PDZ Domain Interaction Controls the Endocytic Recycling of the Cystic Fibrosis Transmembrane Conductance Regulator*

The C terminus of CFTR contains a PDZ interacting domain that is required for the polarized expression of cystic fibrosis transmembrane conductance regulator (CFTR) in the apical plasma membrane of polarized epithelial cells. To elucidate the mechanism whereby the PDZ interacting domain mediates the polarized expression of CFTR, Madin-Darby canine kidney cells were stably transfected with wild type (wt-CFTR) or C-terminally truncated human CFTR (CFTR-ΔTRL). We tested the hypothesis that the PDZ interacting domain regulates sorting of CFTR from the Golgi to the apical plasma membrane. Pulse-chase studies in combination with domain-selective cell surface biotinylation revealed that newly synthesized wt-CFTR and CFTR-ΔTRL were targeted equally to the apical and basolateral membranes in a nonpolarized fashion. Thus, the PDZ interacting domain is not an apical sorting motif. Deletion of the PDZ interacting domain reduced the half-life of CFTR in the apical membrane from ∼24 to ∼13 h but had no effect on the half-life of CFTR in the basolateral membrane. Thus, the PDZ interacting domain is an apical membrane retention motif. Next, we examined the hypothesis that the PDZ interacting domain affects the apical membrane half-life of CFTR by altering its endocytosis and/or endocytic recycling. Endocytosis of wt-CFTR and CFTR-ΔTRL did not differ. However, endocytic recycling of CFTR-ΔTRL was decreased when compared with wt-CFTR. Thus, deletion of the PDZ interacting domain reduced the half-life of CFTR in the apical membrane by decreasing CFTR endocytic recycling. Our results identify a new role for PDZ proteins in regulating the endocytic recycling of CFTR in polarized epithelial cells.

The selective expression of transport proteins in either the apical or basolateral membrane is essential for polarized epithelial cells to carry out vectorial transport of ions and water (1)(2)(3)(4). For example, polarization of the cystic fibrosis transmembrane conductance regulator (CFTR) 1 to the apical plasma membrane is required for vectorial Cl Ϫ secretion across a variety of epithelial cells including those in airway, kidney, intestine, and pancreas (4 -7). In the genetic disease cystic fibrosis, the most common mutation in the CFTR gene, ⌬F508, causes CFTR to fold incorrectly and to be retained in the endoplasmic reticulum (6 -9). Because ⌬F508-CFTR does not reach the apical plasma membrane, epithelial cells in the airway, pancreas, and intestine do not secrete chlorine (6,7,10).
Transport proteins contain amino acid motifs that direct and/or localize proteins to the appropriate membrane domains (1)(2)(3)(4). Highly conserved motifs that direct the polarized expression of transport proteins to the basolateral membrane of epithelial cells include tyrosine-and dileucine-based motifs (1)(2)(3)(4). PDZ domains, which are named for three proteins in which this domain was first described (PSD-95, Dlg, and ZO-1), also determine the polarized expression of proteins in epithelial cells and neurons (11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). PDZ domains are modular 70 -90-amino acid domains that bind to short peptide sequences at the C termini of other proteins, called PDZ interacting domains (15,21,(23)(24)(25). Previously, we demonstrated that a C-terminal, PDZ interacting domain is required for the polarization of CFTR to the apical plasma membrane in airway and kidney epithelial cells (16,18,19). Deletion of the PDZ interacting domain (TRL, using the single letter amino acid code) abrogated the polarized expression of CFTR in the apical membrane and eliminated CFTR-mediated transepithelial Cl Ϫ secretion. However, the mechanism whereby the PDZ interacting domain directs the polarized expression of CFTR to the apical plasma membrane is unknown.
The objectives of the present study were to test whether the PDZ interacting domain of CFTR is an apical membrane sorting motif that directs the trafficking of CFTR to the apical membrane or is a retention motif that results in the polarization of CFTR by selectively retaining CFTR in the apical membrane by interacting with an apical PDZ protein. Pulse-chase studies in combination with domain-selective cell surface biotinylation revealed that newly synthesized wt-CFTR and CFTR-⌬TRL were targeted to the apical and basolateral membrane domains in a nonpolarized fashion. Thus, the PDZ interacting domain is not an apical membrane sorting motif. To determine whether the PDZ interacting domain is a membrane retention motif, we measured the half-life of wt-CFTR and CFTR-⌬TRL in the apical and basolateral membranes. Deletion of the PDZ interacting domain reduced the half-life of CFTR in the apical membrane from ϳ24 to ϳ13 h but had no effect on the half-life of CFTR in the basolateral membrane. Thus, the PDZ interacting domain is an apical membrane retention motif. Deletion of the PDZ interacting domain did not affect apical membrane endocytosis of CFTR. By contrast, deleting the PDZ interacting domain decreased apical endocytic recycling of CFTR. Thus, deletion of the PDZ interacting domain reduced the half-life of CFTR in the apical membrane by decreasing CFTR endocytic recycling. Our results identify a new role for PDZ proteins in regulating the endocytic recycling of CFTR in polarized epithelial cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Stable Cell Lines-MDCK cells stably expressing GFP-CFTR fusion proteins were established and maintained in culture at 37°C in MEM complete medium containing penicillin, streptomycin, L-glutamine, fetal bovine serum (10%), and G418 (150 g/ml) as described previously (16,26). Addition of GFP to the N terminus of CFTR had no effect on CFTR localization, trafficking, its function as a Cl Ϫ channel, or its degradation (26,27).
Pulse-Chase and Selective Cell Surface Biotinylation Studies-Pulsechase and selective cell surface biotinylation studies were conducted essentially as described by Lisanti et al. (28) to determine whether the PDZ interacting domain of CFTR is a sorting and/or a membrane retention motif. The day before pulse-chase studies G418 was removed from the cell culture medium, and sodium n-butyrate (5 mM) was added to stimulate CFTR expression as described previously (26). Confluent, polarized MDCK cells grown on Transwell permeable growth supports (24-mm diameter, 0.4-m pore size; Corning Corporation, Corning, NY; number 3412) were washed extensively in phosphate-buffered saline and incubated in complete MEM without methionine and cysteine for 60 min at 37°C. Subsequently, the cells were metabolically labeled with Tran 35 S-Label Reagent (250 Ci/ml: ICN Pharmaceuticals, Inc., Cosa Mesa, CA) in complete MEM without Met and Cys for 30 min at (37°C). The filters were washed at 4°C in complete MEM containing an excess (10 mM) of unlabeled Met and Cys and chased for variable periods of time in complete MEM containing an excess (10 mM) of unlabeled Met and Cys at 37°C.
To measure the half-life of wt-CFTR and CFTR-⌬TRL in cell lysates, the cells were metabolically labeled with Tran 35 S-Label Reagent as described above, cooled to 4°C, and lysed in lysis buffer as described previously (26). CFTR was immunoprecipitated by incubation with either a GFP polyclonal antibody (5 g: CLONTECH, Palo Alto, CA; number 8372-2) or a mixture of monoclonal antibodies M3A7 and L12B4A (29) (2 g each; Upstate Biotechnology, Inc., Waltham, MA) followed by a second incubation with protein A or protein G (as appropriate) conjugated to Sepharose beads (Pierce). Immunoprecipitated CFTR was eluted from the protein A-or G-Sepharose beads by incubation at 95°C for 5 min in SDS sample buffer and centrifuged for 1 min at 14,000 ϫ g. Immunoprecipitated and 35 S-labeled wt-CFTR and CFTR-⌬TRL were separated by 7.5% SDS-PAGE.
To test the hypothesis that the PDZ interacting domain is a sorting motif, the cells were pulse-labeled with Tran 35 S-Label Reagent (Met/ Cys: 250 Ci/ml) for 30 min, chased for 0, 30, 60, 90, or 120 min, and then cooled to 4°C. The proteins in the apical and basolateral membranes were selectively biotinylated (EZ-Link™ Biotin-LC-Hydrazide; Pierce) as described in detail previously (16,26). Subsequently, the cells were lysed, biotinylated proteins were isolated by streptavidin beads, and biotinylated CFTR was immunoprecipitated using a GFP monoclonal antibody (26) or a mixture of M3A7/L12B4 monoclonal antibodies (29). Immunoprecipitated and biotinylated 35 S-labeled wt-CFTR and CFTR-⌬TRL were separated by 7.5% SDS-PAGE.
Detection of 35 S-labeled wt-CFTR and CFTR-⌬TRL was conducted by placing gels on a general purpose storage phosphorus screen (Molecular Dynamics, Sunnyvale, CA) and, 24 -72 h later, detection using the Storm 860 PhosphorImager system (Molecular Dynamics). Measurement of 35 S-labeled wt-CFTR and CFTR-⌬TRL was conducted using a Dell OptiPlex GX1 computer and ImageQuant 5.1 software (Molecular Dynamics).
Transcytosis Assay-Studies were conducted to determine whether apical polarization of CFTR was mediated in part by transcytosis of CFTR from the basolateral to the apical plasma membrane according to a method described in detail previously (30). In brief, basolateral membrane proteins were biotinylated at 4°C using a derivative of biotin (EZ-Link™ Sulfo-NHS-SS-Biotin: Pierce) that can be reduced by GSH. Subsequently, the cells were warmed to 37°C for 0, 60, or 120 min, and the disulfide bonds on Sulfo-NHS-SS-biotinylated proteins in the apical or basolateral membranes were reduced by GSH added to the apical or basolateral solutions, respectively, for 30 min at 4°C. In preliminary studies we demonstrated that GSH only reduces the disulfide bonds of biotinylated proteins in plasma membranes and does not cross membranes or monolayers of polarized MDCK cells. Biotinylated CFTR was analyzed by Western blotting using a GFP monoclonal antibody (26) or a mixture of M3A7/L12B4 monoclonal antibodies (29) and an antimouse horseradish peroxidase antibody using the Western Lightning™ Chemiluminescence Reagent Plus detection system (ECL).
Endocytic Assay-Studies were conducted to determine whether the reduced half-life of CFTR-⌬TRL compared with wt-CFTR in the apical membrane was due to a difference in endocytosis according to a method described in detail previously (26,31). In brief, apical membrane proteins were biotinylated at 4°C using EZ-Link TM Sulfo-NHS-SS-Biotin (Pierce). Subsequently, the cells were warmed to 37°C for 1, 3, 5, or 10 min, and the disulfide bonds on Sulfo-NHS-SS-biotinylated proteins remaining in the apical membrane were reduced by GSH added to the apical solution for a total of 90 min at 4°C. At this point in the protocol, biotinylated proteins reside within the endosomal compartment. Subsequently, the cells were lysed, and the biotinylated proteins were isolated by streptavidin-agarose beads, eluted into SDS sample buffer, and separated by 7.5% SDS-PAGE. Biotinylated CFTR was analyzed by Western blot analysis as described above.
Endocytic Recycling Assay-Studies were conducted to determine whether the reduced half-life of CFTR-⌬TRL compared with wt-CFTR in the apical membrane was due to a difference in endocytic recycling according to a method described in detail previously (32,33). Briefly, apical membranes were biotinylated at 4°C and then warmed to 37°C for 3 min to load endocytic vesicles with biotinylated proteins, including CFTR. Subsequently, the cells were cooled to 4°C, and the disulfide bonds on Sulfo-NHS-SS-biotinylated proteins in the apical membranes were reduced by GSH, as described for the endocytic assay. Subsequently, the cells were either lysed or warmed again to 37°C for 3 or 5 min (to allow internalized, biotinylated CFTR to recycle to the apical membrane). The cells were then cooled again to 4°C, the disulfide bonds on Sulfo-NHS-SS-biotinylated proteins in the apical membranes were reduced with GSH, and the cells were lysed. The biotinylated proteins were isolated by streptavidin-agarose beads, eluted into SDS sample buffer, and separated by 7.5% SDS-PAGE. Biotinylated CFTR was analyzed by Western blot analysis as described above. Endocytic recycling of CFTR was expressed as the difference between the amount of biotinylated CFTR after the first and second warming to 37°C. Biotinylated CFTR was analyzed as described above.
Data Analysis and Statistics-Each experiment was repeated a minimum of three to six times. In each experiment three to six filters were studied at each time point. Calculation of the half-life of CFTR was performed using GraphPad Prism version 3.0a for Macintosh (Graph-Pad Software, San Diego, CA). Statistical analysis of the data was performed using GraphPad Instat version 3.0a for Macintosh (Graph-Pad Software). The means were compared by the unpaired t test. A p value of Ͻ0.05 was considered significant. The data are expressed as the means Ϯ S.E.

The PDZ Interacting Domain of CFTR Is Not an Apical
Membrane Sorting Motif-To test the hypothesis that the PDZ interacting domain (TRL) is a sorting motif, studies were conducted in MDCK cells using a pulse-chase and domain-specific cell surface biotinylation protocol. MDCK cells stably expressing GFP-wt-CFTR or GFP-CFTR-⌬TRL were washed in a Metand Cys-free MEM solution for 60 min, then pulsed with Tran 35 S-labeled Met/Cys for 30 min to label newly synthesized proteins, and chased for 0, 30, 60, 90, and 120 min in MEM containing an excess of unlabeled Cys/Met. Subsequently, apical and basolateral membrane proteins were selectively biotinylated at 4°C, the cells were lysed, and the biotinylated proteins were isolated by streptavidin beads. Biotinylated CFTR was immunoprecipitated using either a GFP monoclonal antibody or a mixture of M3A7/L12B4 monoclonal antibodies.
Similar results were obtained using either the GFP or the M3A7/L12B4 antibodies. 35 S-Labeled and biotinylated CFTR was separated by SDS-PAGE and detected by PhosphorImager analysis (see "Experimental Procedures" for details). If the PDZ interacting domain is an apical membrane-sorting motif, newly synthesized wt-CFTR should be sorted directly from the trans-Golgi network (TGN) to the apical membrane with little or no wt-CFTR appearing in the basolateral membrane. By contrast, CFTR-⌬TRL should be delivered to the apical and basolateral membranes in equal amounts at each time point.
Newly synthesized wt-CFTR appeared in the apical and basolateral membranes at the same rate at 0, 30, 60, and 90 min into the chase period (Fig. 1A). The ratio of wt-CFTR in the apical/basolateral membrane was ϳ1 at all time points between 0 and 90 min into the chase period (Fig. 2). Thereafter (120 min) wt-CFTR accumulated preferentially in the apical plasma membrane. At 120 min into the chase period, the ratio of wt-CFTR in the apical to basolateral membrane was ϳ2 (Figs. 1A and 2). Similar results were obtained in pulse-chase studies on wt-CFTR in Calu-3 cells, a human airway epithelial cell line expressing native wt-CFTR (data will be presented in a separate study). These data suggest that the polarized expression of wt-CFTR to the apical membrane in the steady state does not result from the selective sorting of wt-CFTR from the TGN directly to the apical plasma membrane.
Similar pulse-chase and selective cell surface biotinylation studies were conducted in MDCK cells stably expressing CFTR-⌬TRL, which does not polarize in the steady state to the apical or basolateral membranes (16,18). Newly synthesized CFTR-⌬TRL appeared in the apical and basolateral membranes at the same rate at 0, 30, 60, 90, and 120 min into the chase period (Fig. 1B). The ratio of CFTR-⌬TRL in the apical/ basolateral membrane was ϳ1 at all time points between 0 and 120 min into the chase period (Fig. 2). Taken together, our studies with wt-CFTR and CFTR-⌬TRL indicate that the PDZ interacting domain is not an apical membrane sorting motif.
Thus, the polarized expression of wt-CFTR in the apical membrane of MDCK cells in the steady state does not result from the selective sorting of wt-CFTR from the TGN directly to the apical plasma membrane.
The PDZ Interacting Domain of CFTR Is an Apical Membrane Retention Motif-Additional studies were conducted to test the hypothesis that the PDZ interacting domain is an apical membrane retention motif and that selective retention in the apical membrane of wt-CFTR leads to its apical polarization in the steady state. According to this hypothesis, the half-life of wt-CFTR in the apical membrane should be greater that its half-life in the basolateral plasma membrane. Moreover, deletion of the PDZ interacting domain should reduce the half-life of CFTR in the apical but not in the basolateral membrane. To test this hypothesis, pulse-chase and domain-specific cell surface biotinylation studies were conducted as described above, except that the chase periods were 2, 6, 18, and 24 h.
As illustrated in Figs. 3 and 4, the half-life of wt-CFTR in the apical membrane (24.1 h) was significantly longer than the half-life of wt-CFTR in the basolateral membrane (12.9 h). Deletion of the PDZ interacting domain significantly reduced the half-life of CFTR in the apical membrane from 24.1 to 12.6 h. By contrast, deletion of the PDZ interacting domain had no effect on the half-life of CFTR in the basolateral membrane (12.9 h for wt-CFTR and 11.2 h for CFTR-⌬TRL). Taken together, these data are consistent with the view that the PDZ interacting domain in CFTR is an apical membrane retention motif and that wt-CFTR is selectively retained in the apical membrane via interaction with a PDZ protein(s).
To examine the role of the PDZ interacting domain in the degradation and stability of CFTR in the intracellular compartment(s), pulse-chase studies were conducted, essentially as described above, except that CFTR was immunoprecipitated from cell lysates after biotinylated proteins had been removed by streptavidin isolation. Thus, nonbiotinylated CFTR was immunoprecipitated and separated by SDS-PAGE. The half-life of maturely glycosylated (C band) wt-CFTR was 12.5 h, and the half-life of maturely glycosylated CFTR-⌬TRL was 11.9 h (Fig.  5). These results confirmed previous studies demonstrating that short truncations of the C terminus of CFTR (Ͻ26 amino acids) have no effect on the degradation of the maturely glycosylated (C band) CFTR (34,35).
Apical Membrane Polarization of wt-CFTR Is Not Mediated by Transcytosis from the Basolateral to the Apical Plasma Membrane-The data presented above do not rule out the pos-  (26). The molecular mass of the maturely glycosylated C band of CFTR was ϳ210 kDa (26). As described previously, the maturely glycosylated C band of GFP-tagged CFTR (evident in this as well as subsequent figures) runs as a doublet on Western blots, possibly because of aggregation of CFTR (26). Neither band is sensitive to endoglycosidase H digestion but is digested to the unglycosylated A band by peptide-N-glycosidase F (PN-Gase F) (26). sibility that apical membrane polarization of wt-CFTR results in part from the transcytosis of wt-CFTR from the basolateral to the apical membrane. Transcytosis of wt-CFTR may be a mechanism that, in addition to apical retention, leads to the apical polarization of wt-CFTR. To test the hypothesis that CFTR is transcytosed from the basolateral to the apical plasma membrane, we conducted a transcytosis assay as described previously (30). Plasma membrane proteins were biotinylated at 4°C with EZ-Link™ Sulfo-NHS-SS-Biotin. After reduction by GSH, the biotin is no longer attached to membrane proteins. After biotinylation at 4°C, the cells were warmed to 37°C for 0, 60, or 120 min, a time at which newly synthesized wt-CFTR begins to polarize to the apical membrane ( Figs. 1 and 2). Subsequently, the cells were cooled again to 4°C, and then GSH (or vehicle) was added to the solutions bathing either the apical or basolateral side of the monolayers to reduce the disulfide bond of NHS-SS-Biotin attached to proteins in the plasma membrane on the cis but not the trans side of monolayers. Thus, in experiments in which basolateral membrane proteins were biotinylated, GSH added to the basolateral solution will reduce only the disulfide bond of NHS-SS-Biotin attached to proteins in the basolateral membrane but will not reduce the disulfide bond in NHS-SS-Biotin attached to proteins that were endocytosed to an intracellular compartment or to proteins that were biotinylated in the basolateral membrane and transcytosed to the apical membrane. By contrast, addition of GSH to the apical solution will only reduce the disulfide bond of NHS-SS-Biotin attached to proteins that were biotinylated in the basolateral membrane and transcytosed to the apical membrane and not reduce the disulfide bond in NHS-SS-Biotin attached to proteins that were retained in the basolateral membrane or endocytosed. Subsequent to GSH (30 min) or vehicle treatment at 4°C, the cells were lysed, the biotinylated proteins were isolated with streptavidin beads, and the biotinylated proteins were separated by SDS-PAGE; the proteins were Subsequently, the monolayers were cooled to 4°C, and the apical or basolateral membranes were biotinylated at the times indicated. Biotinylated proteins were pulled down with streptavidin beads, and biotinylated CFTR was immunoprecipitated with a GFP monoclonal antibody or a mixture of monoclonal antibodies M3A7/L12B4 (29). Immunoprecipitated CFTR was separated by SDS-PAGE and detected by PhosphorImager analysis. In this experiment it is evident that the half-life of wt-CFTR in the apical membrane was significantly longer than the half-life of wt-CFTR in the basolateral membrane and than the half-life of CFTR-⌬TRL in the apical and basolateral membranes.

FIG. 4. Summary of studies conducted to determine the halflife of wt-CFTR and CFTR-⌬TRL in the apical and basolateral membranes.
The data are reported as the percentages of wt-CFTR and CFTR-⌬TRL remaining in the apical and basolateral membrane as a function time after the end of the 30-min pulse with Tran 35 S-labeled Met/Cys. Based on half-life of CFTR in the apical (24.1 h) and basolateral membrane (12.9 h), we estimate that, at steady state, the ratio of CFTR in the apical/basolateral membrane should be ϳ2:1. This value is similar to the range of values measured for wt-CFTR previously in MDCK and human airway epithelial cells (i.e. 3:1 to 8:1) (16)). Moreover, the value of 2:1 is similar to the apical/basolateral ratio of a GFP-tagged fusion protein (3:1) recently reported in MDCK cells by Simons and co-workers (37). The data are expressed as the means Ϯ S.E. where the number of experiments was six for wt-CFTR and CFTR-⌬TRL.

FIG. 5. Summary of studies conducted to determine the halflife of the maturely glycosylated (C band) of wt-CFTR and CFTR-⌬TRL in cell lysates (minus the plasma membrane biotinylated CFTR).
In A the data are reported as the percent of wt-CFTR and CFTR-⌬TRL remaining as a function of time after the end of the 30-min pulse with Tran 35 S-labeled Met/Cys (100% value is the amount of CFTR at time 0, which is at the beginning of the chase period). The data are expressed as the means Ϯ S.E. where the number of experiments was six for wt-CFTR and CFTR-⌬TRL. B illustrates a representative experiment examining the amount of wt-CFTR or CFTR-⌬TRL is cell lysates as a function of time (in h) after a 30-min pulse with Tran 35 S-labeled Met/Cys. As determined by pulse-chase analysis, there appears to be two pools of CFTR: one in the apical plasma membrane (ϳ20% of CFTR in the cell) with a half-life of ϳ24 h (Fig. 4) and one in an intracellular compartment(s) (80% of CFTR in the cell) with a half-life of ϳ12.5 h (Fig. 5). The difference in half-life suggests that CFTR in the apical membrane, endocytic trafficking pathway does not enter the degradative pathway as readily as does CFTR in the intracellular compartment(s). Thus, it can be predicted that deletion of the PDZ interacting domain should increase the amount of CFTR in the intracellular compartment(s), and thus the half-life of CFTR-⌬TRL (11.2 h) should be less than half-life of wt-CFTR (12.9 h.) in the intracellular compartment(s). The difference in half-life did not achieve statistical significance, most likely because the endocytic recycling pool of CFTR is relatively small (ϳ20%) and because the decrease in endocytic recycling with deletion of the PDZ interacting domain is also relatively small. transferred to a polyvinylidene difluoride membrane, and biotinylated CFTR was detected using a GFP monoclonal antibody.
The representative experiment illustrated in Fig. 6 demonstrates that transcytosis of wt-CFTR could not be detected after 60 or 120 min. In monolayers in which basolateral membranes were biotinylated, addition of GSH to the solution bathing the basolateral side of the monolayers reduced the amount of wt-CFTR pulled down by streptavidin by more than 95% (compare lanes a-c with lanes d-f in Fig. 6A). Moreover, addition of GSH to the solution bathing the apical side of the monolayers had no effect on the amount of wt-CFTR pulled down by streptavidin (compare lanes a-c with lanes g-i in Fig. 6A). Fig. 6B demonstrates that the total wt-CFTR expression was similar in each set of monolayers. These data suggest that wt-CFTR was not transcytosed from the basolateral to the apical membrane. Thus, apical polarization of wt-CFTR does not involve transcytosis of wt-CFTR from the basolateral to the apical membrane.
The PDZ Interacting Domain Does Not Regulate Apical Membrane Endocytosis of CFTR-To determine the mechanism whereby deletion of the PDZ interacting domain reduced the half-life of CFTR in the apical membrane, we monitored the endocytosis of wt-CFTR and CFTR-⌬TRL as described under "Experimental Procedures." As illustrated in Fig. 7A, we observed a linear increase in the endocytic uptake of wt-CFTR between 0 and 3 min. Thereafter, the amount of CFTR in endocytic vesicles failed to increase because, as shown below, CFTR is rapidly recycled back to the plasma membrane. The amount of CFTR-⌬TRL endocytosed between 0 and 3 min was similar to wt-CFTR. Thus, deletion of the PDZ interacting domain had no effect on the apical membrane endocytosis of CFTR.
The PDZ Interacting Domain of CFTR Is an Apical Membrane Endocytic Recycling Motif-Additional studies were conducted to determine whether deletion of the PDZ interacting domain reduced the half-life of CFTR in the apical membrane by decreasing its endocytic recycling. In preliminary studies we observed that endocytic recycling of CFTR was linear between 0 and 3 min. Thereafter, the amount of CFTR that recycled back to the apical plasma membrane failed to increase because, as shown above, CFTR is rapidly endocytosed. The endocytic recycling of CFTR was dramatically reduced by deletion of the PDZ interacting domain (Fig. 7B). Thus, the PDZ interacting domain of CFTR is an endocytic recycling motif. Taken together, these data reveal that the PDZ interacting domain selectively retains CFTR in the apical membrane by facilitating the endocytic recycling of CFTR. DISCUSSION Our data demonstrate that the PDZ interacting domain of CFTR is an apical membrane, endocytic recycling motif. CFTR polarizes to the apical membrane in MDCK cells because a CFTR-PDZ protein interaction facilitates the endocytic recycling of CFTR to the apical membrane and thereby dramatically and selectively increases the half-life of CFTR in the apical plasma membrane. When the CFTR-PDZ protein interaction is eliminated by deleting the PDZ interacting domain of CFTR, endocytic recycling is less efficient, the half-life of CFTR in the apical membrane is reduced, and CFTR no longer polarizes to the apical plasma membrane. Because deletion of the PDZ interacting domain did not affect the delivery of CFTR from the TGN to the apical or basolateral membranes, our data also demonstrate that the PDZ interacting domain of CFTR is not an apical targeting or sorting signal. Moreover, the PDZ interacting domain is not involved in transcytosis of CFTR from the basolateral to the apical plasma membrane.
Three basic mechanisms are utilized to direct the polarized expression of proteins in the apical membrane of epithelial cells: 1) sorting from the TGN directly to the apical membrane; FIG. 6. Transcytosis assay to determine whether wt-CFTR is transcytosed from the basolateral to the apical plasma membrane. A, basolateral membranes were biotinylated (EZ-Link™ Sulfo-NHS-SS-Biotin) at time 0 at 4°C and then warmed to 37°C for 0, 60, or 120 min as indicated. Subsequently, monolayers were cooled to 4°C, and GSH was added to the apical or basolateral bathing solutions as indicated by a plus sign above the blot or to neither side as indicated by the minus sign above the blot. Subsequently, the cells were lysed, and the biotinylated proteins were pulled down with streptavidin and separated by SDS-PAGE. The biotinylated proteins were transferred to polyvinylidene difluoride membranes, and CFTR was detected by a GFP monoclonal antibody and ECL. The letters below each lane are to enhance the description under "Results." B, CFTR in cell lysates demonstrating that monolayers in each set of experiments expressed similar amounts of CFTR. 2) sorting from the TGN directly to the basolateral membrane and then transcytosis to the apical membrane; and 3) random sorting to the apical and basolateral membranes with selective retention in the apical membrane (2). Most apical membrane proteins expressed in MDCK cells are sorted directly to the apical membrane (2,36,37). Our data indicate that CFTR is randomly sorted to the apical and basolateral membranes and retained in the apical membrane via interaction with a PDZ protein. Like CFTR, the Na ϩ -K ϩ -ATPase is randomly sorted from the TGN to the apical and basolateral membranes in MDCK cells. The Na ϩ -K ϩ -ATPase is selectively retained in the basolateral membrane via interaction with the actin-based cytoskeleton, which retains the Na ϩ -K ϩ -ATPase in a polarized state (38). Apical Na ϩ -K ϩ -ATPase does not interact with actin and is removed from the apical membrane and degraded. Transcytosis of wt-CFTR from the basolateral to the apical membrane does not appear to be a major mechanism whereby wt-CFTR is polarized to the apical plasma membrane in MDCK cells (Fig. 6). Because MDCK cells are a well described model for studying transcytosis, it is reasonable to conclude that our inability to detect significant transcytosis of wt-CFTR from the basolateral to the apical membrane is not due to the inability of MDCK cells to transcytose proteins. However, we cannot completely exclude the possibility that a small amount of wt-CFTR, below the detection limit of our assay, may undergo transcytosis.
Our data are consistent with a growing body of evidence supporting the view that PDZ domains direct the polarized expression of transport proteins in epithelial cells and neurons (11-20, 22, 39). Previously, we demonstrated that deletion of the PDZ interacting domain of CFTR abrogates binding to EBP50 and CAP70, eliminates the polarized expression of CFTR to the apical membrane of kidney (MDCK) and human airway epithelial cells (16HBE14o Ϫ ), and eliminates CFTRmediated transepithelial Cl Ϫ secretion in MDCK cells (16,18,19). The polarized expression of another apical membrane protein, podocalyxin, also requires an interaction between its Cterminal PDZ interacting domain and an apical PDZ protein in MDCK cells (39). Moreover, deletion of the PDZ interacting domain of podocalyxin decreased its stability in the apical membrane. In addition, the polarized expression of the ␥-aminobutyric acid transporter (BGT-1) to the basolateral membrane in MDCK cells requires an interaction between its Cterminal PDZ interacting domain and the basolateral PDZ protein Lin-7 (12). Deletion of the PDZ interacting domain does not effect BGT-1 targeting from the TGN to the basolateral membrane but decreases BGT-1 retention in the basolateral membrane. Similarly, the expression of the inwardly rectifying potassium channel, Kir 2.3, in the basolateral membrane of MDCK cells also involves stabilization in the membrane via interaction with the PDZ protein hLin-7b (40). Thus, PDZ proteins determine the polarized expression of polytopic membrane proteins in MDCK epithelial cells.
It is becoming increasing clear that PDZ proteins regulate endocytosis and endocytic recycling of membrane proteins in epithelial cells (41)(42)(43). The PDZ interacting domains in both the ␤ 2 -adrenergic and the -opioid receptors interact with PDZ domains in EBP50 and regulate endocytic recycling (41,42). Deletion of the PDZ interacting domain eliminates the endocytic recycling of the ␤ 2 -adrenergic receptor without altering the recycling of the transferrin receptor (41). In addition, fragmentation of actin or expression of a dominant negative EBP50 that lacks the ezrin-binding domain also reduces endocytic recycling of the ␤ 2 -adrenergic receptor. Moreover, addition of a PDZ interacting domain to the C terminus of the ␦-opioid receptor, which is normally endocytosed and degraded, causes the receptor to enter the endocytic recycling pathway (43). Another PDZ protein, PSD95, inhibits the endocytosis of Nmethyl-D-aspartate (NMDA) receptors by tethering the receptors to the cytoskeleton (44). Taken together, these observations demonstrate that PDZ proteins regulate both endocytosis and the endocytic recycling of a variety of membrane proteins including CFTR. The mechanism whereby PDZ proteins regulate endocytic trafficking is not clear. However, recent studies suggest that PDZ proteins may provide the molecular link between receptors (i.e. cargo, including CFTR) in endocytic vesicles and myosin VI molecular motors, which bind directly to F-actin and produce unidirectional movement along actin filaments in an ATP-dependent manner. Myosin VI and the PDZ protein SAP97 form a complex with the alpha-amino-3hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptor subunit, GluR1 (45). The myosin VI-SAP97 complex has been implicated in the endocytic trafficking of GluR1 (45). In addition, the PDZ protein GIPC binds both to the C terminus of myosin VI and to the glucose transporter Glut-1 (46). By linking Glut-1-containing vesicles to myosin VI and actin filaments, GAIP-interacting protein (GIPC) may be the molecular link that determines cargo selection for Glut-1 vesicles to enter the endocytic trafficking pathway. Thus, PDZ proteins may be a molecular link between cargo in vesicles and myosin motors, which bind to F-actin and produce unidirectional vesicular movement along actin filaments.
Because there are PDZ proteins in the apical and basolateral membrane of MDCK cells, it is reasonable to ask why CFTR is selectively retained in the apical membrane. There are at least four explanations that may not be mutually exclusive. First, EBP50, E3KARP and CAP70, PDZ proteins that interact with CFTR, are abundantly expressed in the apical, but not in the basolateral membrane of Calu-3, T84 (47-49) and MDCK cells. 2 Second, CFTR may bind to apical PDZ proteins with a higher affinity than to basolateral PDZ proteins. Third, although the PDZ interacting domains of CFTR (DTRL), an apical protein, and BGT-1 (ETHL), a basolateral protein, are similar, it is important to note that the C-terminal eight amino acids, which differ in CFTR and BGT-1, can determine the specificity of PDZ protein interactions (21). Finally, our data indicate that PDZ proteins in the apical membrane that interact with CFTR regulate endocytic recycling, whereas PDZ proteins that may interact with CFTR in the basolateral membrane may not be involved in endocytic trafficking. Thus, selective retention of CFTR in the apical membrane of MDCK cells may result from a number of factors including the polarization of PDZ proteins, the binding affinity of PDZ proteins to CFTR, the amino acids upstream of the PDZ interacting domain, and the function of the PDZ proteins that interact with CFTR. It will be interesting to identify which of the aforementioned possibilities are responsible for the polarized expression of CFTR in the apical membrane.
Protein targeting and sorting motifs are differentially interpreted by various cell types (50). For example, the ␤ subunit of the H ϩ -K ϩ -ATPase is expressed in the apical membrane of LLC-PK1 cells and in the basolateral membrane of MDCK cells (51). Aquaporin 2 is expressed in the apical and basolateral membrane of MDCK cells but only in the apical membrane of LLC-PK1 cells (52). These observations demonstrate that it is important to study sorting and retention motifs within the context of the cellular environment in which the protein is normally expressed. Whereas we observed that CFTR is randomly sorted from the TGN to the apical and basolateral mem-branes in MDCK cells, we found that CFTR is sorted directly from the TGN to the apical membrane in LLC-PK1 cells, a cell line derived from the renal proximal tubule. 2 These data are consistent with the recent observation that LLC-PK1 cells lack the 1B subunit of the AP1 complex, a protein that is required for proteins to traffic from the TGN to the basolateral membrane (53). Thus, it is not surprising that addition of an epitope tag to the C terminus of CFTR (which should block interaction of CFTR with PDZ proteins) did not affect the apical polarization of CFTR in LLC-PK1 cells (54). Because CFTR is also expressed in airway epithelial cells, we conducted pulse-chase studies in Calu-3 cells, a well characterized human airway epithelial cell line that express CFTR in the apical plasma membrane (47,55,56). Similar to our data in MDCK cells, wt-CFTR was sorted in a nonpolarized manner from the TGN to the apical and basolateral membrane (ratio apical/basolateral ϭ 1.0 from 0 to 90 min after the pulse; data to be presented in a separate manuscript). Our results in MDCK and Calu-3 cells suggest a common mechanism for CFTR trafficking in polarized kidney (MDCK) and airway epithelial cells (Calu-3) but not in LLC-PK1 cells.
In conclusion, the data in this manuscript demonstrate that the PDZ interacting domain of CFTR is an apical membrane, endocytic recycling motif and that polarization of CFTR to the apical membrane in kidney (MDCK) epithelial cells is mediated by interaction with PDZ protein(s) that facilitate CFTR endocytic recycling.