Effects of lovastatin on trafficking of cystic fibrosis transmembrane conductance regulator in human tracheal epithelium.

Genetic defects in the cystic fibrosis transmembrane conductance regulator (CFTR), a cAMP-activated chloride channel, cause cystic fibrosis. Most defective forms of CFTR show improper intracellular trafficking. Because isoprenylated, small GTP-binding proteins are involved in the vesicular trafficking of other integral membrane proteins, we have investigated the role of isoprenylation in the trafficking of CFTR to the apical membranes of primary cultures of human airway epithelium and of Calu-3 cells, a human lung carcinoma cell line. CFTR function was measured as short circuit current, 125I efflux, and conductance of cell sheets with permeabilized basolateral membranes. Lovastatin, an inhibitor of isoprenyl lipid biosynthesis, markedly inhibited all measures of CFTR function. The lovastatin-induced declines in CFTR function were corrected by the simultaneous addition of mevalonate or the isoprenyl lipids geranylgeranyl and farnesyl but not cholesterol. Lovastatin reduced total cellular CFTR as assessed by immunoprecipitation. Mevalonate or isoprenyl lipids protected CFTR levels from the actions of lovastatin. Together, these results suggest a role for isoprenyl lipids, presumably through the actions of small GTP-binding proteins, in the trafficking of CFTR to the apical membrane of human airway epithelium.

Cystic fibrosis (CF) 1 is the commonest lethal genetic disease in Caucasians. The gene responsible for CF has been identified (1) and found to encode a 170-kDa glycoprotein known as the cystic fibrosis transmembrane conductance regulator (CFTR). Several lines of evidence have suggested that CFTR is a cAMPregulated chloride channel in the apical membrane of airway (2,3) and other epithelial cells (4). A relationship between mutations in the CF gene and defective chloride conductance has been clearly demonstrated (5).
Although a variety of mutations in the CFTR gene lead to CF, the commonest involves a deletion of the phenylalanine residue at the 508 position (⌬F508) (6). In transfected cells, many of these CF gene mutations, including the ⌬F508 variant, result in protein products that are not completely processed, are retained in the ER, and are trafficked into a degradation pathway associated with the ER rather than to the plasma membrane (7,8). At reduced temperatures (Ͻ30°C), CFTR ⌬F508 and other mutant forms are correctly trafficked to the plasma membrane and confer a cAMP-activated chloride conductance of similar magnitude to that produced by wild-type CFTR (9). Less frequently observed defective forms of CFTR are trafficked properly to the apical plasma membrane at 37°C but show reduced capacity as chloride channels (10).
Little is known about the trafficking pathways for wild-type CFTR. Previous studies have demonstrated that a class of isoprenylated small GTP-binding proteins (termed Rab proteins) are involved in the trafficking of intracellular vesicles (such as those that would contain CFTR) to the plasma membrane of polarized epithelial cells (11). A recent report suggests that isoprenylated proteins may be involved in trafficking of CFTR in a colonic carcinoma cell line (12). The results reported here strongly support a role for isoprenylated proteins in the trafficking of the CFTR to the apical membrane of human airway epithelium.

MATERIALS AND METHODS
Cell Culture-Primary cultures of human tracheal epithelium (HTE) or a human lung carcinoma cell line (Calu-3) were plated at 10 6 cells/ cm 2 and grown to confluency on Transwell inserts (pore size, 0.45 m; surface area, 1 cm 2 ; Costar, Cambridge, MA) coated with human placental collagen (13). High levels of differentiation of HTE were achieved by adding media containing 2% Ultraser G to the basolateral side of the insert only, leaving an air-interface at the apical surface (14). HTE cultures were used once the transepithelial resistance (R te ) was Ͼ250 ohms⅐cm 2 , which occurred approximately 5-7 days after plating. Calu-3 cells were used once a R te of Ͼ100 ohms⅐cm 2 was achieved, typically after 10 -14 days of culture.
Additions of lovastatin (Merck Sharp & Dohme) were made from a 50-fold concentrated ethanolic solution to produce a final concentration of 50 M. Mevalonic acid (Sigma), farnesyl pyrophosphate (F-PP), and geranylgeranyl pyrophosphate (GG-PP) (American Radiolabeled Chemicals; St. Louis, MO) were added from stocks prepared in 70% ethanol and 0.075 M NH 4 HCO 3 . Cholesterol (Sigma) was added from an ether stock. The same volumes of vehicles alone were added to control cells.
Measurement of Short Circuit Current (I sc ) and Transepithelial Resistance (R te )-Cell sheets were mounted in Ussing chambers and bathed with bicarbonate-buffered Krebs-Henseleit solution (pH 7.4) mixed by a gas lift of 95% O 2 and 5% CO 2 at 37°C. Transepithelial potential differences were clamped to zero, and the resulting I sc was displayed continuously on a pen recorder. Every 20 s the R te was determined from the size of the current deflections resulting from 0.2-s voltage pulses of constant amplitude (0.2-1 mV). Stimulation of cAMPdependent chloride secretion was achieved by the addition of 10 Ϫ5 M isoproterenol or 10 Ϫ5 M forskolin.
Halide Efflux-Effluxes were performed as described (15,16). Cell sheets were loaded with 125 I by placing them in serum-free medium containing Na 125 I (10 Ci/ml, 3 nM, and 17.4 Ci/mg; DuPont NEN) for 2 h. Residual surface-associated tracer was removed with two 200-ml washes (15 s each) using Krebs-Henseleit solution, and monolayers were placed, mucosal side up, in the top half of a Swinnex 25-mm filter holder (Millipore Corp. Bedford, MA) containing a 0.65-m pore, type D, cellulose ester filter. Oxygenated Krebs-Henseleit solution was passed over the mucosal surface of the monolayers and through surrounding pores in the filter holder at 1 ml/min using a peristaltic pump. Effluent fractions were collected at 1-min intervals. The perfusate temperature was raised to 37°C using a heated copper wire wrapped around the glass inlet tubing. This wire connected to a temperature controller, which responded to a thermocouple inside the Swinnex filter holder.
Stable base-line efflux was achieved after an initial 7-min period at room temperature. Over the next 2 min, perfusion temperature was raised to 37°C. We have shown elsewhere (16) that the temperatureinduced increase in 125 I efflux occurs via CFTR. We collected 13 further effluent samples after the initial seven. All samples were counted on a ␥ counter. The filter and medium remaining in the Swinnex filter holder at the end of the experiment were pooled and also counted. Efflux was expressed as the fractional loss of radiolabel over time (i.e. total cell counts lost/min of sampling period divided by the average total counts present in the cells during that time period).
Nystatin Permeabilization-Cell sheets were bathed with a Krebs-Henseleit solution on their serosal side (120 mM NaCl) and on their mucosal sides with a solution in which all but 20 mM NaCl was replaced by sodium gluconate. Nystatin (0.72 mg/ml; Sigma) was added to the serosal bath to permeabilize the basolateral membranes. Under these circumstances, changes in R te and I sc reflect apical membrane chloride channel activity (17).
Immunoprecipitation of CFTR-Our methods are described in full elsewhere (13). In brief, cell sheets were lysed with 0.1% SDS, Triton X-100, and sodium deoxycholate; samples of lysate were incubated with a mouse monoclonal antibody raised against a C terminus peptide of human CFTR (Genzyme, Cambridge, MA). Antibody-antigen complexes were then adsorbed onto pansorbin cells (Calbiochem) and pelleted. CFTR was then phosphorylated by adding protein kinase A catalytic subunit (Sigma) and [␥-32 P]ATP (30 Ci/mmol; 2 mCi/ml; DuPont NEN). Phosphoproteins were separated by 6% SDS-PAGE, and levels of 32 P were determined by densitometric scans of autoradiographs.

RESULTS
Lovastatin, an inhibitor of 3-hydroxy-3-methylglutaryl coenzyme A reductase, blocks one of the earliest dedicated steps of the cholesterol synthetic pathway and inhibits the production of several metabolically important intermediates (18) (Fig. 1). In order to determine whether lovastatin inhibited isoprenylation of proteins in our cells, our first studies investigated the extent to which post-translational modification of p21 ras had occurred. The primary translation product of p21 ras is cytosolic. On isoprenylation it becomes membrane-bound and shows slightly increased mobility on SDS-PAGE (19,20). 12 h after the addition of 50 M lovastatin to HTE, a retardation of p21 ras electrophoretic mobility in SDS-PAGE was seen in Western blots (Fig. 2), reflecting a reduction in the levels of membraneassociated Ras and an accumulation of the cytosolic form as described previously (19). These results show that these conditions of lovastatin exposure resulted in a reduction in the extent of isoprenylation of HTE. Similar results were obtained on Calu-3 cells (data not shown). Mevalonate is a pathway intermediate whose production is blocked by lovastatin. When added at 2.5 mM at the same time as lovastatin, conversion of the cytosolic to the membrane-associated form of p21 ras was restored (Fig. 2).
Ussing chamber studies provided an initial assessment of the effects of lovastatin on chloride secretion. Base-line short circuit current (I sc ) of HTE was ϳ5 A/cm 2 but reached 15-20 A/cm 2 after stimulation with 10 M isoproterenol, an increase reflecting induction of chloride secretion (14). Treatment with 50 M lovastatin completely blocked the chloride secretion induced by isoproterenol after approximately 12 h of exposure (Fig. 3). Washout of the lovastatin resulted in complete recovery of the isoproterenol-induced increase in I sc after a further 24 h of incubation (Fig. 3). Lovastatin at all exposure times had no significant effect on R te (data not shown).
HTE treated for 12 h with 50 M lovastatin showed reduced increases in 125 I efflux in response to a temperature change from 22 to 37°C (Fig. 4A). To evaluate the relative importance of reduced levels of the 15-carbon farnesyl and 20-carbon geranylgeranyl isoprenoid intermediates of the cholesterol synthetic pathways, the pyrophosphate forms of these intermedi-  (Fig. 4B). Similar results were obtained using Calu-3 cell monolayers (Fig. 5). By contrast, 250 M cholesterol did not block the inhibition induced by lovastatin (data not shown).
Permeabilizing the basolateral plasma membrane of airway epithelial cells with nystatin has been used previously as a means of directly investigating anion transport across the apical membrane (17). In the presence of a transepithelial chloride gradient, I sc of Calu-3 cells increased on exposure to nystatin (Fig. 6), reflecting the formation of a chloride concentration gradient across the apical membrane. Further stimulation of I sc occurred following the addition of forskolin. This stimulation of I sc was inhibited by the addition of 2 mM diphenyl amino carboxylic acid, a blocker of the CFTR chloride conductance (4). Pretreatment of Calu-3 cells with 50 M lovastatin dramatically inhibited the increases in I sc associated with addition of nystatin or forskolin (Fig. 6). Calu-3 cells simultaneously exposed to lovastatin plus 2.5 mM mevalonate were nearly completely protected from the actions of lovastatin; 10 M GG-PP was slightly less protective, and 10 M F-PP offered only a moderate amount of protection (Fig. 6).
Disruption of the cholesterol synthetic pathway by lovastatin has previously been shown to impede intracellular trafficking by removing the farnesyl and geranylgeranyl lipids required for effective membrane association of small GTP-binding proteins (21). Because treatment with lovastatin resulted in a significant decrease in the cAMP-dependent iodide efflux from Calu-3 cells, it was not surprising that lovastatin (50 M, 24 h) reduced total CFTR levels of Calu-3 cells to 40 Ϯ 7% of control, as determined by immunoprecipitation (Fig. 7). The observed net reduction of cellular CFTR content under conditions induced by lovastatin suggests that the degradation route(s) for this chloride channel is still competent. The effect of lovastatin on total CFTR content was reversed by 2.5 mM mevalonate (Fig. 7), 10 M GG-PP, or 10 M F-PP (data not shown). DISCUSSION Trafficking pathways for CFTR are rapidly being elucidated. CFTR is synthesized and partially glycosylated in the ER. During passage through the Golgi, the protein becomes fully glycosylated and is transported to the apical membrane. However, only ϳ25% of wild-type CFTR successfully passes from the ER to the Golgi as revealed by pulse-labeling with [ 35 S]me- thionine in nonpolar cells transfected with CFTR cDNA and in epithelial cells naturally expressing CFTR (24,25). The remaining 75% of CFTR protein is believed to be degraded in the ER (24,26). By contrast, essentially all ⌬F508 CFTR is degraded in the ER (25,(27)(28)(29), a defect in trafficking that is probably due to altered folding of the mutant protein (27). Interestingly, the fractions of wild-type and mutant protein passing from the ER to the Golgi can be altered pharmacologically. In nonpolar human airway epithelial cell lines, stimulation of the heteromeric G protein, G␣ 1-2 , reduced the amount of wild-type CFTR trafficking to the apical membrane (30). Conversely, inhibition of G␣ 1-2 caused the appearance of a cAMPactivated chloride conductance in the membranes of CF cell lines (30).
There is also evidence that CFTR is trafficked to and from the apical membrane by exo-and endocytosis. First, endosomes contain functional CFTR (31)(32)(33)(34). Second, forskolin-induced increases in 125 I efflux and transepithelial chloride secretion in T 84 cells are reduced by inhibitors of microtubule formation (35). Third, Prince et al. (36) have shown that CFTR is rapidly internalized in T 84 cells or in CFPAC cells transfected with wild-type CFTR and that the rate of internalization is inhibited by forskolin. Of interest was the finding that the inhibitory effect of forskolin on CFTR internalization required passage of chloride through CFTR; it was not seen in chloride-free (gluconate) medium or in CFPAC cells transfected with nonfunctional, correctly trafficking mutant CFTR (36).
It is almost certain that all of these various routes for trafficking of CFTR involve small GTP-binding proteins (21). Most of these are Rab proteins, which are 21-25 kDa in mass and are cytosolic immediately following their manufacture. However, addition of farnesyl or geranylgeranyl moieties to C-terminal cysteine residues leads to their association with cell membranes (37). Specifically, the isoprenylated Rab proteins bind GDP and GDP dissociation inhibitor proteins. Binding of GTP displaces GDP and GDP dissociation inhibitor, resulting in the GTP-associated form of the isoprenylated Rab protein, which interacts with a Rab receptor and becomes associated with the membrane (37,38). Rab proteins on the surface of vesicles interact with GTPase-activating proteins and specific Rab receptors on the target membrane, leading to membrane fusion. Conversion of GTP to GDP leads to the return of the Rab protein to the cytosol.
A knowledge of which GTP-binding proteins regulate the various pathways for CFTR movement within airway epithelial cells could lead to therapies for CF. As an initial approach to this area, we have determined the changes in CFTR distribution in polarized airway epithelial cells when the functions of multiple Rab proteins are disrupted by incubation with lovastatin, which acts to inhibit synthesis of farnesyl and geranylgeranyl. First, we established that the dose of lovastatin used (50 M, 12 h) inhibited farnesylation of p21 ras . We then showed that this dose inhibited isoproterenol-induced I sc across HTE with a t1 ⁄2 of ϳ5 h. This was not a nonspecific toxic effect because I sc recovered with a t1 ⁄2 of ϳ12 h, R te was not altered, and similar effects were obtained by treating cells with Nacetyl-S-geranylgeranyl-L-cysteine, a specific blocker of methyl esterification of geranyl geranylated proteins (Ref. 39 and data not shown).
Transepithelial chloride secretion (measured as isoproterenol-induced I sc ) reflects the coordinated function of several transport proteins in both the apical and basolateral membranes (40), and changes in net chloride secretion need not reflect changes in CFTR content of the apical membrane. Therefore, we measured CFTR activity as the temperatureinduced increases in 125 I efflux. Iodide is not transported by the basolateral NaK 2 Cl cotransporter (41,42) and exits cells only through chloride channels. Iodide efflux from HTE does not respond to elevation of cAMP (15,16), probably because CFTR is activated under base-line conditions in these cells. However, the temperature-dependent increase in 125 I efflux is inhibited by diphenyl amino carboxylic acid and blockers of protein kinase A but not by DIDS or 1,2-bis(2-aminophenoxyl) ethane-N,N,NЈ,NЈ-tetraacetic acid-acetoxymethyl ester. Thus, iodide flux from these cells is via CFTR rather than calcium-activated halide channels (16). When the basolateral membrane of Calu-3 cells is permeabilized with nystatin in the presence of a transepithelial chloride gradient, then changes in I sc and conductance reflect changes in the apical membrane chloride efflux (17). In this preparation, there is no evidence for calciumactivated chloride channels (17), and therefore cAMP-induced changes in I sc across nystatin-permeabilized Calu-3 cells provide another assay of CFTR function.
We found that treatment with lovastatin for 12 h reduced temperature-induced increases in 125 I efflux to 42 Ϯ 10 and 29 Ϯ 5% of control in Calu-3 cells and HTE, respectively. When Calu-3 cells were exposed to 50 M lovastatin for 24 h, the I sc of nystatin-permeabilized cells in the presence of forskolin was 23% of control. Detailed time courses for the effects of lovastatin on CFTR function were not obtained in these studies. However, assuming an exponential decline toward zero, these data suggest that with vesicular trafficking inhibited, the t1 ⁄2 for CFTR in the apical membrane is 7-11 h, which is in reasonable agreement with the estimated turnover of ϳ14 h determined for CFTR in T 84 cells treated with antisense oligonucleotides to CFTR mRNA (43). All of the effects of lovastatin on CFTR function could be wholly or partially reversed by the addition of mevalonate, GG-PP, or F-PP which allow cells to regain their protein isoprenylation capabilities. By contrast, the addition of cholesterol (200 M, 24 h), which is downstream of farnesyl and geranylgeranyl in the metabolic pathway (Fig. 1), had no effect on the inhibitory action of lovastatin (data not shown).
Similar effects of lovastatin on chloride transport have been reported for T 84 cells (12). In these cells lovastatin also abolished I sc , though at longer exposures (3 days) than those used by us. However, unlike our results, lovastatin markedly decreased R te (to zero with 10 M lovastatin for 3 days) in those studies. Furthermore, recovery from lovastatin exposure was not assessed. When function of apical membrane cAMP-activated chloride conductance was assessed in nystatin-permeabilized T 84 cells, exposure to 2.5 mg/ml for 2 days reduced the cAMP-stimulated change in I sc to ϳ50% of control (12).
Alterations in the cAMP-dependent apical membrane chloride conductance could reflect changes in the properties of individual channels (e.g. unitary conductance or probability of opening) or could be due to changes in channel density. These possibilities can only be definitively distinguished by patchclamp analysis. However, because lovastatin markedly decreased total CFTR in both Calu-3 and HTE (Fig. 7), we speculate that inhibition of trafficking pathways resulted in decreased CFTR content of the apical membrane.
In summary, a variety of approaches suggest that lovastatin reduces the numbers of cAMP-dependent chloride channels (CFTR) in the apical plasma membrane of human airway cells. These effects could be reversed by mevalonate, farnesyl, or geranylgeranyl (intermediates of the cholesterol synthetic pathway) but not by cholesterol. The most likely explanation for these results is that lovastatin lowers levels of isoprenyl lipids, thereby inhibiting the function of small GTP-binding proteins and disrupting the trafficking of CFTR to the apical membrane. Future studies will be directed at identifying the individual GTP-binding proteins involved in these events.