A dual gene reporter system was used to quantify the readthrough efficiency directed by the most frequent stop mutations in the French population (Y122X, G542X, R1162X and W1282X) [10], in the presence or absence of gentamicin. It has been previously demonstrated that for a given mutation, the readthrough levels obtained with this experimental system are similar to those obtained in vivo[11]. With this dual reporter, all the ribosomes that initiate translation give rise to β-galactosidase activity, while only those that read through the stop codon lead to luciferase activity. The use of a dual reporter allows internal calibration of transfection efficiency, transcription rate of the reporter gene, stability of the message and translation initiation efficacy between independent experiments. A 100% activity control was provided by a construct (TQ) with no stop codon between the coding sequences of the two reporters. Corresponding complementary oligonucleotides were annealed and cloned between the lacZ and luc coding sequences in the pAC99 vector (Table 1) [11]. NIH3T3 cells were electroporated with 20 μg of amplified plasmid and incubated with or without gentamicin at a concentration of 600 μg/ml. This concentration has previously been shown to be optimal for readthrough induction while being non-toxic during the length of the treatment in a pilot experiment where concentrations ranging from 100 to 1,200 μg/ml where tested on NIH3T3 cells ([11] and Bidou L and Rousset JP, unpublished results). Following transfection, cells were rinsed every day and fresh medium supplemented with antibiotic was added. After three days of expression, cells were harvested and lysed, and β-galactosidase and luciferase enzymatic activities were assayed. The readthrough efficiency was estimated by the ratio of luciferase to β-galactosidase activity obtained with the test construct and normalised with the ratio obtained with the in frame control construct.

The study was approved by the Necker-Enfants Malades Ethics Committee, and written informed consent was obtained for each subject. Exclusion criteria were abnormal baseline hearing or kidney function, nasal polyps, upper respiratory tract infection and treatment by either parenteral or inhaled aminoglycosides in the previous month. The ethics committee refused to authorise inclusion of a control group of patients treated with placebo, holding the opinion that, given the obvious lack of benefit to these patients, the benefits of having a placebo group did not justify their exposure to the risk of parenteral perfusions.

CF patients were treated for 15 days with intravenous gentamicin, administered once daily at 10 mg/kg infused for 30 minutes. Gentamicin trough (24 hours after infusion) and peak (after 30 minutes infusion ended) levels were measured on the third day. The dose was then adjusted to achieve peak serum levels between 20 and 40 μg/ml and trough levels <2 μg/ml. Gentamicin toxicity was monitored before the study and at midpoint by measuring serum creatinine concentrations and by audiometric studies, including pure-tone air conduction threshold and high-frequency tones.

Clinical evaluation took place the day before gentamicin treatment began (day 0) and the day after it ended (day 15). The Cystic Fibrosis Clinical Score (CFCS), based on pulmonary, nutritional evaluation and temperature [12], was used to assess clinical status. This score has been validated to evaluate modification of clinical parameters. It includes five common symptoms (cough, sputum production, appetite, shortness of breath, and energy level) and five physical findings (temperature, respiration rate, weight, air exchange and crackles). It is graded on a five-point scale for a maximum severity score of 50 points. A decrease of 10 points is considered clinically relevant. Forced expiratory volume (FEV1), forced vital capacity (FVC), and forced expiratory flow at 25 to 75 percent of vital capacity (FEF_25–75) were expressed as percentages of predicted values for age, sex, and height. Patients had not been treated for 8 hrs with β_2-mimetics before performing respiratory function explorations. An increase of 7.5% for any of the three tests was considered significant.

Transepithelial nasal potential difference (NPD) was measured according to the procedures described by Knowles and collaborators [13]. Baseline NPD was measured after perfusion of nasal epithelium with saline solution, and NPD changes were recorded after perfusion with the following solutions: 100 μM amiloride in saline solution to block sodium current (Δamiloride), chloride-free solution with 100 μM amiloride (ΔCl^-free) and chloride-free solution with 100 μM amiloride and 10 μM isoproterenol (Δisoproterenol). The sum of ΔCl^-free and Δisoproterenol (ΔCl^-free-isoproterenol) reflects the cAMP activation of nasal mucosa chloride permeability and served as the indicator of transepithelial CFTR-dependent chloride transport.

CFTR expression in nasal epithelial cells was assessed with 24-1 (R & D Systems, Lille, France) and MATG 1061 (gift from Transgène, Strasbourg, France) monoclonal antibodies before and after gentamicin treatment. These antibodies recognise amino acids 1377–1480 at the C-terminus and amino acids 503–515 in the first nucleotide-binding domain of CFTR, respectively and were chosen as they are accurate to detect CFTR in nasal epithelial cells by immunocytochemistry [14]. Cells were obtained by nasal brushing below the middle turbinate from the CF patients who agreed to furnish these samples and from two healthy controls thoroughly genotyped for the CFTR gene. Cells were spread on slides and fixed at 4°C in acetone for 10 min and then in 4% paraformaldehyde for 20 min. They were then incubated in 10% acetic acid for 10 min. After blockage of nonspecific binding sites with 0.1% TBS, 3% Tween 20, 10% bovine serum albumin-normal human AB serum, cells were incubated for 1 hr with 24-1 or MATG 1061 monoclonal antibodies diluted 1:100 and 1:1,000, respectively. They were then incubated with biotinylated goat-antimouse IgG (Dako, Trappes, France) diluted 1:600 and then with streptavidin-alkaline phosphatase conjugate (Dako, Trappes, France). Enzymatic activity was visualised with Fast Red/naphtol (Sigma Aldrich, St Quentrin Fallavier, France). Slides were then washed, counterstained with hematoxylin and mounted in aqueous medium. Negative controls were obtained by using an isotype-matched mouse nonimmune IgG2 instead of the primary antibody. Two investigators blinded to the patients' data examined all slides. Patients were considered positive for CFTR expression when positive staining at the membrane was observed in at least 10% of the cells examined with at least one antibody.

Data are presented as mean (SD). Data between day 0 (D 0) and day 15 (D 15) were compared with the nonparametric Wilcoxon test. Groups were compared with the nonparametric Mann-Whitney test for quantitative variables. Correlation coefficients were calculated with the Spearman correlation test.

Basal and gentamicin-induced readthrough levels are shown in Table 1. The R1162X and the G542X mutations yielded basal readthrough of less than 0.03% and increased by a factor of 10 and 15 respectively after gentamicin. The basal readthrough of W1282X was ~10 times higher than those of R1162X and G542X and tripled after gentamicin. Y122X had a basal readthrough level five times higher than that for the W1282X mutation, 22 times that for R1162X and 30 times that for G542X. After gentamicin incubation, Y122X readthrough efficiency remained at least 4.5 times higher than that for W1282X, six times that for G542X and 7.3 that for R1162X. We therefore decided to focus the clinical trial on patients with the Y122X mutation.

Table 2 summarises the characteristics of the 18 study patients. Nine carried the Y122X mutation (eight were Y122X homozygous and one was Y122X/F508del) (Group A). Four had another stop mutation: one was homozygous for G542X, one for R1162X, and two were compound heterozygous for W1282X/F508del and R553X/CFTRdele17b (Group B). In addition, five patients without stop mutations were studied according to the same protocol and served as a control group. Three were F508del homozygous, and the other two were compound heterozygous for F508del and class 2 mutations (Group C). There were no significant differences between the three groups for pulmonary status or gentamicin sensitivity of the bacterias in the patient's sputum.

Modification of clinical scores and respiratory function are shown in Tables 3 and 4. Clinical scores for the nine patients with the Y122X mutation (Group A) improved significantly at the end of the study, mainly because of improvements in coughing, sputum production, dyspnea and energy level. Even among the patients with the most severe symptoms, these changes were noted as early as the fourth day of treatment. Pulmonary function also improved significantly for FEV1 and almost at the significant level for FVC and FEF_25–75 (p = 0.09). These changes were not correlated with microbiological sensitivity to gentamicin. Four of the five patients with significant improvement had bronchial colonisation with gentamicin-resistant bacterias. The most impressive effects were observed in a patient with severe pulmonary disease and bronchial colonisation with a gentamicin-resistant Burkholderia cepacia strain. He showed substantial clinical improvement (far greater than with the standard antibiotic treatment) with much less coughing and dyspnea and a considerable fluidification of his sputum. At the end of the trial, he had gained 1.5 kg, and his FEV1 had improved by 19%, FVC by 24%, and FEF_25–75 by 150% (line 7 in Table 3).

Neither clinical scores nor respiratory functions changed significantly among the patients with other stop mutations (Group B) or among controls (Group C).

Mid-study renal function testing and audiometric studies showed no significant changes from baseline. The mean peak and trough serum gentamicin levels were in the target range for all the patients as early as the third day.

Initial sweat chloride levels were typical for CF patients, that is > 60 mM/L. These levels decreased significantly after treatment among patients with the Y122X mutation (Group A) (Tables 3 and 5; Figure 1). Values normalised in two patients to levels < 60 mM/L, and for another, the value fell from 114 mM/L to 65 mM/L. One week after the study ended in these three patients, sweat chloride concentrations had returned to levels above 75 mM/L, thereby indicating that the gentamicin-induced change did not persist after discontinuation of treatment. There was no significant modification of the sweat chloride values in the two other groups (Table 5).

NPD values before treatment were typical of CF patients, i.e., highly negative basal NPD, strong depolarisation in response to inhibition of sodium current by 100 μM amiloride (Δamiloride) and no significant response to CFTR activation by isoproterenol 10 μM in chloride-free solution (ΔCl^-free-isoproterenol) [13] (Figure 2a, Tables 3 and 5). Treatment with gentamicin for 15 days decreased the basal NPD (p = 0.12) and the amplitude of the amiloride-induced depolarisation (p = 0.09), but this did not reach the significance level (Group A, Table 5). Isoproterenol perfusion in chloride-free solution significantly hyperpolarised NPD, from -0.8(1.3) mV to -4.6(6) mV (p = 0.04) (Table 3 and Group A, Table 5). Recordings taken at the end of the study generally showed a stable pre-isoproterenol baseline during low chloride solution perfusion, followed by hyperpolarisation after isoproterenol was added (Figure 2b). There was a significant negative correlation between the changes of Δamiloride and of ΔCl^-free-isoproterenol. (ρ = -0.9; p = 0.01; Spearman correlation test). These results are consistent with the restoration of a CFTR-dependent secretory chloride transport in the nasal epithelium. Changes in the sweat chloride concentration and response to isoproterenol were not correlated (ρ = 0.17; p = 0.66).

In contrast to the patients with the Y122X mutation, the patients with other stop mutations (Group B) and those without any stop mutations (Group C) had no significant modifications of their basal potential difference or response to amiloride or isoproterenol (Table 5).

The analysis of CFTR expression with the 24-1 and MATG 1061 antibodies were all concordant. The effect of parenteral gentamicin on CFTR was analysed in patients who agreed to nasal brushing, i.e., seven Y122X homozygous patients, one compound F508del/Y122X patient, one R1162X homozygous patient, and the five patients without stop mutations. Before gentamicin treatment, no CFTR labeling was observed in any of the Y122X homozygous patients (see representative picture in Figure 2c), while afterwards, five patients expressed CFTR at the membrane in 10 to 80% of cells (Figure 2d). These five patients were considered responders. The staining pattern did not change for two patients after treatment. Before treatment, cytoplasmic labeling in the compound heterozygous Y122X/ΔF508 patient was found in 20% of the cells with 24-1 antibody and 70% of the cells with MATG 1061 antibody and, after treatment, in 70% and 100% of the cells, respectively. This patient was considered a responder. Among the Y122X responders, CFTR-dependent chloride secretion increased after treatment from -1.1(1) to -5.8(6) (p = 0.06) and remained absent for the two non-responders. Disease also improved among responders. CFCS variation between the two evaluations was -10(7) in the responders versus -2(1) for the non-responders (p = 0.04).

The R1162X patient had no CFTR labeling either before or after treatment (data not shown). The patients without any stop mutation (group C) had only cytoplasmic CFTR staining, a pattern that did not change after treatment (data not shown). Both the R1162X homozygous patient and the group C patients were considered non-responders.

Seric and sputum concentrations were higher in the responder patients (respectively 25.8(2) μg/mL versus 20.6(3.3) μg/mL, p = 0.05; and 2.4(1.4) μg/mL versus 1.7(1.3) μg/mL, NS).

Overall, the pattern of the in vitro readthrough, clearly most efficient for the Y122X mutation, was strongly correlated with the immunocytochemically determined CFTR expression in nasal cells, as assessed by the CFTR staining after treatment for the Y122X patients, compared with the lack of staining in both the R1162X patient and the patients without stop mutations.