Monoclonal Antibodies against MDRl P-glycoprotein Inhibit Chloride Conductance and Label a 65-kDa Protein in Pancreatic Zymogen Granule Membranes*

The regulation of C1- and cation conductances by the nonhydrolyzable ATP analog adenosine B’-(P,y-methyl-ene)triphosphate (AMP-PCP) was characterized in isolated zymogen granules (ZG) from pancreatic acinar cells. ZG were purified from rat pancreas homogenate by Percoll gradient centrifugation. C1- conductance was assayed by suspending ZG in isotonic KC1 buffer and measuring osmotic lysis induced by maximal permeabil- ization of ZG membranes (ZGM) for K’ with the K+ ionophore valinomycin (Val). This resulted in influx of K’ through the artificial pathway and of C1- through endog- enous channels. To measure cation conductances ZG (pH, - 6) were suspended in pH 7 buffered isotonic monovalent cation acetate salts. The pH gradient was converted into an outside-directed H’ diffusion potential by maximally increasing H’ conductance of ZGM with the protonophore carbonyl cyanide p-chlorophe-nylhydrazone. Osmotic lysis of ZG was induced by H’ diffusion potential driven influx of monovalent cations through endogenous channels and non-ionic diffusion of the counterion acetate. In the absence of Val, ZG were stable in KC1 buffer up to 2 h. “PCP enhanced os- motic lysis =&fold compared milk in TBSTN, were incubated with the monoclonal antibodies against MDRl or CFTR (5 pg/ml) in TBSTN plus 0.1% milk, overnight at room temperature, washed again and incubated for 1 h with horseradish peroxidase-con- jugated sheep anti-mouse immunoglobulin G (1:6,000 dilution in TBSTN with 0.1% milk). Following washing, blots were developed in en- hanced chemiluminescence reagents for 1 min and visualized on x-ray (Kodak X-Omat A R ) film aRer 3-240 min of exposure. Protein Determination-Protein was assayed, as described by Brad-ford (30), using bovine serum albumin as a standard.

The regulation of C1-and cation conductances by the nonhydrolyzable ATP analog adenosine B'-(P,y-methylene)triphosphate (AMP-PCP) was characterized in isolated zymogen granules (ZG) from pancreatic acinar cells. ZG were purified from rat pancreas homogenate by Percoll gradient centrifugation. C1-conductance was assayed by suspending ZG in isotonic KC1 buffer and measuring osmotic lysis induced by maximal permeabilization of ZG membranes (ZGM) for K' with the K+ ionophore valinomycin (Val). This resulted in influx of K' through the artificial pathway and of C1-through endogenous channels. To measure cation conductances ZG (pH, -6) were suspended in pH 7 buffered isotonic monovalent cation acetate salts. The pH gradient was converted into an outside-directed H' diffusion potential by maximally increasing H' conductance of ZGM with the protonophore carbonyl cyanide p-chlorophenylhydrazone. Osmotic lysis of ZG was induced by H' diffusion potential driven influx of monovalent cations through endogenous channels and non-ionic diffusion of the counterion acetate. In the absence of Val, ZG were stable in KC1 buffer up to 2 h. "PCP enhanced osmotic lysis =&fold compared to control, due to activation of C1-conductance by AMP-PCP and K' influx through an AMP-PCP-insensitive nonselective cation pathway, which could be blocked by 0.1 m~ Ba2+, 0.5 m~ quinine, or 0.2 m~ flufenamate. In addition, a K' and Rb+ selective cation conductance was found which was completely blocked by 0.5 m~ AMP-PCP or 0.5 m~ quinine. AMP-PCP induced C1conductance was strongly inhibited by two monoclonal antibodies against MDRl P-glycoprotein (JSB-1 and C219; 5-10 pg/ml), but not by a monoclonal antibody against the cystic fibrosis transmembrane conductance regulator (M3A7; 5 pg/ml) or by mouse IgG. The "PCP insensitive nonselective cation conductance was not blocked by monoclonal antibodies against MDRl P-glycoprotein (MDR1). Immunoblot studies of ZG membranes revealed the presence of a major immunoreactive protein band of 455 kDa with both monoclonal antibodies against MDRl, but no protein of the approximate size of MDRl (-170 kDa) was detected. W e propose that the C1-channel or a regulator of the channel, that is activated by the non-hydrolyzable ATP analog AMP-PCP in ZG membranes, is a member of the ATP binding cassette superfamily of transporters and may have homology to MDRl P-glycoprotein.
project "Sonderforschungsbereich 246-C6". The costs of publication of * This work was supported by "Deutsche Forschungsgemeinschaft" this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Upon stimulation by secretagogues, such as acetylcholine or cholecystokinin, or their second messengers diacyglycerol and inositol triphosphatelCa", pancreatic acinar cells secrete NaC1, fluid, and digestive enzymes. The mechanisms of regulated NaCl secretion in exocrine gland cells have been intensively investigated (1). NaCl secretion is thought to involve "active," cytoplasmic accumulation of C1-above its electrochemical equilibrium through three interlinked transporters at the basolat-era1 plasma membrane acting together as a C1-pump: the (Na' + K+)-ATPase, the Na+K+ 2C1-cotransporter and, depending on the exocrine gland under study, a K+-selective or nonselective cation channel (2,3). NaCl and fluid secretion is controlled by cytosolic Ca2+ signals which regulate the opening of the cation channels required for recirculation of K+ through the cotransporter. C1-secretion occurs through activation of luminal C1channels, which are regulated by internal Ca2+ as well (4, 5). Na+ flow may occur through paracellular pathways and the leaky tight-junctions into the electrically negative lumen.
Regulated secretion of digestive enzymes takes place through poorly understood processes that are morphologically characterized by fusion of zymogen granules with the apical plasma membrane and release of granular contents into the acinar lumen (6). Using permeabilized pancreatic acinar cells we have demonstrated that the Ca2+ and CAMP signaling pathways stimulate secretion of digestive enzymes (7). This type of studies also revealed an absolute dependence of regulated enzyme secretion on the ionic environment of the secretory granules: isosmotic replacement of C1-or of K+ by impermeant ions, or application of C1-and K' channel blockers abolished hormone-or second messenger-induced enzyme secretion (8).
Based on these observations, the presence of hormonally regulated C1--and K+-selective channels in the membranes of ZG' has been postulated: upon fusion of ZG with the luminal plasma membrane, the increased influx of salt and water through the granule C1-and K+ channels would promote enzyme secretion, possibly by enhanced decondensation and "flushing-out" of macromolecular enzymes (8,9). C1-and K+ permeabilities have been characterized in isolated ZG (9)(10)(11)(12). In addition to regulation by protein kinase-mediated protein phosphorylation (11,121, the C1-and K' permeabilities in ZG are modulated by ATP binding. The K+ conductance is blocked by ATP and its non-hydrolyzable analogs (12). These properties together with its pharmacological characteristics (inhibition by glibenclamide and activation by diazoxide) (12) suggest that the K+ conductance is mediated by channels similar or identical to ATP-sensitive K+ channels found in the plasma membrane of islet cells, neurons, muscle, and renal cells (13). The C1-conductance is activated by ATP and non-hydrolyzable analogs of ATP, such as AMP-PCP or AMP-PNP (14).
Recently, two members of the ATP binding cassette superfamily of transporters with properties of C1-channels have been cloned: the cystic fibrosis conductance transmembrane regulator (CFTR) (15) and the multidrug resistance P-glycoprotein (MDR1) (16). They are activated by ATP, the mechanisms of activation, however, are only partially understood. The CFTR C1-channel is activated by protein kinase A-and C-mediated phosphorylation (171, but activation may also require ATP binding and hydrolysis (18)(19)(20). The MDRl associated C1-current is induced by swelling and requires allosteric interaction with ATP for activation (21,22).
Along-term goal of our studies is the molecular identification of the major transporters present in the ZG membrane and to understand their role in stimulus-secretion coupling. Monoclonal antibodies (mAb) are potentially useful tools in the identification strategy. They can detect proteins of interest in cell membranes and localize them in intact cells. Occasionally, they also alter the function of the protein studied, providing information linking structure and function. In the present study, we have examined the effect of mAb against MDRl and CFTR on the ion conductance pathways of pancreatic ZG and performed immunoblots of purified ZG membranes. The data indicate that the two mAb directed against cytosolic epitopes of MDR1, JSB-1 and (2219, are functional blockers of the C1-conductance of pancreatic ZG membranes that is activated by the non-hydrolyzable ATP analog AMP-PCP. A ZG membrane protein of -65 kDa was labeled by both antibodies against MDRl and could represent the C1channel or a regulator of the channel.

Methods
Isolation of ZG and Purification of ZG Membranes-Zymogen granules were isolated from the pancreatic glands of male Wistar rats (180-300 g, Charles River Wiga GmbH, Sulzfeld, Germany) as described earlier (12). Granule membranes were purified as described previously (26), with slight modifications: ZG were diluted about 10-fold in an ice-cold hypotonic lysis buffer containing 0.1 m M MgSO,, 5 m~ EDTA, 10 m M HEPES, 50 m M KSCN, adjusted to pH 7.0 with Tris, plus a protease inhibitor "mixture" (10 pg/ml leupeptin, 1 m~ benzamidine, 0.2 m M Pefabloc@SC, and 50 pg/ml trypsin inhibitor), and lysis was allowed to proceed on ice for 1 h. The clear suspension was centrifuged at 200,000 X g for 90 min. The pellet containing purified ZG membranes was stored in liquid nitrogen.
Assays for Ion Conductances of ZG-Cl-and cation conductances of pancreatic ZG were assayed according to a previously reported method for quantitative evaluation of macroscopic ion fluxes through endogenous conductance pathways of ZG membranes (10,12). This assay relies on the measurement of osmotic lysis of ZG resuspended in buffered isotonic salt solution following addition of electrogenic ionophores for counterions. If the membrane permeability for counterions is low, ZG will remain stable in isotonic solutions for several hours in the absence of ionophore, because no net salt accumulation can occur into the in-tragranular space. Addition of an electrogenic ionophore maximally permeabilizes the ZG membranes for the counterion, and net salt influx can take place through the endogenous ion conductance and the artificial pathway. The osmotic load of the intragranular space attracts water, the granules swell and lyse. The end point of salt influx through endogenous conductance pathways in the presence of ionophores is osmotic lysis of granules which decreases the absorbance of the granule suspension. It has previously been shown that the granule concentration is linearly related to the absorbance (10). Therefore osmotic lysis can be monitored by measuring the time-dependent absorbance change of the granule suspension at 540 nm. Bulk salt influx into the intragranular space and the resulting granule lysis is limited by the flux of ions through the endogenous conductance pathway, but not by the flux of counterions through the shunt pathway. Consequently, the slope of the decrease in absorbance with time will represent an estimate of the rate of ions transported through the endogenous conductance pathway.
Anion conductance was measured by resuspending ZG in a solution buffered with 20 m M HEPES to pH 7.0 (iso-osmotic 150 mM KC11 in a cuvette, adding 5 p~ valinomycin, which selectively and maximally permeabilizes ZG membranes to the major cation K'. Under these conditions, gradient-driven influx of KC1 and water takes place through the valinomycin K+ shunt pathway and the endogenous C1-permeability, resulting in granular swelling and lysis. In this assay, the rate of granular lysis is limited by the endogenous C1-permeability. To measure cation conductance, zymogen granules were suspended in 150 m M monovalent catiodacetate solutions containing 1 m M EDTA, buffered with 50 m~ imidazole, and titrated to pH 7.0 with acetic acid. Since the intragranular pH is more acidic than the incubation medium, an inside-to-outside directed proton concentration gradient between intragranular space and incubation solution was generated. To control the magnitude of this pH gradient, the permeant buffer imidazole was employed. Under these conditions, the transmembrane pH gradient was about 0.5 pH units (12). Cation influx through endogenous cation pathways was initiated by addition of 16 p~ CCCP, an electrogenic protonophore. CCCP maximally permeabilizes the granular membrane to protons and converts the proton concentration gradient to an insidenegative proton diffusion potential. The inside-negative granular membrane potential, in turn, energizes cation influx through endogenous cation permeabilities of the ZG membrane. Acetate does not permeate the ZG membrane. When the buffer acetate concentration is high, anion influx occurs through the uncharged molecule acetic acid, which permeates through the lipid membrane by non-ionic diffusion and dissociates to provide the intragranular space continuously with protons for protonation of imidazole as well as for proton efflux from the acid interior (12). In this assay H' permeation and non-ionic diffusion of imidazole acetate are not rate-limiting. Intragranular accumulation of cation acetate salts, water influx, and osmotic lysis of the granules are determined by cation influx through endogenous cation permeabilities. Typically, 50-70 pg of granules were suspended in a cuvette containing 3 ml of buffered salt solution, which corresponds to an initial absorbance of 0.3-0.4. Absorbance was usually recorded continuously for 20-50 min at 37 "C with a Beckman DU-64 spectrophotometer equipped with a Peltier constant-temperature chamber, an automatic six-unit sampler, and a kinetics Soft-Pac Module. Data were stored using a Beckman Light Scatter program and analyzed using a Symphony spreadsheet program (Lotus Development GmbH, Munich, Germany).
Analysis and Validation of D-ansport Data-The half-time of granular lysis was estimated from the slope of the decrease in absorbance with time between ionophore addition and either experimental halftime, or the entire observation period if the half-time was not reached. The slope of the absorbance change with time was estimated by linear regression of the digitized data. Lysis rates were expressed as halftimes of granular lysis or its reciprocal value, i.e. the inverse half-time of lysis which was considered proportional to the rate constant of lysis. Linear regression analysis of the data actually represents an oversimplification of the actual kinetics of lysis. For instance, in experiments with AMP-PCP carried out in the absence of ionophore, lysis occurred after some latency (see Fig. 2). For these curves linear regression analysis of the data was carried out either from the onset of the measurements or from the onset of lysis. Lysis rates calculated from the onset of lysis were -2-fold higher than those calculated from the entire observation period. However, the dose response curves for AMP-PCP (Fig. 3) or for the effect of antibodies on lysis (Fig. 5) were not different. In these experiments values for lysis rates were calculated from the data of the entire observation period.
Unless otherwise indicated, experiments were repeated at least with three different granule preparations and data expressed as means S.D. of different preparations. When drugs were added in dimethyl sulfoxide or ethanol, the control experiments received the same amount of carrier solvent. Statistical analysis of data was carried out with the Statgraphics program using paired and unpaired Student's t test. Results with levels of p < 0.05 were considered significant.
Cell Culture and Preparation of Cell Lysates-The colchicine-resistant Chinese hamster ovary cell line (CHRC5) was cultured essentially as described by Ling and Thompson (27), using 5 pg of colchicine/ml of medium. Cells grown to confluency were harvested and solubilized by sonication for 30 s in Tris-buffered saline containing 0.1% Triton X-100, 10 pdml leupeptin, 1 m benzamidine, and 0.2 m~ Pefabloc@SC.
Detergent-insoluble material was removed by ultracentrifugation and the supernatant stored until use at -20 "C.
SDS-PAGE and Western BZotting-Rat pancreas homogenate, ZG membranes, or CHRC5 cell lysates were dissolved in sample buffer and heated for 30 min at 37 "C. SDS-polyacrylamide gel electrophoresis (SDS-PAGE) of membrane proteins was carried out on 7.5-15% linear gradient acrylamide Laemmli (28) minigels. Proteins were transferred to polyvinylidine difluoride membranes in a Bio-Rad mini-electrotransfer apparatus (Munich, Germany) at 20 V overnight according to Towbin et nl. (29) with 192 mM glycine, 250 m M Tris, and 5% methanol as transfer buffer. Blots were blocked with 3% milk in Tris-buffered saline containing 50 m M Tris (pH 7.0), 150 m M NaCI, 0.2 m M NaN,, 0.05% (v/v) Tween 20 (TBSTN) for 6 h at room temperature. Polyvinylidine difluoride membranes, washed three times with 0.1% milk in TBSTN, were incubated with the monoclonal antibodies against MDRl or CFTR (5 pg/ml) in TBSTN plus 0.1% milk, overnight at room temperature, washed again and incubated for 1 h with horseradish peroxidase-conjugated sheep anti-mouse immunoglobulin G (1:6,000 dilution in TB-STN with 0.1% milk). Following washing, blots were developed in enhanced chemiluminescence reagents for 1 min and visualized on x-ray (Kodak X-Omat A R ) film aRer 3-240 min of exposure.
Protein Determination-Protein was assayed, as described by Bradford (30), using bovine serum albumin as a standard.

RESULTS
The Non-hydrolyzable ATP Analog AMP-PCP Blocks K+ Conductance-When ZG were isolated from exocrine pancreas as described under "Methods," they were osmotically stable in isotonic K-acetatehmidazole solution up to 2 h, unless the protonophore CCCP (16 p~) was added to the cuvette, which resulted in enhanced lysis of the granules (Fig. L4). This behavior is explained as follows. In the presence of the protonophore CCCP, the proton concentration gradient across the membrane is converted to an inside-negative diffusion potential that serves as driving force for cation influx through the endogenous K+ permeability. Continuous acetic acid influx maintains the proton gradient and membrane potential. The endogenous K+ conductance present in the membrane becomes rate-limiting for K' influx and the osmotic load (net uptake of K-acetate) induces first granular swelling and finally granular lysis.
Incubation of granules in the presence of the non-hydrolyzable ATP analog AMP-PCP inhibited K+ conductance in a dosedependent fashion (Fig. 1, A and B ) . A maximal reduction of inverse half-time of lysis to 52 * 14% of the controls without AMP-PCP was observed at a concentration of 0.25 mM AMP-PCP; higher concentrations (up to 1 mM) did not further inhibit K+ conductance (Fig. 1B). The experiments were carried out in the presence of 5 m~ M e . Qualitatively similar results were found in the absence of M e , although inhibition by AMP-PCP was less pronounced (not shown). This confirms a previous report, where concentrations of AMP-PCP up to 5 m~ in the absence of Mg2' reduced K+ conductance by about 40% (12). The Non-hydrolyzable ATP Analog AMP-PCP Activates Cl-Conductance and Induces Osmotic Lysis in the Absence of Valinomycin-When ZG were incubated in isotonic HEPESbuffered KC1 solution, they were osmotically stable up to 4 h. Addition of the electrogenic potassium ionophore valinomycin (5 p~) enhanced lysis of the granules (Fig. 2) allowing influxes of K and C1-down their respective electrochemical gradients (through the artificial K+ conductance and the endogenous C1conductance), followed by water influx. As the valinomycin- induced K+ conductance was high, the endogenous C1-conductance present in the membrane became rate-limiting for KC1 flux and subsequent lysis of granules. We had previously demonstrated that adenine nucleotides, such as ATP (above 50 p~) , or the non-hydrolyzable analogs AMP-PCP and AMP-PNP increase C1-conductance in pancreatic and parotid Z G the most pronounced effects were observed with AMP-PCP which in- Interestingly, AMP-PCP in the absence of valinomycin enhanced the rate of ZG lysis after an initial lag time of 5-10 min (Fig. 2). This ionophore independent osmotic lysis apparently required activation of specific ion transporters, since no lysis occurred, when KC1 was replaced by sucrose or mannitol mole by mole (not shown). The increase in osmotic lysis observed with AMP-PCP in the absence of valinomycin showed a halfmaximal effect at -50 p~ and was maximal above 0.25 mM (Fig.  3). Inverse half-times of lysis in controls without valinomycin were calculated to 0.40 f 0.05 h" and increased to 1.46 2 0.06 h-' by addition of 0.25 mM AMP-PCP ( p < 0.001). In other words, 0.25 m~ AMP-PCP stimulated osmotic lysis in the absence of valinomycin by -4-fold.
Valinomycin Independent ZG Lysis Induced by AMP-PCP Involves a Nonselective, AMP-PCP Insensitive, Cation Conductance Pathway-Following replacement of K-acetate by Na-acetate, CCCP also induced granule lysis, suggesting Na' flux through a Na+ permeable electrogenic ion pathway (Table I).
AMP-PCP at concentrations ranging between 0.1 and 1 mM did not significantly inhibit the Na' permeable conductive pathway (5.6 2 3.5 h" in controls versus 4.8 2 3.0 h-' in the presence of 0.5 mM AMP-PCP). We tested in a similar manner the inhibitory effect of AMP-PCP on monovalent cation conductance pathways in the presence of Cs+, Li', K+, or Rb' (Table I). Similarly as for Na' conductance, no significant inhibition of lysis rates by 0.5 m~ AMP-PCP was observed for Cs+ conductance (from 5.4 2 3.2 h-' to 3.7 2 1.8 h-'), and for Li' conductance (from 2.5 2 1.0 h-' to 2.1 2 0.8 h-'). In contrast, with K+ the rate constants of lysis were significantly reduced from 12.4 -c 5.7 h-l to 5.1 2 1.8 h-l ( p < 0.005). Rate constants for Rb' conductance were also significantly reduced by 0.5 m~ AMP-PCP from 76.2 2 42.2 h" to 39.9 2 13.7 h" ( p < 0.05). These data suggest that at least two monovalent cation conductance pathways are present in ZG: 1) a large (-50%) Rb+-and K+-selective pathway which is inhibited by AMP-PCP (AMP-PCP sensitive component of Table I), hereafler referred to as K+-selective pathway.

Selectivity of AMP-PCP-sensitive and -insensitiue components of pancreatic ZG cation permeabilities
Granules were suspended in imidazole-buffered medium containing 150 m~ of the acetate salt listed in the Table. Inverse half-times of lysis (rates of lysis; h-l) were measured after addition of 16 p~ CCCP. Experiments were performed in the absence (total cation conductance) or presence of 0.5 m M of the nucleotide AMP-PCP (AMP-PCP-insensitive cation conductance). The calculated difference between both conditions is equivalent to the AMP-PCP-sensitive cation conductance of ZG. 2) An AMP-PCP-insensitive pathway which is permeable to all monovalent cations tested in the sequence Rb' > K+ = Na' 2 Cs' > Li' (see Table I), hereafter referred to as nonselective cation pathway.
The K+-selective a n d Nonselective Cation Conductance Pathway Have Distinct Inhibitor Sensitivities"l'0 characterize both cation conductance pathways, we tested various inorganic and organic pharmacological agents which block other K+ channels andor nonselective cation channels. We compared the inhibitory effect of these agents on the cation conductance of granules incubated in K-acetate, which predominantly permeates the K+-selective pathway, and of granules incubated in Na-acetate, which almost exclusively permeates the nonselective cation pathway (see Table I).
In this set of experiments AMP-PCP at the maximal inhibitory concentration of 0.5 m~ significantly decreased K+ conductance to 58 * 15% of the controls without nucleotide ( p < 0.001), Le. it mainly blocked the K+-selective pathway (Fig. 4). Na+ conductance was not inhibited by AMP-PCP (91 T 11% of control; not significant). Quinine, a lipophilic inhibitor of certain types of K+ channels, inhibited K+ conductance in a dosedependent way. Half-maximal inhibition was obtained at -0.1 m~ quinine and maximal inhibition at 0.5 mM (not shown). This concentration significantly reduced K+ conductance to 9 2 4% of controls ( p < 0.001). A maximal inhibitory concentration of quinine of 0.5 mM also significantly reduced the nonselective cation conductance measured with Na' as major cation to 21 * 11% of controls ( p < 0.001) (Fig. 4). This indicates that quinine blocks both the K+-selective pathway and the nonselective cation conductance. Ba2+, depending on the type of K channel, blocks from the outside or from the cytosolic side of the channel (31). In the experimental conditions of this study, putative channel proteins of zymogen granule membranes have their cytosolic domains exposed to the incubation medium. Surprisingly, micromolar concentrations of Ba2+ (ranging between 10 and 100 p~) selectively inhibited nonselective cation conductance. 100 p Ba2+ inhibited Na+ conductance to 14 +-2% of controls ( p < 0.001), whereas the same concentration of Ba2+ reduced K+ conductance to 56 2 15% of controls ( p < 0.001) (Fig.   4). Since the AMP-PCP-sensitive (K+-selective) and -insensitive (nonselective) components represent about 50% of total K+ conductance each (see Table I and Fig. 4), the almost complete inhibition of Na' conductance (-90% inhibition) and the inhibition of K+ conductance by -45% with 100 p~ Ba2+ indicate that Ba2+ at low micromolar concentrations blocks the nonselective component of K+ conductance only. Higher, millimolar concentrations of Ba2+ also block the K+-selective component of ZG cation conductance (12). The nonsteroidal anti-inflammatory drug flufenamic acid has been used as a blocker of nonselective cation channels in inside-out patches from rat exocrine pancreas (32). As shown in Fig. 5 , 200 p~ flufenamic acid reduced K+ conductance to 62 * 14% of controls. Inhibition of Na+ conductance was more potent although not complete (inhibition to 41 5 2% of controls), suggesting that 200 PM flufenamic acid partially inhibits the nonselective cation conductance of ZG.
The results shown in Fig. 4 provide further evidence for the presence of a K+-selective conductance which is blocked by AMP-PCP and quinine, and of a n AMP-PCP-insensitive, nonselective cation conductance which is selectively blocked by micromolar concentrations of Ba", flufenamic acid, and quinine as well.
Activation of ZG C1-Conductance by AMP-PCP Is Abolished by Monoclonal Antibodies against MDRl P-glycoprotein-Activation of ZG C1-conductance by the non-hydrolyzable ATP analog AMP-PCP (Fig. 2), is reminiscent of a property of a volume-regulated C1-channel associated with MDRl P-glycoprotein which is activated by ATP, ATP-yS, AMP-PNP, and AMP-PCP (21,22). We therefore tested two different monoclonal antibodies against MDRl P-glycoprotein, which are directed against cytosolic domains of the protein, on AMP-PCP induced ZG lysis in the absence of valinomycin (Fig. 5). 0.1 m~ AMP-PCP stimulated lysis of ZG incubated in KC1 buffer in the absence of valinomycin 3.8-5.2-fold. The commercially available mouse monoclonal antibody against MDRl P-glycoprotein C219, which binds to the two sequences VQAALD and VQEALD found 6 residues away from the consensus sequence of the B site of the two ATP-binding domains of MDRl (24), reduced AMP-PCP induced lysis of ZG in a concentration-dependent manner. In three different experiments the maximally tested concentration of 5 pg/ml C219 reduced ZG lysis induced by 0.1 mM AMP-PCP from 1.7 '' 0.3 h" to 1.1 ' . 0.2 h" ( p < 0.025). Another monoclonal antibody, the JSB-1, which binds to cytosolic domains of MDRl P-glycoprotein (23), was even more inhibitory. JSB-1 reduced ZG lysis as a function of its test * L pg/ml JSB-1 (p < 0.01; means 3 S.D. of three different experiments). CFTR is another C1-channel with nucleotide binding domains, which is activated by protein kinase A-mediated phosphorylation and binding of hydrolyzable ATP analogs, which possibly involves ATP hydrolysis (18)(19)(20). The monoclonal antibody against the cytosolic, carboxyl-terminal domain of CFTR (M3A7) (24) was tested on AMP-PCP induced ZG lysis at the same concentrations as the two antibodies against MDRl P-glycoprotein. Even at the concentration of 5 pg/ml M3A7 had no effect on the enhanced ZG lysis observed with 0.1 mM AMP-PCP (Fig. 5). Also mouse IgG (5 pglml) did not reduce the rate of ZG lysis induced by 0.5 mM AMP-PCP (1.5 2 0.1 h" in controls with AMP-PCP compared to 1.7 t 0.2 h" for experiments with AMP-PCP and IgG, means * S.D. of four different experiments).
Inhibition of AMP-PCP induced lysis of ZG incubated in KC1 buffer by the monoclonal antibodies JSB-1 and C219 (Fig. 5 ) could be the result of a block of the AMP-PCP activated C1conductance pathway, of the nonselective cation conductance, or both. We have therefore tested the effect of 5 pg/ml of the monoclonal antibodies JSB-1 and C219 (MDR1 P-glycoprotein) on the lysis of ZG incubated in Na-acetate buffer after addition of CCCP, i.e. on the Na' permeant, AMP-PCP-insensitive nonselective cation conductance pathway (see Table I). JSB-1 and C219 (5 pg/ml) had no inhibitory effect on Na+ conductance (not shown). Similar results were obtained with 5 pg/ml of the monoclonal antibody against CFTR, M3A7. In contrast, when the C1-conductance was investigated by incubating ZG in KC1 buffer and adding 5 p~ valinomycin, both monoclonal antibodies against MDRl P-glycoprotein JSB-1 (Fig. 6 A ) and C219 (Fig. 6B) (5 and 10 pg/ml) inhibited the C1-conductance activated by 0.5 mM AMP-PCP, although the monoclonal antibody C219 was less inhibitory. Mouse IgG (5 and 10 pg/ml) did not affect the C1-conductance activated by AMP-PCP.
MDRl P-glycoprotein Antibodies Label a Major Immunoreactive band of -65 kDa in Western Blots of ZG Membranes-Western blots of ZG membrane proteins, which had been separated by SDS-PAGE on 7.5-15% linear gradient gels and transferred to polyvinylidine fluoride membranes, revealed the presence of a major immunoreactive protein band of -65 kDa with both monoclonal antibodies against MDRl P-glycoprotein JSB-1 (Fig. 7A) and C219 (Fig. 7 B ) (both 5 pg/ml). Minor immunoreactive bands were also observed with both monoclonal antibodies at about 55 and 30 kDa (Fig. 7, A and B ) . However, no protein band was found at the location of MDRl P-glycoprotein. For comparison, when cell lysates of colchicine-resistant CHRC5 cells, which overexpress MDRl P-glycoprotein (331, were subjected to Western blotting, both antibodies detected an immunoreactive protein band of approximately 170 kDa, as expected for MDRl P-glycoprotein (33). In addition, Western blots of pancreas homogenate and ZG membranes with C219 antibody demonstrated that the -65-kDa band was more abundant in ZG membranes, as compared to pancreas homogenate (Fig. 7C). No immunoreactive band was detected in ZG membranes with the monoclonal antibody against CFTR, M3A7 (not shown).
In summary, these results have demonstrated that the two monoclonal antibodies against MDRl P-glycoprotein JSB-1 and C219 are functional and specific blockers of a C1-conductance pathway in ZG membranes which can be activated by the nonhydrolyzable ATP analog AMP-PCP. Western blot analysis of ZG membranes using both JSB-1 and C219 revealed a major immunoreactive band of about 65 kDa molecular mass, but no band was detected at the approximate molecular mass of MDRl P-glycoprotein. valinomycin must occur as a consequence of both C1-and K+ influx through endogenous pathways, resulting in osmotic lysis of ZG. As illustrated in Fig. 1, however, the K+ conductance present in ZG membranes is blocked by 0.5 mM AMP-PCP. This K+ conductance pathway therefore cannot account for granular lysis observed in KC1 solutions after addition of AMP-PCP and in the absence of valinomycin. An AMP-PCP insensitive K+ permeability is also present in ZG membranes, which accounts for -50% of total K+ conductance (see Fig. lB) and may explain K+ influx and osmotic lysis of granules in the absence of valinomycin in conditions where the C1-conductance has been activated by AMP-PCP (see model of Fig. 8). The major criteria to discriminate between these conductance pathways are their sensitivity to AMP-PCP, their cation selectivity, and their inhibitor sensitivity ( Fig. 4 and Table I).

The AMP-PCP Insensitive Cation Conductance of ZG Has
The AMP-PCP-sensitive pathway is selective to K+ and Rb' only and can be blocked by glibenclamide (12). The AMP-PCPinsensitive, nonselective cation conductance pathway has similar properties to those of nonselective cation channels found in apical and basolateral plasma membranes of secretory tissues and cultured cell lines, but exhibits two major differences. Like other nonselective cation channels it is equally permeable to Na+ and K+ (3,34,35). In agreement with previous reports it is inhibited by quinine (34) and selectively blocked by flufenamic acid (32) (see Fig. 4). In contrast to published data (3), adenine nucleotides, e.g. AMP-PCP, did not inhibit the nonselective cation channel of ZG (Fig. 4). Furthermore, Ba2+ was found to specifically block the nonselective component of ZG cation conductance at concentrations ranging between 10 and 100 p~, an effect which has not been described for nonselective cation channels in other cells.
The Cl-Conductance of Pancreatic Zymogen Granules Has Functional Properties of a Volume-regulated Cl-Channel Associated with MDRl P-glycoprotein"MDR1 is an ATPase that is responsible for active efflux of various hydrophobic drugs out of cells (36). Valverde and co-workers (21) have demonstrated that hypotonicity causes the appearance of a volume-regulated C1current only in cell lines overexpressing MDR1, suggesting that MDRl is a C1-channel. This interpretation has been questioned by several reports, which indicate that MDRl expression may not be correlated to C1-currents involved in volume regulation (37)(38)(39). Despite this controversy the C1-conductance of ZG has characteristics of the C1-current described by Valverde and co-workers (21). The MDRl associated C1-current is blocked by DIDS, verapamil, and tamoxifen a t micromolar con- centrations (21,40). Similarly, the ZG C1-conductance is blocked by DIDS with a half-maximal effect at 2 p~ (10). 100 PM verapamil reduces ZG lysis induced by 0.1 m M AMP-PCP from 1.8 2 0.2 h" to 1.2 f 0.2 h" and 5 p~ tamoxifen from 2.6 2 0.6 h" t o 2.0 2 0.3 h-l (both p < 0.03; n = 4-5) (not shown). Furthermore, ZG C1-conductance is activated by ATP above 50 PM and by the nonmetabolizable ATP analogs AMP-PCP and AMP-PNP (141, in the absence or presence of Mg2'. Similar effects are observed on the MDRl C1-channel, when cytosolic ATP is replaced by equimolar concentrations of ATPyS, AMP-PNP, or AMP-PCP, in the presence or absence of Mg2' (20). ATPyS, however, inhibits ZG C1-conductance (14).
Another C1-channel which displays activation by adenine nucleotides is the CFTR C1-channel. It is activated by ATP, but this process involves phosphorylation by CAMP-dependent protein kinase, requires Mg2' and possibly ATP hydrolysis (18)(19)(20). Another difference is the sensitivity to anion channel blockers: the C1-current associated with CFTR is not sensitive to DIDS, but rather to diphenylamine-2-carboxylic acid (41). A C1-conductance in vesicles of pancreatic endoplasmic reticulum also has similar properties to the ZG C1-conductance: it is activated by ATP and AMP-PCP, but not by ATPyS, and is inhibited by 100 p~ DIDS or SITS and by 10 p~ indanyloxyacetic acid 94-95 (42). A 64-kDa C1-channel protein from kidney microsomes (43) has been detected in the pancreatic endoplasmic reticulum membranes, suggesting that it is the C1channel of pancreatic endoplasmic reticulum. This protein has no homology to MDRl or MDR1-related ATP-binding proteins (43). Its relation to the AMP-PCP activated C1-channel of pancreatic zymogen granules remains unclear.
Inhibition of Pancreatic Zymogen Granule Cl-Conductance by Monoclonal Antibodies against MDRl P-glycoprotein Is Specific-So far, few antibodies raised against C1-channel proteins were functional and therefore could be used as tools to identify and purify C1-channels. 1) Finn et al. (44) immunopurified a 200-220-kDa protein from Necturus gallbladder with a functional monoclonal antibody which inhibits apical C1-conductance in this tissue and a C1-channel of 62 pS when reconstituted into planar bilayers (45). 2) Polyclonal antibodies against a CFTR-derived synthetic peptide corresponding to amino acids 505-511 of CFTR abolished CAMP-dependent C1currents in T84 cells and attenuated swelling activated current, whereas Ca2+-dependent C1-current remained unaffected (46).
Several lines of evidence support our interpretation that the inhibitory effects of the antibodies against MDRl are specific. I) Nanomolar concentrations of both antibodies blocked C1-conductance (Figs. 5 and 61, suggesting direct, stoichiometric interaction of the antibodies with the C1-channel protein. 2) Two control antibodies, the CFTR monoclonal antibody M3A7 and mouse IgG, had no inhibitory effect on the C1-conductance of ZG (Figs. 5 and 6). 3) The MDRl antibodies JSB-1 and C219 had no inhibitory effect on the nonselective cation conductance pathway of the granule membrane, which excludes a nonspecific interaction of this antibody with the granule membrane. 4) Inhibition of AMP-PCP activated C1-conductance by the antibodies JSB-1 and C219 described in Fig. 6 was not caused by nonspecific block of the K+ ionophore valinomycin, since the antibody also blocked AMP-PCP-activated C1-conductance in the absence of valinomycin (Fig. 5). 5 ) A nonspecific effect of the antibodies (as well as of activators and inhibitors of C1-conductance) on H,O permeability of the granule membranes or alteration of the colloidal properties of the granule content by these substances can also be excluded, since granule lysis induced by incubation in hypotonic solutions was unaffected by these compounds (not shown).
Is the Cl-Channel of ZG Membranes an MDRl P-glycoprotein Retated Protein?-Immunocytochemical studies have detected MDRl predominantly in the apical membrane of transport epithelia, such as liver, kidney, pancreas, colon, and jejunum (47). Very little MDRl was found in intracellular organelles, e.g. Golgi stacks and endoplasmic reticulum (48). Interestingly, C219 reactivity was also found in rat pancreatic acini by peroxidase immunocytochemistry (49). Recently, a 65-kDa MDRl related protein ("mini-P-glycoprotein") that is overexpressed in multidrug-resistant cells has been identified in plasma membranes by immunoblot studies with the monoclonal antibody C219 (50). Northern blot analysis with an MDRl cDNArevealed the presence of a 2.4-kb mRNA, suggesting a truncated or alternatively spliced transcript of MDRl (50). Its function is not known. Several members of the MDRIrelated ATP binding cassette superfamily of transporters have been described in intracellular organelles, such as the 70-kDa peroxisomal protein of mammalian liver (511, or the 170-kDa P-glycoprotein homologue detected in the digestive vacuole of Plasmodium falciparum (52). It is therefore likely that the -65-kDa protein detected in rat pancreatic ZG membranes also represents a member of the MDRl related ATP-binding protein superfamily. It is rather improbable that this 65-kDa protein band is a degradation product of MDR1, since the mAb C219 (which is directed against N-and C-terminal sites of MDR1; Ref. 24) should also detect an equally abundant, complementary fragment of E= 100 kDa. Since both monoclonal antibodies against MDRl (C219 and JSB-1) specifically inhibit AMP-PCPactivated C1flux, we suggest that this -65-kDa MDR1-like protein is the C1-channel protein or a regulator of the C1channel.