Biogenesis of cystic fibrosis transmembrane conductance regulator (CFTR) starts with its cotranslational insertion into the membrane of the endoplasmic reticulum (ER) and core glycosylation. These initial events are followed by a complex succession of steps with the main goal of checking the overall quality of CFTR conformation in order to promote its exit from the ER through the secretory pathway. Failure to pass the various checkpoints of the ER quality control targets the most frequent disease-causing mutant protein (F508del-CFTR) for premature degradation. For wild-type CFTR that exits the ER, trafficking through the Golgi is the major site for glycan processing, although nonconventional trafficking pathways have also been described for CFTR. Once CFTR is at the cell surface, its stability is also controlled by multiple protein interactors, including Rab proteins, Rho small GTPases, and PDZ proteins. These regulate not only anterograde trafficking to the cell surface, but also endocytosis and recycling, thus achieving fine and tight modulation of CFTR plasma membrane levels. Exciting recent data have related autophagy and epithelial differentiation to the regulation of CFTR trafficking. Herein, we review the various checkpoints of the complex quality control along the secretory trafficking pathway and the associated pathways that are starting to be explored for the benefit of cystic fibrosis patients.
The cystic fibrosis transmembrane conductance regulator (CFTR) protein, which, when mutated, causes the autosomal genetic disease cystic fibrosis (CF), is an integral membrane glycoprotein that functions as a cAMP-activated and phosphorylation-regulated Cl− channel at the apical membrane of epithelial cells. CFTR (also called ABCC7) is a member of the ATP-binding cassette transporter superfamily, and its structure contains two membrane-spanning domains (MSD1 and MSD2), two nucleotide-binding domains (NBD1 and NBD2), and a regulatory domain (RD), which is a unique feature among ATP-binding cassette transporters (Fig. 1). As happens with other multidomain glycoproteins, CFTR is cotranslationally inserted into the endoplasmic reticulum (ER) membrane and concomitantly N-linked to glycosyl groups. Generally, these post-translational modifications, as well as interactions with chaperones and folding, occur through an iterative process until the protein acquires its fully folded native conformation [1, 2]. Being a membrane protein, CFTR follows the general route of the secretory pathway, i.e. through the Golgi, to reach the plasma membrane (PM).
The most common disease-causing mutation consists of the deletion of three nucleotides that encode Phe508 in the polypeptide chain (F508del), and this leads to the protein failing to reach the PM. This mutation is present in 90% of CF patients in at least one of their two mutated CFTR alleles.
CFTR biogenesis and glycosylation at the ER – folding checkpoints of ER-to-Golgi trafficking
The biogenesis of CFTR begins in cytosolic ribosomes that are targeted through the signal recognition particle to the ER membrane translocon (Sec61 complex) via the signal recognition particle receptor [1, 3]. The CFTR polypeptide chain emerging from the ribosome is cotranslationally inserted into the ER through the sequential and coordinated action of signal and stop transfer signals through the translocon .
Immediately after insertion into the ER membrane, the newly synthesized CFTR polypeptide chain emerging into the ER lumen, either wild type or mutant (F508del), is core-glycosylated at Asn894 and Asn900, located in the fourth extracellular loop (Fig. 1). N-glycosylation occurs by the addition of a glycoconjugate with 14 osidic residues (reviewed in [4, 5]). N-glycosylation plays a pivotal role in glycoprotein intracellular processes such as folding, sorting, and trafficking . Generally, these oligosaccharides serve multiple functions by acting mainly as stabilizers and protective shields for the glycoprotein outside the cell and as recognition targets in adhesion and immune response modulation.
However, at the ER, glycosylation plays an important role in the quality control process that assesses the folding status of proteins destined for the secretory pathway. For CFTR, an integral membrane protein, glycosylation at the ER creates the core-glycosylated (immature) form of the protein (also called band B), which is used to assess folding by the lectin/chaperone calnexin. For proteins undergoing secretion (and thus being biosynthesized and translocated into the ER lumen), the folding status at the ER is assessed by calreticulin instead. From the ER, wild-type CFTR traffics through the Golgi complex, where it is processed by multiple Golgi glycosyltransferases, creating the fully mature form of CFTR (also termed band C) . Usually, these bulky glycans located at the surface of secretory and cell membrane glycoprotein molecules are mostly exposed to the extracellular space. Interestingly, however, a recent study manipulated the 6–7 N-glycan sites of another human membrane glycoprotein, tyrosinase, exposing these glycans to the lumen of subcellular organelles. The results of this study demonstrated that the removal of two close glycosylation sites arrests the post-translational productive folding process, resulting in terminally misfolded mutants that are subjected to degradation through the mannosidase-driven ER-associated degradation pathway .
The most frequent mutant protein, F508del-CFTR, is almost completely retained at the ER, owing to its incorrect folding, and from there it is degraded via the ubiquitin–proteasome pathway (UPP) (reviewed in [4, 5]). The machinery responsible for this retention is generally called the ER quality control (ERQC)  and, despite the large amount of data already gathered from multiple studies on the ERQC, it is still unclear how it distinguishes a native protein conformation from a misfolded one. However, it is generally accepted that the folding status of proteins depends both on intrinsic structural motifs (‘protein-autonomous’), which are determined by its primary sequence, and on extrinsic factors (‘protein-nonautonomous’) present in the cellular milieu, e.g. molecular chaperones . Understanding the factors that correct folding and/or prevent ER retention of F508del-CFTR may ultimately be of benefit for the treatment of CF. As the protein-autonomous factors determining CFTR folding and, in particular, F508del-CFTR misfolding were recently reviewed , we focus here on protein-nonautonomous factors and the respective folding assessment checkpoints.
Chaperone complexes containing Hsc70/Hsp70 and Hsp90 are the two major cytosolic folding machines that monitor the folding status of cytoplasmic proteins or those possessing large cytoplasmic domains. Both have been reported to participate in CFTR folding [11-14] (reviewed in  and ).
Although Hsp70 has long been known to play a key role in the machinery that distinguishes between wild-type CFTR and F508del-CFTR, recent data have shown that F508del-CFTR (but not wild-type CFTR) is present as a stalled folding intermediate in stoichiometric association with the core heat-shock proteins Hsp40, Hsp70, and Hsp90, referred to as the ‘chaperone trap’ . In fact, a previous study characterized such Hsp70 interactions in vitro to clarify the mechanism that senses misfolded F508del-CFTR in vivo. Interestingly, it was found that isolated F508del-NBD1 binds Hsc70 with higher affinity than wild-type NBD1, and that the Hsp70–NBD1 binding affinity can be reduced by either ATP or ADP. These nucleotides, however, increase the difference between the Hsp70-binding affinities for F508del-NBD1 and wild-type NBD1, whereas one small-molecule CFTR corrector can reduce it .
These and other findings are in full support of a previously proposed model that describes two major checkpoints for the early ERQC of CFTR . According to this model, whereas wild-type CFTR passes through this first Hsc70-mediated ERQC checkpoint, F508del-CFTR is kinetically trapped here (Fig. 2), because Hsp/Hsc70 recognizes (and strongly binds to) the exposed hydrophobic residues of mutant protein, as its newly synthesized polypeptide chain fails to acquire the native conformation. Although Phe508 is located at the outer surface of NBD1, without having a major impact on the structure of the isolated domain , it is now evident from full-length CFTR modelling data that it makes crucial contacts during the interdomain folding process [19, 20]. This is also supported by earlier data showing that formation of Hsc70–Hdj-2 complexes with nascent wild-type CFTR is greatly reduced after expression of the RD, suggesting that the intramolecular NBD1–RD interaction, catalyzed by Hdj-2/Hsc70, is a critical step in the CFTR folding pathway and that it is defective in the biogenesis of F508del-CFTR .
The second ERQC checkpoint (Fig. 2) involves the core glycosylation of CFTR (see above) and the calnexin cycle . In contrast to F508del-CFTR, which is mostly retained for UPP degradation at the Hsp70 folding checkpoint, wild-type CFTR proceeds along the folding pathway through this cycle, where it undergoes successive rounds of deglucosylation and release from and reglucosylation and rebinding to the ER membrane chaperone/lectin calnexin . According to this model, some wild-type CFTR that is unable to adopt a folded conformation, and possibly a small amount of F508del-CFTR that escapes degradation at the first ERQC checkpoint, undergo proteolytic glycan-mediated endoplasmic reticulum-associated degradation (glycoprotein endoplasmic reticulum-associated degradation) at this second (calnexin-dependent) checkpoint .
A third ERQC checkpoint occurs when CFTR is transported to the Golgi via coat protein II (COPII)-coated vesicles that form at the ER exit sites . At this point, misfolded F508del-CFTR fails to exit the ER, owing to exposure of ER retention motifs – four arginine-framed (RXR) tripeptides (AFTs). Indeed, upon replacement of one Arg by Lys in each of these motifs, F508del-CFTR is released to the cell surface [22, 23]. Besides the AFT retention motifs, active export of wild-type CFTR from the ER also relies on the presence of a di-acidic code (the DAD motif located in NBD1; Fig. 2), which acts as a positive cargo signal that is necessary for Sec24-mediated COPII packing , and whose disruption reduces both Sec24–CFTR association and ER exit . The mechanism described above for the early stages of CFTR trafficking constitutes the so-called ‘conventional’ secretory pathway.
Nonconventional CFTR trafficking
Another mechanistic model describes an ‘unconventional’ trafficking route for CFTR transport to the PM that is insensitive to blocking of the conventional trafficking pathway from the ER to the Golgi . This route involves tubular structures migrating peripherally to the central Golgi, and appears to be dependent on syntaxin 13 . According to this model, folded CFTR and misfolded CFTR are captured with equal efficiency by the COPII machinery. However, only CFTR that has undergone a degree of maturation will reach the cis-Golgi, and the rest is recycled back to the ER in COPi (coat protein I) vesicles . These recycled proteins re-enter the folding process, and are either sent once more to the cis-Golgi, or converge towards ubiquitination and are degraded via the UPP .
More recently, ER stress was reported to induce cell surface trafficking of the ER core-glycosylated wild-type CFTR and F508del-CFTR via a Golgi reassembly stacking protein (GRASP)55-dependent pathway . In this study GRASPs were proposed to be among the tethering factors that: (a) are involved in the ER stress-induced nonconventional secretion; (b) specifically associate with cargo molecules through their PDZ domains; and (c) are activated by specific upstream kinases .
The major final difference between these two ‘unconventional’ trafficking routes for the transport of CFTR to the cell surface is actually the protein glycosylation status. In the former, the protein still travels back to the early Golgi for oligosaccharide processing into the complex form . In contrast, when trafficking occurs through the GRASP55-dependent route, CFTR reaches the PM in its core-glycosylated form . As complex glycosylation is a critical factor in the stabilization of membrane proteins, the therapeutic significance of this alternative route as a strategy to rescue mutant CFTR is probably limited.
Autophagy and CFTR traffic
As mutant CFTR is targeted for degradation at the proteasome, the formation of protein aggregates, termed ‘aggresomes’, has been reported to occur . Such aggresomes are then degraded mainly through autophagy through the lysosomal pathway. Recently, this pathway was shown, for the first time, to also play a role in the mechanisms responsible for discarding mutant protein. In fact, dysregulation of the autophagy machinery in F508del-CFTR-expressing cells was implicated in the impairment of ‘aggresome’ clearance . Luciani et al. have shown that upregulation of reactive oxygen species production (also a characteristic of cells expressing defective CFTR) leads to aggresome sequestration of phosphatidylinositol-3-kinase complex III and accumulation of p62, which regulates aggresome formation, a mechanism mediated by aggresome sequestration of the autophagy-related protein beclin 1 . Interestingly, both restoration of beclin 1 and treatment with either cystamine or antioxidants are able to rescue F508del-CFTR trafficking to the membrane, an effect later shown to increase the activation of F508del-CFTR activity by the potentiators genistein, VX-770 (Kalydeco), and VRT-532 .
This autophagy defect was also reported to impair the clearance of Burkholderia cepacia vacuoles in F508del-CFTR-expressing macrophages. Remarkably, stimulation of autophagy by rapamycin decreases B. cepacia infection, both in vitro by enhancing clearance via induced autophagy, and also in vivo by contributing to a decrease in the bacterial burden in the lungs of CF mice .
Regulation of CFTR at the PM – the role of the Ras superfamily of small GTPases
Regulation of CFTR trafficking, turnover and retention at the surface of epithelial cells is also a complex process in which the folding status of the protein is assessed . Several members of the Rab and Rho subfamilies of the Ras superfamily of small GTPases have been implicated in the regulation of these late stages of PM trafficking. This superfamily, the founding members of which are the Ras oncoproteins, comprises over 150 human members, divided into five major branches on the basis of sequence and functional similarities: Ras, Rho, Rab, Ran, and Arf [35, 36]. These low molecular mass proteins (~ 200 amino acids) have a structurally and mechanistically preserved GTP-binding core, despite considerable divergence in sequence and function. These variations dictate specific subcellular locations, which, in conjunction with a complex network of the proteins that serve as their regulators and effectors, allow these small GTPases to modulate a remarkably complex and diverse range of cellular processes [36, 37].
Rab GTPases constitute the largest branch of the Ras superfamily, with > 60 members so far having been characterized . Rab proteins are key regulators of both vesicular transport and trafficking of proteins between different organelles of the endocytic and secretory pathways . They allow motor proteins to interact with membranes, facilitating vesicle motility, and, through interactions with a multitude of other proteins, they coordinate the correct docking and fusion of vesicles with the appropriate target membranes [38, 39]. Several Rab GTPases have been implicated in the regulation of the intracellular transport and the PM delivery of CFTR (Fig. 3). Mature, fully glycosylated CFTR exits the trans-Golgi network (TGN) in vesicles that enter the exocytic pathway, ultimately leading to the insertion of the channel into the PM . This process is facilitated by myosin Vb (Myo5b), a molecular motor that drives vesicle transport through actin cytoskeleton fibres, and is controlled by Rab11, which physically interacts with both CFTR and Myo5b .
As CFTR internalization at the cell surface is a rapid process , and its biosynthesis and maturation occur slowly, the recycling of internalized channels is considered to be a key process in maintaining a functional pool of CFTR at the PM . Indeed, several studies have shown that CFTR is continuously recycled from early endosomes back to the PM via Rab11/Myo5b-driven recycling endosomes [41, 44]. Interestingly, Rab4, which has been implicated in rapid recycling of surface proteins , and Rab27a, a critical regulator of the exocytic pathway , were both shown to physically interact with CFTR (Fig. 3), restraining its localization to intracellular compartments, and thus limiting channel expression at the PM [46, 47]. The trafficking of CFTR from the PM to early endosomes is controlled by Rab5 , a critical regulator of the fusion of endocytic vesicles with these compartments . Another Rab protein, Rab7, regulates the movement of CFTR away from the recycling pathway and into late endosomes . Rab7 is also thought to participate in the transport of CFTR from late endosomes to lysosomes for degradation (Fig. 3). Indeed, Rab7 overexpression reduced not only the surface levels but also the endosomal pools of CFTR . In contrast, Rab9 can move CFTR away from lysosomal degradation by mediating its transport from late endosomes back to the trans-Golgi, from which CFTR may re-enter the secretory pathway leading to PM insertion (Fig. 3). Although some in vitro studies have shown that F508del-CFTR cell surface expression can be increased through the manipulation of key Rab GTPases [48, 49], the mechanisms involved are still unclear. However, a better understanding of these processes may yet provide valuable targets to be used to identify new drugs for the treatment of CF.
Although recycling of internalized CFTR to the PM has been considered to be the main mechanism for sustaining a functional pool of CFTR at the cell surface, a study using multiple cell types, including airway epithelial cells, showed that up to 50% of surface CFTR exists in an immobile pool, tethered to filamentous actin (F-actin) . Tethering of CFTR to the PM involves the interaction of its C-terminal domain with the PDZ adaptor protein Na+/H+-exchanger regulatory factor isoform-1 (NHERF-1) [50-52]. NHERF-1 is important for targeting of exosome- and endosome-associated CFTR to the apical membranes of polarized epithelial cells and for the anchoring of CFTR at the PM to the apical actin cytoskeleton. The latter involves the interaction of CFTR-bound NHERF-1 with the ezrin/radixin/moesin (ERM) family protein ezrin, which locks CFTR in an immobile, actin-tethered complex, preventing its endocytosis [50, 52]. Interestingly, annexin 5A was also shown to augment CFTR whole-cell currents independently of the CFTR PDZ-binding domain, indicating that this effect is related to the annexin 5A scaffolding role and not to CFTR-regulated PM traffic .
Notwithstanding this, the importance of the actin cytoskeleton in CFTR recycling and immobilization at the PM was highlighted when a dramatic decrease in the surface CFTR pool, with a corresponding increase in the amount of intracellular CFTR, was observed upon N-WASP inhibition and actin cytoskeleton disruption . WASP proteins promote actin polymerization in response to signalling molecules, namely those of the Rho family of small GTPases. Rho GTPases are found in all eukaryotic organisms, and are divided into three subfamilies, grouped according to their functional and structural similarity to their three founding members, RhoA, Rac1, and Cdc42. Rho proteins are key regulators of actin cytoskeleton dynamics [55, 56], but have been also implicated in the regulation of cell polarity and membrane trafficking through their modulation of F-actin remodelling [57-59]. Consistently, TC10, a member of the Cdc42 subfamily, was shown to increase CFTR levels at the cell surface by facilitating the actin-driven targeting of CFTR-containing vesicles to the PM  (Fig. 3). Moreover, activated TC10 also reduces CFTR degradation by favouring the binding of NHERF-1 to the C-terminus of CFTR, thus preventing its binding to CAL, another PDZ protein that facilitates the trafficking of CFTR to lysosomes [60, 61] (Fig. 3). NHERF-1 overexpression is also sufficient to overcome CAL-induced lysosome targeting of CFTR , and increase polarized expression of CFTR on the apical membranes of airway cells, stimulating the vectorial transport of Cl− . Importantly, overexpression of NHERF1 was also shown to promote apical expression of the F508del-CFTR mutant channel, resulting in significant rescue of CFTR-dependent Cl− secretion in bronchial epithelial cell lines . These findings are consistent with the report that some F508del-CFTR is able to escape the proteolytic ERQC pathway and reach the PM . Furthermore, NHERF1 depletion was shown to enhance degradation of temperature-rescued F508del-CFTR from the PM , thus suggesting that NHERF-1 overexpression contributes to the rescue of mutant CFTR by retaining it on the PM, decreasing its susceptibility to degradation. These data strongly indicate the involvement of actin cytoskeleton remodelling and ezrin-mediated actin anchoring of F508del-CFTR to the PM. Notably, the reciprocal regulation of Rho GTPases and ERM proteins during actin remodelling plays a key role in the distribution and anchorage of macromolecular protein complexes to PM microdomains, which are essential for the maintenance of cell polarity [56, 59]. Consistently, NHERF1 overexpression was shown to stimulate the activation of endogenous RhoA and of RhoA-activated kinase (ROCK), thus leading to reorganization of the actin cytoskeleton. The latter occurs with concomitant phosphorylation of ezrin at Thr567 and stabilization of the multiprotein complex F508del-CFTR–NHERF-1–ezrin–actin at the apical PM, thus rescuing CFTR-dependent Cl− secretion . Remarkably, NHERF-1-induced formation of this multiprotein complex was also shown to enhance the epithelial architecture of F508del-CFTR cell monolayers by favouring CFTR-mediated tight-junction organization, suggesting a role for RhoA in restoring barrier function in CF epithelia . The role of adhesion-related proteins in CF pathophysiology was also demonstrated in recent studies showing a preferential association of cytokeratins (in particular, cytokeratin-8) with F508del-CFTR, and that disruption of this interaction contributes to the rescue of mutant protein .
More recently, we have gathered evidence that the effect of NHERF-1 overexpression on CFTR surface levels relies on the activation of both RhoA and Rac1 endogenous signalling . We found, however, that NHERF-1 relies on Rac1 signalling to promote ezrin-mediated PM anchoring of CFTR, whereas its stimulation of RhoA favours CFTR recycling to the PM via ROCK and possibly actin fibre-associated vesicle transport. Indeed, Rac1 acts by stimulating phosphatidylinositol 4-phosphate 5-kinase (PIP5K)-mediated production of phosphatidylinositol 4,5-bisphosphate [PtdIns(4,5)P2] at the PM. PtdIns(4,5)P2 binding activates ezrin by disrupting a head-to-tail inhibitory conformation that prevents its interaction with actin and ERM-binding proteins such as NHERF-1 . PtdIns(4,5)P2 binding is also required for, and precedes, Thr567 phosphorylation by kinases such as ROCK, which further stabilizes the open conformation of ezrin . Moreover, CFTR PM anchoring also requires Rac1-mediated de novo polymerization of F-actin, as interference with this process prevents CFTR surface anchoring . We further demonstrated that NHERF-1 coaxing of F508del-CFTR to the cell surface relies primarily on Rac1-mediated anchoring and retention. Treatment of CF airway cells with the Rac1 selective inhibitor NCS23766 completely prevents F508del-CFTR rescue by NHERF1 overexpression, and, in contrast to Rac1, expression of constitutively active RhoA produced no significant increase in F508del-CFTR PM levels . Data have also shown that activation of endogenous Rac1 signalling through treatment with hepatocyte growth factor (HGF) is sufficient to induce the cell surface anchoring and retention of F508del-CFTR. Most importantly, HGF treatment dramatically enhances the modest efficacy of the small-molecule CFTR corrector C4a and the investigational drug VX-809, restoring apical expression and function of the mutant channel to nearly 30% of those of wild-type CFTR in primary human bronchial epithelial cells . These findings reveal surface anchoring and retention as a major target pathway to be considered in CF pharmacotherapy, namely to achieve maximal restoration of F508del-CFTR in patients in combination with correctors and potentiators.
CFTR trafficking in proliferating and in differentiated cells
Besides the specific mechanisms controlling ERQC and membrane stability described above, CFTR trafficking is also controlled by the highly specific and tight regulation of the epithelial tissue. First, the protein is only expressed at the apical membranes of well-differentiated epithelial cells . In fact, intracellular expression of CFTR and exclusion from the PM have long been reported not only for trafficking mutants, such as F508del-CFTR, but also for wild-type CFTR in basal cells of non-CF tissue during remodelling or in nondifferentiated epithelium . The targeting of CFTR to the apical PM is thus directly and tightly linked to the process of epithelial differentiation/polarization .
More recent studies have re-examined the causal effects of airway damage and remodelling in the progression of CF lung pathology, through the use of wound-healing experiments. Interestingly, wound-healing airway cells from CF patients were found to repair more slowly than non-CF cells, and CFTR inhibition or silencing in non-CF primary airway cells significantly inhibited wound closure. However, these experiments were performed with cells grown on plastic, so whether this also occurs in well-differentiated and polarized airway epithelia remains to be demonstrated . Notwithstanding this, transduction of wild-type CFTR into CF airway cell lines or correction of CFBE-F508del and primary CF bronchial monolayers with VRT-325 significantly improved wound healing. Altogether, these observations demonstrate not only that functional CFTR plays a critical role in wound repair, but also that differentiation/polarization is somehow related to CFTR trafficking .
Among CFTR's vast interactome, quality control mechanisms play a key role in assessing its folding and conformation, sorting and processing its glycan moieties, promoting its traffic through the secretory pathway, regulating its delivery to the cell surface, and controlling its endocytosis and recycling. These quality control mechanisms discriminate between folded/functional and misfolded/dysfunctional CFTR, with the latter being sent for degradation. CFTR has been shown to reach the cell surface by nonconventional secretory pathways, and exciting recent data have related autophagy and epithelial differentiation to the regulation of CFTR trafficking. These complex pathways are starting to be explored and novel proteins are being identified as potential targets for treatment of the disease, to the benefit of CF patients.
Work in the authors' laboratory has been supported by strategic grant PEst-OE/BIA/UI4046/2011 (Bio-FIG) from FCT, Portugal; grants FCT/MCTES PTDC/SAU-GMG/122299/2010 (to M. D. Amaral) and FCT/MCTES PTDC/BIA-BCM/112635/2009 (to C. M. Farinha] from FCT, Portugal; grant Ref. 7207534 from the Cystic Fibrosis Foundation, USA (to M. D. Amaral); and the ERS Romain Pauwels Research Award (to C. M. Farinha). The authors are grateful to L. Clarke for critically reading the manuscript.