Fixing cystic fibrosis by correcting CFTR domain assembly

CF mutations have been grouped into six categories based on their cellular/molecular pathogenesis (Zielenski, 2000). Class I mutations include nonsense mutations (G542X and R553X), generating premature termination codons and frame-shift mutations that lead to truncated and/or and nonfunctional protein (Fig. 1). Class V mutations cause mRNA mis-splicing or interfere with the promoter activity. Both classes impair CFTR protein production and plasma membrane expression, causing a severe CF phenotype.

Class II mutations, despite normal transcript levels, have little or no detectable CFTR at the plasma membrane as a consequence of misfolding of the newly translated polypeptide. This category includes the most common mutation, deletion of phenylalanine 508 (ΔF508) in the NBD1, identifiable in one or both alleles in ∼90% of CF patients (Riordan, 2008). ΔF508-CFTR is largely retained in the ER and degraded by the ubiquitin–proteasome system (Fig. 1; Cheng et al., 1990; Ward et al., 1995).

Class III (e.g., G551D, ∼4%) and class IV (e.g., R117H) mutations impair the CFTR channel opening-closing (or gating) cycle and conductance, respectively, without recognizable conformational or trafficking defects. Class III mutations are primarily associated with NBD1-2, whereas class IV mutations are localized to the channel pore (Riordan, 2008). Class VI mutations reduce CFTR expression by facilitating the channel removal from the plasma membrane. Notably, some mutations have a mixed phenotype. For example the ΔF508 mutation causes folding, and gating, as well as plasma membrane stability impairments (Dalemans et al., 1991; Denning et al., 1992; Lukacs et al., 1993).

Because CF is a monogenic disease, it is postulated that the clinical phenotype would be alleviated by correcting the basic defects caused by various mutations impeding or preventing CFTR function, expression, or both (Cai et al., 2011). Efforts to correct the basic defects of CFTR biogenesis and function have been primarily focused on the most prevalent mutations: ΔF508, G551D, and premature termination codons (Fig. 1).

Aminoglycosides such as gentamycin interact with eukaryotic rRNA within the ribosomal subunits and reduce the fidelity of translation by interrupting the normal proofreading function (Burke and Mogg, 1985). Consequently, aminoglycosides allow insertion of a near-cognate amino acid at a premature termination codon and the translation of the entire coding region. Aminoglycosides have been used to suppress premature termination codons, resulting in the synthesis of full-length CFTR in CF patients with class I mutations (Wilschanski et al., 2003). Ataluren (PTC124), an orally bioavailable drug with diminished toxicity, was developed by a cell-based high-throughput screening assay (Welch et al., 2007). Although Ataluren selectively suppresses the premature termination codon in a mouse model (Welch et al., 2007), it showed variable efficiency among patients with different genotypes (Rowe et al., 2007). Orally administered Ataluren has been reported to rescue the activity of CFTR with premature termination codons in phase II trials (Kerem et al., 2008; Sermet-Gaudelus et al., 2010). A phase III clinical study is currently underway to evaluate long-term efficacy and safety (http://clinicaltrials.gov/ct2/show/NCT00803205).

In principle, the CFTR folding defect could be counteracted by pharmacological chaperones (PCs), similar to other misfolding diseases, where a variety of ligands or substrates can stabilize the target protein functional conformation (Bernier et al., 2004). Although this approach would be highly specific, and maintain the endogenous regulation and expression pattern of CFTR, high affinity CFTR ligands are not available. Altering the cellular folding environment could also be exploited to overcome the mutant misfolding/misprocessing and has shown some success in preclinical settings (Balch et al., 2011). Chemical chaperones similar to reduced temperature can also counteract the ΔF508-CFTR misfolding and elicit modest accumulation of partially functional but unstable channels at the plasma membrane (Denning et al., 1992; Sato et al., 1996; Sharma et al., 2001).

The revelation that ΔF508-CFTR misfolding can be rescued prompted the development of a cell-based high-throughput screening assay using the yellow fluorescent protein–based halide indicator, which monitors the cAMP-activated plasma membrane chloride permeability, including CFTR activity, by sensing changes in the cytoplasmic halide concentration (Galietta et al., 2001). This assay became instrumental in the identification of not only “correctors” that improve ΔF508-CFTR plasma membrane expression, but also “potentiators” to activate plasma membrane–resident CFTR channels. Interestingly, the assay also identified inhibitors of chloride channels, which may be potential therapeutics in secretory diarrheas (Verkman and Galietta, 2009).

Screening of diverse chemical libraries has produced several classes of small-molecule ΔF508-CFTR correctors, including corr-4a and its analogues (Pedemonte et al., 2005a). Additional correctors were also obtained by high-throughput screening and computational methods (Kalid et al., 2010; Robert et al., 2010; Sampson et al., 2011). However, the efficacy of these correctors in restoring chloride conductance was limited and reached only <10% of normal human primary epithelia, which is significantly lower than the predicted requirement for therapeutic efficiency (Pedemonte et al., 2005a, 2010; Van Goor et al., 2006). A similar approach by Vertex Pharmaceuticals, Inc. with Cystic Fibrosis Foundation Therapeutics’s (CFFT) support has yielded new classes of correctors, including VX-809, the most promising compound being evaluated in phase III clinical trials (Van Goor et al., 2006, 2011). VX-809 restores ∼15% CFTR channel activity in primary respiratory epithelia expressing ΔF508-CFTR, but appears to have marginal clinical benefits (Van Goor et al., 2011; Clancy et al., 2012).

Cell-based functional high-throughput screening assays have also isolated several potentiators that improve the channel function of class II and III mutants (Pedemonte et al., 2005b; Van Goor et al., 2006). The most promising potentiator, VX-770 (Ivacaftor), isolated by Vertex Pharmaceuticals, Inc., restores G551D-CFTR activity to ∼50% of wild-type level (Van Goor et al., 2009). Clinical studies confirmed short-term safety and clinical benefits, including 55% reduced pulmonary exacerbation frequency and 10% increased lung function (Ramsey et al., 2011). Ivacaftor is the first FDA-approved drug for treatment of G551D-CFTR patients (∼4% of CF population), representing a landmark translational achievement, exploiting the basic biology of CFTR and years of research and development in both academia and industry. Remarkably, Ivacaftor also restores the gating defect of several other class II mutations; therefore, it may benefit ∼10% of CF patients (Yu et al., 2012).

Combination therapy of potentiators and correctors could be useful for improving ΔF508-CFTR function given the persisting gating defect of rescued ΔF508-CFTR at the plasma membrane. Indeed, interim results of a phase II clinical trial suggest that ΔF508 CF patients treated with a combination of VX-809 and Ivacaftor seem to display better lung function than those treated with either drug alone (http://clinicaltrials.gov/ct2/show/NCT01225211).

Despite recent advances of CFTR research, further improvement in functional expression of ΔF508-CFTR, the most common mutation in CF patients, is necessary because correction of the CF phenotype likely requires restoring ∼35% of wild-type CFTR plasma membrane activity. Elucidating CFTR folding/misfolding and the available corrector mechanisms should help to achieve this goal. Here, we focus on efforts to understand and correct the folding defects of the ΔF508-CFTR.

Compelling evidence supports the coupled domain-folding model of CFTR. Accordingly, individual domains can fold cotranslationally to metastable states; but attaining the CFTR native fold requires post-translational domain assembly and inter-domain interactions that are critical to proper folding (Fig. 2 A; Du et al., 2005; He et al., 2008, 2010; Du and Lukacs, 2009). The slow post-translational conformational maturation is assisted by chaperones (Rosser et al., 2008) and reflected by the delayed formation of NBD–MSD interfaces in the mature wild-type CFTR (He et al., 2008; Serohijos et al., 2008). The energetic instability of individual domains and the slow domain assembly with the fast ER-associated degradation kinetics of folding intermediates all contribute to the inefficient folding (∼30%) of wild-type CFTR (Fig. 2 B) and are further sensitized by point mutations in CF (Rabeh et al., 2012).

How does the ΔF508 mutation affect the channel? Homology modeling and cysteine cross-linking experiments have revealed the unique three-dimensional architecture of CFTR and the possible role of F508 (He et al., 2008; Mornon et al., 2008; Serohijos et al., 2008). In native CFTR, the F508 residue and surrounding area in the NBD1 forms an interface with the coupling helix of cytoplasmic loops 4 (CL4) and 1 (CL1) in MSD2 and MSD1, respectively, which creates a hydrophobic patch. NBD2 associates with CL2 and CL3 of MSD1 and MSD2, respectively. These interfaces relay conformational changes of the NBDs to the MSDs during channel gating, and are essential for CFTR folding (Wang et al., 2007; He et al., 2008, 2010; Mornon et al., 2008; Serohijos et al., 2008; Loo et al., 2010; Thibodeau et al., 2010; Grove et al., 2011). Destabilization of the interface by missense mutations in the CLs or by mutagenesis of the F508 side chain disrupts folding (Du et al., 2005; Mornon et al., 2008; He et al., 2010; Loo et al., 2010; Thibodeau et al., 2010). These observations are in support of the emerging model of co- and post-translational conformational maturation of CFTR that involves energetic and/or kinetic domain stabilization during coupled-domain folding (Du and Lukacs, 2009), similar to that of certain soluble multi-domain proteins and the BtuCD transporter (Fig. 2 A; Han et al., 2007; DiBartolo and Booth, 2011).

Recent studies revealed that the ΔF508-NBD1 is thermodynamically and kinetically destabilized at physiological temperature and suggested that the NBD1 stabilization would effectively counteract ΔF508-CFTR misprocessing (Protasevich et al., 2010; Wang et al., 2010). Surprisingly, this was not the case. Even substantial conformational stabilization of ΔF508-NBD1 by second site mutations led to modest rescue (<20%) of ΔF508-CFTR processing, plasma membrane expression, and function, and failed to reinstate coupled domain folding (Fig. 2 B; Mendoza et al., 2012; Rabeh et al., 2012). Likewise, reversing the NBD1–MSD2 interface instability by second site mutations (e.g., R1070W) only marginally rescued the ΔF508-CFTR phenotype (Fig. 2 B). Remarkably, simultaneous genetic stabilization of NBD1 energetics and the NBD1–MSD2 interface led to robust, synergistic rescue (65–80%) of ΔF508-CFTR folding and function (Fig. 2 B; Rabeh et al., 2012). A similar conclusion was reached by the analysis of evolved sequences coupled to the F508 residue (Mendoza et al., 2012). These unexpected findings suggest that correction of two primary structural defects is necessary and sufficient to restore CFTR function in most CF patients.

The discovery of two primary folding defects in ΔF508-CFTR highlighted three pharmacological implications (Fig. 2 C): (1) the plural folding defects provide a reasonable explanation for the modest efficacy of single correctors that may target only one of them with a presently unknown mechanism(s) (Sampson et al., 2011; Van Goor et al., 2011); (2) second site mutations counteracting one of the primary folding defects could promote mechanistic classification of existing corrector molecules, as well as the identification of new ones by second generation of structural defect–targeted high-throughput screening assays; (3) correction of both NBD1 energetic and interface instability is likely required to robustly normalize ΔF508-CFTR processing, expression, and function (Fig. 2 C). Whether this could be achieved by one or two small molecules alone, or in combination with proteostasis network regulators that indirectly modulate ΔF508-CFTR folding, trafficking, and function (Balch et al., 2011) awaits further experimentation.

In this paper we have described the cellular consequences of CFTR mutations and recent efforts to understand the folding defects underlying the ΔF508-CFTR mutation in order to improve channel folding, stability, and function. However, many critical issues remain. Our understanding of how the loss of channel function results in CF, particularly the lung symptoms, is incomplete (Ashlock and Olson, 2011) and compounded by the fact that mouse models fail to recapitulate the CF lung disease, though the development of transgenic pig and ferret might address this issue (Rogers et al., 2008; Sun et al., 2010). Mechanistic studies of the channel are hampered by the low copy number and instability of mutants and the difficulties to monitor their structural alteration, protein–protein interaction, and trafficking at high spatiotemporal resolution in the appropriate cellular environment (Riordan, 2008; Balch et al., 2011). In addition to the drug discovery programs described in this paper, there are ongoing efforts to replace the mutant by gene therapy and/or activating alternative chloride secretion. However, multiple cellular mechanisms impede the nuclear delivery of CFTR transgene packaged either into cationic lipid complexes or viruses, though overcoming these processes may allow genotype-independent therapy (Griesenbach and Alton, 2009). Identification of the epithelial isoform of the Ca²⁺-activated chloride channel (TMEM16A) has opened the possibility to pharmacological activation of an alternative chloride secretory pathway (Ferrera et al., 2010).

The ultimate success of translational research most often relies on our detailed understanding of the basic biological problem at hand. We hope that this short perspective will help inspire further biological research, a prerequisite for translational successes in curing basic defects in CF and other genetic diseases such as diabetes insipidus and familial hypercholesterolemia.

We apologize to those authors whose contribution could not be cited in this manuscript owing to space limitation.

We are thankful for the generous support of NIH-NIDDK, Cystic Fibrosis Foundation Therapeutics Inc., Cystic Fibrosis Canada, Canadian Institute of Health Research, and the Canadian Foundation for Innovation. G. Lukacs is holder of a Canada Research Chair. Illustrations were provided by Neil Smith, www.neilsmithillustration.co.uk.