Dysfunction of CFTR bearing the AF508 mutation

Cystic fibrosis transmembrane conductance regulator (CFTR) (Riordan et al., 1989) is a regulated Cl^- channel located in the apical membrane of several Cl^--secretory epithelia (for a review, see Welsh et al., 1992). Mutations in the gene encoding CFTR cause cystic fibrosis (CF) (Kerem et al., 1989; Tsui, 1992). These two observations begin to explain the best characterized physiological defect in CF, namely affected epithelia lack a cAMP-regulated apical membrane Cl^- conductance (Quinton, 1990). The loss of cAMP-regulated Cl^- permeability is manifest in a number of CF epithelia including the pancreas, the intestine, the sweat gland secretory coil, the sweat gland absorptive duct and the pulmonary airways. In each organ, abnormal transepithelial electrolyte transport is thought to contribute to the pathogenesis of the disease. However, it is lung disease that is the major cause of morbidity and mortality in CF (Boat et al., 1989). In the lung, defective electrolyte transport is thought to alter the quantity and composition of the respiratory tract fluid, thereby contributing to the impaired mucociliary clearance observed in patients with CF.

In order to understand the pathogenesis of CF it is important to understand how mutations cause a loss of CFTR Cl^-channel function. The most common CF-associated mutation is deletion of a phenylalanine at residue 508 (AF508) (Kerem et al., 1989; Tsui, 1992). This mutation accounts for approximately 70% of CF chromosomes. Here I discuss some of the mechanisms that cause dysfunction of CFTR bearing the ΔF508 mutation.

Amino acid sequence analysis and comparison of the sequence with that of other proteins suggested that CFTR consists of five domains (Riordan et al., 1989) (Fig. 1). Beginning at the amino terminus there is a putative membrane-spanning domain (MSD1), composed of six possible membrane-spanning α-helices. MSD1 is followed by a nucleotide-binding domain (NBD1) in which there is sequence similarity with nucleotide-binding sequences from a number of other proteins. Then comes the R (regulatory) domain, which contains multiple phosphorylation sites for cAMP-dependent protein kinase (PKA) and protein kinase C. After the R domain, the protein re-enters the membrane with a second membrane-spanning domain (MSD2) followed by a second nucleotide-binding domain (NBD2).

Sequence similarity between the NBDs and the topology of CFTR (with the exception of the R domain) suggest that CFTR belongs to a family of proteins called the traffic ATPases (Ames et al., 1990), or ATP-Binding Cassette (ABC) transporters (Hyde et al., 1990). Most members of this family are ATP-dependent transporters; they include periplasmic permeases in prokaryotes and P-glycoprotein responsible for multiple drug resistance (MDR) in eukaryotes. However, recent studies indicate that CFTR is a regulated Cl^- channel. Studies of CFTR containing site-directed mutations have begun to provide some information about the function of the individual domains (Welsh et al., 1992). The MSDs are thought to contribute to the formation of the channel pore. Mutation of specific basic residues to acidic residues within the first MSD alters the anion selectivity of the channel. The R domain regulates channel activity. Phosphorylation of the R domain at several different sites opens the channel. Regulation by phosphorylation with cAMP-dependent protein kinase (PKA) is complex: phosphorylation of multiple different PKA-consensus sequences regulates the channel. Moreover, deletion of part of the R domain produces a channel that is open even without phosphorylation. The NBDs also control channel activity through an interaction with cytosolic nucleotides. ATP interacts with both NBDs and hydrolysis is probably required for channel opening. It is also possible that CFTR may have other activities in addition to being a regulated channel.

The NBDs and the R domain distinguish CFTR from the structure of known voltage- and ligand-gated ion channels, indicating that CFTR may represent the first identified member of a new family of ion channels. Recent work showing that expression of MDR is associated with volume-regulated Cl^- channels (Mostov et al., 1992; Gill et al., 1992) suggests that MDR may also be an ion channel and that other channels with this motif may be discovered.

The topology of the model shown in Fig. 1 is not known with certainty, but the location, with reference to the membrane, of several sites is known (Denning et al., 1992b). The amino terminus is likely to be intracellular because of the lack of a signal sequence; the first predicted extracellular loop is likely to be so because it is recognized by antibodies directed to that epitope in nonpermeabilized cells; the NBD1 and R domains are intracellular, as assessed by the regulation of these two domains by ATP and phosphorylation, respectively, and because an antibody directed against the R domain only stains permeabilized cells; the third extracellular loop is extracellular because it contains sites that are glycosylated; NBD2 is intracellular, as assessed by the regulation of this domain by ATP; and finally, the C terminus is likely to be intracellular because it is recognized by antibodies directed against this epitope only after cells have been permeabilized.

For CFTR CD channels to govern transepithelial Cl^- secretion, they must be located in the apical membrane. A number of antibodies have been used to immunolocalize CFTR. CFTR has been found in the apical region of several epithelia including small pancreatic ducts, intestinal epithelia and Cl^--secreting epithelial cell lines (Marino et al., 1991; Crawford et al., 1991; Denning et al., 1992b,c; Kartner et al., 1992; Engelhardt and Wilson, 1992).

To directly determine whether CFTR is located in the apical membrane, we turned to intestinal epithelial cells that express high levels of endogenous CFTR and have cAMP-activated apical membrane Cl^- permeability (Denning et al., 1992b). We used T84 cells, CaCo2 cells, and HT29 clone 19A cells. In order to identify CFTR that is localized specifically in the apical membrane, the cells were grown on permeable filter supports so that they polarized, segregating apical from basolateral membrane proteins, and developed a transepithelial resistance and the ability to secrete Cl^-from the basal to the apical surface. We used monoclonal antibodies directed against different regions of the protein: the R domain (Ml3-1), the COOH terminus (Ml-4) and a predicted extracellular domain (M6-4). All three antibodies immunoprecipitated and immunostained recombinant CFTR expressed in heterologous cells.

We immunostained the epithelial cells and examined the cellular distribution of CFTR using confocal laser-scanning microscopy. We found that the pattern of staining for CFTR resembled the staining pattern observed with several apical membrane markers. However, it differed from the staining pattern for basolateral membrane proteins. The majority of CFTR staining was observed at the apical pole. In thin sections of cell monolayers, we also observed staining specifically at the apical membrane.

However, such immunocytochemical studies cannot distinguish between CFTR that is located in the apical membrane and CFTR that is located in a vesicular pole just beneath the apical membrane. This distinction is important for understanding the function and regulation of CFTR. Evidence that CFTR is located in the apical membrane came from studies in nonpermeabilized cells. We found that antibody M6-4, directed against an extracellular epitope, stained nonpermeabilized epithelia. In contrast, antibodies directed against intracellular epitopes (Ml3-1 and Ml-4) only stained permeabilized monolayers.

The conclusion that CFTR is a regulated Cl^- channel and the observation that it is located in the apical membrane, places it in a location where its activation by PKA-depen-dent phosphorylation would directly mediate Cl^- exit from the cell during Cl^- secretion. It is also, however, possible that CFTR is located beneath the apical membrane and functions on intracellular membranes (Barasch et al., 1991).

Studies of the biosynthesis of CFTR in transfected cells identified a defect associated with the AF508 mutation (Cheng et al., 1990; Gregory et al., 1991). The progress of wild-type CFTR through the biosynthetic pathway can be followed by its state of glycosylation (Fig. 2). The nascent CFTR protein migrates at approximately 150 kDa on a polyacrylamide gel (this form is referred to as band A) (Gregory et al., 1990; De Jonge et al., 1989; Gregory et al., 1991). A more slowly migrating form is called band B. This form represents core glycosylated protein that is endoglycosidase H-sensitive, suggesting partial glycosylation in the endoplasmic reticulum. The mature form of the protein migrates as band C, a broad diffuse band of approximately 170 kDa. The band C form is endoglycosidase H-insensitive, but can be shifted to band A with N-Glycanase treatment.

In contrast to results with wild-type CFTR, when CFTRAF508 was expressed in heterologous cells, it migrated only as bands A and B. Based on these results, Cheng and his coworkers (Cheng et al., 1990; Gregory et al., 1991) proposed that the AF508 mutant version of CFTR is misfolded. The abnormal glycosylation pattern of CFTRAF508 suggested that the mutant protein did not reach the Golgi complex and was not delivered to the plasma membrane. Instead, it was recognized as abnormal and targeted for retention and degradation in the endoplasmic reticulum. Incomplete glycosylation does not in itself cause CD impermeability, because protein that lacked glycosylation sites (constructed by site-directed mutagenesis) was present in the plasma membrane and had normal Cl^- channel activity.

A direct test of the hypothesis that CFTRAF5O8 is not at the apical membrane of CF epithelia requires several things. First, it requires the use of nonrecombinant normal and CF epithelia. Because the studies of protein glycosylation were done in recombinant cells, the results could have been an artifact of overexpressing CFTR, or of expressing it in nonpolarized cells. Second, it requires a method of assessing whether CFTR is actually in the apical membrane, because that is where the CF defect in Cl^- permeability resides (Quinton, 1990; Boat et al., 1989). Third, it requires a detection method, which is very sensitive and which clearly distinguishes signal from background. This is essential because CFTR is often present at low levels in nonrecombinant cells (Riordan et al., 1989; Trapnell et al., 1991). To assess the location of CFTR, we used primary cultures of CF airway epithelial cells because they are the main site of disease in patients with CF. We cultured the cells on permeable filter supports so that they developed a transepithelial resistance and polarized with a distinct apical membrane that expresses the CF Cl^- transport defect.

To localize CFTR that is in the apical membrane and provide a quantifiable method for assessing antibody binding, we developed a new technique (Denning et al., 1992c). We incubated nonpermeabilized airway epithelia with the antibody directed against the extracellular epitope (M6-4). We then incubated with a biotinylated secondary antibody followed by streptavidin. Finally, we incubated the cells with a suspension of biotinylated fluorescent beads. The beads were easy to count because of their large size (approximately 1 pm) and high fluorescence intensity. In principle, this method is similar to immunogold electron microscopy. As controls we used antibodies against intracellular epitopes (Ml3-1 or Ml-4), nonspecific mouse IgG, or no primary antibody.

To verify the technique we tested it with T84 cells, which express CFTR in the apical membrane, and with NIH 3T3 fibroblasts expressing recombinant CFTR. The data showed that nonspecific binding was identical for all of the antibodies except antibody M6-4, directed against an extracellular epitope. That antibody showed at an increased number of beads bound per field in cells expressing wild-type CFTR, but not in cells expressing CFTRΔF5O8.

We used this technique to examine primary cultures of CF airway epithelial cells. Fig. 3A shows binding of anti-body/bead complexes to the apical membrane of nonpermeabilized airway epithelia. In nonCF epithelia, the number of beads per field with antibody M6-4 was always greater than with control antibodies, including antibody Ml-4, M13-1 and nonimmune mouse IgG. In contrast, in CF epithelia there was no difference in the number of beads per field when epithelia were exposed to antibody M6-4 or the various controls. This result suggests that CFTR is in the apical membrane of normal airway epithelia, but is missing from or present at a greatly reduced amount in the apical membrane of CF epithelia. The CF epithelia studied in Fig. 2 were derived from patients bearing the ΔF508 mutation, other mutations that are also misprocessed, nonsense mutations that would be expected to fail to produce a complete protein, or unidentified mutations.

The difference between normal and CF epithelia is illustrated more clearly in Fig. 3B, which shows the data from Fig. 3A normalized to the average number of beads/field observed with antibody Ml-4. In normal epithelia, binding with antibody M6-4 averaged 4.65 ± 1.00 times the binding with antibody Ml-4. In contrast in CF epithelia, binding with antibody M6-4 was 1.02 ± 0.03 times the binding with antibody Ml-4.

We also studied permeabilized airway epithelial cells, immunostained them with antibodies against intracellular epitopes, and then used confocal laser scanning microscopy to localize CFTR. In normal airway epithelia, we found the most intense staining in the area of the apical membrane. In contrast, in CF epithelia containing the AF508 mutation, the brightest staining was encountered beneath the apical membrane in a pattern spread throughout the cytoplasm. Because of the low level of immunostaining, we have not been able to localize mutant CFTR to a specific intracellular organelle: however, in some CF epithelia the pattern appeared to be primarily perinuclear.

Our finding that CFTR containing the AF508 mutant is not present in the plasma membrane is consistent with earlier functional studies, which have failed to detect cAMP-stimulated Cl^- currents in the apical membrane of CF epithelial cells (Quinton, 1990), as well as with previous studies that have failed to detect cAMP-stimulated CD channels in cells expressing CFTRAF508 (Rich et al., 1990). Cl^- channel activity was detected, however, when CFTRAF508 was expressed in Xenopus oocytes (Drumm et al., 1991), Vero cells (Dalemans et al., 1991) and Sf9 insect cells (Bear et al., 1992). Because oocytes and Sf9 cells are typically maintained at lower temperatures than mammalian cells, and because processing of nascent proteins can be sensitive to temperature, we tested the effect of temperature on the processing of CFTRAF508.

As a marker of processing, we measured the amount of CFTR that is present in the mature, fully glycosylated band C form (Denning et al., 1992a). We found that when temperature was reduced from 37°C to lower temperatures, processing of wild-type CFTR was not appreciably affected. However, in the AF508 mutant the reduction in temperature produced an increase in the amount of mature CFTR (Fig. 4). This observation is consistent with the finding in other systems that the conformation of nascent proteins and their subsequent processing and transport can be temperature-sensitive.

We found that the amount of mature, fully glycosylated band C CFTR increased in amount as the duration of incubation temperature increased and that the effect was reversible. When cells were incubated at a low temperature to allow CFTR to mature and then were switched back to the nonpermissive 37°C, the amount of CFTR progressively decreased with a half-time of approximately seven hours. Moreover, in pulse-chase experiments we found that if the protein was synthesized at 37°C and then the temperature was reduced to 26°C, mutant CFTR that had been synthesized at 37°C was chased into the mature form. This result shows that correct processing of the mutant is not the result of an effect of temperature on protein synthesis.

These results suggested that at a reduced temperature, CFTRAF5O8 was delivered to the plasma membrane, where it could mediate Cl^- transport. To test that hypothesis, we used the whole-cell patch-clamp to measure cAMP-activated Cl^- currents in the plasma membrane of cells expressing CFTRAF508. We found that after incubation at 30°C for two days, cAMP agonists stimulated large CD-selective currents. All of the properties of these currents were the same as those of wild-type, with the exception that when they were studied at the single channel level they had a reduced probability of being in the open state (P_o). In excised, cell-free patches addition of PKA and 1 mM ATP produced a P_o of 0.34 for wild-type CFTR and 0.13 for CFTRAF5O8. The reduced P_o of CFTRAF508 is similar to a value previously reported from CFTR expressed in Vero cells (Dalemans et al., 1991).

These data suggest that the defective processing of CFTR can be corrected when temperature is reduced. They also indicate that when the mutant CFTRAF508 is correctly localized, it retains at least partial function. Thus, the possibility is raised that a therapeutic maneuver designed to deliver mutant protein to the plasma membrane could be of potential benefit in CF.