Cystic Fibrosis Transmembrane Conductance Regulator

The cystic fibrosis transmembrane conductance regulator (CFTR) forms a Cl channel that is an essential component of epithelial Cl transport systems in many organs, including the intestines, pancreas, lungs, sweat glands, and kidneys. In the Cl secretory intestinal epithelium, Cl enters the cells through a Na-K-2Cl cotransporter in the basolateral membrane and exits through CFTR in the apical membrane; water follows osmotically (1). Absorptive epithelia use similar transporters and channels, but their polarized distribution between the apical and basolateral membranes is usually reversed. A major determinant of the transepithelial Cl transport rate is the level of activation of CFTR (2, 3), which depends on the extent to which it is phosphorylated. This is determined by the relative activities of kinases and phosphatases, the activities of which are often hormonally regulated (1). Defects in the gene encoding CFTR that reduce either its Cl transport capacity or its level of cell surface expression cause cystic fibrosis (CF) (4–6) as well as a form of male sterility due to congenital bilateral absence of the vas deferens (7). CF is the most common lethal genetic disease in Caucasians, with about 30,000 CF patients in the United States. In contrast, in intestinal epithelial cells overstimulation of CFTR because of the activation of protein kinases by bacterial enterotoxins causes secretory diarrhea (1, 8). Secretory diarrhea is the second largest cause of infant mortality in the developing world, causing 3 million deaths per year of children under the age of 5. Thus, although CFTR was named because of its association with CF, as a cause of disease, its relationship to secretory diarrhea is a more widespread public health problem. The cloning of CFTR in 1989 (9) has facilitated studies of its structure, function, regulation, biogenesis, and degradation, which will be reviewed in this article. Issues reviewed elsewhere and not discussed here include the mechanisms by which mutations in CFTR cause CF (5, 6) and the possible role of CFTR in regulating the pH within intracellular organelles (10).

The cystic fibrosis transmembrane conductance regulator (CFTR) 1 forms a Cl Ϫ channel that is an essential component of epithelial Cl Ϫ transport systems in many organs, including the intestines, pancreas, lungs, sweat glands, and kidneys. In the Cl Ϫ secretory intestinal epithelium, Cl Ϫ enters the cells through a Na ϩ -K ϩ -2Cl Ϫ cotransporter in the basolateral membrane and exits through CFTR in the apical membrane; water follows osmotically (1). Absorptive epithelia use similar transporters and channels, but their polarized distribution between the apical and basolateral membranes is usually reversed. A major determinant of the transepithelial Cl Ϫ transport rate is the level of activation of CFTR (2,3), which depends on the extent to which it is phosphorylated. This is determined by the relative activities of kinases and phosphatases, the activities of which are often hormonally regulated (1).
Defects in the gene encoding CFTR that reduce either its Cl Ϫ transport capacity or its level of cell surface expression cause cystic fibrosis (CF) (4 -6) as well as a form of male sterility due to congenital bilateral absence of the vas deferens (7). CF is the most common lethal genetic disease in Caucasians, with about 30,000 CF patients in the United States. In contrast, in intestinal epithelial cells overstimulation of CFTR because of the activation of protein kinases by bacterial enterotoxins causes secretory diarrhea (1,8). Secretory diarrhea is the second largest cause of infant mortality in the developing world, causing 3 million deaths per year of children under the age of 5. Thus, although CFTR was named because of its association with CF, as a cause of disease, its relationship to secretory diarrhea is a more widespread public health problem.
The cloning of CFTR in 1989 (9) has facilitated studies of its structure, function, regulation, biogenesis, and degradation, which will be reviewed in this article. Issues reviewed elsewhere and not discussed here include the mechanisms by which mutations in CFTR cause CF (5, 6) and the possible role of CFTR in regulating the pH within intracellular organelles (10).

Structural Basis of Channel Function
CFTR is a member of the ATP-binding cassette (ABC) membrane transporter gene superfamily that includes both eukaryotic and bacterial proteins, such as the multiple drug resistance protein (MDR), the sulfonylurea receptor, the transporter for antigen presentation, and the bacterial periplasmic permeases (9,11). Being an ion channel makes CFTR unique in this gene superfamily in which most other members are ATP-driven membrane transporters.
CFTR contains 1480 amino acids and consists of two homologous halves (Fig. 1). Each half contains six membrane-spanning segments and a nucleotide-binding domain (NBD). The two halves of CFTR are linked by a cytoplasmic regulatory domain (R-domain) that contains a number of consensus phosphorylation sites (9). In the proposed transmembrane topology ( Fig. 1), which is supported by experimental evidence (9,12,13), 77% of the protein is in the cytoplasm, 19% in membranespanning segments, and 4% in extracellular loops, which (except for the M1-M2 and M7-M8 loops) are very short. The M7-M8 loop contains two N-linked glycosylation sites that are used in vivo (12,13). Covalent linkage of the two halves is not required for assembly and function. When expressed as separate proteins in the same cells the halves assembled into functional channels and could be co-immunoprecipitated. This assembly required interactions within the membrane-spanning domains (13,14).
CFTR was inferred to be a monomer because following solubilization, with either ionic or nonionic detergents, biochemically distinct forms of CFTR, expressed in the same cells, could not be co-immunoprecipitated (15). In contrast, two studies concluded that CFTR is a homodimer in situ based on the size of CFTR particles in freeze-fracture electron micrographs (16) and on the functional effects of coexpressing mutants with distinct functional properties (17). In the homologous MDR transporter, however, dimers were not seen in cryoelectron microscopic images (18).
Channel-lining Residues-The channel lining is formed, at least in part, by residues from among the 12 membrane-spanning segments. The major determinants of the channel's functional properties are likely to be among the channel-lining residues. Water-accessible residues inferred to line the channel were identified in the M1, M3, and M6 segments using the substituted cysteine accessibility method (19 -21). These segments' secondary structure was inferred to be largely ␣-helical based on the patterns formed by the channel-lining residues (19 -21). Further support for an ␣-helical secondary structure of the membrane-spanning segments was provided by studies of synthetic peptides, the sequences of which corresponded to those of the M1-M6 segments. These peptides were largely ␣-helical in liposomes and in detergent micelles (22,23).
The minimum channel diameter was inferred to be ϳ5.3 Å based on the size of the largest permeant anion (24). Further studies using patches with larger numbers of channels showed that anions as large as lactobionate (10 -13 Å in diameter) were slightly permeable. Thus, at least transiently, the diameter of the channel must be 10 -13 Å (25). In the presence of cytoplasmic ATP the large anions were only permeable from the cytoplasmic side (25). The molecular basis for asymmetric perme-ation by large anions is unknown. There is no asymmetry in the conduction of small anions; the channel has a linear currentvoltage relationship in Cl Ϫ solutions (13). Although some studies suggested that ATP was permeable through CFTR (reviewed in Ref. 26), it has been convincingly demonstrated that there is no measurable electrogenic ATP flux through CFTR (27)(28)(29).
Mechanism of Charge Selectivity-The ability to discriminate between Cl Ϫ and cations is essential for the role of CFTR in epithelial Cl Ϫ transport. The extent of anion selectivity, as measured by the Cl Ϫ to Na ϩ permeability ratio (30), ranges between 10 and 300 (31,32). We showed that the charge selectivity filter that discriminates between anions and cations is located at the cytoplasmic end of the channel and that both anions and cations can enter the extracellular end of the channel (33). Arg-352, at the cytoplasmic end of M6, is a major determinant of charge selectivity. Removal of the positive charge at this position reduced the Cl Ϫ to Na ϩ permeability ratio approximately 10-fold (34). We hypothesized that the positive charge of Arg-352 creates an electrostatic barrier to cation permeation. Near Arg-352 the channel must be narrow enough so that the electrostatic barrier extends across the channel lumen (34).
Halide Selectivity and Multiple Ion Occupancy-The channel displays modest selectivity between halides. The halide permeability sequence is indicative of weak interactions between permeating anions and channel binding sites (30,32). Multiple anions can simultaneously occupy the channel, thereby resulting in anomalous mole fraction effects in the single channel conductance in mixtures of Cl Ϫ and SCN Ϫ . These effects were eliminated in the M6 mutant R347D, and Arg-347, therefore, was hypothesized to be at or near an anion binding site (35). Arg-347, however, has been shown to form a salt bridge with Asp-924 (36). Therefore, it is possible that rather than Arg-347 being a binding site itself, structural perturbation caused by the loss of the salt bridge disrupted an anion binding site elsewhere (30).
Channel Inhibitors-CFTR has few inhibitors and none are of high affinity or specificity (13). The affinity of the channel blocker diphenylamine 2-carboxylate was reduced by mutating two water-accessible residues in M6. Mutations of the aligned M12 residues increased diphenylamine 2-carboxylate affinity suggesting that the channel is lined by residues from both halves of CFTR (37).
In contrast to their actions on other Cl Ϫ channels and transporters, disulfonic stilbenes such as DIDS inhibit only when applied to the cytoplasmic side (38). Cytoplasmic application of other large anions also caused flickery block of CFTR. This is further evidence that the channel is asymmetric and contains an anion binding site in the cytoplasmic vestibule (39). The residues forming the cytoplasmic vestibule remain to be identified, but the cytoplasmic loops probably do not contribute to the vestibule because none of the CF-related loop mutations affected ion conduction (40).
The picture that is emerging is of a channel with a large extracellular vestibule that extends into the plain of the membrane and is accessible to anions and cations from the extracellular side (Fig. 2). The channel narrows toward the cytoplasmic end where the charge selectivity filter is located. Finally, there is a short cytoplasmic vestibule that contains an anion binding site. The channel lining is formed, in part, by residues from the M1, M3, M6, and M12 segments. Little is known about whether the other eight membrane-spanning segments contribute to the channel lining. The location of the gate that blocks ion conduction through the channel in the closed state is unknown. The gate, however, is controlled by conformational changes in the cytoplasmic domains.

Regulation of CFTR Channel Gating
Two separate processes control the gating of CFTR: 1) phosphorylation and 2) binding and hydrolysis of ATP. Phosphorylation is necessary for activation, but it is not sufficient. After phosphorylation, gating between the closed and open states is controlled by ATP hydrolysis (12,13,41).
Phosphorylation Is a Prerequisite for Channel Activation-The R-domain contains multiple consensus phosphorylation sites for cAMP-dependent protein kinase (PKA), protein kinase C (PKC), and Type II cGMP-dependent protein kinase (9,12,13,41,42). The more extensive the phosphorylation, the greater the channel open probability (41,43). There appears to be extensive redundancy in the regulatory PKA phosphorylation sites. Five sites are phosphorylated in vivo, but removal of these sites or all 10 dibasic PKA sites reduced but did not eliminate PKA activation (44). CFTR is still phosphorylated after removal of the 10 dibasic sites indicating that phospho- rylation at non-dibasic PKA sites is sufficient for channel activation; one such site has been identified (12,13,41). PKC phosphorylation may be a prerequisite for activation by PKA (45). Finally, although genistein, a tyrosine kinase inhibitor, activates CFTR this occurs through a direct interaction with CFTR and does not involve a tyrosine kinase (reviewed in Ref. 41).
After phosphorylation, CFTR is deactivated by protein phosphatases (41). Dephosphorylation is largely mediated by protein phosphatases PP2C and PP2A (46,47). PP2C is closely linked to CFTR in the membrane as it could be co-immunoprecipitated with CFTR as well as cross-linked to CFTR by bifunctional cross-linkers, whereas PP1, PP2A, and PP2B could not (48).
Channel Gating Requires ATP Hydrolysis-Following phosphorylation, opening and closing of the channel is controlled by ATP binding and hydrolysis (12,13,41). Non-hydrolyzable ATP analogs do not support channel opening (49). Purified, phosphorylated, reconstituted CFTR binds ATP with a K m of 0.3 mM and hydrolyzes ATP at a rate of about 1 molecule/s (50), similar to the rate of channel gating. ATP hydrolysis occurs in the NBDs, which contain highly conserved Walker A and B motifs found in other ATPases (9,13,41,49).
A construct lacking the R-domain, CFTR-⌬R, did not require phosphorylation to open and close in the presence of ATP (13). Adding soluble, phosphorylated R-domain peptide to CFTR-⌬R enhanced its ATP sensitivity resulting in an increased channel open probability (51,52). Thus, the R-domain interacts with the NBDs and regulates their ATP affinity.
Structure of a Bacterial NBD-In the homologous bacterial periplasmic permeases the NBDs and the transmembrane domains are usually separate gene products that assemble during biosynthesis (11,53). The histidine periplasmic permease contains two copies of its NBD protein, HisP; each copy has a single nucleotide binding site. The x-ray crystal structure of HisP was determined (54). The sequence conservation between the bacterial and eukaryotic NBDs implies that their structures are similar and that aligned residues perform similar functions in HisP and CFTR.
HisP has a novel structure (54). It contains only small regions of structural similarity with RecA and the ␣ and ␤ F 1 -ATPase subunits. HisP is shaped like the letter L (Fig. 2). One arm (Arm I) of the L contains the ATP binding site, including the Walker A and B motifs. Most interactions between the Walker A residues and ATP are via hydrogen bonds between the backbone amides and the ATP phosphates. An aspartate at the C terminus of the Walker B motif, aligned with CFTR residues Asp-572 and Asp-1370, in NBD-1 and -2, respectively, was inferred to bind the Mg 2ϩ ion required for hydrolysis. CFTR residues Gln-493 and Ser-573 in NBD-1 and Gln-1291 and Glu-1371 in NBD-2 align with the catalytically important residues in HisP that hydrogen bond to a water molecule that in turn hydrogen bonds to the ␥-phosphate. This water molecule is thought to attack ATP because its hydrogen bond lies parallel to the P␥-O␤ bond that is broken during ATP hydrolysis (54). The adenine base forms a hydrophobic interaction with a single aromatic residue, perhaps explaining the lack of specificity among nucleotide triphosphates that can gate CFTR (49).
The other arm in the HisP structure, Arm II, contains a LSGGQ motif that forms the foundation around which Arm II is built. Arm II was inferred to interact with the membranespanning domains (54). The residue aligned with Phe-508 in CFTR, the site of the most common CF-associated mutation, lies on an exposed surface of an ␣-helix in Arm II.
Channel Gating and the Sites of ATP Hydrolysis-The cur-rent hypothesis that ATP hydrolysis at NBD-1 opens the channel and ATP hydrolysis at NBD-2 closes the channel to terminate a burst is consistent with the functional effects of mutations in the NBDs (13,41). Mutations in the NBD-2 Walker motifs that decrease its ATP hydrolysis rate slow channel closing (55)(56)(57)(58). Similar mutations in NBD-1 reduce channel opening. The NBD-1 mutations also altered channel closing rates, suggesting there are functional interactions between the NBDs (55)(56)(57). In both the crystal structure and in solution HisP forms a dimer, the interface of which is located along the edge of Arm I (54). Thus, direct contacts between the NBDs may be the structural basis for the functional interaction. Mixtures of ATP and non-hydrolyzable analogs lock channels in the open state. Presumably ATP hydrolysis at NBD-1 opens the channels and binding of a non-hydrolyzable analog at NBD-2 prevents closing (13,41). The gating behavior of CFTR has been described by several kinetic models. Cyclic models with essentially irreversible steps corresponding to the ATP hydrolysis steps provide the best descriptions of channel gating (reviewed in Refs. 13, 41, and 59).

Interactions with Other Proteins
In addition to its function as a Cl Ϫ channel, CFTR acts as a regulator of other channels and transporters (26). CFTR mediates cAMP regulation of amiloride-sensitive Na ϩ channels (60), outward rectifying Cl Ϫ channels (61,62), Cl Ϫ /HCO 3 Ϫ exchanger (63), and the ROMK K ϩ channel (26). Whether these regulatory functions result from direct interactions between CFTR and the channels and transporters or through indirect interactions via other proteins remains to be determined.
Direct interactions with CFTR have been shown for several proteins. Syntaxin 1A interacts with the cytoplasmic N-terminal domain of CFTR and inhibits channel activity (64). A family of PDZ-binding domain proteins, including the Na ϩ /H ϩ exchanger regulatory factor (65) and ezrin-binding protein 50 (66), bind to the C terminus (D-(S/T)-X-L) (Fig. 2). These proteins presumably couple CFTR to the cytoskeleton and may be involved in apical membrane localization. Finally, as mentioned above, CFTR is closely associated with protein phosphatase 2C (48).

Synthesis and Degradation
Attention has focused on the synthesis and degradation of CFTR because many mutations that cause CF, such as ⌬F508, lead to the misfolding of CFTR in the endoplasmic reticulum (ER) and subsequent degradation of CFTR (reviewed in Refs. 4 and 67). The synthesis of wild-type CFTR is inefficient. Only about 30% of newly synthesized CFTR develops endoglycosidase H-resistant mature glycosylation suggestive of translocation from the ER to the Golgi complex (68,69). In contrast, 100% of the homologous MDR matures (70). CFTR molecules that fail to mature are polyubiquitinated and degraded via a cytoplasmic, proteasome-dependent pathway (71,72). In the ER, CFTR transiently associates with several chaperones including the cytoplasmic Hsp70 and Hsp90 and the ER membrane chaperone calnexin (67,73). The molecular basis for the inefficient processing of CFTR in the protein synthetic pathway remains to be elucidated.

Conclusions
CFTR has an important role in epithelial Cl Ϫ transport both as a Cl Ϫ channel and as a regulator of other channels and transporters. Studies of the channel have begun to provide a picture of its structure and regulation. Understanding how CFTR interacts with other channels and transporter will provide insights into the cellular regulation of epithelial transport and the pathophysiology that results from defects in CFTR or from its overstimulation by bacterial toxins. One hopes that further studies will provide a basis for the design or identification of specific, high affinity CFTR inhibitors that would be useful for basic studies of the role of CFTR in epithelial Cl Ϫ transport and potentially as a therapy for secretory diarrhea, a major cause of morbidity and infant mortality in the developing world.