Main

Cystic fibrosis (CF) is caused by mutations in the gene encoding the Cystic Fibrosis Transmembrane conductance Regulator (CFTR), a member of the ATP Binding Cassette (ABC) transporter superfamily. The well-established function of CFTR is to act as a cAMP-activated chloride channel. However, CFTR has many other roles, such as regulating epithelial transport proteins (1–3), vesicle trafficking (4) and acidification of intracellular organelles (5). CF, characterized by bronchopulmonary disease and pancreatic insufficiency, presents a wide range of disease severity, especially in terms of lung disease, even among patients harboring the same mutation.

We have previously shown that the cAMP-dependent chloride channel function of CFTR might be relatively conserved in some patients with a ΔF508 homozygous deletion, which is the most frequent mutation among Caucasians (6). We have shown that this heterogeneity of CFTR function is linked to the clinical course of the disease. Therefore, certain factors may modulate or complement CFTR function in some CF patients. The factors responsible for this residual CFTR function have not yet been elucidated, but might be exogenous factors such as environmental factors, or endogenous factors such as so-called “modifier” genes. Whether this residual chloride channel function per se causes less severe disease or whether another residual nonchannel CFTR function may be more relevant in CF pathophysiology remains to be determined.

Multidrug Resistance-associated Proteins (MRPs), which form a family of at least nine recently identified membrane proteins, share the closest structural homology with CFTR, belonging to the same subfamily C of the ABC transporter superfamily (7–8). Some MRPs, especially MRP1 and MRP5, are expressed in most human tissues, while others, such as MRP2 and MRP3, have a more limited expression pattern mainly in liver and gut (9–10). MRPs are characterized by their ability to export organic anions, either glutathione, glucuronate and sulfate conjugated molecules or unmodified substances, out of cells. Although the physiologic functions of these proteins are still under investigation, there is evidence to suggest that these proteins might be involved in drug resistance, detoxification, inflammation and oxidative stress (11–12).

In addition to the structural homology between CFTR and MRPs, there might be an overlap in functions between these proteins. The hypothesis that CFTR could be complemented by MRPs was driven from the case of a CF patient whose lung function improved after receiving anticancer drugs known to induce MRPs (13). In in vitro models, CFTR has been recently shown to efflux organic anions out of cells (14) and transport glutathione (15), which are known to be two main functional characteristics of MRPs.

In the present study, we therefore investigated MRPs gene expression and chloride conductive pathways in nasal respiratory epithelial cells from CF patients. We examined MRP1,-2,-3,-4 and -5 transcript levels by semiquantitative RT-PCR in ΔF508 homozygous patients and control subjects. In CF patients, we also examined the relationship between MRPs gene expression and clinical status on one hand and the existence of a chloride conductive pathway, on the other, by using a halide-sensitive SPQ (6-methoxy-N-3-sulfopropylquinolinium) fluorescent dye assay (6).

METHODS

Subjects

This study was carried out on healthy control subjects (n = 20) and ΔF508 homozygous patients (n = 19). All CF patients (9 girls and 10 boys; mean age: 12 y, range: 5–30) presented pancreatic insufficiency. Control subjects had a normal sweat test and had been negatively screened for the 30 most frequent CFTR mutations. Exclusion criteria were recent upper respiratory tract infection, smoking and nasal polyps.

Clinical status was assessed by measuring the forced expiratory volume (FEV1) and the forced vital capacity (FVC) to assess pulmonary function and by the Shwachman score (16) to assess overall disease severity. FEV1 and FVC were expressed as percentages of predicted values for age and sex. Shwachman score is based on four criteria, which are given equal weight, namely 25 points: the general activity, the findings on physical examination, the nutritional status and the findings on chest roentgenograms (maximum score is therefore 100).

This study was approved by our institutional review board. Before the study, written informed consent was obtained from the older children and from the guardian of each child.

Cells

Nasal respiratory epithelial cells (approximately 1 × 106 cells per subject) were obtained by gentle brushing of the middle turbinate. Brushes were immediately placed in cold Dulbecco's Modified Eagle's Medium (DMEM). An aliquot of scraped cells was cytospun on slides (60 g, 7 min, Cytospin Shandon, France) to check the cell type of the samples under light microscopy. The remaining cells were centrifuged (600 g, 2 min, 4°C) and the pellet was placed in Trizol™ reagent (GIBCO-BRL, Les Ulis, France) for further RNA extraction.

Each sample was assessed for cell type by taking into account the presence of necrotic cells, squamous, inflammatory (polymorphonuclear cells) and respiratory epithelial cells (Table 1). Samples containing necrotic cells, squamous cells or no respiratory epithelial cells were eliminated (i.e. classes 1, 2, or 3).

Table 1 Criteria for the cell type class of nasal samples
Full size table

RT-PCR Analysis

RNA isolation and cDNA synthesis.

Total RNA was isolated using Trizol™ reagent (GIBCO-BRL) by acid guanidine thiocyanate-phenol-chloroform and precipitated by isopropanol at −20°C. RNA (2 μg) was placed in a final volume of 20 μL containing 100 units of MMLV Superscript II reverse transcriptase (GIBCO-BRL), 500 μM dNTP, 10 mM DTT, and 200 ng of random hexanucleotide primers (Pharmacia Biotech, Saint Quentin en Yvelines, France). cDNA was then diluted 5-fold in sterile water.

PCR amplification.

The specificity of the primers used (Table 2) has been previously reported by our laboratory (17). Amplification of β2microglobulin gene was used as an internal control.

Table 2 Sequences of the forward and reverse primers used for PCR amplifications
Full size table

cDNA (5 μL) was added to a 25 μL final volume containing 2 U of Taq DNA polymerase (Boehringer Mannheim, Meylan, France), 500 μM dNTP, variable MgCl2 concentration (Table 2), 20 pmoles of each forward and reverse primer, 10 pmoles of β2microglobulin forward and reverse primers and 1 μCi of [αP32]dCTP (Amersham, Les Ulis, France). The samples were amplified for the adequate number of cycles (30 for all genes except MRP5 and 32 for MRP5) corresponding to the exponential range of amplification reaction and consisting of 1 min of denaturation at 94°C, 1 min of annealing at 55°C and 1 min 30 s of elongation at 72°C. The positive control used for MRP1 and MDR1 genes was the KB8.5 cell line. The positive controls for MRP2, -3, -4, and -5 genes were human liver for MRP2 and -3 genes, human lung for MRP4 gene and human skeletal muscle for MRP5 gene. Negative controls were included for each amplification (water instead of RNA for reverse transcription and water instead of cDNA for PCR).

Semiquantitative analysis of PCR products.

Amplification of the β2microglobulin gene was performed for each sample to account for variability in mRNA extraction, mRNA quantitation and cDNA synthesis. Amplification products were submitted to electrophoresis on 8% polyacrylamide gels at 750 V for 3.5 h. After exposure of gels to X OMAT films (Eastman Kodak), autoradiographs were scanned using a Bio-Rad densitometer (Bio-Rad, Ivry sur Seine, France). Each gene mRNA content was expressed as the ratio of the densitometric value of each gene amplification product to that of β2microglobulin amplification product.

Functional Assay for Chloride Conductive Pathways

Transmembrane chloride conductive pathways were measured in nasal ciliated cells either under basal conditions or after addition of cAMP agonists using a halide-sensitive SPQ fluorescent dye assay, as previously reported (6). Briefly, this method measures the rate of chloride transport as the change in fluorescence of intracellular SPQ in response to exchange of extracellular chloride with nitrate (NO3−), an anion that passes through CFTR, but does not quench indicator fluorescence unlike chloride. cAMP agonists consisting of 25 μM forskolin, 100 μM isobutylmethylxanthine and 500 μM 8-(4-chlorophenylthio)-adenosine 3′:5′-cyclic monophosphate were added to activate CFTR resulting in chloride efflux, NO3− influx and increased fluorescence. To enhance the sensitivity of the assay, chloride was replaced by iodide (I−), as CFTR is permeable to I− and SPQ fluorescence is more strongly quenched by I− than by chloride.

Basal anion conductance was defined as the maximum fluorescence change (ΔFbasal/Δt) after replacing I− with NO3−. The cell was considered to have a basal chloride conductive pathway when the maximum fluorescence change ΔFbasal/Δt was >0.1. cAMP-dependent anion conductance was defined as the greatest fluorescence change (ΔFcAMP/Δt) after adding cAMP agonists. The cell was considered to have a cAMP-stimulated chloride conductive pathway when the maximum fluorescence change after addition of the cAMP agonists ΔFcAMP/Δt was >0.1. Results were expressed for each subject studied as the percentage of ciliated cells with basal chloride conductive pathway and the percentage of ciliated cells with cAMP-stimulated chloride conductive pathway. The experimental protocol used is shown in Fig. 1.

Figure 1
figure 1

Experimental protocol for SPQ fluorescence assay to measure chloride conductive pathways in nasal ciliated cells. The cells were superfused with I−, NO3− and then with cAMP agonists.

Full size image

Statistics

Statistical calculations and tests were performed using Stat View 5.0 software (Abacus, CA, U.S.A.). The limit of statistical significance was defined as p ≤ 0.05.

Means were compared using t tests. Results are expressed as mean ± SEM. Among CF patients, a subgroup with lower MRP1 levels and another subgroup with lower MRP5 levels were found (Fig. 2) and relationship between MRP1 or MRP5 transcript levels and clinical status (age, FEV1, FVC and Shwachman score) was therefore investigated. The highest MRP1 and MRP5 values in the subgroups with lower levels were 0.145 and 0.077, respectively. The values 0.15 and 0.08 were therefore chosen as cut-off values to analyze differences in clinical status according to MRP1 or MRP5 levels, respectively.

Figure 2
figure 2

Semiquantitative analysis of MRP1 (A), MRP3 (B), MRP4 (C), MRP5 (D), and MDR1 (E) mRNA levels in controls (C) and ΔF508 homozygous patients (CF) by densitometric evaluation of each gene PCR product compared with β2microglobulin PCR product. Horizontal line: mean level of expression for each data set.

Full size image

To investigate whether there was a difference in MRP1 or MRP5 transcript levels (≤0.15 or >0.15 and ≤ 0.08 or >0.08 for MRP1 and MRP5, respectively) and cell type (inflammatory or non inflammatory), the Fisher exact test was used.

Correlation coefficients were calculated for each combination of mRNA levels.

RESULTS

Cell type of the analyzed nasal samples.

Nasal samples containing necrotic cells, squamous cells or no respiratory epithelial cells were eliminated for further analysis (Table 1, classes 1–3). No difference in cell type, as assessed by the cell class, of the remaining samples, was observed between control subjects and ΔF508 homozygous patients (Table 3). In both groups, the majority of the samples (90%) contained epithelial cells (i.e. ciliated, goblet and basal cells) with no inflammatory cells and were classified 7 or 8 depending on the proportion of ciliated cells.

Table 3 MRPs and MDR1 transcript levels in healthy controls (C, n = 20) and ΔF508 homozygous patients (CF, n = 19)
Full size table

MRPs and MDR1 transcript levels.

Individual results are presented in Table 3. Variable MRP1 transcript levels were observed in control subjects (0.87 ± 0.12) and ΔF508 homozygous patients (0.56 ± 0.09) with significantly lower levels in CF patients (p < 0.05) (Fig. 2A). A subgroup of CF patients (n = 5, about 25%) with undetectable or very low transcript levels (≤0.15) could be distinguished. No difference in cell type was observed between samples from CF patients with MRP1 mRNA levels lower or higher than 0.15 as percentages of samples without inflammatory cells (i.e. classes 7–8) were similar (80%versus 92%, respectively; Fisher exact p value = 0.5).

We found no detectable MRP2 transcripts in any of the nasal samples from all subjects analyzed.

MRP3 transcript levels were similar in controls (0.15 ± 0.04) and CF patients (0.16 ± 0.04) with undetectable or low MRP3 transcript levels in both groups (Fig. 2B).

Variable MRP4 transcript levels, including low or undetectable mRNA levels, were observed in controls (0.43 ± 0.07) and CF patients (0.28 ± 0.06) (Fig. 2C) with no statistical difference between the two groups.

Analysis of MRP5 mRNA showed variable levels among controls (1.37 ± 0.14) and CF patients (0.73 ± 0.12) (Fig. 2D). However, MRP5 transcripts (>0.08) were observed in all controls, but were undetectable or very low (≤0.08) in about 25% of CF patients (n = 5) and this difference between CF patients and controls was statistically significant (p < 0.01). Cell type was similar in samples from CF patients with MRP5 mRNA levels lower or higher than 0.08 (percentages of samples without inflammatory cells: 100%versus 85%, respectively; Fisher exact p value > 0.9). Three patients had low mRNA levels for both MRP1 and MRP5 genes (Table 3).

MDR1 gene encodes for Pgp which is an ABC transporter involved in efflux of various compounds and belonging to another subfamily (B) than subfamily C which includes CFTR and MRPs. CFTR and MDR1 genes have been shown to exhibit complementary patterns of epithelial expression, suggesting analogous roles in epithelial cells (18). Therefore, in addition to MRPs, we also analyzed MDR1 gene expression in CF patients. Similar low MDR1 levels were detected in controls (0.06 ± 0.03) and CF patients (0.07 ± 0.03) with undetectable transcripts in about one half of subjects in each group (Fig. 2E).

Correlation analysis of gene transcript levels.

Correlation coefficients (r) for the correlation of MRPs and MDR1 mRNA levels were calculated taking into account all subjects (n = 39) (Table 4). Among the various possible combinations of mRNA data sets, MRP1 and MRP5 transcript levels demonstrated the highest correlation (r = 0.71, p < 0.0001) (Fig. 3).

Table 4 Correlation coefficients (r) for the correlation of transcript levels in controls and ΔF508 homozygous patients (n = 39)
Full size table
Figure 3
figure 3

Correlation between MRP1 and MRP5 transcript levels in controls and ΔF508 homozygous patients (n = 39). r = 0.71, p < 0.0001.

Full size image

Relation with clinical status.

As an overlap in MRP1 and MRP5 transcript levels was found between some CF patients and some controls, we investigated whether CF patients exhibiting lower MRP1 (≤0.15, n = 5) or lower MRP5 transcript levels (≤0.08, n = 5) had a different clinical status on the basis of age, FEV1, FVC and Shwachman score (Table 5). No association was found between MRP1 or MRP5 transcript levels and age, FEV1 or FVC (Table 6). However, patients with lower MRP1 transcript levels had significantly lower Shwachman scores, although in this subgroup one patient had a maximal score of 100. No such relation was observed for MRP5 gene.

Table 5 Clinical status (according to age, FEV1, FVC, and Shwachman score) and chloride conductive pathways in CF patients
Full size table
Table 6 Age, FEV1, FVC, and Shwachman score according to MRP1 (A) and MRP5 transcript levels (B) in CF patients ( n = 19). Results are expressed as mean ± SEM.
Full size table

Relation with basal and cAMP-dependent chloride conductive pathways.

Given that the cAMP-dependent chloride channel function of CFTR varies between ΔF508 homozygous patients and that this variation is related to the severity of CF (6), we investigated whether MRP1 or MRP5 gene expression could be related to CFTR function. We measured cAMP-dependent chloride conductive pathway using a halide-sensitive SPQ fluorescent dye assay in nasal ciliated cells from CF patients. We also measured chloride conductive pathway under basal conditions, i.e. before addition of cAMP agonists. As we previously reported (6), the percentage of ciliated cells exhibiting basal chloride conductance varies among healthy controls as it is the case for cAMP-dependent chloride conductance. The lowest percentages were 63% and 50% for basal and cAMP-dependent chloride conductances, respectively (6). These values were therefore chosen as cut-off values to analyze conductances in CF patients (see Table 5 for individual results).

As expected, there was a significant difference in the percentage of ciliated cells exhibiting a cAMP-dependent chloride conductive pathway between controls and CF patients (83 ± 5%versus 45 ± 7%, p < 0.01). There was also a significant difference in the percentage of ciliated cells exhibiting a basal chloride conductive pathway between controls and CF patients (98 ± 2%versus 68 ± 7%, p < 0.01).

When considering cAMP-dependent chloride conductive pathway, no difference in MRP1 or MRP5 transcript levels was observed between CF patients with less than or more than 50% cells showing cAMP-dependent chloride conductive pathway. However, when considering basal chloride conductive pathway, significantly lower MRP1 transcript levels were found for CF patients with less than 63% cells exhibiting basal chloride conductive pathway (p < 0.01) (Fig. 4). Although lower MRP5 transcript levels were also found for CF patients with less than 63% cells with basal chloride conductive pathway (0.49 ± 0.19 versus 0.84 ± 0.15%), this difference was not significant.

Figure 4
figure 4

MRP1 transcript levels according to the percentage of nasal ciliated cells exhibiting a ΔFbasal/Δt fluorescence change >0.1. The 63% value was the lowest percentage observed in controls as previously reported (6) and was therefore chosen as cut-off value. Significantly lower MRP1 transcript levels were found for CF patients with less than 63% cells exhibiting basal chloride conductance: 0.23 ± 0.12 vs 0.7 ± 0.09%, p < 0.01.

Full size image

DISCUSSION

The main findings of the present study are:i) the lower MRP1 and MRP5 transcript levels in ΔF508 homozygous patients compared with control subjects and ii) the associations of MRP1 transcript levels with both clinical status and presence of a cAMP-independent chloride conductive pathway in CF patients. These results suggest a potential interrelation between MRPs and CFTR and that MRPs, especially MRP1, might play an important role in the pathophysiology of CF.

The lower expression of MRP1 and MRP5 genes observed in CF patients compared with controls, which was not due to differences in cell type of the analyzed nasal samples, raises the question of the mechanisms of this down-regulation. Regulation of expression of MRPs genes is still poorly known. MRP1 expression has been shown to be modulated by drugs and toxins (19), oxidative stress (20), Nmyc (21) and p53 (22–23) and down regulated by elevated glutathione levels (20) and wild-type p53 (23). CF lung disease, which is the major cause of death, is characterized by chronic inflammation and severe oxidative stress in the airways. Oxidative stress may therefore be involved in MRP1 regulation in CF patients. Even fewer data are available concerning the mechanisms of MRP5 gene regulation. In contrast with MRP1, MRP5 does not appear to be regulated by pro-oxidants (24).

The highest correlation among MRPs genes between MRP1 and MRP5 transcript levels for control subjects and CF patients (Fig. 3) suggests that these two genes may be co-ordinately regulated under physiologic conditions and in CF. Interestingly, in another non-CF pathologic setting, namely lung cancer, no such strong correlation was found between MRP1 and MRP5 mRNA levels, while the highest correlation was observed between MRP1 and MRP3 mRNA levels (25). The mechanisms of MRPs gene regulation and consequently their co-regulation may therefore be different in different diseases.

Defective cAMP-dependent chloride transport in the apical membrane of secretory epithelia has been considered to be the underlying source of CF pathophysiology. However, there is growing evidence that CF pathogenesis might be not solely related to the defective chloride channel function, but also to other nonion channel functions of CFTR and to “modifier” genes (26–27). An interesting finding of the present study is the association between MRP1 transcript levels and clinical course of CF disease, as low MRP1 levels were associated with more severe disease, as assessed by the Shwachman score, than high MRP1 levels. Although this result may be incidental and needs further investigation on a larger series of patients, it suggests that MRP1 could be involved in CF pathophysiology and could constitute a “modifier” gene in CF. Although MRP1 is localized to the basolateral membrane of epithelial cells, especially respiratory ciliated cells (10), as opposed to CFTR which is on the apical membrane, CFTR and MRP1 may have similar functions, such as efflux of organic anions (14) or glutathione (15). The function of MRP1 possibly involved in the clinical course of CF might be related to either a nonchloride channel function and/or the chloride channel function of CFTR. MRP1 is thought to be involved in various physiologic or pathologic processes, some of which may be potentially relevant in CF pathophysiology, such as inflammation and antioxidant defense.

Among MRPs, MRP1 has the highest affinity for leukotriene C4 (28), a potent mediator of the inflammatory response causing edema and increased vascular permeability. A decreased inflammatory response was observed in a MRP1-deficient mouse model, consistent with a role of MRP1 in mediating leukotriene C4 export (29). As bronchial cells are known to produce various types of leukotrienes, especially leukotriene C4 (30), one of the functions of MRP1 in respiratory cells could be efflux of endogenously synthesized leukotriene C4. The absence of an effective inflammatory response via MRP1 could therefore be involved in the more severe clinical course of CF in patients with low MRP1 transcript levels.

Oxidative stress is increased in CF patients due to chronic inflammation and evidence for inadequate antioxidant defense has been reported (31). In particular, glutathione (GSH), which is a major antioxidant in the lung, is decreased in the apical fluid from CF airway epithelial cells (32). As MRP1 can transport GSH out of cells into the interstitial compartment, MRP1 might play a protective role against toxic compounds generated by oxidative stress (11). Oxidative stress has been shown to increase paracellular permeability (33). GSH might therefore reach the airway lining fluid from the interstitial compartment via the paracellular route during oxidative stress. CF patients with lower MRP1 levels might have lower GSH in airway fluid via lower efflux of GSH by MRP1 and subsequently they might have a more severe disease via a lack of adequate protective role in the lung against oxidants due to defective MRP1.

The striking relation between MRP1 transcript levels and the existence of a basal chloride conductive pathway in CF patients (Fig. 4) strongly suggests that MRP1 may have a chloride channel-related function. Therefore, MRP1 could be involved in the better clinical status of CF patients with high MRP1 levels via this chloride channel-related function in addition to the possible functions discussed above.

A chloride channel-related function has been previously reported in human cell lines overexpressing MRP1 (34–35). Whether MRP1 might act as a chloride channel by itself or regulate, directly or indirectly, a chloride channel remains to be determined. Given the basolateral localization of MRP1 and its possible implication in CF pathophysiology, we hypothesize that MRP1 rather indirectly regulates a chloride channel localized to the apical membrane as CFTR. Our results suggest that the channel regulated by MRP1 would be different from CFTR, as no association was demonstrated between MRP1 mRNA levels and the existence of cAMP-dependent chloride conductance characteristic of CFTR.

Although we found a correlation between MRP1 and MRP5 transcript levels, we found no association of MRP5 with either clinical status or chloride conductive pathway. A possible explanation is that the group of patients with low MRP1 mRNA levels was not the same as the group with low MRP5 levels.

As no effective treatment for CF is yet available, many different treatment strategies are being developed in addition to transfection of normal CFTR gene, especially activation of defective CFTR or non-CFTR chloride channels by pharmacological agents. Although there is need to further evaluate MRPs in CF at the protein level, our results suggest that MRP1 could constitute a new target for pharmacotherapy for CF. Based on the findings that (i) low MRP1 gene expression was associated with more severe disease and (ii) levels were positively associated with the existence of a cAMP-independent chloride conductive pathway, a new therapeutic approach might consist of inducing MRP1 in CF patients. Such pharmacological modulation of MRP1 in CF patients, associated with restored chloride conductance, has been very recently reported in an abstract (36).