Molecular and Functional Identification of a Ca2+ (Polyvalent Cation)-sensing Receptor in Rat Pancreas*

The balance between the concentrations of free ionized Ca2+ and bicarbonate in pancreatic juice is of critical importance in preventing the formation of calcium carbonate stones. How the pancreas regulates the ionic composition and the level of Ca2+ saturation in an alkaline environment such as the pancreatic juice is not known. Because of the tight cause-effect relationship between Ca2+ concentration and lithogenicity, and because hypercalcemia is proposed as an etiologic factor for several pancreatic diseases, we have investigated whether pancreatic tissues express a Ca2+-sensing receptor (CaR) similar to that recently identified in parathyroid tissue. Using reverse transcriptase-polymerase chain reaction and immunofluorescence microscopy, we demonstrate the presence of a CaR-like molecule in rat pancreatic acinar cells, pancreatic ducts, and islets of Langerhans. Functional studies, in which intracellular free Ca2+concentration was measured in isolated acinar cells and interlobular ducts, show that both cell types are responsive to the CaR agonist gadolinium (Gd3+) and to changes in extracellular Ca2+ concentration. We also assessed the effects of CaR stimulation on physiological HCO3 −secretion from ducts by making measurements of intracellular pH. Luminal Gd3+ is a potent stimulus for HCO3 − secretion, being equally as effective as raising intracellular cAMP with forskolin. These results suggest that the CaR in the exocrine pancreas monitors the Ca2+ concentration in the pancreatic juice, and might therefore be involved in regulating the level of Ca2+ in the lumen, both under basal conditions and during hormonal stimulation. The failure of this mechanism might lead to pancreatic stone formation and even to pancreatitis.

Pancreatic juice in humans and other species is an alkaline secretion, containing up to 140 mM bicarbonate ions (1). The juice also contains millimolar quantities of Ca 2ϩ ions, which are released from secretory granules along with pancreatic zymogens (1). The presence of these ions means that the pancreatic juice is at risk of precipitating calcium stones, which are a major cause of chronic pancreatitis (1). This is especially true when the residence time of the juice within the ductal compartment is increased, i.e. at low pancreatic secretory rates. Because the incidence of pancreatic stone formation is surprisingly low, it has been postulated that homeostatic responses must be activated to reduce lithogenic potential. A number of suggestions have been advanced for candidate mechanisms, all within the ductal system, including increased H ϩ and fluid secretion and a reduction in bicarbonate production (1). However, thus far there is no direct experimental evidence for any of these hypotheses, and the question of how the pH and the free Ca 2ϩ concentration in pancreatic juice are regulated is yet to be understood.
Recently Brown et al. (2) have identified a cell-surface, G protein-linked Ca 2ϩ (polyvalent cation)-sensing receptor (CaR) 1 which is capable of monitoring even minute changes in extracellular calcium concentration ([Ca 2ϩ ] o ) and responds with an increase in intracellular calcium concentration ([Ca 2ϩ ] i ). The receptor is sensitive to changes in [Ca 2ϩ ] o in the millimolar range, compatible with [Ca 2ϩ ] o in the plasma. Intriguingly, the concentration of free ionized Ca 2ϩ in human pancreatic juice under conditions of basal secretion has been estimated to be ϳ1 mM (1). Since hypercalcemia stimulates pancreatic secretion (3), and is associated with pathological effects including acinar and ductal cell necrosis, formation of intraductal precipitates, and clinical pancreatitis (3)(4)(5)(6), we tested the hypothesis that a CaR-like mechanism was also expressed in the rat exocrine pancreas.
Reverse-transcriptase PCR on total rat pancreas mRNA, using intron-spanning primers, established the presence of CaRlike transcripts, and immunological localization of CaR protein employing an anti-CaR polyclonal antiserum revealed strong immunoreactivity at the luminal side of pancreatic ducts and a more diffuse, cellular and basolateral staining pattern in acinar cells. The receptor is also strongly expressed in clusters of cells representing the islets of Langerhans. Functional studies in isolated pancreatic acinar cells, and in microperfused interlobular ducts, using the Ca 2ϩ -sensitive fluorescent dye Fura-2 showed that the CaR agonists [Ca 2ϩ ] o and Gd 3ϩ raised [Ca 2ϩ ] i in both acini and ducts. Finally, we tested the hypothesis that the CaR might play a role in regulating ductal HCO 3 Ϫ secretion. Luminal microperfusion of Gd 3ϩ resulted in enhanced ductal HCO 3 Ϫ secretion, comparable to that evoked by the secretagogue forskolin (7).
Our findings indicate that the CaR in the pancreas plays a significant role in the regulation of pancreatic juice secretion, in particular by constantly monitoring the level of free Ca 2ϩ in secreted fluid within the acinar and duct lumen. Alterations of * This work was supported in part by grants from the Royal Society (to D. R.), the Wellcome Trust and Biotechnology and Biological Sciences Research Council (to A. C. E., R. M. C., M. C. S., and D. R.), and the UK Cystic Fibrosis Research Trust (to M. C. S. and R. M. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  receptor function could help explain how ion sensing by the exocrine pancreas can go awry, resulting in pancreatic stone formation and acute pancreatitis.

EXPERIMENTAL PROCEDURES
Reverse Transcriptase-PCR (RT-PCR)-Male Sprague-Dawley rats weighing between 100 and 300 g were killed by cervical dislocation. The pancreas was removed, frozen immediately in liquid N 2 , and stored at Ϫ80°C until use. Total RNA was extracted using the guanidinium thiocyanate/acid phenol method (8). First strand DNA was synthesized from 1 to 2 g of total RNA using Superscript Reverse Transcriptase (Life Technologies) according to the manufacturer's instructions and the resultant first-strand DNA was then used for PCR amplification. In order to perform "hot start" PCR, the Taq DNA polymerase was added during the initial 3-min denaturation, followed by 35 cycles of amplification (1 min denaturation at 92°C, 30 s annealing at 47°C, and 1 min extension at 72°C), with final extension for 10 min at 72°C. The 383 base pairs expected PCR product was visualized with ethidium bromide after electrophoretic separation on a 1% agarose gel, and the specificity of the PCR reaction was assessed by high stringency Southern blotting using a 32 P-labeled 3.2-kilobase XhoI-ApaI fragment corresponding to the coding region of the rat kidney CaR (9, 10), as described elsewhere (9). To avoid genomic DNA amplification, the primer sequences were based on the known rat CaR sequence (10) and were designed to span one intron: forward primer, 5Ј-ACCTTTACCTGTCCCCTGAA-3Ј; reverse primer, 5Ј-GGGCAACAAAACTCAAGGTG-3Ј. As negative and positive controls the DNA template was replaced with an equivalent volume of water and with an equivalent amount of total RNA reversetranscribed from rat kidney, respectively.
Immunoreactivity of CaR Protein in Rat Pancreas-CaR polyclonal antibodies raised against a 20-amino acid sequence in the predicted hydrophilic amino-terminal region of the bovine CaR (11) were generously provided by Drs. S. C. Hebert (Vanderbilt University, Nashville, TN) and by Dr. E. M. Brown (Harvard Medical School, Boston, MA). Prior to use, the antibodies were purified using an affinity column conjugated with the antigenic peptide (11). For immunohistochemistry in pancreatic sections, rats were killed by cervical dislocation and the pancreas perfused with 4% paraformaldehyde in phosphate-buffered saline. The pancreas was then isolated, post-fixed for 5 min, and snapfrozen by immersing it in N-methylbutane using standard embedding medium (OCT compound) as mounting medium. Four micron-thick cryosections were cut using a Leica CM3050 cryostat (Leica, Nussloch, Germany) and processed as described elsewhere (11). The intensity of immunostaining was enhanced using an antigen-retrieval protocol employing citrate buffer (BioGenex, San Ramon, CA) and 1% sodium dodecyl sulfate (12). Nonspecific binding of proteins was prevented with the DAKO serum-free protein block reagent (DAKO Corp., Carpinteria, CA). Staining was performed using a horseradish peroxidase-conjugated secondary antibody (DAKO LSAB 2 Kit, HRP) according to the manufacturer's instructions. The affinity-purified anti-CaR polyclonal antibody was diluted in phosphate-buffered saline containing 1% (w/v) bovine serum albumin and 3% (v/v) normal goat serum and applied at a 1:400 dilution with an overnight incubation at 4°C. Negative control (preabsorption of the antibody with 5 g/ml of the antigenic peptide prior to incubation) was performed under the same conditions. For examination of CaR immunoreactivity in isolated pancreatic ducts we utilized single interlobular ducts isolated as described below for [Ca 2ϩ ] i experiments (n ϭ 23 ducts from 3 rats). Each individual duct was partly split open and acetone-fixed for 10 min. Positive and negative CaR immunoreactivity was tested as described above, except that the antigen retrieval protocol was omitted. Fluorescent secondary antibody (goat anti-rabbit IgG conjugated with rhodamine red, Jackson Immunochemicals, Luton, Beds, United Kingdom) diluted 1:200 was added and the sections incubated for 1 h at room temperature. Both horseradish peroxidase-stained whole pancreas sections and fluorescently stained single isolated ducts were visualized with a Zeiss Axioskop microscope, and images were acquired using a digital camera (Photonic Science, Cambridge, UK) using the Openlab software package (Improvision, Coventry, UK).
Acinar Cell Isolation-Small clusters of rat pancreatic acinar cells were isolated by enzymatic digestion of the pancreas with collagenase as described previously (13,14), except that all media contained a reduced concentration of CaCl 2 (0.5 mM). This value, which is less than half the physiological concentration, was selected to avoid possible activation of CaRs, which is known to occur at [Ca 2ϩ ] concentrations of 1 mM and above (2). Following isolation, the cells were resuspended in a HEPES-buffered physiological saline (pH 7.4) containing (mM): 104 NaCl, 5 KCl, 1.2 MgCl 2 , 0.5 CaCl 2 , 25 HEPES, 15 glucose, 2 L-glutamine. The solution was supplemented with 2% (v/v) minimal essential medium amino acids, 0.12 mg ml Ϫ1 trypsin inhibitor (type II-S, Sigma) and 1% (w/v) bovine serum albumin (Fraction V, Sigma). This cell suspension was either used immediately or maintained on ice and top-gassed with 100% O 2 until use.
Loading of Pancreatic Acinar Cells with Fura-2-AM and Recording of [Ca 2ϩ ] i -Acinar cells were loaded with 4 M Fura-2/AM (Molecular Probes) for 40 min at room temperature, and then transferred to a perfusion chamber as described previously (14). Once cells had adhered to the chamber base they were continuously superfused with HEPESbuffered physiological saline from a gravity-fed perfusion system at a rate of 2 ml min Ϫ1 . Fluorescence in a field of up to 30 acinar cells was imaged using a Nikon Diaphot microscope and a slow-scan CCD camera (Digital Pixel Ltd, Brighton, UK) as described previously in detail (14). Background-subtracted 340:380 images were calculated off-line and used to generate records of the 340:380 ratio in each of the individual acinar cells within the microscope field.
The composition of the physiological saline used in acinar cell imaging experiments was similar to that in the isolation procedure, but omitting amino acids, glutamine, trypsin inhibitor, and bovine serum albumin. In addition, SO 4 2Ϫ and HPO 4 2Ϫ ions were replaced with Cl Ϫ in experiments employing gadolinium and high [Ca 2ϩ ] o in order to prevent precipitation of insoluble gadolinium or calcium salts. In most experiments, CaCl 2 was reduced to 0.1 mM to prevent possible activation of the CaR. Preliminary experiments established that 0.1 mM CaCl 2 was the minimum necessary to maintain [Ca 2ϩ ] i oscillations evoked by the physiological secretory agonist cholecystokinin (CCK). As an index of viability only cells which responded to 50 pM CCK were utilized. All solutions and experimental media were equilibrated with 100% O 2 and all experiments were carried out at room temperature.
Isolation and Culture of Interlobular Ducts-Interlobular ducts were prepared as described previously (15). Animals were killed by cervical dislocation, the body and tail of the pancreas were removed and injected with a digestion buffer consisting of Dulbecco's modified Eagle's medium (Flow Laboratories, Irvine, UK) supplemented with 80 units ml Ϫ1 collagenase (Worthington Biochemical Corp., Freehold, NJ), 400 units ml Ϫ1 hyaluronidase (Sigma), 0.2 mg ml Ϫ1 soybean trypsin inhibitor, and 0.2% (w/v) bovine serum albumin. The tissue was chopped coarsely with scissors into approximately 1-mm 3 pieces, gassed with 5% CO 2 , 95% O 2 and incubated at 37°C for 35 min and then in fresh digestion buffer for a further 30 min. The digested tissue was washed with Dulbecco's modified Eagle's medium and resuspended in Dulbecco's modified Eagle's medium containing 0.2 mg ml Ϫ1 soybean trypsin inhibitor and 3% (w/v) bovine serum albumin. Interlobular ducts (diameter 100 -130 m) were microdissected from samples of this tissue suspension under a dissection microscope using sharpened needles. The ducts were then placed on polycarbonate membrane filters (Cyclopore) floating on McCoy's 5A tissue culture medium supplemented with 10% (v/v) fetal calf serum, 2 mM glutamine, 0.1 mg ml Ϫ1 soybean trypsin inhibitor, 0.1 IU ml Ϫ1 insulin and 4 g ml Ϫ1 dexamethasone, and were cultured at 37°C in 95% O 2 , 5% CO 2 for up to 24 h, during which time the ends of the ducts seal spontaneously.
Duct Microperfusion and Measurement of [Ca 2ϩ ] i and pH i -After overnight culture, the ends of a duct were cut open using sharpened needles and the duct was incubated for 1 h at room temperature in HEPES-buffered saline containing either 5 M Fura-2/AM (for measurements of [Ca 2ϩ ] i ) or 1 M BCECF/AM (for measurements of intracellular pH). The duct was then transferred to a perfusion chamber (volume 200 l) mounted on the stage of a Nikon Diaphot inverted microscope for simultaneous luminal microperfusion and measurement of [Ca 2ϩ ] i . The duct was gently aspirated into a holding pipette (tip diameter 50 -80 m), and a perfusion pipette (tip diameter 10 m) was then advanced into the duct lumen for luminal perfusion. The other end of the duct was attached to the surface of the glass coverslip, which formed the bottom of the perfusion chamber. The duct lumen was then perfused at a flow rate of 10 -20 l/min, while the chamber was perfused at a flow rate of 3.0 ml/min. Duct cell [Ca 2ϩ ] i or intracellular pH (pH i ) was estimated by dualexcitation wavelength microfluorometry as described previously (15,16). An optical diaphragm in the emitted light path limited fluorescence collection to a small region of the ductal epithelium (10 -20 cells). Records of [Ca 2ϩ ] i are displayed as the uncalibrated Fura-2 fluorescence ratio, while records of pH i were calibrated using nigericin/high The Ca 2ϩ -and Mg 2ϩ -free HEPES-buffered perfusion fluid used in experiments measuring [Ca 2ϩ ] i in isolated ducts contained (mM): 140 NaCl, 5 KCl, 10 D-glucose, and 10 HEPES, and was equilibrated with 100% O 2 . When the perfusate contained 0.5, 5, or 8 mM Ca 2ϩ , NaCl concentration was reduced appropriately to maintain isosmolarity. All HEPES-buffered solutions were adjusted to pH 7.4 at 37°C. For experiments assessing ductal HCO 3 Ϫ secretion, the bath was perfused with a medium containing 25 mM HCO 3 Ϫ (replacing Cl Ϫ ) and equilibrated with 95% O 2 , 5% CO 2 to give a pH of 7.4, while the lumen was perfused with the HEPES-buffered perfusion fluid described above.

Expression of CaR-related Transcripts in Rat Pancreas
Reverse transcription and PCR amplification was carried out on total mRNA from rat pancreas and rat kidney as positive control. Using CaR-specific primers, we were able to amplify a product of the expected size (383 base pairs) for the CaR from both samples. High-stringency Southern analysis using a 32 Plabeled CaR-specific cDNA probe (9) confirmed that the 383base pair amplified PCR products were genuine CaR transcripts ( Fig. 1, arrow). No products were detected when DNA was replaced with water in the PCR reaction ( Fig. 1).

Immunofluorescence and Immunohistochemistry
In order to identify which cells in the rat pancreas expressed the CaR, we carried out immunolocalization with an affinitypurified anti-CaR polyclonal antibody raised against the bovine parathyroid CaR (11). Light microscope images (Fig. 2) revealed that specific CaR immunoreactivity, visualized using peroxidase staining, was present in ducts, acini, and also in cells within the islets of Langerhans. Similar results were also seen when a fluorophore-coupled, rather than horseradish peroxidase-conjugated, secondary antibody was used (data not shown). No signal was detected when the primary antibody was preincubated with the antigenic peptide ( Fig. 2A). Distinct staining patterns were observed in the different cell types. A punctate-like staining clearly above background was observed in groups of acinar cells (Fig. 2, B-D), with a rather heterogeneous intensity. At higher magnification (Fig. 2D) the staining seemed to be localized at the basolateral membrane and/or intracellularly in the acinar cells, although apical staining was also visible. An intense signal was detected in the islet of Langerhans, predominantly at the periphery of the islets, so that the staining clearly delimited the islets (Fig. 2C). Fig. 2C also shows strong CaR immunoreactivity in an interlobular duct (arrow). The intensity of the staining observed in acinar cells was clearly weaker than in both islet of Langerhans and duct cells (Fig. 2C), suggesting a reduced expression of the CaR in acinar cells relative to the other two cell types.
Since unambiguous identification of ducts in tissue sections was often difficult, we went on to define the subcellular localization of the CaR in the same isolated interlobular duct preparation used for functional studies. As shown in Fig. 2F, a strong immunofluorescence signal was observed inside the ducts, representing the luminal membrane of the duct cells.

Functional Evidence for a CaR in Exocrine Pancreas
Response of Isolated Acini to Gd 3ϩ -In order to test whether acinar cells express a functional CaR, we stimulated fura-2loaded acinar cells with CaR agonists. Initially we tested 1 mM Gd 3ϩ , which produced changes in [Ca 2ϩ ] i when applied either before or after stimulation with 50 pM CCK. Treatment with 1 mM Gd 3ϩ after CCK stimulation triggered responses in approximately half (53%) of the acinar cells tested, with a range of response profiles being observed (Fig. 3). A rapid and irreversible increase in [Ca 2ϩ ] i was evoked in 6% of cells (Fig. 3A), a large transient increase in [Ca 2ϩ ] i in 9% of cells (Fig. 3B), oscillations in [Ca 2ϩ ] i in 31% of cells (Fig. 3C), and a slow rise in [Ca 2ϩ ] i in 7% of cells (Fig. 3D)  Reverse transcriptase PCR using specific, intron-spanning primers was followed by high stringency Southern analysis. The expected 383-base pair size product was amplified from rat pancreas (P), rat kidney (positive control, ϩ) but not when H 2 O replaced DNA samples in the PCR reaction (negative control, Ϫ).

FIG. 2. CaR immunoreactivity in rat pancreas.
A and B, phasecontrast images (ϫ 88) of CaR specific immunoreactivity (B) and negative control peptide preabsorbed immune serum (A). C, phase-contrast images of sparse populations indicating densely stained CaR-positive cells. An islet of Langerhans is clearly visible while an arrowhead indicates a rounded structure, almost certainly an interlobular duct. D, higher magnification (ϫ 510) view of acinar cells exhibiting basolateral/ intracellular CaR immunoreactivity. In addition, weaker apical staining seems to be present. E and F, CaR immunofluorescence microscopy of isolated interlobular ducts (ϫ 510). Positive staining is detected in the lumen of the duct (F) but not in control ducts stained with CaR antibody preabsorbed with the antigenic peptide (E).

Response of Isolated Acini to Elevated [Ca 2ϩ
] o -We went on to test acinar cells with the physiological agonist of the CaR by examining the effects of raising [Ca 2ϩ ] o from 0.1 to 8 mM (Fig.  4). This produced a variety of changes in [Ca 2ϩ ] i . However, the pattern of changes was noticeably different from the profile of responses observed with Gd 3ϩ (Fig. 3). Only 20% of cells produced oscillations in [Ca 2ϩ ] i , and the oscillations were superimposed over a slowly rising baseline (Fig. 4A), which did not occur with Gd 2ϩ (compare Fig. 3). In 16% of cells a transient increase in [Ca 2ϩ ] i was observed (Fig. 4B). However, the kinetics of this change in [Ca 2ϩ ] i were different from the equivalent Gd 3ϩ -evoked response (Fig. 3B) (Fig. 4C) while 47% of cells failed to produce a response ( Fig. 4D; n ϭ 140 cells, 5 rats). Overall, it was clear that fewer cells showed typical receptormediated responses when stimulated with elevated [Ca 2ϩ ] o than when stimulated with Gd 3ϩ . In fact the predominant response seemed to be a slow increase in baseline [Ca 2ϩ ] i , seen in around 50% of cells, which is suggestive of increased Ca 2ϩ entry.
Finally, we tested the polycationic antibiotic neomycin (500 M and 1 mM), another known CaR agonist (2). In our hands neomycin failed to produce any change in [Ca 2ϩ ] i in all acinar cells tested (n ϭ 48 cells from two rats for 500 M neomycin, 156 cells from six rats for 1 mM) (not shown).
Response of Isolated Interlobular Ducts to Gd 3ϩ -We examined the effects of Gd 3ϩ on interlobular ducts with the same experimental protocol used to test for CaR expression in studies on heterologous expression systems (2). In this protocol Gd 3ϩ is applied in the absence of all other divalent cations (i.e. in the absence of extracellular Ca 2ϩ and Mg 2ϩ ). We first performed preliminary experiments to ensure the viability of the isolated ducts in the absence of divalent cations. This was achieved by perfusing both bath and lumen with Ca 2ϩ -and Mg 2ϩ -free solution for at least 10 min and then testing duct viability by the response to bath application of 10 M acetylcholine (ACh, not shown). Having established that the ducts remained viable in the absence of divalent cations, we determined the functional localization of the CaR (apical versus basolateral). Ducts were challenged with 600 M Gd 3ϩ using three different approaches: simultaneous luminal and bath perfusion, luminal perfusion only, or basolateral perfusion only. In all experiments we could only detect changes in [Ca 2ϩ ] i when Gd 3ϩ was applied luminally (n ϭ 31 ducts from four rats). Bath perfusion with Gd 3ϩ failed to produce an increase in [Ca 2ϩ ] i (n ϭ 8 ducts from four rats).
Under these conditions, all of the ducts tested responded to luminal Gd 3ϩ with an increase in [Ca 2ϩ ] i . Fig. 5 (16), and are thought to be due to Ca 2ϩ entry, presumably across the basolateral membrane. We found that, in order to obtain reproducible responses to changes in luminal [Ca 2ϩ ] o , it was necessary to ensure that intracellular Ca 2ϩ stores remained loaded with Ca 2ϩ during the period of low [Ca 2ϩ ] o perfusion. This was achieved by perfusing the bath with 1 mM [Ca 2ϩ ] o , to maintain loading of Ca 2ϩ into the intracellular stores, while perfusing the lumen with a lower concentration of [Ca 2ϩ ] o (0.5 mM). Under these conditions (Fig. 6), increasing luminal [Ca 2ϩ ] o to 8 mM evoked an increase in [Ca 2ϩ ] i which was qualitatively and quantitatively similar to that stimulated by addition of Gd 3ϩ (Fig. 6) (n ϭ 9 ducts from two rats).
Finally, we tested whether the CaR in the ducts could be activated by neomycin (500 M and 1 mM, not shown). We did not observe a change in [Ca 2ϩ ] i in any of the ducts tested (n ϭ 9 ducts from two rats), although the same ducts remained responsive to luminal Gd 3ϩ (600 M, n ϭ 4 ducts from two rats) and/or bath ACh (10 M, n ϭ 9 ducts from two rats). These data agree with those obtained in acinar cells and indicate that neomycin does not act as an agonist of the CaR in rat pancreas.
Stimulation of Ductal Secretion by Luminal Gd 3ϩ -In order to test the hypothesis that the luminal CaR can stimulate fluid secretion in the ductal system, we measured the rate of cellular

HCO 3
Ϫ efflux in microperfused ducts. Pancreatic ducts secrete a Na ϩ -and HCO 3 Ϫ -rich fluid, so that the rate of ductal HCO 3 Ϫ secretion also gives an index of the rate of fluid secretion (7,15,17). We have previously shown that stimulation of isolated ducts with the cAMP-elevating hormone secretin causes little change in intracellular pH (pH i ) because the increased rate of luminal (secretory) HCO 3 Ϫ efflux is balanced by increased basolateral HCO 3 Ϫ uptake (15). This basolateral HCO 3 Ϫ uptake is mediated by the basolaterally located Na ϩ -HCO 3 Ϫ co-transporter and Na ϩ -H ϩ exchanger (15). Consequently, inhibition of these basolateral transporters with a combination of H 2 DIDS and amiloride reveals a marked intracellular acidification which can be attributed to luminal HCO 3 Ϫ efflux (7, 15). Fig. 7A shows typical records of such cellular acidification ( HCO 3 Ϫ secretion) in control ducts and in ducts treated with forskolin or with luminal Gd 3ϩ . The averaged data in Fig. 7B show that forskolin, which raises intracellular cAMP by stimulating adenylate cyclase, and evokes secretion of fluid and HCO 3 Ϫ (15), caused an approximate doubling of the cellular acidification rate ( HCO 3 Ϫ secretory rate). This is similar to the increase in HCO 3 Ϫ secretory rate that we have previously reported for stimulation with secretin (15). Perfusion of the duct lumen with 600 M Gd 3ϩ increased the acidification rate to approximately the same extent as forskolin. This demonstrates that activation of luminal CaRs can indeed activate ductal HCO 3 Ϫ (and hence fluid) secretion. DISCUSSION In this study we have investigated the hypothesis that regulation of pancreatic juice ionized Ca 2ϩ concentration involves an extracellular CaR within the pancreas. The first molecular evidence for the presence of such a receptor came from RT-PCR. As shown in Fig. 1, high-stringency Southern analysis using a CaR probe confirmed the specificity of the PCR amplification product. Immunofluorescence and confocal microscopy subsequently revealed expression of CaR protein in acinar, duct, and islet cells, in each case with a different and characteristic staining pattern at the subcellular level: basolateral/ intracellular in the acini, luminal in the ducts, and at the periphery of cells composing the islets of Langerhans. CaR in Acinar Cells-Acinar cells exhibit diffuse, punctate staining, indicating that the receptor is not homogeneously distributed at the cell surface (see Fig. 2). Staining was predominantly observed at the basolateral side of acinar cells, consistent with the presence of the CaR on the basolateral membrane. This agrees well with the functional studies showing agonist effects of Gd 3ϩ and Ca 2ϩ . In addition, however, significant intracellular and more apical staining was also detected. There are two possible explanations for this. First, pancreatic acinar cells have a high rate of luminal exocytosis (zymogen secretion). Endocytosis accompanies this exocytosis as part of the normal process of membrane recycling (18). It is therefore conceivable that, if the CaR is present on the acinar cell apical membrane, immunolocalization would reveal primarily intracellular staining as the receptor is recycled.
Alternatively, the intracellular staining may indicate a true intracellular localization of the receptor. This might reflect a high rate of receptor biosynthesis, or extensive post-translational modification occurring prior to receptor insertion into the plasma membrane, or even some functional role. An intracellular punctate distribution pattern is not unusual for the CaR, and has previously been observed in chief cells of the bovine parathyroid gland (19). Interestingly, in these cells the receptor is localized within caveolin-rich intracellular membrane domains where it has been suggested to regulate parathyroid hormone secretion via phosphorylation/dephosphorylation (20).
It is now well established that the CaR stimulates Ca 2ϩ release from intracellular stores via a phosphoinositide pathway (2, 10). When we exposed pancreatic acinar cells to 1 mM Gd 3ϩ , changes in [Ca 2ϩ ] i were evoked in between 50% (CCK prior to stimulation with Gd 3ϩ ) and 75% (Gd 3ϩ prior to CCK administration) of all acinar cells, with oscillations in [Ca 2ϩ ] i being observed in about a third of the cells. The lack of any response to Gd 3ϩ or high [Ca 2ϩ ] o in between 50 and 25%, respectively, of all cells could be ascribed to the possibility that the enzymatic digestion procedure employed to isolate single cells damages some of the cell-surface receptors. This is a particular concern with the CaR, because it possesses a very large extracellular domain (2,10). Indeed, damage of this kind might explain why Gd 3ϩ was a more effective stimulus than high [Ca 2ϩ ] o or neomycin, since Hammerland et al. (21) showed by mutagenesis that removal of the extracellular domain of the CaR abolishes the response to Ca 2ϩ and neomycin but partly retains the response to Gd 3ϩ . In addition, previous observations in human pancreatic insulinoma cells are consistent with the presence of a Ca 2ϩ -sensing mechanism activated by Gd 3ϩ but not by neomycin (22).
Overall ؊ efflux in isolated, microperfused, pancreatic ducts. Single interlobular ducts were loaded with the pH-sensitive fluorophore BCECF and microperfused. The bath was perfused throughout the experiment with a medium containing 25 mM HCO 3 Ϫ . The duct lumen was perfused with a HCO 3 Ϫ -, sulfate-, and phosphate-free medium. A combination of 0.5 mM H 2 DIDS (inhibitor of basolateral Na ϩ -HCO 3 Ϫ co-transporter) and 0.2 mM amiloride (inhibitor of basolateral Na ϩ -H ϩ exchange) was added to the bath as indicated by the black bars. Forskolin (0.5 M) was added to the bath, while Gd 3ϩ (600 M) was added to the luminal perfusate, in both cases 6 min prior to the start of the record. Panel A shows typical experimental traces, each representative of a total of 10 ducts, while Panel B shows the average (mean Ϯ S.E.) rate of HCO 3 Ϫ efflux (measured as the initial rate of intracellular acidification on application of H 2 DIDS and amiloride).
Asterisks indicate values significantly different from control (t test, p Ͻ 0.05). observed with Gd 3ϩ are characteristic of responses to a Gprotein/phosphoinositide-linked agonist in acinar cells, and notably resemble responses to submaximal doses of CCK (23). Under these circumstances some cells, representing those least sensitive to CCK, produce no response, while a small proportion of cells, representing the cells with the highest CCK sensitivity, produce a large sustained or transient increase in [Ca 2ϩ ] i . The majority of cells, possessing intermediate CCK sensitivity, produce oscillations in [Ca 2ϩ ] i . Although the coupling of the CaR to the phosphoinositide/Ca 2ϩ pathway is well established, to our knowledge this is the first demonstration that activation of the CaR evokes [Ca 2ϩ ] i oscillations.
Treatment with Gd 3ϩ was somewhat more effective in raising [Ca 2ϩ ] i when cells had not previously been stimulated with CCK. It is possible that treatment with CCK prior to Gd 3ϩ may partially deplete the intracellular Ca 2ϩ stores, especially since [Ca 2ϩ ] o in these experiments was only 0.1 mM, which might reduce store re-filling. This is consistent with the observation that the stimulatory effect of hypercalcemia on feline pancreatic secretion could be prevented when a large dose of CCK was infused prior to induction of hypercalcemia (3).
Consistent with previously reported observations (2, 10), elevated [Ca 2ϩ ] o appeared less effective than Gd 3ϩ as an agonist of the CaR in pancreatic acinar cells (see "Results"). Similar results were also obtained in isolated ducts, where Gd 3ϩinduced responses were more consistent than those evoked by increased [Ca 2ϩ ] o . Interpretation of experiments with high [Ca 2ϩ ] o is complicated by the fact that Ca 2ϩ will leak into the cell via Ca 2ϩ channels. It is unlikely, however, that this would produce rapid spike-like increases in [Ca 2ϩ ] i or [Ca 2ϩ ] i oscillations. The most likely effect of increased Ca 2ϩ entry would be a much slower rise in [Ca 2ϩ ] i , as was indeed observed in around 40% of acinar cells and in three-quarters of the isolated ducts. This kind of very gradual rise in [Ca 2ϩ ] i was never observed with the cell-impermeant CaR agonist Gd 3ϩ .
The characteristics of the CaR in the pancreas appeared slightly different from those previously described in heterologous expression systems (2,10). First, we found that it was necessary to use 1 mM Gd 3ϩ to elicit [Ca 2ϩ ] i responses in acinar cells, rather than the 600 M used in most heterologous expression systems. Second, the response to 8 mM Ca 2ϩ was also rather poor compared with that reported in cells overexpressing the CaR (2, 10). Finally, neomycin failed to produce any change in [Ca 2ϩ ] i in either acinar or duct cells. It is not entirely clear whether these data imply that the receptor in pancreatic acinar and duct cells is distinct from the CaR isolated from parathyroid or kidney cells, since damage to the extracellular domain of the receptor during enzymatic digestion (see above) could perhaps explain the reduced agonist sensitivity. In addition, the failure of neomycin to increase [Ca 2ϩ ] i in acinar and duct cells could reflect the fact that millimolar levels of neomycin are known to inhibit phospholipase C activity and therefore inositol trisphosphate production (24,25). Another possibility would be that the pharmacology of the CaR might be influenced by other regulatory proteins with tissue-specific distribution. However, the hypothesis of a different CaR subtype with a reduced affinity for Ca 2ϩ and polyvalent cations in the pancreas cannot be entirely ruled out, particularly given (i) the recent identification of a novel CaR splice variant in keratinocytes (26) and (ii) the possible existence of multiple CaR genes (27).
Although our data provide the first functional evidence for the existence of a CaR in pancreatic acinar cells, data preceding the cloning of the CaR in 1993 are consistent with our findings. For instance, elevating perfusate [Ca 2ϩ ] to 5 mM greatly potentiates amylase secretion from the cat pancreas (28), while Ma-ruyama (29) showed in whole cell patch-clamp experiments that extremely high concentrations (Ͼ50 mM) of divalent cations (Ca 2ϩ , Sr 2ϩ , Ba 2ϩ , Ni 2ϩ , and Mg 2ϩ ) activated Ca 2ϩ -dependent currents in pancreatic acinar cells by triggering intracellular Ca 2ϩ release. These latter results were ascribed by the author to the activation of a nonspecific cell surface receptor or receptor-effector complex via the indirect action of cations on membrane surface charge (29). In retrospect, the data are quite consistent with the presence of the CaR, although a nonspecific effect on membrane charge cannot be ruled out.
The presence of the CaR in pancreatic acinar cells may have physiological implications, as it suggests that increased [Ca 2ϩ ] o may be able to stimulate exocrine secretion in the absence of hormonal or neuronal stimulation. The exact concentration of Ca 2ϩ required to stimulate secretion in vivo would be very difficult to deduce from this study. However, given that CaR(s) are sensitive to values of [Ca 2ϩ ] o in the physiological range, it is tempting to speculate that stimulation of the CaR might play a role in producing the basal in vivo pancreatic secretion which is observed in a number of species (3). In addition, several lines of evidence indicate that hypercalcemia induces acinar cell damage and eventually pancreatitis (6). The largest effects are obtained with Ca 2ϩ concentrations ranging between 0.6 and 5 mM (5), consistent with the range of activation of the CaR (2). Our data suggest the intriguing hypothesis that these effects might be mediated by CaR present at the basolateral surface of acinar cells.
A further possible functional role for the CaR in acinar cells might be in controlling cell proliferation. Proliferation-associated pathways are known to be present in pancreatic acinar cells and are thought to be activated by CCK (30,31). Given the similar [Ca 2ϩ ] i responses evoked by CCK and Gd 3ϩ , it may be that the CaR in acinar cells helps to transduce a proliferative stimulus driven by extracellular Ca 2ϩ .
CaR in Exocrine Duct Cells-In contrast to acinar cells, duct cells exhibit a strong CaR immunostaining pattern along the apical surface delimiting the lumen of the ducts, suggesting that the receptor is localized to the luminal membrane. This staining pattern was observed both in tissue sections and in the isolated duct preparation used for the functional studies. It was also confirmed by the functional studies, which showed that Gd 3ϩ only elevated duct [Ca 2ϩ ] i when included in the luminal perfusate.
The effect of luminal Gd 3ϩ on [Ca 2ϩ ] i in ducts was very consistent, namely a gradual increase. No oscillations were observed, but this is not surprising given that oscillations in [Ca 2ϩ ] i have never been reported in agonist-stimulated duct cells. Although the gradual increase in [Ca 2ϩ ] i could conceivably also be attributed to an inhibition of Ca 2ϩ extrusion by Gd 3ϩ , this seems unlikely given that Gd 3ϩ only raised [Ca 2ϩ ] i when applied to the luminal membrane. Previous studies on ducts have generally employed only bath perfusion of agonists and other agents, with the result that most receptor systems identified in the duct cells are considered to be present on the basolateral plasma membrane.
The effects of raising luminal [Ca 2ϩ ] o in ducts were in general qualitatively and quantitatively similar to those obtained with Gd 3ϩ . However, the actions of [Ca 2ϩ ] o were less clear-cut than in acinar cells, mainly because basolateral (and probably also luminal) [Ca 2ϩ ] o also affects [Ca 2ϩ ] i in ducts via changes in Ca 2ϩ entry (16).
Possible Physiological Role of CaRs in the Pancreatic Duct System-The unequivocally luminal location of the CaR in the ducts suggests a role for CaRs in the regulation of ductal fluid secretion. This is strongly supported by our experiments showing that luminal Gd 3ϩ increased ductal HCO 3 Ϫ secretion. Lu-minal Gd 3ϩ -stimulated HCO 3 Ϫ efflux as effectively as forskolin, which we have previously shown to be a potent ductal secretagogue (7,15).
What might be the physiological significance of the stimulation of ductal secretion by a luminal Ca 2ϩ -sensing receptor? We wish to suggest a possible role in the prevention of pancreatic stone formation. Zymogen granules in acinar cells contain a high concentration of Ca 2ϩ and Mg 2ϩ (32,33). During exocytosis Ca 2ϩ is therefore released into the lumen, so that pancreatic enzyme secretion is tightly correlated with Ca 2ϩ and Mg 2ϩ release into the duct lumen (28,34). Excess Ca 2ϩ is potentially dangerous (see Introduction), and we propose that the CaR on the duct cells protects the pancreas by effectively "sensing" the local luminal Ca 2ϩ concentration, and stimulating ductal fluid secretion. Increased ductal fluid secretion would dilute the Ca 2ϩ derived from acinar secretion, thereby preventing precipitation of insoluble Ca 2ϩ salts. A similar role for a luminal CaR in the regulation of fluid transport in the presence of saturating concentration of calcium has been identified in the terminal part of the renal inner medullary collecting ducts, where the receptor may help prevent calcium stone formation by blunting the cellular response to vasopressin (35).
The proposal that ductal fluid secretion is regulated by an apical CaR is supported by the clinical evidence that pancreatitis accompanies familial hypocalciuric hypercalcemia (36,37). Familial hypocalciuric hypercalcemia is an autosomal, dominant disorder associated with an inactivating mutation of the CaR (36). A (partial) loss of CaR function in the duct system of the exocrine pancreas might conceivably, through stone formation, give rise to ductal hypertension or autodigestion (see above), which could account for the pancreatic damage during familial hypocalciuric hypercalcemia-induced pancreatitis.
Occurrence of CaR in Islet Cells-Our immunofluorescence studies also revealed strong staining at the surface of cells in the islets of Langerhans (Fig. 2), consistent with the presence of the receptor in glucagon-secreting ␣ cells and/or insulinsecreting ␤ cells. The presence of the CaR in pancreatic islets, specifically in ␤ cells, has previously been suggested by molecular and functional studies of human insulinoma cells (22,38). The presence of the receptor in normal rat tissue is interesting, since the selective arterial calcium injection test, which is used clinically to identify insulinomas (38), relies on the fact that only insulinoma cells, and not normal ␤ cells, respond to Ca 2ϩ infusion by increased insulin secretion. However, other studies have shown that raising [Ca 2ϩ ] o causes an increase in [Ca 2ϩ ] i in normal rodent ␤ cells (39,40). The immunolocalization of the CaR in islet cells in the present study, and the previous functional evidence, together suggest that the CaR is present on the ␤ cell plasma membrane and probably mediates Ca 2ϩ o -evoked insulin secretion.
Summary-We have shown that the extracellular Ca 2ϩsensing receptor is expressed in acinar and duct cells of the exocrine pancreas and in islet cells. Functional studies of acinar and duct cells confirmed the presence of the receptor, and showed that the receptor in the ducts can stimulate HCO 3 Ϫ efflux. This strongly suggests that the receptor may have a physiological role in controlling secretion. In the ducts this may be critical in ensuring that fluid secretion is sufficient to prevent the luminal free ionized Ca 2ϩ concentration from rising to levels which precipitate pancreatic stone formation in the presence of the high concentrations of bicarbonate that are typical of the pancreatic juice. In agreement with previous work, the receptor expressed in islet cells probably has a modulatory role in insulin secretion. Further studies are currently in progress to establish the precise nature and cellular distribution of the CaR in the pancreas, as well as the role of the receptor in pancreatic physiological and pathophysiological states.