Cystic Fibrosis Transmembrane Conductance Regulator Regulates Luminal Cl−/HCO3 −Exchange in Mouse Submandibular and Pancreatic Ducts*

We have demonstrated previously the regulation of Cl−/HCO3 − exchange activity by the cystic fibrosis transmembrane conductance regulator (CFTR) in model systems of cells stably or transiently transfected with CFTR (Lee, M. G., Wigley, W. C., Zeng, W., Noel, L. E., Marino, C. R., Thomas, P. J., and Muallem, S. (1999)J. Biol. Chem. 274, 3414–3421). In the present work we examine the significance of this regulation in cells naturally expressing CFTR. These include the human colonic T84 cell line and the mouse submandibular gland and pancreatic ducts, tissues that express high levels of CFTR in the luminal membrane. As in heterologous expression systems, stimulation of T84 cells with forskolin increased the Cl−/HCO3 − exchange activity independently of CFTR Cl− channel activity. Freshly isolated submandibular gland ducts from wild type mice showed variable Cl−/HCO3 −exchange activity. Measurement of [Cl−] i revealed that this was largely the result of variable steady-state [Cl−] i . Membrane depolarization with 5 mm Ba2+ or 100 mmK+ increased and stabilized [Cl−] i . Under depolarized conditions wild type and ΔF/ΔF mice had comparable basal Cl−/HCO3 − exchange activity. Notably, stimulation with forskolin increased Cl−/HCO3 − exchange activity in submandibular gland ducts from wild type but not ΔF/ΔF mice. Microperfusion of the main pancreatic duct showed Cl−/HCO3 − exchange activity in both the basolateral and luminal membranes. Stimulation of ducts from wild type animals with forskolin had no effect on basolateral but markedly stimulated luminal Cl−/HCO3 − exchange activity. By contrast, forskolin had no effect on either basolateral or luminal Cl−/HCO3 −exchange activity of ducts from ΔF/ΔF animals. We conclude that CFTR regulates luminal Cl−/HCO3 − exchange activity in CFTR-expressing cells, and we discuss the possible physiological significance of these findings regarding cystic fibrosis.

؊ exchange activity of ducts from ⌬F/⌬F animals. We conclude that CFTR regulates luminal Cl ؊ /HCO 3 ؊ exchange activity in CFTR-expressing cells, and we discuss the possible physiological significance of these findings regarding cystic fibrosis.

HCO 3
Ϫ secretion is a primary function of many CFTR 1 -ex-pressing cells (1)(2)(3). Most of the HCO 3 Ϫ is secreted by duct or duct-like cells to the lumen and thus requires transductal HCO 3 Ϫ transport. Little is known about the pathways mediating HCO 3 Ϫ entry in the basolateral membrane (BLM). The best studies available to date are in the pancreatic duct, in which Case and co-workers (4 -6) provided strong evidence that HCO 3 Ϫ influx is mediated largely by a BLM Na ϩ -HCO 3 Ϫ cotransport. HCO 3 Ϫ efflux across the luminal membrane (LM) and its regulation are equally poorly understood. Most models assume that the electroneutral portion of HCO 3 Ϫ secretion is mediated by a luminal Cl Ϫ /HCO 3 Ϫ exchanger (AE, anion exchanger) (1)(2)(3)6). This function is also believed to mediate part of Cl Ϫ absorption by the duct. Cl Ϫ is supplied to the duct lumen in the plasma-like primary fluid secreted by acinar cells (1)(2)(3)7). The pathophysiology of cystic fibrosis indicates that CFTR plays a critical, but poorly defined, role in HCO 3 Ϫ secretion and Cl Ϫ absorption. In tissues such as the salivary glands, in which acinar cells secrete the bulk of the fluid, CFTR is assumed to mediate the electrogenic part of Cl Ϫ absorption (2). The same role is attributed to CFTR in sweat glands (7) and intestinal epithelia (8,9). Recent work suggests that airway epithelia absorb Na ϩ and Cl Ϫ to produce a hypotonic airway surface fluid (Ref. 10 and references within, but see Ref. 11 and references within). Because Na ϩ and Cl Ϫ concentrations in airway surface liquid produced by cystic fibrosis airway epithelium are isotonic (10,11), CFTR may mediate electrogenic Cl Ϫ absorption in airway epithelia (12,13). In glands like the pancreas, fluid secretion by acinar cells is limited, and the duct secretes the bulk of the fluid in pancreatic juice (1). In this type of gland the limited supply of Cl Ϫ secreted by acinar cells led to the proposal that CFTR mediates Cl Ϫ secretion to the lumen of duct cells to fuel the Cl Ϫ /HCO 3 Ϫ exchanger (1). However, a recent work showed that agonist-and cAMP-stimulated HCO 3 Ϫ secretion in guinea pig pancreatic duct is independent of luminal Cl Ϫ (6). Hence, the role of CFTR in ion transport by these tissues remains obscure.
To date direct evidence in support of the two models is meager indeed. Localization of CFTR in the luminal membrane of all CFTR-expressing epithelia is well documented (14 -16). Cl Ϫ /HCO 3 Ϫ exchange activity was found in the BLM and LM of pancreatic (17) and submandibular gland (SMG) ducts (18). In SMG ducts the AE isoform 2 (AE2) was localized in the BLM (19). The isoform(s) expressed in the LM is not known. * This work was funded in part by National Institutes of Health Grants DE12309 and DK38938 (to S. M.) and DK49835 (to P. J. T.). 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 U.S.C. Section 1734 solely to indicate this fact.

A relationship between HCO 3
Ϫ secretion and CFTR was documented in two cell lines and intestinal epithelia. In a human airway epithelial cell line CFTR-dependent HCO 3 Ϫ conductance (20,21) was proposed to be mediated by CFTR itself. By contrast, similar studies in a human pancreatic duct cell line concluded that electrogenic Cl Ϫ and HCO 3 Ϫ secretions are mediated by independent proteins (22). In duodenal epithelium basal and acid-stimulated HCO 3 Ϫ secretions were reduced or absent in CFTR Ϫ/Ϫ mice (23). Surprisingly, in a recent study Seidler et al. (24) showed that all forms of HCO 3 Ϫ secretion stimulated by agonists or agents that elevate cAMP, cGMP, and, in particular, [Ca 2ϩ ] i were impaired in the intestinal epithelia of CFTR Ϫ/Ϫ mice. These studies suggest the likely involvement of CFTR in the electrogenic component of HCO 3 Ϫ secretion, which is particularly prominent in the intestine (25,26). However, a large fraction of HCO 3 Ϫ secretion in tissues such as salivary glands (2) and the rat and mouse pancreas (1) is mediated by an electroneutral HCO 3 Ϫ transport mechanism. The role of CFTR in this critical component of HCO 3 Ϫ secretion is unknown.
The intimate relationship between CFTR expression and HCO 3 Ϫ secretion seen in intestinal (23,24) and airway epithelia (27) raises the question of whether and how CFTR modulates HCO 3 Ϫ secretion in other CFTR-expressing tissues. In addition to its possible function as a regulator of a HCO 3 Ϫ conductive channel, CFTR may also regulate luminal Cl Ϫ /HCO 3 Ϫ exchange activity. In a recent study (28) we used cells stably transfected with CFTR and transient transfection of WT CFTR and several CFTR mutants to demonstrate regulation of AE activity by CFTR. To evaluate the physiological relevance of these findings, in the present work we report the regulation of the luminal AE activity by CFTR in the mouse SMG and pancreatic ducts.
The SMG and pancreatic ducts were selected as model systems for several reasons. The fraction of electroneutral HCO 3 Ϫ secretion in these tissues is relatively high (1,2). Among all CFTR-expressing tissues, the mechanism of fluid and electrolyte secretion is understood best in the SMG (2). The SMG and pancreatic ducts express high levels of CFTR (15), and the abundance of ducts in the SMG (2, 29) facilitates experimentation. Although the mouse is not the ideal species to study HCO 3 Ϫ secretion (1,6,30), it was selected for the present work because of the availability of ⌬F/⌬F mice (31). To supplement the studies in the native ducts we also used the human colonic cell line T84 because these cells have been used extensively to study the properties of naturally occurring CFTR (32,33).
We show here that stimulation of CFTR with cAMP increased Cl Ϫ /HCO 3 Ϫ exchange activity in T84 cells. More importantly, stimulation of CFTR by cAMP resulted in selective activation of luminal Cl Ϫ /HCO 3 Ϫ exchange activity in SMG and pancreatic ducts of WT mice, which was absent in ducts prepared from ⌬F/⌬F mice. We conclude that CFTR regulates luminal Cl Ϫ /HCO 3 Ϫ exchange activity of SMG and pancreatic ducts and probably other CFTR-expressing cells.
Culture of T84 Cells-T84 cells were purchased from American Type Culture Collection (ATCC CCL 248, Rockville, MD) and maintained in a 1:1 mixture of Ham's F-12 medium and Dulbecco's modified Eagle's medium supplemented with 5% fetal calf serum. The cells were plated on a sterile 22 ϫ 40-mm coverslip at a density of 2.5 ϫ 10 5 cells/cm 2 for intracellular pH measurements.
Animals and Preparation of Ducts-A cystic fibrosis mouse model in which the ⌬F508 mutation was introduced in the mouse CFTR by gene targeting in ES cells (31) was obtained from Dr. Kirk R. Thomas (Eccles Institute of Human Genetics, HHMI, University of Utah School of Medicine, Salt Lake City). The mice were maintained on a standard diet, and genotyping was carried out on day 14 postpartum as described previously (16).
Duct fragments from the mouse SMG were prepared by a slight modification of our published procedure (16). Mice were sacrificed by exposure to a methoxyflurane-saturated atmosphere and subsequent cervical dislocation, and the SMGs were removed to a cold pancreatic solution A (PSA). The composition of PSA was (in mM) 140 NaCl, 5 KCl, 1 MgCl 2 , 1 CaCl 2 , 10 Hepes (pH 7.4 with NaOH), 10 glucose, 10 pyruvate, 0.1% bovine serum albumin, and 0.02% soybean trypsin inhibitor. Each gland was cleaned by injection of 5 ml of PSA and minced. The minced tissue was transferred to 8 ml of PSA containing 2.5 mg of collagenase (CLS4, 254 units/mg; Worthington Biochemicals) and digested for 8 -10 min at 37°C. The dissociated cells were then washed twice with PSA, resuspended in 2 ml of PSA, and kept on ice until use.
Microperfusion experiments were performed with microdissected pancreatic ducts from WT and ⌬F/⌬F mice. The procedure for preparation and perfusion of the main pancreatic duct was identical to that used for perfusion of the rat pancreatic duct (17). The ducts were dissected in PSA, cannulated, and perfused through the lumen and the bath with solution A. After completion of BCECF loading the ducts were perfused with HCO 3 Ϫ -buffered solution B for at least 10 min prior to manipulation of Cl Ϫ gradients.
Intracellular pH (pH i ) Measurements-The procedure of pH i measurement in T84 cells was identical to that described in detail in our recent work (28). In the case of SMG cells, the dissociated cells were loaded with BCECF by a 10-min incubation at room temperature in PSA containing 1 M BCECF-AM. The cells were then washed with PSA and plated on a polylysine-coated coverslip that was assembled into a perfusion chamber. The chamber was placed on an inverted microscope, and intralobular ducts were identified by morphology. The BCECF fluorescence of 10 -16 cells of a duct fragment was recorded at excitation wavelengths of 440 and 490 nm. Fluorescence ratios of 490/ 440 were calibrated using the procedures described previously (28). In the case of the perfused pancreatic duct BCECF loading was accomplished by including 2.5 M BCECF-AM in the luminal perfusate for 10 min.
Changes in Cl Ϫ /HCO 3 Ϫ exchange activity were estimated from the initial rate of pH i changes (T84 cells and pancreatic ducts) or from the extent of pH i changes (SMG ducts). Initial rates of pH i changes were obtained from the first derivative of the traces using a single exponential fit. The extent of pH i changes was estimated by averaging the pH i changes measured as a result of Cl Ϫ removal and addition. All results are given as mean Ϯ S.E. of the indicated number of experiments.
was measured with the aid of the Cl Ϫ -sensitive dye MQAE using the procedure described before for 6-methoxy-N-(3-sulfopropyl)quinolinium (SPQ; 18) with minor modifications. SMG cells were suspended in PSA containing 10 mM MQAE and incubated for 20 min at room temperature and 40 min at 0°C before plating on coverslips. About 2 min after plating, unattached cells and external MQAE were washed by starting the perfusion with solution A. MQAE fluorescence was measured at an excitation wavelength of 360 nm with the dichroic mirror and emission cut-off filter set normally used to monitor Fura-2 fluorescence. At the end of each experiment a two-point calibration procedure was performed. To obtain the maximal fluorescence the cells were perfused with high K ϩ , Cl Ϫ -free solution containing 5 M nigericin and 10 M tributyltin cyanide. Incubation in a Cl Ϫ -free solution without ionophores did not result in complete depletion of intracellular Cl Ϫ . To obtain the minimal fluorescence the cells were then exposed to a solution containing 127 mM KSCN. Significant dye leak, in particular after exposure to tributyltin cyanide, precluded a more extensive in vivo calibration. A Stern-Volmer constant of 12.4 M Ϫ1 reported before for rabbit SMG ducts (34) The results of multiple experiments with each cell type and under the different conditions were analyzed using paired or nonpaired Student's t test, as appropriate.

RESULTS
Regulation of AE Activity in T84 Cells-We have described previously the regulation of AE activity by CFTR in cells stably or transiently transfected with CFTR (28). The purpose of the present work was to determine whether such regulation exists in a cell line and in native cells naturally expressing CFTR. The first set of experiments was performed with the human colonic cell line T84, which can serve as a suitable model system in future studies. This cell line has been used in the past in several studies as a model system to characterize natively expressed CFTR (32,33). Ϫ -dependent changes in pH i indicate the expression of relatively modest AE activity in T84 cells. Stimulation of the cells with 5 M forskolin caused a reproducible reduction in pH i . This reduction in pH i was less pronounced than that observed in NIH 3T3 and HEK 293 cells expressing high levels of CFTR. Removal and addition of Cl Ϫ to the incubation medium showed that forskolin increased the rate of AE activity by about 2.2-fold or 0.051 ⌬pH unit/min. As was found in NIH 3T3 and HEK 293 cells expressing CFTR (28), the AE activity stimulated by forskolin was not affected by inhibition of CFTR-mediated Cl Ϫ current with 0.1 mM N-phenylanthranilic acid (DPC in Fig. 1a) or 0.1 mM glibenclamide (Glib in Fig. 1b). On the other hand, the AE activity was nearly abolished by 0.5 mM DIDS (Fig. 1b), as was found in 293 cells expressing modest levels of CFTR (see Fig. 7b of Ref. 28).
The potential regulation or function of CFTR as a HCO 3 Ϫ channel (20,(35)(36)(37) raised the possibility that the rate and extent of HCO 3 Ϫ influx during Cl Ϫ removal are underestimated because of the CFTR-dependent efflux of HCO 3 Ϫ which entered the cells through the anion exchanger. To test this possibility we measured the effect of membrane depolarization on HCO 3 Ϫ fluxes. In Figs. 1c and 2, T84 cells were depolarized by raising the external K ϩ concentration from 5 to 100 mM. This had a minor effect on pH i . Membrane depolarization nearly doubled the initial rate of HCO 3 Ϫ influx observed upon Cl Ϫ removal (Figs. 1c and Fig. 2, first and third bars from left). Stimulation with forskolin of cells bathed in high K ϩ medium increased AE activity by about 1.9-fold (compare third and fourth bars of Fig.  2), similar to the stimulation found in the presence of 5 mM external K ϩ (first and second bars in Fig. 2). However, in the presence of high external K ϩ , forskolin stimulation increased the absolute rate of HCO 3 Ϫ influx by 0.082 ⌬pH/min, which was approximately 1.6-fold higher than that found in normal K ϩ medium (Fig. 2, inset). The simplest interpretation of these results is that membrane depolarization increased the steadystate level of intracellular Cl Ϫ (see below). If the internal Cl Ϫ site of the AE was not saturated with Cl Ϫ present in the cells under normal conditions, the increased [Cl Ϫ ] i caused by membrane depolarization will increase the rate of Cl in Ϫ /HCO 3 out Ϫ exchange. Another contributing factor can be reduction in a potential HCO 3 Ϫ permeability under depolarized conditions. An additional implication of the findings in Fig. 1c is that most of the HCO 3 Ϫ fluxes induced by changes in transcellular Cl Ϫ gradients are caused by the electroneutral AE activity.
AE Activity in the SMG of WT and ⌬F/⌬F Mice-A critical aspect of the regulation of AE activity by CFTR is to determine whether it occurs in native CFTR-expressing cells. We elected to study the relationship between the two proteins in the mouse SMG and pancreatic ducts because of the availability of the ⌬F/⌬F mouse strain. The rat and mouse SMG and pancreatic ducts express a high level of CFTR protein in the luminal membrane (16) and Cl Ϫ /HCO 3 Ϫ exchange activity in the basolateral and luminal membranes (17,18). However, a previous study in the perfused main duct of the mouse SMG suggested very low, if any, Cl Ϫ /HCO 3 Ϫ exchange activity in either membrane of this duct (30). We reevaluated these findings by measuring Cl Ϫ /HCO 3 Ϫ exchange activity in the intralobular duct of the mouse SMG. As illustrated in Fig. 3, a and b, we found high variability in Cl Ϫ /HCO 3 Ϫ exchange activity in isolated SMG ducts. The traces in Fig. 3, a and b, represent the range of high and low AE activity in SMD ducts from WT mice, respectively. Fig. 3c shows AE activity in a SMG duct fragment from a ⌬F/⌬F mouse. The AE activity of ducts from mutant mice was less variable, as reflected in the averaged results (Fig. 3d). In a significant number of experiments it was difficult to estimate accurately the initial rate of HCO 3 Ϫ efflux. Therefore in these experiments we elected to evaluate AE activity from the extent of pH i changes caused by Cl Ϫ removal, which gave more reproducible results and allowed us to include all of the experiments performed in the statistical analysis. Fig. 3d shows that removal of Cl Ϫ increased pH i by about 0.078 Ϯ 0.006 pH unit in ducts from WT animals. The same protocol was used to evaluate AE activity in SMG ducts from ⌬F/⌬F mice. Because of the regulation of AE activity by CFTR demonstrated before (28), we expected to find the same or lower AE activity in the SMG duct of ⌬F/⌬F mice. Surprisingly, AE activity in ducts isolated from SMG of ⌬F/⌬F mice was significantly higher than in ducts isolated from the SMG of WT mice (Fig. 3, c and d).
[Cl Ϫ ] i in SMG Ducts of WT Mice-A potential explanation for the paradoxical findings above is illustrated in the model in Fig. 3. If CFTR was at least partially active in unstimulated cells from WT mice, and resting membrane potential was similar in ducts from WT and ⌬F/⌬F mice, the steady-state [Cl Ϫ ] i is expected to be variable and lower in SMG from WT mice. To test this possibility we measured [Cl Ϫ ] i in SMG ducts with the aid of the Cl Ϫ -sensitive dye MQAE. The results are summarized in Table I Ϫ of about 72 mM H ϩ or HCO 3 Ϫ /pH unit (38), this amounts to 6.3 Ϯ 0.5 mM base equivalents. Hence, the increase in steady-state [Cl Ϫ ] i can account for about 75% (4.8/6.3) of the increased HCO 3 Ϫ influx. The remaining portion can be explained by reduced HCO 3 Ϫ leakage under the depolarized conditions. The possible contribution of a HCO 3 Ϫ leak pathway is consistent with our previous reports of a Ba 2ϩ -sensitive and agonist-regulated H ϩ /HCO 3 Ϫ leak pathway in the luminal membrane of the SMG duct (18,29,38). Not only did membrane depolarization increase [Cl Ϫ ] i , but it also reduced the variability of [Cl Ϫ ] i in unstimulated cells (Table I), which facilitated evaluation of Cl Ϫ /HCO 3 Ϫ exchange activity in cells from WT and ⌬F/⌬F mice. Hence, in most subsequent experiments Cl Ϫ /HCO 3 Ϫ exchange activity was measured under depolarized conditions.
AE Activity under Depolarized Conditions-The effect of membrane depolarization with Ba 2ϩ on Cl Ϫ /HCO 3 Ϫ exchange is illustrated in Fig. 4. In SMG ducts from WT mice, inhibition of K ϩ conductance is expected to depolarize the membrane and reduce Cl Ϫ efflux, increasing and stabilizing steady-state [Cl Ϫ ] i . As a consequence, the maximal Cl Ϫ /HCO 3 Ϫ exchange activity of these cells could be measured. In SMG ducts from ⌬F/⌬F mice membrane depolarization is expected to have smaller, if any, effect on the Cl Ϫ -dependent changes on pH i . Importantly, membrane depolarization minimizes the contribution of all Cl Ϫ channels expressed in these cells (39) to allow a better comparison between the electroneutral Cl Ϫ /HCO 3 Ϫ exchange activity of ducts from WT and ⌬F/⌬F mice. The traces of an individual experiment (Fig. 4, a and b) and the averages obtained from multiple experiments (Fig. 4c) show that this was indeed the case.
In these experiments Ba-Hepes (0.5 M, pH 7.4) was added to HCO 3 Ϫ -buffered solutions in which Cl Ϫ was replaced by gluconate. These solutions were clear for about 30 min, after which precipitates, probably of Ba-gluconate, started to form. At this time old solutions were replaced with fresh Ba 2ϩ -containing, Cl Ϫ -free solutions. Exposing SMG ducts bathed in a HCO 3 Ϫbuffered medium to 5 mM Ba 2ϩ caused a small and reproducible reduction in pH i , the cause of which was not investigated in the present work. Membrane depolarization with Ba 2ϩ increased the apparent AE activity by more than 2-fold in SMG ducts from WT mice. By contrast, Ba 2ϩ had a small, statistically insignificant effect on AE activity of SMG ducts from ⌬F/⌬F mice. Furthermore, in the presence of Ba 2ϩ , the AE activity of SMG duct from WT mice tended to be higher than that of ducts from ⌬F/⌬F mice, although it did not reach statistical significance (p ϭ 0.083). Finally, the Cl Ϫ -dependent changes in pH i were blocked completely by DIDS (Fig. 4, a and b).
Regulation of AE Activity by Forskolin-The effect of forskolin stimulation on AE activity of SMG duct from WT and ⌬F/⌬F mice is shown in Fig. 5. Fold stimulation was determined from the extent of pH i changes caused by Cl Ϫ removal and addition before and after forskolin stimulation of the same duct fragments. Even in the absence of Ba 2ϩ , stimulation of SMG duct from WT mice with forskolin increased AE activity by about 1.7-fold. By contrast, and as expected, forskolin had no effect on the AE activity of ducts isolated from the SMG of ⌬F/⌬F mice. We noticed that multiple removals and additions of Cl Ϫ to the incubation medium resulted in a slightly reduced AE activity in each successive round. Thus, the small, frequently observed  reduction in AE activity in forskolin-stimulated ducts from ⌬F/⌬F mice is probably the result of this artifact. Because in most experiments a control test preceded an experimental test, the effect of forskolin on AE activity in SMG duct from WT mice is probably underestimated. Nevertheless, it was sufficiently large to be highly statistically significant (Fig. 5c).
To determine the actual stimulation of AE by CFTR in SMG duct we measured the effect of forskolin in ducts treated with 5 mM Ba 2ϩ . Fig. 6 shows that forskolin stimulation of SMG ducts from WT mice, but not from ⌬F/⌬F mice, incubated in Ba 2ϩcontaining solutions increased AE activity. In SMG ducts from WT mice, forskolin increased pH i caused by Cl Ϫ removal by 0.06 pH unit above that measured in the same unstimulated ducts, which consists of an approximately 1.3-fold stimulation. However, when pH i changes caused by AE activity are compared in forskolin-stimulated ducts from WT and ⌬F/⌬F mice, pH i changes in WT ducts are higher by about 0.12 pH unit, which is 1.9-fold above that measured in ducts from ⌬F/⌬F mice. This stimulation is comparable to that found in the absence of Ba 2ϩ as illustrated in Fig. 5.
To avoid the possibility of a nonspecific effect of Ba 2ϩ , we tested the effect of high K ϩ on AE activity in SMG ducts from WT mice. Fig. 7 shows that depolarizing the membrane potential with 100 mM K ϩ was as effective as 5 mM Ba 2ϩ in unmasking the maximal AE activity. As expected, the Cl Ϫ -dependent pH i changes were inhibited by DIDS (Fig. 7a). Stimulation of the ducts with forskolin increased the AE activity of these ducts. In these experiments we compared the AE activity measured in the first exposure of all ducts to Cl Ϫ -free medium (Fig.  7, b and c). The forskolin-stimulated increases in pH i in the presence of 100 mM K ϩ or 5 mM Ba 2ϩ were similar.
Membrane-specific Regulation of Cl Ϫ /HCO 3 Ϫ Exchange by CFTR in Pancreatic Ducts-CFTR-expressing cells are likely to express more than one anion exchanger, the housekeeping AE2 in the basolateral membrane and as yet unidentified AE isoform or other exchanger protein involved in transcellular HCO 3 Ϫ transport in the luminal membrane. Indeed, AE activity was measured previously in both membranes of the SMG (18) and pancreatic ducts (17). To extend our findings to another native CFTR-expressing tissue and determine the membrane localization of the AE activity regulated by CFTR, we measured luminal and basolateral AE activity in the microperfused pancreatic ducts of WT and ⌬F/⌬F mice. Fig. 8 summarizes the results of multiple experiments. It was satisfying to find that stimulation with forskolin exclusively increased the activity of the luminal AE without affecting the basolateral AE in pancreatic ducts from WT mice. Furthermore, such regulation was absent in ducts from ⌬F/⌬F animals. In the first protocol we measured AE activity under polarized conditions (Fig. 8a). As we reported before for the rat duct, luminal AE activity was higher than basolateral AE activity in the mouse pancreatic Ϫ -buffered solutions containing 5 mM Ba 2ϩ throughout the experiment. As indicated by the bars, the ducts were exposed to Cl Ϫ -free solutions before and after stimulation with 5 M forskolin. Panel c shows the summary of all experiments performed in terms of mean Ϯ S.E. Note that forskolin-stimulated ducts from WT mice showed higher AE activity than nonstimulated ducts from WT mice and forskolin-stimulated ducts from ⌬F/⌬F mice.
FIG. 7. Effect of high K ؉ on pH i changes in SMG duct from WT mice. SMG ducts from WT mice were incubated in HCO 3 Ϫ -buffered solutions. As indicated by the bars, the ducts were exposed to Cl Ϫ -free solutions in the presence of 5 or 100 mM K ϩ before (panels a and b) or after stimulation with 5 M forskolin (panel c). Panel d illustrates the mean Ϯ S.E. of the indicated number of experiments. duct (compare the rate and extent of pH i changes caused by the Cl Ϫ removal from the bath and lumen in panels a and b before forskolin stimulation). Stimulation with forskolin caused the typical initial acidification, had no effect on basolateral AE activity, and prominently increased luminal AE activity in ducts from WT mice. Forskolin had no effect on luminal or basolateral AE activity of ducts from ⌬F/⌬F mice (data not shown).
To obtain a better estimate of the AE activity in each membrane and under resting and stimulated conditions, we measured separately the basolateral and luminal AE activity under depolarized conditions. The left part of Fig. 8b shows the higher rate (and extent) of pH i changes caused by Cl Ϫ removal and addition to the lumen (and thus the higher AE activity) of pancreatic duct from WT mice. The right part of the trace shows that stimulation with forskolin had no effect on the basolateral AE activity but increased the initial rate (and extent) of pH i changes because of removal of luminal [Cl Ϫ ] out in the same duct. Both activities were inhibited completely by 0.5 mM DIDS. By contrast, forskolin had no effect on basolateral and luminal AE activity of ducts from ⌬F/⌬F mice (Fig. 8c). Hence, we can conclude that CFTR regulates the luminal but not the basolateral AE activity. DISCUSSION In the present work we used the SMG and pancreatic ducts of WT and ⌬F/⌬F mice to study regulation of Cl Ϫ /HCO 3 Ϫ exchange activity by CFTR. Along with our experience with these cells, the SMG and pancreatic ducts offer other advantages as model systems. Both ducts express high levels of CFTR (16) and basolateral and luminal AE activity (17,18). In addition, the SMG offers the availability of several experimental preparations from microperfused ducts (2,18) to isolated single cells (2,16); and most importantly, among all CFTR-expressing tissues, the mechanisms of fluid and electrolyte transport are understood best in the SMG (2).
In agreement with our previous work in the rat SMG duct (18), in the present work we were able to demonstrate AE activity in the mouse SMG duct. The evidence includes a DIDSsensitive, Na out ϩ -independent, electroneutral Cl in Ϫ -dependent HCO 3 Ϫ influx and Cl out Ϫ -dependent HCO 3 Ϫ efflux. By contrast, a work with the main SMG duct concluded low or no AE activity in the mouse duct (30). This discrepancy may be the result of sufficiently high Cl Ϫ and/or HCO 3 Ϫ conductance in the previous study (30) which masked the AE activity.
We had a particular interest in evaluating HCO 3 Ϫ conductance in native CFTR-expressing cells and its possible regulation by CFTR. In a recent work, Ishguro et al. (6) reported the intriguing finding that luminal AE mediated HCO 3 Ϫ transport in resting but not cAMP-stimulated guinea pig pancreatic ducts. In addition, CFTR-dependent HCO 3 Ϫ conductance was reported in pancreatic duct cells (35) and in an airway epithelial cell line expressing CFTR (20). This would imply that stimulation of CFTR should dramatically increase the HCO 3 Ϫ conductance of the luminal membrane. In the present work we were unable to obtain evidence in support of such a conclusion. Our results in the mouse are in agreement with the finding that HCO 3 Ϫ conductance is at least 6-fold lower than that of Cl Ϫ (20, 35) and the absence of HCO 3 Ϫ conductance in sweat duct (40). Hence, in the presence of physiological Cl Ϫ and HCO 3 Ϫ gradients it is not likely that CFTR mediates or modulates the HCO 3 Ϫ conductance in the luminal membrane of the rat and mouse secretory epithelia. Accordingly, the electrogenic component of HCO 3 Ϫ transport in T84 cells (Fig. 1) and SMG ducts (Figs. 4 and 7) was rather small and not affected by inhibitors of CFTR Cl Ϫ channel activity ( Fig. 1 and not shown). Furthermore, stable and transient transfection of CFTR in NIH 3T3 and HEK 293 cells, respectively, did not increase HCO 3 Ϫ conductance even when the cells were stimulated with forskolin (compare the effect of high K ϩ in Fig. 10 of Ref. 28 and Figs. 1c and 7 of this paper).
Low luminal HCO 3 Ϫ conductance in the mouse SMG and pancreatic duct cannot be extended to other species and tissues. This is because, most likely, different mechanisms mediate HCO 3 Ϫ transport in different tissues and species. For example, the electrogenic component of HCO 3 Ϫ transport is much higher in the duodenum than in the colon (3) or the SMG (2). The pancreatic juice of the guinea pig contains a much higher HCO 3 Ϫ concentration than that of the rat (1, 6). However, despite this variability we can safely conclude that in tissues secreting HCO 3 Ϫ in a mechanism similar to that of the mouse SMG or the pancreatic ducts (a) HCO 3 Ϫ conductance is of secondary importance in HCO 3 Ϫ secretion, and (b) regulation of luminal AE activity by CFTR may be the major mechanism by which CFTR regulates HCO 3 Ϫ secretion. In this respect it would be of particular significance to test the effect of CFTR stimulation on luminal AE activity and HCO 3 Ϫ conductance in guinea pig SMG and pancreatic ducts.
The major finding of the present work was extending the finding of regulation of AE activity by CFTR to native CFTRexpressing tissues such as the SMG and pancreas. Even though the increase in AE activity after forskolin stimulation of SMG duct was not as prominent as that observed in the transfected cell lines (28), it could be clearly demonstrated even when the cells were not depolarized (Fig. 5). It is important to note that the extent of stimulation of AE activity by CFTR in SMG ducts may be significantly underestimated. Studies on the rat SMG duct showed the presence of AE activity in both the BLM and LM (18). Expression of CFTR in the LM of SMG ducts (16) indicates that CFTR should stimulate AE activity present in the LM but not in the BLM of SMG ducts. Hence, AE activity in the BLM, although lower than that in the LM (18), increases the background against which the stimulation of AE activity by CFTR is evaluated. This reasoning is reinforced by the finding Ϫ -buffered solution B. Where indicated by the bars, the luminal and basolateral AE activities were measured by perfusing the bath and the lumen with a HCO 3 Ϫ -buffered, Cl Ϫ -free solution C (panel a). In panels b and c the ducts were perfused with the high K ϩ , depolarizing solution D before exposure to high K ϩ , Cl Ϫ -free solution E. In all experiments, as indicated by the bars, the ducts were stimulated with 5 M forskolin. Forskolin was included in the luminal solution. Panel d summarizes the results of multiple experiments with ducts from WT and ⌬F/⌬F mice. that CFTR stimulates the luminal but not the basolateral AE activity of the pancreatic duct. Notably, in the SMG duct, when the AE activity of both membranes contributed to the measurement, stimulation of CFTR increased AE activity by about 43% (Fig. 7). Under the same depolarized conditions CFTR increased luminal AE activity of the pancreatic duct by about 96% (Fig. 8).
The exclusive expression of CFTR in the luminal membrane of the SMG duct and other CFTR-expressing cells indicates that also in these tissues CFTR regulates the luminal AE. To date the protein responsible for the luminal AE activity of CFTR-expressing cells has not been identified. Good immunocytochemical evidence indicates that the housekeeping AE2 is expressed exclusively in the basolateral membrane of SMG duct and acinar cells (19). This excludes AE2 as the isoform regulated by CFTR. It has been suggested that a variant of AE1 is expressed alternatively in the basolateral or luminal membrane of intercalated cells of collecting duct based on the metabolic state of the animal (41). However, preliminary reverse transcriptase-polymerase chain reaction analysis of the AE isoforms expressed in the cell lines used in the present paper and in Ref. 28 indicates that these cells express only AE2 and AE3 (not shown). Hence, it is possible that CFTR regulates AE3 and that AE3 is the isoform expressed in the luminal membrane of the SMG duct. Another alternative is that a protein other than the known AE isoforms mediates the luminal AE activity. Using expression systems we are attempting to test the effect of CFTR on the activity of each AE isoform.
Our findings provide new insight into the mechanisms of fluid and electrolyte secretion by CFTR-expressing cells and into the pathophysiology of cystic fibrosis. An important function of CFTR-expressing cells is the secretion of HCO 3 Ϫ (1-3, 7,9,12,[25][26][27]. HCO 3 Ϫ is a chaotropic anion that is commonly used to dissolve, and thus, strip membranes of peripheral proteins. Moreover, HCO 3 Ϫ regulates the pH of biological fluids. Fluids secreted by CFTR-expressing cells are rich in proteins, in particular mucins (1-3, 7, 25-27). Mucin solubility may be increased in high HCO 3 Ϫ , high pH fluids (42). In the special case of the exocrine pancreas, the high pH is also needed to prevent premature activation of harmful digestive enzymes (1). Impaired HCO 3 Ϫ secretion may cause precipitation of mucins and thus play a major role in obstruction of almost all ductal systems in cystic fibrosis (43). Every model of luminal HCO 3 Ϫ secretion in CFTR-expressing cells suggests that AE activity mediates the electroneutral portion of HCO 3 Ϫ secretion (1-3, 7). By demonstrating regulation of AE activity by an activated CFTR, our findings indicate that CFTR directly regulates the entire process of HCO 3 Ϫ secretion. Furthermore, the suggested regulation of luminal Na ϩ channel by CFTR (44) allows the regulation of Na ϩ absorption by CFTR. Several CFTR-expressing cells, such as the SMG duct, while absorbing Na ϩ secrete isotonic K ϩ (2). The luminal pathway responsible for K ϩ efflux is believed to be an inward rectifier K ϩ channel such as ROMK II in the kidney (3). CFTR has also been implicated as a regulator of ROMK II (45). Hence, CFTR must be viewed as a global regulator of epithelial fluid and electrolyte transport through its ability to function as a Cl Ϫ channel (46) and regulate Cl Ϫ absorption, regulate the epithelial Na ϩ channel to regulate Na ϩ absorption, regulate the luminal K ϩ channel to regulate K ϩ secretion, and regulate AE activity to regulate HCO 3 Ϫ secretion. The global role of CFTR indicates that its action is not likely to be replaced solely by activation of luminal Cl Ϫ channels through stimulation of selective P 2 purinoceptors. Again, the finding that cholinergic stimulation of intestinal HCO 3 Ϫ secre-tion is abolished in the intestines of CFTR Ϫ/Ϫ mice (24) supports the global role of CFTR in epithelial fluid and electrolyte secretion. Cholinergic stimulation is expected to activate the luminal Ca 2ϩ -activated Cl Ϫ channel found in many epithelial cells (47). Activation of this Cl Ϫ channel was not sufficient to cause normal HCO 3 Ϫ secretion in cholinergically stimulated intestinal epithelium (24). Hence, although stimulation of luminal purinoceptors may induce Cl Ϫ secretion, it is not likely to be sufficient in itself in alleviating the symptoms of cystic fibrosis. Thus, improved expression of CFTR in the luminal membrane seems to be the best option in this endeavor.