MATERIALS AND METHODS

Patients and families

In total, 77 children with IB from 73 unrelated families were included in the study. Eight of the patients were sib pairs. Forty-five of the patients were girls and 32 were boys. DNA samples were available from the great majority of patients’ parents and 51 healthy sibs. Thirty-eight patients had no sibs available for study, while 23 had one healthy sib each, nine had two, two had three, and one had four. There were no twins in the study group. Appropriate informed consent was obtained from the families.

The status of parental consanguinity was known in 52 of the families: 27 declared consanguinity (19 were first cousin marriages) and 25 denied it. Two parent pairs in the latter group had originated from the same village, thus some degree of consanguinity could not be excluded.

Clinical findings

Bronchiectasis was diagnosed by computed tomography scan or bronchography. Primary ciliary dyskinesia, α₁-antitrypsin deficiency, and immunodeficiency were excluded as the cause of the disease. Patients had no other clinical findings such as malabsorption or sinus disease and had normal to borderline sweat chloride values (<60 mEq/l). Eleven of the patients were diagnosed as borderline CF late in the study. Broad clinical information was available for 46 of the patients, 30 girls and 16 boys. These patients had a mean age of 10.3 (SD 3.9) years at the time of the last clinical examination, and the disease manifested at 1 month to 12 years (mean 31.7, SD 44 months). In 12 of them the disease was disseminated. In the rest, it was localised most commonly in the lower left lobe (15), followed by lower right (9), and both lower lobes (8). Bronchiectasis was not localised in an upper lobe in any patient. Seven of the 46 patients had undergone lobectomy, while three others exhibiting a severe clinical course had been assessed in need of but unsuitable for operation. Eleven of the patients had chronic persistent cough and in 25 patients coughing was productive. Ten patients complained only of sputum production. One patient had haemoptysis, five had chest deformity, and 13 had clubbing. Pulmonary function tests were performed in 33 patients above the age of 6 years. The average forced vital capacity (FVC) was 72.6% (SD 23.8) and the forced expiratory volume in one second (FEV1) 68.5% (SD 24.1).

Mutation analysis

The methods have been described in detail by Kýlýnç et al.¹⁵ Briefly, all 27 exons of the CFTR gene and the flanking intronic sequences were amplified by polymerase chain reaction (PCR) and analysed by denaturing gradient gel electrophoresis (DGGE).^16,17 Any pattern variation was investigated by comparing it to known DNA variant marker patterns, and, when necessary, by subsequent DNA sequence analysis. Amplification primers were kindly supplied by Professor M Goossens on behalf of the European Concerted Action for Coordination of Cystic Fibrosis Research and Therapy (ECACF). In addition, patients were screened for five mutations not detectable by the DGGE analysis described above. They were intronic mutations 3849+10kbC>T¹⁸ and 1811+1.6kbA>G,¹⁹ deletion mutations CFTRdele2,3²⁰ and CFTRdele19²¹, and −33A>G in the minimal promoter region.²² Marker DNA samples for the latter region were kindly provided by Dr M Claustres. E1228G was identified by sequence analysis in an ABI 310.

Haplotype analysis

Patients and family members were assayed for a total of six intragenic DNA polymorphisms, five intronic and one exonic, as described previously.¹⁵ The three alleles (T5, T7, and T9) of T_n were amplified by allele specific PCR.³ Reliability of the technique was ascertained by verification by another method that involved nested PCR amplification, cleavage with a restriction enzyme at the created site, and size determination on 8% polyacrylamide gels.⁹ The alleles for the (TG)_m tract upstream of T_n that were associated with alleles T5 and T9 were determined by allele specific PCR amplification and size determination on 8% polyacrylamide gels. Polymorphism 470M/V (A/G variation at nucleotide 1540 in exon 10) was assayed by either DGGE or restriction enzyme digestion.²³ The alleles for the biallelic GATT repeat in intron 6 and the three multiallelic microsatellites (IVS8CA in intron 8 and IVS17bTA and IVS17bCA in intron 17b) were resolved on polyacrylamide gels.^24–26 Haplotypes were constructed by segregation analysis using the available DNA samples from parents and sibs.

Statistical analysis

χ² test was applied to contingency tables to detect statistically significant differences in allele frequencies²⁷; p values of less than 0.05 were considered to indicate significance and less than 0.01 high significance.

RESULTS

In total, 69 unrelated children and four sib pairs with the IB phenotype were screened for mutations in the CFTR gene. Mutations were identified in 28 of the unrelated patients. Nine different mutations were identified in 34 (23.3%) of the 146 CFTR chromosomes analysed. Five patients were homozygous for CFTR mutations, one each for 3272-26A>G, N1303K, and CFTRdele19, and two for K68E. Another patient was a compound heterozygote for −33A>G and CFTRdele2,3. Twenty-two patients carried one mutation each. The chromosomal backgrounds were determined, and CFTR haplotypes were constructed to assess identity by descent. The frequencies of T9 and T5 as well as whether any alleles of 470M/V and (TG)_m were associated significantly were investigated by comparing our CF patient group and the normal control group reported previously.¹⁵

Spectrum of mutations

The mutations detected at the CFTR locus in the IB patients and their frequencies are shown in table 1. The most frequent mutation was K68E (nine of the total 34 mutant chromosomes), a mutation we recently identified in a Turkish CF patient.¹⁵ The next most frequent mutations were −33G>A and N1303K, observed on seven and six chromosomes, respectively, and 3272-26A>G was found on three chromosomes. Large deletion mutations CFTRdele19 and CFTRdele2,3 were observed in four and two chromosomes, respectively. One patient had W1282X and another F1052V, while the last one had the novel mutation E1228G, which resulted from an A to G transition at nucleotide 3815 (sequence data available on request). An uncharged polar amino acid was substituted for an acidic one at residue 1228 in exon 19 in the second nucleotide binding domain of the protein. The residue has been conserved in human, bovine, Xenopus, and dogfish.²⁸ No other mutant allele was identified upon screening for the two intronic mutations 3849+10kbC>T and 1811+1.6kbA>G, which were reported to be quite common in southern Mediterranean populations.^18,19

The spectrum of mutations and their frequencies differed from those in our CF patient group. F508del, 1677-1678delTA, 2183AA>G, and G542X, the most common four mutations in our CF patients comprising 52% (64/125) of all mutant chromosomes,¹⁵ were not observed at all in the IB group. Only four of the total of nine different mutations we found in the IB group were also found in the 166 CF chromosomes we had analysed.¹⁵ The numbers in CF patients were as follows: one K68E, four N1303K, one CFTRdele2,3, and five W1282X. Three of the remaining IB mutations (3272-26A>G, CFTRdele19, F1052V) were identified in other studies in CF patients,^16,21,29 one (−33G>A) in a CBVAD patient,²² and the last one (E1228G) in this study.

Significance of the allelic frequencies at three polymorphic loci

Certain alleles of T_n and (TG)_m and allele 470M, alone or in association with others, have been implicated in the aetiology of bronchiectasis.^1,3,4 We investigated whether any alleles were associated with IB in our patients. We had found the frequencies of T9, T7, and T5 in the normal chromosomes to be 28, 162, and 10, respectively, in a total of 200.¹⁵ We found the frequencies of T9, T7, and T5 in the IB chromosomes in which no mutation was identified (no mutation chromosomes) to be 71, 30, and 11, respectively, in 112. The frequency of T5 with respect to T7 in the IB no mutation chromosomes was found to be highly significant compared to the normal population (11/41 versus 10/172). The frequency of T9 with respect to T7 in no mutation IB chromosomes also was significantly higher than in the normals (71/101 versus 28/190). Regarding the IB chromosomes harbouring mutations, we found the frequencies of T9, T7, and T5 to be 20, 13, and 1, respectively, in 34. Thus, the frequency of T9 with respect to T7 was highly significant (20 versus 13) as compared to the CF mutant chromosomes (17 versus 102).¹⁵

The frequency of 470M was similar in the normal T9 chromosomes (13/28) and the normal T7 (66/157). Similarly, the difference in the association of 470M between the T7 IB no mutation chromosomes (16/29) and the T7 normals (66/157) did not reach significance. However, T9 no mutation IB chromosomes had a highly significant association with 470M in comparison to the T9 normals (44/70 versus 13/28). In addition, while all of the 10 normal T5 alleles were associated with 470M, five of 11 IB T5 no mutation chromosomes were on 470V background.

The frequencies of the (TG)_m alleles were similar in T9 IB no mutation chromosomes (58 TG9 and 13 TG11) and normal T9 chromosomes (24 TG9 and four TG11). In contrast, the association with the T5 chromosomes was different: seven of the T5 IB no mutation chromosomes were on TG11 background and four on TG13, while all of the 10 normals were on TG11.

In summary, T9 showed a highly significant association with the disease, and 470M was highly significant in the T9 no mutation IB chromosomes. Allele T5 also showed a highly significant association with the disease and increased association with both TG13 and 470V.

Haplotypes

Haplotypes could be determined in 71 of the unrelated patients with respect to the alleles at the polymorphic loci (GATT)_n, T_n, and 470M/V, rare polymorphisms detected in the course of mutation screening, and mutations. Also, the (TG)_m alleles associated with the T5 and T9 alleles were determined. We later refined the haplotypes by analysing three microsatellite loci. A large number of different haplotypes were observed: 22 among the total of 34 with mutations and 53 among the total of 112 without mutations.

The T5 IB chromosomes also showed great variation as compared to normals. There were 10 T5 chromosomes among the 200 normals, and they were all on GATT7-TG11-470V background. Moreover, they had in total only three different haplotypes with respect to the three microsatellite polymorphisms IVS8CA, IVS17bTA, and IVS17bCA: 14-30-13, 17-30-13, and 17-33-13 with frequencies of 3, 3, and 4, respectively. In contrast, T5 IB chromosomes were mostly on backgrounds GATT7-TG11-470M (five of 12) and GATT6-TG13-470V (four of 12). The remaining three haplotypes were observed once: GATT6-TG11-470M, GATT7-TG11-470V-1001+11T-2694T-4002G, and GATT7-TG13-470V-F1052V-2694T. When the three microsatellites were also taken into account, no two of the T5 IB chromosomes had the same haplotype.

Comparison of the genotypes of the sibs

We compared the CFTR haplotypes of the patients to their sibs (table 2). All of the four affected sib pairs shared both of their haplotypes, indicating that both of the CFTR chromosomes contributed to the phenotype. The genotypes were T9-N1303K homozygous, T9/T9-N1303K, T9/T9-3272-26A>G, and T7/T5. However, six affected-healthy sib pairs also shared both of the haplotypes, excluding the gene as the sole locus responsible for the disease phenotype. The genotype of one pair was T7/T7-CFTRdele19, three were T9/T9, and two were T9/T7.

Identity by descent

The haplotypes were evaluated to assess identity by descent. In total, nine patients exhibited haplotype homozygosity, three with mutations and six with no mutation detected. The family of one of them had declared parental consanguinity and of four others had denied it. Patients from 26 other families who had claimed parental consanguinity (including 18 first cousin marriages) were not homozygous.

DISCUSSION

CFTR mutations were identified in 23.3% (34/146) of the CFTR chromosomes in our 73 unrelated idiopathic bronchiectasis patients. This frequency is very high, as the carrier frequency in our population was assessed as 1/50.¹⁵ It was intriguing that F508del was not observed in our IB patients, although it is the most common mutation (23.5% of the mutant chromosomes) in our CF patients.¹⁵ This mutation comprised three of the mutant alleles among the total of 19 detected in the adult Italian disseminated bronchiectasis patients,³⁰ three of 16 in the French patients,² and two of seven in the Greek mostly adult patients.⁵ Conversely, three of the mutations in our IB patients (3272-26A>G, CFTRdele19, and −33A>G) were not carried by any of our CF patients¹⁵ (this study). Also, none of the mutations detected in our IB patients was found in the French, Italian, or Greek IB patients. This was surprising since N1303K and 3272-26A>G are both common CF mutations in most Mediterranean countries. K68E, the most common mutation in our IB group, was observed only once in the 166 CF chromosomes we had analysed. It manifested a very mild phenotype.¹⁵ It has been reported in only one other person, a CF patient in north eastern Italy.³¹ The next most common mutation/gene variation, −33A>G, had been identified on a CBAVD chromosome, out of 159 CBAVD, 376 CF, and 238 normal chromosomes, and was proposed to be either a rare polymorphism or a mutation that affected the regulation of the gene.²² We detected it in seven of the 146 IB chromosomes, but not in any of the 41 CF no mutation chromosomes or 44 control samples. Therefore, we propose that it is a pathogenic mutation, but perhaps associated with CBAVD, IB, or mild CF. It was associated in trans with either T5 (two patients), T7 (four patients), or T7-CFTRdele2,3 (one patient), but in no case with T9. The next most common mutations were N1303K and CFTRdele19, both of which manifest severe CF phenotypes.^21,32 These five mutations make up 85.3% (29/34) of all mutant IB chromosomes, and 20 of these 29 chromosomes were on TG9-T9 background. It is worth mentioning here that in our population F508del is mostly on T7 background: 38 of the total 39 F508del chromosomes we had analysed were on this background.¹⁵ Also interesting was that all of the three 3272-26A>G mutant alleles in our IB patients were on T9 background in contrast to the European alleles which were all on T7 background.^33,34 All of these observations point to a role of T9 in the aetiology of IB.

In assessing the significance of T9 and T5, we compared their frequencies to those of T7, since it is the most common allele world wide and has not been implicated in any kind of pathogenesis. Alleles T5 and T9 both showed a highly significant association with IB, and 470M was similarly associated with the T9 IB chromosomes. T5 was shown to be high also in the Italian patients (mean age 53 (SD 15.8)),⁴ but not in two French and one Greek mostly adult patient group.^2,5,35 In addition, the T5 chromosomes in our IB patients were mostly on backgrounds TG11-470V and TG13-470M, whereas the normal T5 chromosomes were all on TG11-470M. Thus, the IB chromosomes all deviated from TG11-470M. This was not surprising, since the association of both 470V and a low number of TG repeats have been shown to lead to lower gene activity.³⁶ The 470M allele leads to a higher protein activity, 470M protein having 1.7-fold intrinsic chloride channel activity compared to that of 470V in transfected cells, and the lower the (TG)_m repeat number, the less the proportion of the mRNA lacking exon 9 sequences transcribed from T7 CFTR chromosomes in transfected cells. Noone et al³⁷ studied a patient who had CF type lung disease with normal to borderline sweat chloride values and was homozygous for haplotype T5-TG12-470V. She had defective CFTR mediated chloride conductance in epithelia.

In addition to confirming the previous findings on increased frequencies of CFTR mutations and T5 in IB patients,^1–5 this report highlights T9, an allele that had not been reported previously in association with any disease. Despite the fact that this allele results in normal transcripts (not lacking exon 9), it is not the most common allele in the normal population world wide, thus is not the most common allele. The high frequency of T9 chromosomes in our patients cannot be attributed to a possible association with an as yet unidentified common mutation, because the allele was on a large number of different haplotypes. We also observed a significant association of 470M with the T9 no mutation IB chromosomes. Molecular studies are necessary to elucidate the basis of the pathogenesis.

Two opposing hypotheses could be proposed to explain the role of T9 in the aetiology of IB. T9 could be a predisposing genotype that does not lead to disease on its own, but leads to IB when in association with defects in an as yet unidentified gene. Alternatively, T9 could be an attenuator for CF instead of a mutation with a role in the aetiopathogenesis of IB. A higher CFTR activity conferred by T9 (more so in association in cis with 470M) could dampen the effect of an in cis mutation and manifest a milder form of CF. This fits in well with the model that mild CFTR mutations lead to milder forms of disease, such as disseminated bronchiectasis and obstructive azoospermia.³⁸ As for the no mutation patients, T9 would be expected to compensate for defects in the hypothetical gene. The hypothetical gene would be expected to exert its effect in an autosomal recessive fashion, and the frequency of its defective form in the population would be lower than CF mutations. IB associated with this gene would thereby be noticeable only in populations with high consanguinity, similar to the Turkish population.

The attenuator hypothesis would explain why 12 of the patients developed a more CF-like disease several years after IB diagnosis. One such patient had mutations on both CFTR chromosomes (−33G>A/CFTRdele2,3). He had the lower left lobe removed at the age of 4 and was referred to our clinic at the age of 5 with borderline sweat test values of 52 and 50 mEq/l. Eighteen and 20 months later the values were high (64 and 63 mEq/l, respectively), and he had developed CF-like gastrointestinal problems. The remaining five patients with two mutations had normal sweat test values (<40 mEq/l) and no gastrointestinal complaints at the ages of 7, 11, 13, 13, and 14. It will be interesting to follow whether these patients also develop CF-like clinical findings in the future.

Identification of an IB modifier gene would be of much clinical value, as no gene other than CFTR has been implicated in the aetiology of this common disease. A modifier gene has already been identified for idiopathic chronic pancreatitis associated with CFTR mutations.³⁹ The four IB sib pairs and their six unaffected sibs would give sufficient genetic information in a genome scan study aiming at the identification of the locus for the hypothetical gene. Identification of the gene would also increase our knowledge on the pathogenesis of CFTR defects and shed light on other diseases associated with CFTR.