|Home | About | Journals | Submit | Contact Us | Français|
Background/Aims: Mutations in the CFTR gene cause Cystic Fibrosis (CF) the most common life-threatening autosomal recessive disease affecting Caucasians. We identified a CFTR mutation (c.120del23) abolishing the normal translation initiation codon, which occurs in two Portuguese CF patients. This study aims at functionally characterizing the effect of this novel mutation. Methods: RNA and protein techniques were applied to both native tissues from CF patients and recombinant cells expressing CFTR constructs to determine whether c.120del23 allows CFTR protein production through usage of alternative internal codons, and to characterize the putative truncated CFTR form(s). Results: Our data show that two shorter forms of CFTR protein are produced when the initiation translation codon is deleted indicating usage of internal initiation codons. The N-truncated CFTR generated by this mutation has decreased stability, very low processing efficiency, and drastically reduced function. Analysis of mutants of four methionine codons downstream to M1 (M82, M150, M152, M156) revealed that each of the codons M150/M152/M156 (exon 4) can mediate CFTR alternative translation. Conclusions: The CFTR N-terminus has an important role in avoiding CFTR turnover and in rendering effective its plasma membrane traffic. These data correlate well with the severe clinical phenotype of CF patients bearing the c.120del23 mutation.
Mutations in the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR, MIM# 602421) gene cause Cystic Fibrosis (CF, MIM# 219700) the most common life-threatening autosomal recessive disease among Caucasians . The major CF symptoms are progressive pulmonary dysfunction, pancreatic insufficiency and elevated sweat electrolyte concentration . CFTR protein, a Cl- channel in the apical membrane of epithelial cells comprises several domains, namely: two membrane-spanning domains (MSD1/2), two nucleotide binding domains (NBD1/2) and a regulatory domain (RD), besides the N- and C-termini which are both also functionally relevant [2, 3].
Although ~1,600 variants of the CFTR gene have been identified , F508del, the most frequent mutation (~70% of CF chromosomes worldwide) is associated with a severe clinical phenotype. Most F508del-CFTR is retained and degraded at the level of the endoplasmic reticulum (ER) . In Southern Europe the incidence of F508del is lower, in Portugal it only accounts for ~60% of CF chromosomes . Hence, there is a higher prevalence of non-F508del alleles among Portuguese CF patients and this necessitates close assessment of other predicted mutations to determine their functional consequences at the cellular and physiological levels. The functional characterization of such naturally occurring CFTR mutations is of high relevance for genetic counselling and clinical prognosis, not just for such gene variants, but also for mutations in other disease-associated genes which perturb the same molecular mechanism.
Here, we describe a novel CFTR mutation in exon 1, c.120del23, which we reported previously , that was found in two CF patients from the Azores islands, both with F508del in the other allele. As deletion of these 23 nucleotides (120-142) removes the translation initiation codon (133-135), we asked whether this mutation still allows for synthesis of CFTR protein through usage of alternative internal initiation codon(s), as shown for other CFTR variants . We demonstrate that truncated CFTR proteins are produced and show that M150/M152/M156 (but not M82) can mediate this alternative translation initiation. By functionally characterizing these N-truncated protein(s) we show that at least one of these reaches the cell membrane, but it is unstable and has drastically reduced Cl- channel activity. These findings correlate well with the severe clinical phenotype of the two patients bearing this mutation.
The 27 CFTR exons, intronic flanking regions, 5’ and 3’ UTRs and [TG]mTn were analysed by sequencing as described elsewhere .
Following approval of this study by the hospital ethical committee and written informed consent from the parents, nasal epithelial cells from one of the CF patients with the F508del/ c.120del23 genotype were obtained. RNA extraction and cDNA synthesis were performed as described before . Briefly, quantitative (log-phase) RT-PCR analysis of CFTR transcripts from nasal epithelial cells was performed in the region of exons 8-10, using a FAM-labelled primer . After separation in the automatic sequencer, quantitative analysis of RT-PCR products was performed with the GeneScan® software as before .
The CFTR mRNA sequence (GenBank: M28668, position 1-4575) was assessed for in-frame AUG codons as possible alternative initiation codons using the “AUG evaluator” and Open-Reading-Frame-Finder online software .
By site-directed mutagenesis (QuikChange, Stratagene, La Jolla, CA, USA) c.120del23 was introduced into the wt-CFTR-cDNA in the pNUT mammalian expression vector [12, 13] using as primer the patient's RT-PCR product (positions 96-265). The exon-1-lacking-CFTR-cDNA-construct for in vitro translation resulted from cloning (CFTR-cDNA, exons 2-24) into the pSP73 vector. Other CFTR mutants were produced by mutagenesis on c.120del23-CFTR-cDNA-pNUT or on the exon2-24-CFTR-cDNA-pSP73 (primers shown in Table Table11).
The c.120del23-CFTR cDNA pNUT plasmid and its methionine mutants were used (2 µg) to produce novel stable baby hamster kidney (BHK) cell lines by transfection with lipofectin (Invitrogen, Paisley, UK) according to the manufacturer's instructions. The novel cell lines stably expressing these constructs were cultivated under metothrexate (500 µM) selection as previously described [14, 15].
CFTR protein expression assessment by Western blot (WB) was performed as described  using the anti-CFTR monoclonal antibody (Ab) M3A7 (Millipore, Hampshire, UK) and the SuperSignal® West Pico chemiluminescent substrate system (Thermo Scientific, Rockford, IL, USA). Metabolic pulse-labelling of cells, CFTR immunoprecipitation (IP), electrophoresis, fluorography, and densitometry were also performed as described before .
Immunofluorescence was also performed as previously . Briefly, CFTR-transfected BHK cells grown on chamber-slides (Nalgene Nunc, Roskilde, Denmark) were fixed in methanol, permeabilized and incubated for 1h with 24-1 monoclonal anti-CFTR antibody (R&D Systems, Minneapolis, MN, USA) and after washing, incubated with FITC-secondary antibody (Amersham Biosciences Corp, Piscataway, NJ, USA) and mounted in Vectashield (Sigma-Aldrich Corp, St. Louis, MO, USA) with DAPI (blue) for nuclear staining. Preparations were observed in a Zeiss Axioskop fluorescence microscope and microphotographs obtained with a Photometrics image point cooled CCD video camera.
For the functional assessment of CFTR (the c.120del23, F508del mutants and wt as a control) the CFTR-mediated iodide (I-) efflux was measured at room temperature, as previously  using an I--selective electrode (ThermoElectron Corporation, Waltham, MA) and forskolin (10 µM), genistein (50 µM) as CFTR cAMP agonist and potentiator, respectively. Data are expressed as means ± SEM (standard error of the mean).
Statistical analyses were performed using Student's t test, with a value of p<0.05 considered statistically significant.
For in vitro transcription and translation, the TnT Coupled Reticulocyte Lysate System (Promega, Madison, WI, USA) was used with [35S]methionine, according to the manufacturers’ instructions. SDS-PAGE (3%/6%) separation of protein products was followed by autoradiography.
A Portuguese female patient from the Azores islands was diagnosed with CF at the age of 1 month by accepted criteria , including significant pulmonary involvement (bronchiectasis and bronchial obstruction), pancreatic insufficiency (PI) and elevated Cl- in sweat (72 mmol/L, mean of two independent measurements); maxillary and ethmoid sinusitis; and suspicion of (but not confirmed) meconium ileus. As only one mutation (F508del) was identified by routine CFTR genotyping, we sequenced all CFTR exons and flanking intronic regions and found c.120del23 mutation (CFTR mRNA numbering as in GenBank: M28668). This mutation which had not been reported previously, except by ourselves  is localized in exon 1 and deletes 23 nucleotides (position 120 onwards) including the translation initiation codon (Fig.(Fig.1a).1a). This patient was previously analysed for the cAMP-mediated chloride (Cl-) secretion in the colon and no secretion was detected . Subsequently, another unrelated Portuguese CF male patient, also with a severe clinical involvement (diagnosed at the age of 11 years; PI; 119 mEq/L Cl- in sweat; mild bronchiectasis and lung colonization by Pseudomonas aeruginosa) was identified with the same genotype. To understand the disease-causing mechanisms of this novel mutation and to establish a correlation with the clinical phenotype, we studied its molecular, cellular and functional consequences.
To evaluate whether c.120del23 alters the levels of CFTR transcripts, we quantified RNA from nasal epithelial cells from one of the two above c.120del23/F508del patients. To this end, we applied a previously described quantitative (q) RT-PCR strategy (see Methods) which allows allele-specific quantification of CFTR transcripts when F508del is present in one allele and also enables quantification of exon 9 skipping . The latter occurs to a lesser or greater extent depending on the intron 8 polymorphism [TG]mTn . This qRT-PCR approach enables detection of 4 different PCR products which in this patient are (Fig.(Fig.1B):1B): the two exon 9+ transcripts from c.120del23 (391 nt) and F508del (381 nt) alleles, both at the far right of the electropherogram; and the two exon 9- transcripts from c.120del23 (208 nt) and F508del (205 nt) alleles, at the far left. Fig.Fig.1C1C summarizes the quantification of these data (n =3) represented as the percentage of total CFTR transcripts from each allele (F508del, left; c.120del23, right) where values for exon 9+ and exon 9- transcripts are shown in black and in white, respectively. These results show that although levels of full-length (exon 9+) c.120del23 transcripts are lower than those from the F508del allele, this is because skipping of exon 9 (exon 9- transcripts) is higher for c.120del23- than for F508del-transcripts. However, this enhanced exon skipping in c.120del23 transcripts is plausibly not induced by the c.120del23 mutation per se, but rather by the [TG]11T7 polymorphism, which we found to be in cis with c.120del23 (data not shown), and which is described to cause more exon 9 skipping than the [TG]10T9 polymorphism which is in cis with F508del . It should nevertheless be noted that the difference between the total levels of F508del and c.120del23-transcripts was not found to be statistically significant.
Next, we used “AUG evaluator”  to identify downstream AUG codons in CFTR mRNA (positions 1-4575 of GenBank: M28668) that could be used as alternative initiation codons according to the Kozak mRNA scanning model. This analysis revealed 82 AUGs of which only 38 are in-frame with M1 (Table (Table2).2). It also observed that all the out-of-frame AUGs identified would result in proteins smaller than in-frame CFTR (the largest having only 51 amino acids (aa)), due to the presence of premature termination codons (PTCs) (data not shown). The scores of the first four in-frame methionines downstream of M1 (M82, M150, M152 and M156) are shown in Table Table2.2. Interestingly, all four have lower scores than M1, with the lowest (i.e, the methionine with the lowest probability of initiating CFTR translation) being M82 (exon 3). The scores of the other three methionines (M150, M152 and M156 all in exon 4) are very close to each other (Table (Table2)2) and truncated CFTR initiated at any of them would lack TM1 and TM2 (of MSD1), and the N-terminus. Sequence alignment of CFTR protein from several species  demonstrates that M1 and M150 are fully conserved through evolution and that M82 is the least conserved of the four downstream methionines.
To determine whether c.120del23-CFTR mRNA uses any of these alternative codons in vivo, we produced a stable BHK cell line expressing the c.120del23-CFTR-cDNA-pNUT construct. Western blot analysis (Fig.(Fig.2A)2A) shows that this construct generates two CFTR-specific products (lane 2, Fig.Fig.2A,2A, arrows D and E of ~133 and ~128 kDa, respectively), both of lower molecular mass than the immature ER form (known as band B ~140 kDa) of wt-CFTR (lane 1, Fig.Fig.2A).2A). The slower migrating of these two forms (D) predominates. Data in Fig.Fig.2A2A also indicate that expression levels of the mutant forms (lane 2) at steady-state are lower than those of wt-CFTR (lane 1). This quantitative difference is not the result of clonal variation in expression levels because lower expression levels were consistently observed for multiple c.120del23-CFTR clones in comparison to wt-CFTR clones (data not shown). These low levels may result from either decreased translation efficiency of the mutant protein, higher turnover or both.
To distinguish between these possibilities, we analysed the turnover rate of c.120del23-CFTR by pulse-chase experiments followed by IP (Fig.(Fig.2B).2B). The graph in Fig.Fig.2C2C shows that the turnover rate of immature c.120del23-CFTR is higher than that of immature wt-CFTR. Results in Fig.Fig.2B2B also show that c.120del23-CFTR is very inefficiently processed, as only a very faint band corresponding to its fully-glycosylated form is detected in comparison to that of wt-CFTR for which, after a 0.5h-chase, the fully-glycosylated form (band C) can already be observed. However, these data also suggest that a small fraction of the c.120del23-CFTR may traffic to the cell membrane.
To confirm that c.120del23-CFTR is present at the plasma membrane, we performed immunocytochemistry. Data in Fig.Fig.3,3, show that, whereas wt-CFTR is mostly detected at the membrane, c.120del23-CFTR (Fig.(Fig.3C)3C) evidences a predominant intracellular staining, although with a more widespread distribution throughout the cytoplasm than F508del-CFTR (Fig.(Fig.3B),3B), but also with some faint staining appearing at the plasma membrane (arrows in Fig.Fig.3C).3C). To further confirm the presence of c.120del23-CFTR at the cell membrane, we assessed its activity as a Cl- channel by the iodide efflux technique  which assesses the function of a population of CFTR Cl- channels in intact cells (see Methods). When cells expressing c.120del23-CFTR were treated with forskolin (FSK)/ genistein (Gen), a detectable activity peak of iodide efflux was elicited (Fig. (Fig.4A,4A, open squares) which is significantly lower (~25%, Fig.Fig.4B)4B) than that of wt-CFTR (Fig. (Fig.4A4A black circles) but also significantly higher (~20%, Fig.Fig.4B)4B) than that of F508del-CFTR (Fig.(Fig.4A,4A, grey triangles). Moreover, a delay of ~3 min in the peak response of 120del23-CFTR to agonists is also observed in comparison to wt-CFTR (Fig. (Fig.4A).4A). Nevertheless, these data confirm that some c.120del23-CFTR is present at the cell membrane, albeit at very low levels.
To determine which of the alternative AUG codons (Table (Table2)2) is used to initiate translation of the c.120del23-CFTR mRNA, resulting in the protein with the above-described properties, the four methionine codons (M82, M150, M152, M156) downstream of M1 (translation is a 5’->3’ mechanism) were individually or jointly mutated into valines and the respective stable cells were analysed by immunoblot for CFTR. Results in Fig.Fig.5A5A reveal the presence of two proteins (D and E) for single mutants of M82V, M150V, M152V and M156V (lanes 3-6, respectively) and also for the double mutants M150V/ M152, M150V/M156V and M152V/M156V (lanes 7-9). These two forms exhibit similar mobility to those detected for c.120del23-CFTR (lane 2). The upper of these (D ~133 kDa) is also observed when the wt-CFTR construct is analysed (lane 1) so CFTR alternative translation also occurs when M1 is present at the sequence. However, when the triple mutant (M150V/M152V/M156V) was analysed (lane 10, Fig.Fig.5A),5A), only the lower form (E ~128 kDa) could be detected. The failure to detect the larger form (arrow D) for this variant demonstrates that it corresponds to usage of M150, M152 or M156 (lane 10, Fig.Fig.5A).5A). The shorter CFTR protein (arrow E) observed when the M150/M152 and M156 are simultaneously mutated probably results from usage of an AUG codon downstream of M156. In exon 6a four additional AUG codons (Table (Table2)2) are observed which can be used to initiate translation of this shorter CFTR version.
To confirm these data by in vitro translation (IVT), we used a cDNA construct lacking M1 (containing only CFTR exons 2-24) in the pSP73 vector and generated mutations in the same methionine residues. The cDNA CFTR exon 2-24 pSP73 construct mimics the alternative splicing CFTR transcripts lacking exon1 observed during the human and sheep lung development . The IVT data shown in Fig.Fig.5B5B demonstrates that a number of smaller protein products were generated from the CFTR exon 2-24-pSP73 construct, including two promi- nent proteins (lane 2, arrows D and E), likely corresponding to the same species observed in vivo. The full-length immature form of wt-CFTR migrates at about 140 kDa (lane 1, arrow A, left). Additional products of CFTR exon 2-24-pSP73 construct IVT are observed (Fig.(Fig.5B),5B), likely resulting from usage of AUGs that are not used in vivo (Fig.(Fig.5A)5A) due to a more rigorous quality control of translation or from premature termination events. Individual and double mutations of M82V, M150V and M152V (lanes 5-8) did not cause loss of either protein species D or E, consistent with the corresponding constructs in the in vivo assay. The mutant M82V/M150V/M152V (lane 4) does not alter the production of proteins D and E either. However, the simultaneous deletion of all four methionines (M82, M150, M152 and M156) abolishes the production of the upper migrating protein at D (lane 3). Thus, this species likely corresponds to the CFTR variant resulting from usage of M150/M152 or M156 (and not of M82 because the M82V mutation alone did not abolish the appearance of this band).
Our aim here was to assess the functional impact at the molecular and cellular levels of the c.120del23 mutation, a novel mutation in the CFTR gene which we describe here for the first time in two Portuguese CF patients in compound heterozygosity with F508del.
We first assessed the total levels of c.120del23 mRNA in native nasal cells from one of the above CF patients and found them to be at equivalent levels to F508del transcripts. This suggested that the mutation must generate CFTR protein(s), as translation usually precludes mRNA from degradation . Moreover, our bioinformatic analysis indicates that all out-of-frame protein products (i.e., those resulting in open-reading-frames other than CFTR) bear premature stop codons (PTCs, data not shown). Hence, if these out-of-frame proteins were produced in vivo, significant degradation of the c.120del23-CFTR transcripts by nonsense-mediated mRNA decay (NMD) would be expected . Our c.120del23 mRNA quantitative data thus predicts that out-of-frame proteins must not be produced at all in vivo.
Our data also reveal that higher levels of exon 9 skipping occur for c.120del23 transcripts than for F508del transcripts. However, these cannot be directly related to c.120del23 but rather to the [TG]11T7 polymorphism which in cis with the c.120del23-allele (not shown). Indeed, this polymorphism is known to induce higher levels of exon 9 skipping  than [TG]10T9 which is in cis with the F508del-allele.
To determine whether downstream in-frame AUG codons can be used to initiate CFTR translation in the absence of M1 in the c.120del23 construct, we performed WB analysis. Two CFTR alternative species (D and E, ~133 and ~128 kDa, respectively) with lower molecular mass than wt band B are detected, thus demonstrating usage of alternative initiation codons, consistently with previous studies with other CFTR variants . These truncated forms of CFTR were consistently found at lower expression levels than wt-CFTR, possibly reflecting their lower efficacy in translation initiation , as predicted from the respective in silico efficacy scores.
An additional explanation for the low expression levels of c.120del23-CFTR at steady-state could be a faster degradation of the protein, as recently described for the Δ264 N-truncated CFTR . Moreover, by WB only core-glycosylated forms of the c.120del23 were detected which also suggests low processing efficiency. Indeed, our results from metabolic pulse-labelling show that immature c.120del23-CFTR turns over faster than immature wt-CFTR (band B). This is an interesting finding since immature F508del-CFTR evidences no significant reduction in synthesis  or turnover [31, 32] relative to immature wt-CFTR. Since these experiments also show that maturation efficiency of c.120del23-CFTR is very low (mature c.120del23-CFTR form is very faint), the high turnover observed probably results from enhanced degradation, as indicated by previous studies [29, 33]. These reduced levels of processed c.120del23-CFTR can also result from reduced stability at the cell surface, as it was proposed that CFTR N-terminus plays a role in endocytosis through protein interactions, namely with cytoskeletal proteins [34, 35].
Altogether, these results for c.120del23-CFTR suggest an important role of the N-terminus in CFTR folding, stability and processing, which was also evidenced by other studies demonstrating that point mutations in this region - S50P; E60K; G85E/V; E92K - prevent CFTR maturation [35, 36]. Notwithstanding, since c.120del23-CFTR completely lacks the N-terminus it thus misses the arginine-framed-tripeptide (AFT) motif at residues 29-31 acting as a retention/retrieval signal for CFTR at the ER exit checkpoint . The absence of this AFT would be predicted to enhance ER-to-Golgi transport of c.120del23-CFTR and this may actually account for the detection of the truncated protein at the plasma membrane.
Next, we investigated which of the methionines with the highest probability of initiating translation (M82; M150; M152; and M156) originate the truncated forms CFTR of proteins detected here. Our in vivo and in vitro data indicate that M150, M152 and/or M156 may all be used to produce the major N-truncated CFTR protein (~133 kDa). It is not possible to determine which one is preferentially used, since in single-mutant experiments they appear as equivalent. In contrast, both in vivo and in vitro data demonstrate that M82 (the one with the lowest score and also the least conserved of the four), is not used to initiate CFTR translation. Since M150 has the highest score of the four and it is also the most conserved one, we predict that it is the codon which is selected in vivo to initiate translation when M1 is missing (Fig.(Fig.5A,5A, lane 7 and 8). However, our data show that M152 and M156 can similarly be used to produce the form D (~133 kDa), at least when M150 is missing (Fig.(Fig.5A5A lanes7-8). Although the most crucial nucleotides for AUG context are purine at −3 and guanine at +4 [28, 37], none of the CFTR AUG codons (including M1) analysed here has the latter. The fact that all three M150, M152 and M156 can be used to initiate alternative translation of CFTR, may result from a secondary structure on the mRNA at this region that enables the ribosome to recognize any one of them as initiation codon . Indeed, these codons are used even in the presence of the M1 although at lower efficiency. Moreover, the fact that another truncated CFTR, of even greater mobility on SDS/PAGE (~128 kDa), is also produced when these three methionines are mutated, suggests the existence of another initiation codon further downstream of M156. Four additional AUG codons (in exon 6a) could initiate translation of this shorter CFTR protein, the first being M212. Other CFTR mutations reported in CF patients  that abolish M1 (M1V; M1K; M1T) will probably result in the production of the same N-truncated CFTR proteins as c.120del23 mutation. Moreover, it was previously showed that a transcript lacking exon 1, as a result of the recruitment of alternative 5’ exons joined to exon 2, shows developmental regulation in the human and sheep lung . This form also initiates at the methionines in exon 4  suggesting a common function for this N-truncated protein.
Despite its intrinsic instability and the very low membrane levels of this mutant, we find here that c.120del23-CFTR, still has residual Cl- channel activity, in contrast to other CF-causing mutants like F508del-CFTR (this study) and R560T-, or A561E-CFTR [14, 15]. This may reflect a smaller impact on the CFTR Cl- channel activity by the most abundant N-terminal truncation product resulting from c.120del23-CFTR (i.e., M150/M152/M156 which remove TM1; TM2) in comparison to the above NBD1 mutations. Consistently, it was previously reported in Xenopus oocytes  that segments TM1-TM4 are not essential components of the conduction pore or the selectivity filter of CFTR, although their removal causes lower conductance of the channel. However, in another study by Ge et al , segments M1 and M6 are defined as major contributors to the CFTR channel pore and Sheppard et al  demonstrated that the first half of CFTR forms a regulated Cl- channel with conduction properties similar to those of wt-CFTR.
Interestingly, recordings of c.120del23-CFTR by single-channel analysis were of two different types (André Schmidt and David Sheppard, personal communication), probably reflecting the presence of the two truncated CFTR forms detected here by Western blot (M150/M152/M156 and the shorter form). Both these types evidenced a detectable conductance, although of much lower open probability and burst durations than those of wt, consistent with the study by Carroll et al .
Overall, the biochemical data indicating that the prevalent N-truncated CFTR protein has decreased stability and very low processing efficiency together with these functional results showing that the c.120del23-CFTR has drastically reduced function, demonstrate a good correlation with the severe clinical CF phenotype of the two patients bearing this mutation. In previously analysis by Ussing chamber of native tissues from one of these patients, CFTR activity could not be detected . This apparent discrepancy with our current results may simply derive from the higher CFTR expression levels in the systems employed here which allow enhanced sensitivity in functional measurements. Nevertheless, this study demonstrates the relevance of the molecular and functional characterization of mutations for the clinical setting.
Work supported by PIC/IC/83103/2007 (FCT, Portugal) and NIH:R01 HL094585 (AH). ASR was a recipient of postdoctoral fellowship SFRH/BPD/20622/2004 (FCT, Portugal). We are grateful to André Schmidt and Dr. David Sheppard (University of Bristol, UK) for sharing their single-channel data on c.120del23-CFTR and also to Dr S Cheng (Genzyme) for the 936C plasmid.