Regulation of Cl−/ HCO3 −Exchange by Cystic Fibrosis Transmembrane Conductance Regulator Expressed in NIH 3T3 and HEK 293 Cells*

A central function of cystic fibrosis transmembrane conductance regulator (CFTR)-expressing tissues is the secretion of fluid containing 100–140 mmHCO3 −. High levels of HCO3 − maintain secreted proteins such as mucins (all tissues) and digestive enzymes (pancreas) in a soluble and/or inactive state. HCO3 −secretion is impaired in CF in all CFTR-expressing, HCO3 −-secreting tissues examined. The mechanism responsible for this critical problem in CF is unknown. Since a major component of HCO3 − secretion in CFTR-expressing cells is mediated by the action of a Cl−/HCO3 −exchanger (AE), in the present work we examined the regulation of AE activity by CFTR. In NIH 3T3 cells stably transfected with wild type CFTR and in HEK 293 cells expressing WT and several mutant CFTR, activation of CFTR by cAMP stimulated AE activity. Pharmacological and mutagenesis studies indicated that expression of CFTR in the plasma membrane, but not the Cl− conductive function of CFTR was required for activation of AE. Furthermore, mutations in NBD2 altered regulation of AE activity by CFTR independent of their effect on Cl− channel activity. At very high expression levels CFTR modified the sensitivity of AE to 4,4′-diisothiocyanatostilbene-2,2′-disulfonate. The novel finding of regulation of Cl−/HCO3 − exchange by CFTR reported here may have important physiological implications and explain, at least in part, the impaired HCO3 −secretion in CF.

In most CFTR-expressing tissues, HCO 3 Ϫ secretion has electrogenic and electroneutral components (4 -6). The electrogenic component is assumed to be mediated by an unknown HCO 3 Ϫ channel or due to HCO 3 Ϫ transport through CFTR itself (11,12). The electroneutral component is assumed to be mediated by a Cl Ϫ /HCO 3 Ϫ exchange activity. However, direct evidence for a Cl Ϫ /HCO 3 Ϫ exchange activity in the luminal membrane is limited to the perfused pancreatic (13) and submandibular ducts (14).
Most models of HCO 3 Ϫ secretion assume that CFTR and the luminal Cl Ϫ /HCO 3 Ϫ anion exchanger (AE) are indirectly coupled. In these models Cl Ϫ absorbed by the AE across the luminal membrane is secreted into the lumen by CFTR to support further HCO 3 Ϫ secretion (4,15). However, if Cl Ϫ /HCO 3 Ϫ exchange is unaltered in CF, such a mechanism cannot adequately explain the concomitant acidity of the secreted fluid and the impaired Cl Ϫ absorption observed in CF (16 -18). If Cl Ϫ /HCO 3 Ϫ exchange is responsible for the bulk of Cl Ϫ absorption and HCO 3 Ϫ secretion and CFTR function is required only for return of Cl Ϫ to the lumen, then in CF Cl Ϫ absorption should be normal (normal Cl Ϫ /HCO 3 Ϫ exchange) and the secreted fluid should be acidic due to the limited supply of luminal Cl Ϫ . This is not the case (16 -18). Alternatively, if Cl Ϫ reabsorption is singularly impaired in CF, the model predicts that the high Cl Ϫ concentration in the luminal fluid should increase HCO 3 Ϫ secretion by AE to produce an alkaline fluid with high Cl Ϫ concentration. Again, this is not observed.
The observation that the luminal fluid is acidic with high Cl Ϫ concentration in CF (16 -18) suggests that CFTR regulates HCO 3 Ϫ secretion in CFTR-expressing tissues. CFTR could regulate the electrogenic, electroneutral, or both components of HCO 3 Ϫ secretion. In the present work, we explored the existence of these regulatory mechanisms in cells stably or transiently expressing wild type (WT) or several mutated CFTR constructs. We report that a cAMP-activated CFTR regulates Cl Ϫ /HCO 3 Ϫ exchange activity in several experimental systems. Expression of CFTR in the plasma membrane was required for regulation of Cl Ϫ /HCO 3 Ϫ exchange, as expression of several folding mutants, including ⌬F508, had no effect on Cl Ϫ /HCO 3 Ϫ exchange activity. Surprisingly, the Cl Ϫ conductive function of CFTR was not required for activation of Cl Ϫ /HCO 3 Ϫ exchange. Furthermore, mutations in NBD2 altered regulation of AE activity by CFTR independent of their effect on Cl Ϫ channel activity. At very high expression levels, CFTR modified the sensitivity of Cl Ϫ /HCO 3 Ϫ exchange to DIDS. The novel finding of regulation of Cl Ϫ /HCO 3 Ϫ exchange by CFTR reported here * This work was supported 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 1 The abbreviations used are: CFTR, cystic fibrosis transmembrane conductance regulator; CF, cystic fibrosis; AE, Cl Ϫ /HCO 3 Ϫ exchanger; WT, wild type; NBD, nucleotide binding domain; DIDS, 4,4Ј-diisothiocyanatostilbene-2,2Ј-disulfonate; GFP, green fluorescent protein; BCECF-AM, 2Ј7Ј-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester; DPC, N-phenylanthranilic acid; NMDG, N-methyl-D-glucamine; Ab, antibody; DMEM, Dulbecco's modified Eagle's medium; ER, endoplasmic reticulum; ENaC, epithelial Na ϩ channel. may have important physiological implications and explain, at least in part, the impaired HCO 3 Ϫ secretion in CF.

EXPERIMENTAL PROCEDURES
Culture of NIH 3T3 Cells-Mock-transfected NIH 3T3 cells or NIH 3T3 cells stably transfected with WT or ⌬F508 CFTR were kindly provided by Dr. Michael J. Welch (University of Iowa, Iowa City, IA). The cells were maintained in Dulbecco's modified Eagle's medium containing 10 mM glucose (DMEM-HG) and 10% fetal calf serum and plated on a sterile 22 ϫ 40-mm coverslip at a density of 2.5 ϫ 10 5 cells/cm 2 for intracellular pH (pH i ) measurements.
Site-directed Mutagenesis-The pCMVNot6.2 plasmids containing human WT or ⌬F508 CFTR cDNA were a generous gift from Dr. Johanna Rommens (Hospital for Sick Children, Toronto, Canada). Oligonucleotide-directed mutagenesis using the GeneEditor mutagenesis kit (Promega, Madison, WI) was performed to generate the mutant CFTR in the expression vector pCMVNot6.2. Briefly, mutants were selected based upon the incorporation of a second-site mutation in ␤-lactamase, which alters its substrate specificity allowing resistance of transformed bacteria to cefotaxime and ceftriaxone in addition to ampicillin. Incorporation of the mutation was verified by DNA sequencing. The mutagenesis primers were as follows: P205S primer, 5Ј-CGT GTG GAT CGC TTC TTT GCA AGT GGC-3Ј; W846term, 5Ј-GAG CAT ACC AGC AGT GAC TAC ATA GAA CAC ATA CCT TCG ATA TAT TAC-3Ј;  G1247D/G1249E, 5Ј-GTG GGC CTC TTG GGA AGA ACT GAT TCA  GAG AAG AGT ACT TTG TTA TCA GC-3Ј; K1250M, 5Ј-CTT GGG AAG  AAC TGG ATC AGG GAT GAG TAC TTT GTT ATC AGC-3Ј; D1370N,  5Ј-GTA AGG CGA AGA TCT TGC TGC TTA ATG AAC CCA GTG CTC  ATT TGG ATC-3Ј. Transfection-quality plasmid DNA was prepared using reagents supplied by Qiagen (Valencia, CA).
Expression of WT and Mutant CFTR in HEK 293 Cells-HEK 293 cells were maintained in DMEM-HG supplemented with 10% fetal calf serum, and plated on coverslips. On the following day, WT or mutant CFTR plasmids and green fluorescent protein (GFP)-expressing plasmids (Life Technologies, Inc.) were transfected into 293 cells using the Fugene mammalian transfection kit (Boehringer Mannheim) according to instructions provided by the manufacturer. Briefly, the mixture of plasmids and Fugene solution (pCMVNot6.2, 1.5 g; pCMVGFP, 1.5 g; Fugene, 12 l) was incubated in 100 l of DMEM for 30 min before addition to the culture media. The cells were used for immunocytochemistry or pH i measurements 48 -72 h after transfection.
Immunocytochemistry-HEK 293 cells transfected with expressing vectors were stained with a rat polyclonal anti-C-terminal CFTR antibody (Ab) R3194 (19) and/or mouse monoclonal anti-rat Grp78 (BiP) antibody (StressGen Biotechnologies, Victoria, BC, Canada) to determine their expression patterns using a published procedure (20). For double-labeling, primary and secondary incubations were repeated with antibodies against the second protein of interest. Images were obtained using a Bio-Rad MRC 1024 confocal microscope.
Intracellular pH Measurements-The coverslips with cells attached to them were washed once with a Hepes-buffered solution and assembled to form the bottom of perifusion chamber. The Hepes-buffered solution contained (in mM) 140 NaCl, 5 KCl, 1 MgCl 2 , 1 CaCl 2 , 10 glucose, 10 Hepes (pH 7.4 with NaOH). In the case of 293 cells, the level of transfection was estimated from GFP fluorescence. High GFP-expressing cells were identified by viewing GFP fluorescence at excitation wavelength of 475 nm. GFP fluorescence was recorded and used to compare CFTR expression in different experiments. Subsequently, cells were loaded with BCECF by a 10-min incubation at room temperature in Hepes-buffered solution containing 2.5 M BCECF-AM. BCECF fluorescence was at least 10-fold higher than the original GFP fluorescence. After BCECF loading the cells were perfused with a HCO 3 Ϫbuffered solution and pH i was measured by photon counting using the recording setup (PTI Delta Ram, Brunswick, NJ) and the conditions described previously (21). The HCO 3 Ϫ -buffered solution contained (in mM) 120 NaCl, 5 KCl, 1 MgCl 2 , 1 CaCl 2 , 10 glucose, 5 Hepes, 25 NaHCO 3 (pH 7.4 with NaOH) and was continuously gassed with 95% O 2 and 5% CO 2 . Cl Ϫ free solutions were prepared by replacing Cl Ϫ with gluconate. When desired 5 M forskolin was added to the perfusate from a stock solution of 10 mM in Me 2 SO. BCECF fluorescence was recorded at excitation wavelengths of 490 and 440 nm at a resolution of 2/s. The 490/440 ratios were calibrated intracellularly by perfusing the cells with solutions containing 145 mM KCl, 10 mM Hepes, 5 M nigericin with pH adjusted to 6.2-7.6, as described previously (13). The results of multiple experiments were analyzed using paired or non-paired Student's t-test as appropriate.
Patch Clamp-Cl Ϫ current was recorded using the whole cell config-uration of the patch-clamp technique (22) as described before (23). NIH 3T3 cells were released from culture dishes by a 30-s treatment with trypsin-EDTA, washed twice with DMEM, and placed in a perfusion chamber. Cl Ϫ current was isolated by using Cl Ϫ as the only permeant ion in the pipette and bath solutions. In all experiments the bath solution contained (in mM) 140 N-methyl-D-glucamine chloride (NMDG-Cl), 1 MgCl 2 , 10 glucose and 10 Hepes (pH 7.4 with Tris), and the pipette solution contained 140 NMDG-Cl, 5 EGTA, 5 Tris-ATP, 5 MgCl 2 and 10 Hepes (pH 7.2 with Tris). All recordings were made at room temperature. Seals of 5-8 gigohms were obtained on the cell surface prior to establishing the whole-cell configuration. Macroscopic currents were recorded using the Axopatch-1B patch clamp amplifier (Axon Instruments). Results were collected at 5 kHz after filtering at 2 kHz. The membrane potential was held at -40 mV to record the inward current.

AE in Cells Stably
Transfected with CFTR-The first set of experiments to study regulation of AE activity by CFTR was performed in NIH 3T3 cells stably transfected and expressing high levels of CFTR protein. This particular model system has been used to extensively characterize the properties of CFTR Cl Ϫ channel activity (17). Mock-transfected cells of the same parental line were used as controls. Significantly, results identical to mock-transfected cells were obtained in cells stably transfected with ⌬F508 CFTR (data not shown). A standard protocol of removal and addition of Cl Ϫ to the incubation medium buffered with HCO 3 Ϫ was used to follow Cl Ϫ /HCO 3 Ϫ exchange activity. All the changes in pH i reported here were dependent on the presence of HCO 3 Ϫ in the incubation media (data not shown). Fig. 1 illustrates the basic observation that CFTR-expressing cells exhibited a forskolin-dependent activation of the AE. Fig. 2 summarizes the results of 5-17 experiments under each condition. Removal of Cl Ϫ from the incubation medium of mock-transfected cells resulted in a slow and modest increase in pH i , which was completely reversed on addition of Cl Ϫ to the medium. Stimulation of control cells with 5 M forskolin had no effect on basal level of pH i or the pH i changes observed upon removal and readdition of Cl Ϫ . Finally, treating the cells with 0.5 mM DIDS, a blocker of Cl Ϫ /HCO 3 Ϫ exchange activity (24), nearly abolished pH i changes resulting from changes in transcellular Cl Ϫ concentration. These properties are commonly used to demonstrate Cl Ϫ /HCO 3 Ϫ exchange activity in cells (13,14,24). CFTR-expressing cells showed marginal statistical difference in Cl Ϫ /HCO 3 Ϫ exchange activity under resting conditions when compared with mock-transfected cells (p ϭ 0.11), or cells stably transfected with ⌬F508 CFTR (data not shown). Interestingly, Cl Ϫ /HCO 3 Ϫ exchange activity observed in resting cells expressing CFTR was inhibited by DIDS to the same extent as that measured in control cells (Fig. 2).
Stimulation of CFTR-expressing 3T3 cells with forskolin caused a time-dependent intracellular acidification that was complete after 3 min of incubation at 37°C. This acidification was observed only in cells expressing CFTR in all experiments tested (n ϭ 17) and was not inhibited by DIDS (n ϭ 9). Furthermore, removal of Cl Ϫ from the incubation medium of forskolinstimulated, CFTR-expressing cells caused a rapid and a large increase in pH i that was reversed upon readdition of Cl Ϫ to the medium (Fig. 1b). Fig. 2 shows that after forskolin stimulation the rate of pH i change due to changes in transcellular Cl Ϫ gradient in CFTR-expressing cells is 8-fold faster than that before forskolin stimulation in the same cells, or before and after forskolin stimulation in control cells. Thus, the increased rate of pH i changes required both expression of CFTR and activation of the protein by cAMP-dependent mechanisms. It is well established that CFTR-mediated Cl Ϫ channel activity is regulated by a cAMPdependent phosphorylation (2,7,8).
The finding that after forskolin stimulation, the pH i changes due to changes in transcellular Cl Ϫ gradient are resistant to inhibition by DIDS (Figs. 1b and 2) was unexpected. Since the same NIH 3T3 cell line was used to suggest that CFTR may function as a Cl Ϫ and a HCO 3 Ϫ -permeable channel (25), we considered the possibility that the pH i changes illustrated in Figs. 1 and 2 are due to CFTR functioning as a HCO 3 Ϫ channel. Several lines of evidence indicate that this is not the case. In contrast to Poulsen et al. (25), in more than 10 experiments, we did not see any effect of HCO 3 Ϫ addition on pH i of forskolinstimulated acidified NIH 3T3 cells (using the protocol of Fig. 1b in Ref. 25). Furthermore, depolarization of the plasma membrane with 5 mM Ba 2ϩ (n ϭ 5; data not shown), 100 mM K ϩ (see Fig. 10) or 125 mM K ϩ (data not shown) had no effect on the changes in pH i observed on removal and addition of Cl Ϫ as would be expected is CFTR was functioning as a HCO 3 Ϫ channel. Additional evidence that CFTR conductance was not responsible for the pH i changes in Figs. 1 and 2 is provided by testing the effect of inhibitors of CFTR Cl Ϫ channel activity. Fig. 3 shows the effect of 100 M DPC and 100 M glibenclamide on Cl Ϫ channel activity of CFTR-expressing cells. At 100 M these blockers inhibited CFTR-dependent Cl Ϫ current by at least 90% (Fig. 3c). Notably, these blockers had no effect on the ability of CFTR to stimulate pH i changes upon Cl Ϫ removal or addition in a forskolin-dependent manner (Fig. 4).
The lack of effect imparted by changes in membrane potential and inhibitors of CFTR Cl Ϫ channel activity strongly suggested that the pH i changes observed on removal and addition of Cl Ϫ are not mediated by an electrogenic pathway. Rather, it appears that expression and stimulation of CFTR by cAMP activated an electroneutral HCO 3 Ϫ transport mechanism. If this pathway transports Cl Ϫ in exchange for HCO 3 Ϫ , then the pH i changes should be a function of intracellular Cl Ϫ content. Fig.  5 shows the protocol used to test this prediction. The cells were first treated with DIDS to prevent the initial changes in pH i

FIG. 1. Effect of forskolin on AE activity in NIH 3T3 cells. NIH 3T3 cells attached to glass coverslips were loaded with BCECF and perfused with HCO 3
Ϫbuffered solutions. As indicated by the bars, the cells were perfused with a Cl Ϫfree solution before and after stimulation with 5 M forskolin. In panel a, the cells were transfected with empty vectors. Identical results were obtained in cells stably transfected with ⌬F508 CFTR (data not shown). In panel b, the cells were stably transfected with WT CFTR.
In both experiments the cells were treated with 0.5 mM DIDS after the stimulation with forskolin. Upper deflection in all traces indicates increase in pH i . Fig. 1 were used to measure the rate and extent of pH i changes due to removal and addition of Cl Ϫ to mock transfected or WT CFTRexpressing NIH 3T3 cells before (Con, control) and after stimulation with 5 M forskolin and before and after treatment with 0.5 mM DIDS. The results of 5-17 experiments were summarized to calculate the mean Ϯ S.E. After forskolin stimulation all pH i changes in WT CFTR-expressing cells were much higher than those in mock transfected cells. Before forskolin stimulation the pH i changes in cells expressing WT CFTR trended to be higher than those in control cells, although the differences did not reach statistical significance (p ϭ 0.11). due to Cl Ϫ removal. Then the cells were incubated in a HCO 3 Ϫbuffered, Cl Ϫ -free medium for 1 min (Fig. 5a), 60 min (Fig. 5b), or various times between 5 and 30 min (data not shown) to deplete intracellular Cl Ϫ . The cells were then stimulated with forskolin to activate CFTR and, thus, Cl Ϫ /HCO 3 Ϫ exchange. Progressive depletion of intracellular Cl Ϫ resulted in a graded inhibition of forskolin-activated pH i increase (Fig. 5). Readdition of Cl Ϫ to the incubation medium resulted in a pronounce acidification, as expected from HCO 3

FIG. 2. AE activity in CFTR-stimulated cells. The protocols of
Ϫ i /Cl Ϫ o exchange. Removal and readdition of Cl Ϫ in these cells showed the expected changes in pH i (Fig. 5). In additional experiments we incubated the cells in a HEPES-buffered, Cl Ϫ -free medium for 30 -60 min to deplete intracellular Cl Ϫ . Such incubations were as effective in inhibiting the effect of forskolin on pH i in the presence of HCO 3 Ϫ as the incubation in HCO 3 Ϫ -buffered, Cl Ϫ -free medium shown in Fig. 5b (data not shown).
Expression and Localization of WT CFTR and CFTR Mutants in HEK 293 Cells-All the experimental protocols used to identify the HCO 3 Ϫ transporter activated by forskolin in CFTR-expressing 3T3 cells except for the lack of inhibition by DIDS point to a Cl Ϫ /HCO 3 Ϫ exchanger. These include (a) requirement for a HCO 3 Ϫ gradient, (b) requirement for a Cl Ϫ gradient, (c) independence from Na ϩ o , (see below), (d) electroneutrality, and (e) insensitivity to Cl Ϫ channel blockers. A possible explanation for these observations is that high level expression of CFTR in 3T3 cells activated the anion exchanger and modified its sensitivity to DIDS. To test this hypothesis and provide additional evidence for regulation of anion exchange by CFTR, we examined the effect of transient expression of WT and mutant CFTR on anion exchange in HEK 293 cells. To identify the transfected cells and evaluate the extent of protein expression, the cells were co-transfected with GFP and the various CFTR plasmids.
Many CFTR mutants, including some used in the present work, are known folding mutants (26,27) that are rapidly degraded by the ubiquitin-dependent proteasome system (28) before substantial amount of the protein reaches the plasma membrane. Therefore, we first determined the expression and localization of the CFTR mutants used in the present work. To

FIG. 3. Stimulation of Cl ؊ current in NIH 3T3 cells expressing WT CFTR.
The whole cell configuration of the patchclamp technique was used to measure Cl Ϫ current in cells internally perfused through the patch pipette and bathed in Na ϩ -and K ϩ -free, NMDG-Cl-containing medium. Cl Ϫ current was stimulated by exposing cells to 5 M forskolin. When the current reached maximal value, the cells were perfused with a solution containing 100 M DPC (a) or 100 M glibenclamide (b). Panel c summarizes the results from the indicated number of experiments to give the mean Ϯ S.E. of the current as percentage of the maximal current measured before addition of the drugs.

FIG. 4. Effect of CFTR Cl ؊ channel inhibition on AE activity. NIH 3T3 cells stably transfected with WT CFTR and incubated in HCO 3
Ϫ -buffered media were stimulated with 5 M forskolin and transiently exposed to Cl Ϫ -free medium before and after incubation with 100 M DPC (a) or 100 M glibenclamide (b). Similar results were obtained in at least three experiments under each experimental condition.
distinguish plasma membrane and ER localized CFTR, CFTR localization was compared with that of the ER-resident chaperone BiP (29). Fig. 6 (a and b) shows that WT CFTR is expressed in the plasma membrane and CFTR expression correlated with expression of GFP. In numerous cells examined, expression of WT CFTR and all mutants correlated very well with expression of GFP. Fig. 6 (e and f) shows the correlation between expression of GFP and ⌬F508 CFTR. In agreement with previous reports (30), it can be seen that ⌬F508 CFTR is retained in the ER and excluded from the plasma membrane. Fig. 6, c and d shows the localization of WT CFTR relative to that of BiP, and g and h show the localization of P205S CFTR. It is clear that WT CFTR was present in the ER and plasma membrane whereas P205S CFTR localized in the ER. Fig. 6 (i-k) shows the plasma membrane localization of K1250M CFTR, D1370N CFTR, and the double mutant G1247D/ G1249E CFTR, respectively. Another mutant used in the present work is W846term CFTR, which includes amino acids 1-845 of WT CFTR. This construct could not be localized with the C-terminal specific antibodies used to detect the other constructs. However, a similar C-terminal truncation at Asp-836 has been shown to be expressed in plasma membrane of HeLa cells as a functioning Cl Ϫ channel (31).
WT CFTR and Anion Exchange Activity in HEK 293 Cells-The WT CFTR and GFP constructs were used to determine whether expression of CFTR in 293 cells affected AE activity as observed in stably transfected NIH 3T3 cells (Figs. 1-5). Fig. 7 shows representative traces, and Figs. 8 and 9 summarize the results of multiple experiments. In these experiments GFP fluorescence was measured prior to loading with BCECF. Based on intensity of GFP fluorescence, the transfected cultures were divided into two groups: those that express low to moderate levels and those that express high levels of the transgene.
Expression of CFTR in 293 cells was sufficient to increase the DIDS-inhibitable pH i increase upon removal of external Cl Ϫ and prior to stimulation with forskolin. The increase in AE activity in unstimulated cells was statistically significant only at high levels of WT CFTR expression ( Fig. 8; p ϭ 0.018). Stimulation of cells expressing moderate or high levels of WT CFTR with forskolin caused an initial acidification, as was observed in NIH 3T3 cells (Fig. 1). Stimulation with forskolin FIG. 6. Localization of WT CFTR and mutant CFTR in HEK 293 cells. HEK 293 cells were transfected with plasmids carrying the indicated CFTR-expressing genes and cotransfected with plasmids carrying the gene for GFP. About 48 h after transfection, the cells were fixed and stained with Ab specific for CFTR and the ER-resident chaperone BiP. The transfected cells were identified in the confocal microscope by measuring GFP fluorescence. Examples for such fluorescence are given in panels b and f. In all experiments examined, there was excellent correlation between expression of GFP and WT CFTR or any of the mutant CFTR. The cellular distribution of CFTR in GFP-expressing cells was identified by a rhodamine staining. When the correlation between BiP and CFTR mutants was studied, BiP was detected with Ab specific to BiP, which were stained with fluorescein-coupled secondary Ab. The Ab used are listed below each image . Panels a and b, c and d, e and f, and g and h are from the same cells. Please note the expression of WT CFTR in the plasma membrane. Panels e and g show the predominant ER localization of ⌬F508 and P205S CFTR, respectively. Panels i, j, and k show the expression of the indicated CFTR mutants in the plasma membrane. Localization similar to that shown in each panel was seen in virtually every cell expressing the respective construct. (Original magnification in panels c-f, ϫ400; original magnification in all other panels, ϫ600.)

FIG. 5. Dependence of pH i changes on intracellular Cl ؊ content. NIH 3T3 cells stably transfected with WT CFTR and incubated in HCO 3
Ϫ -buffered media were treated with 0.5 mM DIDS before incubation in Cl Ϫ -free medium (a and b). After 1 min (a) or 1 h (b) of incubation in Cl Ϫ -free medium, the cells were stimulated with 5 M forskolin while still in Cl Ϫ -free medium. Approximately 3 min after forskolin stimulation, the cells were incubated in Cl Ϫ -containing medium, which caused a rapid reduction in pH i . Subsequently the cells were subjected to another round of incubation in Cl Ϫ -free and Cl Ϫ -containing media while still incubated with 0.5 mM DIDS and stimulated with 5 M forskolin. Similar results were observed in at least three experiments under each experimental condition.
dramatically increased AE activity and the increased activity (n ϭ 16) correlated with the extent of WT CFTR expression (Fig. 8). Figs. 7d and 9 show that inhibitors of CFTR Cl Ϫ current, DPC and glibenclamide, had no measurable effect on AE activity after stimulation with forskolin. Again, these results are similar to those found in NIH 3T3 cells stably expressing WT CFTR (Fig. 4).
Of all known HCO 3 Ϫ transport pathways (including HCO 3 Ϫ conductance and the Na ϩ -HCO 3 Ϫ cotransporters), only the AE is electroneutral and its activity is independent of Na ϩ (24). Hence, as a further test for the HCO 3 Ϫ transport activity stimulated by CFTR we determined the effect of membrane potential and external Na ϩ on this activity using two experimental conditions. In the first set of experiments, cells incubated in HCO 3 Ϫ -buffered solutions in which all NaCl was replaced with KCl and all NaHCO 3 was replaced with choline-HCO 3 Ϫ . In these solutions, as needed, Cl Ϫ was replaced with gluconate using K ϩ -gluconate. To prevent intracellular acidification due to incubation of the cells in these Na ϩ -free solutions, all solutions also contained 5 M Na ϩ /H ϩ exchange inhibitor, ethylisopropyl-amiloride. Under these conditions removal of external Na ϩ still caused substantial intracellular acidification, probably due to the activity of a Na ϩ -HCO 3 Ϫ cotransporter (data not shown). However, the effect of removal and addition of Cl Ϫ was identical to those illustrated in Fig. 10 using the second experimental protocol. In these experiments external Na ϩ was reduced from 140 to 40 mM, which removed the need to include ethyl-isopropyl-amiloride in the incubation medium and almost eliminated the initial acidification on reduction in external Na ϩ . The membrane potential was strongly depolarized by increasing external K ϩ from 5 to 100 mM. Fig. 10 shows that membrane depolarization had no effect on the pH i changes observed on removal and addition of Cl Ϫ in NIH 3T3 (Fig. 10a) or HEK 293 cells (Fig. 10b) expressing high levels of CFTR.
The complete (Fig. 10a) or partial (Fig. 10b) resistance of the Cl Ϫ -dependent pH i changes to 0.5 mM DIDS was preserved under high K ϩ conditions. However, a detailed examination of the result in 293 cells show that the sensitivity of AE activity to inhibition by DIDS was a function of the level of WT CFTR expression. At moderate expression levels of WT CFTR, DIDS nearly abolished AE activity. However, at high expression levels of WT CFTR DIDS inhibited only about 60% of AE activity (Figs. 7, b and c, and 9). This may account, at least in part, for the resistance of CFTR-stimulated AE activity in NIH 3T3 to DIDS (Figs. 1 and 2). In conclusion, we believe that the combined results in NIH 3T3 and HEK 293 cells provide strong evidence for regulation of AE activity by CFTR. It is important to reiterate that such regulation required the activation of CFTR by cAMP but did not require Cl Ϫ transport by CFTR. The increased activity observed in non-stimulated cells expressing high levels of CFTR probably reflects tonic activation of CFTR in resting cells as a result of routine cell handling during an experiment.
CFTR Mutants and AE Activity-To begin to elucidate the mechanism by which CFTR domains regulate AE activity, the effect of several mutations in CFTR that have been previously characterized in terms of CFTR Cl Ϫ channel activity were assessed. Fig. 11 shows the results obtained with CFTR mutants that did not affect AE activity. ⌬F508 and P205S CFTR are known maturation mutants (26,27) that do not reach the plasma membrane of 293 cells (Fig. 6). Hence, it was not surprising that they had no effect on AE activity. CFTR truncated at Asp-836 (between the R domain and NBD2) was reported to maintain Cl Ϫ channel activity when expressed in HeLa cells (31). However, expression of a similar construct truncated at Trp-846 in 293 cells was insufficient to activate AE (Fig. 11c).
Another series of mutations in NBD2 that are known to affect channel activity (Fig. 12) indicate that there is no correlation between Cl Ϫ channel activity and activation of AE, as predicted from the lack of effect of Cl Ϫ channel blockers. For example, the G1247D/G1249E CFTR double mutant was reported to have no Cl Ϫ channel activity (32), was expressed in the plasma membrane (Fig. 6k), and had no effect on AE activity (Fig. 12a). The K1250M CFTR mutant had increased channel activity (32), was expressed in the plasma membrane (Fig. 6i) and activated AE similar to WT CFTR (Fig. 12b). However, D1370N CFTR had nearly normal Cl Ϫ channel activity (32) and was expressed in the plasma membrane (Fig. 6j), but was unable to activate AE (Fig. 12c).
Taken together, the results presented here show that CFTR regulates Cl Ϫ /HCO 3 Ϫ exchange activity in stably transfected NIH 3T3 cells and transiently transfected HEK 293 cells. The anion exchange activity stimulated by CFTR has all the kinetic properties associated with anion exchange reported in many cell types (24). The only deviation was the relative insensitivity of the exchange activity to DIDS. However, expression of CFTR at high level was apparently responsible for this behavior. This finding highlights the need for caution when using cell lines and overexpression of CFTR to reach conclusions as to its function in native tissues. Ϫ -buffered solutions. After strong membrane depolarization by increasing external K ϩ to 100 mM, the cells were incubated in Cl Ϫ -free and then Cl Ϫ -containing high K ϩ medium before and after stimulation with 5 M forskolin. The cells were then incubated with 0.5 mM DIDS. As was found in normal K ϩ (5 mM) medium, DIDS had no effect in NIH 3T3 cells and only partially inhibited AE activity in 293 cells expressing high WT CFTR levels. Identical results were obtained when cells were incubated in Na ϩ -free media containing 125 mM K ϩ , 25 mM choline-HCO 3 Ϫ , and 5 M ethyl-isopropyl-amiloride to inhibit Na ϩ /H ϩ exchange activity (data not shown). Similar results were obtained in at least three experiments under each condition.
FIG. 11. Folding CFTR mutants and N-terminal half CFTR had no effect on AE activity. HEK 293 cells were cotransfected with plasmids carrying GFP and P205S CFTR (a), ⌬F508 CFTR (b), or Trp-846 termination codon CFTR (c). Immunocytochemical assays verified that cells expressing high levels of GFP also expressed high levels of the mutants. Cells expressing high levels of GFP were used to test the effect of forskolin stimulation on AE activity by the standard protocol of Cl Ϫ removal and addition. Expression of the above mutants did not activate AE before or after forskolin stimulation. Similar observations were seen in at least three experiments with each construct.
FIG. 12. Effect of CFTR mutations in NBD2 on AE activity. HEK 293 cells were transfected with the indicated constructs, all of which carrying mutations in NBD2. Cells expressing high levels of GFP were used for experimentation. The G1247D/G1249E double mutant was expressed in the plasma membrane but had no effect on AE activity (a). The K1250M mutant was at least as effective as WT CFTR in stimulating AE activity (b). The D1370N mutant had minimal effect on AE activity (c). Similar results were obtained in at least three experiments with each mutant.
Kinetic, pharmacological, and molecular data indicate that it is highly unlikely that the Cl Ϫ /HCO 3 Ϫ exchange activated by CFTR was mediated by CFTR itself. This is concluded from the findings that inhibitors and mutants of CFTR Cl Ϫ channel activity had no effect on Cl Ϫ /HCO 3 Ϫ exchange activity stimulated by CFTR. More importantly, this lack of correlation demonstrates that transport of Cl Ϫ by CFTR was not needed to observe increased exchange activity, although CFTR had to be activated by a cAMP-dependent mechanism to exert its effect on the AE. Thus, an activated conformation of the protein was needed for activation of AE. This is further supported by the findings with the ⌬F508 and P205S CFTR maturation mutants, which showed that expression of CFTR in the plasma membrane, rather than mere expression of CFTR in the cells, was required for activation of AE. In this respect the results obtained with D1370N CFTR are of particular interest since this mutation in NBD2 did not ablate channel activity (32) but eliminated regulation of AE activity by CFTR. Recently, it was reported that the D1506A mutation of the sulfonylurea receptor 1 protein, which corresponds to D1370N of human CFTR, similarly failed to stimulate the K ATP channel (33). This points to the importance of NBD2 in regulating other ion channel or transport proteins. It will be important to determine whether NBD2 of CFTR by itself can regulate AE activity. These experiments are in progress.
Previously, CFTR has been shown to modulate the activity of several ion channels and transporters. The best studied example is regulation of epithelial Na ϩ channels (ENaC) (34). Co-expression of CFTR and ENaC in Madin-Darby canine kidney cells resulted in attenuation of ENaC activity upon stimulation of CFTR with cAMP (34). This regulation appears to be by direct interaction between the two proteins as it can be reproduced with purified proteins reconstituted into planar lipid bilayers (35). CFTR was also proposed to regulate the inward rectifying K ϩ channel ROMK2 (36) in a mechanism similar to the regulation of the ATP-sensitive K ϩ channel by the sulfonylurea receptor in pancreatic ␤ cells (37). Such a regulation may account for the cAMP-dependent membrane repolarization in colonic crypt base cells (6,38). Finally, expression of CFTR in Xenopus oocyte and stimulation with cAMP increased water permeability (39), as if CFTR functions as or regulates a water channel. Similar to our findings with AE, in all these cases, activation of CFTR by cAMP, but not Cl Ϫ channel activity, was required for regulation of the transporters. This conclusion is reinforced by our findings that mutations in or truncation of NBD2 resulted in loss of regulation of AE by CFTR, even when Cl Ϫ channel activity was retained.
The regulation of AE by CFTR demonstrated here may be of particular physiological significance to understanding the pathophysiology of cystic fibrosis. It is, therefore, of interest to determine which of the known AE isoforms is regulated by CFTR and whether such regulation exists in a native CFTR-expressing tissue. Recently, we obtained evidence for regulation of Cl Ϫ /HCO 3 Ϫ exchange activity by CFTR in the intestinal cell line T84 and, more importantly, in duct cells of the mouse submandibular gland and pancreas. 2 In these studies, we also discuss the physiological significance of regulation of AE by CFTR.