Cystic fibrosis mutations lead to carboxyl-terminal fragments that highlight an early biogenesis step of the cystic fibrosis transmembrane conductance regulator.

Inefficient delivery of the cystic fibrosis transmembrane conductance regulator (CFTR) to the surface of cells contributes to disease in the majority of cystic fibrosis patients. Analysis of cystic fibrosis-associated missense mutations in the first nucleotide binding domain (NBD1), including A455E, S549R, Y563N, and P574H, revealed reduced levels of mature CFTR with elevated levels of carboxyl-terminal polypeptide fragments of 105 and 90 kDa. These fragments appear early in biogenesis and degrade rapidly in four distinct cell types tested including the bronchial epithelial IB3-1 cell line. They were detected at highest levels with CFTRA455E where the 105-kDa fragment accounted for 40% of newly synthesized polypeptide but for only 20 and 7% of nascent wild type and mutant DeltaF508 proteins, respectively. The bands represent core- and unglycosylated forms of the same CFTR fragment supporting that precursor forms are correctly inserted into the membrane of the endoplasmic reticulum. Proteolytic cleavage would be predicted to occur on the cytosolic face of the endoplasmic reticulum within the NBD1-R domain segment, but pharmacological testing did not support involvement of the 26 S proteasome. The examined missense mutations in NBD1 manifest differently than the major mutant, DeltaF508, and highlight a critical conformational aspect of biogenesis of CFTR.

Inefficient delivery of the cystic fibrosis transmembrane conductance regulator (CFTR) to the surface of cells contributes to disease in the majority of cystic fibrosis patients. Analysis of cystic fibrosis-associated missense mutations in the first nucleotide binding domain (NBD1), including A455E, S549R, Y563N, and P574H, revealed reduced levels of mature CFTR with elevated levels of carboxyl-terminal polypeptide fragments of 105 and 90 kDa. These fragments appear early in biogenesis and degrade rapidly in four distinct cell types tested including the bronchial epithelial IB3-1 cell line. They were detected at highest levels with CFTRA455E where the 105-kDa fragment accounted for 40% of newly synthesized polypeptide but for only 20 and 7% of nascent wild type and mutant ⌬F508 proteins, respectively. The bands represent core-and unglycosylated forms of the same CFTR fragment supporting that precursor forms are correctly inserted into the membrane of the endoplasmic reticulum. Proteolytic cleavage would be predicted to occur on the cytosolic face of the endoplasmic reticulum within the NBD1-R domain segment, but pharmacological testing did not support involvement of the 26 S proteasome. The examined missense mutations in NBD1 manifest differently than the major mutant, ⌬F508, and highlight a critical conformational aspect of biogenesis of CFTR.
Cystic fibrosis (CF) 1 is a recessive disorder with impaired vectorial transport in the epithelia of sweat gland ducts and of the respiratory, gastrointestinal, and genitourinary tracts. Over 800 mutations (1) identified in patients lead to a spectrum of phenotypes that include the classical picture of CF with chronic pulmonary disease, pancreatic insufficiency, elevated sweat electrolytes, and male sterility. Some of the mutations lead to the pancreatic sufficient form or to the notably milder form that involves congenital bilateral absence of the vas deferens as its major clinical manifestation (2). The affected gene product, the cystic fibrosis transmembrane conductance regulator (CFTR), functions as a protein kinase A and ATP-regulated, multidomain chloride channel (3)(4)(5)(6). Associated activities include regulatory roles in ATP transport, in amiloride sensitive Na ϩ transport, and in the action of outwardly rectifying chloride channels (7)(8)(9).
Both family studies (10) and CF mouse models (11) have emphasized the importance of genetic background in modulation of disease, but the nature of mutations or their combinations are major contributors to the tissue phenotype. The precise defects have not been elucidated for many mutations, but deficiencies can be broadly classified into at least six groups (2,13,14) resulting in the reduction of mature CFTR at the cell surface or impaired regulation and/or conduction of chloride.
A multidomain membrane protein, such as CFTR, attains its overall native conformation by co-and post-translational folding in the endoplasmic reticulum (ER) (15). These processes have been shown to be inefficient for wild type CFTR (CFTRwt), as only 20 -50% of nascent polypeptides mature beyond the ER (16,17). Studies of the major mutation, ⌬F508, have revealed near complete degradation of the core-glycosylated mutant protein such that no mature protein is detectable under normal growth conditions (16 -19).
A number of the components of the ER and associated proteins have been implicated in monitoring the production and elimination of misfolded wild type (wt) and mutant CFTR. Prolonged interactions between CFTR⌬F508 and two chaperones, Hsp70 and calnexin, have been shown to occur in comparison to wt (20,21). More recently, it has also been demonstrated that Hsp40 and Hsp90 contribute to the folding of CFTR (22)(23)(24). Whether the interactions of these molecular chaperones prevent the removal of specific forms of incompletely folded polypeptides or help to promote degradation is not known, although it was noted that interruption of Hsp90 binding seems to lead to accelerated degradation of both the wt and ⌬F508 forms (22).
Misfolded proteins are commonly targeted for degradation via the lysosomal proteolytic system or the cytoplasmic degradative system involving the proteasome (25, 26). Lysosomal degradation does not appear to contribute substantially to the degradation of either core-glycosylated wt or ⌬F508 CFTR (16). Proteasomal degradation of most proteins is dependent on the covalent attachment of multiple ubiquitin molecules and elim-* This work was supported by the Canadian Cystic Fibrosis Foundation and NIDDK, National Institutes of Health Grant SCOR-DK49096. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ination by the 26 S proteolytic complex (for review see Ref. 27). Proteasome inhibitor sensitivity and the detection of ubiquitin CFTR conjugates has demonstrated that immature wt and ⌬F508 CFTR are degraded, at least in part, by the ubiquitinproteasome pathway (28,29).
Although the observations with the major mutation outline the significance of CFTR maturation and provide an explanation for the absence of mutant protein at the cell surface, many aspects of CFTR biogenesis remain unknown. Further, although many CF mutations show trafficking problems (30 -33), few have been examined in detail. To gain additional insight into the determinants of CFTR expression we have analyzed the production and processing of a series of CF-associated NBD1 missense mutations and have identified a group that show defects that appear distinct from CFTR⌬F508. The identified pathway also appears with CFTRwt and thus highlights a mechanism that contributes to the notable inefficiency of CFTR maturation.

EXPERIMENTAL PROCEDURES
Construction of CFTR Expression Plasmids-The CFTR mutants A455E, S549R, P574H, and Y563N were generated from pBQ6.2 (34) as described previously (35). To generate carboxyl-terminal HA (CHA)-CFTRwt, polymerase chain reaction mutagenesis (36,37) was carried out with the oligonucleotides C1-1Z (5Ј-ATGACTGTCAAAGATCTCAC) that corresponds to native CFTR sequence and CHArev (5Ј-CGCGTA-CGGCCGTTACAAGCTAGCGTAGTCTGGGACGTCATATGGGTACA-TAAGCCTTGTATCTTGCAC) to incorporate the hemagglutinin (HA) epitope (YPYDVPDYA) inframe, prior to the natural stop codon (amino acid 1480) of the CFTR cDNA. All polymerase chain reaction-generated fragments were confirmed by sequencing after insertion into recipient vectors. Expression vector (pCMV) constructions and CHA-CFTR mutants were generated by shuttling appropriate restriction fragments between untagged and tagged cDNA versions and confirmed by sequencing or restriction digestion analysis.
Cell Culture and Transfections-HEK293, COS-7, and CHO-duk Ϫ cells were grown at 37°C with 5% CO 2 in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (FBS) and antibiotics. IB3-1 cells were grown under the same conditions in LHC-8 medium containing 10% FBS and antibiotics. Subconfluent cells grown in 100-mm plates were transfected according to the SuperFect reagent protocol provided by the manufacturer (Qiagen), using 10 g of the appropriate CFTR expression plasmid, 250 ng of pCMV ␤-galactosidase reporter plasmid, and 30 l of the SuperFect reagent in 3 ml of medium containing FBS. After an 8-h incubation at 37°C with 5% CO 2 , the cells were rinsed with PBS and incubated for an additional 16 h in normal culture medium prior to harvest or as indicated in individual experiments.
Cell Lysis, Deglycosylation, Electrophoresis, and Immunoblotting-Transfected cells were washed with ice-cold PBS and solubilized in 2 ml of RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 0.5% deoxycholate, 1 mM phenylmethylsulfonyl fluoride, and 1 mg/ml trypsin inhibitor) for 20 min on ice. Following centrifugation at 15,000 ϫ g for 15 min at 4°C, the supernatant was removed and used for immunoblot analysis or immunoprecipitation. Protein concentrations were determined (Bradford assay; Bio-Rad) using BSA as a standard, and the total protein concentrations were adjusted to 1 mg/ml. 25 g of protein were loaded on each gel lane for the Western blots. To assess the amount of RIPA insoluble CFTR, the insoluble remains were resuspended in 90 l of RIPA (minus SDS) and treated with 100 units of DNase (Roche Molecular Biochemicals) for 15 min at room temperature followed by incubation in the presence of 1% SDS for 15 min at 55°C. The SDS concentration was then diluted to 0.1% with RIPA (minus SDS), and any remaining insoluble material was removed by centrifugation. Proteins from equal proportions of cells were used in the comparison of the pellets to the supernatants.
Endoglycosidase H and N-glycosidase F (New England Biolabs) digestions were performed as specified by the manufacturer. In brief, 100 g of total protein was digested with 1000 units of enzyme for 2 h at 37°C. The deglycosylated protein was then prepared for SDS-PAGE (with 50% reaction loaded per lane) by precipitating with trichloroacetic acid and sodium deoxycholate as described (38).
Metabolic Labeling, Proteasome Inhibition, and Immunoprecipitation-Metabolic labeling and immunoprecipitation were carried out essentially as described (21). Cells transfected on 100-mm plates were split 24 h post-transfection onto four 60-mm plates and grown for an additional 24 h. After incubation in methionine-and cysteine-free ␣-MEM for 30 min at 37°C, the transfected cells were pulsed in the same medium containing 140 Ci/ml [ 35 S]methionine and [ 35 S]cysteine (Ͼ1000 Ci/mmol; Amersham Pharmacia Biotech) for 15 min at 37°C. After the pulse period, cells were washed twice with growth medium and chased at 37°C in complete ␣-Minimal essential media supplemented with 10% FBS for the times indicated.
Proteasome inhibitor studies were performed using 50 M each of MG-132 and ALLN (Calbiochem) and 10 M lactacystin (Calbiochem). HEK293 cells transfected with the indicated CFTR expression constructs were preincubated with or without inhibitors for 60 min and metabolically labeled in the presence or absence of the inhibitors as described above.
Metabolically labeled CFTR was isolated by immunoprecipitation with M3A7 and L12B4 antibodies as a mixture as described previously (16) or individually. Immunoprecipitated proteins were separated by SDS-PAGE as above; the gels were then fixed in 10% acetic acid and 40% ethanol, soaked for 30 min in Amplify (Amersham Pharmacia Biotech), and dried. Labeled proteins were visualized by exposing the gels to Kodak BIOMAX MR film at Ϫ70°C for 1-3 days. The radioactivity associated with the immunoprecipitated CFTR bands was quantified using a PhosphorImager with ImageQuant software (Molecular Dynamics).
Indirect Immunofluorescence of Epitope-tagged CFTR-COS-7 cells transfected with CHA-CFTR were split 24 h post-transfection and cultured on 12-mm circular coverslips in Dulbecco's modified Eagle's medium supplemented with 10% FBS and antibiotics for an additional 24 h at 37°C. The cells were fixed and permeabilized in 4% paraformaldehyde and 0.1% Triton X-100, respectively, for 30-min durations at room temperature. Nonspecific binding sites were blocked with 1% BSA in phosphate-buffered saline (PBS-BSA) for 30 min, and the cells were subsequently incubated with HA.11 and diluted 1:1000 in PBS-BSA for 1 h at room temperature. Cells were washed with PBS-BSA and incubated with 1:1000 donkey anti-mouse fluorescein isothiocyanate-conjugated secondary antibody (Jackson ImmunoResearch Laboratory Inc.) for 1 h at room temperature. Cells were washed twice with PBS-BSA, once with PBS, and mounted with Vectashield mounting medium (Vector Laboratories). Indirect immunofluorescence was examined with a Nikon E1000 fluorescence microscope. Images were captured using a CCD camera and IPLab Spectrum software (Scanalytics).

Biosynthetic Processing of Epitope-tagged CFTR-N-Linked
oligosaccharide modification of CFTR can be followed by the rapid conversion of the nascent polypeptide (band A) to a coreglycosylated form (band B) with subsequent conversion to a complex-glycosylated mature form (band C) (18). Complex glycosylation reflects the trafficking of core-glycosylated CFTR through the cis-medial-Golgi to the plasma membrane. Band B and the predominant band C forms of CFTRwt are recognized in both heterologous and endogenous expression systems. Many CFTR polypeptides with CF-associated mutations fail to undergo correct processing such that only the core-glycosylated forms are identified (30 -33, 42). To rapidly assay the fate of mutants and analyze their patterns of degradation, a HA epitope for high affinity monoclonal antibodies was introduced into the coding region of CFTR at the carboxyl terminus (Fig.  1A). A similarly positioned epitope has been introduced by others to achieve large scale purification of CFTR leading to functional chloride channels in reconstituted systems (43).
Immunoblot analysis was used to directly compare the glycosylation states of wt and mutant CFTR versions (Fig. 1B) to those with the CHA epitope (Fig. 1C). At steady state, the level of expression of bands B and C in HEK293 cells was comparable for tagged and untagged CFTRwt. Analysis of the S549R mutant showed measurable but intermediate levels of band C, whereas A455E, Y563N, and P574H mutants showed markedly reduced levels using both tagged and untagged CFTR. All mutants were expressed at a high level as revealed by the prominent band B. As expected, both CFTR⌬F508 and CHA-⌬F508 failed to be expressed as the complex glycosylated form (18).
We also examined the subcellular distribution of the CHA-CFTRs in COS-7 cells using indirect immunofluorescence microscopy (Fig. 2). Predominant levels of CHA-CFTRwt appear at the cell surface, in contrast to CHA-⌬F508 and CHA-A455E, as described in alternate reports with untagged versions (18,44). The mutants reside intracellularly, with a reticular, ERlike distribution. Consistent with the observed steady state glycosylation pattern, CHA-S549R shows both cell surface and ER-like staining. These results affirm that wt and mutant CHA-tagged CFTRs reflect the behavior of their untagged counterparts.
Identification of Carboxyl CFTR Derivatives-In addition to full-length CFTR forms, both the M3A7 and the anti-HA antibodies revealed products with relative molecular masses of 105 and 90 kDa that were most prominent in detergent extracts of HEK293 cells expressing the A455E and S549R mutants (Fig. 1, B and C, lanes 4 and 5). These polypeptides were detected at lower steady state levels in cells expressing CFTRwt or CHA-CFTRwt (lanes 2) and were negligible for CFTR⌬F508 or CHA-⌬F508 (lanes 3). The placement of the epitope tag establishes that the fragments contain the carboxyl terminus with an apparent molecular mass that would include at least the complete R domain, the second series of transmembrane spanning segments, and the second nucleotide binding domain (NBD2) (Fig. 1A). The results with the monoclonal antibody M3A7 (Fig. 1B) are consistent with this interpretation as its epitope resides in NBD2 (40).
Three sets of experiments were designed to confirm that the identified and presumed degradation intermediates do not reflect a peculiar facet of overexpression in HEK293 cells. The first involved an analysis of the extract preparation to examine the compartmentalization of CFTR, the second involved analysis of the intermediates at reduced expression levels with the A455E mutant, and the third involved expression in alternate cell types. CFTRwt is solubilized and can be extracted with RIPA buffer containing a mixture of ionic and non-ionic detergents (16,21). Because the formation of protein aggregates may differ between mutants (45), we used a high concentration of SDS to extract the RIPA insoluble polypeptides. Immunoblot analysis indicated that less than 10% of total immunoreactive band B or carboxyl-terminal polypeptides were present in the original CHA-A455E pellet (Fig. 3A). In addition, overexposure of an immunoblot revealed that CHA-CFTRwt and ⌬F508 pellets did not contain appreciable levels of either the 105-and 90-kDa bands (Fig. 3B). The occurrence of the 90-kDa band, most prominent with the A455E mutant, does suggest that it is relatively less soluble.
A series of parallel transfections were carried out using the standard conditions described for Fig. 1 and using 1/10 and 1/20 the amount of the CHA-A455E expression plasmid (Fig. 3C). A prolonged immunoblot exposure was necessary to detect the expressed CFTR polypeptides at the lower plasmid levels, but the occurrence and steady state proportions of the 105-and 90-kDa fragments and band B remained consistent at all levels of expression. Together, these results confirm that solubility differences between the CFTRs and overexpression do not account for the varied accumulation of the carboxylterminal fragments.
Transient transfections were carried out with CHA-tagged CFTRwt, ⌬F508, and A455E in COS-7, CHO-duk Ϫ , and CF bronchial epithelium IB3-1 cell lines. Immunoblot analysis of detergent extracts indicated that although the levels varied between the cell types, CHA-A455E consistently revealed the most prominent accumulation of the 105-and 90-kDa bands (Fig. 4, A-C). The 90-kDa band was difficult to detect in the detergent extracts of CHO-duk Ϫ and IB3-1 cells reflecting both a lower steady state level and the reduced solubility of this fragment (Fig. 4, B and C, and data not shown). Consistent with the observations made in HEK293 cells, reduced levels of the intermediates were identified for CFTRwt in all cell types. Accordingly, immunoblot analysis of CHA-⌬F508 did not readily reveal the carboxyl-terminal fragments; they could be detected but only with prolonged exposure of the blots. To further examine the prevalence of the intermediates in the IB3-1 cell line, untagged wt and mutant CFTR constructs were transfected and metabolically labeled with [ 35 S]methionine and [ 35 S]cysteine for 3 h (Fig. 4D). The increased sensitivity clearly shows the accumulation of the 105-kDa intermediate in both the wt and A455E extracts. These results are consistent with the steady state immunoblot analyses and suggest that the 105-and 90-kDa fragments reflect a feature of CFTR biogenesis that is bypassed by the major mutant because of its specific folding defect and degradation fate (16 -18, 20). Together, these data suggest that the observed carboxyl fragments reflect common processes in multiple cell types including those of epithelial origin and that differences in accumulation of the carboxyl fragments between CHA-CFTRs are intrinsic to the polypeptide being expressed.
Analysis of Biogenesis and Turnover of Untagged Wild Type and Mutant CFTR-Given their carboxyl-terminal constitution, the 105-and 90-kDa immunoreactive polypeptides were suspected to be degradation products, although it was not immediately apparent when or how they were generated. CFTR degradation events may occur during the folding of the nascent polypeptide at the ER, upon modification in the Golgi, during vesicular transport to the cell surface and/or subsequent to its incorporation into the plasma membrane. Rapid turnover of band B and band C forms have been shown (14,16,17). We therefore analyzed the intermediates using metabolic pulsechase analyses to determine appearance and turnover kinetics.
CFTRwt is initially synthesized as a 140-kDa core-glycosylated, immature protein and is inefficiently converted to the 160-kDa complex-glycosylated, mature protein (16 -18). In HEK293 cells (Fig. 5), using a 15-min pulse, mature protein became evident only after a minimum chase period of 1 h (left panel). In contrast, no mature forms were detected for the A455E mutation (right panel) or with prolonged pulse and/or chase periods (data not shown). The 105-and 90-kDa carboxylterminal fragments were evident for both CFTRwt and A455E as early as the completion of the pulse (15 min) and rapidly disappeared during the chase. These untagged proteins were detectable with a CFTR antibody mixture (M3A7 and L12B4) and were consistently most prominent for the A455E mutant. Detection within the pulse period of 15 min clearly reveals that the carboxyl-terminal fragments result from an immature CFTR intermediate as opposed to a mature band C form.  (16,17) emphasizing that this fragment is susceptible to proteolysis.
Analysis of the Glycosylation State of the Carboxyl-terminal Fragments-To further characterize the 105-and 90-kDa fragments we examined their glycosylation status via glycosidase digestion of whole cell lysates. Results with CHA-S549R at steady state expression are shown (Fig. 6) as the 105-and 90-kDa intermediates as well as both core-and complex-glycosylated full-length protein forms are detectable. The treatment of CHA-S549R with N-glycosidase F (left panel) shifted both complex-glycosylated (band C) and core-glycosylated (band B) CFTR to the unglycosylated form (band A), as expected. Digestion with endoglycosidase H (right panel) removed only the high mannose-type asparagine-linked glycans from the band B form of CHA-S549R with minimal effect on band C forms of the protein (18,46). Both enzymes shifted the molecular mass of the 105-kDa fragment to 90 kDa, whereas the 90-kDa fragment remained unaltered and showed a corresponding increase in abundance (left and right panels). Epitope-tagged versions of wt and the A455E mutant gave consistent results, as did radioactively labeled and immunoprecipitated untagged CFTR polypeptides (data not shown). These investigations reveal that the 105-and 90-kDa carboxyl-terminal intermediates represent core-glycosylated and unglycosylated forms, respectively, of the same CFTR fragment. As CFTR has only two adjacent N-glycosylation sites between the sixth and seventh transmembrane segments, this implies correct ER membrane orientation and insertion of their precursor form and suggests that a cytoplasmic degradation system would be responsible for their generation.
Effect of Proteasome Inhibitors on the Generation and Turnover of the 105-and 90-kDa Intermediates-The earliest steps in the recognition and removal of misfolded membrane proteins from the ER are largely unknown. Constituents of the ER lumen may be responsible but there is growing evidence that ATP binding cassette family members, including CFTR, are targeted for proteolysis by the prevalent cytoplasmic degrada-tive system using the proteasome (28,29,47). To investigate the role of the 26 S proteasome in the generation of the 105and 90-kDa intermediates, we performed metabolic labeling analyses in the absence or presence of various proteasome inhibitors (Fig. 7). MG-132, a reversible and cell-permeable proteasome inhibitor, has been shown to block the conversion of CFTR to a maturation competent form (29). The results of Fig. 7A show that immature CFTRwt (band B) and the 105and 90-kDa fragments were detected in the absence or presence of inhibitor, whereas mature CFTRwt was detected only in the absence of MG-132, as predicted, after the 3-h chase period. 50 M MG-132 did not effect the presence of the 105-and 90-kDa intermediates in HEK293 cells.
Subsequently, we examined the influence of MG-132, ALLN, and lactacystin on CFTRA455E degradation, because different peptide aldehyde inhibitors have distinct effects on CFTR processing (28,29). We have shown (Fig. 5) that the 105-and 90-kDa CFTR fragments appear early in biogenesis and therefore analyzed CFTRA455E biosynthesis in the absence or presence of inhibitors using a 15-min pulse-labeling period. RIPA soluble CFTR peptides were immunoprecipitated demonstrating that the inhibitors had no effect on the prevalence of the 105-and 90-kDa polypeptides formed during A455E biosynthesis (Fig. 7B). The accumulation of a 60-kDa band in the presence of the inhibitors likely reflects the degradation of alternate CFTR intermediates.
To investigate the role of proteasomal degradation in the turnover of these polypeptides, we performed a pulse-chase analysis in the absence or presence of inhibitors (Fig. 8). HEK293 cells expressing CFTRA455E were pulse-labeled for 15 min and chased for up to 2 h. In the absence of inhibitors (Ϫ lanes), band B and the 105-and 90-kDa intermediates were rapidly degraded over the duration of the chase. The presence of MG-132, ALLN, and lactacystin (ϩ lanes) resulted in a general increase in radioactive label incorporated but no change in the half-lives of the 105-or 90-kDa polypeptides occurred. These results are consistent with those obtained with the wt protein and demonstrate that the 26 S proteasome does not contribute to the generation or turnover of the identified degradation intermediates.

Carboxyl Fragments Reflect Early Recognition at the ER
Membrane-The carboxyl-terminal fragments that have been identified reveal remnants of a feature of early CFTR biogenesis that limits the delivery of CFTR to the cell surface. The timing of appearance, localization, and glycosylation features indicate that the process is active when CFTR is associated with the ER. It is imposed on the wt protein but is more prominent in the presence of missense mutations such as A455E or S549R. These altered amino acids are both within NBD1 but are separated by nearly 100 residues so it is unlikely FIG. 5. Carboxyl-terminal fragments appear early in biogenesis. Transfected cells were metabolically labeled with radioactive amino acids using a 15-min pulse and chase times as indicated. Cells were lysed, and CFTR was immunoprecipitated with a mixture of the M3A7 and L12B4 antibodies. Proteins were separated by SDS-PAGE and visualized by fluorography. Identical results were obtained by immunoprecipitation with the anti-HA antibody (not shown). The relative levels of the 105-kDa band with CFTRwt, A455E, and ⌬F508 were determined by phosphorimage analysis. The percentage of radioactivity incorporated was calculated as the counts of the 105-kDa band over the sum of the counts of band B plus the 105-kDa band from five experiments. The half-life of the B and 105-kDa bands were determined by monitoring their rate of disappearance over time (16).
FIG. 6. Deglycosylation analysis of the degradation bands. RIPA soluble protein extracts of CHA-S549R were treated with Nglycosidase F (N-Glyc F) and endoglycosidase H (Endo H). Immunoblot analysis using the anti-HA antibody as described revealed that the 105-kDa band is the core glycosylated form of the 90-kDa band. that specific cis proteolytic sites have been directly affected. Further, it is evident that for enhanced proteolytic activity, the mutated amino acid need not be present in the resulting carboxyl fragments as size estimation would preclude this for at least the A455E mutant. Our findings are consistent with the subjection of CFTR to a "quality control" surveillance mechanism that eliminates improperly or unfavorably folded protein (48). A number of conformational changes are imposed on nascent ER membrane-bound polypeptides during the folding of the individual domains, the assembly of interdomain interactions, or during post-translational modification. We predict that the mutations lead to an increased fraction of folding intermediates or of blocked conformations that are recognized and subsequently removed by a proteolytic mechanism that leads to the carboxyl remnants.
Newly synthesized proteins including apolipoprotein B100 have been shown in vivo to be exposed to the cytosol and targeted to the proteasome co-translationally (49). This is presumably related to the exposure of sequences during folding and/or membrane dispersion. Given evidence of the effects of proteasome inhibitors on immature wt or ⌬F508 mutant CFTR (28,29) and in vitro studies suggesting that ubiquitination of CFTR can occur co-translationally (50), we hypothesized that the proteasome may be directly involved in the process leading to the formation of the carboxyl fragments. Our pharmacological studies, however, do not support a role for the major cytoplasmic degradation system in either their formation or direct elimination and thus imply that an alternate mechanism is responsible. We cannot determine whether the mechanism employs the proteasome at later stages of degradation, subsequent to the loss of identifying epitopes. The protease involved is unknown but the absence of an effect with ALLN would also indicate that it is not a neutral cysteine protease.
Degradation of misfolded ER membrane proteins via a cytoplasmic pathway would require retrograde transport from the membrane of the ER back into the cytosol (51). To enable such processes, the Sec61 membrane protein complex, which is the major constituent of the co-translational protein translocation machinery, may be utilized (51)(52)(53). Recent evidence suggests that Sec61␤, a subunit of the Sec61 complex, does interact with CFTR and may participate in the elimination of both wt and ⌬F508 CFTR (46). It would therefore be interesting to closely examine its interaction with the S549R and A455E CFTR mutants.
Distinction of the ⌬F508 Mutation and Formation of the CFTR Carboxyl Fragments-The degradation process and fate of the carboxyl fragments appear interesting from several perspectives. Foremost, the formation of the carboxyl fragments from nascent wt and mutant CFTR indicates a fate distinct from nascent ⌬F508 molecules. The carboxyl fragments were difficult to detect for CFTR⌬F508 in all cell lines tested. The proximity of the amino acid deletion to a proteolytic site may account for this; however, analysis of cells expressing CFTR versions with both the ⌬F508 mutation and either S549R or A455E mutations retain the formation of prominent levels of the carboxyl-terminal fragments (data not shown). It is possible that for the major mutation, discrete conformational intermediates or aggregates form with consequent burial of the proteolytic site or sites such that the generation of the carboxyl polypeptides is diminished unless provoked by an alternate mutation.
The nature or number of steps of the proteolytic process involved in the formation of the carboxyl fragments remains unknown. The placement of the carboxyl-terminal epitope tag together with the sites of N-glycosylation provide two definitive inclusion features for the observed fragments, but the large size and the distorted migration of the immature CFTR polypeptides in SDS-PAGE limits a precise determination of the amino end. The observed sizes do indicate that at least one cleavage occurs within the latter portion of NBD1 or at the NBD1-R domain boundary and would thus be predicted to occur on the cytoplasmic face of ER-associated CFTR. Predicted proteolytic cleavage sites for general proteases do occur within this region, but preference of one over another is not readily rationalized.
Mutations and Complexity of CFTR Maturation-Our results indicate the specificity of the processes involved in the folding pathway of the cytosolic domains but also outline the complexity of the effects of individual mutations on CFTR biogenesis. In attempting to correlate the abundance of the carboxyl degradation bands with the formation of band C, it was evident from the S549R mutation that moderate levels of complex-glycosylated CFTR can be generated even though the production of 90-and 105-kDa products is prominent. We interpret this to indicate that conformational or folding defects that manifest very early are not necessarily problematic at later maturation stages. This is consistent with the findings of the wt protein but less discernible for some mutations such as A455E.
The A455E, Y563N, and P574H mutations do appear to be able to achieve at least nominal levels of chloride conduction at the cell surface based both on the presenting phenotype in CF patients (54) and on single channel and whole cell current measurements (55) in heterologous expression systems. In contrast to the Y563N and P574H mutations, where low levels of mature band C forms were detectable with long exposure and/or long label incorporation times in HEK293 cells (data not shown), fully glycosylated protein could not be detected for A455E using our assay systems. Of the NBD1 missense mutations surveyed, A455E degradation products appeared at the highest levels relative to full-length nascent CFTR such that immature band B may be too low to lead to detectable levels of band C. It is not without interest that channel activity has been measured with the carboxyl half of the CFTR molecule (12); however, there was no evidence to support that the carboxyl fragments can be delivered to the cell surface based on immunofluorescence (Fig. 2). Further, the half-lives of the 105-and 90-kDa bands are comparable to that of the immature band B such that very limited rescue of activity would be anticipated.
In summary, we have used both steady state and metabolic pulse-chase analysis with heterologous expression systems to reveal the appearance and rapid turnover of carboxyl-terminal fragments of CFTR in four cell types including bronchial epithelial cells. We predict that these fragments correspond to the proteolysis of unstable folding intermediates of immature CFTR. Their prominence is enhanced with missense mutations in NBD1 but are notably reduced with the major CF mutation, ⌬F508, emphasizing distinction in the manifestation of CFassociated trafficking mutations. The identified process would appear to contribute to the overall inefficient maturation of wild type CFTR, whereas increased formation of these fragments because of NBD1 missense mutations would likely also contribute to the clinical condition of CF patients.