Interference with Ubiquitination in CFTR Modifies Stability of Core Glycosylated and Cell Surface Pools

Abstract

It is recognized that both wild-type and mutant CFTR proteins undergo ubiquitination at multiple lysines in the proteins and in one or more subcellular locations. We hypothesized that ubiquitin is added to specific sites in wild-type CFTR to stabilize it and at other sites to signal for proteolysis. Mass spectrometric analysis of wild-type CFTR identified ubiquitinated lysines 68, 710, 716, 1041, and 1080. We demonstrate that the ubiquitinated K710, K716, and K1041 residues stabilize wild-type CFTR, protecting it from proteolysis. The polyubiquitin linkage is predominantly K63. N-tail mutants, K14R and K68R, lead to increased mature band C CFTR, which can be augmented by proteasomal (but not lysosomal) inhibition, allowing trafficking to the surface. The amount of CFTR in the K1041R mutant was drastically reduced and consisted of bands A/B, suggesting that the site in transmembrane 10 (TM10) was critical to further processing beyond the proteasome. The K1218R mutant increases total and cell surface CFTR, which is further accumulated by proteasomal and lysosomal inhibition. Thus, ubiquitination at residue 1218 may destabilize wild-type CFTR in both the endoplasmic reticulum (ER) and recycling pools. Small molecules targeting the region of residue 1218 to block ubiquitination or to preserving structure at residues 710 to 716 might be protein sparing for some forms of cystic fibrosis.

INTRODUCTION

The most common disease-causing mutation in CFTR is a deletion of F508 (F508del CFTR), resulting in a protein that is misfolded, ubiquitinated, and degraded prematurely by the endoplasmic reticulum-associated degradation (ERAD) system. F508del CFTR can be rescued by enhancing cellular chaperone functions through reduced temperature or small molecules such as 4-phenylbutyrate (1). However, the rescued F508del CFTR mutant undergoes accelerated retrieval or recycling from the plasma membrane (2, 3) and degradation in the lysosome. Therefore, F508del CFTR requires at least two different stabilizing mechanisms to become accessible to a potentiator that can activate chloride transport.

Posttranslational modification can occur at multiple intracellular sites and modify the fate of native and damaged proteins. Typically, covalent conjugation of one or more ubiquitins (Ubs) to misfolded proteins directs protein degradation through the ERAD system (4, 5). There are seven potential lysines on ubiquitin through which specific linkages are associated with specific signaling pathways. For example, K48-linked polyubiquitinated proteins tend to be directed to proteasomal degradation, whereas K63-linked polyubiquitination plays a role in regulation of the protein recycling process on the cell surface (6). We showed that imposing monoubiquitination on wild-type (wt) and F508del CFTR leads to increased preservation of band B core glycosylated CFTR (7). Ubiquitination of CFTR is important for its regulation and quality control process (8–10), including ubiquitin-mediated proteasomal degradation (8, 9) and recycling (11, 12). However, the ubiquitination sites on wt CFTR, mechanism of degradation, and regulation of recycling are largely unknown. Here, we identify the specific lysine residues in wt CFTR that regulate maturation and trafficking of wt CFTR.

MATERIALS AND METHODS

Materials.

M3A7 mouse monoclonal antibody to C-tail amino acids 1370 to 1380 (Millipore), H-182 rabbit polyclonal antibody to N-terminal amino acids 1 to 182 (Santa Cruz Biotechnology), CF3 mouse monoclonal antibody to first extracellular loop amino acids 103 to 117 (Abcam), and 169 rabbit polyclonal antiserum to the regulatory (R) domain (13) were used for immunoblotting, immunoprecipitation, and immunofluorescence of CFTR. Anti-Ub and anti-K63 Ub polylink were purchased from Cell Signaling, and Na/K ATPase was purchased from Millipore for immunoblotting.

Cell culture.

CFTR cDNA vectors were studied in immortalized human airway epithelial cells (IB3-1; F508del/W1282X heterozygous mutation in CFTR) (14), which were cultured in LHC-8 medium (Gibco, Grand Island, NY) containing 10% fetal bovine serum (FBS) and 1% antibiotic/antimycotic at 37°C in 4% CO₂.

Site-directed mutagenesis.

Site-directed mutagenesis was performed by PCR with the forward and reverse primers listed in Table 1; PCR products were treated with DpnI to digest template plasmid and transformed into JM109 (Promega). Selected colonies were amplified, and purified plasmids were sequence confirmed.

TABLE 1.

Primers for site-directed mutagenesis

Mutation	Directiona	Sequence (5′–3′)b
K14R	F	GGCCAGCGTTGTCTCCAGACTTTTTTTCAGCTGGACC
	R	GGTCCAGCTGAAAAAAAGTCTGGAGACAACGCTGGCC
K68R	F	GGCTTCAAAGAAAAATCCTAGACTCATTAATGCCCTTCGGCG
	R	CGCCGAAGGGCATTAATGAGTCTAGGATTTTTCTTTGAAGCC
K710R	F	CCAATCAACTCTATACGAAGATTTTCCATTGTGCAAAAG
	R	CTTTTGCACAATGGAAAATCTTCGTATAGAGTTGATTGG
K716R	F	GAAAATTTTCCATTGTGCAAAGGACTCCCTTACAAATGAATGG
	R	CCATTCATTTGTAAGGGAGTCCTTTGCACAATGGAAAATTTTC
K710/716R	F	CCAATCAACTCTATACGAAGATTTTCCATTGTGCAAAGGACTCCCTTACAAATGAATGG
	R	CCATTCATTTGTAAGGGAGTCCTTTGCACAATGGAAAATCTTCGTATAGAGTTGATTGG
K1041R	F	CCTCACAGCAATTCAGACAACTGGAATCTGAAG
	R	CTTCAGATTCCAGTTGTCTGAGTTGCTGTGAGG
K1080R	F	GAAACTCTGTTCCACAGAGCTCTGAATTTACATAC
	R	GTATGTAAATTCAGAGCTCTGTGGAACAGAGTTTC
K1218R	F	GATCTCACAGCAAGATACACAGAAGG
	R	CCTTCTGTGTATCTTGCTGTGAGATC

^aF, forward; R, reverse.

^bThe codons at the sites for K-to-R mutations are underlined.

Cell culture and treatment.

Wild-type or mutant CFTR cDNA constructs were transfected in IB3-1 cells by using transIT-2020 (Mirus) for 48 h. The proteasomes or lysosomes were inhibited by using MG132 (25 μM for 12 h) or a combination of the inhibitors E64d and pepstatin A (25 μg/ml for 12 h), respectively. To study the stability of CFTR, at 2 days posttransfection, protein synthesis was inhibited by treatment with 100 μg/ml of cycloheximide (CHX) for increasing times. Cell surface-localized CFTR was isolated by surface biotinylation as described previously (15).

Confocal analysis.

Cell surface CFTR was detected by confocal microscopic analysis. IB3-1 cells (1 × 10⁵) were grown on microscope cover glasses and transfected with wild-type or mutant CFTR constructs. At 2 days posttransfection, cells were fixed in 4% paraformaldehyde and blocked with 3% bovine serum albumin (BSA) without permeabilization to prevent antibody penetration below the plasma membrane. The cover glasses were incubated with primary antibody against the first extracellular loop of CFTR (CF3; Abcam) and with secondary antibody labeled with Alexa Fluor 568 (Invitrogen). Imaging was performed on a Zeiss LSM 510 instrument using a 40×/1.3-numerical-aperture objective.

MS.

Two different cell lines and institutional mass spectrometry (MS) core laboratories were used to analyze native CFTR. At the University of Alabama at Birmingham, single-cell clonal cultures of human CFTR with 10 copies of a histidine tag, expressed in an N-acetylglucosaminyltransferase I-deficient HEK293S cell line, were expanded and analyzed for CFTR expression and activity as described previously (16). His-CFTR was isolated using nickel affinity chromatography, followed by size exclusion chromatography and SDS-PAGE for Western blotting and Coomassie staining. The CFTR band was excised and proteolytically digested by trypsin, and aliquots were loaded onto a C₁₈ NanoCHIP column. Eluted peptides were passed into a modified ion spray interface of a 5600 TripleTOF (time of flight) instrument for tandem mass spectrometry (MS/MS) or multiple reaction ion monitoring (MRM)-MS analysis. Resulting spectra were processed, and potential peptides were identified using AB Sciex Protein Pilot software. Search parameters were set against the NCBInr Homo sapiens database. Potential modified peptides were further evaluated using Protein Pilot and PeakView software (AB SCIEX).

At the Johns Hopkins University School of Medicine, we immunopurified CFTR protein from wild-type CFTR-transfected IB3-1 cells to compare lysine modification of CFTR in bronchial epithelial cells with that of the HEK293S cells. Wild-type CFTR was immunoprecipitated by using 169 rabbit anti-CFTR antibody from wild-type CFTR-transfected IB3-1 cell lysates. The protein was treated with peptide N-glycosidase F (PNGase F) for deglycosylation, and CFTR was separated by SDS-PAGE. The band corresponding to CFTR was excised from a SimplyBlue-stained gel and further subjected to trypsin digestion. Peptides were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) using a linear trap quadrupole (LTQ) Orbitrap Velos MS interfaced with a two-dimensional (2D) nano-LC system. Peptide sequences were identified using Proteome Discoverer (Thermo Scientific, Cambridge, MA) and MASCOT (Matrix Science, Inc., Boston, MA) software. Scaffold (Proteome Software, Portland, OR) software was used to validate protein and peptide identification.

Limited proteolysis of wild-type or mutant CFTR.

Wild-type or mutant CFTR was immunoprecipitated by using 169 rabbit anti-CFTR antibody and protein G beads (GE Healthcare) from CFTR-transfected cell lysates. Proteins were eluted by the addition of 0.5 ml of 0.1 M glycine, pH 2.5, to the tube containing equal volumes of 1 M Tris-HCl, pH 8.0, and radioimmunoprecipitation assay (RIPA) buffer. Protein solution was concentrated, and buffer was exchanged into RIPA buffer by using centrifugal filters (100-kDa cutoff value; Millipore). Sequencing-grade trypsin (Promega) was prepared in the following concentrations: 200, 50, 12.5, and 3.125 μg/ml. Five microliters of trypsin solution was incubated with 10 μl of CFTR solution at 37°C for 30 min. Final trypsin concentrations were 66, 16.7, 4, 1, and 0 μg/ml. Undigested CFTR proteins were detected by immunoblotting.

Computer-modeled structure of CFTR.

The N tail and other regions of CFTR were modeled by i-Tasser (17) and assembled into a previously built CFTR structure (18) by using Jmol software (http://www.jmol.org/).

RESULTS

Two lysines in the R domain of wt CFTR are ubiquitinated at steady state.

From the wt CFTR that was overexpressed in HEK293S cells and purified for mass spectrometry analysis, two lysine residues in the R domain were identified at positions 710 and 716 (K710 and K716), with a 96% confidence level and 70% sequence coverage to CFTR (Table 2 and Fig. 1; lysine residues in CFTR are shown in Fig. 2A). From the wt CFTR-transfected cystic fibrosis (CF) bronchial epithelial cells (IB3-1), the same two sites, residues K710 and K716, were detected, with a 96% confidence level and 22% sequence coverage (data not shown). We hypothesized that ubiquitin might be required at these sites to stabilize the nascent protein, and the residues were mutated to arginine (K710R or K716R and the double mutant K710R K716R [K710/716R]). Arginine interferes with posttranslational modifications, such as ubiquitination or acetylation, but leaves the gross structural backbone relatively intact. We used the i-Tasser software and website to compare the predicted structures of the R domains alone with either lysine or arginine at these positions (Fig. 3A). Interestingly, small perturbations in the predicted structure were seen with the K710R or K716R mutation alone but not when both residues were mutated in the same vector. We transiently expressed these mutants, and the fate of CFTR was analyzed by SDS-PAGE. Band A is nascent CFTR in the ER. Band B is core glycosylated CFTR in the ER. Band C is fully glycosylated CFTR that resides anywhere from the Golgi compartment to the plasma membrane and vesicles in between. At steady state and 37°C, the K710R and K716R mutations decrease the level of CFTR C band expression in whole-cell lysates, which suggests that the normally ubiquitinated K710 and K716 residues might be required to help process CFTR beyond the Golgi compartment (Fig. 2B).

TABLE 2.

Information for all spectra from three different trials^a

Residueb	Peptide	MS method	Unmodified mass (Da)	Observed m/z	y and b ion coverage
K68*	(Formyl)NPK(UGG)LINALR	MS/MS	1,179.69	590.82+	y1-y7, b2-b5, b7
K68	(Formyl)NPK(UGG)LINALR	MRM-MS	1,179.69	590.82+	y1-y7, b2-b5
K710*	K(UGG)FSIVQK(CAM)TPLQ(Dea)M(Oxi)NGIEEDSDEPLER	MS/MS	3,090.48	773.64+	y1-y10, b2-b3, b5-b11
K710	(Formyl)K(UGG)FSIVQK	MRM-MS	990.56	496.32+	y1-y6, b2-b6
K710/K716*	K(UGG)FSIVQK(UGG)TPLQM(Oxi)N(Dea)GIEEDSDEPLER	MS/MS	3,147.50	787.94+	y1, y3-y11, y13, b3-b8, b10
K716*	(Formyl)FSIVQK(UGG)TPLQMNGIEEDSDEPLER	MRM-MS	2,916.40	973.13+	y1-y6, y8-y10, b2-b7, b9-b10
K716	KFSIVQK(UGG)TPLQM(Oxi)N(Dea)GIEEDSDEPLER	MS/MS	3,033.46	750.44+	y1-y10, y12-y13, b2, b4-b8, b10-b13
K716	(Formyl)FSIVQK(UGG)TPLQMNGIEEDSDEPLER	MS/MS	2,933.40	978.83+	y1-y2, y4-y5, y7-y12, b3-b7, b9
K716	(Formyl)FSIVQK(UGG)	MRM-MS	862.47	432.22+	y1-y5, b2-b4
K1041*	(Formyl)AYFLQTSQQLK(UGG)QLESEGR	MRM-MS	2,267.14	756.73+	y1-y9, b2-b4
K1041	AYFLQTSQQLK(UGG)QLESEGR	MS/MS	2,239.12	747.43+	y1-y14, b2-b6, b8-b11
K1041	AYFLQTSQQLK(UGG)QLESEGR	MS/MS	2,239.13	747.43+	y4-y13
K1080*	(Gln→Pyro-Glu)QPYFETLFHK(UGG)	MS/MS	1,405.67	703.82+	y2-y8, b2-b6

^aAll spectra achieved a Protein Pilot confidence score of ≥98.

^bRepresentative spectra are shown in Fig. 1 for proteins marked with an asterisk.

FIG 1.

Ubiquitination sites in CFTR. Individual lysine residues (K14, K68, K710, K716, K1041, K1080, and K1218) were identified as ubiquitination sites by LC-MS/MS, MRM-MS, and/or mutagenesis. A covalent ubiquitin attachment is indicated by a 114-Da increase in mass. Full-length purified CFTR was analyzed with a 5600 TripleTOF mass spectrometer following trypsin digestion. All spectra achieved a Protein Pilot confidence score of ≥98.

FIG 2.

Mutations to block ubiquitination in key lysines in CFTR modify CFTR expression. (A) Molecular models of the relevant domains in wild-type CFTR. The N-tail region spanning residues 1 to 80 was submitted to the i-Tasser website (17) and assembled with previously built CFTR structural information (18) by using Jmol (http://www.jmol.org/). Each of the subsequent domains was visualized by using ViewerLite (Accelrys, Inc., San Diego, CA). We have marked the lysines chosen for mutation to block posttranslational modification. The table lists the K-to-R mutations and the residues spanning each domain in the model. (B) Representative SDS-PAGE gel in which whole-cell lysates were generated after expression of a vector control, wt CFTR, and the indicated K-to-R mutations in the wt CFTR backbone. This immunoblot demonstrates increased levels of band C CFTR expression from the K14R, K68R, and K1218R cDNA vectors and decreased levels of band C CFTR expression in K710R, K716R, K710/716R, and K1080R vectors. At steady state, the K1041R mutation in TM10 is almost completely lacking band C CFTR expression. Densitometry was performed (n = 5). Expression in the C band is graphed on the left, and the sum of A and B band expression as a percentage of wt C band is shown on the right. Data are means ± standard errors of the means. Differences compared to wt CFTR were considered significant (*) at a P value of ≤0.05 by an unpaired Student t test (n = 5).

FIG 3.

Mutations in key lysines in CFTR do not affect conformational structure, as determined by modeling and limited proteolysis. (A) Wild-type and mutant models were created by using i-Tasser (17). A K-to-R mutation does not alter the conformational structure of each domain except for the K68R mutant. (B) Immunoblot of trypsin-digested wild-type or mutant CFTR proteins. Proteins were incubated with increasing concentrations of trypsin at 37°C for 30 min. Concentrations of trypsin were 0, 1, 4, 17, and 66 μg/ml, respectively. (C) We surveyed the consequences of Lys-to-Arg mutations throughout CFTR and demonstrate here that most do not affect protein expression. Total protein (10 μg) from the transfected whole-cell lysate was resolved by SDS-polyacrylamide gels, and CFTR protein was detected by Western blotting using M3A7 antibody. Representative SDS-polyacrylamide gels demonstrate a lack of major changes in the expression/maturation of CFTR as a result of K-to-R mutations. The K1041R mutant was included to highlight the differences that will be explored further.

HEK293S cells expressed wt CFTR that also contained ubiquitinated lysines at K1041 in the 10th transmembrane domain (TM10) and K1080 in the fourth intracellular loop (ICL4) (Table 2 and Fig. 1). IB3-1 cells expressed wt CFTR with acetylated residues at K1284 and K1317 in nucleotide-binding domain 2 (NBD2) (data not shown). Again, we compared the predicted structures for each of the domains with a substitution of arginine for lysine (Fig. 3A). Neither substitution appears to alter the predicted structure. We generated K1041R and K1080R mutants and expressed them in IB3-1. At steady state, prevention of the ability to ubiquitinate residue 1041 almost completely eliminated CFTR expression, except for a faint band B (Fig. 2B). This suggests that 1041 is a critical residue, either for posttranslational modification or some other structural requirement. Interference with residue 1080 drastically reduced CFTR but, in contrast to the K1041R mutation, significant band C continued to appear, similar to results with the R domain mutants. C-band expression was quantified by densitometry, as shown in Fig. 2B, and K1041R was the lowest although band B expression was higher than in the wt, further supporting a maturational block after band B and before C.

Given that we did not consistently observe the full coverage of CFTR through overlapping peptides, we also generated individual K-to-R mutations in each of 87 lysines in CFTR. Most of these mutants left CFTR maturation unaltered, as assessed by banding patterns on SDS-PAGE gels, and were comparable to wt CFTR (representative examples are shown in Fig. 3C). However, several were strikingly altered. To help us understand these mutants, we constructed several structural models of different regions in CFTR in order to find lysines that might be exposed to cellular enzymes that could be modified with ubiquitin. Figure 2A shows several such models generated with i-Tasser (17) and the identities of the sites chosen for mutation. Lysine-to-arginine mutations caused minor changes in CFTR conformation, except for the K68R mutation which caused changes in the orientation of third alpha helix, where 68R is located (Fig. 3A).

Beginning with the N-tail region which is the first region to be translated, we individually mutated all seven lysines alone and then together in one construct. K14R and K68R mutations result in increased levels of immature (A/B band) and mature (C band) forms of CFTR expression (Fig. 2B). Interference with modifications of these residues improves production, suggesting that ubiquitination at either K14 or K68 might signal eventual proteolysis, possibly in a post-Golgi compartment, given the efficient maturation from band B. In addition to the lysine residues in the TM10 (K1041) and ICL4 (K1080), a site in NBD2 (K1218) was similarly critical to CFTR expression and/or maturation. The K1218R mutation increases CFTR B and C bands even more significantly than N-tail mutants (Fig. 2B). These results indicate that K1218 in the NBD2 domain negatively regulates expression/maturation of CFTR and that blocking modification at this site restores CFTR to levels greater than the wild-type level.

The observed reduction in expression of K710R, K716R, K710/716R, K1041R, and K1080R CFTRs might be the consequence of disordered conformational structure or folding/assembly. We therefore explored limited proteolysis of the wt and mutant CFTR. The data in Fig. 3B show similar patterns of proteolytic digestion for wt and mutant CFTR; K1041R is the only mutant which shows significantly less CFTR at the final trypsin concentration. These data suggest that there is not a major protein folding defect (Fig. 3B).

Proteasomal and lysosomal degradation signaled by different sites on CFTR.

Whole-cell experiments cannot discriminate the subcellular compartments within which CFTR is stabilized. To study this further, we started with the N-tail mutants, where early translational events can signal later proteasomal degradation. To test this hypothesis for K14R and K68R, we expressed these N-tail mutants in the presence and absence of proteasomal inhibition. As expected and as a positive control, treatment with MG132 increased total wt CFTR expression (Fig. 4A), but proteasome inhibition during expression of the K14R and K68R mutant CFTR cDNAs led to even higher expression levels. This supports the hypothesis that interfering with N-tail ubiquitination may allow a positive feedback loop leading to accumulation of band C and ongoing band B production. We next hypothesized that blocking N-tail lysine modifications might increase the load on lysosomal proteolysis and lead to even higher levels. We expressed the N-tail mutants in cells treated with lysosomal inhibitors. As expected, treatment with lysosomal inhibitors (E64d/pepstatin A) increased wt CFTR expression because terminal proteolysis was limiting. However, unexpectedly, expression levels in K14R- and K68R-transfected cells were not increased by lysosomal inhibition, perhaps because they were already slow to recycle. Densitometry analysis showed that a proteasomal inhibitor increased C bands of wt, K14R, and K68R compared to control levels, but changes in A/B bands were insignificant. Lysosomal inhibitor did not change C bands, but A/B bands of K14R and K68R were decreased compared to control levels. These data suggest that K14R and K68R mutations allow CFTR to bypass the lysosome for degradation so that preventing lysosomal activity does not further increase the expression level of the mutant compared to wild-type CFTR.

FIG 4.

Identification of the lysine residues in wt CFTR that are critical for proteasomal and/or lysosomal degradation. IB3-1 cells were transfected with the indicated cDNA CFTR vectors for 48 h and then treated with 50 μM proteasomal inhibitor MG132 or lysosomal inhibitor cocktail (E64d/pepstatin A, 25 μg/ml) for 12 h. CFTR was detected by immunoblotting from whole-cell lysates. Densitometry was performed (n = 5), and results are shown in the graphs on the right. Data are means ± standard errors of the means. Differences compared to wt CFTR were considered significant (*) at a P value of ≤0.05 by an unpaired Student t test. (A) The results from expression of control vector, wt CFTR, K14R, and K68R with and without proteasomal or lysosomal inhibition are shown in representative blots. MG132 increased total A/B/C bands of CFTR in wild-type, K14R, and K68R proteins. Lysosomal inhibition reduced accumulation of A/B bands without lowering the C band. (B) The results from expression of control vector, wt CFTR, K710R, K716R, and K710/716R with and without proteasomal or lysosomal inhibition are shown in representative blots, similar to the results show in panel A. MG132 increased wt CFTR band C as expected but did not affect band C of the mutants. Lysosomal inhibition led to accumulation of band C for wt and R mutants. Thus, any CFTR that could escape ERAD was rescued from the lysosome. (C) The results from expression of control vector, wt CFTR, K1041R, K1080R, and K1218R with and without proteasomal or lysosomal inhibition are shown in representative blots similar to the results show in panel A. MG132 rescues band B from K1041R to some extent, as well as a small portion of band C. Lysosomal inhibition recovers some band C. K1080R is rescued by both MG132 and lysosome inhibition. K1218R shows the most robust rescue by both. Thus, blocking modification of lysine residues at 1041 destabilizes CFTR and sends it to the proteasome, whereas blocking lysine residues at 1218 improves CFTR stability and trafficking. Blocking modification at residue 1080 limits the amount of CFTR processed to the surface, which can be improved by lysosomal inhibition.

We also expressed each of the R domain mutants (K710R, K716R, and K710/716R) during inhibition with MG132 and E64/pepstatin A. Treatment with MG132 did not significantly increase A/B/C bands in the mutants, suggesting that K710R and K716R were not able to improve trafficking just because premature degradation was blocked and/or because lysosomal degradation was accelerated (Fig. 4B). Next, treatment of cells with lysosomal inhibitors in contrast to proteasome inhibition did rescue C band expression in the R domain mutants, suggesting that nonubiquitinated K710 and K716 could be collected and stored in the surface recycling compartment if proteolysis at this late stage was blocked (Fig. 4B).

The TM10, ICL4, and NBD2 domain mutants were also expressed with proteasomal or lysosomal inhibitors. MG132 treatment increased A/B band expression in wild-type and mutant forms of CFTR. However, there was no significant increase in C band expression in these mutant forms of CFTR compared to the wild type (Fig. 4C). Interestingly, MG132 treatment rescued only B band expression in the K1041R mutant form of CFTR, but this was not sufficient for trafficking beyond the ER. K1080R and K1218R mutants displayed both bands B and C after proteasomal inhibition. This supports the hypothesis that the K1080R and K1218R mutants undergo the same degree of proteasomal degradation as wild-type CFTR and are still trafficked. Prevention of proteolysis at the lysosome accumulates more C band K1041R and K1080R. However, the lysosomal inhibitors restored reduced expression of K1080R C band and improved the K1218R mutant, indicating that the K1218R mutant can avoid lysosomal degradation. Therefore, these data suggest that K1041R and K1080R mutants are degraded at the proteasome and lysosome, respectively. The K1218 residue may be important for cell surface localization.

K710R and K716R CFTRs reduce K63-linked polyubiquitination.

The specific type(s) of ubiquitin linkages to CFTR residues is unknown. Ubiquitin ligases have been implicated in CFTR degradation by the ERAD system (9, 19–21). J. M. Younger et al. detected ubiquitination in NBD1 R fragments by in vitro coincubation with chaperones and ubiquitination enzymes (22). However, the ubiquitination of specific lysine residues in the R domain is uncertain. Using an antibody specific for K63-linked polyubiquitin, we show that immunopurified CFTR has visibly less K63-linked polyubiquitination in the K710R and K716R mutants than the wild type (Fig. 5). In addition to K710 and K716, the K68 and K1218 residues are required for K63 polyubiquitination. For K1041R the reduced Ub-K63 immunoblot signal likely derives from the reduced amount of immunoprecipitated product. Therefore, K1041 might not be a site for K63 poly-Ub. Immunoblot detection for all ubiquitin showed no differences in total ubiquitination between mutants and wild-type CFTR (Fig. 5), suggesting that other linkages and sites are unaffected by the single-site mutations.

FIG 5.

K63-linked ubiquitin is detected on wt and mutant CFTR. Control vector, wt CFTR, K14R, K68R, K710R, K716R, K710/716R, K1041R, K1080R, and K1218R were expressed in IB3-1 cells for 48 h. CFTR was immunoprecipitated (IP) with M3A7 as described in Materials and Methods and separated by SDS-PAGE. On the immunoblot on the left, the top half demonstrates the ubiquitin conjugated by K63. The bottom half was blotted with antibody to CFTR (H182). K63-linked ubiquitin on CFTR is reduced in the K68R, K710R, K716R, and K1218R mutants. K-63-linked ubiquitin appears to be increased in the K1080R mutant. In the immunoblots shown on the right, CFTR was immunoprecipitated with antibody 169, separated by SDS-PAGE, and blotted for ubiquitin (top half) or CFTR (with M3A7; bottom half). CFTR remains polyubiquitinated in all forms. WB, Western blotting.

To compare the protein stability of wild-type and mutant CFTR, the biosynthesis of new proteins was inhibited by cycloheximide (CHX) treatment and chased for sequential time points up to 24 h, and CFTR was measured (Fig. 6A). A CHX chase assay shows no further synthesis of immature A/B bands, and the level of C band expression decreases over time. The K710R, K716R, and K710/716R mutants showed faster degradation of the C band of CFTR than the wild type and K1218R. Therefore, ubiquitination in the K710 and K716 residues again appears critical for the stability of matured CFTR, and ubiquitination might be the key for this matured CFTR stabilization. The initial expression level of K1218R is higher than that of wt CFTR, and the degradation rate of K1218R is similar to that of the wt CFTR.

FIG 6.

K-to-R mutation alters the stability of CFTR protein, and wt CFTR, K14R, K1080R, and K1218R, but not K1041R, are detectable at the cell surface. (A) Relative half-lives of each form of CFTR were determined using a cycloheximide (CHX) chase assay. Cells transfected for 24 h were then exposed to 100 μg/ml of CHX for the indicated times before preparation of whole-cell lysates and immunoblotting for CFTR with antibody M3A7. Representative blots are on the left, and the densitometry (n = 4 to 5) results are shown on the right. Wild-type and K1218R CFTR maintain the highest stability with little turnover in 24 h. (B) Confocal analysis for cell surface localization of CFTR. IB3-1 cells were transfected with the indicated constructs for 48 h and then fixed without permeabilization. Cells were stained with the anti-CFTR antibody that is specific for the first extracellular loop (CF3). The red signal indicates CFTR, and the blue (4′,6′-diamidino-2-phenylindole) marks nuclei. Scale bar, 20 μm. As expected, CFTR is virtually undetectable in cells expressing control vector or K1041R. CFTR is easily detected from the surface of cells expressing wt CFTR, K14R, K1080R, and K1218R. A lighter signal is visible from cells expressing K68R and the R domain mutants. (C) Densitometric analysis of CFTR (red) signals. Mean fluorescence intensity was measured in arbitrary units. Differences compared to wt CFTR were considered significant (*) at a P value of <0.05 by an unpaired Student t test (n = 4). As in panel A, the densitometric semiquantitative analysis is consistent with the visible signals. (D) Biochemical detection of cell surface CFTR. Cells treated as described for panel A underwent cell surface protein biotinylation, precipitation with streptavidin beads, and SDS-PAGE. CFTR was detected from the isolated cell surface fraction by immunoblotting with M3A7. Band C was most reduced in K1041R and highest in the wt, K1080R, and K1218R, consistent with data in panels A and B. (E) Densitometric analysis of immunoblots for cell surface CFTR located on the plasma membrane (PM). Data are means ± standard errors of the means (n = 3). (F) To assess the effects of proteasomal or lysosomal inhibition on cell surface CFTR, the cells were treated with MG132 or E64d/pepstatin A to inhibit the proteasome or lysosome prior to surface biotinylation and analysis. Representative immunoblots and the corresponding densitometry are shown. Data are means ± standard errors of the means (n = 3). Cell surface band C, prominent for wt CFTR and K1218R located on the PM, was unaffected by MG132 and increased by lysosomal inhibition. Interestingly, higher-molecular-weight forms of K1041R were pulled down from the biotinylated surfaces of cells treated with proteasomal or lysosomal inhibition. A similar pattern was seen with lysosomal inhibition of cells expressing K1218R, suggesting that there is a ubiquitination step between the cell surface and the lysosome occurring through an alternate site.

K14R, K1080R, and K1218R but not K1041R reach the cell surface.

Immunodetection of band C CFTR in whole-cell protein lysates is of insufficient resolution to confirm cell surface localization and instead represents the sum of terminally glycosylated CFTR beyond the ER. To determine whether mutant forms of CFTR are localized at the cell surface or remain in subcellular compartments, we undertook a confocal analysis of nonpermeabilized, fixed cells transfected with the wild-type or mutant forms of CFTR. Immunofluorescence analysis was performed after staining with the CF3 antibody (specific for the first extracellular loop of CFTR). Cells were prepared without permeabilization in order to detect cell surface-localized CFTR exclusively. Figure 6B shows representative digital micrographs of each condition, with CFTR is indicated by a red signal. There were decreased levels of cell surface CFTR detection in the K68R, K710R, K716R, K710/K716R, and K1041R mutant forms of CFTR, but the wild-type, K14R, K1080R, and K1218R mutants accumulated similar levels of visible cell surface CFTR. Analysis of the Z stack of images for the wild type, K68R, K710R, K1080R, and K1218R confirmed that the CFTR immunostaining, shown in red, resided at the apical surface in these nonpermeabilized cells (data not shown). Figure 6C contains the semiquantitative assessment of CFTR signal strength, and the data correspond well with Fig. 6B.

To confirm the qualitative immunofluorescence data, we also used biochemical methods. The cell surface membrane fraction was labeled by biotinylation of transfected IB3-1 cells, and the proteins were collected by streptavidin resin precipitation. The cell surface membrane fraction was then separated by SDS-PAGE. Again, there was a reduction in immunodetectable CFTR (K14R, K68R, K710R, K716R, K710/716R, and K1080R forms) in cell surface membranes compared to wild-type levels. As expected, increased amounts of K1218R were detected from the cell surface fraction (Fig. 6D and E). Next, transfected cells were treated with proteasomal or lysosomal inhibitors prior to labeling and harvesting of cell surface proteins. Interestingly, smaller amounts of the rescued C band in mutants after proteasomal inhibition (Fig. 4) were detected in the cell surface fraction (Fig. 6F). The amount of rescued C band in mutants from lysosomal inhibitor-treated whole-cell lysate (Fig. 4) was increased in the biotinylated cell surface fraction (Fig. 6F). These data demonstrate that by preventing ubiquitination at K710R and K716R, the CFTR mutants are then degraded in the lysosome. Preventing ubiquitination at K14R, K1080R, and K1218R allows the mutant protein to gain access to the cell surface. On the other hand, prevention of ubiquitination at 1041 by making the K1041R construct promotes degradation primarily by the proteasome. Interference at a single ubiquitination site does not prevent modification at other sites.

If, as we have shown previously (23), valosin-containing protein (VCP) interacts with CFTR to extract it from the ER and if HDAC6 must interact with CFTR to promote trafficking beyond the ER, then we predict that K1041R would interact preferentially with VCP and that K1218R would interact preferentially with HDAC6. Furthermore, ZO-1 is often cited as a key interacting protein for plasma membrane components (24). In the experiment shown in Fig. 7A, we analyzed the immunoprecipitated CFTR complexes for the two interacting cargo proteins and ZO-1. VCP was pulled down from all mutants with CFTR antibody but not from the control vector-transfected cells. HDAC6 was abundant in all cases, and ZO-1 was conspicuously absent from the K1080R and K1218R mutants. This might suggest an alternative trafficking pathway, independent of the C-tail PDZ domain.

FIG 7.

Lysine modification affects CFTR interaction through its PDZ binding domain. (A) CFTR was immunoprecipitated from whole-cell lysates expressing each of the indicated CFTR mutants for 48 h with mouse anti-CFTR antibody (M3A7). The resultant proteins were electrophoresed and immunoblotted for VCP, HDAC6, ZO-1, and CFTR. Whereas HDAC6 was abundant in the CFTR complex of both the wt and mutants, VCP was at a lower level overall and not detected in the wt CFTR complex. We did not detect ZO-1 in the K1080R or K1218R complex. Since ZO-1 interacts with PDZ domains, the reduced interactions in K14R, K68R, K710R, K716R, K710/716R, and K1041R and the absence of interactions with K1080R and K1218R raise the issue of alternative trafficking routes that are not dependent on the C-tail PDZ domain. Representative immunoblots from triplicate experiments are shown. (B) Schematic presentation of trafficking and expected subcellular localization of CFTR. Poly- or monoubiquitination in K14, K68, K710, K716, 1041, K1080, and K1218 sites allows CFTR to pass through its regular quality control process, producing adequate band C. K710R, K716R, K710/K716R, and K1080R mutants undergo some proteasome- and some lysosome-dependent degradation, whereas K1041R is primarily degraded by the proteasome. K14R and K68R experience some proteasome degradation, but a sizeable fraction is localized near the plasma membrane. K1218R bypasses the proteasome fairly efficiently and further is slow to undergo lysosomal degradation.

In summary, replacement of lysine with arginine causes minimal structural perturbation while at the same time preventing the covalent attachment of a ubiquitin. N-tail K14 and K68 can be tagged with ubiquitin early during translation, feeding the nascent CFTR to the proteasome. Blocking this modification allows accumulation of band B and trafficking to the cell surface compartment. The R domain K710 and K716 sites appear to stabilize CFTR if allowed to accept ubiquitin chains, but without additional help, such as proteasomal and lysosomal inhibition, they do not accumulate enough C band to restore CFTR to wild-type levels. K1218R is the most interesting construct since this mutant CFTR does better than the wild type in remaining at the cell surface in large-mass quantities. Of course, we cannot rule out functional deficits in any of our K-to-R mutants without further study; nevertheless, they provide clues as to where to target future protein sparing small molecule screens.

DISCUSSION

Much is known about ER quality control of CFTR biogenesis. The heat shock 70-kDa protein family and related cochaperones regulate biogenesis and quality control of CFTR (8–10), and both wild-type and common mutated forms of CFTR are degraded by ubiquitin-mediated proteasomal degradation (8, 9). CFTR peripheral quality control occurs after the Golgi compartment, where CFTR localized to the cell surface undergoes endocytosis and recycling by a ubiquitin-dependent endosomal sorting system. The F508del CFTR that is rescued to the cell surface by lower temperature or small molecules is quickly retrieved from the cell surface and trafficked to the lysosome for degradation (11, 12). Prolongation of recycling and cell surface residence time is desirable to enable potentiation of chloride channel openings by molecules such as ivacaftor (25). Our study suggests that interference with ubiquitination at K14, K68, K1080, and K1218 or maintaining Ub at K710 and K716 is one potential therapeutic strategy to accumulate cell surface CFTR.

Previous studies revealed that CFTR interacts with microfilaments (filamins) through its N-tail region (S13 is the key residue) and that this interaction stabilizes CFTR at the cell surface (26). Other groups reported that D47/E51/E54/D58 residues in the N-tail region of CFTR are important for protein kinase A-dependent channel activity through the interaction with the R domain (27, 28). It has been shown that the R domain of CFTR regulates its trafficking (29). Mutations encompassing the K710 and K716 residues are found in the Toronto CFTR mutation database. Recent research revealed that the ICL4 in CFTR interfaces with the NBD1 domain and that the F508del CFTR mutation interrupts ICL4 and NBD1 domain interactions by altering NBD1 domain folding (30, 31). The NBD2 domain is important in ATP binding and hydrolysis for gating (32). The role of ICL4 and the NBD2 domain in the regulation of CFTR expression/maturation is unexplained. However, K1080 and K1218, if ubiquitinated, could disrupt the natural pathway of trafficking and/or function of CFTR.

It has been shown that CFTR interacts with cargo proteins such as VCP and HDAC6 to deliver CFTR to the proteasome or lysosome (23, 33), and it is also known that the C tail of CFTR has a PDZ binding domain (34). Previous research shows that ubiquitination is one of the key steps in determining the CFTR destination during trafficking (8, 35). Our eight mutants do not show any differences in the ability to complex with these cargo proteins (Fig. 7A). However, immunoprecipitates of wt and mutant CFTR showed different affinities to ZO-1. Interestingly, ZO-1 was highly abundant in complexes formed with wt CFTR; K14R, K68R, K710R, K716R, K710/716R, and K1041R mutants bound less ZO-1, whereas K1080R and K1218R mutant complexes had none. This result indicates that interrupting ubiquitination of CFTR in lysine residue 1218 weakens its interaction with ZO-1 complex and that this decreased affinity to PDZ binding proteins might decelerate the degradative pathway, whereas reduced interaction with ZO-1 in the K1080R mutant accelerated degradation. Overall, these opposing results indicate that CFTR interaction with ZO-1 is not the primary determinant of cell surface stability.

We searched the Toronto cystic fibrosis mutation database (http://www.genet.sickkids.on.ca/app) and the Hopkins Clinical and Functional Translation of CFTR (CFTR2) website (http://www.cftr2.org) to ask whether our mutations had been defined as disease causing. Table 3 lists the findings: K14X (stop), K68E, K68N, K710X (stop), K1080Q, K1080R, and K1080I mutations already exist in human genes. The stop codon mutants likely lead to unstable mRNA transcripts or short proteins. The missense mutants may resemble ours, such as the K1080R mutant. It is also interesting that the K1080R mutant is capable of C band expression. It remains a possibility that these mutations cause structural or protein folding changes. It is predicted, however, that lysine residues in CFTR undergo posttranslational modification and that this affects the interaction between CFTR and its binding proteins in subcellular compartments, resulting in disrupted expression, maturation, stabilization, and localization of CFTR during membrane trafficking. Figure 7B shows a hypothetical schematic presentation for the fate of ubiquitinated or nonubiquitinated CFTR during maturation and trafficking.

TABLE 3.

Toronto cystic fibrosis database results

Domaina	Mutation(s)
N tail	K14X (stop), K68E, K68N
R	K710X (stop)
ICL4	K1080Q, K1080R, K1080I

^aR, regulatory domain; ICL4, intracellular loop 4.

In summary, lysine residues in key domains of the CFTR protein accessible to the cytoplasm are important for the expression, maturation, polyubiquitination, and cell surface localization/trafficking of CFTR. We expect that our results will contribute to strategies implemented to increase the half-life of CFTR so that CFTR proteins remain longer on the cellular plasma membrane, resulting in overall enhanced activity of CFTR protein. Therefore, these research data may contribute to the design of therapeutic approaches for rescuing CFTR mutants and eventually treating CF patients.