Endocytic Trafficking of CFTR in Health and Disease

Abstract

The cystic fibrosis transmembrane conductance regulator (CFTR) is a Cl-selective anion channel expressed in epithelial tissues. Mutations in CFTR lead to the debilitating genetic disease cystic fibrosis (CF). Within each epithelial cell, CFTR interacts with a large number of transient macromolecular complexes, many of which are involved in the trafficking and targeting of CFTR. Understanding how these complexes regulate the trafficking and fate of CFTR, provides a singular insight not only into the patho-physiology of cystic fibrosis, but also provides potential drug targets to help cure this debilitating disease.

INTRODUCTION

Coordinated cellular responses to hormones or neurotransmitters depend not only on the presence of the appropriate protein molecules within the cell, but also on the appropriate location of those proteins within the cell. For many genetic diseases, point mutations (resulting in protein sequence variants) lead not to a complete absence of the affected protein, but rather the production of a “misfolded” protein that fails to reach its appropriate location within the cell (1, 2). Disease pathology resulting from such mutations can result either from loss of protein and/or function at the appropriate subcellular domain, or the deposition of “misfolded” aggregates in inappropriate locales. Aggregation of “misfolded” proteins is observed in such diverse pathological conditions as Parkinson's and Alzheimer's diseases, prion diseases such as the spongiform encephalopathies and specific variants of α1-antitrypsin (α1-AT). In contrast, “loss-of-targeting” pathologies are associated with diseases such as retinitis pigmentosa, familial hypercholesterolaemia and cystic fibrosis. In these latter cases, failure of the mutant protein to fold appropriately is tightly associated with defects in endoplasmic reticulum (ER) export competency, protein trafficking and localization.

Cystic fibrosis (CF) (MIM 219700), the most common life-threatening genetic disease of Caucasians (3) is caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR; ABCC7). This anion selective ion channel is required for the normal function of epithelia lining the airways, intestinal tract as well as ducts in the pancreas and sweat glands. Characteristic of CF disease are an exocrine pancreatic insufficiency, an increase in sweat NaCl concentration and male infertility; however the major cause of morbidity and mortality is pulmonary disease (4). Although chronic inflammation and infection form a vicious cycle leading to fibrosis and loss of lung function, there is evidence of airway inflammation already present in CF neonates (5, 6). The loss of CFTR from airway epithelial cells also leads to altered regulation of other ion channels, significant changes in the composition of airway surface liquid (ASL) (7) and dysregulation of inflammatory elements (8, 9). Over 1,400 mutations have been identified in the CFTR gene (10) (http://www.genet.sickkids.on.ca/cftr/), however the most common mutation is loss of a Phe residue at position 508 (ΔF508-CFTR). Approximately 70% of individuals with CF are homozygous for the ΔF508-CFTR mutation, and almost 90% of patients have at least one ΔF508-CFTR allele. Based on the predominance of the ΔF508-CFTR mutation, most of the basic research and drug-development initiatives have focused on ΔF508-CFTR.

In the CFTR “life cycle”, there are four broad trafficking pathways that can be identified (i) biosynthesis, conformational maturation and trafficking from the endoplasmic reticulum (ER) to the plasma membrane, (ii) endocytic retrieval from the plasma membrane to early or sorting endosomes, (iii) recycling of endocytosed CFTR back to the cell surface (either directly or via recycling endosomes) and (iv) targeting of endocytosed CFTR for degradation (Figure 1). ΔF508 CFTR is recognized as misfolded by the ER quality control machinery and targeted for proteosomal degradation (Figure 1 (v)). Export of CFTR from the ER is dependent upon the presence of both acidic and basic exit codes (11, 12). These codes play a key role in linking CFTR to COPII-vesicles that deliver newly synthesized CFTR to the Golgi apparatus (11), prior to delivery of CFTR to the cell surface. Depending upon the cell system investigated, anywhere from 60 to ∼100% of wild-type CFTR folds appropriately in the ER and becomes “export competent”, trafficking to the Golgi and cell surface (13-15). In contrast, the ΔF508 variant fails to fold properly in the ER, is “ER export incompetent” and is rapidly degraded by the ER quality control machinery and never reaches the cell surface (16-18). Interestingly, early studies showed that growth of ΔF508 CFTR expressing cells at reduced temperature allowed the mutant CFTR molecules to exit the ER and reach the cell surface (19). However, this is likely due to temperature-dependent inefficiencies in the ER quality control machinery rather the acquisition of a “properly folded conformation” by ΔF508 CFTR (20). The search form pharmacological agents that facilitate the exit of ΔF508 CFTR from the ER is currently a major endeavour in both academic and pharmaceutical company laboratories, and of clear import in developing potential therapeutic interventions in CF patients. Although the biosynthetic trafficking of CFTR is an extremely important field, the present review will focus on the trafficking of CFTR after it has reached the cell surface. Details of current research on the biosynthetic trafficking of CFTR are available in several excellent reviews (21-23).

FIGURE 1.

Model showing main trafficking pathways taken by wild-type and ΔF508-CFTR

(i) CFTR is translated in the endoplasmic reticulum (ER) where core sugars are added to the protein. Most ΔF508-CFTR is recognized as misfolded by the ER quality control and targeted for proteosomal degradation (v). Wild-type CFTR traffics to the Trans Golgi network where the cores sugars are modified into complex carbohydrates, and then trafficked to the apical plasma membrane (i). (ii) CFTR is efficiently removed from the cell surface by clathrin mediated endocytosis using trafficking signals embedded in the amino acid sequence of CFTR. (iii) From endosomes, CFTR can recycle back to the cell surface in a direct manner, or via recycling endosomes. (iv) Internalized CFTR can be directed to lysosomes for degradation.

ENDOCYTIC PATHWAYS

Endocytosis is classically thought of as a process whereby nutrients and other molecules are taken up from the extracellular milieu (24). However, it is becoming increasingly clear that endocytosis is also a key element in regulating the cell surface density of a variety of integral membrane proteins, including G-protein coupled receptors (25), receptor tyrosine kinases (26), transport proteins (27) and ion channels (28-33). Indeed, the number of ion channels whose cell surface expression, and hence cellular function is governed by endocytosis and recycling has risen rapidly in the past few years. Moreover, there is accumulating evidence that mutations that alter the endocytic traffic of ion channels can have devastating clinical consequences. For example Liddle's syndrome is associated with reduced endocytic traffic of the epithelial sodium channel (ENaC) leading to sodium hyperabsorption (33, 34), whereas mutations in ROMK1, a channel internalized by clathrin mediated endocytosis (35), lead to Barter's syndrome (36), a salt-wasting disease. It is not surprising then, that the cell surface density of CFTR is dependent upon endocytic and recycling processes.

Although there are several endocytic mechanisms in cells (37), the best characterized uses the coat protein clathrin along with several accessory or adaptor proteins (37-39) to internalize integral membrane proteins from the plasmalemma. When clathrin coated vesicles (CCV) are isolated from either epithelial cells grown in culture or native intact epithelial tissues, large amounts of CFTR are seen to be present (40). Treatment of these CCV with coated vesicle “uncoatase' to strip off the clathrin allows fusion of the resultant vesicles (endosomes) with planar lipid bilayers (an electrophysiological technique to study the biophysical properties of isolated ion channels) leading to the incorporation of fully active CFTR channels. In contrast, it is not possible to detect any CFTR in caveolae (a non-clathrin mediated endocytic organelle (41-43)) purified from airway epithelial cells (44). Interestingly when overexpressed in heterologous cell systems, small amounts of CFTR can be detected in caveolae (44); likely an over expression artifact. Several tools are available that differentially affect clathrin and caveolae mediated endocytic pathways. Exposure of cells to chlorpromazine, K+ depletion or hypertonic sucrose causes an inhibition in clathrin mediated internalization (45) and a marked reduction in CFTR endocytosis (29, 44). In contrast, pharmacological agents that inhibit caveolae such as filipin, and treatment of cells with extracellular cholesterol oxidase (46) were unable to affect cell surface levels of CFTR. Taken together, these data strongly argue that CFTR is removed from the cell surface exclusively by clathrin mediated pathways. Although it is not possible to exclude fully the possibility of very low levels of CFTR entry into caveolae, the data argue that if such caveolar CFTR exists, it is a minor component of the total cell surface CFTR pool. However, it should be noted that recently Kowalski and Pier (47) argued that entry of CFTR into caveolae is critical for its function as a cellular receptor for Psuedomonas aeruginosa (48). These authors proposed that Pseudomonas-dependent entry of CFTR into caveolae triggers the internalization of a CFTR/Pseduomonas complex. Data presented in support of this hypothesis was based upon finding CFTR and caveolin-1 (a marker for caveolae) in similar fractions when Triton X-100 insoluble pellets were subject to a sucrose density gradient centrifugation protocol. However, such data merely indicates similar buoyant densities of the CFTR and caveolin-1 fractions, not necessarily co-localization within the same vesicle.

ENDOCYTIC SIGNALS ARE PRESENT IN CFTR

Whilst no-one would question that CFTR would be removed from the plasma membrane at some point during its life cycle, when early studies on CFTR endocytosis were performed it was surprising how rapidly CFTR was removed from the cell surface. Using a cell-surface biotinylation approach, Prince et al demonstrated that CFTR was efficiently removed from the apical membrane of polarized epithelial cells (49), with 25% of cell surface CFTR being internalized in 2.5 minutes; an uptake rate significantly greater than non-specific bulk uptake of other surface glycoproteins (49). In general, CFTR appears to be removed from the cell surface with rates not dissimilar from that of nutrient and signal molecule receptors, proteins typically associated with efficient clathrin mediated endocytosis. Such efficient endocytosis through clathrin mediated pathways argues for the presence of internalization motifs within CFTR. A complete review of endocytic signals is beyond the scope of this review; for a more in-depth analysis of sorting motifs, the reader is directed to a number of excellent review articles (38, 50-54). Briefly, endocytic sorting signals are generally short linear arrays of amino acids in the cytoplasmic domains of integral membrane proteins that, although not entirely conserved between proteins, are usually four to seven amino acids in length of which two to three are critical for function. The critical amino acids are usually bulky and hydrophobic. Two major classes of endocytic signal have been characterized, “tyrosine-based” and ‘”dileucine-based” depending upon which are the critical amino acids in the motif. Tyrosine-based motifs come in two flavours, NPXY where X is a variable amino acid, and YXXϕ where ϕ is a bulky hydrophobic amino acid. The NPXY endocytic signal was the first to be identified by Brown and Goldstein in the LDL receptor, (55) for which they won the 1985 Nobel prize in Medicine, yet this is a fairly rare motif and the more common sequence is the YXXϕ sequence. For other membrane proteins, di-leucine motifs (D/EXXXLL/I and DXXLL) are important for targeting proteins that bear them (e.g., LIMP11 and CD3-γ) to endosomal/lysosomal compartments (56).

Taking a chimeric approach, Prince and colleagues generated proteins containing the extracellular and transmembrane domains of the transferrin receptor fused to either the amino or carboxyl cytoplasmic domains of CFTR (31). In this context, both the amino and carboxyl terminal domains of CFTR were sufficient to direct efficiently the internalization of the fusion proteins from the cell surface. Using a similar chimeric approach in which domains of CFTR were fused to the interleukin 2 receptor α-chain (Tac) (57), Hu et al also demonstrated that the carboxyl terminal domain of CFTR contained sequences sufficient for efficient endocytosis of the resultant chimeras. Although the amino terminal domain was able to promote endocytosis of transferrin receptor chimeras, it has been difficult to demonstrate such a role within the context of intact full-length CFTR, since mutation of this region results in a misfolded protein that fails to reach the cell surface. Therefore, precisely what role the amino terminus plays in CFTR endocytosis remains unclear. Moreover, the issue of whether the amino terminus of CFTR plays a role in endocytic trafficking is complicated by the evidence that this region of CFTR also interacts with other components of the intracellular trafficking machinery, where syntaxin/CFTR interactions modulate CFTR movement and/or channel activity (58-61). If the interaction between the amino terminus of CFTR and syntaxins is necessary for physiological regulation of CFTR channel activity at the cell surface, an intriguing possibility is that regulated syntaxin/CFTR interactions would prevent the interaction of CFTR's amino terminus with the clathrin/adaptor endocytic machinery increasing the residence of CFTR at the cell surface. Despite the uncertainty of the role of the amino terminus of CFTR in modulating CFTR endocytosis, a clear consensus on the role of the carboxyl terminus of CFTR in endocytic trafficking has emerged.

Comparison of the amino acid sequences of CFTR carboxyl-termini from various species reveals the presence of a conserved tyrosine-based (YXXϕ motif) (Table 1). Mutational analysis in both CFTR chimeras and intact CFTR constructs confirms that ¹⁴²⁴YDSI is an authentic tyrosine internalization motif (31, 57, 62). Interestingly, both dog (Canis familiaris) and dogfish (Squalus acanthus) have a phenylalanine residue rather than tyrosine at the equivalent position to Y1424 in human CFTR. Phenylalanine residues are able to substitute for tyrosine residues and maintain wild-type internalization activity of the transferrin receptor (63), and mutation of tyrosine to phenylalanine within the context of human CFTR does not inhibit CFTR endocytosis (64), arguing that the endocytic signal is intact in dog and dogfish. Whether or not there are subtle differences in CFTR trafficking kinetics for proteins bearing a Phe residue compared to a Tyr residue at position 1424 are not known.

TABLE 1.

Comparison of the amino acid sequences of various CFTR carboxyl-terminal tails. Amino acid sequence alignment of carboxyl tails of CFTR from human (P13569), orangutan (Q2QL83), macaque (Q9TUQ2), lemur (Q2Ql83), dusky titi (Q2QLB4), horse (Q2QLA3), cow (P35071), pig (Q6PQ22), sheep (U20418), rabbit (Q00554), mouse (M60493), rat (1901178A), Xenopus (U60209), killifish (AF000271), dog (Q5U820) and dogfish (P26362). All sequences except dog and dogfish conform to the YXXϕ motif common to internalization signals, where X can be any amino acid and ϕ is a hydrophobic residue. Phenylalanine has been shown to substitute for tyrosine and still maintain wild-type internalization of CFTR (64) and the transferrin receptor (63).

Human	AFADCTVILCEHRIEAMLECQQFLVIEENKVRQYDSIQKLLNERSLFRQAISPSDRVKLFP	HRNSSKCKSK	PQIAALKEETEEEVQDTRL
Orangutan	AFADCTVILCEHRIEAMLECQQFLVIEENKVRQYDSIQKLLNERSLFQQAISPSDRVKLFP	HRNSSKCKSK	PQIAALKEETEEEVQDTRL
Macaque	AFADCTVILCEHRIEAMLECQQFLVIEENKVRQYDSIQKLLNERSLFRQAISPSDRVKLFP	HRNSSKCKTQ	PQIAALKEETEEEVQDTRL
Lemur	AFADCTVILCEHRIEAMLECQRFLVIEENNVRQYDSIQKLLSEKSLFRQAISPSDRMKLFP	RRNSSKHKSR	PPITALKEETEEEVQDTRL
Dusky	AFADCTVILCEHRIEAMLECQQFLVIEENKVEQYDSIQKLLNEKSLFQQAISHSDRVKLFP	HRNSSKYKSR	PQIASLKEETEEEVQETRL
Horse	AFADCTVILSEHRIEAMLECQRFLVIEENKVRQYDSIQKLLSEKSLFQQAISSSDPLKLFP	HRNSSKHKSR	SKIAALQEETEEEVQETRL
Cow	AFANCTVILSEHRIEAMLECQRFFVIEENKVRQYDSIQRMLSEKSLFRQAISPADRLKLLP	HRNSSRQRSR	SNIAALKEETEEEVQETKL
Pig	AFADCTVILSEHRIEAMLECQRFLVIEENKVRQYDSIQRLLSEKSLFRQAISPLDRLKLLP	HRNSSKQRSR	SKIAALKEETEEEVQETRL
Sheep	AFADCTVILSEHRIEAMLECQRFLVIEENKVRQYDSIQRMLSEKSLFRQAISPADRLKLLP	HRNSSRQRSR	ANIAALKEETEEEVQETKL
Rabbit	AFADCTVILCEHRIEAMLECQRFLVIEENTVRQYDSIQKLLSEKSLFRQAISSSDRAKLFP	HRNSSKHKSR	PQITALKEEAEEEVQGTRL
Mouse	AFAGCTVILCEHRIEAMLDCQRFLVIEESNVWQYDSIQALLSEKSIFQQAISSSEKMRFFQ	GRHSSKHKPR	TQITALKEETEEEVQETRL
Rat	AFAGCTVVLCEHRIEAMLDCQRFLVIEQGNVWQYDSIQALLSEKSVFQRALSSSEKMKLFH	GRHSSKQKPR	TQITAVKEETEEEVQETRL
Xenopus	AFADCTVILSEHRLEAMLECQRFLVIEDNTVRQYDSIQKLVNEKSFFKQAISHSDRLKLFPLHRRNSSKRKSR		PQISALQEEQEEEVQDTRL
Killifish	SFSGCTVILSEHKVEPLLECQSFLVIEKSSVRQYDSIQKLMNEMSHLKQAISPADRLHLFPTPHRLNSIKRPQPQTTKISALQEETEEEVQDTRL
Dog	AFADCTVILSEHRIEAMLECQRFLVIEDSRLRQFESIQRLLSERSAFRQAUGPPERPGLL	PHRLSSRQRSP	SRIAALKEETEDEVQDTRL
Dogfish	TFSNCTVILSEHRVEAILECQSFLVIEGCSVKQFDALQKLLTEASLFKQVFGHLDRAKLFTAHRRNSSKRKTR		PKISALQEEAEEDLQETRL

Fusing segments of CFTR with a truncated interleukin 2 receptor α chain (TacT) (57), Lukacs et al argued for the presence of three endocytic motifs in the carboxyl terminus of CFTR; a phenylalanine-based motif (¹⁴¹³FLVI), the previously identified tyrosine-based motif (¹⁴²⁴YDSI) and a di-leucine based motif (¹⁴³⁰LL). Whilst the Phe-based motif was not addressed by Collawn et al, their transferrin receptor chimeras also identified a potential di-leucine based motif at ¹⁴³⁰LL (31). Mutation of either the tyrosine motif alone ¹⁴²⁴YDSI→ADSI or the di-leucine motif alone ¹⁴³⁰LL→AL results in similar levels of inhibition in CFTR endocytosis (66% and 61% respectively) (57). Interestingly, in these studies, inhibition of CFTR endocytosis did not lead to increases in macroscopic analysis of halide efflux (57). It is possible that macroscopic assays such as iodide efflux assays are not sensitive enough to allow correlations between changes in cell surface density and channel function, therefore further assays such as whole cell patch clamping likely will be needed to address this issue. As with mutations in putative endocytic motifs in the amino terminus of CFTR, mutation of the Phe based signal (¹⁴¹³FLVI) also result in a protein that fails to fold appropriately and exit the endoplasmic reticulum, thus evaluation of ¹⁴¹³FLVI is not possible. The failure of ¹⁴¹³FLVI mutations to properly mature is likely due to the location of ¹⁴¹³F at the boundary between NBD-2 and the carboxyl terminus of CFTR. Studies by Gentzch et al (65) argue that a hydrophobic patch ¹⁴¹³FLVI is necessary for the structure and stability of NBD-2; the same sequence that was suggested to be an internalization motif based on chimera studies (57). Since endocytic sequences need to be exposed to the cytosol in order to interact with the endocytic machinery, it is not clear how a hydrophobic patch within the context of intact CFTR would fulfill the role of both a domain stabilizer and an internalization motif. Thus it is possible that the identification of ¹⁴¹³F as an internalization sequence may be a chimeric artifact. Based on structural considerations, homology with NBD-1 and homology with NBDs from other ABC transporters, Gentzsch et al have argued that the β-strand that delineates the last structural element of NBD-2 terminates at amino acid ¹⁴²⁴tyr (65). Crystal structures of the μ2 clathrin adaptor subunit complexed with the endocytic motifs of the EGF receptor and the Golgi protein TGN38 show that the endocytic motif adopts an extended β-strand conformation (66), and that the tyrosine residue cannot be present in a β-turn. Definitive answers to the structure of the ¹⁴²⁴YDSI internalization motif and the boundary of NBD-2 thus await future crystallization studies.

Are there any clinical mutations in endocytic signals?

There are no known point mutations in the endocytic signals of CFTR, however there are premature stop codons that give rise to molecules lacking that region of CFTR containing internalization motifs (67). While such molecules are apparently biosynthetically normal, the mature fully glycosylated molecules are very unstable and are substrates for proteosome-dependent degradation (68). Thus, the importance of inhibiting CFTR endocytosis in the context of a patient cannot be determined. While it is not possible to study ‘loss of endocytosis motif’ mutations, a mutation that results in the gain of an endocytosis signal has been described (69). A patient with a mutation N287Y (991A→T) was identified based on a diagnosis of elevated sweat electrolytes (70). Clinically, the patient presented as pancreatic sufficient, but did have 4-5 upper respiratory tract infections per year. Thus the patient can be considered to have a mild CF phenotype. Biosynthesis and delivery of N287Y CFTR to the cell surface is unaffected, but endocytosis of N287Y CFTR from the cell surface is markedly enhanced (∼ twice the rate of wild-type CFTR), resulting in a 50% reduction in steady-state levels of CFTR at the cell surface. In contrast to many mutations in CFTR, the single channel properties of N287Y CFTR and wild-type CFTR are indistinguishable arguing that disease cannot be attributed to conductance or gating defects, but rather to enhanced CFTR endocytosis. The genotype of the identified patient was ΔF508/N287Y (70). Since little or no cell surface CFTR is produced by the ΔF508 allele, the only CFTR protein is produced from the N287Y allele, this would potentially lead to a 75% reduction in the amount of CFTR at the cell surface compared with individuals homozygous for wild-type CFTR. Such data would argue that at least 25% of wild-type CFTR activity is required to ameliorate the disease symptoms in CF patients. It is of interest to note that the N287Y mutation is a gain of function mutation and appears to generate an endocytic signal that is present within the body of the protein rather than at the termini of the protein, a location not previously identified in polytopic membrane proteins. In the broader context of molecular mechanisms underlying the pathology of human genetic diseases, the significance of the observations on N287Y CFTR lies in the recognition that mutations can reduce the expression level of a membrane protein, not only by impairing its biosynthesis or stability (or in the case of ion channels their biophysical fingerprint), but also by accelerating endocytic retrieval from the plasma membrane.

R31C and R31L are CFTR mutations that also give rise to a mild clinical phenotype (71). Both of these mutations generate a CFTR molecule that has dramatically enhanced endocytic rates (72). Precisely why these mutations affect endocytosis is not clear, however it is possible that they introduce a hydrophobic residue relative to an upstream tyrosine Y, Y²⁸XXC³¹ and Y²⁸XXL³¹, or that such mutations somehow affect the recognition of the tyrosine-based motif in the carboxyl terminus to enhance endocytosis. Expression of R31C and R31L CFTR leads to reduced macroscopic currents compared to expression of wt CFTR; however, since single channel records were not determined it is not possible to determine whether the reduced macroscopic currents are due to endocytic defects alone, or whether there are also altered gating kinetics.

Recycling of CFTR

Many membrane transport proteins are rapidly recycled between intracellular vesicles and the cell surface, whereas others have a long residence on the plasma membrane. Recycling of membrane proteins serves several functions, (i) it allows receptors to internalize ligands, such as nutrients, hormones and toxins, (ii) recycling also allows cells to regulate the steady-state levels of proteins by altering the relative rates of endocytosis and exocytosis and (iii) recycling of membrane proteins also protects them from degradation and allows them to undergo multiple rounds of endocytosis and recycling. For example, the kidney collecting duct recycles the aquaporin water channel AQP2 between intracellular vesicles and the cell surface in response to vasopressin (73-75), and Glut4 glucose transporters recycle between endosomes/recycling vesicles and the cell surface in response to insulin (76-78). Endocytosed material is initially directed to an endosomal sorting compartment from which it is directed to one of two destinations (See Figure 2). Some molecules, such as low-density lipoprotein (LDL) are retained in the sorting endosomes that mature into and/or fuse with late endosomes and lysosomes for degradation (79). Molecules that are destined to be recycled back to the cell surface are initially targeted to the endosomal recycling compartment (ERC) (80).

FIGURE 2.

Model showing involvement of various proteins in CFTR endocytosis and recycling

CFTR (brown rectangle) is endocytosed from the apical plasma membrane in a clathrin-dependent process that requires dynamin (for vesicle fission), the μ subunit of the AP-2 adaptor complex that mediates interaction between the YDSI endocytic motif on CFTR and the clathrin lattice. The endocytosis of CFTR also requires myosin-VI, a molecular motor that drives cargo to the minus end of F-actin (i.e., inwardly directed). Several members of the RabGTPase family have been shown to modulate CFTR trafficking. Rab5 promotes endocytosis of CFTR. Exit of CFTR from early endosomes can be mediated by Rab4 which directly targets CFTR back to the plasma membrane, or by Rab7 which increases CFTR degradation by enhancing the trafficking of CFTR to late endosomes and lysosomes. Rab9 mediates CFTR trafficking from late endosomes to the trans Golgi network (TGN). Rab11 mediates CFTR trafficking from recycling endosomes to either the plasma membrane or to the TGN. RME-1 facilitates exit of CFTR from the recycling endosome. PDZ binding proteins can inhibit CFTR endocytosis from the plasma membrane, as well as facilitate recycling of internalized CFTR from early endosomes.

Given that endocytic internalization kinetics for CFTR are rapid (29, 49, 64) and that internalized CFTR is fully functional (81), the argument can be made that internalized CFTR is likely recycled back to the cell surface. Indeed, given the rapid internalization kinetics of CFTR, biosynthetic kinetics would have to be equally rapid to maintain steady-state surface CFTR were all internalized CFTR to be degraded. However, treatment of CFTR expressing CHO cells with either cycloheximide or brefeldin A does not cause any diminution of forskolin-stimulated CFTR currents up to 24 hours (82). Similarly, our laboratory has also shown that forskolin-stimulated CFTR currents are maintained in polarized T84 epithelial monolayers at least 24 hours after cycloheximide-dependent arrest of CFTR biosynthesis (unpublished observations). Lack of effect of biosynthetic inhibitors on CFTR mediated Cl secretion in the background of rapid CFTR endocytosis strongly argues that internalized CFTR is indeed recycled back to the cell surface. Using a cell surface biotinylation approach (62), our lab, and that of Stanton and colleagues, have shown that internalized CFTR is very efficiently recycled back to the cell surface with approximately 50% of internalized CFTR being recycled back to the cell surface (83, 84). Similarly, Lukacs and colleagues used an extracellular epitope tagged CFTR construct to clearly show that CFTR entering the early endosomal compartment is targeted for recycling back to the cell surface (85).

Whilst there is compelling evidence that CFTR channels undergo endocytic recycling, such is not the case for all ion channels. For example, once inserted into the cell surface, the inwardly rectifying potassium channel Kir3.4 is a stable component of the plasma membrane and when it does undergo endocytosis is degraded rather than recycled (30). Why CFTR should undergo such rapid internalization and recycling is not entirely clear. As with several other transporters, regulation of cell surface density and “gating” of resident transporters may be complementary modes of regulation. Indeed, given the pathological extremes observed with CFTR hyperactivity (as seen in toxin induced secretory diarrheas) and CFTR hypoactivity (as seen in cystic fibrosis), it is clear that CFTR protein/function has to be maintained within a very narrow window of activity. Another intriguing possibility underlying the physiological role of CFTR recycling arises from the notion that CFTR may take an unusual route from the ER to the cell surface. Balch and colleagues (86, 87) have argued that CFTR takes a non-conventional route from the ER to the Golgi. Thus in BHK and CHO cells, but not HeLa and 293T cells, CFTR is initially targeted directly from the endoplasmic reticulum to the late Golgi/endosomal compartment without traversing the medial or cis Golgi compartments. Acquisition of a mature fully glycosylated status would thus require recycling of CFTR through a late Golgi/endosomal system.

Regulated CFTR trafficking

In addition to undergoing constitutive endocytic recycling (83, 85, 88, 89), there is strong evidence that CFTR undergoes regulated trafficking between intracellular compartments and the cell surface (88). Acute regulation of CFTR insertion into and endocytosis from the plasma membrane is mediated by the cAMP-dependent protein kinase (PKA) (27-29, 31, 88, 90-92). The overall flow of ions across a membrane (assuming no other limiting factors) is governed by the equation I = iNPo, where I is the macroscopic current, i is the unitary conductance of the channel (in this case CFTR), N is the number of channel in the plasma membrane and Po is the open probability of the channel. It is therefore theoretically possible that a secretagogue could alter ion flow across a membrane by altering i, N or Po. In practice, i is rarely altered, leaving changes in N and/or Po to account for changes in transmembrane ion flow. When secretagogues stimulate Cl secretion through the cAMP/PKA mediated pathway, they do so on two levels. Firstly there is a direct activation of cell surface resident CFTR mediated by R-domain phosphorylation (increase in Po), and secondly there is a trafficking of CFTR from intracellular storage vesicles into the plasma membrane (increase in N). Interestingly, in some cell types, the ability of PKA to regulate CFTR trafficking depends upon the expression of wild-type CFTR. In cells from patients with cystic fibrosis, PKA fails to alter endocytosis (93).

In order to undergo regulated trafficking to the plasma membrane two critical requirements for CFTR must be fulfilled: the first is that CFTR is present in and stored in organelles beneath the plasma membrane and secondly, those storage vesicles should mobilize rapidly or translocate to the plasma membrane for insertion upon appropriate secretagogue or second messenger stimulation. Once in the plasma membrane, the newly inserted CFTR proteins (which may be intrinsically active or require activation in the plasma membrane) contribute to the overall chloride transport of the plasma membrane (27). The initial suggestion that CFTR may undergo regulated trafficking arose from observations based on morphologic studies characterizing the subcellular distribution of CFTR in a number of epithelial tissues. Though CFTR was recognized to be an apical membrane chloride channel, immunofluorescence localization studies of a variety of epithelial tissues suggested that in addition to its presence on the apical plasma membrane, CFTR could be detected in the sub apical cytoplasm in a punctate distribution, suggesting its location in a vesicular storage pool beneath the plasma membrane (94-96). Despite the suggestive immunofluorescence data, the precise subcellular localization for CFTR requires the use of electron microscopic techniques since fluorescence localization lacks the sensitivity to discriminate between CFTR that is localized to sub apical vesicles in close proximity (less than 100nm) to the apical membrane, and CFTR present in the apical membrane. Immunoelectron microscopic localization is the most sensitive method to distinguish clearly CFTR on the plasma membranes from that in vesicles. Sadly, few studies have localized successfully CFTR by immunoelectron microscopy because (i) CFTR is a low abundance protein in most cells (ii) its detection in tissues and cells is dependent on the use of high affinity specific antibodies (iii) its detection is sensitive to routine methods used for fixation and preparation of samples for electron microscopy (iv) the technique is labour intensive, expensive, and usually successful only in very experienced centers.

Nevertheless, successful characterization of the subcellular distribution of CFTR by immunoelectron microscopy has been accomplished in two endogenous CFTR-expressing tissues that express high levels of CFTR, using well characterized high affinity antibodies (87, 91, 97). Pools of endosomal CFTR have been documented by electron microscopic localization in the sub apical cytoplasm of rat submandibular glands, that co-localized with markers of receptor mediated endocytosis, confirming that CFTR undergoes receptor mediated endocytosis in that tissue (97). We used cryoimmunoelectron microscopy to examine the subcellular distribution of CFTR in rat intestinal tissues. We confirmed that CFTR was present on the membranes of the apical microvilli and on the membranes of sub apical vesicles within crypt cells. Quantitative analysis of the subcellular distribution of CFTR by immunogold labeling revealed that CFTR was equally distributed between the apical plasma membrane and intracellular vesicles in crypt cells, a distribution that was consistent with its role in regulated trafficking. This distribution suggested that like other solute transporters that regulate ion transport by regulated trafficking, cAMP-dependent CFTR mediated chloride and fluid secretion in the crypt may be regulated by acute insertion of cytoplasmic vesicles into the apical plasma membrane. This was confirmed in a recent study in rat small intestine. Intravenous administration of vasoactive intestinal peptide (a potent physiologic cAMP agonist) to rats resulted in an acute redistribution of CFTR to the apical surface of crypt enterocytes that was associated with fluid secretion (98).

Although there is compelling evidence for PKA-dependent trafficking of CFTR in intestinal cells, there is no direct evidence for such trafficking of CFTR in airway epithelial cells (88, 99, 100). In fact, direct measurements of CFTR in the plasma membrane of airway epithelial cells show no change in CFTR levels upon PKA stimulation (99, 100). Why should there be a difference in CFTR trafficking patterns between intestinal and airway epithelial cells? It is possible, even probable, that the proteins which interact with and regulate CFTR activity and trafficking display cell-type specificity. Thus the complement of proteins necessary to impart PKA sensitivity to CFTR trafficking may be expressed in intestinal but not airway cells. The difference between PKA dependent CFTR trafficking in airway and intestinal cell may also reflect a fundamental difference in the physiology of airway and intestinal epithelial tissues. Although Cl- is the dominant ion passed through CFTR in intestinal cells, there is a strong argument that CFTR may pass bicarbonate (HCO₃-) in airway cells (101, 102). For a more comprehensive discussion on regulated trafficking of CFTR and epithelial secretion, the reader is directed to an excellent comprehensive review on this topic (88).

GTPases, myosin-VI and PDZ sequences modulate CFTR trafficking

Several molecules are now known to regulate vesicular trafficking through the endosomal and recycling compartments, including Rabs, Rme-1 and myosins (80, 103-112). Rab GTPases are molecular switches that have several cellular functions, including cargo selection during vesicle transport, tethering of vesicles to motor proteins, and facilitating vesicle docking and fusion (113). Several Rab GTPases have been shown to modulate the intracellular trafficking of CFTR (89, 114) (Figure 2). Rab5 facilitates the trafficking of CFTR from the cell surface to early endosomes, whereas Rab11 controls trafficking of CFTR from early endosomes to the trans Golgi network (TGN) and also trafficking of endocytosed CFTR back to the cell surface. Rab7 regulates movement of CFTR from early endosomes to late endosomes and from late endosomes to lysosomes. Alternatively, CFTR in late endosomes can undergo Rab9-dependent trafficking to the TGN and thence back to the cell surface. Given the importance of Rabs in regulating CFTR trafficking, it is not surprising that manipulation of Rab activity can be used to alter the cell surface levels of both wt and ΔF508 CFTR (89, 114). Indeed, it is possible that Rab GTPases may be potential therapeutic targets for treating CF.

RME-1 is GTPase that regulates exit from the endosomal recycling compartment (ERC). Rme-1 plays no role in endocytosis, or in trafficking of material from early endosomes to lysosomes. A mutation near the EH domain of RME-1 (G429R) has a dominant-negative phenotype and causes the protein to associate tightly with the ERC, preventing exit but not entry of endocytosed material into the ERC. Trafficking of material such as LDL from the cell surface to early endosomes to lysosomes is unaffected by G429R Rme-1, since such trafficking bypasses the recycling pathway. For molecules, such as transferrin, that recycle, G429R Rme-1 has no impact upon endocytosis, transport to early endosomes or transport into the ERC; yet once in the ERC the exit of molecules such as transferrin is blocked leading to an accumulation in, and an expansion of, the ERC (80, 110). Expression of G429R Rme-1 has no effect on CFTR endocytosis, which continues uninterrupted. However when endocytosed CFTR enters the recycling compartment, it is trapped and cannot exit back to the cell surface. The block in recycling without a block in endocytosis of CFTR leads to a marked steady-state redistribution of CFTR such that almost no CFTR is present in the cell surface, and the majority of CFTR is trapped in an expanded ERC (83). This redistribution of CFTR away from the cell surface to recycling compartments in the presence of G429R Rme-1 argues that maintenance of steady-state cell surface levels of CFTR is primarily due to recycling of internalized CFTR rather than insertion of newly synthesized CFTR.

Myosin motors are a large superfamily of proteins that use the energy from ATP hydrolysis to move themselves (and vesicles bound to them) along actin filaments (112, 115). Myosin VI is distinct from other myosins, in that in moves cargo towards the F-actin minus end (i.e., away from the plasma membrane into the cell), a direction consistent with a role in endocytosis (116, 117). Expression of the tail domain of myosin VI, a dominant-negative recombinant fragment that retains cargo binding ability but no motor head domain, increases the expression of CFTR at the plasma membrane, by inhibiting endocytosis of CFTR. Consistent with its function in endocytic traffic, tail domains of myosin VI have no effect on the recycling of CFTR back to the cell surface (118).

PDZ domains are a common module found in mammalian proteins (119-121). PDZ domains are 80-100 amino acid sequences that mediate protein-protein interactions by binding to short peptide sequences that are often, but not always, at the carboxyl termini of proteins. In this way, PDZ proteins are able to form macromolecular complexes, cluster and colocalize transport proteins, channels and signaling molecules in specific subcellular domains, and modulate protein trafficking. The carboxyl terminus of CFTR contains a PDZ domain binding motif (DTRL) that plays several potential roles in CFTR trafficking. NHERF-1 (also known as EBP50), NHERF-2, NHERF-3 (also known as CAP70), NHERF-4 and CAL (CFTR associated ligand) are all able to bind the DTRL motif on CFTR. NHERF-1 and 2 also possess regions that allow them not only to interact with CFTR, but also with the apical actin cytoskeleton. Such interactions may tether CFTR in a stabilized network in specific apical compartments, preventing its removal from the cell surface by clathrin mediated endocytosis. Since PDZ proteins are also able to target PKA to macromolecular complexes with CFTR (122), such interactions may facilitate the PKA dependent inhibition of CFTR internalization as well as PKA dependent activation of CFTR's ion channel activity. NHERF-1 possesses two PDZ domains, leading to the speculation that NHERF-1 may facilitate CFTR dimerization (123, 124). Moreover, dimerization of CFTR has been suggested to be dependent upon PKA activation (125). Although the dimerization of CFTR is still controversial (126), an intriguing model has been suggested, in which dimerization of CFTR may prevent its endocytic retrieval from the cell surface, leading to an increase in the number of CFTR molecules on the cell surface. The model further proposes that PKA activation would lead to a dimerization of CFTR (inhibiting its endocytosis) and an activation of CFTR channel activity at the cell surface (126), consistent with the observations that PKA reduces the endocytosis of CFTR from the plasma membrane (29, 49). Following PKA inactivation, dephosphorylation of CFTR would lead to channel inactivation, and a switch from dimeric to monomeric CFTR would then lead to a loss of CFTR protein from the cell surface by allowing monomeric CFTR to be removed by endocytosis.

There is also some evidence that the PDZ binding domain of CFTR plays a role in CFTR recycling. Removal of the last 3 amino acids of CFTR (CFTR-ΔTRL) causes a reduction of cell surface stability of the resultant protein, not by altering endocytosis but rather by reducing the ability of the CFTR-ΔTRL molecule to recycle (84). Whether the loss of the PDZ binding domain leads to a decreased ability of CFTR-ΔTRL to enter the recycling pathway or to an increased ability of CFTR-ΔTRL to enter a degradative pathway is not known. In addition, what PDZ proteins CFTR interacts with to undergo endocytic recycling are not known. Interestingly, the PDZ binding domain of CFTR appears to be a self-contained recycling element, and is able to function as a transplantable signal to facilitate the efficient recycling of G-protein coupled receptors to which it is appended (127). For a detailed description of the role of macromolecular complexes in regulating CFTR trafficking through the biosynthetic pathway, and the regulation of CFTR activity at the cell surface, the reader is directed to an excellent review by Guggino and Stanton (119).

Recycling of ΔF508 CFTR

Given the low levels of ΔF508 in the plasma membrane of cells grown at 37°C (but see also (128)) it has been difficult to perform trafficking assays on ΔF508-CFTR. Experimentally, the most robust means of increasing the cell surface expression of ΔF508-CFTR has been to grow cells at reduced temperature (19), so-called ‘temperature-rescued’ ΔF508-CFTR. Clues to the endocytic trafficking of ‘temperature-rescued’ ΔF508 CFTR first came from studies by Marino and colleagues (129) who showed a markedly reduced cell surface stability of ΔF508 compared to wt CFTR. Later quantitative studies showed that there was almost a 10-fold decrease in the surface half life of mutant CFTR (85). Such a decrease could be due to accelerated internalization, reduced recycling and/or targeting of endocytosed CFTR to degradation. Sharma and colleagues showed that the endocytic rate constant for ΔF508 CFTR was not significantly different from that of wt CFTR (85), arguing for abnormalities in ΔF508 recycling. Internalized ΔF508 did not enter the recycling compartment, but rather was “tagged” as misfolded in the sorting endosome by the addition of ubiquitin and targeted for lysosomal degradation. In contrast to the studies of Sharma, those of Stanton and colleagues argue that the cell surface stability of ΔF508 CFTR is reduced due to accelerated endocytosis from the cell surface, with no alteration in endocytic recycling (89). The molecular mechanisms whereby ΔF508 undergoes enhanced endocytosis were not elaborated in these studies. The basis for the disparity in the mechanisms underlying the reduced cell surface stability of ‘temperature-rescued’ ΔF508, decreased recycling vs. increased endocytosis, is not entirely clear. Although both studies used over expression of ΔF508 in epithelial cells, different cell types were employed. Studies using Panc-1 cells (a pancreatic duct cell line) (85) showed ΔF508 endocytosis to be normal and recycling abnormal, whereas studies CFBE41o⁻ cells (an airway epithelial cell) showed ΔF508 endocytosis to be enhanced (89). In addition, different methodologies were used to monitor CFTR recycling (epitope-tagged CFTR (85) and cell surface biotinylation (89)). Thus, while there is general agreement on the observation that the cell surface stability of ΔF508 is significantly reduced compared to wtCFTR, the precise mechanisms responsible for this reduction are far from clear.

Future Directions

Given that CFTR is an integral membrane protein, one can easily assert that CFTR is obliged to traffic through the cell within membrane limited organelles. We are beginning to understand the compartments through which CFTR traverses on its way into, and out of the plasma membrane as well as the molecules which play a role in mediating the entry and exit of CFTR between these compartments. However, it is likely that we have underestimated the complexity and the number of proteins involved in CFTR trafficking. For example, many of the studies on CFTR trafficking have been performed in heterologous expression systems, and it is now critical to move such studies forward into fully polarized epithelial cells such as airway, intestine and exocrine pancreas cells that are affected by cystic fibrosis. A thorough understanding of how CFTR traffics within its native environment is essential to identifying potential therapeutic targets for cystic fibrosis therapy. Indeed, it is also important to understand how mutations in CFTR affect its trafficking. A large amount of time, effort and money is being focused on pharmacological strategies to increase the exit of ΔF508 CFTR from the ER and its subsequent insertion into the plasma membrane. Given the present review, such an approach is clearly only half the therapeutic battle in maintaining mutant CFTR at the plasma membrane. The question of how much “correction” ΔF508 needs to achieve in order to cure cystic fibrosis is one that is central to drug development, and will become increasingly important as we move into clinical trials. Underlying the question of “how much ΔF508 correction do we need?” is the complication of what it is that needs correcting in ΔF508. ΔF508 has a folding problem in the ER, leading to its degradation rather than export to the Golgi. Since the phenylalanine residue at position 508 will always be missing (absent gene therapy), ΔF508-CFTR can never fully achieve a wild-type conformation. The question then is how close an approximation of wild-type CFTR structure can ΔF508 attain? The answer to that question lies in what physiological role ⁵⁰⁸Phe plays. Although it has been proposed that ⁵⁰⁸Phe is involved in intramolecular domain-domain interactions, there is still no definitive evidence in support of this. In heterologous expression systems, the efficiency with which wild-type CFTR is exported from the ER is ca. 60%. If the export efficiency of ΔF508 CFTR is ca. 1%, then we must increase ΔF508 CFTR ER export efficiency by ca. 60-fold. In native expression systems where the ER export efficiency of wild-type CFTR is ca. 100%, we must potentially increase ΔF508 CFTR ER export by up to 100-fold. Facilitating ΔF508-CFTR export by growth of cells at reduced temperature suggests that even when ΔF508-CFTR reaches the plasma membrane it is still dysfunctional. Open probability estimates argue that ΔF508-CFTR has a Po ca. 10-fold less than wild-type CFTR (130), arguing that we made need to increase rates of biosynthetic insertion of ΔF508-CFTR into the plasma membrane to 10-fold above the rate for wild-type CFTR. Once in the plasma membrane, ΔF508-CFTR has a significantly reduced stability compared to wild-type CFTR; due to an inability of ΔF508-CFTR to recycle properly. Whether a single drug will be able to facilitate ΔF508 exit from the ER and enhance ΔF508 recycling is debatable. Perhaps a combination of drugs targeted at different aspects of ΔF508 trafficking will be required to provide therapeutic benefit. Although ΔF508 is by far the most common mutation in CFTR, there are many hundreds of other mutations described, the trafficking consequences of which have not been evaluated. For example, there is nothing known about the trafficking of G551D CFTR, a mutation common in Scottish and Scandinavian populations. Nevertheless, as we begin to unravel the mysteries and intricacies of CFTR's endocytic travels, we are likely to come across important new pharmacological targets for maintaining cell surface CFTR.