receptor requires further investigations. Regarding the P2Y family of purinergic receptors, we did not include primers for P2Y 3 , as the sequence available from GenBank TM is non-mammalian, or primers for P2Y 5 and P2Y 7 , as their identity as purinoceptors is doubtful (1). In the case of the P2Y receptors, we were able to detect transcripts of P2Y 2 and P2Y 4 , but not of P2Y 6 . As a control for duct epithelium, we show expression of mRNA for carbonic anhydrase II. The samples were free of RNA contaminants from acini and blood vessels since no products for amylase and thrombin were detected. The primers that gave no product in the ducts (P2X 1 , P2X 2 , P2X 3 , P2X 5 , P2X 6 , P2Y 1 , and P2Y 6 ) were tested on total RNA from testis, blood vessels, and whole pancreas to verify the ability of these primers to give products (Fig. 11B). DISCUSSION The key observations are that UTP/ATP evoked large Ca 2ϩ signals, a small depolarization of the membrane voltage, and a decrease in the whole-cell conductance, effects due to an inhibition of K ϩ channels. BzATP/ATP increased cell Ca 2ϩ and caused a large depolarization of the membrane voltage and an increase in the whole-cell membrane conductance, effects due to opening of cation channels and influx of Na ϩ and Ca 2ϩ . Thus, the two types of electrophysiological and Ca 2ϩ responses are due to at least two types of purinoceptors in pancreatic ducts. RT-PCR analysis showed that native pancreatic ducts possess mRNAs for P2Y 2 , P2Y 4 , P2X 4 , and P2X 7 purinoceptors. These are the first molecular biological data on native pancreatic duct epithelium.
P2Y Receptors-Recently, it became evident that pyrimidines can exert their action via several P2Y purinoceptors (1). In expression systems, the P2Y 2 receptor is stimulated equally well with UTP and ATP; the human P2Y 4 receptor has a greater selectivity for UTP over ATP; and the P2Y 6 receptor is stimulated preferentially with UDP (42,44,45). In pancreatic ducts, we did not find expression of mRNA for P2Y 6 , which is the luminal receptor on respiratory and intestinal epithelia FIG. 11. P2 purinoceptor subtype expression in pancreatic duct cells analyzed by RT-PCR. A, total RNA from pancreatic duct cells was reverse-transcribed, and the resulting cDNA was PCR-amplified using the indicated primers (see Table I). Lane M w contains molecular weight markers. B, PCR primers that gave no signal in pancreatic ducts are shown here as positive controls obtained from testis, blood vessels, and whole pancreas. The gels shown are representative of at least three independent experiments. CAII, carbonic anhydrase II. responsible for Ca 2ϩ -mediated Cl Ϫ secretion (12,13). However, we detected transcripts for both P2Y 2 and P2Y 4 receptors (Fig.  11). Unfortunately, rat P2Y 2 and P2Y 4 receptors show similar pharmacology regarding agonists and accept ATP and UTP equally well (46). Both cell Ca 2ϩ and electrophysiological responses are slightly more pronounced with UTP compared with ATP. However, this may simply be due to the ability of ATP to cross-react with the second type of receptors (P2X) discussed below. Thus, we cannot on this basis conclude which receptor is associated with the physiological response we observed in rat ducts. Nevertheless, stimulation of P2Y 2 /P2Y 4 leads to a large Ca 2ϩ transient and inhibition of K ϩ channels (Figs. 4 -6). The inhibition of K ϩ channels is an unexpected finding, as in many similar epithelia, including cultured human and dog pancreatic duct epithelia (see the Introduction), P2Y 2 -like receptors stimulate K ϩ fluxes as well as Cl Ϫ fluxes (2,3,7,10,47). Some differences between our results and these data may be accounted for by different experimental conditions, as mentioned in the Introduction, or by different transport properties of epithelia derived from small intralobular ducts compared with main/large pancreatic ducts used for the tissue culture (48). Nonetheless, there are a few recent reports indicating that P2Y 2 receptors can inhibit M-type K ϩ currents in neurons (49), IsK/KvLQT1 K ϩ channels in vestibular dark cells (50), and Ca 2ϩ -activated K ϩ channels in spermatogenic cells (51). It is not yet clear what cellular mechanisms cause, on one hand, an increase in cell Ca 2ϩ and, on the other hand, an inhibition of K ϩ channels. Inhibitory mechanisms could be indirect, involving calmodulin II, protein kinase C, or membrane-bound phosphatases; or they could be directly exerted by intracellular Ca 2ϩ , e.g. the eag K ϩ channel and renal K ϩ channels, or by acidic pH i (52)(53)(54)(55)(56). Our earlier studies on ducts show that the major cell K ϩ conductance resides on the basolateral as opposed to the luminal membrane (32,33); but except for a few patch-clamp studies on pancreatic ducts (31,57), the identity of K ϩ channels is unknown. However, from our earlier studies, we know that the basolateral K ϩ channels are inhibited by acidic pH i ; and consequently, ducts cannot secrete (33). Interestingly, in this study, we have shown that ATP actually decreases pH i (Fig. 7). Thus, a fall in pH i or changes in cell Ca 2ϩ and associated messengers could close the K ϩ channels in pancreatic ducts. Given that P2Y 2 /P2Y 4 receptors inhibit K ϩ channels in pancreatic duct cells, they cannot support secretion, as Cl Ϫ efflux cannot be followed by K ϩ efflux (see the Introduction). Perhaps then P2Y 2 /Y 4 receptors could be involved in a downregulation of secretion evoked by the true secretagogues.
P2X Receptors-In our recent study, we identified the BzATP-responsive receptor pharmacologically as the P2Z (P2X 7 -like) receptor (36). BzATP or high ATP levels evoked Ca 2ϩ transients resulting from a Ca 2ϩ influx, which was sensitive to DIDS, Mg 2ϩ , La 3ϩ , and extracellular pH (36,38). This study shows that BzATP activates an inward Na ϩ current, supporting the idea that this is a ligand-gated receptor. The P2Z receptor would in many cases correspond to the cloned P2X 7 receptor (58). Since we also found transcripts for the P2X 7 receptor in pancreatic ducts, the simplest interpretation is that the P2Z receptors we have described earlier are also of this type. However, the receptor does not form a pore either in our preparation or in other exocrine glands (4, 11, 17, 19 -23, 26), as it does in immunoreactive cells. One possibility is that the P2X 7 receptor expressed in glands does not contain the carboxyl-terminal domain necessary for the pore formation (58). Another possibility is that our cells do not express ancillary factors necessary for the P2X 7 pore formation. This possibility is not remote, as the P2X 7 -associated pore formation depends on the expression system (59). Finally, P2X receptors are two transmembrane-spanning proteins that can form heteromers, e.g. P2X 2 and P2X 3 (60); and given the expression of P2X 4 in ducts, it is possible to speculate that the P2X 4 and P2X 7 proteins could form heteromers responsible for the physiological properties we observed in glands. Nevertheless, we cannot exclude that the P2X 4 homomer could also be responsible for some of the characteristics we observed. This receptor displays a higher affinity for ATP than P2X 7 (43,61,62), consistent with the observed effects on Ca 2ϩ transients (Fig. 2). However, since ATP also cross-reacts with the P2Y receptors described above, one cannot use the EC 50 values conclusively. Nevertheless, the lower sensitivity of P2X 4 to DIDS and ␣,␤-methylene-ATP (43,62) compared with the sensitivity we observed (36) indicates that the P2X 7 response may be dominant.
The present finding that the stimulation of the P2X 7 receptor is associated with increased Na ϩ /Ca 2ϩ conductance is in agreement with earlier findings on salivary gland and lacrimal gland acini, which express the P2X 7 -like receptors (17, 19 -23, 26). Moreover, in these glands, ATP also increases protein secretion, cell volume changes, and ion fluxes, all indicative of secretory activity. With respect to pancreatic ducts, if the P2X 7 receptors increase only the Na ϩ /Ca 2ϩ conductance, it is difficult to envisage how this could support secretion. Clearly, there must be an opening of the luminal Cl Ϫ channels and the basolateral K ϩ channels as well as activation of exchangers, as observed with true secretagogues (30,33). Therefore, on the basis of the present data, this receptor on its own could probably not evoke secretion; rather, by virtue of its effect on Ca 2ϩ , it could regulate secretion evoked by other secretagogues. Since, in the perfused duct, ATP had the strongest effects on the lumen, reminiscent of those BzATP had on the duct fragments, the most likely localization of the P2X 7 receptor is on the luminal membrane. In another type of ductal epithelium, namely ducts from submandibular glands, the luminal P2X 7like receptor seems to have the unusual effect of opening Cl Ϫ channels, without reported effects on the cation conductances (4,11). One interpretation is that salivary gland ducts do not secrete fluid as do pancreatic ducts; and thus, coupling of receptors and transporters may be different.
In conclusion, our studies demonstrate that native pancreatic ducts express mRNAs for P2Y 2 and P2Y 4 receptors as well as for P2X 4 and P2X 7 receptors. Stimulation of P2Y 2 (and possibly P2Y 4 ) receptors leads to an increase in cell Ca 2ϩ , but a decrease in cell K ϩ conductance, a process that is not compatible with secretion. Thus, UTP/ATP acting on these receptors would down-regulate secretion evoked by other secretagogues. In contrast, stimulation of P2X 7 (and possibly P2X 4 ) receptors is associated with increased cation conductance, a process that may modulate action of other secretagogues.
Note Added in Proof-After acceptance of our paper, Luo and coworkers published an article (Luo, X., Zheng, W., Yan, M., Lee, M. G., and Muallem, S. (1999) Am. J. Physiol. 277, C205-C215) showing a greater variety of P2X and P2Y receptors in pancreatic tissue. Differences between these and our results may be due to purity of duct tissue used for RT-PCR.