Cystic fibrosis (CF) is a fatal genetic disease caused by mutations in cftr, a gene encoding a PKA-regulated Cl− channel. The most common mutation results in a deletion of phenylalanine at position 508 (ΔF508-CFTR) that impairs protein folding, trafficking, and channel gating in epithelial cells. In the airway, these defects alter salt and fluid transport, leading to chronic infection, inflammation, and loss of lung function. There are no drugs that specifically target mutant CFTR, and optimal treatment of CF may require repair of both the folding and gating defects. Here, we describe two classes of novel, potent small molecules identified from screening compound libraries that restore the function of ΔF508-CFTR in both recombinant cells and cultures of human bronchial epithelia isolated from CF patients. The first class partially corrects the trafficking defect by facilitating exit from the endoplasmic reticulum and restores ΔF508-CFTR-mediated Cl− transport to more than 10% of that observed in non-CF human bronchial epithelial cultures, a level expected to result in a clinical benefit in CF patients. The second class of compounds potentiates cAMP-mediated gating of ΔF508-CFTR and achieves single-channel activity similar to wild-type CFTR. The CFTR-activating effects of the two mechanisms are additive and support the rationale of a drug discovery strategy based on rescue of the basic genetic defect responsible for CF.
- high-throughput screening
- human bronchial epithelia
- Cl− channels
cystic fibrosis (CF) is caused by mutations in the gene encoding the CF transmembrane conductance regulator (CFTR), a PKA-regulated Cl− channel (41, 43). Analysis of the ∼1,000 known CFTR mutations indicates that impaired CFTR function in airway epithelia is highly correlated with severity of lung disease in CF patients and supports the hypothesis that restoration of mutant CFTR activity to between 5 and 30% of wild-type (wt)-CFTR would lead to improved lung function (9, 34–36). Approximately 90% of CF patients in North America have an in-frame deletion resulting in a loss of phenylalanine at position 508 (ΔF508-CFTR) with ∼50% homozygous for the ΔF508-CFTR mutation (7). This mutation interferes with CFTR folding, trafficking, membrane stability, and channel gating (4, 8, 27, 28, 59), resulting in greatly reduced CFTR density and activity in the apical membrane and impaired epithelial cell function (2, 33, 51, 52). Because modest repair of this mutation would be expected to benefit the majority of CF patients, a number of attempts have been made to identify pharmacological agents that rescue folding and/or gating defects in ΔF508-CFTR as candidates for CF therapy (46, 57).
Pharmacological agents that increase PKA-regulated Cl− channel gating of CFTR are called potentiators and include the isoflavone genistein (24). The efficacy of genistein in vivo, however, may ultimately be limited because of its low potency, rapid in vivo metabolism, and lack of selectivity (3, 24, 31). With the use of high-throughput screening (HTS), several different structural classes of potentiators have been identified (23, 29, 32, 62), including phenylglycine (38), sulfonamides (38), and tetrahydrobenzothiophenes (62). In contrast to potentiators, there are relatively few correctors, small molecules that rescue the folding and/or trafficking of ΔF508-CFTR to increase the cell surface density of ΔF508-CFTR. The most studied corrector to date is 4-phenylbutyrate (4-PBA), which produces a small increase in ΔF508-CFTR cell surface density in vitro (44) but has shown poor efficacy in clinical trials (45). HTS efforts identified 1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid diamides as potent (∼10 nM) modulators of CFTR activity in expression systems (22); however, these compounds have not been shown to be active in primary epithelia. Recently, Egan et al. (14) proposed that curcumin was a corrector of ΔF508-CFTR; however, the activities of curcumin in vitro and in vivo are inconsistent (53). HTS approaches by Verkman and colleagues (37) have recently identified other structural classes of small-molecule correctors of ΔF508-CFTR trafficking with potency in the micromolar range. Indeed, there is a continued need for potent and efficacious correction agents that act either alone or in combination with potentiators of channel gating to restore ion channel function of ΔF508-CFTR. Here, we describe efficacious small molecules that rescue either the defective trafficking or gating of ΔF508-CFTR in both recombinant cells and human bronchial epithelia (HBE) isolated from ΔF508-homozygous patients.
MATERIALS AND METHODS
NIH/3T3 mouse fibroblasts stably expressing wt-CFTR (wt-NIH/3T3) or ΔF508-CFTR (ΔF508-NIH/3T3) were cultured as previously described (11). HEK-293 cells expressing ΔF508-CFTR or wt-CFTR were grown as previously described (51). Assay medium used for NIH/3T3 and HEK-293 experiments was HyQ CCM5 (HyClone, Logan, UT) with 1% heat-inactivated FBS. Non-CF and CF airway epithelia were isolated from bronchial tissue, cultured as previously described (19), and plated onto Costar Snapwell filters that were precoated with NIH/3T3-conditioned medium. After 4 days, the apical medium was removed, and the cells were grown at an air-liquid interface for >14 days before use. This resulted in a monolayer of fully differentiated columnar cells that were ciliated, features that are characteristic of airway epithelia. Non-CF HBE were isolated from nonsmokers who had no known lung disease. CF-HBE were isolated from patients homozygous for ΔF508-CFTR.
NIH/3T3 cells expressing ΔF508-CFTR were used to develop a HTS assay to identify modulators of ΔF508-CFTR. To monitor compound-induced changes in anion flux through CFTR, we used voltage-sensitive assays based on the change in fluorescence resonance energy transfer between a membrane-soluble voltage-sensitive dye, bis-(1,2-dibutylbarbituic acid)trimethine oxonol [DiSBAC2(3)] and a plasma membrane-localized fluorescent, coumarin-linked phospholipid (CC2-DMPE; Invitrogen, Madison, WI) (20). We excited the coumarin-lipid donor at 405 nm using a 300-W xenon arc lamp directed toward the cells with a 425-nm dichroic; the emitted light was passed through 460-nm and 580-nm filters, which we collected at 1 Hz using photomultiplier tubes. The data were normalized to a starting ratio of 1 and reported as RF/RI (where RF is final response and RI is initial response).
For the corrector HTS assay, ΔF508-CFTR-NIH/3T3 cells were preincubated with compound to allow for de novo synthesis and processing and trafficking to the membrane. After the preincubation, the compound was removed, and membrane potential changes in response to ΔF508-CFTR activation by genistein and forskolin were monitored with a cell-based fluorescence assay of membrane potential (Fig. 1A). ΔF508-CFTR-NIH/3T3 cells were seeded at a density of 10,000/well into poly-l-lysine-coated 384-well microtiter plates and incubated in assay medium (HyQ CCM5 medium with 1% FBS) with test compound (10 μM in 0.1% DMSO) or 0.1% DMSO (control) for 16 h at 37°C and 5% CO2. Because low-temperature incubation is known to correct the trafficking of ΔF508-CFTR (11), we used temperature correction to validate the assay and then as a positive control for correction. Microtiter plates containing ΔF508-CFTR-NIH/3T3 cells were incubated for 16 h at 37°C with or without compound or at 27°C and 5% CO2. Before activity was monitored, the compound was washed out with three changes of bath medium containing (in mM) 160 NaCl, 4.5 KCl, 2.0 CaCl2, 1 MgCl2, 10 d-glucose, and 10 HEPES (ph 7.4 with NaOH). The cells were then incubated with 20 μM CC2-DMPE for 30 min, rinsed three times, and then incubated with 3 μM DiSBAC2 (3) for 20 min with bath medium in the dark at room temperature. After the fluorescence resonance energy transfer-based membrane potential probes were loaded, the microtiter plates were placed into an integrated liquid handler and fluorescent detector (VIPR-II) to monitor changes in fluorescent intensity in response to membrane potential changes. After baseline readings were obtained, Cl−-free medium and genistein were added to the bath to promote Cl− efflux and potentiate channel gating, respectively. Forskolin (10 μM) was then added to activate ΔF508-CFTR; the maximum increase in absolute ratio of the RF and RI was monitored and expressed as RF/RI. As expected, the response in temperature-corrected cells was significantly higher than that in uncorrected cells (Fig. 1B). No response was observed in temperature-corrected NIH/3T3 cells not expressing CFTR. The statistical robustness of the assay was tested by running six plates on 3 days, with one-half of the plates incubated for 16 h at 27°C and the other one-half at 37°C. From data shown in Fig. 1C, the 1-z′ screening window parameter (63) was 0.64, indicating that this assay is appropriate for HTS.
The activity of compounds was calculated by normalizing to the temperature-corrected response and the uncorrected vehicle (DMSO-treated) controls using Eq. 1 (1) where X is RF/RI in the presence of compound and where 27°C and DMSO is RF/RI for the positive and negative controls, respectively.
For the potentiator HTS assay, ΔF508-CFTR-NIH/3T3 cells in assay medium were seeded at 30,000 cells/well into 384-well microtiter plates and incubated at 27°C for 16 h. The cells were subsequently washed, loaded with the fluorescent membrane potential probes, and placed into the VIPR-II as described above. After baseline readings were obtained, Cl−-free medium and compound were added 15 s before 10 μM forskolin addition. The maximum increase in RF/RI was measured, and the data were normalized to the positive (genistein) and negative (DMSO-treated) controls using Eq. 2 (2) where X is RF/RI in the presence of compound and where genistein and DMSO is RF/RI for the positive and negative controls, respectively. In this assay, genistein potentiated the forskolin-induced membrane depolarization by 73 ± 7% above the untreated control with an EC50 of 19.2 ± 1.9 μM (n = 35).
The level of cAMP in NIH/3T3 cells after forskolin or test compound application was determined with an adenylyl cyclase activation FlashPlate assay, according to the manufacturer’s directions (Perkin-Elmer Life Sciences, Boston, MA). Briefly, NIH/3T3 cells were incubated for 30 min with forskolin or test compounds in the presence or absence of IBMX. The cells were then lysed and transferred along with the medium to a 96-well FlashPlate precoated with anti-cAMP antibody. A fixed concentration of radiolabeled cAMP was added to the plates and incubated overnight at 22°C. The FlashPlates were read in a microplate scintillation counter, and the cAMP concentrations were determined with a cAMP standard curve that was present in each plate.
Ussing chamber recordings.
Fischer rat thyroid (FRT) epithelia expressing ΔF508-CFTR or G551D-CFTR were grown on Costar Snapwell cell culture inserts for 4 days. The epithelia were then mounted in an Ussing chamber (Physiologic Instruments, San Diego, CA), and the monolayers were continuously short circuited with a voltage-clamp system (Department of Bioengineering, University of Iowa). Transepithelial resistance was measured by applying a 2-mV pulse. For measurement of short-circuit current (Isc), the basolateral bath solution contained (in mM) 135 NaCl, 1.2 CaCl2, 1.2 MgCl2, 2.4 K2HPO4, 0.6 KHPO4, 10 HEPES, and 10 dextrose (titrated to pH 7.4 with NaOH). To set up a basolateral-to-apical Cl− concentration gradient, the basolateral membrane was permeabilized with nystatin (360 μg/ml). All experiments were performed 30 min after nystatin permeabilization. The apical NaCl was replaced by equimolar sodium gluconate (titrated to pH 7.4 with NaOH) to give a large Cl− concentration gradient across the epithelium. The solutions were maintained at 37°C and bubbled with air. The electrode offset potential and fluid resistance were corrected using a cell-free insert. Under these conditions, the current reflects the flow of Cl− through CFTR expressed in the apical membrane. The Isc was digitally acquired using an MP100A-CE interface and AcqKnowledge software (version 3.2.6; BIOPAC Systems, Santa Barbara, CA). Forskolin (10 μM) and all test compounds were added to both sides of the cell culture inserts.
HBE grown on Costar Snapwell cell culture inserts were mounted in an Ussing chamber (Physiologic Instruments), and the transepithelial resistance and Isc in the presence of a basolateral-to-apical Cl− gradient were measured with a voltage-clamp system (Department of Bioengineering, University of Iowa). Briefly, HBE were examined under voltage-clamp recording conditions (holding voltage = 0 mV) at 37°C. The basolateral solution contained (in mM) 145 NaCl, 0.83 K2HPO4, 3.3 KH2PO4, 1.2 MgCl2, 1.2 CaCl2, 10 glucose, and 10 HEPES (pH adjusted to 7.35 with NaOH), and the apical solution contained (in mM) 145 sodium gluconate, 1.2 MgCl2, 1.2 CaCl2, 10 glucose, and 10 HEPES (pH adjusted to 7.35 with NaOH).
Total Cl− current in ΔF508-NIH/3T3 cells was monitored with the perforated-patch recording configuration as previously described (42). Voltage-clamp recordings were performed at 22°C using an Axopatch 200B patch-clamp amplifier (Axon Instruments, Foster City, CA). The pipette solution contained (in mM) 150 N-methyl-d-glucamine (NMDG)-Cl, 2 MgCl2, 2 CaCl2, 10 EGTA, 10 HEPES, and 240 μg/ml amphotericin B (pH adjusted to 7.35 with HCl). The extracellular medium contained (in mM) 150 NMDG-Cl, 2 MgCl2, 2 CaCl2, and 10 HEPES (pH adjusted to 7.35 with HCl). Pulse generation, data acquisition, and analysis were performed with a personal computer equipped with a Digidata 1320 A/D interface in conjunction with Clampex 8 (Axon Instruments). To activate ΔF508-CFTR, 10 μM forskolin and 20 μM genistein were added to the bath, and the current-voltage relation was monitored every 30 s.
Gating activity of wt-CFTR and temperature-corrected ΔF508-CFTR expressed in NIH/3T3 cells was observed with excised inside-out membrane patch recordings as previously described (8) using an Axopatch 200B patch-clamp amplifier (Axon Instruments). The pipette contained (in mM) 150 NMDG, 150 aspartic acid, 5 CaCl2, 2 MgCl2, and 10 HEPES (pH adjusted to 7.35 with Tris base). The bath contained (in mM) 150 NMDG-Cl, 2 MgCl2, 5 EGTA, 10 TES, and 14 Tris base (pH adjusted to 7.35 with HCl). After excision, both wt-CFTR and ΔF508-CFTR were activated by adding 1 mM MgATP, 75 nM of the catalytic subunit of PKA (Promega, Madison, WI), and 10 mM NaF to inhibit protein phosphatases, which prevented current rundown. The pipette potential was maintained at 80 mV.
Mutant hERG trafficking assay.
The cell surface expression of a trafficking deficient mutant human ether-a-go-go (hERG) channel (G601S-hERG) was monitored with a hERG-Lite assay developed by ChanTest (Cleveland, OH) as described by Wible et al. (61). Briefly, HEK-293 cells expressing G601S-hERG with an HA epitope in the extracellular loop spanning transmembrane domains S1 and S2 were incubated for 16 h at 37°C with or without test compound. After preincubation, the cells were first incubated with rat anti-HA (1:500; Roche) for 2 h, after which they were rinsed three times and then incubated for 1 h with a secondary antibody cocktail containing; horseradish peroxidase-conjugated goat anti-rat IgG (Jackson Laboratories; 1:1,000) in blocking buffer plus SYBR green (Molecular Probes; 1:10,000). After cells were rinsed, SYBR green fluorescence was measured in a ThermoElectron Fluoroskan Ascent microplate reader (wavelengths: excitation, 485 nm; emission, 537 nm). To capture chemiluminescent signals, SuperSignal ELISA Femto maximum sensitivity substrate (Pierce, Rockford, IL; 100 ml/well) was added to each well. For each compound, the chemiluminescence signal in each well was normalized to the control (DMSO) to give a relative surface expression level. A significant increase in cell surface expression was observed when the test compound mean was above the vehicle control, mean ± 3 control SD.
HEK-293 cells expressing ΔF508-CFTR or HBE natively expressing wt-CFTR or ΔF508-HBE were incubated for 16 h at 37°C with or without test compound in assay medium. After incubation, cells were harvested in ice-cold PBS (without calcium and magnesium) plus 1 mM EDTA and pelleted at 1,000 g at 4°C. Cell pellets were lysed in 1% NP-40, 0.5% sodium deoxycholate, 200 mM NaCl, 10 mM Tris, pH 7.8, and 1 mM EDTA plus protease inhibitor cocktail (Sigma) used 1:250 for 30 min on ice. Lysates were spun clear for 10 min at 10,000 g at 4°C to pellet nuclei and insoluble material. Approximately 10 μg (HEK-293 cells) or 150 μg (HBE cells) total protein were heated in sample buffer without bromophenol blue at 37°C for 5 min and loaded onto a 3–8% Tris-acetate gel (Invitrogen). The gel was transferred to nitrocellulose and processed for Western blotting with monoclonal CFTR antibody, M3A7 (Upstate). Blots were developed by enhanced chemiluminescence. We quantified the relative amounts of bands B and C using NIH Image analysis of scanned autoradiograms.
Metabolic pulse-chase analysis.
HEK-293 wt-CFTR and ΔF508-CFTR cells were incubated for 16 h in assay medium with DMSO or compound. Metabolic labeling was performed as described (55) with some modifications. Cells were starved 30 min in DMEM without cysteine and methionine with 1% dialyzed FBS in the presence of compound. Cells were then pulsed with [35S]methionine and cysteine EXPRES35S35 label (Perkin-Elmer) for 15 or 60 min (labeling times for individual experiments are noted in the figure legends). Cells were washed and chased in assay medium with compound for 0–23 h. At each time point, cells were harvested and lysed in RIPA buffer, and CFTR was immunoprecipitated with M3A7. Samples were separated by SDS-PAGE and analyzed by autoradiography. Radioactivity was quantified by Phosphorimager analysis (GE Healthcare, Piscataway, NJ).
Cell surface biotinylation.
HEK-293 wt-CFTR and ΔF508-CFTR cells were incubated for 16 h in assay medium with DMSO or compound. Cell surface glycoproteins were labeled with biotin-LC-hydrazide (Pierce) as described by Prince et al. (40). Cells were immediately harvested and lysed after labeling. Postnuclear supernatants were immunoprecipitated for CFTR by M3A7. Samples were separated by SDS-PAGE, and proteins were transferred to nitrocellulose. Biotinylated proteins were visualized by horseradish peroxidase-streptavidin (GE Healthcare) via enhanced chemiluminescence.
Real-time quantitative PCR.
Cells were pretreated for 16 h with DMSO or compound in assay medium and harvested, and total RNA was extracted with RNeasy (Qiagen, Valencia, CA). cDNA was synthesized by iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). Real-time PCR was carried out for CFTR and 18S transcripts with iQ SYBR Green Supermix (Bio-Rad) using the iCyler (Bio-Rad). CFTR primers used were forward (5′- GCAGCCTTACTTTGAAACTC-3′) and reverse (5′AACAGCAATGAAGAAGATGAC-3′). Cycle thresholds were determined for CFTR and normalized to total mRNA.
Analysis of ubiquitin proteasome activity in vitro.
The activity of the ubiquitin proteasome was monitored with a model substrate as previously described (54). Briefly, Jurkat cells expressing 2XUb-β-lactamase reporter were plated in 96-well plates at a density of 1.5 × 106 cells/ml and incubated with the various concentrations of compounds. The cells were treated for 1 h with 100 μg/ml cycloheximide and then loaded with 1 μM CCF2-acetoxymethylester (AM) (Invitrogen). β-Lactamase activity was measured by quantifying the fluorescence emission from the cells with a CytoFluor 4000 plate fluorimeter (PerSeptive Biosystems, Foster City, CA). Background-subtracted emission values at 460 and 530 nm were expressed as a ratio of 460 to 530 nm, where a high ratio represents high β-lactamase activity.
HTS assays for potentiators and correctors of ΔF508-CFTR.
To identify small-molecule modulators of ΔF508-CFTR density and/or gating in the plasma membrane, we developed cell-based assays of membrane potential for either potentiators or correctors using NIH/3T3 cells expressing ΔF508-CFTR (Fig. 1). The assays report ΔF508-CFTR activity as a forskolin-stimulated depolarization in the presence of a Cl− gradient. Two separate HTS assays were validated to detect potentiators or correctors using genistein and temperature correction, respectively (see materials and methods for details).
Identification of ΔF508-CFTR correctors.
To identify novel chemical scaffolds that correct the trafficking of ΔF508-CFTR to the cell surface, we screened ∼164,000 synthetic compounds at a final concentration of 10 μM. The library consisted of chemically diverse druglike and leadlike compounds obtained from several commercial vendors and internal medicinal chemistry programs. Individual compounds were selected for further testing if their activity was >2.5 SD from the mean, which corresponded to 32% of the 27°C control. This resulted in the identification of 1,028 compounds, 185 of which confirmed in retests at 10 μM, giving a positive hit rate for the assay of 0.11%. Of these, 67 compounds were removed because of impurities (<85% purity by LC/MS) and poor chemical attractiveness, including 1) high molecular weight and/or cLogP, 2) known toxic or reactive functionalities, 3) chemotypes with limited protein interaction sites, and 4) structures not amenable to relatively rapid analog synthesis. The remaining compounds were selected for dose-response analysis with the HTS assay and were counterscreened in NIH/3T3 cells that did not express ΔF508-CFTR. A total of 108 compounds, comprising 13 structurally distinct scaffolds, were selected for further study. As independent confirmation of corrector activity, these compounds were then tested in the FRT epithelial cell line expressing ΔF508-CFTR using Ussing chamber measurement of Isc. Six of the 13 structurally distinct scaffolds identified in the HTS were confirmed in FRT-ΔF508 epithelia (data not shown). None of the compounds was active in untransfected FRT cells that did not express ΔF508-CFTR (data not shown). The quinazolinone shown in Fig. 2A (VRT-422) was among the most potent (EC50 = 3.7 ± 0.9 μM) and efficacious (47 ± 4%, 27°C) compounds identified in the HTS (Fig. 2B; Table 1).
Initial chemistry efforts to evaluate the structure activity relationship of the quinazolinone scaffold were undertaken with a parallel synthesis approach. This resulted in the synthesis and testing of >500 analogs in the optical assay, which led to the discovery that the para-bromo substituent could be replaced with a para-methoxy group without affecting potency and efficacy and that 4-alkoxyquinazolines had enhanced efficacy. Focused optimization of the 4-alkyl moiety resulted in the identification of VRT-325 having a cyclohexyloxy moiety at position 4 of the quinazoline (Fig. 2A). Although the potency of VRT-325 in the optical assay was similar to VRT-422, the efficacy (i.e., the magnitude of correction) of VRT-325 was significantly improved (Table 1; Fig. 2B) compared with VRT-422, as well as the known correctors of ΔF508-CFTR, 4-PBA, and 1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid diamides (compound 9; Ref. 22). Unlike other known correctors, curcumin was not active in the optical assay (Table 1). No response was observed in VRT-325-treated NIH/3T3 or FRT cells not expressing ΔF508-CFTR (data not shown), indicating that the compound effects were not due to modulation of endogenous channels or fluorescent artifacts.
We next investigated whether the increase in ΔF508-CFTR activity in response to either VRT-422 or VRT-325 corresponded to changes in ΔF508-CFTR processing and transport to the cell surface. We conducted many of the biochemical studies in HEK-293 cells expressing ΔF508-CFTR. This system yields a high expression of ΔF508 CFTR, making it well suited for antibody-labeling studies (59). To evaluate effects on processing, we compared the glycosylation pattern of ΔF508-CFTR using immunoblots of cell lysates from HEK-293 cells expressing ΔF508-CFTR in the presence and absence of compound. Incubation of HEK-293 cells expressing ΔF508-CFTR for 16 h in the presence of either VRT-422 or VRT-325 resulted in a dose-dependent increase in the steady-state levels of extensively glycosylated (150–170 kDa), mature ΔF508-CFTR (so-called band C; Fig. 3A), which is indicative of passage through the Golgi complex (27). In addition, there was an increase in the steady-state levels of the core glycosylated form (so-called band B; Fig. 3A). Quantification of the amounts of ΔF508-CFTR in the core-glycosylated (band B) and mature band C forms yielded an approximate EC50 of 4.0 μM for VRT-422 and 0.8 μM for VRT-325 (Fig. 3B). Similar results were obtained in NIH/3T3 cells expressing ΔF508-CFTR (data not shown).
To determine whether the compound-stimulated increase in ΔF508-CFTR maturation resulted in increased cell surface density, we performed cell surface biotinylation/immunoprecipitation and electrophysiological experiments. HEK-293 cells expressing ΔF508-CFTR were treated with VRT-325 for 18 h, cell surface proteins were biotinylated, and CFTR was immunoprecipitated from cell lysates. These experiments demonstrate that the ΔF508-CFTR band C induced by compound treatment was present on the cell surface, albeit at levels lower than wt-CFTR (Fig. 4A). In whole cell patch-clamp experiments, incubation of NIH/3T3 cells expressing ΔF508-CFTR for 16 h with VRT-422 or VRT-325 increased the ΔF508-CFTR current density (IΔF508) stimulated by forskolin and genistein (Fig. 4, B and C). The compound-corrected IΔF508 was anion selective and time and voltage independent and was inhibited by glibenclamide, all of which are characteristic features of CFTR-mediated currents (50). Acute application (<10 min) of the compounds to temperature-corrected cells did not increase IΔF508 (data not shown), indicating that they do not directly potentiate channel gating. Together with the biochemical steady-state (Fig. 3) and cell surface biotinylation experiments (Fig. 4A), these data confirm that VRT-422 and VRT-325 increased cell surface expression of mature ΔF508-CFTR, leading to increased functional activity.
The modest increase in the core-glycosylated band B form (Fig. 3A) could be due to increased expression of the cftr gene or inhibition of degradation by the proteosome. To investigate these possibilities, we tested whether the corrector compounds increase CFTR transcript levels. CFTR mRNA levels were monitored in NIH/3T3 cells stably expressing ΔF508-CFTR using real-time quantitative RT-PCR. In contrast to the positive control, sodium butyrate, no increases in ΔF508-CFTR mRNA levels were observed after 16-h incubation with 6.7 μM VRT-422 or 6.7 μM VRT-325 (Fig. 5A). We also found that the corrector compounds do not inhibit the ubiquitin proteasome-mediated degradation of a ubiquitin-β-lactamase fusion protein (Fig. 5B). These results indicate that the persistence of core-glycosylated ΔF508-CFTR was not due to direct compound inhibition of the ubiquitin-proteasome pathway, which has been demonstrated to regulate CFTR degradation (60).
The lack of effect of the correctors on CFTR transcription or the proteasome suggests that the increase in core-glycosylated ΔF508-CFTR may be due to its stabilization and/or retention in the endoplasmic reticulum (ER). To confirm this, we monitored the degradation rate of core-glycosylated ΔF508-CFTR after 16-h incubation of HEK-293 cells with 6.7 μM VRT-422, 6.7 μM VRT-325, or DMSO (Fig. 5, C and D). In DMSO-treated cells, the band corresponding to core-glycosylated ΔF508-CFTR decayed rapidly with a half-life (t1/2) of 28 min, similar to previously reported values (27). After 16-h incubation with VRT-422 or VRT-325, the kinetics of immature ΔF508-CFTR degradation were significantly slowed (t1/2 = 38 and 46 min, respectively), and the mature form was clearly present after 90 min (Fig. 5, C and D). Together, the data suggest that the corrector compounds act, at least in part, at the level of the ER to facilitate the proper folding of a proportion of ΔF508-CFTR, resulting in decreased ER degradation and more efficient ER export. Further studies using in vitro or structural approaches will be required to confirm that the correctors facilitate proper folding.
In addition to rapid degradation in the ER, ΔF508-CFTR also exhibits increased cell surface turnover relative to wt-CFTR (60). To further investigate the effects of the quinazolinone correctors on the maturation efficiency and stability of ΔF508-CFTR, we used pulse-chase experiments to examine ΔF508-CFTR maturation in the presence and absence of 6.7 μM VRT-325 (Fig. 6). In HEK-293 cells expressing ΔF508-CFTR, 16-h incubation with VRT-325 increased the amount of mature ΔF508-CFTR, with a maximum conversion of core-glycosylated ΔF508-CFTR to the mature form of 6.8 ± 2% in 6.7 μM VRT-325-treated cells vs. 3.4 ± 0.8% for DMSO treatment (Fig. 6A). The conversion of core-glycosylated ΔF508-CFTR to the mature form in compound-treated cells represents ∼15% of the maturation efficiency observed for wt-CFTR (Fig. 6A). In addition, 6.7 μM VRT-325 (n = 3) increased t1/2 of band C from 4.9 ± 0.1 to 7.7 ± 1.1 h (n = 3) compared with the DMSO-treated controls (Fig. 6, B and C), suggesting that the compound-corrected form of ΔF508-CFTR is more stable at the cell surface.
Selectivity of the quinazolinone correctors for ΔF508-CFTR.
Because it is possible that corrector compounds may act on other ER-arrested misfolded proteins in addition to ΔF508-CFTR, we assessed the effects of the correctors on hERG cardiac potassium channels containing a single-point mutation (G601S) known to cause hereditary human long-QT syndrome type 2 (18). This mutation generates a trafficking-deficient channel that is largely retained (90%) in the ER, resulting in reduced cell surface expression. Like CFTR, Hsp90 and Hsp70 mediate maturation of the hERG K+ channel (16) and trafficking-deficient mutants are restored by low-temperature incubation (18) and chemical chaperones (17). The cell surface expression of G601S-hERG expressed in HEK-293 cells was monitored with an antibody-based detection system that recognizes an epitope introduced into an extracellular loop of the G601S-hERG mutant. Incubation of G601S-hERG-HEK-293 cells for 16 h with VRT-422 or VRT-325 increased the cell-surface expression of G601S-hERG with a similar rank order efficacy as that observed for ΔF508-CFTR (Fig. 7). These data indicate that the action of these corrector compounds is not restricted to ΔF508-CFTR and raise the possibility that the compounds act on a target in a folding pathway shared by the two misfolded proteins rather than directly on CFTR itself.
Identification of novel ΔF508-CFTR potentiators using HTS.
Because restoration of normal lung function may require rescue of defective gating in addition to correcting the trafficking defect of ΔF508-CFTR, we screened 122,000 synthetic compounds from our in-house screening library in the potentiator HTS assay at a final concentration of 20 μM. Individual compounds were selected for further testing if their activity was >2.5 SD from the mean, which corresponded to >65% of the genistein control. This resulted in the identification of 1,535 compounds, 278 of which were confirmed in retests at 2 and 20 μM, giving a positive hit rate for the assay of 0.23%. Of these, 145 compounds were removed because of impurities (<85% purity by LC/MS) and poor chemical attractiveness. The remaining compounds were selected for dose-response analysis using the HTS assay and were counterscreened in NIH/3T3 cells that did not express ΔF508-CFTR. Fifty-three compounds, comprising 10 structurally distinct scaffolds, were selected for further study. As independent confirmation of potentiator activity, these compounds were tested in temperature-corrected ΔF508-CFTR-FRT using Ussing chamber measurement of Isc. All 10 structurally distinct scaffolds identified in the HTS were confirmed in the ΔF508-CFTR-FRT (data not shown). The pyrazole hit, VRT-532, shown in Fig. 8A was among the most potent and efficacious potentiators identified in the HTS using ΔF508-CFTR-NIH/3T3 cells (Table 1; Fig. 8B). The response to VRT-532 is observed only after forskolin addition, indicating that VRT-532 is a potentiator and not an activator of ΔF508-CFTR under these conditions. No response to forskolin and VRT-532 addition was observed in either NIH/3T3 cells or FRT epithelia that do not express ΔF508-CFTR (data not shown). Recently, HTS strategies have identified potentiators that are structurally distinct from those presented here and are ∼10-fold more potent against ΔF508-CFTR (38, 62).
In addition to ΔF508-CFTR, other mutations in the CFTR gene result in defective gating. This includes the missense mutation G551D, which results in defective gating but does not impair trafficking to the apical membrane (9). In FRT epithelial monolayers expressing G551D-CFTR, VRT-532 potentiated forskolin-stimulated Cl− secretion with an EC50 of 20 ± 3 μM (n = 3), which was about fivefold less potent than the EC50 observed for ΔF508-CFTR in temperature-corrected FRT monolayers (EC50 = 3.8 ± 0.5 μM; n = 46).
Potentiator activity was confirmed by inside-out patch-clamp recording in ΔF508-NIH/3T3 cells. Treatment with 20 μM VRT-532 in the presence of ATP and PKA increased the open probability (Po) of ΔF508-CFTR from 0.09 ± 0.04 to 0.39 ± 0.1 (n = 3; P < 0.05, paired t-test), primarily by increasing the open burst duration, and had no effect on single-channel conductance (Fig. 8C). The Po of ΔF508-CFTR in the presence of VRT-532 (0.39 ± 0.1) was similar to that of wt-CFTR (0.36 ± 0.04; n = 3) under identical recording conditions. VRT-532 did not increase the total cAMP concentration in NIH/3T3 cells in the presence or absence of the phosphodiesterase inhibitor IBMX, nor did it directly inhibit phosphodiesterases in in vitro enzyme assays (data not shown). Together, these results suggest that VRT-532 acts on ΔF508-CFTR to potentiate its gating activity.
Rescue of ΔF508-CFTR activity in primary CF airway cultures by correctors and potentiators.
To test the activity of the corrector VRT-325 and the potentiator VRT-532 in a physiologically relevant system, we used monolayers of HBE obtained from the airways of ΔF508-homozygous CF patient donors (ΔF508-HBE). HBE cultures represent highly differentiated, polarized surface airway cells and allow assessment of CFTR-mediated Cl− secretion with Isc under voltage-clamp control conditions (19). To monitor CFTR-dependent Isc in ΔF508-HBE cultures, amiloride was added to the apical side to block epithelial Na+ channels, and a basolateral-to-apical Cl− gradient was established. Figure 9, A and B, shows validation of the ΔF508-HBE model with low-temperature correction, potentiation with genistein, and inhibition by the known CFTR blockers, CFTRinh-172 (29) and glibenclamide (47). A small forskolin-stimulated current corresponding to ∼5% of wt-CFTR activity was observed in uncorrected cells and appears to be mediated by ΔF508-CFTR on the basis of its forskolin sensitivity, potentiation by genistein, insensitivity to Ca2+-activated Cl− channel inhibitor DIDS (data not shown), and inhibition by CFTRinh-172 or glibenclamide (Fig. 9).
Incubation of ΔF508-HBE with VRT-325 for 48 h increased the forskolin-stimulated Isc by about twofold compared with DMSO controls, with an EC50 of 1.4 μM (n = 5; Fig. 9, C–E) and a t1/2 of 40 h (Fig. 10A). The compound-corrected Isc was blocked by CFTR inhibitors (Fig. 9, C and D) and by the basolateral Na+-K+-2Cl− cotransporter inhibitor bumetanide (data not shown). To confirm that the increase in forskolin-stimulated Isc was due to increased Cl− secretion across the apical membrane, the basolateral membrane was permeabilized with nystatin. The percent increase in the amount of correction compared with untreated controls in response to 48-h pretreatment with VRT-325 in intact and nystatin-permeabilized ΔF508-HBE was 197 ± 23% (n = 6) and 210 ± 15% (n = 17), respectively. In addition to increasing apical membrane Cl− secretion, preincubation with 6.7 μM VRT-325 increased ΔF508-CFTR maturation in ΔF508-HBE (Fig. 9F). These results indicate that VRT-325 increases ΔF508-CFTR maturation and apical membrane density of functional channels in primary ΔF508-HBE cells, similar to the results in recombinant HEK-293 cells (Fig. 3A). Acute addition of the pyrazole potentiator (10 μM) VRT-532 to VRT-325-corrected ΔF508-HBE further increased Isc (Fig. 9, C and D), demonstrating that the effects of these potentiator and corrector compounds are additive. The additive effect of the potentiator was dose dependent, with an EC50 of 2.7 ± 0.2 μM (n = 13), which was similar to that in the potentiator HTS assay using ΔF508-NIH/3T3 cells (Table 1). In addition to potentiating ΔF508-CFTR activity, VRT-532 increased the forskolin-stimulated Isc in HBE isolated from non-CF bronchial tissue with an EC50 of 0.9 ± 0.4 μM (n = 3), which is similar to that observed in ΔF508-HBE.
One of the hallmarks of misfolded ΔF508-CFTR is reduced residence time in the plasma membrane compared with wt-CFTR (27), possibly because of decreased recycling to the cell surface after endocytosis. To determine whether VRT-325 altered the ΔF508-CFTR residence time in the apical membrane, we monitored the persistence of correction after washing out the compound from the HBE cultures (Fig. 10, B and C). Interestingly, the CFTR-mediated forskolin response was maintained for up to 36 h after the compound was removed (t1/2 of 18 h). In contrast, the forskolin-stimulated response in temperature-corrected ΔF508-HBE was abolished within 2 h after the cells were returned to 37°C (half time of 45 min). These results indicate that the residence time of compound-corrected ΔF508-CFTR in the apical membrane of CF-HBE is significantly improved to a level similar to wt-CFTR (27) and that compound correction produces a more stable form of the protein in the membrane than temperature correction. The persistence of the functional response is consistent with the biochemical evidence of increased stability of mature ΔF508-CFTR observed in pulse-chase experiments (Fig. 6).
Estimating potential clinical efficacy of correctors and potentiators.
Estimates of the amount of ΔF508-CFTR rescue believed to ameliorate the decline of lung function in CF patients range from 5% to 30% of wt-CFTR, based on correlations of CFTR genotype and Cl− secretion in nasal potential difference and sweat Cl− measurements with the clinical phenotype of CF patients with mild, moderate, and severe CF disease (9, 34–36). To evaluate the potential for the corrector VRT-325 to ameliorate lung disease in CF patients alone or in combination with the potentiator VRT-532, we compared the amount of Cl− secretion in compound-treated ΔF508-HBE with that in non-CF HBE endogenously expressing native wt-CFTR (Fig. 11). This comparison of in vitro activities is important because of the lack of clinically predictive animal models of CF lung disease (21). In non-CF HBE, forskolin stimulated a large, biphasic increase in Cl− secretion with net changes in Isc for the peak and steady-state levels of 53.0 ± 6.3 and 34.1 ± 4.5 μA/cm2, respectively (n = 30). A 48-h incubation with VRT-325 increased the forskolin response in CF-HBE from 2.0 ± 0.3 to 3.9 ± 0.5 μA/cm2 (n = 17), which is >10% of the sustained CFTR-dependent Cl− secretion in non-CF HBE (Fig. 11, A and B). In the absence of a basolateral-to-apical Cl− gradient and in the presence of amiloride, the peak forskolin response in untreated and VRT-325-treated ΔF508-HBE was 1.2 ± 0.2 and 2.0 ± 0.2 μA/cm2 (n = 6; P < 0.05, paired t-test), respectively. For comparison, the peak forskolin response in non-CF HBE under identical recording conditions was 13.9 ± 1.5 μA/cm2 (14 ± 1.5% of non-CF HBE; n = 4). These data indicate that the amount of correction is 14.1 ± 1.5% of the response observed in non-CF HBE, which is similar to the amount of correction observed in the presence of a basolateral-to-apical Cl− gradient (Fig. 11, A and B). Subsequent addition of the potentiator VRT-532 further increased ΔF508-CFTR-dependent Cl− secretion to levels >20% of that observed in non-CF HBE (Fig. 11, A and B), indicating that the two mechanisms can be combined to achieve additional efficacy and possibly synergy. In addition, VRT-325-induced correction was significantly greater than that with 1.5 mM 4-PBA and with temperature correction in ΔF508-HBE (Fig. 11, C and D). Other reported correctors (curcumin and compound 9) were not active in ΔF508-HBE (Fig. 11D). These results indicate that VRT-325 is significantly more efficacious than other known correctors of ΔF508-CFTR trafficking in HBE isolated from CF patients.
We have used HTS to identify two classes of small-molecule CFTR modulators: one class potentiates PKA-stimulated gating of defective ΔF508-CFTR (VRT-532) and the other class corrects the trafficking defect of ΔF508-CFTR (VRT-422 and VRT-325) through an apparent effect on cell folding and/or trafficking machinery. These compounds have already been made available for research purposes, and Loo et al. (26) have confirmed the activity of VRT-325 (designated CFcor-325 in Loo et al.) to correct the trafficking of ΔF508-CFTR and other misprocessed CFTR mutations. The potentiator and corrector compounds show activity alone and in combination in primary airway cultures generated from cells isolated from CF patients, indicating that they function in a therapeutically relevant cell system. Further optimization of potency, efficacy, selectivity, and evaluation of pharmacokinetic and toxicological parameters will be required to assess the possibility for these compound classes to generate drugs. Together with the efforts of Verkman and colleagues (30, 37, 38), our results provide validation of the use of engineered cell systems to identify CFTR modulators that are active in human primary disease airway epithelia and support the possibility of a drug discovery strategy based on rescue of the basic genetic defect responsible for CF. Because polarized airway epithelia may contain factor(s) or pathways that are not present in NIH/3T3 cells, screening assays employing epithelial cells, such as that described by Yang et al. (62), may identify additional compound classes that could operate in a cell-specific manner.
The mechanistic data on the quinazolinone corrector compounds presented here suggest a model in which the compounds act primarily or initially at the level of the ER to facilitate the folding and export of ΔF508-CFTR. Data supporting compound action in the ER include the increased persistence of core-glycosylated ΔF508-CFTR and lack of effects on CFTR transcription and proteasome activity. The relatively slow increase in functional cell surface ΔF508-CFTR observed in CF-HBE (Fig. 10) is likely due to several compounding actions that result in an accumulation of corrected cell surface ΔF508-CFTR over the course of several days. First, the fact that <10% of newly synthesized ΔF508-CFTR is converted to the mature form (Fig. 6A) suggests that corrector compound action is able to catalyze an export-competent conformation in a relatively small fraction of the ER pool of ΔF508-CFTR. Second, the slow decrease in cell surface CFTR activity in ΔF508-HBE after compound washout (Fig. 10) suggests that the ΔF508-CFTR that escapes the ER due to compound treatment possesses a conformation similar to wt-CFTR. These two effects would result in a situation where a small amount of correctly folded ΔF508-CFTR is produced on compound addition, and this stable form accumulates at the cell surface until ER exit is balanced by cell surface turnover such that steady-state is achieved several days later. The possibility that the corrector compounds result in properly folded ΔF508-CFTR suggests additional parameters for future investigation, including cell surface recycling rate, channel gating efficiency (Po), and CFTR folding in vitro. In addition, further studies in ΔF508-HBE could be used to investigate the functional consequences of ΔF508-CFTR upregulation, including effects on ion transport, airway surface fluid transport, and mucociliary clearance (2, 33). The HBE system also allows the direct comparison of Cl− secretion and HBE function not only between CF and non-CF epithelia but also between CF patients with genotypes other than ΔF508, which could lead to insights regarding the relationship between CFTR activity and clinical phenotype.
Our results showing that VRT-325 increased trafficking of G601S-hERG (Fig. 7), as well as those of Loo et al. (26) showing a similar effect on the G268V P-gp mutant and other CFTR mutations, indicate that VRT-325 is not a specific corrector of ΔF508-CFTR, and it is possible that the molecular target of VRT-325 is a component of the ER folding/quality control machinery. Possible targets include molecular chaperones such as Hsc70 and Hsp90, which have been reported to act on G601S-hERG (16). Further work is needed to characterize the mechanism of action of VRT-325 and to identify which classes of proteins are affected by this compound. However, these data suggest the intriguing possibility that correction compounds may be useful for the pharmacological treatment of other diseases known to be caused by protein misfolding (1, 58).
Although several potent CFTR potentiators have been reported in the literature, CFTR potentiator VRT-532 appears to be distinguished by its ability to potentiate the gating of several forms of CFTR, including ΔF508-, G551D-, and wt-CFTR. Like genistein (38), VRT-532 had an approximately fivefold lower affinity for G551D-CFTR than ΔF508-CFTR. Other structural classes of ΔF508-CFTR potentiators such as phenylglycines show greater potency shifts vs. G551D (∼70 nM and ∼1,100 nM against ΔF508- and G551D-CFTR, respectively) or are inactive against G551D-CFTR (benzothiophenes and sulfonamides) (38, 62). Further characterization of the ability of VRT-532 to potentiate other CFTR mutations is warranted. Patch-clamp recording and cell signaling analysis suggest that VRT-532 acts directly on the channel to increase the channel Po to levels observed for wt-CFTR. Further studies, however, are required to confirm that VRT-532 binds directly to CFTR to alter its activity.
Under our culture and recording conditions, a small forskolin-stimulated Cl− current was observed in uncorrected HBE isolated from ΔF508 homozygous patients, and this current was increased by either genistein or VRT-532. The total Cl− current observed in ΔF508-HBE was ∼5% of that observed in non-CF HBE, was blocked by CFTR inhibitors, as well as by the basolateral Cl− transport inhibitor, bumetanide. In contrast, the Ca2+-activated Cl− channel blocker DIDS had no effect. These data suggest the presence of residual ΔF508-CFTR activity in ΔF508-HBE. The presence of residual ΔF508-CFTR activity in airway is controversial. For example, residual Cl− current has been reported in primary cultures of nasal (48) and airway epithelia (12), and immunohistochemical studies have demonstrated the presence of apical localized ΔF508-CFTR in primary cultures of respiratory tissue (13, 39). However, other studies have failed to demonstrate the presence of ΔF508-CFTR on the apical membrane of cultured HBE (25). The extent to which the presence of residual ΔF508-CFTR activity in vitro is related to the culture conditions is not clear. The ability to detect residual activity in our study could be due to the use of a basolateral-to-apical Cl− gradient, which markedly increased the signal-to-noise ratio compared with that observed in the absence of a Cl− gradient. However, there is growing evidence that residual activity is present in a fraction of the ΔF508-homozygous patient population (15, 23, 56). These patients exhibit a milder CF phenotype, including improved lung function (forced expiratory volume in 1 s) and pancreatic sufficiency. This is an important issue to resolve from a clinical perspective because it is possible that such patients could benefit from a potentiator therapy.
A key question in the design of CFTR modulators to treat CF is this: how much CFTR-mediated Cl− secretion is needed to achieve clinical benefit? Genotype/phenotype correlations of Cl− channel function with disease severity show that mutations resulting in a only modest recovery toward wild-type levels (e.g., 5–30%) increase in CFTR expression or activity (e.g., A455E, 2,789 + 5G→A, 5T, R334W, R347P, and R117H) and are typically associated with pancreatic sufficiency, a slower rate of pulmonary function decline, and a better severity index than that shown in patients with severe disease genotypes (5, 6, 10, 36, 49). We have used HBE cells in an attempt to benchmark the effects of small molecules for potential efficacy. VRT-325 alone or in combination with VRT-532 increased Cl− secretion to >10–20% of non-CF HBE, levels expected to provide therapeutic benefit in CF patients (9, 34–36). These findings support the prospect for small-molecule therapies to restore clinically significant function of CFTR in a majority of CF patients.
Vertex Pharmaceuticals has a financial interest in the discovery and development of small-molecule therapies for CF.
We thank Cystic Fibrosis Foundation Therapeutics for generous financial support, as well as Melissa Ashlock, Robert Beall, Preston Campbell, John Chabala, Kevin Foskett, Ray Frizzell, Eric Gordon, Diana Wetmore, William Guggino, Christopher M. Penland, Bruce Stanton, and Jeffrey Wine for guidance and support. We also thank Dr. Michael Welsh for providing wt-CFTR and ΔF508-expressing NIH/3T3 and FRT cells, Neil Bradbury for providing ΔF508-expressing HEK-293 cells, and National Disease Research Interchange for providing lung samples for HBE cells.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society