Mechanism-based corrector combination restores ΔF508-CFTR folding and function

Abstract

The most common cystic fibrosis mutation, ΔF508 in nucleotide binding domain 1 (NBD1), impairs cystic fibrosis transmembrane conductance regulator (CFTR)-coupled domain folding, plasma membrane expression, function and stability. VX-809, a promising investigational corrector of ΔF508-CFTR misprocessing, has limited clinical benefit and an incompletely understood mechanism, hampering drug development. Given the effect of second-site suppressor mutations, robust ΔF508-CFTR correction most likely requires stabilization of NBD1 energetics and the interface between membrane-spanning domains (MSDs) and NBD1, which are both established primary conformational defects. Here we elucidate the molecular targets of available correctors: class I stabilizes the NBD1-MSD1 and NBD1-MSD2 interfaces, and class II targets NBD2. Only chemical chaperones, surrogates of class III correctors, stabilize human ΔF508-NBD1. Although VX-809 can correct missense mutations primarily destabilizing the NBD1-MSD1/2 interface, functional plasma membrane expression of ΔF508-CFTR also requires compounds that counteract the NBD1 and NBD2 stability defects in cystic fibrosis bronchial epithelial cells and intestinal organoids. Thus, the combination of structure-guided correctors represents an effective approach for cystic fibrosis therapy.

Figure 1. Combination of corrector and suppressor mutations restores ΔF508-CFTR folding, PM expression and function

(a) NBDs-MSDs interfaces in the CFTR open structural model. Some critical interface residues are indicated. (b) Relative PM density of ΔF508-CFTR containing R1S (Y-axis) or R1070W (X-axis) upon corrector treatment was measured by ELISA in BHK cells (n=6–12). Correctors indicated by black symbols are listed in Supplementary Fig. 2a. (c) Structural preference of correctors to NBD1-MSD2 interface over NBD1 stability defect was visualized by plotting sum of correction (Res_R1S + Res_R1070W, Y-axis) as a function of log₂ ratio of the augmented PM density of ΔF508-CFTR with R1S and R1070W (Res_R1S/Res_R1070W, X-axis). (d) Maturation efficiency of ΔF508-CFTR was measured by metabolic pulse-chase experiments. B, immature core-glycosylated; C, mature complex-glycosylated form. (e–f) PM density (e; n=8–20), and function (h; n=3–4) of ΔF508-CFTR with or without suppressor mutation was measured by ELISA and apical Cl⁻ current (ICl(apical)), respectively, in CFBE41o- cells. Na⁺/K⁺-ATPase (ATPase) was used as a loading control. (g) Correlation between T_m of NBD1 variants of 0S, 1S, 3S, R1S and R4S (listed in Table S2) and PM density of the respective CFTR variants in BHK cells (n=8–12) in the presence or absence of correctors. The data were fitted by linear regression analysis and slopes were listed in the result as %/°C unit. Correctors (C3 at 10 μM; C18 or VX-809 at 3 μM) were applied for 24 h at 37°C. Data are means ± SEM.

Figure 2. VX-809 functions as a pharmacological chaperone of CFTR

(a-b) Effect of VX-809 and C18 correctors on thermal inactivation of the ΔF508-CFTR-2RK reconstituted into artificial phospholipid bilayer. (a) Representative channel activity is shown for WT or ΔF508-2RK CFTR at 24°C, 30°C or 36°C during the temperature ramp in the absence or presence of 3 μM VX-809. The processing defect of the ΔF508-CFTR variants was rescued at 26°C prior to microsome isolation. Channel activity is also shown at higher time resolution for each condition at 36°C. The closed (c) state of the channels is indicated. The control gating of ΔF508-CFTR-2RK was derived from separate experiments at 24°C and 27°C plus 36°C. Incorporation of two channels was observed at 27–36°C. The ΔF508-CFTR-2RK activity was recorded in separate experiment at 24°C plus 30°C and 36°C in presence of VX-809. (b) The P_o of PKA-activated CFTRs was analyzed at the indicated temperature as described in panel a. Cumulative duration of single channel experiments for WT and ΔF508-CFTR-2RK was at least 15 min. Number of independent experiments is indicated in the bars (n=8–25). Data represent means ± SEM and significance was tested by paired t-test. *<0.05, **p<0.01. (c) VX-809 in silico docking to open ΔF508-CFTR (top panel) or ΔF508-CFTR-ΔNBD2 model (bottom panel) obtained by AutoDock. The first four VX-809 clusters are ranked based on their lowest binding energy pose in ascending order and are illustrated on the model using PyMOL. For clarity, NBD2 and R-domain are hidden from the full-length model. Red, blue, magenta, cyan: clusters of VX-809 with increasing binding free energy.

Figure 3. Evaluating corrector mechanism by using CFTR variants

(a–c) Effect of correctors (10 μM C3 or C4, 3 μM C18 or VX-809 for 24 h at 37°C) on cellular expression (upper) and PM density (lower) of R1070W (a), F508G (b), and R170G (c) CFTR variants in BHK cells, measured by Western blotting and cell surface ELISA (n=8), respectively. (d) Effect of V510D or R1070W mutation on the PM density of corrector treated ΔF508-CFTR-3S measured by cell surface ELISA in BHK cells and expressed as % of the respective DMSO control (n=8). Same results are also shown in Supplementary Fig. 6a as % of WT. (e–f) Effect of NBD2 deletion (ΔNBD2) on corrector effect measured by the PM density (e, n=6–9) and cellular expression (f) of ΔF508-CFTR-R1S. Correctors (5 μg/ml core-corr-II, 10% glycerol, others are same as in panel a–c) were added for 24 h at 37°C. Data represent means ± SEM.

Figure 4. Effect of correctors on the isolated NBD1 stability in vitro and in vivo

(a–c) Melting temperature (T_m) of human ΔF508-NBD1-1S (a) or ΔF508-NBD1 (b, c) was determined by differential scanning fluorimetry (DSF)^, as described in Methods. The indicated correctors or chemical chaperones (CC) (see Supplementary Table 1) were present during thermal unfolding at the following concentration: 3 μM and 10 μM C1, C2, C3, C4, C5, C6, C7, C9, C11, C12, C13, C14, C17 and CoPo; 10 μM and 20 μM C8, C15 and C16; 1 μM and 3 μM C18 and VX-809, 5 μM and 15 μM RDR1; 2.5 μg/ml and 5 μg/ml core-corr-II; 5% and 10% glycerol; 150 mM and 300 mM TMAO, taurine, myo-inositol and D-sorbitol. WT-NBD1-1S or WT-NBD1 was used as a positive control. (c) T_m of ΔF508-NBD1 was determined by DSF as in panel b. Data represent mean ± SEM (n=3). Change in T_m was considered statistically significant if they were > ± 3SD (30.2 ± 0.4°C) of DMSO treated ΔF508-NBD1 T_m. (d) PM density of CD4T-ΔF508-NBD1-1S in COS7 cells was measured by PM ELISA using anti-CD4 Ab. The ELISA signal of mock-transfected cells is also indicated. Cells were treated with corrector for 24 h at 37°C. CD4T-WT-NBD1-1S (WT-1S) and CD4T-ΔF508-NBD1-R4S (ΔF-R4S) were used as positive control. Data represents mean ± SEM (n=8–12). (eh). Conformational dynamics of wt and ΔF508-NBD1-1S was determined by HDX-MS following 5 min D₂O labelling in the presence or absence of 10 μM VX-809. Data are means ± SEM (n=3). Representative peptides are shown from > sixty peptides obtained by pepsin digestion.

Figure 5. Combination of correctors targeting distinct structural defects completely restores ΔF508-CFTR folding, PM expression and stability in BHK cells

(a) Schematic model shows the proposed targets of available correctors based on in vitro, in vivo and in silico analysis. Chemical chaperones (CC, e.g. glycerol and myo-inositol) preferentially, but presumably not exclusively stabilize the ΔF508-NBD1. VX-809, C18 and C3 target the NBD1-MSD1/2 interface and C4 and core-corr-II target the NBD2. (b–f) ΔF508-CFTR PM density (b; n=6–12), cellular expression (c), ER folding efficiency (d–e; n=3–4) and PM stability after 4 h (f; n=5–12) were determined upon treatment with correctors (10 μM C3 or C4; 3 μM C18 or VX-809, 5 μg/ml core-corr-II; 10% glycerol) or their combinations in BHK cells by cell surface ELISA, Western blotting, metabolic pulse-chase technique and cell surface ELISA, respectively. All data represent means ± SEM.

Figure 6. Corrector combination restores ΔF508-CFTR functional expression in polarized epithelial cell lines and rectal organoids from ΔF508 CF patients

(a–e) ΔF508-CFTR PM density (a, n=7–22), cellular expression (b), PM stability (c, n=16) and function (d, n=3–4) in CFBE41o- cells. Correctors (10 μM C3 or C4, 3 μM C18 or VX-809, 2.5 μg/ml core-corr-II, 5% glycerol [Gly]) were applied for 24 h at 37°C. For comparison, ΔF508-CFTR was rescued by low temperature at 26°C or 30°C for 36–48 h. (d) Representative records of forskolin and 100 μM genistein (gen) activated ΔF508-CFTR ICl(apical) following corrector treatment are shown. (e) Summary of the peak, Inh₁₇₂-sensitive ICl(apical) of ΔF508-CFTR in CFBE41o- cells after corrector treatment (n=3–4). (f) Fluorescence laser confocal images of calcein-green loaded CF rectal organoids before and after forskolin-induced swelling (FIS) for 60 min in the presence of correctors (2 μM, VX-809, C4 and/or 125 mM myo-inositol (myo)), and inhibitor (Inh; 50 μM CFTR inhibitor-172 and GlyH-101). (g) Time course of FIS of CF organoids from a representative experiment. Organoids were treated as described in Methods. FIS was expressed as percentage of initial cell cross-sectional area before forskolin stimulation. (h) FIS was expressed as percentage of the area of 2 μM VX-809 treated cells after 60 min of FIS. Besides the measured FIS (red bar), the predicted sum of the individual corrector effect (2 μM C3, C4, C18; 0.1 or 2 μM VX-809; 125 mM myo-inositol) is indicated, assuming additive effect (blue bar). The mean FIS from three healthy controls (WT) is included. No FIS was observed after treatment of myo-inositol, C4 and VX-809 in organoids carrying E60X/4015ATTdel CFTR mutant. Data represent means ± SEM from six independent experiments performed on three ΔF508 homozygous patients.