Therapeutic Options for Hydrating Airway Mucus in Cystic Fibrosis

Background: In cystic fibrosis (CF), genetic mutations in the CF transmembrane conductance regulator (CFTR) gene cause reduced chloride efflux from ciliated airway epithelial cells. This results in a reduction in periciliary liquid (PCL) depth of the airway surface liquid due to associated reduced water efflux. PCL layer dehydration reduces mucociliary clearance (MCC), leading to airway obstruction (reduced airflow and inflammation due to pathogen invasion) with mucus plug formation. Summary: Rehydrating mucus increases MCC. Mucus hydration can be achieved by direct hydration (administering osmotic agents to set up an osmotic gradient), using CFTR modulators to correct dysfunctional CFTR, or it can be achieved pharmacologically (targeting other ion channels on airway epithelial cells). Key Messages: The molecular mechanisms of several therapies are discussed in the context of pre-clinical and clinical trial studies. Currently, only the osmotic agent 7% hypertonic saline and the CFTR ‘potentiator' VX-770 (ivacaftor) are used clinically to hydrate mucus. Emerging therapies include the osmotic agent mannitol (Bronchitol), the intracellular Ca2+-raising agent Moli1901/lancovutide, the CFTR potentiator sildenafil [phosphodiesterase type 5 (PDE5) inhibitor] and the CFTR ‘corrector' VX-809 (lumacaftor). Other CFTR correctors (e.g. ‘chemical chaperones') are also showing pre-clinical promise.

In CF, there is a reduction in PCL depth linked to reduced water efflux due to reduced Clefflux caused by malfunctioning CFTR [9] . PCL layer dehydration reduces ciliary beat frequency, which slows the MCC rate [10] , allowing mucus plug formation, resulting in the following symptoms of the disease: reduced airflow and inflammation due to pathogen invasion [11] .
One therapeutic approach against mucus accumulation is mucus hydration, because this increases MCC. Mucus hydration can be achieved by three main methods: • Direct hydration • CFTR modulators • Pharmacologically -using drugs which target other ion channels on airway epithelial cells An extensive portfolio of drug compounds is being developed to aid the above approaches to airway mucus hydration ( table 2 ; fig. 3 ) and is described in detail below. Provision of a physical barrier to inhaled airborne irritants, particles and micro-organisms as well as aspirated food and liquid Trapping of irritants, particles, micro-organisms and food Formation of a transportable 'vehicle' upon which irritants are removed from the airways by mucociliary clearance Provision of a 'waterproof' layer to limit desiccation (e.g. of the epithelium)   Administering aerosolised hypertonic saline (HS) or dry powder mannitol (Bronchitol) into the airways sets up an osmotic gradient which draws water through epithelial cells' aquaporins, rehydrating the PCL [12,13] ( fig. 4 ).

Hypertonic Saline
HS is used for experimental sputum induction in respiratory diseases such as chronic obstructive pulmonary disease and asthma, but can also be used therapeutically, usually as a 7% solution [14] .
Therapeutic efficacy can be measured by the MCC rate, for example as the amount of sulphur colloid in expectorated sputum [15] . However, using the sulphur colloid content as a marker for clearance does not differentiate between mucus composition (e.g. hydration) and ciliary beating, or both in combination [16] . Consequently,    [14] (due to a possible negative feedback effect on epithelial Na + channel (ENaC) triggered by increased intracellular Na + concentration as a result of water efflux [17] ). The clinical significance of a 12% increase in MCC was not defined. HS changes the rheology of expectorated mucus in vitro (reducing viscoelasticity by disrupting ionic bonds) [18] , induces inflammatory mediator release such as prostaglandin E 2 [19] (which increases ciliary beat frequency) [20] and induces coughing [14] , which increases total MCC [17] .
Nevertheless, although the exact mechanism of action of HS is unknown, HS is an effective treatment. In clinical trials, the effect of HS was measured focusing on lung function, rate of pulmonary exacerbations and mucus clearance. HS significantly increased forced vital capacity compared to baseline values, and HS on its own produced a sustained increase ( ≥ 8 h) in mucus clearance [21] . Accompanying in vitro experiments (cultures of patients' bronchial epithelium) demonstrated changed ASL volume after HS application [21] . HS has long-term positive effects. The absolute lung function increase was significantly different between the control and the HS group, and the '48-week exacerbation-free survival' rate was 14% significantly higher in the HS group in comparison to the control group [22] . Recently, HS has shown short-term  122 efficacy and a lack of adverse effects in infants [23] . An ISIS trial with 321 <6-year-old participants found no significant difference in the rate of pulmonary exacerbations during the 48-week treatment period between HS and isotonic saline groups [24] . When adjusted for height, the 'lung clearance index' as an endpoint showed a significant difference between HS and isotonic saline groups in children aged <6 years, albeit with a small sample size (n = 25), after 48 weeks' treatment [25] .
In summary, HS is relatively inexpensive and well-tolerated by adults and infants. However, although there is a dose-dependent relationship between the HS concentration and MCC, most efficacious with 7% HS, clinical trials show mixed results for HS improving pulmonary function and reducing exacerbation rates [12] . However, there may be an issue with the endpoints against which the efficacy of CF drugs is currently measured [26] , because trial results do not translate into clinical outcomes. This issue is worsened by problems of an attempted com- parison of CF in infants with CF in adults, where baseline pulmonary function values for very young CF children reflect non-CF values. These endpoints make it difficult to find treatments with best efficacy when administered before CF symptoms are present.

Mannitol
Mannitol is a non-ionic osmotic agent [27] . Inhaled dry mannitol causes mucus hydration by the same mechanism as HS: increasing the osmotic gradient causing water efflux into the lung lumen [28] . The disadvantage of mannitol powder over HS is its relatively large size, so deposition at the intended site, the small airways and the lung periphery [29] , is hard to achieve [30] . Similar to HS, mannitol has rheologic effects on mucus; disruption of hydrogen bonds in the mucus reduces viscoelasticity, which increases mucus clearance, and induces cough [31] .
Phase 3 studies (in ≥ 6-year-old CF children) show improved FEV 1 and ∼ 35% reduction in pulmonary exacerbation occurrence [32] and improved lung function [33] after 26 weeks. It should be noted that the control group for these studies received 50 mg mannitol (a dose determined as subtherapeutic in a previous dose-escalation study [34] ). However, the 2012 study (n = 3,138) [33] questioned this 50-mg dose, where, although there was ∼ 8% improvement in FEV 1 from baseline in the mannitol group, statistically significant results were not obtained between the 400-mg mannitol and the control group, suggesting that the 50-mg dose may have had some efficacy. Another reason for statistically insignificant results may be that only one data point was collected for baseline FEV 1 .
In summary, mannitol's easier administration via inhaler (compared to HS nebuliser), similar adverse event rates, more consistent clinical efficacy and longer epithelial retention time may make mannitol the leading treatment in the future. Mannitol may cause bronchoconstriction in patients with hyperresponsive airways, but this is not proven [32] . Although a crossover trial in 6-to 17-year-olds [35] is currently underway, additional studies are needed to assess mannitol in patients <6 years old as well as its inflammation-inducing effects.

CFTR Modulators
Another way to increase mucus hydration in CF is to correct dysfunctional CFTR.

Mutation Classes
Different CFTR mutations can cause problems with synthesis, gating, conductance or stability of the channel ( fig. 5 ).

Class I: Reduced Synthesis of CFTR
These mutations synthesise unstable or truncated CFTRs due to prematurely coding a termination codon into the amino acid sequence. The most common class I mutation is G542X [6] and W1282X in the Ashkenazi Jewish population [36] . If a splicing error occurs instead of a nonsense or a frameshift mutation, some functional CFTRs are synthesised, but total CFTR expression is suboptimal.

Class II: Defective Processing of CFTR
The most common mutation (F508del) results in misfolded CFTR synthesis due to deletion of the phenylalanine codon at the 508th amino acid position [4] . Usually, dysfunctional CFTR is degraded before localisation to the apical membrane, but when F508del-CFTR is 'rescued' (trafficked and inserted into the membrane) its gating is defective, classifying it simultaneously as a class III mutation [37] .

Class III: Defective Gating of CFTR
Amino acid codons are substituted for other amino acids. While transcription and processing events are successful, these channels have defective regulation of ion transfer upon insertion into the membrane (e.g. G551D, glycine-to-aspartate missense mutation) [5] .
Class III mutations confer a conformational change to the NBD, negatively affecting ATP-binding capacity, failing to be stimulated by ATP [38] .

Class IV: Reduced Cl -Conductance of CFTR
CFTR is successfully inserted into the cell membrane but the channel's Clconductance is reduced by decreased open state frequency, linked to reduced phosphorylation capacity of cAMP-dependent protein kinases, due to membrane-spanning domain (MSD) mutations [4] .

Class V: Reduced CFTR Synthesis
Defective CFTR transcription forms both correctly and incorrectly spliced mRNA, resulting in suboptimal amounts of functional CFTRs [5] .

Class VI: Decreased Stability of CFTR at the Apical Membrane
CFTRs which are functional but have a high turnover at the cell surface are produced; they are degraded 5-6 times faster than normal due to deletion of the C-terminal tail which stabilises folded CFTR [39] .

Therapy
Correction of the underlying defect of CFTR mutations by gene replacement therapy is progressing slowly and has significant efficacy and safety problems [40] . Instead, mutation-class-specific drugs aim to correct the molecular mechanism of CFTR dysfunction to regain CFTR function [40] .

Cl -Cl -Cl -(Cl -)
Cl -Cl - Fig. 5. Summary of the effects on Clefflux of the different CFTR mutation classes. WT-CFTR are abundant at the apical surface and permit effective Clefflux into the airway lumen. Class I mutations produce truncated CFTR which are degraded before reaching the apical surface. Class II mutations produce misfolded CFTR which are degraded before reaching the apical surface. Class III muta-tions produce CFTR with defective activation (do not permit Clefflux). Class IV mutations produce CFTR with low ion channel conductance (suboptimal Clefflux). Class V mutations produce suboptimal amounts of functional CFTR. Class VI mutations produce unstable CFTR which are quickly degraded.
Color version available online 4.2.1 Class I Drugs which correct class I CFTR mutations are termed 'premature stop codon suppressors' or 'readthrough agents' [41] .
Aminoglycoside antibiotics disrupt translation by binding to rRNA, thereby inhibiting premature termination by inserting a different amino acid into the sequence [40] . G418 (Geneticin) increased expression of full-length CFTR to 25% of wild-type CFTR (WT-CFTR) levels in vitro in cells containing R553X-mutated CFTR [42] . Treatment with gentamicin or tobramycin (separately) in vivo in intestinal tissues of mice caused full-length hCFTR localisation to the apical surface [43] . However, longterm use of aminoglycoside antibiotics can cause nephroand ototoxicity [40] , and gentamicin and tobramycin are risk factors for acute renal failure in patients with renal/ hydration problems [44] .
PCT124 (ataluren) allows 'read-through' of stop codons, allowing transcription to continue, with synthesis of full-length CFTR [45] . Nasal potential difference (NPD) was the biomarker of Cltransport and drug activity/efficacy. PCT124 induced a ≥ 5 mV change in NPD in children [46] and adults [47] with few/mild adverse effects. PCT124 is undergoing a phase 3 extension study to observe long-term effects [48] .

Class II
Due to the close interaction between CFTR channel gating and cellular processing, and uncertainties of drug mechanisms, the distinction between CFTR correctors (targeted at cellular processing and trafficking defects [49] ) and CFTR potentiators (which correct Clchannel activity in mutant CFTRs at the cell surface [49] ) is ambiguous [50] .

CFTR Correctors
CFTR synthesis, co-translational folding and interaction with chaperone proteins occur in the endoplasmic reticulum (ER). If CFTR folding time is increased and CFTR is retained in the ER, molecular chaperones transport improperly folded CFTRs to the cytoplasm, where the 26S proteasome degrades them. This process is termed ER-associated degradation. WT-CFTR is transported from the ER to the Golgi where it undergoes N-glycosylation before transportation to the apical membrane [51,52] .
The mechanism underlying the primary therapeutic method against F508del was to reduce CFTR retention in the ER. In vitro, low temperature [53] , glycerol [54] and trimethylamine N-oxide [55] correct the processing defect of F508del-CFTR but had cytotoxic effects.
Organic solutes (myoinositol alone/with taurine and betaine) stabilised and promoted glycosylation of F508del-CFTR by increasing hydration of the peptide in IB3 (bronchial) and NIH 3T3 (fibroblast) cells [56] . However, merely releasing F508del-CFTR from the ER was not sufficient to treat CF; protein generation and maintenance by interaction with chaperone proteins are important. CF may be due to incorrect recognition of F508del-CFTR by the 'protein homeostasis network' [57] . However, mutated CFTR interaction with certain chaperone proteins (expressed during cell stress response [58] ) promotes degradation [44] .
The following chaperone proteins have been investigated: • Hsc70, which associates more strongly with F508del-CFTR than with WT-CFTR and promotes degradation [59] . • Hsp70, which enhances F508del-CFTR maturation [60] . • Hsp90, which provides ATPase activity for the proteasome [61] , integral to its function, thereby contributing to mutated CFTR degradation. Chemical and pharmacological chaperones have been investigated. Chemical chaperones are defined as nonspecific small molecules which stabilise cellular proteins by improving the rate or fidelity of protein folding, by acting on chaperone proteins or the proteasome rather than directly on CFTR [62] . Pharmacological chaperones act directly on mutated CFTR.

Chemical
Chaperones. Sodium 4-phenylbutyrate (Buphenyl) with genistein were used in a phase 1/2 clinical trial in 2001 [63] . A similar study in heterozygous F508del patients was conducted in 2007 [64] but was terminated in 2008; the data were not analyzed due to the identification of more clinically efficacious compounds. This may seem reasonable, because Buphenyl only showed statistically significant improvement in NPD (e.g. sweat Clconcentration was not significantly reduced) in its pilot study [65] , although it seemed promising, since in vitro it trafficked mutated CFTR to the apical membrane [66] and downregulated Hsc70 [67] .
Ruling the drug as inefficacious was disputed, suggesting that NPD is not the best measure of drug efficacy, and measurements were prone to error [84] . Ultimately, it was agreed that the effect size was correctly determined (accounting for the small sample size) and the results could not be classed as statistically significant [85] .
VX-809 (lumacaftor) binds to mutated CFTR's MSD1, inducing a protein-stabilising conformational change, protecting it from degradation [94] . VX-809's specificity depends on the MSD structure but was shown to rescue CFTR in F508del and other mutations [94] . VX-809 produced a significant change in NPD in a phase 2 clinical trial [95] and is investigated with concomitant VX-770 (CFTR potentiator) and other CF drugs which prevent infection [96] .

CFTR Potentiators
CFTR potentiators are effective for (class III/IV) mutations where the CFTR has been trafficked to the apical membrane but has a defective ion channel function.
VRT-532 corrects the gating defect of several CFTR mutants, with the highest affinity for F508del-CFTR [89] . It increases ATP-binding affinity of mutated CFTR after protein kinase A activation [97] . VRT-532 corrects G551D-CFTR by increasing ATP binding and hydrolysis, with no effect on WT-CFTR where ATPase activity is already maximal [98] .
The CFTR corrector Buphenyl has been investigated in conjunction with genistein because genistein alone was shown to be a CFTR potentiator [104] . Genistein acts at the NBD, showing 5-fold higher affinity for F508del-CFTR than G551D-CFTR [89] , increasing ATP binding capacity and so increasing channel open probability [105] . Capsaicin (pungent essence of hot chillies) was shown to have a similar mechanism [106] . Genistein in conjunction with curcumin demonstrated a significant synergistic effect [107] . It also increases ciliary beat frequency [108] .
The inhaled PDE5 inhibitor, sildefanil, inhibits cGMP breakdown in vivo in mice [112] . The relevance of cGMP to cAMP was addressed: there is 'crosstalk' between signalling pathways, with increased cGMP inhibiting PDE3 (PDE3 degrades cAMP) [113] , suggesting sildefanil increases intracellular cAMP. Sildefanil's potent analogue, KM11060, is effective, but its safety in vivo is unknown [114] . Sildenafil, vardenafil and tadalafil each corrected Clconductance in mice orally [115] and when inhaled [112] . Sildenafil had both exocytotic effects on F508del-CFTR and ion channel potentiating function, making it both a CFTR corrector and potentiator. However, sildenafil's efficacy is only achieved at high doses [116] , but a current phase 2 clinical trial [117] , following a phase 1/2 safety trial in 2008 [118] , suggests a potentially positive clinical future.

Direct ENaC Inhibitors
These compounds block open ENaC directly.

Amiloride and Amiloride Derivatives
Amiloride and early amiloride derivatives (phenamil and benzamil) had limited therapeutic usefulness due to rapid absorption/clearance from the lungs [124] . A thirdgeneration amiloride derivative, 552-02, which in vitro in human bronchial cells was 60-fold more potent and 2 times less reversible than amiloride, showed a significant positive synergistic effect with HS (unlike amiloride-HS treatment, see section on Hypertonic Saline above) and in vivo in sheep increased MCC for >5 h [125] . Since a 2006 phase 1/2 clinical trial [126] , development has not advanced, except for a 2009 trial in xerostomia [127] .
Another amiloride analogue, GS-9411, had greater solubility in HS, was 100-fold more potent in vitro , maintained 75% more ENaC block following apical surface washes than amiloride, and had less permeability and increased metabolism in airway epithelium than amiloride, suggesting a better safety profile [128] . However, in phase 1 ascending-dose trials in healthy people [129,130] , GS-9411 also blocked renal ENaCs, causing clinically significant hyperkalaemia [131] . Hence, the phase 1 trial in CF patients was withdrawn in 2009 [132] . Pyrazinoyl quaternary amines are structurally different from pyrazinoyl guanidine amiloride. Some show efficacy in vitro in human bronchial cells and in vivo in guinea pigs, having decreased cellular permeability and increased solubility (significant for administration methods, i.e. dry powder) [133] . 'Compound 12g' (α-branched quaternary amine) blocked ENaC and produced a guinea-pig tracheal potential difference 16 times greater than that induced by amiloride [134] .
Camostat is a low-molecular-weight CAP inhibitor which inhibits prostasin in vitro [135] . By inhibiting prostasin, camostat inhibits ENaC for 6 h after washing in vitro in human bronchial cells and increases MCC in vivo in sheep for 5 h after dosing [136] . In a phase 2 clinical trial in CF patients (n = 9) [137] , camostat (nasal spray) reduced Na + efflux, causing a change in NPD of +13.1 mV from baseline, but 3 patients experienced serious adverse effects [138] . Concerns exist about camostat's specificity for certain CAPs and there is uncertainty regarding tolerance/safety with administration to lower airways by nebuliser [138] .

Small Peptide Inhibitors
The small peptide RFSHRIPLLIF inhibits endogenous ENaC in vitro in human airway epithelia by blocking an ENaC γ-subunit domain [139] .
In a second phase 3 trial [146] (TIGER-2; n = 466), denufosol caused a mean change in FEV 1 from baseline to 0.040 litre (at week 48; placebo was 0.032 litre): lung function change by denufosol, relative to placebo, was now just 8 ml. The statistically insignificant difference may be attributed to less patient adherence in TIGER-2 than TI-GER-1 or to the fact that denufosol fails to prevent longterm lung function decline (TIGER-2 lasted twice as long as TIGER-1) [149] .
A phase 2 trial of denufosol [150] was initiated in 2010 in 2-to 4-year-old CF patients based on the concept that drugs with 'disease-modifying capability' be trialled in young children with low disease severity, where small short-term effects should be expected, but which have possible long-term efficacy [151] .
The bacterial polycyclic peptide duramycin (Moli1901/ lancovutide) activates CaCC in airway epithelium by causing Ca 2+ influx via Ca 2+ -permeable pores and releases Ca 2+ from intracellular stores [156] . However, in vitro in human colonic epithelial cells, duramycin disrupted cell membrane integrity by forming non-specific ion channels [157] . In a phase 2 trial (n = 24), nebulised duramycin was well-tolerated, had a long half-life, and did not enter the systemic circulation. FEV 1 was increased (on day 26) by 8% from baseline, with the highest concentration of duramycin administered (2.5 mg/day) [158] . Another phase 2 (estimated n = 160) dosage-finding study was conducted in 2008 [159] , the increased sample size showing promise.

Conclusions
Properly hydrated mucus is a critical component of airway homeostasis. Failure to maintain adequate mucus hydration results in reduced MCC, which leads to mucus accumulation in the airways, causing impaired gas exchange and inflammation due to pathogen and proteolytic enzyme accumulation [11] .
A 2013 meta-analysis of randomised clinical trials of CFTR gene replacement therapy concluded that transferring the correct CFTR gene via topical administration fails to produce clinical benefit [160] . Other ways to correct mutated-CFTR-induced abnormal ion concentration include correction of the processing of mutated CFTR using CFTR modulators, directly applying osmotic agents onto the ASL, or targeting other ion channels on the ciliated airway epithelial cell surface.
Of the therapies discussed herein, only 7% HS and CFTR potentiator VX-770 (ivacaftor) are used clinically to hydrate mucus. Ivacaftor is currently only used with G551D-CFTR patients (clinical trials are underway testing its efficacy with other mutations). Therapies which may also succeed clinically are the osmolytic agent, mannitol, which shows more consistent efficacy than HS in clinical trials, the intracellular Ca 2+ -raising agent Moli1901/lancovutide, the PDE5 inhibitor sildefanil and the pharmacological chaperone VX-809 (lumacaftor), which is in clinical trials in combination with other CF drugs. Also, research into some chemical chaperones is relatively new, with SAHA, Aha1 inhibitors and matrine showing positive results in vitro. Nevertheless, HS surprisingly remains the best treatment today, despite the CFTR gene mutation being first described in 1989 [3] .