Background: Primary ciliary dyskinesia (PCD) is characterised by recurrent infections of the upper respiratory airways (nose, bronchi, and frontal sinuses) and randomisation of left–right body asymmetry. To date, PCD is mainly described with autosomal recessive inheritance and mutations have been found in five genes: the dynein arm protein subunits DNAI1, DNAH5 and DNAH11, the kinase TXNDC3, and the X-linked retinitis pigmentosa GTPase regulator RPGR.
Methods: We screened 89 unrelated individuals with PCD for mutations in the coding and splice site regions of the gene DNAH5 by denaturing high performance liquid chromatography (DHPLC) and sequencing. Patients were mainly of European origin and were recruited without any phenotypic preselection.
Results: We identified 18 novel (nonsense, splicing, small deletion and missense) and six previously described mutations. Interestingly, these DNAH5 mutations were mainly associated with outer + inner dyneins arm ultrastructural defects (50%).
Conclusion: Overall, mutations on both alleles of DNAH5 were identified in 15% of our clinically heterogeneous cohort of patients. Although genetic alterations remain to be identified in most patients, DNAH5 is to date the main PCD gene.
Statistics from Altmetric.com
Primary ciliary dyskinesia (PCD) (OMIM 242650) or immotile cilia syndrome is a clinically heterogeneous disease that affects all ciliated cells in 1 in 20 000 live births (1/12 500 to 1/30 000).1 Defective ciliary function in the respiratory system impairing mucociliary clearance results in chronic sinusitis and upper respiratory tract infections, leading later to bronchiectasis and nasal polyps.2 When PCD is associated with situs inversus, which occurs in 50% of patients, the condition is called Kartagener syndrome (KS) (OMIM 244400).3
The most common ultrastructural abnormalities observed in PCD involve the outer dynein arm (ODA) and/or inner dynein arm (IDA), which can be absent or shorter. Dynein arms are multisubunit complexes composed of several light, intermediate and heavy chains, encoded by distinct loci dispersed throughout the genome. Other abnormalities may include microtubule transpositions and radial spoke defects.4
PCD, which generally follows an autosomal recessive inheritance pattern, is also highly genetically heterogeneous as first suggested by linkage studies.5–8 So far, five PCD-causing genes have been identified including the two heavy axonemal chain dyneins DNAH5 and DNAH11, and the intermediate chain axonemal dynein DNAI1.9–15 Another candidate gene, encoding thioredoxin-nucleoside diphosphate kinase TXNDC3, was recently characterised in a few patients.16 An additional gene, RPGR, is responsible for an X-linked form associated with retinis pigmentosa in one family.17
Previous studies identified mutations in either DNAI1 or DNAH5 in approximately 38% of a cohort of mostly preselected PCD/KS patients. Of the five currently known genes associated with PCD, the DNAI1 gene has been the object of most studies,9–12 17 18 although mutations in the coding or splicing sequences have been found in only 10% of mostly preselected PCD patients and more recently in only 2% in a cohort of unselected individuals.18 In the current study we assessed the frequency of mutations in the DNAH5 gene in a large cohort of PCD patients without any preselection based on axoneme ultrastructure and/or genotype. DNAH5 had been identified using homozygosity mapping and candidate gene approach.19 This 79 exon gene (with one alternative first exon) codes for a heavy chain that localises in the outer dynein arm and is the homolog of the dynein γ-heavy chain of Chamydomonas reinhardtii. Previous studies found DNAH5 mutations in 28% of the analysed families with PCD,15 and a total of 42 mutations have been reported (table 1).14 15 20 Recessive homozygous mutations in Dnahc5, the mouse ortholog of DNAH5, confirmed the involvement of this gene in PCD. Homozygous embryos had normal organ situs in 25%, situs inversus totalis in 35%, and heterotaxy in 40%. Embryos with heterotaxy had complex structural heart defects.21
The rationale to assess the DNAH5 gene in an independent cohort of patients included: (1) the DNAH5 protein is localised within the ODA of the ciliary axoneme arm which is often (60–70%) absent or shortened in PCD patients22 23; (2) to date, DNAH5 mutations represent the most common known cause of PCD (28%). However, prevalence of DNAH5 mutations found in PCD might have been biased since patients were mostly recruited through preselection for linkage to the DNAH5 locus15 or mainly for ODA defect.14
PATIENTS AND METHODS
Case recruitment and selection
We collected blood samples (n = 108) from patients and families (unaffected parents and affected/unaffected siblings). Patients originated from various geographical locations as described previously but were mainly Caucasian (92%).18
Diagnosis of PCD/KS was based on criteria previously established.18 Among the 108 patients, inclusion criteria were fulfilled in the 89 individuals that constituted the study population (supplemental table E1) including seven families with a declared parental consanguinity. Fifty-five per cent of individuals had situs inversus (Kartagener syndrome). Ultrastructural information on electron microscopy of ciliary axonemal section was available for 74% (n = 66) of the patients. Among the patients showing dynein arms defects, 50% had a combined ODA+IDA defect, 18% had an isolated ODA defect, and 12% had an isolated IDA defect. The remaining 14% had other ultrastructural features, whereas 6% had normal ultrastructure of the ciliary axoneme.
Written informed consent was obtained from all patients, and the research protocol was approved by the ethics committee of the respective institutions.
Polymerase chain reaction amplification
Genomic DNA was isolated from peripheral blood using standard extraction procedures. The entire DNAH5 gene was analysed including the alternative first exon and the 5′- and 3′- untranslated regions (UTR). Specific primer pairs were designed for the 80 exons at least 50 bp away from exon limits to allow polymerase chain reaction (PCR) amplification and sequence analysis of the entire exon and the flanking intronic splicing sites.
Denaturing high performance liquid chromatography
Denaturing high performance liquid chromatography (DHPLC) analysis was performed as described previously.18 We analysed all the samples included in the cohort for all the exons of DNAH5, except those analysed by direct sequencing (see sequence analysis for details). Briefly, DNA heteroduplexes were produced by denaturation of PCR products at 95°C for 10 min, followed by a slow and gradual annealing of single strands from 95°C to 25°C over a 30 min period. Heteroduplexes were resolved from homoduplexes using a WAVE 3500 HT DNA DHPLC fragment analysis system and with the Navigator raw data interpretation software (Transgenomic, Omaha, Nebraska, USA).
All changes in DHPLC were further confirmed by direct sequencing (Applied Biosystems, 3130 xl Genetic Analyzer, Foster City, California, USA). In addition, all exons previously described to harbour a cluster of mutations (34, 50, 63, 76 and 77)15 or with isolated mutations (14, 25, 28, 32, 48, 53, 62, 75)14 were directly sequenced. The entire DNAH5 gene was directly sequenced in patients from families in whom consanguineous marriage had been documented (n = 7), and in patients with mono-allelic mutation. Exons from homozygous regions that had been identified by single nucleotide polymorphism (SNP) genotyping were also analysed by direct sequencing in the respective patients.
Whenever a missense variant was found, the possibility that we were dealing with a non-pathological polymorphism was considered by, in the following order: (1) interrogating the SNP database (www.ncbi.nlm.nih.gov/dbSNP/); (2) verifying the amino acid conservation among species using the UCSC genome browser (www.genome.cse.ucsc.edu/); (3) screening for the variant in at least 160 chromosomes of same ethnic origin by direct sequencing or minisequencing (Pyrosequencing, Biotage AB, Uppsala, Sweden). All identified mutations were confirmed by repeating sequencing on the original DNA sample source tube.
SNP analysis on the six linkage disequilibrium (LD) blocks encompassing gene DNAH5 in Caucasians (www.hapmap.org) was performed as described previously18 by minisequencing (Pyrosequencing, Biotage AB, Uppsala, Sweden). The following SNPs were genotyped: rs10513151, rs6860899, rs17278234, rs4702001 and rs4702002 (LD block 1); rs10057950, rs6859484, rs1445823, rs7721634 and rs962138 (LD block 2); rs11748811, rs13170062, rs1900162, rs17275618 and rs6554820 (LD block 3); rs980897, rs795542, rs924630, rs1596790 and rs795540 (LD block 4); rs6554810, rs4701984, rs6554811, rs6867796 and rs4351149 (LD block 5); rs3734110, rs6862469, rs10513151, rs1502046 and rs2896104 (LD block 6). Briefly, PCRs were performed using 50 ng of template DNA and checked for size and yield on a 2% agarose gel. Amplicons were purified and minisequenced using 30 pmol of sequencing primer per reaction under standard conditions according to the manufacturer’s instructions (Pyrosequencing, Biotage AB, Uppsala, Sweden) in a Pyrosequencing PSQ HS 96 System. Data were captured using PSQ HS 96 SNP software (Pyrosequencing, Biotage AB, Uppsala, Sweden).
In vitro splicing assay
When RNA was not available, an in vitro splicing assay was used to characterise the effect of the splicing mutation on the DNAH5. As described previously,24 the wild-type and the mutated constructs were generated by PCR amplification using oligonucleotides in DNAH5 introns 6–7 and 9–10 (6195-F: ATTGGAAGCA TGGAATACGC/6195-R: TTGAACTGA GCCAATGTGGT), 23–24 and 26–27 (170-F: TGGGTTAGAG GGCAATAAGC/170-R: GGGCCCTCTAT CTTACAAAGAA) and 33–34 and 35–36 (167-F: AGGAAACAATGAGAA ACGTGAC/167-R: AAAGAGCCTATAAACCCTAAGAGAC). The PCR fragments were cloned into the pcDNA3.1/V5-His TOPO-TA mammalian expression vector (Invitrogen, San Diego, California, USA), and their integrity verified by sequencing. The constructs were transfected into HEK-293T cell line using Lipofectamine 2000 (Life Technologies, Basel, Switzerland). Total RNA was extracted using the RNeasy kit (Qiagen, Valencia, California, USA), and reverse transcriptase (RT)-PCR was performed (Invitrogen, San Diego, California, USA). The primers used for the PCR as well as for sequencing were as follows: Exon 7F: GAGCTGGAGCACTGGAAAAA and Exon 9R: CAGAGATG TGATCTTCTCAGAGGT. Exon 24F: ACCGGAGTGAGAT GGAAAAC and Exon 26R: CGCTAGCCATTGGACCAT. Exon 34F: CACCACGAGGGATCTGAGTT and Exon 35R: CCTGTCAGTGCAGCCTAAAA.
Mutational analysis of DNAH5
We carried out mutation analysis using a combination of DHPLC and sequencing on a cohort of 89 patients with confirmed PCD/KS, recruited without preselection. Analysis of all coding exons, including splice site junction and UTR regions, revealed 18 novel DNAH5 mutations in 16 patients in addition to six previously reported mutations that we also identified in six patients.15 We observed missense (43%), nonsense (40%), indels, and splice site mutations (17%) (table 2 and supplemental table E2 for clinical details).
Localisation of these mutations within the specific domains of DNAH5 is depicted in supplemental fig E1. We identified two mutations in 15% of the patients, while 18% had at least one mutated allele. The 18 novel variants included 10 missenses (p.I370F, p.F540L, p.N549D, p.D556G, p.R1716W, p.R2630W, p.V2829F, p.R2833G, p.E3455H and p.R3539C) found in seven patients (tables 2 and 3). Four missenses—p.I370F, p.F540L, p.N549D, p.D556G—were localised within the N-terminal region 1 domain (DHC_N1) of DNAH5 which is known to form dimers with other heavy chains, and with intermediate chain–light chain dynein complexes involved in a basal cargo binding unit.25 The functional importance of this domain has been demonstrated recently in a mouse model in which a targeted homozygous deletion resulted in PCD as well as structural heart defects in 40% of the embryos.21
A substitution at amino acid 1716 was previously reported as a pathogenic alteration.15 The novel substitution p.R1716W, which is not conservative, may similarly affect the function of the protein as described by the other study. The following changes that cause missenses within conserved functional domains of the protein—p.R2630W lying within the AAA3 domain (ATPases associated diverse cellular activities)26; p.V2829F and p.R2829G lying between the AAA3 and AAA4 domains; p.E3455H located in the microtubule binding site (MTB); p.R3539C localised between the MTB and the P5 loop—are strongly suspected to be disease causing mutations. Finally, the previously identified frameshift mutation p.D4398EfsX16 was found in two patients, either as a compound heterozygote with the known missense p.S2264N mutation (patient 8177), or as heterozygote (mono-allelic mutation) in patient 6476. The previously described missense mutation p.R2501P was found in one patient (6191) who carries two other novel missense mutations (p.F540L and p.R3539C).15
We have also identified five novel nonsense mutations (p.R224X, p.Q1450X, p.K1853X, p.R1883X and p.E4133X) in five patients and the previously described nonsenses p.R1761X, p.R2677X and p.R4496X were found in three other patients.15 Of the 16 patients in whom DNAH5 mutations were found in this study, three patients had mono-allelic mutations (table 2).
In order to verify that any homozygous mutations were not missed because of the detection technical limitations of the DHPLC assay, we genotyped a series of SNPs in the six linkage disequilibrium blocks (LD) that encompass the whole length of the DNAH5 gene (according to HapMap study in CEU sample, www.hapmap.org). We genotyped patients who could carry such a hidden homozygous mutation because they (1) were born from consanguineous marriages, or (2) had no identified heterozygous variants within the region(s).
As previously described,18 a total of 30 informative SNPs (n = 5 per LD block) were selected for genotyping (data not shown). We then directly sequenced all the exons that belonged to LD blocks in which genotyping was showing total homozygosity. None of the genotyped patients showed heterozygosity in all six LD blocks. Furthermore, individuals (total n = 72) without heterozygosity in one or more LD blocks (n = 16 for block 1, n = 31 for block 2, n = 28 for block 3, n = 15 for block 4, n = 22 for block 5, and n = 21 for block 6) were directly sequenced for the respective exons of DNAH5. No additional mutations were detected.
Characterisation of three novel splice site mutations
In three patients (167, 170 and 6195), we detected three novel heterozygous variants (c.5710-2A>G, c.3876_4053+158del, and c.1089+1G>A, respectively) predicting altered splicing of their respective exons and subsequent aberrant transcripts of DNAH5. To test whether these nucleotide changes could alter the sequence of mature RNA, in vitro splicing assay of the mutant and wild type (WT) constructs in subsequent HEK-293T cells were performed (fig 1, supplemental table E3).
Variant c.5710-2A>G resulted in a PCR fragment of 220 bp instead of the 238 bp observed in WT (fig 1-A1). The sequence of this 220 bp transcript cDNA revealed that the regular acceptor splicing site was not recognised, but was replaced by a cryptic splicing site within the following exon. This results in an in-frame deletion skipping the beginning of exon 35 (p.C1904-K1909del) (fig 1-A2, 1-A3), which might result in a deleterious functional effect on the protein.
In vitro splicing analysis of the c.3876_4053+158del mutation in patient 170 (fig 1-B1) resulted in a single fragment of 180 bp, while two products of unequal intensity were obtained in the WT situation (a weak band at 180 bp and a stronger one at 399 bp) (fig 1-B2). Sequencing of the 180 bp fragment resolved a p.E1279-K1351del, while the 399 bp was confirmed as the WT product (fig 1-B3). The relatively lower intensities of the 180 bp fragment amplified may reflect the low abundance of the partially skipped exon 25 of DNAH5 in non-pathological conditions.
Two different outcomes resulted from the c.1089+1G>A mutation splicing assay analysis in patient 6195 with abnormal transcripts of equal intensity, while the wild-type transcript was only obtained in WT construct (fig 1-C1). In mutant constructs, one transcript (the shorter) revealed a deletion of the entire exon 8 (219 bp), resulting in p.T326-P363del, while the other predicts a premature termination of translation secondary to disruption of the original reading frame (421 bp) (p.T326VfsX25) (fig 1-C2, C3).
Since the initial characterisation of DNAH5,14 60 mutations have been detected in gene DNAH5 in a total of 223 families, including the ones reported here.15 We find that DNAH5 mutations are to date the most common of the known causes of PCD/KS with ODA±IDA defect which is partially in agreement with Hornef et al.15 Our findings showed that, as suggested by Hornef et al, mutations in DNAH5 are the most frequent causes of PCD in ODA, but the frequencies we found are lower.
In contrast to previous studies, our cohort was screened without any type of preselection based on subtype of phenotypes.18 Patients included had abnormal ultrastructure and/or multiple clinical characteristics linked to PCD. This cohort is thought to be representative of PCD patients in the Caucasian population since the rate of occurrence of the various ultrastructural defects did not deviate from that in previously reported studies.9 18
In three out of the 16 patients carrying mutations, we could not find the second altered allele although all the exons (including the alternatively transcribed first exon), and the entire length of the 3′- and 5′- known UTR regions were directly sequenced. As described previously,18 the second mutation in these three patients could lie in regions that were not screened such as introns, potential functional/regulatory sequences that are located remotely from coding regions, large deletions or insertions that are not detectable by the techniques used in this study.
Another hypothesis would be a digenic or triallelic inheritance as described in some cases of Bardet Biedl syndrome.27 Despite the mutation screening on the DNAI1 gene, we cannot completely rule out this hypothesis before other genes encoding for dyneins contributing to the ODA complex have been thoroughly screened for mutations. Finally, DNAH5 may not be the gene causing PCD in these three patients with only a single mutation.
Overall in our cohort, alterations of the sequence of DNAH5 mutations are responsible for PCD in 15% of the 89 unrelated patients analysed here. This contrasts with the study of Hornef et al,15 in which DNAH5 mutations were identified in 28% of PCD patients. Most PCD patients have ODA defects (≈60–70%),22 28 29 and to date DNAH5 was the most frequently mutated gene in patients with a documented ODA defect. Considering ODA defect alone as a distinct phenotype from the combined ODA+IDA, we found that 33% of our ODA patients carry a mutation in DNAH5 which is less than expected according to the study of Hornef et al (49% of the patient with documented ODA defects have at least one DNAH5 mutation).15 In our cohort, it represents 24% of patients with ODA+IDA defect who have mutations in DNAH5 and 27% when considering ODA±IDA ultrastructure defect. No mutations are found in patients with EM defects different from ODA±IDA.
All the patients studied here were also included in our previous DNAI1 mutation analysis in a cohort of 104 individuals.18 Taken together, 89 individuals have been screened for both the DNAI1 and DNAH5 genes (supplemental table E1). In 17% of all the patients we were able to identify two mutations in either DNAI118 or DNAH5 (table 2). There were 22% of patients harbouring at least one mutant allele in either one of these two genes. Our finding deviates from a previous study in which mutations in either DNAI1 or DNAH5 genes represented 38% of all PCD patients.9 Since both cohorts were mainly composed of Caucasian individuals, the inclusion of 21 consanguineous families and the preselection by linkage to DNAH5 locus and/or ultrastructural phenotype may explain this deviation between studies. According to our results the molecular diagnostics of PCD should not rely only on DNAI1 and DNAH5 genes, since diagnostic sensitivity does not exceed 17%.
Most of the candidate genes for PCD—mainly heavy and intermediate chains for axonemal dynein—have been investigated with low success rates despite the large number studied. This means that the aetiology of PCD and Kartagener syndrome remains unknown in most cases in our cohort (83%). The identification of most, if not all, genes involved in the pathogenesis of PCD, and estimation of the real impact of each of these remains necessary to advance our knowledge of the molecular basis of PCD. This is mandatory for improved diagnostic testing, finally tuned medical management, and innovative treatment of this clinically and genetically heterogeneous disease.
Authors are grateful to all the patients and their families for participation in this research study as well as Dr M Jaspers and Professor M Jorissen, Leuven, Belgium for their contribution in the analysis of the electronic microscopy analysis. The authors would also like to thank warmly the following physicians who contributed to this study by recruiting one to several patients: A Brauchlin, A Bitton, R Aebi, C Barazzone, E Bierens de Haan, G Schwarz, J-Y Berney, P-A Guerne, S Tarab, T Rochat, Geneva; P Eng, M Künzli, Aarau; H Wacker, Allschwill; W Schäppi, Andelfingen, J Hammer, M Ruthishauser, Basel; W Graf, Bethlehem; E Horak, S Schibli, HM Schöni, Davos; K Brugger, Deissenhofen; P Terrier, Delémont; B Zimmermann, Emmenbrüke; O Brändli, Faltigberg-Wald; J C Heili, Grub; D Stefanutti, La Chaux-de-Fonds; H-U Dubach, Langenthal; M Gehri, A Calame, P Marguerat, Lausanne; P Eng, Luzern; J-G Frey, Montana; C Meili, RM Kaelin, Morges; A Wallmeroth, Muttenz; B Thiévent, Porrentruy; M Bigler, Rorschach; O Schoch, J Barben, St Gallen; A Brauchlin, J Wildhaber; H Walt, Zürich; (Switzerland); BA Afzelius, Stockolm (Sweden); M Jorissen, Leuven; L van Maldergem, Loverval (Belgium); C Le Pommelet, Avignon (France); S Ala-Mello, Helsinki (Finland); G Piatti, Milano (Italy); M Wessels, Rotterdam, (Netherlands); M Armengot, Valencia (Spain); H Cox, Birmingham (UK); D Chitayat, Toronto; R Hodder, Ottawa (Canada); R Gershoni, Haifa (Israel); E Reich, New York, D Schidlow, Philadelphia (USA).
Review history and Supplementary material
▸ Additional tables and figure are published online only at http://jmg.bmj.com/content/vol46/issue4
Funding: This work was supported by grants from the Swiss National Science Foundation (#3200BO-105838, to JLB), the Carvajal ProKartagener Foundation (to JLB), and the Société Académique de Genève on behalf of Fonds Dr E Rapin (to MF and JLB). LB was supported by a Marie Heim-Vögtlin grant from Swiss National Science Foundation. The Swiss Group for Interstitial and Orphan Lung Diseases was supported by the Swiss Academy of Medical Sciences, the Swiss Respiratory Society, the Swiss Pulmonary League, the Geneva Pulmonary League, Geneva University Hospitals, Bern University Hospital, the Carvajal ProKartagener Foundation, GlaxoSmithKline, Actelion and Aventis.
Competing interests: None declared.
Patient consent: Obtained.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.