The Role of the C Terminus and Na (cid:1) /H (cid:1) Exchanger Regulatory Factor in the Functional Expression of Cystic Fibrosis Transmembrane Conductance Regulator in Nonpolarized Cells and Epithelia*

The conserved C-terminal peptide motif ( 1476 DTRL) of the cystic fibrosis transmembrane conductance regulator (CFTR) ensures high affinity binding to different PSD-95/Disc-large/zonula occludens-1 (PDZ) domain-containing molecules, including the Na (cid:1) /H (cid:1) exchanger regulatory factor (NHERF)/ezrin-radixin-moesin-bind-ing phosphoprotein of 50 kDa. The physiological rele-vance of NHERF binding to CFTR is not fully under-stood. pH 7.2. The holding potential was (cid:5) 60 mV. For analysis of current-voltage relationships, the potential was stepped between (cid:5) 90 and (cid:3) 90 mV in 15-mV increments. Voltage steps were applied for 50–300 ms at 800-ms intervals. To monitor the time course of CFTR activation, whole cell current values were measured at (cid:3) 75 mV following the addition of PKA mixture in every 20 s until the current reached a steady state. Currents were recorded using AXO-PATCH at room temperature (22–24 °C) and analyzed by the CLAMP- FIT software. Currents were normalized per unit capacitance to account for variations in cell size.

the cAMP-dependent protein kinase-activated chloride channel conductance in polarized pancreatic epithelia. On the other hand, electrophysiological studies demonstrated that NHERF is able to stimulate the cAMP-dependent protein kinase-phosphorylated CFTR channel activity in intact cells. These results help to reconcile the discordant genotype-phenotype relationship in individuals with C-terminal truncations and indicate that apical localization of CFTR involves sorting signals other than the C-terminal 26 amino acid residues and the PDZ-binding motif in differentiated epithelia.
The cystic fibrosis transmembrane conductance regulator (CFTR), 1 a member of the adenine nucleotide binding cassette family, is an ATP-and cAMP-dependent protein kinase (PKA)regulated chloride channel (1,2). CFTR consists of two homologous halves, each comprising a transmembrane (TM) domain and a nucleotide binding domain, connected by the highly charged regulatory domain (3). Based on 8-azido-ATP-binding/ hydrolysis assays, the C-terminal boundary of nucleotide binding domain 2 seems to extend to amino acid 1420 (4), implying that the C-terminal tail encompasses approximately the last 60 amino acid residues (1420 -1480) of CFTR.
Compelling evidence suggests that the C-terminal tail is engaged in a variety of intra-and intermolecular interactions. The C terminus contains a tyrosine-based ( 1424 YDSI) and a dileucine-based ( 1430 LL) internalization motif, ensuring clathrin binding via the AP-2 adaptor protein and constitutive internalization via a clathrin-dependent pathway (5)(6)(7). The ␣ 1 catalytic subunit of the adenosine 5Ј-monophosphate-activated protein kinase has been recognized as a regulator of channel activity (8). The C-terminal tail is also involved in folding and conformational stabilization of CFTR. Deletion of a hydrophobic patch ( 1413 FLVI) ablates posttranslational folding (9), whereas truncating 70 (⌬70 CFTR) or more amino acid residues destabilizes the mature, complex-glycosylated CFTR (10,11). Finally, high affinity association of multivalent PDZ domain-containing molecules with the conserved C-terminal peptide ( 1476 DTRL) has been demonstrated (12)(13)(14).
PDZ domain-containing proteins bind to short C-terminal peptides of membrane receptors, ion channels, and signaling complexes and are grouped into two classes based on the target sequence. Class I PDZ domains recognize the sequence motif (S/T)X(V/I/L)T, identified also at the C terminus of CFTR ( 1476 DTRL), whereas Class II domains recognize the motif (F/ Y)X(F/V/A) (15,16). The role of the PDZ domain-containing CFTR-associated ligand, the 70-kDa CFTR-associated protein, and the Na ϩ /H ϩ exchanger regulatory factor (NHERF)/ezrinradixin-moesin-binding phosphoprotein of 50 kDa in the regulation of CFTR has been extensively studied (17). CFTR-associated ligand down-regulates CFTR expression by impeding its biosynthetic processing at the Golgi compartment (14). 70-kDa CFTR-associated protein is co-localized with CFTR in native epithelia and was proposed to stimulate channel activity by promoting CFTR oligomerization in excised patches (13).
NHERF may impact on the CFTR-dependent chloride conductance of the apical membrane by several means. Channel activation is facilitated by a conformational change of CFTR induced by binding to the multivalent NHERF and 70-kDa CFTR-associated protein, as found in excised inside-out patches (13,30). Furthermore, NHERF may promote the phosphorylation of CFTR by tethering PKA and protein kinase C⑀ to the vicinity of the channel (28,31).
NHERF and the C terminus tail were proposed to have a central role in the apical expression of CFTR (32)(33)(34)(35). This hypothesis was supported by the observation that deletion of the last 26 amino acids of CFTR provoked the accumulation of the channel at the lateral membrane in MDCK II and bronchial epithelia (32,34) and caused almost a complete (Ͼ97%) loss of the PKA-activated CFTR function (34). These observations, however, contradict the clinical phenotype of the ⌬26 CFTR mutation. Individuals harboring the ⌬26 CFTR on one allele and a nonsense mutation on the second allele have only moderately elevated sweat chloride concentration without obvious pancreatic and pulmonary phenotype (36), implying the involvement of other factors in addition to NHERF and the C terminus in the apical targeting of CFTR.
In an attempt to gain more insight into the role of the C terminus in CFTR physiology and to provide an explanation for the contrasting clinical and cellular phenotype, biochemical and functional aspects of the truncated CFTR were examined. Whereas association of NHERF with the C-terminal tail of CFTR could be confirmed in vitro and in vivo, disruption of the complex had no discernible effect on the apical localization of CFTR in epithelia derived from the trachea, pancreatic duct, intestine, as well as the distal tubule of the kidney. No significant alteration in the apical halide conductance of pancreatic duct cells expressing the truncated CFTR was observed, providing a plausible explanation for the mild clinical phenotype of the truncated CFTR at the cellular level.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-Carboxyl-terminal fragments of the CFTR cDNA, comprising the deletion of the last 6, 40, 50, and 60 amino acid residues (designated as ⌬6, ⌬40, ⌬50, and ⌬60 CFTR) were generated by PCR (antisense primers, 5Ј-C CCC CCC TCG AGC TAC TCT TCT   TCT GTC TCC-3Ј, 5Ј-AA AGG GCC CTA GGC TTG CCG GAA GAG  GC-3Ј, 5Ј-AA AGG GCC CTA CAG TTT CTG GAT GGA ATC G-3Ј, 5Ј-AA  AGG GCC CTA TTT GTT CTC TTC TAT GAC C-3Ј, and 5Ј-TGC AAG  AAT GGC CAA CTC TCG CC-3Ј, respectively, and a sense primer, 5Ј-A  AAG GGC CCG CTA GCA TTC CAG CAT TGC TTC-3Ј) and inserted into the PmlI/ApaI sites of the pCDNA3 plasmid, encoding the N terminus HA-tagged CFTR as described previously (11). To stably transfect BHK cells, mutants were subcloned into pNUT-CFTR expression plasmid. All mutants were sequence-verified. To express CFTR in PANC-1 epithelia, WT and ⌬6 CFTR were subcloned into the pMT/EP episomal expression cassette, provided by Dr. J. Ilan (37). CFTR-GFP fusion was constructed by shuttling the HpaI/SalI fragment of the pEGFP-N3 vector (BD Biosciences) into pNUT-CFTR-HAC t after eliminating the hemagglutinin (HA) tag. To confer NHERF binding capacity, short flexible linker and the last six amino acids of CFTR (GGG-VQDTRL) were fused by PCR mutagenesis to the C terminus of the CFTR-GFP, resulting in the pNUT-CFTR-GFPDTRL expression cassette. The influenza HA epitope (YPYDVPDYA) was either attached to the last amino acid of CFTR via linker sequence containing an XhoI restriction site (CFTR-HAC t ) or inserted into the C-terminal tail after Phe 1450 (CFTR-HAC i ) by replacing amino acid residues between 1451 and 1460. Insertion of three tandem HA epitopes in the fourth extracellular loop of CFTR will be described elsewhere. GST-Ct61 expression plasmid was constructed by subcloning the cDNA of the last 61 amino acids of CFTR into the XhoI/EcoRI sites of pGEX-4T3. The pGEX-GST-Ct61⌬5 was constructed by the same approach without the last five amino acids of the C-tail. Mammalian expression constructs, containing the HA-tagged NHERF, D1, and D2 constructs were kindly provided by Drs. Robert Lefkowitz (Howard Hughes Medical Institute, Duke University, Durham, NC) and Randy Hall (Emory University, Atlanta, GA) and have been described (18,38).
Cell Lines and Transfection-BHK-21 (CCL 10; American Type Culture Collection (ATCC)), COS-1 (CRL-1650; ATTC), MDCK II (CCL-34; ATTC), and CaCo-2 (HTB-37; ATTC) cells were cultured as described previously (10). Pancreatic duct cells PANC-1 (CRL-1469; ATCC) were grown in human tracheal epithelia (9HTE160 Ϫ , designated as HTE), a generous gift of Dr. Dieter Gruenert (Human Molecular Genetics Unit, Department of Medicine, University of Vermont, Burlington, VT), were cultured in Eagle's minimal essential medium and 10% fetal calf serum. BHK cells, expressing WT or mutant CFTR, were generated by the calcium phosphate transfection method, and clones were selected in methotrexate as described (11). BHK cells expressing CFTR were also transfected with the pBK vector encoding the HA-tagged NHERF, D1, and D2 domain by the calcium phosphate method. Stable cell lines were selected in 500 M methotrexate and 500 g/ml G418 (Invitrogen). PANC-1 cells, transfected with the pMT/EP expression cassette encoding the WT or ⌬6 CFTR, were selected in the presence of hygromycin (0.2 mg/ml). COS-1 cells, maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, were transiently transfected by FuGene (Roche Applied Science) or LipofectAMINE (Invitrogen) at 60% confluence and analyzed after 48 h. MDCK II cells were transfected by FuGene (Roche Applied Science) or Effectene (Qiagen) at 60 -70% confluence, trypsinized after 24 h, and seeded on 12-or 25-mm Transwell polycarbonate filters (0.4-m pore size) at 80 -90% confluence. Sweat duct cells, kindly provided by Dr. Paul Quinton, were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and growth factors. To ensure the development of differentiated monolayers, experiments were performed after culturing the epithelia for 3-5 days at confluence. Differentiation of the epithelia was verified by the accumulation of occludin and zonula occludens-1 (ZO-1) at tight junctions, by transepithelial electrical resistance measurement and polarized release of iodide (39). Domain-specific fluorescence labeling of filter-grown MDCK II by rhodamine-conjugated wheat germ agglutinin also ruled out the presence of paracellular leak.
Retroviral Infection-Retroviral expression plasmid was constructed by shuttling the cDNA encoding the CFTR-EGFP into the NotI site of the retroviral expression vector pFBneo (Stratagene). The pFB-CFTR-EGFPDTRL was constructed by overlapping PCR mutagenesis inserting the coding sequence of amino acids QVDTRL. Virus stock was generated by transient cotransfection of the CFTR expression plasmid together with the pVpack GP and pVpack vesicular stomatitus virus glycoprotein (Stratagene) packaging plasmids into the HEK 293T cell, using the FuGene transfecting reagent. Viral supernatant was harvested 24 h later. MDCK II, CaCo-2, and HTE epithelia were infected at 50% confluence by incubating the cells in 2 ml of viral supernatant supplemented with 8 g/ml polybrene for 12 h and then in normal medium for 48 h. The infected cells were seeded on semipermeable filter and examined by fluorescence microscopy after 3-14 days.
Isolation of Recombinant Proteins-GST and GST fusion protein containing the C-terminal tail of CFTR were expressed in E. coli (HB101) upon induction with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C. Cells were lysed by sonication (model XL 2015 Heat Systems; Ultrasonics Inc.) in 150 mM NaCl, 20 mM HEPES, 10% glycerol, and 0.1 mM EDTA, supplemented with protease inhibitors (0.2 M leupeptin, 0.2 M pepstatin, and 0.5 mM phenylmethylsulfonyl fluoride) and 1 mM DL-dithiothreitol. The lysate was cleared by centrifugation (10,000 ϫ g, 30 min) and applied to a glutathione Sepharose 4B (Amersham Biosciences) column. After washings, GST and GST fusion proteins were dissociated by the elution buffer (50 mM reduced glutathione, 150 mM NaCl, 50 mM Tris-HCl, and 0,1 mM EDTA, pH 8.0) and dialyzed against PBS. For pull-down assays, fusion proteins were immobilized on glutathione-Sepharose 4B beads (1 h at 4°C), washed with PBS, and incubated for an additional 2 h in the presence of the indicated protein extract. Beads were washed three times with PBS containing 0.1% Triton X-100 and eluted by 2ϫ Laemmli sample buffer.
Immunofluorescence Microscopy-CFTR expression in COS-1 cells was visualized by the mouse monoclonal anti-HA Ab (Covance Research Products 16B12, 1:1000 dilution) and fluorescein-conjugated goat antimouse secondary Ab as described (40). Fluorescence staining of MDCK II cells grown on polycarbonate filters was performed after fixing (3% paraformaldehyde in PBS for 10 min), permeabilizing (0.2% Triton X-100 in PBS for 5 min), and incubating the cells for 10 min in 100 mM glycine in PBS at room temperatures. Filters were cut out, blocked in 1% bovine serum albumin for 1 h and incubated in primary and secondary antibodies for 2 h at 37°C. Primary antibodies were as follows: mouse monoclonal anti-HA (Covance Research Products 16B12) at 1:250 dilution, rat polyclonal anti-ZO-1 (Chemicon, Inc.) at 1:50 dilution, mouse monoclonal anti-E-cadherin (kindly provided by Dr. W. Gallin, University of Alberta, Edmonton) at 1:250 dilution, and mouse monoclonal anti-gp135 (kindly provided by Dr. S. Hansen, Boston Biomedical Research Institute) at 1:250 dilution. Rhodamine-and fluorescein-conjugated goat anti-mouse and anti-rat secondary antibodies were applied at 1:500 dilution (Jackson ImmunoResearch Laboratories). Fluorescence micrographs were obtained by a Zeiss LSM 510 laser fluorescence confocal microscope as described (40). Confocal images were scanned using a 0.6 -0.9-m pinhole and the multitrack scanning mode. No spillover of the fluorescein and rhodamine signals could be detected. X-Z projections were reconstructed from the horizontal optical sections.
Metabolic Labeling and Immunoprecipitation-The protocol of pulsechase labeling was similar to our previously published ones (10,11). Briefly, transfected BHK cells were depleted in cysteine-and methionine-free ␣-minimal essential medium, pulse-labeled in the presence of 0.1-0.2 mCi/ml [ 35 S]methionine and [ 35 S]cysteine (Amersham Biosciences) for 20 min, and then chased for the indicated time at 37°C in complete medium. Metabolically labeled CFTR was isolated by immunoprecipitation, separated by SDS-PAGE, and visualized by fluorography. Radioactivity associated with CFTR was measured by phosphorimaging analysis as described (10).
Iodide Efflux Measurements and Electrophysiology-Iodide efflux was performed essentially using the protocol described for BHK and MDCK I monolayers (39,42). Whole cell currents were recorded by the method of O. P. Hamill et al. (43) as detailed previously (44). The pipette filling solution contained 110 mM sodium gluconate, 20 mM NaCl, 8 mM MgCl 2 , 5 mM EGTA, 10 mM glucose, 2 mM ATP, 10 mM HEPES, pH 7.2. The bath solution was 137 mM NaCl, 3 mM MgCl 2 , 1 mM CaCl 2 , 10 mM glucose, 10 mM HEPES, pH 7.2. The holding potential was Ϫ60 mV. For analysis of current-voltage relationships, the potential was stepped between Ϫ90 and ϩ90 mV in 15-mV increments. Voltage steps were applied for 50 -300 ms at 800-ms intervals. To monitor the time course of CFTR activation, whole cell current values were measured at ϩ75 mV following the addition of PKA mixture in every 20 s until the current reached a steady state. Currents were recorded using AXO-PATCH at room temperature (22-24°C) and analyzed by the CLAMP-FIT software. Currents were normalized per unit capacitance to account for variations in cell size.

CFTR-NHERF Interaction Is Dispensable for the Stability of the Complex-glycosylated CFTR in Nonpolarized Cell and Ep-
ithelia-To evaluate the impact of selective loss of the NHERFbinding motif as well as other parts of the C-terminal tail, the effect of progressively larger deletions, encompassing the last 6 -70 amino acid residues, was determined in transiently transfected COS-1 cells. Immunodetection of CFTR was facilitated by attaching the HA-epitope to the N terminus (see Fig.  1). We could not observe any qualitative difference in the membrane targeting of CFTR lacking the last 6, 26, 40, or 50 (⌬6, ⌬26, ⌬40 or ⌬50 CFTR) amino acids relative to WT CFTR ( Fig.  2A). In contrast, the staining intensity of ⌬60 and ⌬70 CFTR expressors appeared to be attenuated ( Fig. 2A). Quantitative immunoblot analysis confirmed the immunostaining results and showed comparable expression levels of the complex-glycosylated WT, ⌬6, ⌬26, ⌬40, and ⌬50 CFTR, whereas a significant reduction in abundance of ⌬60 and ⌬70 CFTR was noticed (Fig. 2B). Similar data were obtained in BHK cells stably expressing the truncated CFTR variants (data not shown).
The reported metabolic destabilization of the ⌬3 and ⌬26 CFTR in MDCK II cells (34,35), contrasting with the WT-like expression level of ⌬6, ⌬26, ⌬40, and ⌬50 in BHK and COS-1 cells, has several plausible explanations. First, NHERF binding may have a dual role in nonpolarized cells. NHERF may participate not only in the stabilization of the complex-glycosylated CFTR, but also in tonic inhibition of its biosynthetic processing at the Golgi compartment as proposed previously for CFTR-associated ligand (14). Thus, disruption of NHERF binding could simultaneously facilitate processing and destabilize the complex-glycosylated form. This scenario was tested by measuring the turnover of the complex-glycosylated form using the pulse-chase technique in BHK cells, stably expressing CFTR. Deletion of the last 26, 40, or 50 amino acids had no apparent effect on the slow turnover of complex-glycosylated CFTR (t1 ⁄2 Ϸ 12-14 h, Fig. 2C and data not shown). The shortest destabilizing truncation (⌬60 CFTR) accelerated the turnover rate of CFTR to a similar extent than reported for the ⌬70, ⌬82, and ⌬98 CFTR mutants (11). This was reflected by a Ͼ6-fold shorter half-life (t1 ⁄2 Ϸ 2 h) of the ⌬60 CFTR compared with the WT form (t1 ⁄2 Ϸ 14 h). Since truncating 6 -70 amino acids does not alter the translational rate and maturation efficiency of CFTR (data not shown; see Ref. 11), these data imply that neither the DTRL motif nor the last 50 amino acids are required to maintain the metabolic stability of the complex-glycosylated CFTR in BHK and COS-1 cells.
The lack of stabilizing effect of NHERF in BHK cells could be attributed to inefficient complex formation between NHERF and CFTR due to cell-specific variations and/or the presence of competing binding partners (45). Pull-down assays were performed with a recombinant GST-C-terminal tail fusion protein, containing the last 61 amino acids of CFTR (GST-Ct 61 ) to assess NHERF-CFTR interaction in vitro. Immunoblot analysis shows that the C-terminal tail associates with HA-NHERF, expressed in BHK cells (Fig. 3A). Deletion of the PDZ-binding motif abolished complex formation (Fig. 3A). Co-immunoprecipitation of HA-NHERF with an EGFP fusion protein containing the last 61 amino acids of CFTR (EGFP-Ct 61 ), but not with EGFP, confirmed the association in BHK cells (Fig. 3B).
To further evaluate the interaction between CFTR and NHERF in vivo, we generated BHK cell lines constitutively co-expressing the WT or the ⌬26 CFTR in the presence of the HA-tagged NHERF, PDZ-1 (D1), or the PDZ-2 (D2) domain. Expression of NHERF, D1, and D2 was verified by immuno-blotting (Fig. 3C, lower panel). Detection of CFTR in immunopreciptates of NHERF, D1, or D2 suggests that not only NHERF but D1 and D2 as well can associate with CFTR, as observed previously (Fig. 3C, upper panel) (30). Importantly, ⌬26 CFTR was unable to co-precipitate NHERF, underscoring the role of the C terminus in NHERF binding (Fig. 3D). Together these data demonstrate that a CFTR-NHERF interaction exists in BHK cells and that disrupting this interaction has no effect on the metabolic stability of CFTR.
To test the hypothesis that complex formation between NHERF, CFTR, and the cytoskeletal network is required to target/retain CFTR at the apical membrane in polarized epithelia, the metabolic stability, localization, and channel function of truncated and WT CFTR were determined next. Since the degree of correlation between CFTR genotype and cystic fibrosis phenotype is highest for the exocrine pancreatic status and lowest for the pulmonary disease (for recent reviews, see Refs. 46 -48), the NHERF-dependent targeting of CFTR was examined in pancreatic duct epithelia. To this end, we selected a human adenocarcinoma cell line derived from pancreatic duct (PANC-1), which lacks endogenous CFTR and forms highly polarized monolayers (49). PANC-1 cells were transfected with WT and ⌬6 CFTR, and more than 50 drug-resistant clones were pooled for subsequent studies. The disappearance kinetics of the complex-glycosylated forms was monitored after inhibition of protein synthesis by cycloheximide (Fig. 4A). According to the densitometric analysis of immunoblots, the half-lives of the complex-glycosylated WT and ⌬6 CFTR were similar (t1 ⁄2 Ϸ 16 h), suggesting that the metabolic stability of the complexglycosylated CFTR is independent of NHERF binding not only in nonpolarized cells but in PANC-1 epithelia as well (Fig. 4B).
Considering that the complex-glycosylated form stability reflects the overall characteristic of post-Golgi, cell surface, and endosomal CFTR pools, and NHERF may facilitate recycling of CFTR as observed for G-protein-coupled receptors (24,26,35,50), next we evaluated the effect of C-terminal truncation on the apical localization and channel function of CFTR.
Apical Localization of Truncated CFTR Is Not Influenced by the Cell Culture Model and Epitope Tag-Polarized expression of CFTR was evaluated by indirect immunostaining, using laser confocal fluorescence microscopy (LCFM) on differentiated monolayers of PANC-1 epithelia. Polarization of PANC-1 epithelia was also verified by occludin immunostaining, one of the constituents of tight junctions (Fig. 4C). Serial horizontal optical sections, as well as x-z image reconstruction demonstrated that both the WT and the ⌬6 CFTR were confined predominantly to the apical domain of PANC-1 epithelia, demarcated by occludin, and were virtually absent below the level of the tight junctions (Fig. 4, C and D).
The functional consequences of the truncation were determined by measuring the halide conductance of the apical plasma membrane by an iodide efflux assay (39). The iodide efflux measurement was performed on differentiated PANC-1 monolayers. No significant difference could be detected in the PKA-stimulated iodide release from epithelia expressing WT or ⌬6 CFTR (Fig. 4E). This observation suggests that the deletion of the DTRL motif has no significant effect on functional channel expression at the apical membrane.
Our results contrast with the partial or complete redistribution of truncated GFP-CFTR to the lateral membrane upon deletion of the last 4 or 26 amino acid residues, respectively, reported by Moyer et al. (32) in MDCK II and bronchial epithelial cell. Since polarized targeting may be influenced by the epitope tag used as well as the cell culture model (51,52), the impact of these factors was further investigated.
The localization of truncated CFTR was analyzed in differentiated MDCK II monolayers, grown on a semipermeable support. Polarized monolayers transiently expressing WT and truncated CFTR were immunostained for CFTR and ZO-1. ZO-1 is a tight junctional membrane protein, delineating the boundary of the apical and lateral membrane domains (53). Horizontal optical images, sectioning the cell from the apical

FIG. 3. Interaction of NHERF with the C terminus tail of CFTR in vitro and in vivo.
A, NHERF binds to the C-terminal tail of CFTR in vitro. GST fusion proteins purified from E. coli, containing the C-terminal 61 amino acids of CFTR (Ct 61 ) or lacking its last 5 amino acid residues (Ct 61-⌬5 ), were adsorbed to glutathione-Sepharose beads and incubated with cell lysates of BHK cells, expressing HA-NHERF. Eluted polypeptides were probed by immunoblotting, using monoclonal anti-HA Ab. GST-Ct 61 but not the GST-Ct 61-⌬5 fusion protein is bound to HA-NHERF. B, association of NHERF with the C-terminal tail of CFTR in BHK cells. BHK cells, stably expressing EGFP or EGFP-Ct 61 fusion protein were overtransfected with HA-NHERF. NHERF was isolated by immunoprecipitation (IP), using anti-HA Ab. The immunoprecipitate (upper panel) and the cell lysate (lower panel) were probed with polyclonal anti-EGFP and anti-HA Ab, respectively. C, CFTR binds to NHERF, NHERF-PDZ1 (D1), and NHERF-PDZ2 (D2). NHERF or one of its PDZ domains was immunoisolated from cells, co-expressing WT CFTR and NHERF, D1, or D2 (indicated by an arrowhead). The immunoprecipitate (upper panel) and cell lysate (lower panel) were probed with rabbit polyclonal anti-CFTR and anti-HA Ab, respectively. D, C-terminal truncation compromises the association of CFTR with NHERF. BHK cells expressing constitutively the indicated constructs were lysed and immunoprecipitated with M3A7 and L12B4 anti-CFTR antibodies. The precipitate was probed with polyclonal anti-HA Ab (upper panel). Expression of CFTR and NHERF variants were confirmed by immunoblotting the lysate using anti-HA and the M3A7 anti-CFTR Ab (middle and lower panels). W, Western blot. pole to the midline region, show that the ⌬6 and ⌬26 CFTR are found predominantly at or above the ZO-1 staining (Fig. 5A). No significant difference could be resolved between the WT and mutant distribution, implying that the truncations have little effect on the apical expression pattern of CFTR, which was confirmed by the x-z projections (Fig. 5B). Although a minor fraction of both the WT and truncated CFTR is associated with the lateral membrane and intracellular compartments (Fig.  5B), this is not due to incomplete differentiation of MDCK II cells or to the saturation of the sorting machinery. The polarized distribution of gp135 and E-cadherin, specific markers of the apical and lateral membrane domains, was preserved in transfected cells relative to ZO-1 localization (Fig. 5, C-E). Furthermore, inserting the HA epitope into the fourth extra-cellular loop had no effect on the polarized distribution of the WT and ⌬26 CFTR, indicating that the N-terminal HA tag did not interfere with the subcellular localization of CFTR (data not shown). These observations, collectively, suggest that the apical localization of CFTR is independent of NHERF binding not only in PANC-1 but also in MDCK II epithelia.
Masking the PDZ-binding Motif Does Not Interfere with Apical Localization of CFTR in Tracheal, Intestinal, and Kidney Epithelia-As a complementary approach to the deletional mutagenesis, NHERF interaction was inhibited by masking the PDZ-binding motif of CFTR with the HA or the EGFP tag, which were attached to the C terminus (CFTR-HAC t and CFTR-GFP, respectively; see Fig. 1). Complex formation between CFTR and NHERF was assessed in BHK cells co-ex- FIG. 4. Stability, polarized expression, and function of WT and ⌬6 CFTR in PANC-1 epithelia. Biochemical and functional studies were performed on polarized PANC-1 epithelia, cultured at least for 3 days postconfluence. Translation was stimulated with 50 M Zn 2ϩ for 24 h prior to the experiment. A, to monitor the turnover of CFTR following zinc induction, protein synthesis was inhibited by cycloheximide for the indicated time. Equal amounts of cell lysate were separated by SDS-PAGE and immunoblotted with anti-HA and anti-occludin antibody. CFTR is expressed at a very low level in the absence of Zn 2ϩ . B, the disappearance rate of complex-glycosylated WT and mutant CFTR was determined by densitometric analysis of immunoblots, shown in A. Data are means Ϯ S.E. (n ϭ 3) and expressed as a percentage of the initial amount of CFTR. C, WT and ⌬6 CFTR are predominantly confined to the apical domain in PANC-1 epithelia. The tight junction marker occludin (green) and CFTR (red) were immunostained with polyclonal anti-occluding and monoclonal anti-HA Ab, respectively. The approximate location of horizontal optical slices, obtained by LCFM from the apical toward the basal pole, is shown in the inset. D, x-z projections obtained from horizontal optical sections, shown in C. Immunoreactive CFTR could not be detected below the tight junction level (see also C). E, halide permeability of the apical membrane of PANC-1 epithelia. The cAMP-activated halide conductance of the apical membrane of PANC-1 monolayers was measured by the iodide efflux assay in cells expressing WT and ⌬6 CFTR. Iodide release was stimulated by the addition of the PKA-agonist mixture (10 M forskolin, 0.5 mM CPT-cAMP, and 0.2 mM isobutylmethylxanthine) at time 0.
pressing CFTR-HAC t or CFTR-GFP with HA-NHERF. The C-terminal HA and EGFP tag prevented coimmunoprecipitation of NHERF with WT CFTR (Fig. 6, lanes 3, 5, and 7) despite similar level of NHERF expression (Fig. 6, lower panel). Disruption of NHERF binding cannot be attributed to misfolding of tagged CFTRs, since the steady-state expression level, biosynthetic processing, stability, and PKA-stimulated whole cell current densities of the native and tagged CFTRs were comparable (Fig. 7, A-F; data not shown) (11). Importantly, attachment of a second DTRL motif to the C terminus of the CFTR-GFP (CFTR-GFPDTRL) or insertion of the HA epitope upstream of the C terminus (CFTR-HAC i ; see Fig. 1) restored NHERF binding to CFTR (Fig. 6, lanes 4 and 8), confirming the notion that the DTRL-binding motif has to be at the C terminus to mediate NHERF binding (11). These fusion proteins provided additional tools to test the significance of NHERF binding in the polarized expression of CFTR.
Following retroviral infection, the distribution of CFTR-GFP was determined by LCFM in the HTE and intestinal (CaCo-2) epithelia grown on permeable filter supports. CFTR-GFP accumulated virtually exclusively at the apical domain of HTE and CaCo-2 epithelia, indicated by the nonoverlapping staining of E-cadherin and CFTR on the x-z images (Fig. 8, A and B). Comparable apical targeting could be observed for CFTR-GFP and the CFTR-GFPDTRL in MDCK II cells as well, regardless of whether the cells were transduced by retroviral infection or lipid-mediated gene transfer (Fig. 8, C-F). Finally, apical localization of CFTR-HAC t could also be established in MDCK II cells, similarly to CFTR-HAC i (Fig. 8, G and H). These observations are fully consistent with the results obtained by the deletional analysis and suggest that complex formation with NHERF is not an absolute requirement for the apical localization of CFTR in a variety of clinically relevant epithelia. These results are also concordant with the mild cystic fibrosis phenotype of patients harboring the ⌬26 CFTR (36).
NHERF Potentiates the PKA-dependent Activation of CFTR-NHERF may facilitate CFTR activation at the cell membrane by inducing conformational changes, promoting CFTR phosphorylation by anchoring the catalytic subunit of PKA/protein kinase C to the vicinity of the channel (28,29,31), and/or increasing the channel number (32,34). To examine whether NHERF can have a modulatory effect on the channel activity in intact cells, we compared the effect of NHERF and D1 domain on the PKA-stimulated whole cell chloride current FIG. 5. Distribution of WT and truncated CFTR in polarized MDCK II cells. MDCK II cells were transiently transfected with the ⌬6, ⌬26, and WT CFTR construct and cultured on polycarbonate membranes for 3 days postconfluence. Cells were immunostained for the tight junctional marker ZO-1 with a rat monoclonal Ab (green) and for CFTR with the mouse monoclonal anti-HA Ab (red). A, horizontal optical confocal images; B, the x-z projection. Note that no significant difference could be detected in the apicobasolateral distribution pattern of the WT and truncated CFTR. The location of horizontal optical slices, obtained by LCFM, is shown in the inset. C-F, subcellular distribution of apical, lateral, and tight junctional markers in MDCK II cells. Transfected cells were cultured on polycarbonate filter support and co-immunostained for ZO-1 (green) and for apical membrane protein gp135 (C and E) or the lateral membrane protein E-cadherin, (D and F) (red). C and D, horizontal optical slices; E and F, x-z projections.
activity. Since D1 domain binds to the C-terminal tail of CFTR with higher affinity than D2 (30), we argued that expression of the monovalent D1 domain represents a dominant negative mutant. WT or ⌬26 CFTR-expressing BHK cell lines were overtransfected with either NHERF or the D1 domain. Immu-noblot analysis verified that overexpression of NHERF or the D1 domain did not alter the cellular level of complex-glycosylated CFTR, whereas it increased NHERF expression level by 5-fold (data not shown).
The whole cell current-voltage relationship was measured by FIG. 6. Association of NHERF with CFTR is blocked by fusing the HA or the EGFP tag to the C terminus of CFTR. EGFP-or HA-tagged CFTR was immunoprecipitated (IP) from BHK cells co-expressing HA-NHERF by the M3A7 and L12B4 anti-CFTR antibodies (upper and middle panels). The precipitates as well as the cell lysates were immunoblotted (W) with anti-HA Ab and visualized by an enhanced chemiluminescence assay to detect the HA-tagged CFTR and NHERF. Note that the C-terminal HA as well as EGFP tag prevents complex formation between CFTR and NHERF ( lanes 5 and 7), whereas inserting the HA-tag into the C terminus or fusing EGFP-DTRL restores NHERF binding to CFTR (lanes 4 and 8). The expression level of HA-NHERF is comparable in BHK cells harboring the various CFTR constructs (lower panel).

FIG. 7. Biochemical and functional characterization of epitope-tagged CFTR variants.
A, steady-state expression level CFTR was assessed by immunoblotting of cell lysates, prepared from BHK cells stably transfected with the indicated HA-tagged CFTR. CFTR was visualized by the L12B4 anti-CFTR Ab and enhanced chemiluminescence. The coreand complex-glycosylated CFTR are indicated by empty and filled arrowheads, respectively. B and C, biogenesis and stability of tagged CFTRs. Metabolic pulsechase experiments were performed on BHK cells, expressing the indicated CFTR as described under "Experimental Procedures." CFTR was pulse-labeled with [ 35 S]methionine and [ 35 S]cysteine, chased for the indicated time, immunoprecipitated, and visualized by fluorography. D, the stability of the complex-glycosylated CFTR, CFTR-GFP, and CFTR-GFPDTRL was measured by metabolic pulse-chase techniques in BHK cells as shown in C, following a 3-h chase to allow the conversion of the core-into the complex-glycosylated form. Radioactivity remaining in CFTR was measured by Phos-phorImager analysis and expressed as a percentage of the initial amount. Data are means Ϯ S.E. (n ϭ 3). E and F, currentvoltage relationship of BHK cells expressing CFTR, CFTR-N terminus H-tagged, CFTR-GFP, and CFTR-GFPDTRL was determined before (ϪcAMP) and after maximal phosphorylation of CFTR by the PKA agonist mixture (ϩcAMP) using the patch clamp technique in the whole cell configuration. Since the basal currentvoltage relationships are similar, only one of them (WT CFTR) is depicted. F, statistical analysis shows no significant difference in the PKA-stimulated whole cell current densities of the native, HA-and GFP-tagged CFTR. Results are means Ϯ S.E. (n ϭ 19 -25). the patch clamp technique in BHK cells expressing WT CFTR with pCDNA3, NHERF, or the D1 domain. Current values were recorded in the resting state and after maximal stimulation of CFTR. Maximal activation was attained by the agonist mixture, since the current density evoked by forskolin alone could not be superseded by subsequent CPT-cAMP and isobutylmethylxanthine addition (Fig. 9A). NHERF overexpression augmented the PKA-activated whole cell current density by 30% compared with the mock-transfected cells (334.5 Ϯ 23.3 pA/pF (n ϭ 15) versus 258.4 Ϯ 25.7 pA/pF (n ϭ 21), p ϭ 0.05) (Fig. 9B). In contrast, the D1 domain only marginally decreased the current density (220.3 Ϯ 21.9 pA/pF, n ϭ 25). However, the current density of the NHERF-overexpressing cells was significantly (p Ͻ 0.02) higher than the D1-transfected ones (Fig. 9B). Deletion of the last 26 amino acids ablated the effect of NHERF (210.5 Ϯ 37.7 pA/pF, n ϭ 12) and the D1 domain (204.7 Ϯ 33.5 pA/pF, n ϭ 10) (Fig. 9B), indicating that the stimulatory effect of NHERF is specific to the PDZbinding motif of CFTR. These observations demonstrate the potential modulatory role NHERF on the channel function of CFTR in intact cells. DISCUSSION This study was undertaken in an attempt to assess the physiological significance of the C-terminal tail of CFTR, in light of the contrasting clinical and cellular phenotypes reported for the truncated CFTR. Compound heterozygous indi-viduals, harboring the ⌬26 CFTR on one allele and a nonsense mutation on the second allele (deletion of exon 14), have moderately elevated sweat chloride concentration yet have a normal pancreatic and pulmonary phenotype (36). Considering that 10 -20% of the WT CFTR message is sufficient to preserve normal epithelial function (54 -56) and that the ⌬26 and WT CFTR have similar single channel characteristics (9), at least 20 -40% of ⌬26 CFTR channels should be directed to the apical domain to prevent lung and pancreatic disease in individuals harboring ⌬26 CFTR only on one allele. Surprisingly, a ⌬26 GFP-CFTR fusion protein was reported earlier to be predominantly localized to the lateral membrane of MDCK II and human bronchial epithelia (32), retaining 3% of the WT CFTR activity at the apical membrane (34). This level of CFTR activity should be associated with severe pancreatic and lung phenotype (54 -56).
Regardless of whether C-terminal deletion or epitope fusion was used to prevent the complex formation between NHERF and CFTR in vivo, polarized targeting of CFTR was preserved not only in dog kidney (MDCK II), human intestinal (CaCo-2), and tracheal (HTE) but in pancreatic duct epithelia (PANC-1) as well. There were negligible differences in the polarized distribution of WT, truncated (⌬6, ⌬26, and ⌬40) or C-terminally tagged (HA, EGFP, and EGFP-DTRL) CFTR variants. In addition, we could not detect a significant difference in the metabolic stability and the PKA-stimulated apical anion conductance of WT and ⌬6 CFTR in polarized PANC-1 epithelia, ruling out the possibility that truncation provokes the redistribution of CFTR between the apical membrane and subapical endosomes. Based on these observations, we propose that in differentiated epithelia, derived from respiratory, intestinal, pancreatic duct, and kidney tubules, the apical localization of CFTR requires sorting signals additional to the C-terminal 26 amino acid residues and the PDZ-binding motif. Our results help to reconcile the mild clinical phenotype of individuals with the cellular behavior of the truncated CFTR reported here. The data presented are consistent with the lack of effect of the C-terminal vesicular stomatitus virus glycoprotein tag on the PKA-activated short circuit current reported in pig kidney epithelia (LLC-PK1) (57). Although disruption of NHERF binding was not demonstrated by Costa de Beauregard et al. (57), it is more than likely that the vesicular stomatitus virus glycoprotein tag, similarly to the C-terminal HA or the EGFP tags, prevented NHERF association with CFTR. In addition, a recent abstract of Ostedgaard et al. (58) indicates that C-terminal deletions had no effect on the magnitude of the PKA-activated transepithelial chloride current in differentiated primary human respiratory epithelia.
There are number of differences between the experimental conditions utilized in previous studies and in our work that may account for the different localization of the ⌬26 CFTR. In our experiments, CFTR was tagged with the small HA tag or the larger EGFP tag to immunolocalize the channel. While the HA tag was fused at four different locations, the EGFP was attached to the C terminus of CFTR. Biochemical and initial functional analysis suggested that the tags do not alter the intracellular trafficking and chloride conductance of CFTR, precluding potential problems associated with the epitope-tagging approach as demonstrated by aquaporin-2-GFP fusion proteins (51).
Although clonal variations could not be ruled out as a plausible explanation for the discordant results, we tried to minimize cell-specific contribution, which could profoundly alter the cell surface delivery of membrane proteins in complex with PDZ proteins (52,59). Immunolocalizations of CFTR were carried out in both nonpolarized cell lines and four different, x-z projections were reconstructed from horizontal optical sections. G and H, immunolocalization of CFTR-HAC i and CFTR-HAC t in polarized MDCK II cells. Indirect immunolocalization of epitope-tagged CFTR and ZO-1 was achieved on transiently transfected, differentiated MDCK II monolayers, grown on permeable support. Note that both constructs are associated, predominantly, with the apical membrane, and negligible staining is detectable below the tight junction.
clinically relevant epithelia. Since polarization is critical to the apical targeting of CFTR (60, 61), differentiation of epithelia was verified by immunolocalization of tight junctional (ZO-1 and occludin) and lateral membrane markers (E-cadherin) and by domain-specific lectin labeling as well. Adverse effects of lipid-mediated transfection were also ruled out by generating stably transfected PANC-1 epithelia and by using retroviral infection (HTE, MDCK, and CaCo-2 epithelia).
NHERF association with CFTR was disrupted by deletional mutagenesis or by fusing the HA or the EGFP tag to the C terminus of CFTR. The efficiency of all of these methods was verified, lending credence to our assumption these CFTR variants fail to interact with NHERF in vivo. This notion was further supported by preliminary fluorescence recovery after photobleaching studies, suggesting that the immobile fraction of CFTR-GFP increased after attaching the DTRL motif to its C terminus in MDCK cells (data not shown).
While we are confident that the apical targeting of CFTR is independent of NHERF and the C-terminal tail of CFTR in those epithelia examined, the reasons behind the moderately elevated sweat chloride concentration of individuals with the ⌬26 CFTR mutation remain elusive. Considering that CFTR is localized both to the apical and to the basolateral membranes in sweat duct cells (62) and that PDZ domain-containing adaptor molecules are involved in basolateral targeting/retention of some membrane proteins (63), we speculate that the C-terminal truncation may interfere with the basolateral expression of CFTR in sweat duct cells. Alternatively, full activation of CFTR may require the association of NHERF with CFTR in sweat duct. Due to the high flux of chloride resorption, even a modest inhibition of the truncated CFTR activity would lead to elevated sweat chloride concentration. Our attempts to distinguish between these possibilities, however, have failed due to insufficient expression of CFTR in immortalized sweat duct epithelia.
The electrophysiological measurements provided direct evidence for a stimulatory role of NHERF in the PKA-dependent activation of CFTR in intact cells. This could be the consequence of increased channel density, augmented open probability, and/or unitary conductance. NHERF overexpression may increase the recycling efficiency of internalized CFTR, as described for the ␤ 2 -adrenergic and the ␦and the -opioid receptors in nonpolarized cells (24,26,50) and suggested for GFP-CFTR (35). Whereas a role of NHERF in the recycling of CFTR cannot be completely discounted, we were unable to detect attenuated expression, metabolic destabilization, or any functional defect of the truncated CFTR in pancreatic duct and BHK cells (11). Deletion of the last 26 amino acid residues had no impact on the whole cell current densities in NIH 3T3 cells as reported earlier (36). Since defective recycling of the Gprotein-coupled receptor in the absence of NHERF binding is invariably associated with the accelerated degradation of the receptors (24,26,50), our observations, collectively, argue against the role of NHERF in CFTR recycling. Although the functional studies imply that the endogenous level of NHERF is insufficient to influence CFTR activity in BHK and PANC-1 cells, variations in the concentration and/or the compartmentalization of NHERF may ensure a regulatory role for NHERF in CFTR activation in other cells (64), including the sweat duct and Calu-3 epithelia (29,65).
NHERF overexpression may facilitate CFTR phosphorylation by anchoring the catalytic subunit of PKA via the NHERFezrin macromolecular complex to the vicinity of the channel (28,29,31). Since the activation kinetics of the PKA-stimulated whole cell current were not influenced by NHERF or the dominant negative D1 overexpression (Fig. 9C and D), an increased phosphorylation rate of CFTR upon PKA activation is unlikely. We favor the hypothesis that complex formation between CFTR and the two PDZ domains of NHERF exerts its effect via allosteric modulation, as envisioned for both NHERF and CFTR-associated ligand of 70 kDa (13,30). An alternative, but not mutually exclusive possibility is that NHERF promotes the formation of a signaling complex between CFTR and protein kinase C⑀, potentiating CFTR activation by PKA (66). Additional experiments are required to distinguish between these possibilities.