S945L-CFTR molecular dynamics, functional characterization and tezacaftor/ivacaftor efficacy in vivo and in vitro in matched pediatric patient-derived cell models

Abstract

Cystic Fibrosis (CF) results from over 400 different disease-causing mutations in the CF Transmembrane Conductance Regulator (CFTR) gene. These CFTR mutations lead to numerous defects in CFTR protein function. A novel class of targeted therapies (CFTR modulators) have been developed that can restore defects in CFTR folding and gating. This study aimed to characterize the functional and structural defects of S945L-CFTR and interrogate the efficacy of modulators with two modes of action: gating potentiator [ivacaftor (IVA)] and folding corrector [tezacaftor (TEZ)]. The response to these modulators in vitro in airway differentiated cell models created from a participant with S945L/G542X-CFTR was correlated with in vivo clinical outcomes of that participant at least 12 months pre and post modulator therapy. In this participants' airway cell models, CFTR-mediated chloride transport was assessed via ion transport electrophysiology. Monotherapy with IVA or TEZ increased CFTR activity, albeit not reaching statistical significance. Combination therapy with TEZ/IVA significantly (p = 0.02) increased CFTR activity 1.62-fold above baseline. Assessment of CFTR expression and maturation via western blot validated the presence of mature, fully glycosylated CFTR, which increased 4.1-fold in TEZ/IVA-treated cells. The in vitro S945L-CFTR response to modulator correlated with an improvement in in vivo lung function (ppFEV₁) from 77.19 in the 12 months pre TEZ/IVA to 80.79 in the 12 months post TEZ/IVA. The slope of decline in ppFEV1 significantly (p = 0.02) changed in the 24 months post TEZ/IVA, becoming positive. Furthermore, there was a significant improvement in clinical parameters and a fall in sweat chloride from 68 to 28 mmol/L. The mechanism of dysfunction of S945L-CFTR was elucidated by in silico molecular dynamics (MD) simulations. S945L-CFTR caused misfolding of transmembrane helix 8 and disruption of the R domain, a CFTR domain critical to channel gating. This study showed in vitro and in silico that S945L causes both folding and gating defects in CFTR and demonstrated in vitro and in vivo that TEZ/IVA is an efficacious modulator combination to address these defects. As such, we support the utility of patient-derived cell models and MD simulations in predicting and understanding the effect of modulators on CFTR function on an individualized basis.

Keywords: CFTR; airway epithelial cell models; cystic fibrosis; modulators; molecular dynamics; personalized medicine.

Figure 1

Effect of TEZ/IVA in vivo on lung function, nutritional factors and sweat chloride concentration in a patient with S945L/G542X-CFTR. (A) Lung function (ppFEV₁) 12 months pre and 24 months post initiation of TEZ/IVA treatment. Serial (B) weight, (C) height and (D) body mass index (BMI) measurement percentiles from three years of age until 24 months post initiation of TEZ/IVA. Percentiles are based on data from the Centers for Disease Control and Prevention (CDC) growth charts. (E) Table summarizing statistical models applied to clinical outcome measurements and the best-fit values pre and post initiation of TEZ/IVA treatment. A simple linear regression was fit to the ppFEV₁ (A) and BMI percentile (D) measurements from the time periods pre TEZ/IVA (grey line) and post TEZ/IVA (black line). The slope for each time period was estimated and the difference between them was tested. A one phase decay model was fit to the weight (B) and height (C) percentile measurements from the time periods pre TEZ/IVA (grey line) and post TEZ/IVA (black line), with zero as a shared plateau value and separate intercept values. The decay parameter for each time period was estimated and the difference between them was tested. (F) Sweat chloride concentration (mmol/L) measurements taken at baseline and 24 months post initiation of TEZ/IVA treatment. Dotted lines indicate lower limits of CF diagnostic (>60 mmol/L) and CF indeterminate (30–59 mmol/L) value ranges. No statistical modelling for sweat chloride concentration was performed due to single measurements pre and post TEZ/IVA.

Figure 2

Effect of TEZ/IVA in vitro in S945L/G542X-CFTR patient-derived human nasal epithelial cell (hNEC) air-liquid interface (ALI) cultures. (A) Representative Ussing chamber recordings of short circuit current (I_SC) in hNECs from a S945L/G542X participant. The protocol used to measure functional CFTR expression in hNECs pre-treated at 48 h with corrector (5 μM VX-661 or 3 μM VX-445/18 μM VX-661) or vehicle (0.01% DMSO), followed by sequential addition from 0 h with 100 μM apical amiloride (1. Amil), apical 0.01% DMSO or 10 μM VX-770 (2. DMSO, VX-770), 10 μM basal forskolin (3. Fsk), 30 μM apical CFTR-specific inhibitor (4. CFTR_Inh−172) and 100 μM apical ATP (5. ATP). A basolateral-to-apical chloride gradient was used. (B) Bar graphs of total mean CFTR activated currents stimulated by DMSO or VX-770 plus Fsk in hNECs untreated or pre-treated with VX-661 or VX-445/VX-661. Data are from 2 to 3 independent ALI cultures and are represented as Mean ± SEM. Ordinary one-way ANOVA was used to determine statistical significance. *p < 0.05. (C) Western blot in cell lysates from S945L/G542X differentiated nasal epithelial cells. Cell lysates of F508del/F508del and WT/WT hNECs were used as controls to indicate protein size of CFTR bands B and C. To avoid signal saturation, the WT/WT sample was loaded at 7% of the mutant CFTR samples. S945L/G542X CFTR maturation following treatment with LUM/IVA (VX-809/VX-770), TEZ/IVA (VX-661/VX-770) or ELX/TEZ/IVA (VX-445/VX-661/VX-770) was quantified by measuring the level of mature CFTR (band C) in treated cells as a fold increase compared to mature CFTR in untreated cells. Band C represents the mature, complex-glycosylated CFTR. Band B represents the immature, core-glycosylated CFTR. All data were normalized to the calnexin loading control.

Figure 3

Molecular dynamics and free energy calculations to characterise the S945L mutation. (A) The open structure of the CFTR protein is used in all computational work (PDB ID: 6MSM), which is depicted here as coloured ribbons embedded within a grey lipid bilayer. A red blob shows the location of the S945 amino acid between TM8 and the R domain; this region is critical for channel gating functions. In simulations 1, 2 and 3 of WT-CFTR, a network of hydrogen bonds is found that connects Y852 to S945. This network is absent in simulations of S945L-CFTR, as the mutation replaces the polar serine side chain with a hydrophobic leucine side chain that is incapable forming hydrogen bonds. (B) The distance between S/L945 and Y852 alpha carbon atoms measured in three unbiased simulations of WT- and S945L-CFTR at 77°C (350 K). The distance trace is stable in WT-CFTR due to the formation of hydrogen bonds between Y852 and S945, where the alpha carbons of these amino acids remain at a distance of 9.1 ± 0.3 Å. Distances are denoted here by Mean ± SD. Some destabilisation is seen in S945L-CFTR with unbiased MD: in simulations 2 (orange) and 3 (green) the average distance is 9.4 ± 0.5 Å, while in simulation 1 (blue line), a conformational transition occurs after approximately 400 ns, that increases the average distance to 12.0 ± 0.6 Å. (C) Relative free energy as the distance between S/L945 and Y852 is varied in umbrella sampling calculations. In WT-CFTR there exists one deep free energy minimum at 9.5 Å (red arrow). In S945L-CFTR there are instead three shallow free energy minima between 9.0 Å and 13.0 Å (black arrows), separated by small barriers with heights of 2.0 kcal/mol or less. Black horizontal bars indicate regions of two distinct local conformations: one at distances below 10.8 Å, where the R domain and TM8 are properly folded and a Y852-S945 hydrogen bond network exists; the other at distances above 11.2 Å, where the R domain and TM8 separate into a misfolded conformation. The relative free energies obtained through umbrella sampling suggest that such misfolding is more likely to occur in S945L-CFTR.