Evidence against Defective trans -Golgi Acidification in Cystic Fibrosis*

Defective organelle acidification has been proposed as a unifying hypothesis to explain the pleiotropic cellular abnormalities associated with cystic fibrosis. To test whether cystic fibrosis transmembrane conductance regulator (CFTR) participates in trans -Golgi pH regulation, intraluminal trans -Golgi pH was measured in stably transfected Swiss 3T3 fibroblasts (expressing CFTR or (cid:68) F508-CFTR) and CFTR-expressing and nonexpressing epithelial cells. trans -Golgi pH was measured by ratio-imaging confocal microscopy using a liposome injection procedure to label the lumen of trans -Golgi with fluid phase fluorescein and rhodamine chro-mophores J. Biol. Chem. 270, 4967–4970). Selective labeling of trans -Golgi was confirmed by colocalization of the delivered fluid phase fluorophores with N -{6-[(7-nitro-benzo-2-oxa-1,3-diazol-4-yl)amino]caproyl}-sphingosine.InunstimulatedfibroblastsinHCO 3 (cid:50) -free buffer, trans Golgi 6.25 (cid:54) 0.04 (mean 6.30 0.03 ( CFTR) and 0.06 60, F508) significant). After

Although the cystic fibrosis (CF) 1 genotype is clearly associated with defective plasma membrane Cl Ϫ conductance (1-3), there is little known about the fundamental cellular abnormality that links genotype to cellular defect to clinical disease (for review, see Refs. 4 and 5). A provocative hypothesis to account for many of the observed cellular abnormalities in CF is that the CF genotype is associated with defective acidification of intracellular organelles (6). It was proposed that CFTR provides a conductive pathway across the limiting membrane of certain organelles of the endosomal and secretory pathways, which permits acidification by the electrogenic, vacuolar-type proton pump. The "defective organelle acidification" hypothesis is attractive because it may provide a unifying theme to explain the diverse cellular defects in cystic fibrosis involving abnormal post-translational modifications (decreased sialylation and increased fucosylation) of secreted and surface proteins (7)(8)(9) and increased cell adhesion of Pseudomonas aeruginosa (10,11). For example, decreased protein sialylation has been proposed to result from decreased activity of pH-dependent sialyltransferases in the trans-Golgi (12). The principal evidence supporting the hypothesis is (6): (a) the pH in early endosomes and trans-Golgi in CF cells, measured by DAMP immunoelectron microscopy, was 0.2-0.3 pH unit of alkaline relative to normal cells; and (b) ATP-dependent acidification in cell-free endosomal preparations suggested a functional Cl Ϫ channel in endosomes from normal but not CF cells.
A testable prediction of the defective acidification hypothesis is that organellar pH in cells expressing CFTR is lower than that in nonexpressing or CF cells, particularly after CFTR stimulation by cAMP agonists. In addition, CFTR should be present and functional as a Cl Ϫ conductance in the endosomal and trans-Golgi compartments of CFTR-expressing cells. Several studies have been carried out to define the role of CFTR in acidification of vesicular compartments of the endosomal pathway. Using transfected Chinese hamster ovary cells, Lukacs et al. (13) concluded that CFTR is functional in a mixed (early and late) endosomal compartment, but that CFTR expression does not influence endosome pH. Biwersi and Verkman (14) reported a similar conclusion for the early endosomal compartment in transfected 3T3 fibroblasts and T84 colonic epithelial * This work was supported by Grant HL42368 from the National Institutes of Health, Grants I958 and R613 from the National Cystic Fibrosis Foundation, a grant from the California Lung Association, and a grant from the University of California Tobacco Research Fund. 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. ‡ Supported in part by a fellowship from the California Lung Association.
§ To whom correspondence should be addressed. Tel: 415-476-8530; Fax: 415-665-3847; E-mail: verkman@itsa.ucsf.edu. 1 The abbreviations used are: CF, cystic fibrosis; CFTR, cystic fibrosis transmembrane conductance regulator; FS, fluorescein sulfonate; SR, sulforhodamine 101; CCCP, carbonyl cyanide m-chlorophenylhydrazone; C 6 -NBD-ceramide, N-{6-[(7-nitrobenzo-2-oxa-1,3-diazol-4-yl)amino]caproyl}-sphingosine; CPT, 8-(4-chlorophenylthio); SPQ, 3-(6methoxyquinolino)propanesulfonate; Cl 2 Cf-TMR-dextran, 5(and 6)carboxy-2Ј,6Ј-dichlorofluorescein-5(and 6)-carboxytetramethylrhodamine-dextran; MDCK, Madin-Darby canine kidney; PBS, phosphatebuffered saline. cells. Using CFPAC-1 cells, Dunn et al. (15) also found that CFTR and ⌬F508 CFTR expression did not affect endosomal acidification. These reports suggest that although functional CFTR is present in endosomal vesicles, a result supported by immunochemical studies (16 -18), CFTR-independent mechanisms are primarily responsible for the regulation of endosome pH. It is noted that the above mentioned studies were conducted in nonphysiological buffers in the nominal absence of HCO 3 Ϫ , which may be of significance in view of evidence suggesting that cAMP-activated CFTR may have substantial HCO 3 Ϫ conductance (19,20). There have been no studies of the role of CFTR in acidification of the trans-Golgi compartment, the principal compartment responsible for post-translational processing of synthesized proteins. To be able to test whether trans-Golgi acidification is defective in CF, we recently developed a direct method to measure the pH of the trans-Golgi compartment in living cells (21). Liposomes with entrapped, membrane-impermeant fluorescent pH indicators were microinjected into the cell cytoplasm. At 37°C and in the presence of ATP, liposomes of suitable size and composition fused selectively with trans-Golgi and delivered their aqueous-phase contents into the trans-Golgi lumen. The pH was then quantified by fluorescence ratio imaging using confocal microscopy. The initial studies performed in human skin fibroblasts gave an average trans-Golgi pH of 6.17 and showed that trans-Golgi pH was regulated by various cytosolic second messengers.
In this study, we have applied the liposome fusion method to compare pH in Swiss 3T3 fibroblasts expressing CFTR, ⌬F508-CFTR, or no CFTR, as well as in epithelial cells that natively express CFTR (Calu-3 cells; Ref. 22) and those that do not (MDCK and SK-MES-1 cells; Ref. 23). The influence of CFTR expression on trans-Golgi acidification was examined from steady-state pH measurements in unstimulated and cAMPstimulated cells in both HCO 3 Ϫ -free and HCO 3 Ϫ -containing buffers. In addition, because previous studies of endosomal acidification were carried out only in HCO 3 Ϫ -free buffer (13)(14)(15), pH was measured in individual fluorescently labeled endosomes in HCO 3 Ϫ -containing buffers. The results demonstrated that CFTR Cl Ϫ channels are not a significant determinant of trans-Golgi or endosomal pH.
Cell Culture-Control (vector-transfected) and stably transfected Swiss 3T3 fibroblasts expressing CFTR and ⌬F508 CFTR were provided by Dr. Michael Welsh. MDCK cells (ATCC CCL 34) were obtained from the University of California, San Francisco cell culture facility. Calu-3 cells (ATCC HTB 55) were provided by Dr. Jonathan H. Widdicombe, and SK-MES-1 cells (ATCC HTB 58) were provided by Dr. Walt Finkbeiner. Cells were grown at 37°C in Dulbecco's modified Eagle's-H21 medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. For microinjection, cells were grown on 18-mm diameter glass coverslips and used when ϳ80% confluent. For Calu-3, MDCK, and SK-MES-1 epithelial cells, the coverslips were coated with a 10 mg/ml collagen solution overnight, dried, and rinsed with PBS (140 mM NaCl, 3 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 1.5 mM KH 2 PO 4 , and 8 mM Na 2 HPO 4 , pH 7.4). SPQ Measurements-As described previously (14), cells were loaded with 5 mM SPQ added to the cell culture medium overnight. Coverglasses were then washed in PBS and mounted in a 200-l perfusion chamber in which the cell-free glass surface made contact with the immersion objective (Leitz; 40ϫ magnification, glycerol immersion, quartz; numerical aperture, 0.65). SPQ fluorescence was excited at 365 nm and detected by a photomultiplier using a 410-nm dichroic mirror and 420-nm barrier filter in a Nikon inverted epifluorescence microscope. Cell Cl Ϫ efflux was measured by the increase in SPQ fluorescence after addition of 0.5 mM CPT-cAMP to a Cl Ϫ -free buffer in which Cl Ϫ was replaced by nitrate.
Preparation of Liposomes-Multilamellar vesicles were prepared by dispersing dry dioleoylphosphatidylcholine (40 mg/ml) in 25 mM HEPES, 115 mM KCl, and 2.5 mM MgCl 2 , containing SR (5 mM) and FS (0 or 30 mM) or rhodamine B-dextran (40 mg/ml), pH 7.2. Multilamellar vesicles were then frozen in liquid nitrogen and thawed at 40°C for five cycles. Small, uniform-sized liposomes were obtained by passing 100 l of the liposome suspension through polycarbonate membranes (Nuclepore, Pleasanton, CA) of decreasing pore size (20 times through a 100-nm pore membrane then 20 times through a 50-nm pore membrane) using a LiposoFast extruder (Avestin, Ottawa, Ontario, Canada). External SR and FS were removed by Sephadex G-50 size-exclusion gel chromatography. External rhodamine B-dextran was removed on a Sepharose 4B size-exclusion gel in a 1-ml column. The liposome diameter was measured to be 70 Ϯ 1 nm by quasi-elastic dynamic light scattering (model N4; Coulter Electronics, Hialeah, FL). Liposomes were stored at 4°C for up to 3 weeks.
Microinjection Procedure-Cell microinjection was performed using glass needles prepared from thin-walled filament capillaries (FHC, Brunswick, ME) drawn to a fine tip (0.5-m hole diameter) with a vertical needle puller (Kopf, Tujunga, CA) and back-filled with injection solutions. Filled needles were mounted in the holder of an Eppendorf 5170 micromanipulator, and cells were injected with a volume of ϳ4 ϫ 10 Ϫ12 cm 3 using an Eppendorf 5242 microinjector over 0.5 s at an injection pressure of 100 kilopascals. During the procedure, the cells were visualized on a Nikon inverted epifluorescence microscope equipped with a ϫ40 air objective. C 6 -NBD-ceramide Labeling-To stain the trans-Golgi membrane (26), 50 nmol of C 6 -NBD-ceramide was dissolved in 200 l of ethanol and injected into 10 ml of 10 mM HEPES-buffered minimal essential medium containing 0.34 mg of bovine serum albumin. The solution was dialyzed overnight at 4°C against HEPES-buffered minimal essential medium. Cells were incubated with the C 6 -NBD-ceramide-bovine serum albumin complex for 5 min at 37°C and washed with PBS prior to observation.
Ratio-imaging Confocal Microscopy-Fluorescence microscopy was carried out on a Leitz epifluorescence microscope equipped with a Nipkow wheel coaxial-confocal attachment (Technical Instruments, San Francisco, CA). Cells were mounted in a perfusion chamber and viewed with a Nikon Plan-Apo ϫ60 oil immersion objective (numerical aperture, 1.4). Confocal fluorescence images were detected by a cooled CCD camera (AT200; Photometrics, Tucson, AZ) with a back-thinned, 14-bit detector (TK512CB; Tektronix). C 6 -NBD-ceramide or FS and SR or rhodamine B-dextran were visualized with standard fluorescein and rhodamine filter sets, respectively. Image pairs (SR and FS) were acquired (exposure time, 500 ms) for the same field containing one or more cells. Dark current and shading corrections were applied. Analysis was performed using PMIS software (Photometrics) on each image pair: the trans-Golgi contour on the SR image was drawn, and the integrated pixel intensity (I SR ) was calculated over the delimited area. The same area on the FS image was used for determination of integrated intensity (I FS ). For calculation of the FS-to-SR signal ratio, (I FS Ϫ B FS )/ (I SR Ϫ B SR ), the background signals (B FS and B SR ) were determined over a region of cytosol near the trans-Golgi contour.
In Situ pH Calibration-Calibrations were performed as described previously (21). Briefly, after microinjection with liposomes containing FS and SR, cells were incubated for 30 min at 37°C and mounted in the perfusion chamber. Prior to calibration, cells were perfused for 20 min with 10 nM bafilomycin A 1 in PBS at 23°C to inhibit the vacuolar proton pump. To set trans-Golgi lumenal pH equal to extracellular pH, cells were then incubated for 10 min with 10 nM bafilomycin A 1 , 10 M monensin (Na-H exchanger), and 1 M CCCP (protonophore) in a high K ϩ buffer containing 125 mM KCl, 20 mM NaCl, 25 mM HEPES, 0.5 mM MgSO 4 , and 0.5 mM CaCl 2 , titrated to specified pH values. The FSto-SR signal ratio was plotted as a function of trans-Golgi pH. The calibration relation was: trans-Golgi pH ϭ 2.5 ϩ 41.6r Ϫ 154r 2 ϩ 199r 3 , where r is the FS-to-SR signal ratio (correlation coefficient of polynomial regression, 0.99). A two-point calibration (pH 7.8 and 5.8) was performed for each set of experiments; FS-to-SR signal ratios were normalized to the values in the calibration curve and the trans-Golgi pH was calculated.

trans-Golgi Acidification in Cystic Fibrosis
Endosome pH Measurements-pH values of individual fluorescently labeled endosomes were measured by a quantitative imaging method developed previously (25). Endosomes were pulse labeled by a 5-or 10-min incubation with 5 mg/ml Cl 2 CF-TMR-dextran at 37°C and then maintained at 37°C for 5 or 20 min. Cells were viewed by wide field epifluorescence microscopy with a Nikon Plan-Apo ϫ60 oil immersion objective. Image pairs were acquired using fluorescein and rhodamine filter sets. At the end of the experiment, cells were perfused with high-K ϩ buffers containing 5 M nigericin titrated to specified pH values (see above). pH values were determined in individual fluorescent endosomes by ratio imaging software, which identified endosomes in image pairs and calculated background-subtracted, area-integrated pixel intensities (25). Fig. 1 shows the labeling of transfected 3T3 fibroblasts and MDCK epithelial cells after microinjection with liposomes containing fluid phase fluorescent probes. Selectivity of trans-Golgi labeling was demonstrated by colocalization of liposomedelivered SR with the trans-Golgi-specific lipid phase marker C 6 -NBD-ceramide (Fig. 1, A and B). The liposome delivery method worked for every cell type tested; similar micrographs showing colocalization of SR and C 6 -NBD-ceramide were obtained for Calu-3 and SK-MES-1 cells (data not shown). For pH determination, cells were microinjected with liposomes containing the fluorophores FS and SR. Fig. 1, C and D, shows colocalization of FS and SR fluorescence in 3T3 fibroblasts and MDCK cells; similar results were obtained for SK-MES-1 and Calu-3 cells (not shown). There was no differential photobleaching or dye leakage under the conditions of our experiments.

RESULTS
After a 30-min incubation at 37°C to permit delivery of the fluorescent probes, the specific labeling of the trans-Golgi lumen remained stable for more than 3 h at 23°C. However, further incubation at 37°C caused a rapid disappearance of the signal (Ͻ15 min), which could be due to dye leakage or downstream trafficking. To distinguish between these possibilities, liposomes containing rhodamine B-dextran 10,000 (a high molecular weight and membrane-impermeant chromophore) were microinjected, and trans-Golgi was stained with C 6 -NBD-ceramide (Fig. 2). Early after microinjection (5-15 min), little specific labeling was observed. Thirty minutes was found to be the optimal incubation time for labeling. Subsequently, trans-Golgi fluorescence declined over 45-90 min, and the labeling pattern became less distinct. A few cells (ϳ5%) were still observed with good trans-Golgi labeling at 60 min. We note that the ability to follow the fate of high molecular weight fluorescent probes in the lumen of the secretory compartment should enable the study of secretory trafficking by fluorescence microscopy.
The cell types used for subsequent trans-Golgi pH measurements were examined for functional expression of CFTR at the plasma membrane using the SPQ fluorescence method in which cytoplasmic Cl Ϫ was measured in response to Cl Ϫ -nitrate exchange and CPT-cAMP addition (14). SPQ measurements showed strong CPT-cAMP-stimulated Cl Ϫ efflux in CFTR-transfected fibroblasts and Calu-3 cells but little change in Cl Ϫ efflux in the other cell types (Fig. 3). These functional results are consistent with the known CFTR expression in the transfected fibroblasts and Calu-3 cells (22,23,27).
trans-Golgi pH measurements were performed on 3T3 fibroblasts and several types of epithelial cells by ratio-imaging confocal microscopy, as described and validated previously (21). FS (pH sensitive) and SR (pH insensitive) were delivered together in the microinjected liposomes. Absolute trans-Golgi pH was calculated by quantitative image analysis from the FSto-SR signal ratio, followed by a two-point in situ calibration in which the trans-Golgi and solution pH were set equal using buffers (pH 7.8 and 5.8) containing high K ϩ and ionophores (Fig. 4, inset). No significant difference was found between the trans-Golgi pH of mock-transfected (pH 6.25 Ϯ 0.04), CFTRtransfected (pH 6.30 Ϯ 0.03), and ⌬F508-CFTR-transfected (pH 6.23 Ϯ 0.06) 3T3 fibroblasts in HCO 3 Ϫ -free buffer (Fig. 4). Similarly, in 25 mM HCO 3 Ϫ , the trans-Golgi pH of mock-transfected fibroblasts (6.22 Ϯ 0.07) was not significantly different from the trans-Golgi pH of CFTR-transfected fibroblasts (6.28 Ϯ 0.07). However, after addition of CPT-cAMP, the trans-Golgi pH in the three cell types increased in HCO 3 Ϫ -free buffer by 0.1-0.2 pH unit, whereas no significant change was found in HCO 3 Ϫ -containing buffer (see "Discussion"). The trans-Golgi pH values of MDCK, SK-MES-1, and Calu-3 epithelial cells were also determined in HCO 3 Ϫ -containing buffer before and after stimulation of CFTR by CPT-cAMP (Fig. 5)

lung carcinoma cells that are similar in morphology and function to Calu-3 cells (23). The trans-Golgi pH of MDCK cells in HCO 3
Ϫ -free buffer was 6.30 Ϯ 0.03. A small alkalinization was found on addition of CPT-cAMP (0.5 mM) for 20 min at 23°C (pH 6.41 Ϯ 0.05). In HCO 3 Ϫ -containing buffer, the trans-Golgi pH (6.36 Ϯ 0.04) in MDCK cells did not change after CPT-cAMP stimulation (6.34 Ϯ 0.08). CPT-cAMP addition did not change the trans-Golgi pH significantly in Calu-3 and SK-MES-1 cells in HCO 3 Ϫ -containing buffer. These results provide the first data on trans-Golgi pH in epithelial cells and indicate that CFTR expression and activation do not influence the trans-Golgi pH.
The pH was next measured in individual endosomes in control and CFTR-expressing fibroblasts in HCO 3 Ϫ -containing buffers. Cohorts of endosomes were pulse labeled with Cl 2 CF-TMRdextran and incubated at 37°C for a total of 10 or 30 min, and the endosome pH was measured by ratio-imaging microfluorimetry (25). Cl 2 CF was chosen because of its good pH sensitivity at low pH values in the range of 4.5-6.0 (pK a ϳ5.3). Fig.  6 shows images of CFTR-transfected 3T3 fibroblasts after 10and 30-min incubations. Individual fluorescent endosomes are clearly visualized in the fluorescein (left) and rhodamine (right) images. Labeled vesicles (late endosomes and some lysosomes) are larger and located closer to the nucleus at 30 min. Fig. 7A gives averaged endosome pH values measured in the absence and presence of forskolin. Averaged endosome pH values were very similar at each incubation time, with no significant effect of CFTR expression or forskolin activation. To examine the pH values of individual endosomes, the pH distribution was quantified. Fig. 7B shows the pH distribution as the percentage of endosomes in 0.25 (10 min) or 0.2 (30 min) pH unit intervals. There was a unimodal pH distribution for both cell types at 10 and 30 min, with and without forskolin, without evidence of subpopulations of highly acidic endosomes. These results indicate little effect of CFTR expression and activation on endosome pH in HCO 3 Ϫ -containing buffers.

DISCUSSION
The purpose of this study was to test the "defective organelle acidification" hypothesis predicting that the trans-Golgi pH is lower in cells containing functional, activated CFTR than in cells without CFTR. A stably transfected cell line was used, which strongly expressed functional CFTR at the plasma membrane as well as in intracellular vesicles. Previous studies testing whether endosomal acidification is defective in cystic fibrosis were performed in these same cells (14), as well as in stably transfected Chinese hamster ovary (13), T84 (14), and CFPAC-1 (15) cells. CFTR-transfected cells were chosen because direct comparisons could be made with control (vectortransfected) and ⌬F508-expressing cells, and because the high CFTR expression levels were predicted to amplify any pH dif-

trans-Golgi Acidification in Cystic Fibrosis
ferences. In addition, trans-Golgi pH measurements were carried out using several non-CFTR-expressing epithelial cell lines (MDCK and SK-MES-1) and an epithelial cell line that natively expresses CFTR (Calu-3). A liposome injection method developed recently by our laboratory permitted the selective delivery of pH-sensitive fluorophores into the lumen of the trans-Golgi for determination of pH by ratio-imaging confocal microscopy. It was found that CFTR expression and activation did not influence the trans-Golgi pH significantly in any of the cells tested in HCO 3 Ϫ -containing or HCO 3 Ϫ -free buffers, with a S.E. of generally better than 0.1 pH unit. Measurements of endosome pH using HCO 3 Ϫ -containing buffers extended previous observations that the endosome pH is not influenced by CFTR expression and activation. Taken together, these findings provide direct evidence against the defective organelle acidification hypothesis.
Several caveats should be pointed out in evaluating the strength of our conclusion. The studies were performed in stably transfected cells and a CFTR-expressing epithelial cell line, not in vivo in human airway epithelial cells. For the reasons discussed above, we believe that functional effects of CFTR on trans-Golgi acidification should have been apparent in these systems. The accuracy of the pH measurements here are probably not much better than 0.1 unit; therefore, very subtle changes in pH would not be detected. We believe that such small pH differences, if present, are unlikely to be physiologically important based on the intrinsic pH variability in trans-Golgi of different cell types (for example, between 3T3 fibroblasts and MDCK cells) and the sensitivity of organellar pH to various cytosolic factors (21). Last, it is recognized that the conclusion applies to acidification in endosomes and trans- Ϫ -containing buffer (120 mM NaCl, 25 mM NaHCO 3 , 2.5 mM K 2 HPO 4 , 1 mM MgSO 4 , 1 mM CaCl 2 , 20 mM HEPES (pH 7.4), and 5 mM glucose, equilibrated with 5% CO 2 ). Where indicated, cells were incubated with 0.5 mM CPT-cAMP for 20 min at 23°C. Data are mean Ϯ S.E., with numbers of experiments given in parentheses. *, p Ͻ 0.05, comparing with versus without CPT-cAMP. Inset, representative calibration performed after each experiment (see "Materials and Methods"); where indicated, 10 nM bafilomycin A1 was added, and PBS was exchanged for a high K ϩ buffer containing 10 nM bafilomycin A 1 , 10 M monensin, and 1 M CCCP at pH 7.8 and then pH 5.8.

FIG. 5. Effect of CPT-cAMP on trans-Golgi pH of MDCK, Calu-3, and SK-MES-1 epithelial cells in HCO 3
؊ -containing buffer. Measurements were performed as in Fig. 4 at 50 min after microinjection in PBS or a HCO 3 Ϫ -containing buffer. Where indicated, cells were incubated with 0.5 mM CPT-cAMP for 20 min at 23°C. Data are mean Ϯ S.E., with numbers of experiments in parentheses. *, p Ͻ 0.05, comparing with versus without CPT-cAMP.
FIG. 6. Fluorescence micrographs of endosome labeling of CFTR-transfected 3T3 fibroblasts. Endosomes were pulse labeled (5 or 10 min) with Cl 2 CF-TMR-dextran (5 mg/ml) and incubated at 37°C for a total of 10 or 30 min. Cl 2 CF and TMR images were visualized with fluorescein (left) and rhodamine (right) filter sets, respectively. Scale bar, 10 m.
Golgi and not to acidification in other subcellular compartments in which CFTR might be localized.
If defective organellar acidification does not account for the multiple cellular abnormalities associated with the cystic fibrosis genotype, then what function of CFTR does? The possibility remains but seems unlikely that the pleiotropic set of defects in CF is a consequence of defective CFTR Cl Ϫ conductance at the cell apical membrane. Recent data implicate a role for CFTR in regulating other ion channels, including the epithelial Na ϩ channel (28) and the outwardly rectifying Cl Ϫ channel (29). CFTR may itself transport nucleotides (30), water (31), and other substances as well as chloride. The cystic fibrosis phenotype may be associated with defective vesicular trafficking (32) and/or intracellular vesicle fusion (33). Although our negative results here provide no new information about the cellular basis of cystic fibrosis, they underscore the need to investigate the above possibilities, as well as other novel pathophysiological mechanisms.
An interesting finding in these experiments was the effect of CPT-cAMP on trans-Golgi pH. Alkalinization was induced by CPT-cAMP in HCO 3 Ϫ -free buffer in all cell types, whereas the trans-Golgi pH was unaffected in HCO 3 Ϫ -containing buffer. The mechanism of CPT-cAMP-induced alkalinization cannot be deduced from the available data. Possible mechanisms include cAMP inhibition of the vacuolar H ϩ pump or of a trans-Golgi ion channel, or cAMP stimulation of a passive H ϩ (OH Ϫ )coupled transporter. In HCO 3 Ϫ -containing buffer, the absence of a significant cAMP effect on pH might result from the presence of a cAMP-regulated HCO 3 Ϫ transport mechanism, which leads to lumenal acidification which balances the alkalinization. Whatever the mechanistic basis of the cAMP effect, the insensitivity of trans-Golgi pH to CFTR expression and activation indicates that transporters other than CFTR are responsible for the regulation of pH in this compartment.
Our results indicate that the trans-Golgi pH is not affected by CFTR expression and cAMP stimulation in CFTR-transfected and natively expressing cell models. CFTR function as a Cl Ϫ channel in trans-Golgi was not studied, because trans-Golgi vesicles cannot be purified, and because there is no effective approach to access the trans-Golgi membrane in intact cell experiments. We believe that CFTR is likely to be functional in trans-Golgi based on evidence that CFTR is functional in endosomes (13) and nuclear and endoplasmic reticulum membranes (34), as well as data for other polytopic membranetransporting proteins including the related multidrug resistance protein, MDR (35). Finally, it is emphasized here that the principal conclusion of this study rests not on the density and function of CFTR in the trans-Golgi-limiting membrane, but on measured trans-Golgi pH, the penultimate determinant of protein processing.