Positional effects of premature termination codons on the biochemical and biophysical properties of CFTR

Abstract

About one-third of genetic diseases and cancers are caused by the introduction of premature termination codon (PTC). In theory, the location of the PTC in a gene determines the alternative mechanisms of translation, including premature cessation or reinitiation of translation, and read-through, resulting in differential effects on protein integrity. In this study, we used CFTR as a model system to investigate the positional effect of the PTC because of its well-understood structure-function relationship and pathophysiology.

The characterization of three PTC mutations, E60X-, G542X- and W1282X-CFTR revealed heterogenous effects of these PTCs on CFTR function. The W1282X mutation results in both C-terminus truncated and read-through proteins that are partially or fully functional. In contrast, only the read-through protein is functional with E60X- and G542X-CFTR, although abundant N-terminus truncated proteins due to reinitiation of translation were detected in E60X-CFTR.

Single-channel studies of the read-through proteins of E60X- and G542X-CFTR demonstrated that both mutations have a similar single-channel amplitude as WT, and good responses to high-affinity ATP-analogue, suggesting intact ion permeation pathways and NBDs, albeit with reduced open probability (Po). The comparison of the Po of these mutations with the proposed missense mutations revealed potential identities of the read-through products. Importantly, a majority of the functional protein studied responds to CFTR modulators like GLPG1837 and Lumacaftor. These results not only expand current understanding of the molecular (patho)physiology of CFTR, but also infer therapeutic strategies for different PTC mutations at large.

INTRODUCTION

Premature termination codons (PTCs) account for about one-third of the ultimate molecular mechanisms responsible for various genetic diseases and cancers (Frischmeyer & Dietz, 1999). Several defects in protein biogenesis are known to result from PTCs. First, after transcription, a portion of the PTC-harboring mRNA is degraded by the nonsense-mediated mRNA decay (NMD; Maquat, 2004). Even though some mRNA may escape this cellular quality control process, three other detrimental scenarios could take place. The ribosomal complex will stop at the PTC, and a premature cessation of the translation will result in the production of a protein with C-terminal truncation (Keeling & Bedwell, 2011). However, if the PTC is located in the 5’ region of the gene, the translation may be reinitiated from the downstream start codons, leading to a protein with various degrees of N-terminal truncation (Zhang & Maquat, 1997). Finally, PTCs can occasionally be suppressed endogenously or by chemicals known as read-through reagents (e.g., gentamicin and G418) or PTC124 (also known as Ataluren), resulting in significant “read-through” events (Gesteland & Atkins, 1996; Hermann, 2007; Welch et al., 2007; Du et al., 2008). During the read-through, near-cognate aminoacyl-tRNAs are accommodated in the ribosome, leading to the insertion of a range of unintended amino acids into the nascent polypeptide chain; hence missense mutations at the site of the PTC are almost inevitable during the read-through (Xue et al., 2017).

PTCs are found in ~10% of patients suffering from cystic fibrosis (CF), a life-shortening autosomal recessive hereditary disease mostly afflicting Caucasian population (Wilschanski, 2012; De Boeck & Amaral, 2016). CF is caused by the loss-of function mutations of the cftr gene encoding a chloride channel cystic fibrosis transmembrane conductance regulator (CFTR; Riordan et al., 1989; O’Sullivan & Freedman, 2009; Wilschanski, 2012; De Boeck & Amaral, 2016). Because of a fairly large and complete databank that offers clear genotype-phenotype relationship (http://www.cftr2.org), and a comprehensive understanding of the structure and function of the CFTR protein, (Zhang & Chen, 2016; Liu et al., 2017; Zhang et al., 2017; Hwang et al., 2018; Zhang et al., 2018), CF or CFTR can serve as an ideal model system to investigate the consequences of PTCs at a fundamental level.

As a member of the ATP-binding cassette (ABC) transporter protein superfamily, the CFTR protein inherits the similar canonical motifs including two transmembrane domains (TMD1 and TMD2) constructing the ion permeation pathway (Gao & Hwang, 2015; Linsdell, 2016), each conjoined by a nucleotide binding domain (NBD1 and NBD2) which controls gate opening and closing by ATP binding and hydrolysis (Gadsby et al., 2006; Jih & Hwang, 2012). In addition, the CFTR protein possesses a unique regulatory domain (RD) that is flanked by two TMD-NBD complexes; the phosphorylation of multiple consensus serine/threonine residues in the RD by protein kinase A (PKA) is a prerequisite for the channel activation (Gadsby & Nairn, 1999; Ostedgaard et al., 2001; Hwang et al., 2018).

Based on the molecular mechanisms underlying CFTR defects, pathogenic mutations are categorized into six classes (Boyle & De Boeck, 2013; Wang et al., 2014): no functional protein production due to nonsense or other mutations (Class I), defective trafficking of CFTR protein (Class II), abnormal open probability (Class III), decreased CFTR single channel conductance (Class IV), reduced CFTR protein synthesis (Class V), and lowered plasma membrane stability (Class VI). Proper characterization and classification of each pathogenic mutation are crucial not only for the mechanistic understanding of CFTR malfunction, but also for deploying treatment regimen tailored to the specific mutation carried by the patients with CF. To date, only two types of CFTR modulators are approved by the FDA for clinical use: CFTR correctors (e.g., VX-809 a.k.a. Lumacaftor and VX-661 a.k.a. Tezacaftor) that improve surface expression of CFTR by promoting proper protein folding (Van Goor et al., 2011; Taylor-Cousar et al., 2017), and CFTR potentiators (e.g., VX-770 a.k.a Ivacaftor) which increase the open probability (P_o) of the channel (Van Goor et al., 2009; Eckford et al., 2012; Jih & Hwang, 2013). Correctors are mainly approved for treating patients with Class II mutations such as F508del, and potentiators for patients with Class III mutations like G551D (Harrison et al., 2013; Char et al., 2014). However, many mutations possess multiple deficits. For example, the most common pathogenic mutation F508del was first defined as a prototypical Class II mutation, and yet it also manifests severe gating defect (Class III) and decreased plasma membrane stability (Class VI) (Dalemans et al., 1991; Lukacs et al., 1993; Varga et al., 2008; Jih et al., 2011). The multi-faceted dysfunction of these pathogenic mutations demands detailed functional studies for each disease-associated mutation in order to optimize the treatment regimen.

To systematic investigate the positional effects of PTCs, we characterized three Class I mutations with their PTCs located at positions across the protein (from C-terminus to N-terminus): W1282X (c.3846G>A) at the NBD2 with TGA stop codon, G542X (c.1624G>T) at the NBD1 of the protein with a TGA codon and E60X (c.178G>T) at the lasso motif of the CFTR protein with a TAG stop codon (Liu et al., 2017). C-terminus truncated CFTR with a lower molecular weight was detected in W1282X-CFTR using Western blotting, whose expression can be augmented by corrector VX-809. Electrophysiological experiments suggest the chloride current is carried by a mixture of functional C-terminus truncated and read-through proteins. For G542X-CFTR, although Western blot experiments failed to detect any full-length CFTR protein, cAMP-dependent chloride currents likely from a minute amount of read-through products were recorded. In contrast, two CFTR bands were resolved in the Western blot of E60X-CFTR, but further analysis suggests that these proteins are nonfunctional and may represent N-terminus truncated CFTR generated by the reinitiation of the translational process. At a single-channel level, both E60X- and G542X-CFTR present single-channel conductance similar to that of WT channels, and responded to high-affinity ATP-analogue or CFTR potentiator (Yeh et al., 2017), indicating the channels are likely full-length. Interestingly, we also found that the read-through reagent G418 changes the gating behavior of the read-through for E60X. Comparison of the gating parameters between the read-through products and the proposed mutations during read-through revealed insights about the molecular identity of the read-through products. These results thus underscore the highly-varied, position-dependent functional consequences of PTCs. Mechanistic and clinical implications of these results will be discussed.

MATERIALS AND METHODS

Cell culture and transfection

Chinese hamster ovary (CHO) and human embryonic kidneys (HEK) 293 cells were grown in 37°C in Dulbecco’s modified Eagle’s medium (Gibco, Waltham, MA, USA) supplemented with 10% (v/v) fetal bovine serum (Sigma-Aldrich, St. Louis, MO, USA) in a humidified atmosphere of 5% CO₂. Cells were trypsinized and transferred into 35 mm culture dishes 1 – 2 days before the transfection depending on the confluency. For patch-clamp experiments, CHO and HEK293 cells were co-transfected with pcDNA plasmids containing various CFTR constructs and pEGFP-C3 (Takara Bio Inc., Shiga, Japan) using PolyFect transfection reagent (Qiagen, Valencia, CA, USA). After transfection, cells were incubated at 27°C for 2–3 days for inside-out microscopic current experiments, and 3–6 days for inside-out macroscopic and whole-cell experiments. For Western blot studies, CHO and HEK293 cells were transfected with CFTR pcDNA using X-tremeGENE transfection reagent (Roche, Basel, Switzerland) and incubated at 37°C for 2–3 days before the cells were harvested. The reagents used to promote surface expression or read-through of CFTR variants were added to the medium to the desired concentration six hours after transfection. DMSO (0.1% vol/vol) was used as volume control.

Western blotting

Post-transfected CHO and HEK293 cells were lysed using ice-cold RIPA buffer (Santa Cruz Biotechnology, Dallas, TX, USA) containing cOmplete protease inhibitor (EDTA-free; Roche, Basel, Switzerland). Whole cell lysates were separated in 4–20% gradient gels (Bio-Rad Laboratories, Hercules, CA, USA) and transferred onto PVDF membranes (Bio-Rad Laboratories). To minimize non-specific bindings, the membranes were blocked with 5% milk in Tris-buffered saline with Tween-20 (TBST) buffer (20 mM Tris, 137 mM NaCl, 0.1% Tween-20) at room temperature for 1 hour, and then incubated with primary antibody diluted in TBST at 4°C overnight. The primary antibody and dilution are as follows: anti-CFTR monoclonal antibody Ab 596 (dilution 1:3000; CF foundation, Bethesda, MD, USA), anti-CFTR monoclonal antibody Ab 660 (dilution 1:3000; CF foundation, Bethesda, MD, USA), anti-CFTR antibody MM13-4 (dilution 1:500; MilliporeSigma, Burlington, MA, USA) and β-actin antibody AC-15 (dilution 1:1000; Santa Cruz Biotechnology).

After the incubation of the antibody, the membranes were washed with TBST and then incubated with anti-mouse IgG, HRP-linked antibody (dilution 1:5000; Cell Signaling Technology, Danvers, MA, USA) at room temperature for 1 hour. The membranes were washed with TBST and developed with SuperSignal West Pico PLUS chemiluminescence substrate (Thermo Fisher Scientific, Waltham, MA, USA). The signal was detected by Molecular Imager Chemidoc (Bio-Rad Laboratories) and quantified using ImageLab software (Bio-Rad laboratories). The intensity of CFTR bands was normalized against that of the β-actin within the same lane. Of note, whole-cell lysates were used in all the Western blot experiments except for the one shown in Figure 6A, which was done with immunoprecipitated proteins (see below).

Figure 6. Immunoprecipitation (IP) and MS experiments for WT- and E60X-CFTR.

(A) The Western blot of WT- and E60X-CFTR immunoprecipitated by CFTR antibody 596 and probed with CFTR antibody 596. (B) Peptide sequence of WT-CFTR band A and C, and E60X-CFTR LMWB and HMWB identified by MS. Dark blue regions represent the identified peptides. The horizontal axis marks the numbers of the amino acid residues in CFTR.

Glycosidase digestion

To remove the N-glycans in CFTR proteins, whole cell lysates were incubated with 500 U peptide N-glycosidase F (PNGaseF) purchased from New England Biolabs (Ipswich, MA, USA) for 3 h at 33 °C. Post-digested lysates were used for Western blot experiments as described above. The shift of the position of the band to a lower molecular weight in Western blotting suggests deglycosylation of CFTR.

Mutagenesis

CFTR mutations were constructed using QuikChange XL kit (Agilent Technologies, Santa Clara, CA, USA) following the manufacturer’s instructions. The entire coding sequences of mutated CFTR constructs were confirmed by DNA sequencing (DNA Core Facility, University of Missouri, MO, USA).

Electrophysiological recordings

For experiments in the inside-out configuration, glass pipettes were made from borosilicate glass capillaries (Kimble & Chase, Vineland, NJ, USA) using a two-step vertical micropipette puller (PP-81l; Narishige, Tokyo, Japan) and polished to a resistance of 2–4 MΩ with a home-made microforge when placed in a bath solution containing (in mM): 145 NaCl, 5 KCl, 2 MgCl₂, 1 CaCl₂, 5 glucose, 5 HEPES and 20 sucrose with pH adjusted to 7.4 using NaOH. The pipette solution contains (in mM): 140 NMDG-Cl, 2 MgCl₂, 5 CaCl₂, 10 HEPES with pH adjusted to 7.4 using NMDG. The membrane patches were excised after the seal resistance reached >40 GΩ. After the excision, the perfusion solution was changed to the one containing (in mM): 150 NMDG-Cl, 10 EGTA, 10 HEPES, 8 Tris, 2 MgCl₂, with pH adjusted to 7.4 using NMDG, and the pipette was placed close to the outlet of a three-barrel perfusion system controlled by a fast solution exchange device (SF-77B; Warner Instruments, Hamden, CT, USA) with a dead-time of ~30 ms (Tsai et al., 2009). All the single-channel traces shown in the figure are acquired after > 10 mins of phosphorylation by perfusing the solution containing 25 IU PKA and 2 mM ATP to reach a fully-phosphorylated state for the CFTR.

The pipettes used in whole-cell experiments were made as mentioned above, but without the polishing process in order to yield a resistance of 1–2 MΩ in the bath solution containing (in mM): 145 NaCl, 5 KCl, 2 MgCl₂, 1 CaCl₂, 5 glucose, 5 HEPES and 20 sucrose with pH adjusted to 7.4 using NaOH. Whole-cell pipette solution contains (in mM): 10 EGTA, 10 HEPES, 20 TEACl, 10 MgATP, 2 MgCl₂, 85 aspartate, 16 pyruvate, 5.8 glucose with pH adjusted to 7.4 using CsOH. Fast changes of the external perfusion solution were achieved using the same equipment mentioned above.

All of the electrophysiological data were recorded at room temperature with a patch-clamp amplifier (EPC10; HEKA Elektronik, Lambrecht/Pfalz, Germany). For inside-out experiments, the membrane potential (Vm) was held at −50 mV during the entire experiment unless noted otherwise, and the data were filtered online at 100 Hz with an eight-pole Bessel filter (LPF-9; Warner Instruments) and digitized to a computer at a sampling rate of 500 Hz. For whole-cell recordings, a voltage ramp of ± 100 mV over 200 ms was applied every 5 s, and the currents were filtered online at 1 kHz and digitized at 2kHz.

Reagents for electrophysiological and biochemical experiments

MgATP and PKA were purchased from Sigma-Aldrich (St. Louis, MO, USA). MgATP was dissolved into a 500 mM stock solution and stored at −20 °C. The [MgATP] was 2 mM in all inside-out experiments, unless indicated otherwise. CFTR_inh-172, provided by Dr. Robert Bridges (Rosalind Franklin University, Chicago, IL, USA) who directs the Chemical Compound Distribution Program sponsored by Cystic Fibrosis Foundation Therapeutics (Bethesda, MD, USA), was stored at −70 °C as a 5 mM stock solution in DMSO. GLPG1837 was provided by Galapagos NV (Mechelen, Belgium) and stored as a 10 mM stock at −20 °C. Lumacaftor (VX-809) was purchased from Aobious Inc. (Gloucester, MA, USA) and stored as a 10 mM stock solution in DMSO at −70 °C. G418 was purchased from Gibco (Waltham, MA, USA) and stored at −20 °C. P-dATP was custom-synthesized by Biolog Life Science Institute (Bremen, Germany) and stored as 10 mM stock at −20 °C. Forskolin and genistein were purchased from Alexis (Plymouth Meeting, PA, USA) and stored at −20 °C. Reagents used in electrophysiological experiments were freshly thawed and diluted to the desired concentration with the prefusion solution for excised inside-out or whole-cell patches and the pH was adjusted to 7.4 with NMDG.

Sample preparation and in-gel protein digestion for the mass spectrometry (MS)

The following protocols of the MS experiment are validated and done with the help from the staff scientists in the Gehrke Proteomics Center at the University of Missouri, USA. To improve the yield and purity of the CFTR proteins for MS analysis, immunoprecipitation was performed on CHO cells transfected with WT- and E60X-CFTR. CHO cells were lysed using ice-cold RIPA buffer (Santa Cruz Biotechnology) containing cOmplete protease inhibitor (EDTA-free; Roche). Cell lysates were immunoprecipitated with Dynabeads Protein G (Invitrogen, CA, USA) coated with anti-CFTR antibody 596 (CF foundation) for 10 minutes under room temperature. The immunoprecipitated protein complex was then washed using Washing buffer (Invitrogen) and eluted using Elution buffer (Invitrogen) according to the manufacturer’s protocol. The immunoprecipitated sample for each construct was divided in two groups and separated in 4–20% gradient gels (Bio-Rad Laboratories). The first group was visualized using Western blot as described before. The second group was stained with Colloidal Coomassie Brilliant Blue (Charles W Gehrke Proteomics Center, MO, USA) to identify specific protein bands. The gel band sample was minced into 1 mm³ pieces and destained with 50 mM ammonium bicarbonate and 50% acetonitrile. The in-gel protein was reduced with 10 mM dithiothreitol (DTT) in 100 mM ammonium bicarbonate for 30 minutes at 56°C and alkylated with 50 mM iodoacetamide (IAA) in 100 mM ammonium bicarbonate in the dark at room temperature for 30 minutes. Then the IAA solution was removed, and gel pieces were dehydrated by 100% acetonitrile.

Three samples were collected for MS experiments. Each sample was digested by sequencing-grade modified trypsin (V5111; Promega, WI, USA), sequencing-grade Arg-C (V1881; Promega) or mass spec-grade Lys-C (VA1170; Promega) overnight according to the manufacturer’s protocol. The digested peptides were extracted with 60% acetonitrile, 1% TFA, desalted by Pierce C18 Tips according to the manufacturer’s protocol. The desalted peptide was lyophilized and resuspended in 5% acetonitrile and 0.1% formic acid.

Mass Spectrometry (MS)

The dissolved peptides were analyzed with a Bruker timsTOF pro instrument attached to a nanoElute system LC (Bruker, MA, USA). The samples were loaded onto 300 μm interior diameter (i.d.) × 5 mm column with C8 PepMap100 (100 Å, 5 μm; Thermo Fisher Scientific) The peptides were separated by an in-house packed column of 75 μm i.d. × 20 cm with BEH C18 (130Å, 1.7 μm; Waters, MA, USA). The peptide was eluted at a flow rate of 400 nl/min with the initial gradient of 2% B (A: 0.1% formic acid in water, B: 99.9% acetonitrile, 0.1% formic acid), followed by 4.5 min ramp to 17% B, 17–25% B over 8.5min, 25–37% B over 4.5 min, 37–80% B over 2 min, and hold at 80% B for 5.5 min. Total running time was 25 min. TimsTOF pro was operated in PASEF mode. Duty cycle was locked to 100%. Ion mobility coefficients (1/K0) value was set from 0.6 to 1.6 Vs cm⁻². MS data was collected over m/z range of 100 to 1700. During MS/MS data collection, each TIMS cycle of 1.27s contained one MS and ten PASEF MS/MS scan. Exclusion was active after 0.4 min.

Data Analysis for MS experiments

Raw data were searched using PEAKS (version X; Bioinformatics Solutions Inc., Canada) with Uniprot Human protein database. Data were searched with the following parameters: trypsin protease digestion with two missed cleavage allowed, precursor ion tolerance of 50 ppm, and fragment ion tolerance of 0.1 Da. Cysteine carbamidomethylation was set as a fixed modification, and methionine oxidation, protein N-terminus acetylation and asparagine/glutamine deamidation were set as variable modifications. For the protein identification, the following criteria were used: more than or equal to 2 unique peptides per protein, and the false detection rate (FDR) for the peptide spectrum was within 1%. The possible missense mutations were searched using the SPIDER algorithm in PEAKS (Han et al., 2005). The coverage map shown in Figure 6B was plotted by the seaborn Python package (https://seaborn.pydata.org) based on the peptide sequence identified in PEAKS software (version X; Bioinformatics Solutions).

Statistical analysis

Single-channel kinetic analysis was performed using the software developed by Dr. Csanady (Csanady, 2000). Only the kinetics of membrane patches containing fewer than four opening steps were analyzed. In order to accurately estimate the number of channels in the membrane patch, reagents such as P-dATP or GLPG1837 were applied in the end of the recording to increase the P_o.

The statistical tests were performed by Igor Pro (Wavemetrics, Lake Oswego, OR, USA). Paired and unpaired Student’s t tests assuming equal variance were utilized for comparisons between two groups. For comparisons across more than two groups, one-way ANOVA followed by Turkey’s range test was used. P < 0.05 was considered statistically significant. All error bars in the figures represent the S.E.M and the data are shown as the mean ± S.E.M, with N indicating the number of experiments.

RESULTS

Chloride currents carried by a mixture of C-terminus truncated and read-through proteins in W1282X.

Our previous study (Yeh et al., 2019b) on a PTC mutation close to the C-terminus of CFTR (i.e., Q1412X) shows that a significant amount of truncated, but partially functional proteins is produced and expressed in the cell membrane. That report also echoes the functional importance of CFTR’s NBD2 (Hwang et al., 2018) since the Q1412X mutation with only the last two β-strands in NBD2 deleted exhibits severe defects in gating. As the W1282 residue is located further from the C-terminus of CFTR than Q1412, we expected that the truncated proteins of the W1412X mutation should present with gating defects.

Figure 1A shows Western blot results with W1282X-CFTR expressed in CHO cells. We also pre-incubated the cells with a CFTR corrector VX-809 (3 μM) or read-through reagent G418 (150 μM) to probe the pharmacological responses of the CFTR mutant. The effects of VX-809 and G418 on WT-CFTR have been well documented: VX-809 can increase the expression of WT-CFTR (Van Goor et al., 2011), but G418 has no effect on the expression and function of WT-CFTR (Howard et al., 1996). Hence, in this study we mainly focused on investigating the effects of these reagents on the PTC mutations. Using the CFTR antibody 596 that recognizes amino acids 1204–1211 in NBD2 (Tosoni et al., 2013) as a probe, we observed three bands representing non-glycosylated band A (mol wt ~130 kDa), core-glycosylated band B (mol wt ~150kDa) and complex-glycosylated band C (mol wt ~170 kDa) with WT-CFTR. For W1282X-CFTR, the expression pattern of CFTR bands is shifted downward as expected due to the C-terminal truncation. However, the expression of this shorter CFTR can be increased by the incubation of VX-809 (Figure 1C) as reported previously (Haggie et al., 2017). Although the full-length read-through protein was not detected even with G418 treatment (Figure 1A), patch-clamp experiments, which are more sensitive than Western blotting, provided evidence for the existence of read-through protein. The left panel of Figure 1B shows a recording of CFTR currents in an excised inside-out patch from W1282X-CFTR-expressing cells pre-incubated with 3 μM VX-809. Only a small chloride current could be activated by PKA and ATP (Figure 1B, left). However, the application of 20 μM GLPG1837, a CFTR potentiator that shares the same mechanism of action as well as the binding sites with the FDA-approved drug Ivacaftor (Yeh et al., 2017; Yeh et al., 2019a) greatly increased the current by 29.3 ± 3.4-fold (N = 5), indicating the P_o of the channel recorded is extremely low; as the maximal P_o of an ion channel is 1, the P_o of W1282X-CFTR must be < 0.034 or 1/29.3, vs. ~0.4 for WT-CFTR (Yeh et al., 2015).

Figure 1. The biochemical and biophysical properties of W1282X-CFTR.

(A) Effects of CFTR corrector VX-809 and read-through reagent G418 on W1282X-CFTR expression. DMSO treatment serves as a volume control. WT-CFTR in the same gel was used for comparison. Non-glycosylated band A (~130 kDa), core-glycosylated band B (~150 kDa) and complex-glycosylated band C (~170 kDa) of CFTR were probed with CFTR antibody 596. β-actin was probed by β-actin antibody AC-15 as a loading control. (B) Real-time macroscopic recordings of W1282X-CFTR currents with the pre-incubation of VX-809 alone (left) or VX-809 plus G418 (right). The current in both conditions can be activated by 25 IU PKA and 2 mM ATP, potentiated by 20 μM GLPG1837 and inhibited by 10 μM CFTR_inh-172 (Inh-172; Kopeikin et al., 2010). The current of W1282X-CFTR decayed in both condition upon removal of ATP in the presence of GLPG1837. The membrane potential was held at −30 mV. (C) Quantification of the expression of CFTR band C for W1282X-CFTR compared to WT-CFTR. The data from individual experiment are shown as dots. The bar charts represent the mean and error bars are the standard error of mean (S.E.M.) gathered from four independent experiments. Asterisks indicate the expression level is significantly higher than that of WT-CFTR (ANOVA followed by Turkey’s test). (D) Quantification of the percentage of ATP-sensitive current under GLPG1837 in W1282X-CFTR treated with VX-809 alone (N = 4) or VX-809 plus G418 (N = 6). The percentage is significantly higher in VX-809 plus G418 group compared with VX-809 alone (Student’s t test). *P < 0.05; **P < 0.01.

One important biophysical feature of WT-CFTR is that phosphorylated channels are opened by ATP binding-induced dimerization of the two NBDs (Vergani et al., 2005). In contrast, mutations that disrupt this process (e.g., ΔNBD2) abolish the ability of ATP to open CFTR, rendering a defective channel with a low P_o (Wang et al., 2007). Indeed, the current of W1282X-CFTR treated with VX-809 only decreased by 15.9 ± 3.2% (N = 4; Figure 1B left, D) upon the removal of ATP in the presence of GLPG1837 compared with > 90% current drop under the same condition for WT-CFTR (Yeh et al., 2017).

Since the read-through process produces full-length CFTR, which is more likely to have relatively normal ATP-sensitivity due to the functional integrity of NBD2, we speculated that the current seen in Figure 1B represent two populations of CFTR: a small fraction of read-through full-length and a large fraction of C-terminus truncated proteins. This hypothesis predicts that application of G418 may increase the proportion of ATP-dependent current in W1282X by increasing the amount of the read-through proteins. Indeed, the ATP-sensitive current was significantly increased to 32.1 ± 2.2% (N = 6) by the additional treatment with the read-through reagent G418 (Figure 1B right, D).

Different expression patterns between G542X- and E60X-CFTR.

We next characterized two other PTC mutations: G542X located at the NBD1 and E60X situated in the lasso motif of the CFTR protein (Zhang & Chen, 2016). In both cases, the truncated CFTR proteins are unlikely to serve as a functional channel since the ion permeation pathway constructed by the TMDs is likely destructed. However, read-through process is possible for both mutations, whereas only E60X has been proposed to undergo reinitiation upon translation (Sharma et al., 2018).

Figure 2 shows Western blot results with G542X- and E60X-CFTR expressed in CHO cells. Cells were pre-incubated with a CFTR corrector VX-809 or a read-through reagent G418 (DMSO treatment as a control). To ensure the protein we probed is not the C-terminus truncated product, CFTR antibody 596 recognizing NBD2 amino acids 1204–1211 (Tosoni et al., 2013) was used in the experiment. As shown in Figure 2A, three bands representing different glycosylation levels can be clearly seen with WT-CFTR. For G542X-CFTR (Figure 2A) pre-incubated with DMSO, VX-809 (3 μM), G418 (150 μM) or VX-809 plus G418, no band was detected under all tested conditions, which is consistent with previous reports that the read-through process is insufficient to yield proteins detectable by Western blot (Kalin et al., 1999; Roy et al., 2016).

Figure 2. Western blot analysis of E60X- and G542X-CFTR.

(A) Effects of CFTR corrector VX-809 and read-through reagent G418 on E60X- and G542X-CFTR. DMSO treatment serves as a volume control. WT-CFTR in the same gel was used for comparison. Non-glycosylated band A (~130 kDa), core-glycosylated band B (~150 kDa) and complex-glycosylated band C (~170 kDa) of WT-CFTR were probed with CFTR antibody 596. β-actin was probed by β-actin antibody AC-15 as a loading control. (B, C) Effects of stop codon swap on the expression of E60X-CFTR and G542X-CFTR. The reagents and antibodies used are as in (A). Of note, in (B), the signal of the bands of E60X (TGA) may seemed to be stronger than that of E60X (TAG), but there weren’t statistically different.

Surprisingly, for E60X-CFTR, two bands, one with a higher molecular weight (we will refer to this band as HMWB hereafter) that’s similar to band C in WT-CFTR and the other similar to band A (we will refer to this band as LMWB) were detected (Figure 2A). By averaging results from four independent experiments, we found that the percentage of the HMWB of E60X-CFTR compared to band C of WT-CFTR was 12 ± 5% with DMSO, 10 ± 2% with VX-809 (3 μM), 8 ± 3% with G418 (150 μM) and 9 ± 2% with VX-809 plus G418. While the expression levels of HMWB in all treatment condition are significantly smaller (P < 0.001) than that of band C of WT-CFTR, they are not different statistically among themselves (ANOVA followed by Turkey’s test). Similarly, relative to band C of WT-CFTR, the expression levels of LMWB under all tested conditions were significantly smaller: 46 ± 16% with DMSO, 34 ± 11% with VX-809 (3 μM), 34 ± 10% with G418 (150 μM) and 22 ± 6% with VX-809 plus G418. Again, there is no significant difference among these treatment groups (ANOVA followed by Turkey’s test). These two unexpected bands of E60X-CFTR were also seen when CFTR antibody 660 recognizing NBD1 amino acids 576–585 (Figure 3A, van Meegen et al., 2013) was used. Moreover, similar results were obtained in human embryonic kidney (HEK) cell line (Figure 3B).

Figure 3. Confirmation of the expression of E60X-CFTR using different antibodies and expression system.

(A) Similar experiment as Figure 2B. The E60X-CFTR proteins expressed in CHO cells were probed with anti-CFTR antibody 660. (B) The expression pattern remained upon changing the expression system to HEK293 cells. Proteins were probed by anti-CFTR antibody 596. Similar results were observed in three individual experiments.

The differences in the expression pattern between G542X- and E60X-CFTR are not due to the identities of PTC or the possible missense mutations.

There are at least two mechanisms that potentially account for the presence of these two bands in E60X-CFTR. First, the HMWB and LMWB are produced by the read-through process, yielding full-length CFTR protein possibly with the glycosylation profiles similar to the band C and A of WT-CFTR, respectively. However, the conspicuous absence of band B seems at odds with this proposition. The second possibility is that the HMWB and LMWB in E60X-CFTR are not the counterparts of band C and band A with WT-CFTR; they are generated by mechanisms other than read-through, e.g., translation reinitiation reported previously for PTC occurring at the N-terminus of CFTR (Sharma et al., 2018), and just happen to share similar molecular weights.

We first tested the hypothesis that the HMWB and LMWB we observed are the products of the read-through process. Two factors known to influence the susceptibility of stop codons to read-through may account for the different expression levels between G542X and E60X: the exact coding sequence of the stop codon (i.e., TAG in E60X and TGA in G542X), and the sequence context around the stop codon (Dabrowski et al., 2015). We therefore swapped the stop codons of E60X and G542X (Figure 2B&C). The unique expression pattern for E60X persisted when the stop codon was changed from TAG to TGA (Figure 2B). Again, the expression levels of HMWB and LMWB for E60X-CFTR with either TAG or TGA codon are lower than the band C of WT-CFTR, and the treatment failed to cause significant changes (ANOVA followed by Turkey’s test, N = 4). Furthermore, no CFTR could be detected with G542X after swapping the stop codon from TGA to TAG (Figure 2C). Thus, the different sequence of the stop codon in G542X and E60X is not accountable for the differences observed in Figure 2.

Since altering the sequence context around the PTC inevitably introduces changes in amino acid compositions in the CFTR protein, and hence other unpredictable effects on biogenesis, we decided not to further pursue the mechanism underlying the differences of E60X and G542X in Figure 2 at the DNA sequence level. Instead, we next tested if the difference between E60X and G542X is secondary to the different missense mutations they produce after the read-through (Figure 4). We are fortunate that the possible missense mutations for E60X and G542X have been identified using mass spectrometry (MS) techniques by Bedwell group at the University of Alabama-Birmingham: E60Y/Q for E60X (Xue et al., 2017), and G542R/C/W for G542X (Roy et al., 2016; Xue et al., 2017). We thus made these missense mutations and carried out Western blot experiments. Interestingly, although missense mutations E60Y/Q demonstrate various degrees of trafficking defects (i.e., lower expression of band C than WT), they show clear presence of all three CFTR bands (Figure 4A), and VX-809 can improve the biogenesis of these mutants (Figure 4A, C). These observations contradict the data with E60X where band B is absent and VX-809 bears no effect (Figure 2A). Thus, the perplexing behavior of E60X-CFTR cannot be explained by these three missense mutations tested, although mutations other than E60Y/Q remain a possibility. For G542X, robust CFTR expression was detected for G542R and G542C, but G542W did not show any expression in four independent experiments (Figure 4B). Although the expression of both G542C and G542R was increased by VX-809, only the change of G542C reached statistical significance (Figure 4D).

Figure 4. Western blot analysis for predicted missense mutations of the read-through products in E60X and G542X.

(A) Western blot for WT-, E60Y-, and E60Q-CFTR with the incubation of DMSO or VX-809, the antibodies used are the same as in Figure 1. (B) Western blot for WT-, G542C-, G542R- and G542W-CFTR. The protein bands for G542W-CFTR were hardly detectable in four independent experiments. (C) Quantification of the expression of CFTR band C for E60Y/Q-CFTR compared to WT-CFTR. The data from individual experiment are shown as dots, and the bar charts represent the mean and S.E.M. The band C expressions of E60Y/Q are significantly higher when treated with VX-809 compared with DMSO (N = 4, paired t-test). (D) Quantification of the expression of CFTR band C for G542C/R-CFTR compared to WT-CFTR. Results with G542W-CFTR were not quantified. The band C of G542C is significantly higher when treated with VX-809 compared with DMSO (N = 4, paired t-test), but the difference for G542R does not reach statistical significance. *P < 0.05; **P < 0.01; ***P < 0.001.

Two protein bands detected in E60X-CFTR may represent the products of translation reinitiation.

From the experiments above, we demonstrated that the expression pattern of E60X-CFTR is neither dependent on the identity of the stop codon, nor is it likely due to the defective biogenesis of the proposed missense mutations produced by the read-through. As the codon for E60 is located at the third exon of the cftr gene, a stop codon at this position may be subjected to translation reinitiation, in which the ribosome resume the translation at the downstream start codon, leading to the production of N-terminus truncated proteins (Sharma et al., 2018). These N-terminus truncated CFTR are expected to be glycosylated as the sites for glycosylation are further downstream from the N-terminus (Chang et al., 2008). To study the glycosylation profile of the HMWB and LMWB in E60X-CFTR, we incubated the whole cell lysate of the cells transfected with E60X-CFTR with 500U peptide N-glycosidase F (PNGase F), which will remove the N-glycans attached on the glycosylated CFTR protein, causing the CFTR band to shift toward a lower molecular weight position (Farinha et al., 2004; Glozman et al., 2009). As shown in Figure 5A, the band C of WT-CFTR was indeed shifted to a lower molecular weight position after the treatment of PNGase F. In contrast, for E60X with either TAG or TGA stop codon, it was the LMWB that could be deglycosylated (i.e., the low-molecular-weight band shifted further downward), whereas the HMWB remained stationary after the treatment (Figure 5A). This result thus suggests that the LMWB is actually the glycosylated, N-terminus truncated CFTR protein. However, the identity of the HMWB in E60X-CFTR, which seems lack of glycosylation, remains unknown.

Figure 5. The response of E60X-CFTR to PNGaseF and N-terminal antibody suggests the presence of N-terminus truncated protein.

(A) Western blot for WT, E60X with TAG and TGA codon with or without the deglycosylation by 500U PNGase F. The antibodies used were the same with previous figures. The smear-like signal for β-actin was possible due to the interaction with proteasome or other enzyme, which was inevitable during the 3 hours room temperature incubation of PNGase F even with the supplement of protease inhibitor. Similar results were consistently observed in three independent experiments. (B) Western blot of WT- and E60X-CFTR with either TAG or TGA codon probed with treatments and probing antibodies similar to that of previous figures. (C) The same batch of cells were probed with CFTR N-terminal antibody MM13-4 separately. The absence of band A in WT may be due to different sensitivity of the antibody. The results were consistently observed in three independent experiments.

To further verify that the LMWB in E60X-CFTR is indeed N-terminus truncated protein, we compared Western blots from the same batch of cells transfected with E60X-CFTR cDNA and probed separately with CFTR antibody 596 (same as previous figures, Figure 5B) and CFTR antibody MM13-4 recognizing N-terminal amino acids 25–36 (Figure 5C). Indeed, while antibody MM13-4 successfully recognized WT-CFTR, no CFTR signal was detected with E60X-CFTR (Figure 5C), supporting the idea that the bands of E60X-CFTR recognized by antibody 596 (Figure 5B) were N-terminus truncated. Of note, the first possible site for reinitiation is at position 82 (Ramalho et al., 2009), which is further downstream from the epitope for MM13-4. Collectively, these results suggest that the LMWB of E60X-CFTR is likely to be the N-terminus truncated but glycosylated CFTR protein generated through translation reinitiation instead of read-through, whereas the identity of the HMWB remains elusive.

Mass spectrometry (MS) analysis of the CFTR proteins produced in E60X-expressing cells

To validate that the LMWB is indeed an N-terminus truncated protein and investigate the identity of the HMWB, we conducted MS experiments to determine the sequence coverage of these two bands seen with E60X-CFTR. Figure 6A shows the Western blot of the immunoprecipitated protein probed with CFTR antibody 596. Probably due to the vulnerability of CFTR proteins to deglycosylation during the immunoprecipitation, only two bands in WT could be clearly detected, which are band C and band A based on the molecular weights. For E60X, both HMWB and LMWB were resolved (Figure 6A). The peptides identified in MS experiments were integrated to a coverage map shown in Figure 6B. CFTR proteins were identified by MS in all four bands seen in Figure 6A, validating the specificity of the immunoprecipitation. No peptide in the N-terminal region for LMWB before residue 105 was found, supporting the notion that LMWB represents the N-terminus truncated protein produced by translational reinitiation. Although the sequence coverages of WT-CFTR are slightly different between band A and band C, they do contain sequences of a full-length CFTR. Surprisingly, the MS pattern of HMWB of E60X-CFTR is remarkably similar to that of band C of WT-CFTR. Disappointingly, this latter result of HMWB hence failed to shed more light onto its identity (see Discussion).

Whole-cell recordings detected functional CFTR currents in E60X- and G542X-CFTR.

Although the identity of the HMWB in E60X-CFTR is unknown, the following experimental results suggest that neither HMWB nor LMWB is functional CFTR. Firstly, whole-cell patch-clamp experiments were carried out to quantify cAMP-dependent chloride conductance in cells transfected with the E60X-CFTR cDNA (Figure 7A). The application of 10 μM forskolin elicits significant currents even in some DMSO-treated cells (Figure 7A, left panel). The amplitude of forskolin-activated currents could be further increased (P < 0.05 in all condition by paired t test) with 20 μM genistein, a CFTR potentiator (Hwang et al., 1997), and subsequently inhibited by 10 μM CFTR_inh-172 (Kopeikin et al., 2010), confirming the presence of functional CFTR in the cell membrane. However, compared to WT-CFTR with a forskolin-sensitive current density of 755 ± 333 pA/pF (N = 10), the current density of DMSO-treated E60X-CFTR is only ~ 1% of that of WT-CFTR. The whole-cell current amplitude is determined by multiplying the number of functional channel N, the open probability of individual channels P_o, and the single-channel current amplitude i. Since the intensity of the bands seen in the Western blot reflects the amount of proteins, the fact that the expression levels of HMWB and LMWB are respectively ~12% and 46% of band C of WT predicts a current density of E60X-CFTR much higher than just 1% of WT-CFTR, given neither P_o or i obtained from single-channel recordings of E60X-CFTR were altered significantly (see below). Thus, the proteins in the HMWB and LMWB may be non-functional. It follows that the currents recorded in both E60X- and G542X-CFTR (Figure 7) were most likely generated by the minute amount of read-through products that is beyond the detectable level of Western blotting.

Figure 7. Whole-cell recordings of E60X- and G542X-CFTR revealed functional CFTR currents.

(A) Whole-cell recordings of E60X-CFTR with the incubation of DMSO (left), 3 μM VX-809 (middle) and 150 μM G418 (right). The currents were activated by 10 μM forskolin, potentiated by 20 μM genistein and inhibited by 10 μM CFTR_inh-172 (Inh-172). (B) Quantification of the current densities in forskolin (black) and forskolin plus genistein (blue) in cells incubated with DMSO (N = 7), VX-809 (N = 7) and G418 (N = 5). (C) Whole-cell recordings of G542X-CFTR with similar conditions to (A). (D) Quantification of the current densities in forskolin (black) and forskolin plus genistein (blue) in cells incubated with DMSO (N = 7), VX-809 (N = 13) and G418 (N = 8). Multiple comparisons were done by ANOVA followed by Turkey’s test. *P < 0.05, compared to the current density in forskolin plus genistein with DMSO incubation.

Interestingly, incubation of E60X-expressing cells with 3 μM VX-809 or 150 μM G418 significantly increases the current density measured in the presence of forskolin plus genistein, contradicting the lack of response of HMWB and LMWB to these reagents in the Western blot experiments above (Figure 2A). These inconsistencies further support the idea that the CFTR protein assayed with the patch-clamp technique represents the very few read-through, full-length E60X-CFTR channels. Indeed, although Western blot fails to show any bands for G542X-CFTR (Figure 2A), cAMP-dependent chloride currents can also be detected in some cells transfected with G542X-CFTR construct with the whole-cell patch-clamp technique (Figure 7C). The current density of G542X-expressing cells is similar to that of cells transfected with E60X-CFTR; however, the incubation of VX-809 and G418 failed to increase the current density (Figure 7D).

Single-channel recordings of E60X-CFTR suggest that the read-through products possess intact the ion permeation pore and NBDs.

As our whole-cell patch-clamp experiments demonstrated cAMP-dependent chloride currents for both E60X- and G542X-CFTR, we took advantage of this opportunity to study these channels at a single-channel level to gain further insights into the functional properties of these read-through products. We first examined the single-channel current-voltage (I-V) relationship of E60X-CFTR. Figure 8A shows a representative single-channel recording of E60X-CFTR at different membrane potentials in an excised inside-out patch from cells incubated with 3 μM VX-809. The resulting I-V relationship (Figure 8B) is similar to that of WT (Yeh et al., 2015), indicating that the permeation pathway crafted by the TMDs in E60X-CFTR is similar to that of WT channels. Similar single-channel amplitude at −50 mV holding potential were consistently seen in other single-channel traces for the read-through products for both E60X- and G542X-CFTR (Figure 9&11 below), suggesting all the read-through products recorded at the single-channel level in this study retain an intact chloride permeation pathway.

Figure 8. Single-channel I-V relationship of E60X-CFTR treated with 3 μM VX-809.

(A) Representative single-channel current traces of E60X-CFTR treated with 3 μM VX-809 at different holding potentials. The arrowheads indicate the baseline (closed state). (B) The single channel I-V plot of E60X-CFTR, where the horizontal axis is the membrane potential (Vm), and the vertical axis is the single channel amplitude (i). The squares are mean single-channel current amplitudes with the error bars representing S.E.M., the numbers of observation are indicated above the squares. The black line is the single-channel I-V relationship of E60X-CFTR acquired by linear fitting of the data, yielding a single channel conductance of 6.75 pS that is similar to that of WT-CFTR (6.91 pS; Yeh et al., 2015).

Figure 9. E60X-CFTR can be potentiated by high-affinity ATP analogue and the CFTR potentiator GLPG1837.

(A) Single-channel recordings of E60X-CFTR in ATP. The panel is comprised of five subplots arranged vertically. The upper two panels show the raw traces of E60X-CFTR from cells pre-treated with 3 μM VX-809 (black) or 150 μM G418 (blue) in 2 mM ATP. The lower three bar charts are the statistics of open probability (P_o), open time (τ_o), and closed time (τ_c) from 7 membrane patches. (B) Effects of P-dATP on E60X-CFTR. Similar organization as (A) except that the recordings were made in the presence of 20 μM P-dATP. Data were collected from 5 membrane patches for cells treated with VX-809 and 4 patches for ones treated with G418. (C) Effects of GLPG1837 on E60X-CFTR. Similar organization as (A) except that 2 mM ATP and 20 μM GLPG1837 was applied to the E60X-CFTR. Data were collected from 4 membrane patches for cells treated with either VX-809 or G418. *P < 0.05; **P < 0.01.

Figure 11. Single-channel behaviors and responses to channel modulators for G542X-CFTR read-through protein.

(A) Single-channel recordings of G542X-CFTR in the presence of ATP (2 mM). The panel is comprised of five subplots arranged vertically. The upper two panels show the representative raw traces of G542X-CFTR from cells pre-treated with 3 μM VX-809 (black) or 150 μM G418 (blue). The lower three bar charts are the statistics of open probability (P_o), open time (τ_o), and closed time (τ_c) from six membrane patches. (B) Effects of P-dATP (20 μM) on G542X-CFTR. Similar organization as (A) except that the recordings were made in the presence of P-dATP. The kinetic parameters were obtained from three membrane patches for both VX-809 and G418 group. (C) Effects of GLPG1837 on G542X-CFTR. Similar organization as (A) except that the recordings were made in the presence of 2 mM ATP and 20 μM GLPG1837. The kinetic parameters were gathered from three membrane patches for VX-809 group and six patches for G418 group *P < 0.05.

We next investigated the kinetic properties and the responses to pharmacological reagents of E60X- and G542X-CFTR. Specifically, we tested the effects of the high-affinity ATP analogue N6-phenylethyl-2’-deoxyATP (P-dATP), which works on the NBDs (Miki et al., 2010), and the CFTR potentiator GLPG1837 that acts on CFTR’s TMDs (Yeh et al., 2017; Yeh et al., 2019a) on the kinetic parameters of endogenous read-through (preincubated with 3 μM VX-809) or drug-facilitated read-through (preincubated with 150 μM G418) proteins of E60X- and G542X-CFTR.

Figure 9 shows representative single-channel recordings of E60X-CFTR from cells preincubated with VX-809 (black trace) or G418 (blue trace) with their respective responses to P-dATP and GPLG1837. For the E60X-CFTR treated with VX-809, the P_o at 2 mM ATP is 0.25 ± 0.02 (N = 7), which is around 50% of WT-CFTR mainly due to prolonged closed time constant (τ_c) of 1.14 ± 0.07 s (cf. ~0.3 s for WT-CFTR in Yeh et al., 2015). The open time constant (τ_o) of 0.39 ± 0.03 s (N = 7) is similar to that of WT-CFTR. Interestingly, for the G418-treated E60X-CFTR, the P_o in 2 mM ATP increased to 0.47 ± 0.05 (N = 7) with a longer open time constant of 0.83 ± 0.17 s (N = 7) and similar closed time constant of 1.05 ± 0.19 s (N = 7), compared to VX-809 treated E60X-CFTR. The observation that the single-channel behavior can be altered by G418 treatment further supports the notion that the CFTR proteins recorded are generated by read-through. Of note, although a previous study has shown an effect of G418 on the identity and proportion of amino acids inserted at the PTC site for G542X (Roy et al., 2016), here we presented the first piece of evidence that drug-facilitated read-through could alter the functional properties of PTC mutants of CFTR. Moreover, the effectiveness of P-dATP on the gating parameters of E60X-CFTR suggests functional (and likely structural) integrity of the NBDs.

Single-channel studies of the proposed missense mutations for E60X-CFTR

Different single-channel behaviors between VX-809- and G418-treated E60X-CFTR (Figure 9A) suggest G418 can change the identity of missense mutations during read-through. Previous study has predicted that the possible amino acids being inserted during the read-through over a TAG stop codon are tyrosine and glutamine, hence the possible missense mutations for E60X are E60Y/Q (Xue et al., 2017). We therefore introduced tyrosine and glutamine into this position and characterized single-channel properties of these missense mutations E60Y/Q (Figure 10A). While in 2 mM ATP, E60Y-CFTR show a reduced P_o ~50% of that of WT-CFTR, whereas the P_o of E60Q-CFTR is similar to WT-CFTR (Yeh et al., 2015). Multiple comparison by ANOVA followed by Turkey’s test shows that the P_o of E60X-CFTR pre-treated with VX-809 is similar to that of E60Y-CFTR, whereas E60X-CFTR read-through by G418 is similar to E60Q-CFTR (Figure 10C). This is the first functional result supporting the notion that different near-cognate tRNAs are utilized between endogenous and G418-mediated read-through processes, leading to changes in the output missense mutations.

Figure 10. Comparisons of single-channel behavior between E60X- and G542X-CFTR read-through protein and their respective possible missense mutations.

(A) Representative single-channel traces of E60Y/Q in 2 mM ATP. (B) Representative single-channel traces of G542C/R in 2 mM ATP. (C) Quantification of P_o of E60X-CFTR read-through, and that of E60Y (N = 4) and E60Q (N=3). The P_o of E60X-CFTR treated with VX-809 is similar to that of E60Y-CFTR, whereas the P_o of G418 treated E60X-CFTR is similar to that of E60Q-CFTR (ANOVA followed by Turkey’s test) (D) Quantification of P_o of G542X-CFTR read-through, and that of G542C (N = 4) and G542R (N = 5). The P_o of G542X-CFTR treated with either VX-809 or G418 is significantly lower than that of G542C- and G542R-CFTR (ANOVA followed by Turkey’s test).

Single-channel studies of G542X-CFTR read-through products

Similar to E60X-CFTR, single-channel recordings of G542X-CFTR were made possible by pre-treating the cells transfected with G542X-CFTR cDNA with either VX-809 or G418 (Figure 11). As shown in Figure 11A, both the P_o of G542X-CFTR read-through with the treatment of VX-809 (black trace) or G418 (blue trace) were reduced mainly due to prolonged closed time (τ_c) compared to that of WT-CFTR (Yeh et al., 2015). Although the closed time is longer for G418-treated G542X-CFTR, the overall P_o is similar between channels treated with VX-809 and G418. Thus, unlike E60X, G418 does not change the channel behavior of G542X-CFTR read-through proteins. However, the significant increase in P_o by P-dATP (Figure 11B) or GLPG1837 (Figure 11C) indicates intact pharmacological responses of the read-through channels despite a minor decrease of their basal function (Figure 11A).

As the predicted missense mutations at position 542 are cysteine, arginine, and tryptophan, we next investigated the single-channel behaviors of G542C/R/W (Roy et al., 2016). Although some cAMP-dependent whole-cell G542W-CFTR currents were recorded (Figure 12), we failed to observe any single-channel activity in excised inside-out patches probably because of an extremely low channel density in the cell membrane as shown in the Western blot for this mutant (Figure 4B). For both G542C- and G542R-CFTR, the P_o is surprisingly higher than those recorded in cells expressing G542X-CFTR (Figure 10B&D, multiple comparison by ANOVA followed by Turkey’s test). Interestingly, G542R-CFTR is a gain-of-function mutation with a P_o higher than that of WT-CFTR. Several gain-of-function mutations of CFTR have been reported, such as P574H and H949Y, whose open-probability is higher albeit the protein processing is defected (Sheppard et al., 1995; Seibert et al., 1996). The mechanism for the higher function of G542R awaits further investigation. Although these data fail to illuminate the identity of G542X-CFTR, it seems unlikely that G542C or G542R is accountable.

Figure 12. Whole-cell recording of G542W-CFTR.

Although the current of G542W-CFTR couldn’t be recorded using the inside-out configuration, when applying 10 μM foskolin in the whole-cell configuration, miniscule current could be activated with an average current density of 8.51 ± 1.67 pA/pF (N = 6). The current density can be further increased by 20 μM genistein with an averaged fold increase of 6.7 ± 2.7 (N = 6) and inhibited by 10 μM CFTR_inh-172 (inh-172).

DISCUSSION

In the present study, we employed a spectrum of biochemical and biophysical techniques to investigate the functional consequences of premature termination codons (PTCs) at different locations of CFTR. Three Class I CF-inflicting PTC mutations W1282X-, G542X- and E60X-CFTR were characterized, and the heterogenous effects of the PTCs were unveiled. Here we will first take advantage of the comprehensive information in CF pathogenesis as well as CFTR structure/function understanding to offer an overall molecular insight into the positional effects of PTCs presented in the current study. We will then use this insight to synthesize therapeutic strategies that are tailored to individual PTC mutations.

PTCs in the cftr gene cause deficits at multiple levels. At the mRNA level, the transcripts harboring a PTC are usually eliminated by nonsense-mediated mRNA decay (NMD), a cellular surveillance mechanism for quality control (Fatscher et al., 2015). Even if sufficient mRNA is produced, during the translation process a premature cessation or reinitiation of the translation may result in the expression of proteins with varied degrees of truncation depending on the position of the PTC (Zhang & Maquat, 1997; Thermann et al., 1998). Although endogenous read-through does happen, its success rate is low so that the production of a full-length CFTR protein is limited (Gesteland & Atkins, 1996; Howard et al., 2004; Welch et al., 2007; Du et al., 2008). Moreover, the function of the CFTR read-through products is highly variable due to the fact that the read-through frequency and the possible missense mutations differ with the identity of the stop codon, the structural context, and the specific read-through mechanisms (i.e., endogenous or drug-facilitated read-through) (Roy et al., 2016; Xue et al., 2017). Hence, it is difficult if not impossible to predict the consequences of PTC mutations with a fixed set of rules.

These multi-faceted defects associated with PTC mutations present a great challenge to the development of effective treatments for patients carrying PTC mutations in CF or other diseases. Several studies, as well as ongoing clinical trials, have shown the potential of utilizing CFTR modulators such as potentiators (e.g., Ivacaftor) and correctors (e.g., Lumacaftor and Tezacaftor) to enhance the function of the truncated protein. However, it is unclear if these CFTR modulators that are now clinical medicines can be universally applied to PTCs since the relationship between the position of the PTC and the function of the truncated protein has yet to be established (Wang et al., 2007; Haggie et al., 2017; Mutyam et al., 2017; Xue et al., 2017; Birmingham, 2018; Yeh et al., 2019b). In theory, reagents that can promote read-through have the greatest potential for overcoming PTCs, and indeed compounds such as Ataluren and G418 have been shown to increase the expression of read-through CFTR in vitro (Howard et al., 1996; Roy et al., 2016) and in vivo (Wilschanski et al., 2003; Du et al., 2008). However, a clinically-viable drug candidate has yet to be identified (Wilschanski et al., 2000; Wilschanski et al., 2003; Kerem et al., 2008; Wilschanski et al., 2011; Wilschanski, 2012; Kerem et al., 2014). Thus, a better understanding of the basic biology of PTCs in CFTR holds the key for the development of effective therapeutic strategies to outrun PTC mutations in CF.

In the current report, we show variable biophysical/biochemical consequences of PTCs that are position-dependent. For the mutations with the PTCs located at the more C-terminal region of the protein (e.g., W1282X-CFTR), the C-terminus truncated protein due to the premature cessation of translation is partially functional, which is an unsurprising observation since the CFTR protein with the removal of the entire NBD2 plus the C-terminus (i.e., ΔNBD2) can work as a chloride channel (Wang et al., 2007; Yeh et al., 2015). However, as shown in Figure 1B, the open probability of W1282X-CFTR is low as a large portion of NBD2 is removed. Indeed, our previous study demonstrated detrimental effects on CFTR gating once the C-terminal deletion exceeds the last two beta-strands of NBD2 (Yeh et al., 2019b). The ATP-washout experiment in Figure 1B suggests the presence of read-through proteins in W1282X, although the majority of the chloride current is carried by the C-terminus truncated protein. Structurally, the deletion caused by the W1282X mutation results in the loss of important motifs responsible for ATP-binding, namely the ABC signature sequence (a.a. 1346–1350), which partners with ATP-binding motifs in NBD1, and the Walker B motif in NBD2 that participates in coordinating Mg²⁺ as well as the phosphate group in ATP. Thus, the C-terminus truncated protein out of W1282X is unlikely to response to ATP (Hwang & Sheppard, 2009; Hwang et al., 2018). On the other hand, the tiny fraction of ATP-sensitive currents observed in W1282X-CFTR is contributed by the full-length read-through proteins. This interpretation is further supported by the result that G418 significantly increases the ATP-dependent component of the currents (Figure 1B&D).

Contrary to PTCs at the C-terminus of the CFTR protein, G542X with the PTC located in the middle of the CFTR protein shows little evidence for any channel function out of a largely truncated CFTR at position 542 (cf. Roy et al., 2016). Of note, Cormet-Boyaka et al. (2009) reported that the CFTR protein with only a.a. 1–633 is nonfunctional, hence the possibility that the truncated protein of G542X-CFTR composed of a.a. 1–541 is functional is minimal. Indeed, the recorded G542X-CFTR channels (Figure 11) exhibit properties very similar to those of WT-CFTR including single-channel conductance and responses to the high-affinity ATP analog P-dATP or GLPG1837, supporting the notion that these channels represent the full-length read-through proteins with intact TMDs and NBDs.

As E60X is located in the third exon of the cftr gene, it is possible for the ribosomal complex to skip the PTC and reinitiate the translation at downstream start codons, generating N-terminus truncated proteins (Zhang & Maquat, 1997; Sharma et al., 2018). Indeed, both Western blot (Figure 2A and Figure 5C) and MS (Figure 6) results support the notion that the LMWB, which is subjected to PNGase F digestion (Figure 5A) is likely to be the CFTR protein with N-terminal truncation. For HMWB, the Western blot data (Figure 5) indicate that the protein is unglycosylated and N-terminus truncated, but peptides in the N-terminus of CFTR were still detected by the MS (Figure 6). This discrepancy may be due to the presence of minuscule amount of full-length read-through proteins of E60X-CFTR which are included in the sample of HMWB (see Figure 4A). Thus, the exact identify of the HMWB remains elusive, and awaits further studies.

Regardless of the exact identity of HMWB in E60X-CFTR, we conclude that neither HMWB nor LMWB proteins are functional based on following arguments. Functionally, as the open probability (Figure 9) and the single-channel current amplitude (Figure 8) of E60X-CFTR are similar to those of WT, the relative expression level of either LMWB or HMWB (Figure 2A) is too high to explain a macroscopic current density of 1% of WT (Figure 7). Indeed, Carroll et al. (1995) reported that the function of the translationally reinitiated CFTR is severely impaired. From the structural point of view, the possible sites for reinitiation include M82, M150, M152, and M156, which result in N-terminal truncations extending to the TM1 of CFTR (Carroll et al., 1995; Liu et al., 2017). However, numerous studies have established that TM1 (the first transmembrane segment) contributes to the formation of the anion permeation pathway for CFTR (Wang et al., 2011; Gao et al., 2013; Liu et al., 2017; Zhang et al., 2018). Thus, despite the presence of abundant N-terminus truncated CFTR, these proteins are unlike to serve as a chloride channel; the channel activity recorded in both whole-cell (Figure 7) and excised patches (Figure 9) represents read-through, full-length CFTR.

Our studies of PTCs at different locations of CFTR not only unveil a broad variety of biophysical and biochemical properties for the end products of PTCs, but also bear clinical implications on the development of therapeutic strategies targeting PTC mutations in CF and beyond. Together with our previous report (Yeh et al., 2019b), we first conclude that not all PTC mutations are created equal. Many of the PTC mutations (e.g., W1282X and Q1412X) close to the C-terminus of CFTR can retain residual function due to the presence of the TMDs constructing the pore. Although we don’t have a definite answer regarding the shortest functional C-terminus truncated CFTR, since TM12 is an important pore-forming segment (Bai et al., 2011; Qian et al., 2011; Gao & Hwang, 2016; Liu et al., 2017), the boundary should lie in the position close to 1172 the last amino acid defining TM12 (Liu et al., 2017). For the PTC mutations yielding functional C-terminus truncated protein, the application of potentiators (e.g., VX-770) and correctors (e.g., VX-809 and VX-661), both now available in clinics, could enhance the function and the expression of the protein (Yeh et al., 2015; Wang et al., 2016; Haggie et al., 2017; Mutyam et al., 2017; Birmingham, 2018; Yeh et al., 2019b). Aside from these FDA-approved CFTR modulators, future development of novel CFTR stabilizer (e.g., N91115) may further benefit patients carrying these PTCs as some of the C-terminus truncated CFTR variants show accelerated membrane turnover (Haardt et al., 1999; Donaldson et al., 2017).

It is however more challenging to overcome the clinical hurdle with PTCs located proximal to position 1170 (e.g., G542X and E60X) because more drastic changes of the CFTR pore structure (and hence function) are expected for the truncated proteins. Our data suggest that for these PTC mutations, enhancing the read-through process may be the major pathway for generating meaningful CFTR activity. Thus, aside from waiting for the eventual success of gene editing, developing effective read-through reagents should be a priority in drug development. The observation that the single-channel function of E60X-CFTR is better when the read-through is promoted by G418 (Figure 9) is encouraging. The demonstration that the G542R mutation enhances gating function of CFTR also suggests a promising outcome for the development of read-through reagents that can bias the choice of amino acids to optimize the function of the read-through products.

In summary, by detailed characterization of biochemical and biophysical defects in CFTR that are associated with PTC mutations, we show that the multi-faceted functional consequences of PTCs at least in part depend on the location of the mutations. These highly-varied functional perturbations also demand effective therapeutic strategies tailored to individual PTC mutation. We believe that the principles elucidated from our work, in combination with other emerging approaches that are not within the scope of the current study, such as inhibiting the NMD by antisense oligonucleotide (Huang et al., 2018; Keenan et al., 2019), will shed light on future development of effective treatments for CF or other genetic diseases bearing nonsense mutations.

Key points.

• Biochemical and biophysical characterizations of three nonsense mutations of Cystic Fibrosis Transmembrane conductance Regulator (CFTR) associated with severe form of cystic fibrosis (CF) reveal the importance and heterogenous effects of the position of the premature termination codon (PTC) on the CFTR protein function.

• Electrophysiological studies of W1282X-CFTR, whose PTC is closer to the C-terminus of CFTR, suggest the presence of both C-terminus truncated CFTR proteins that are poorly functional and read-through, full-length products.

• For G542X- and E60X-CFTR, the only mechanism capable of generating functional proteins is the read-through, but the outcome of read-through products is highly variable depending on the interplay between the missense mutation caused by the read-through and the structural context of the protein.

• Pharmacological studies of these three PTCs with various CFTR modulators suggest position-dependent therapeutic strategies for these disease-inflicting mutations.