Efficient Endocytosis of the Cystic Fibrosis Transmembrane Conductance Regulator Requires a Tyrosine-based Signal*

We previously demonstrated that the cystic fibrosis transmembrane conductance regulator (CFTR) is rapidly endocytosed in epithelial cells (Prince, L. S., Workman, R. B., Jr., and Marchase, R. B. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 5192–5196). To determine the structural features of CFTR required for endocytosis, we prepared chimeric molecules consisting of the amino-terminal (residues 2–78) and carboxyl-terminal tail regions (residues 1391–1476) of CFTR, each fused to the transmembrane and extracellular domains of the transferrin receptor. Functional analysis of the CFTR-(2–78) and CFTR-(1391–1476) indicated that both chimeras were rapidly internalized. Deletion of residues 1440–1476 had no effect on chimera internalization. Mutations of potential internalization signals in both cytoplasmic domains reveal that only one mutation inhibits internalization, Y1424A. Using a surface biotinylation reaction, we also examined internalization rates of wild type and mutant CFTRs expressed in COS-7 cells. We found that both wild type and A1440X CFTR were rapidly internalized, whereas the Y1424A CFTR mutant, like the chimeric protein, had ∼40% reduced internalization activity. Deletions in the amino-terminal tail region of CFTR resulted in defective trafficking of CFTR out of the endoplasmic reticulum to the cell surface, suggesting that an intact amino terminus is critical for biosynthesis. In summary, our results suggest that both tail regions of CFTR are sufficient to promote rapid internalization of a reporter molecule and that tyrosine 1424 is required for efficient CFTR endocytosis.

Previous studies have demonstrated that CFTR is endocytosed (5, 6) through clathrin-coated vesicles (6,7), suggesting that CFTR internalization may provide a mechanism for controlling the cAMP-stimulated chloride channel activity at the cell surface (6). Others have suggested that CFTR may play additional roles by regulating plasma membrane recycling (5,8) and in clearance of Pseudomonas aeruginosa from the respiratory tract (9,10).
The purpose of this study was to determine the structural features of the CFTR protein required for internalization. Internalization signals identified to date include tyrosine-based motifs (YXX or NPXY, where X is any amino acid and is a bulky hydrophobic residue), dileucine motifs, and acidic cluster/casein kinase II-based motifs (11)(12)(13)(14)(15). Initial studies of type III membrane proteins indicate that the targeting signals occur in the amino-and carboxyl-terminal cytoplasmic tail regions (16 -19).
Our initial studies on the identification of CFTR internalization signals focused on the two tail regions of the CFTR molecule. Here we show that both the amino-and carboxyl-terminal cytoplasmic tail regions of CFTR, residues 2-78 and 1391-1476, are individually sufficient to promote rapid internalization of a reporter molecule, the transferrin receptor (TR). We also demonstrate both in the context of chimeric and native proteins that tyrosine 1424 is important for CFTR endocytosis. Furthermore, we show that the intracellular distribution of the CFTR-TR chimeras is similar to that of the TR, suggesting that endocytosis may regulate CFTR activity at the cell surface.
Expression of Wild-type TR and CFTR-TR Chimeras-Human TR and CFTR-TR chimeras were expressed in chicken embryo fibroblasts as described previously (25) using the BH-RCAS expression vector (26,27). Internalization Assay-The rate of transferrin internalization was determined using the IN/SUR method (28) as described previously (29).
Construction of CFTR Mutants-pMT-CFTR (wild type) and pMT-CFTR-A1440X were kindly provided by Dr. Seng Cheng (Genzyme) (30). pKCTR-CFTR (wild type) was provided by Dr. Eric Sorscher and the Gregory James Cystic Fibrosis Research Center Vector Core (University of Alabama at Birmingham) (31). For mutagenesis of the aminoterminal region of CFTR, a XmaI-XbaI fragment of pKCTR-CFTR was subcloned into pSK-Bluescript (Stratagene). For mutagenesis of the carboxyl-terminal region of CFTR, a BstXI-SgrAI fragment from pKCTR-CFTR was subcloned into pSK-Bluescript. CFTR point mutations or deletions in the amino-or carboxyl-terminal tail regions were prepared from the corresponding pSK-Bluescript vectors (containing either the XmaI-XbaI or BstXI-SgrAI fragments, respectively) from single-stranded DNA as described previously (11) by the method of Kunkel (32). Mutants were selected by restriction mapping or sequencing and then subcloned into the XmaI-XbaI or BstXI-SgrAI site of pKCTR-CFTR. The mutations were verified by dideoxynucleotide sequencing (24) using the Sequenase kit (U.S. Biochemical Corp.) according to the manufacturer's directions.
Transient Expression of Wild Type and Mutant CFTRs in COS-7 Cells-Transient expression of wild type or mutant CFTR in COS-7 cells was performed as described by Cheng et al. (30). The transfected cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal calf serum and incubated at 37°C in humidified air with 5% CO 2 for 48 h.
SPQ Fluorescence Assay of Wild Type and Mutant CFTRs in COS-7 Cells-CFTR function in individual cells was assayed using the halidequenched dye SPQ (30). Briefly, cells were loaded for 10 min with SPQ (10 mM) by hypotonic shock and then mounted in a specially designed perfusion chamber for fluorescence measurements. Fluorescence (F) of single cells was measured with a Zeiss inverted microscope, a PTI imaging system, and Hamamatsu camera. Excitation was at 340 nm, and emission was Ͼ410 nm. All functional studies were at 37°C. At the beginning of the experiments, cells were bathed in a quenching buffer (NaI buffer; 130 mM NaI, 5 mM KNO 3 , 2.5 mM Ca(NO 3 )2, 2.5 mM Mg(NO 3 ) 2 , 10 mM D-glucose, 10 mM HEPES), and following establishment of a stable base line, they were switched to a halide-free (NO 3 ) dequenching buffer at 200 s. Cells were stimulated with agonist unless otherwise indicated at 500 s and then returned to the quenching NaI buffer. Fluorescence was normalized to the base-line (quenched) value (average fluorescence from 100 to 200 s), with increases presented as percentage of increase F over basal level. Each curve was generated from the mean values (ϮS.E.) of either 1) responding cells (defined by increase in dequench slope of Ͼ100% following cAMP stimulation (   (28). A representative experiment is shown. B, a summary of the internalization rates (mean Ϯ S.E.; n ϭ 5 or more) for each of the CFTR-TR chimeras is shown.
with the wild-type CFTR (Fig. 8A) or the mock condition (Fig. 8B). The buffers used in the SPQ assay were 1) NaI buffer, pH 7.3, and 2) NaNO 3 buffer (identical to NaI buffer except that 130 mM NaNO 3 replaces NaI).
CFTR Labeling and Internalization-Cell surface CFTR biotinylation was performed as described previously (5).
Separation and Isolation of Biotinylated CFTR-Biotinylated and nonbiotinylated proteins were separated on an immobilized monomeric avidin column (Pierce) as described previously (5). The unbound fraction and biotin eluent fraction were then quantitated as described below.
Immunoprecipitation and Detection of CFTR-Wild-type or mutant CFTRs were immunoprecipitated from the pooled fractions (either unbound fraction or biotin eluent fraction) with either anti-C-terminal (24 -1) or anti-R domain (13-1) monoclonal antibodies generously supplied by Dr. Seng Cheng (Genzyme). The CFTR immunoprecipitates were phosphorylated with [␥-32 P]ATP and cAMP-dependent protein kinase and then analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography as described previously (5). Quantitation was performed using a Molecular Dynamics PhosphorImager. The percentage of cell surface CFTR was calculated by dividing the number of counts/min detected in the CFTR C-band of the biotin eluent fraction by the total amount of CFTR C-band found in both the unbound and biotin eluent fractions (5).
Indirect Immunofluorescence-COS-7 cells expressing wild-type or mutant CFTRs were trypsinized 12 h after transfection and plated onto glass coverslips. 48 h later, the coverslips were fixed in methanol/acetic acid (3:1) for 30 min at Ϫ20°C, rinsed with 1% BSA in PBS for 10 min at room temperature, and incubated with either monoclonal antibody 24 -1 or 13-1 (diluted 1:50 in PBS/BSA) for 1 h at room temperature. Coverslips were then incubated for 5 min in 1% BSA in PBS (three times), incubated with Texas Red-conjugated goat anti-mouse antibody (diluted 1:1000 in PBS, 1% BSA, 5% normal goat serum) for 1 h at room temperature, and then rinsed for 5 min in 1% BSA in PBS (3 times). Coverslips were mounted in 1 mg/ml p-phenylenediamine in a 1:10 mixture of PBS/glycerol and sealed with nail polish. Slides were analyzed o a Leitz fluorescence microscope equipped with a Vario Orthomat II camera system and a Texas Red epifluorescence filter module. Micrographs were prepared on T-Max 400 film processed at ASA 800. Chicken embryo fibroblasts expressing the CFTR-TR chimeras and wild-type TR were plated onto glass coverslips and cultured overnight. The coverslips were then rinsed with PBS and fixed in 2% formaldehyde in PBS for 15 min at room temperature; rinsed with PBS; quenched with 0.37% glycine, 0.27% NH 4 Cl in PBS; and permeabilized with PHS (PBS, 10% horse serum, 0.1% saponin) for 30 min at room temperature. The coverslips were then incubated with JS8 mouse anti-chicken TR antibody (diluted in PHS) and with rabbit anti-human TR (1:500 in PHS) for 30 min at 37°C and rinsed in PHS at room temperature. The coverslips were next incubated with Texas Red-labeled goat anti-mouse FIG. 3. Co-localization of CFTR-TR chimeras with the endogenous TR. Chicken embryo fibroblasts expressing wild-type human TR (A), CFTR-(2-78) (B), CFTR-(1391-1440) (C), and ⌬3-59 TR (D), were fixed, permeabilized, and stained with rabbit anti-human TR antisera followed by Oregon Green-labeled goat anti-rabbit Ig and with JS8-mouse anti-chicken TR antibody followed by Texas Red-labeled goat anti-mouse IgG 1 . Co-localization of the two proteins is indicated by yellow fluorescence. IgG 1 (1:50 dilution in PHS) and with Oregon Green-labeled goat antirabbit antibody (1:50 in PHS) for 30 min at 37°C and then rinsed, mounted, and analyzed as described above.

CFTR-TR Chimeras Containing the Amino-and Carboxylterminal Tails of CFTR Are Expressed on the Cell Surface and
Rapidly Internalized-To identify the regions of CFTR important for endocytosis, we prepared chimeric molecules consisting of the amino terminus (residues 2-78) and the carboxyl terminus of CFTR (residues 1391-1476), each fused to the transmembrane and extracellular domains of the human TR (Fig. 1). Both chimeras along with a wild-type TR and ⌬3-59 TR (a mutant TR that is very poorly internalized (11)) were expressed in chicken embryo fibroblasts using BH-RCAS, a replicationcompetent retroviral vector derived from the Rous sarcoma virus (26,27). Cell surface expression of both chimeras was confirmed using indirect immunofluorescence and 125 I-labeled transferrin binding at 4°C (data not shown).
Internalization rates of the CFTR-TR chimeras were monitored using the IN/SUR method of Wiley and Cunningham (28). Analysis of residue 2-78 CFTR-TR and residue 1391-1476 CFTR-TR indicated that both were internalized rapidly (k e ϭ 0.126 and 0.061, respectively, similar in fact to the wild-type TR (k e ϭ 0.090; Fig. 2A). For comparison, the ⌬3-59 TR lacking an internalization signal (11) was internalized very slowly (k e ϭ 0.009). This suggested that both cytoplasmic tail regions of CFTR were sufficient to promote TR endocytosis. To determine if the potential acidic cluster/casein kinase II region of CFTR (residues 1469 -1474) in the carboxyl-terminal tail was important for endocytosis, we prepared a deletion mutant lacking this region, CFTR-(1391-1440) (Fig. 1) and compared the internalization rates of the two chimeras. The results indicate that CFTR-(1391-1440) and CFTR-(1391-1476) chimeras were internalized with similar kinetics (k e ϭ 0.063 versus 0.061, respectively; Fig. 2A), suggesting that the membrane-distal portion of the carboxyl-terminal tail of CFTR was not required for efficient endocytosis.
Next, we analyzed CFTR-TR chimeras that contained point mutations in potential internalization signals in both cytoplasmic tail regions: Y38A, L69A, Y1424A, and L1430A (Fig. 1). Analysis of these mutants in internalization assays indicated that only one mutation, Y1424A, affected the internalization rate of the chimeras (Fig. 2B; ϳ40% loss of internalization activity, p Ͻ 0.05), suggesting that this residue might be a part of an internalization signal. Since overexpression of the human wild type transferrin receptor (up to 10-fold over the endogenous receptor) does not affect internalization of the endogenous receptor in this cell type (not shown), the endocytic machinery should not be limiting for analysis of the CFTR chimeras. Therefore, we were unable to determine if the CFTR chimeras competed for the same cytosolic factors as the endogenous TR receptors.
CFTR-TR Chimeras Co-localize with the Transferrin Receptor-To examine the intracellular distribution of the chimeras, we compared their distribution to that of the endogenous transferrin receptor using immunofluorescence microscopy. CFTR-(2-78) was localized predominantly to juxtanuclear structures that largely co-localized with the native chicken TR receptor (Fig. 3B). This is similar to co-localization of the human TR expressed in these cells compared with the endogenous receptor (Fig. 3A) except that the relative surface expression of the wild-type TR appeared to be much higher (shown in green) than that of CFTR-(2-78) chimera. Very few vesicles were CFTR-(2-78)- (Fig. 3A) or CFTR-(1391-1440)- (Fig. 3B) positive only, but there were vesicles containing only TR (shown in red). The intracellular distribution of the CFTR-(1391-1440) chimera (Fig. 3C) appeared similar to that of the CFTR-(2-78) chimera FIG. 4. Schematic diagram of CFTR and CFTR mutants. The CFTR mutants used in this study consisted of point mutations and deletions in the amino-terminal and carboxyl-terminal cytoplasmic tails. The boxes refer to the wild-type (WT) amino-terminal tail (residues 1-80), and the lines represent deleted amino acids from this domain. The wild-type carboxyl-terminal tail consists of residues 1391-1480. A1440X contains a stop mutation at residue 1440, and Y1424A contains an alanine substitution for tyrosine. but very different from a "tailless" TR (⌬3-59 TR) that lacks intracellular sorting signals and is localized primarily to the cell surface (Fig. 3D). Comparison of the CFTR-(1391-1440) with the Y1424A mutant showed little or no difference (not shown). In nonpermeabilized cells, the wild type and tailless TRs are strongly positive by immunofluorescence, whereas the two CFTR chimeras are only weakly positive (not shown). 125 I-Labeled transferrin binding at 4°C indicated that the relative surface expression of both CFTR chimeras is approximately 5-10-fold lower than the wild-type TR (not shown). These results suggest that both chimeras co-localized with the endogenous TR and therefore were a part of the constitutive recycling pathway as is the case for the native TR.
CFTR and 1440X CFTR Are Rapidly Internalized in COS-7 Cells-Having established that both cytoplasmic tail regions were sufficient for endocytosis, we next determined if they were necessary in the context of the CFTR protein. CFTR, unlike the TR, is a type III membrane protein with 12 membrane-spanning domains. In addition to the R-domain and the two nucleotide binding domains, both amino-and carboxyl-terminal tails of CFTR are cytoplasmic in orientation (Fig. 4). Using a cell surface two-step biotinylation assay to monitor CFTR endocytosis (5) that relies on biotinylation of the carbohydrate side chains found in extracellular loop 4 (Fig. 4), we compared the internalization rate of wild-type CFTR to a previously described premature stop mutant, A1440X (33). First, we confirmed that A1440X expressed in COS-7 cells is maturely glycosylated by monitoring band C formation (Fig. 5A). Next, using the surface biotinylation assay, we monitored CFTR and A1440X clearance from the cell surface (Fig. 6A). Interestingly, the A1440X premature stop mutant was internalized faster than the wild-type CFTR, suggesting, as had been seen for the chimeric protein, that the last 41 residues in CFTR were not necessary for rapid endocytosis.
Tyrosine 1424 Is Important for CFTR Endocytosis-Next, we tested the only point mutation that affected internalization of the chimeras, Y1424A. Analysis of this mutation in CFTR revealed that it was maturely glycosylated (Fig. 5B) but internalized 41% more slowly than the wild-type CFTR protein (k e ϭ 0.16 versus 0.27 (p Ͻ 0.01); Fig. 6B). Since slower internaliza-

FIG. 6. Comparisons of the internalization rates of CFTR and CFTR mutants. A, internalization of CFTR and A1440X in COS-7 cells. COS-7 cells transfected with wild-type CFTR (pMT-CFTR) or A1440X (pMT-1440X
) were analyzed 48 h posttransfection. Cell surface CFTR or CFTR mutants were biotinylated using a two-step cell surface periodate/LC-hydrazide biotinylation procedure previously described (5). At zero time, both steps were conducted at 4°C to label the entire surface pool of CFTR. Internalization is monitored by a loss of biotinylation of the cell surface pool by including a 37°C incubation period (shown on the x axis as time in min) between periodate and biotin LC-hydrazide treatments. Biotinylated and nonbiotinylated proteins were separated on a monomeric avidin column, and CFTR and A1440X were in vitro phosphorylated and analyzed by SDS-polyacrylamide gels and autoradiography to quantitate the amount of CFTR remaining on the cell surface during the warm-up step. Each time point represents the mean Ϯ S.E. of 13 experiments for wild-type CFTR and 6 for the A1440X. B, internalization of CFTR and Y1424A in COS-7 cells. Cells transfected with wild-type CFTR (pGT-CFTR) or Y1424A were analyzed as described in A for percentage CFTR internalized from the cell surface during the warm-up period. C, percentage of CFTR or Y1424A at the cell surface under steady-state conditions. Cells transfected with CFTR or Y1424A were analyzed for total CFTR expression by performing the two-step biotinylation reaction without a warm-up step. Biotinylated and nonbiotinylated CFTR or Y1424A was separated on a monovalent avidin column and quantitated as described for A. The percentage of CFTR at the cell surface represents the mean Ϯ S.E. of six experiments. D, wild-type CFTR expression in COS-7 cells using pMT-CFTR and pGT-CFTR. COS-7 cells transfected with CFTR using the pMT-CFTR and pGT-CFTR vector were analyzed 48 h posttransfection. Biotinylated (biotin eluent) and nonbiotinylated (unbound) CFTR was separated on a monovalent avidin column. Biotinylated and nonbiotinylated CFTR were then immunoprecipitated from the two fractions, phosphorylated in vitro with protein kinase A and [␥-32 P]ATP, and analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. The positions of bands B and C are indicated on the left. The relative amount of CFTR expressed using the pMT-CFTR vector was always higher than the pGT-CFTR vector. tion would imply that the steady-state distribution of Y1424A might favor a higher cell surface distribution pattern, we determined the percentage of CFTR at the cell surface using surface biotinylation at 4°C. As predicted, the percentage of Y1424A at the cell surface was higher than the wild-type CFTR protein (36.1 versus 23.1% (p Ͻ 0.01), respectively; Fig. 6C), supporting the idea that tyrosine 1424 was important for CFTR endocytosis.
CFTR Expression Levels in COS-7 Cells Affect Internalization Rates-In our analysis of wild-type CFTR internalization, we were surprised to find that the relative internalization rates varied depending on whether the cells were transfected with the pMT-CFTR expression vector (30) or the pGT-CFTR vector (31). The CFTR internalization rate (k e ) in COS-7 cells transiently transfected with pMT-CFTR was 0.16 (Fig. 6A), whereas the CFTR internalization rate in cells transfected with pGT-CFTR was 0.27. Since the promoters were different in the two expression vectors, we compared the relative expression levels by immunoprecipitating and in vitro phosphorylating CFTR from cells transfected with the two different expression vectors with [␥-32 P]ATP and cAMP-dependent protein kinase and analyzing by SDS-polyacrylamide gel electrophoresis and autoradiography (5). As is shown in Fig. 6D, the protein expression levels in transfected cells using the pMT-CFTR vector were substantially higher, suggesting that CFTR internalization rates were directly affected by protein expression levels. A similar phenomenon occurs when the transferrin receptor is overexpressed in HeLa cells (34), suggesting that the cellular machinery for removing surface receptors is limiting and can be saturated. A similar comparison between wild type CFTR (pGT) and Y1424A (pGT) revealed that the expression levels were similar (Fig. 5B).
Amino-terminal Deletions of CFTR Are Only Core-glycosylated-In order to localize the potential internalization signals in the amino terminus of CFTR, we prepared four deletion mutants of CFTR: ⌬2-79, ⌬2-59, ⌬35-45, and ⌬60 -72. Expression of each of these mutants in COS-7 cells produced only the B form of CFTR (Fig. 5C). This protein was sensitive to endoglycosidase H treatment and could not be biotinylated at the cell surface (data not shown). To determine the intracellular location of the transfected CFTR mutants, we examined the cells using indirect immunofluorescence. As is shown in Fig. 7, the wild-type CFTR protein and A1440X (Fig. 7, A and B, respectively) showed a strong juxtanuclear distribution and evidence of surface staining (a clear outline of the cell borders), whereas all of the deletion mutants had a reticular staining pattern and little if any evidence of surface staining. The results suggest, therefore, that all of the deletion mutants were expressed but failed to exit the endoplasmic reticulum and reach the cell surface.
Finally, as functional proof that the CFTR deletion mutants failed to reach the cell surface, we tested each of these mutants for CFTR Cl Ϫ channel activity using the SPQ halide efflux assay. Fig. 8A shows that wild-type CFTR generated Cl Ϫ channel activity after stimulation with cAMP agonists (20 M forskolin and 100 M 3-isobutyl-1-methylxanthine). However, expression of each of the deletion mutants failed to produce functional CFTR Cl Ϫ channels, indicating either that the CFTR deletion mutants were nonfunctional chloride channels or that they failed to reach the cell surface or both. The A1440X mutant, unlike the amino-terminal deletion mutants, had wildtype chloride channel activity (Fig. 8B). Since the amino-terminal deletion mutants of CFTR failed to reach the cell surface, we were unable to determine if this domain is necessary for CFTR endocytosis.

DISCUSSION
The conclusions of this study are as follows: 1) the amino and carboxyl termini of CFTR are each sufficient to promote the endocytosis of the transferrin receptor; 2) the carboxyl-terminal 40 amino acids of CFTR are not required for endocytosis; 3) tyrosine 1424 is a critical part of an internalization signal; and 4) deletions in the amino-terminal tail region of CFTR result in loss of CFTR surface expression. Our results clearly indicate that both cytoplasmic tail domains of CFTR are sufficient to promote internalization of a reporter molecule, suggesting that these domains are capable of interacting with the cell's endocytic machinery. Recent results, however, indicate that both domains interact with other proteins at the cell surface, suggesting that CFTR internalization and/or function may be regulated by these interactions (see below).
The only mutation that we identified in our studies that affected CFTR internalization was found in the carboxyl-terminal tail, Y1424A. This mutation inhibited endocytosis by approximately 40% in both the native protein and the chimera. For comparison, a similar mutation in the tyrosine-based internalization signal of the TR results in an 80% loss of internalization activity (11). Since the same modification (Tyr 3 Ala) has a more modest effect on CFTR internalization than TR internalization, it is tempting to speculate that the CFTR protein contains more than one internalization signal, such as is the case for the CD3 ␥-chain (13).
To determine if the tyrosine residue was conserved in other species, we compared CFTR carboxyl-terminal tail sequences from 10 species (Fig. 9). In all but dogfish, the tyrosine residue is conserved. In this case, a phenylalanine is found in place of the tyrosine residue. Interestingly, previous studies have demonstrated that phenylalanine can substitute for tyrosine in the TR internalization signal, YTRF, without a loss of internalization activity (35). In seven of 10 CFTR sequences, the entire motif, YDSI, is conserved, suggesting that this sequence conforms to the general pattern of internalization signals, YXX, and may represent a CFTR internalization signal.
Our studies using deletional analysis of the CFTR amino terminus imply that this region of the molecule is required for proper folding of CFTR. Several lines of evidence support the idea that the four deletion mutants failed to reach the cell surface. First, when each mutant was expressed, only the endoglycoside H-sensitive B form of CFTR was detected. Second, we were unable to biotinylate CFTR from the surface of cells expressing any of the mutants. Third, immunofluorescence studies demonstrated that the protein was being made but was localized to a reticular staining pattern. Fourth, SPQ analysis of COS-7 cells demonstrated that cells expressing the wild-type CFTR had chloride channel activity, whereas the mock-transfected cells and cells transfected with each of the deletion mutants did not. Clearly, an intact amino terminus is required for proper CFTR biosynthesis and/or function, as is suggested by the fact that multiple point mutations in the amino-terminal tail result in cystic fibrosis (P13569). For comparison, only one mutation in the carboxyl-terminal tail has been reported to result in CF, V1397E. This finding and our results on the deletion mutants are consistent the idea that the amino terminus may associate with other domains of CFTR for proper CFTR maturation and folding.
Although we were able to show that the amino-terminal tail was sufficient to promote TR endocytosis, we were unable to show that this domain was necessary for CFTR endocytosis. Since modification of Tyr 1424 did not completely eliminate CFTR internalization, the prediction is that another signal exists, but this could be in either of the two tail regions. Therefore, the role of the amino terminus in CFTR endocytosis remains unclear. Further complicating this matter, recent results suggest that the amino terminus interacts with syntaxin 1A (36), and this interaction regulates CFTR activity (37). Although syntaxin 1A is known to be a component of the vesicle fusion machinery, disruption of syntaxin 1A/CFTR interaction potentiates chloride channel activity. Whether the CFTR/syntaxin 1A interaction regulates trafficking of CFTR or, more directly, modulates the channel activity of CFTR remains unclear (36). If the amino terminus of CFTR interacts with syntaxin 1A and/or other domains of CFTR, then the question is whether this domain would still be available for interaction with the clathrin and the adaptor molecules at the cell surface. One interesting possibility is that interaction is regulated in some manner and that the loss of interaction promotes loss of CFTR from the cell surface through the latent targeting information found in the CFTR amino-terminal tail.
The amino terminus may not be the only interacting domain at the cell surface. Recent studies have also proposed that the last four residues of the carboxyl-terminal tail of CFTR, Asp-Thr-Arg-Leu, interact with EBP50 and that this interaction tethers CFTR to the cytoskeleton (38,39). EBP50, through its PDZ domain, has been proposed to either regulate the stability of the CFTR protein at the cell surface or regulate CFTR channel function or perhaps both (38). Although the EBP50 is FIG. 8. Functional analysis of wild type and deletion mutants of CFTR using SPQ fluorescence. The change in SPQ fluorescence is shown for COS-7 cells expressing wild-type CFTR, ⌬2-79 CFTR, ⌬2-59 CFTR, ⌬35-45 CFTR, and ⌬60 -72 CFTR (A) and wild-type CFTR and A1440X (B). Cells were stimulated with 20 M forskolin, 50 M 8-(4chlorophenylthio)adenosine 3Ј:5Ј-cyclic monophosphate, and 100 M 3-isobutyl-1-methylxanthine (at arrow, cAMP). The change in SPQ fluorescence for all of the amino-terminal deletion mutants was not significantly different from the negative control (mock-transfected cells). Curves were generated from either 1) the mean Ϯ S.E. of responding cells (defined by increased rate of dequench following cAMP stimulation of Ͼ100% (wild type CFTR (A), wild type CFTR and A1440X CFTR (B)) or 2) the total of all screened cells expressing a given construct. The values in parentheses are the number of responding cells (numerator) over the total cells screened (denominator). A, The wild type CFTR sample was significantly different from each deletion mutant or mock sample (p Ͻ 0.001 by 2 test for each condition compared with wild-type CFTR). B, the wild type CFTR and A1440X CFTR samples were significantly different from the mock cells (p Ͻ 0.001 and 0.025, respectively).
FIG. 9. Comparison of the amino acid sequences of various CFTR carboxyl-terminal tails. Amino acid sequence alignment of C termini of CFTR from human (P13569), monkey (3057116), rabbit (Q00554), sheep (U20418), bovine (P35071), mouse (M60493), rat (1901178A), dogfish (P26362), killifish (AF000271), and Xenopus (U60209) is shown. The tyrosine residue is conserved among all of the species except the dogfish, which has a phenylalanine residue. Phenylalanine residues have been shown to substitute for tyrosine residues and still maintain wild-type internalization activity for the transferrin receptor (35). All of the sequences except dogfish conform to the YXX⌽ motif common to internalization signals, where X can be any amino acid and ⌽ is a hydrophobic residue. concentrated at the apical surface in human airway epithelial cells and associates with CFTR in in vitro binding assays, a direct linkage has not been established in vivo. Clearly, a direct comparison of the internalization rates of wild-type CFTR and A1440X mutant in a polarized epithelial cell line will be required to clarify the role of the carboxyl terminus in membrane association and/or endocytosis. Although it might not seem appropriate for a protein tethered to the cytoskeleton to be endocytosed through the clathrin-mediated pathway, this is clearly the case for the ␤ 2 -adrenergic receptor (40), which has recently been shown to associate with the first PDZ domain of EBP50 (41).
Our results demonstrate that CFTR has sorting signals that allow it to be rapidly cleared from the cell surface, but what is the physiological relevance of this? Recent studies indicate that ClC-5, the chloride channel mutated in Dent's disease, may be involved in acidification of endosomes (42,43). Such a role was once proposed for CFTR, but this was difficult to confirm, especially given the narrow tissue distribution of CFTR and the widespread phenomenon of endosomal acidification (44,45).
If CFTR is not important for endosome acidification, then why would it be endocytosed? The simplest explanation is that endocytosis may provide a mechanism for regulating the amount of CFTR at the cell surface at any one time. If CFTR is tethered to the cytoskeleton (38,39), then CFTR would be kept out of the recycling pathway. If the interaction at the cytoskeleton is regulated in any manner, then loss of association would result in rapid clearance from the cell surface through interactions with the clathrin-based sorting machinery. Therefore, the possibility exists that some of the CFTR is in a rapidly recycling pool and some is tightly associated with the cytoskeleton and that the distribution of these two pools is somehow regulated. Clearly, careful structure-function studies of CFTR in polarized epithelial cells will be required before we will understand the complex trafficking and regulation of the CFTR molecule.