Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

In primary ciliary dyskinesia (PCD), also known as immotile cilia syndrome (mendelian inheritance in man number [MIM] 242650), recurrent infections of the respiratory tract, sinus, and middle ear are caused by a reduced mucociliary clearance (1). Some patients have reduced fertility, which is explained by dysmotility of spermatozoa or Fallopian tubes. By electron microscopy, in most of the patients with PCD ultrastructural defects of the cilia can be detected (2). The structural defects consist of total or partial absence of dynein arms (70 to 80% of cases), defects of radial spokes, nexin links, or general axonemal disorganization with microtubular transposition. PCD represents a heterogeneous group of genetic disorders affecting one of 20,000 to one of 60,000 individuals at birth (3). Inheritance in most cases is autosomal recessive, but there have been anecdotal reports of autosomal dominant and X-linked inheritance (4). PCD with situs inversus is also referred to as Kartagener syndrome (MIM 244400). In families with Kartagener syndrome, all affected individuals have PCD, but only half of the affected siblings have a complete situs inversus due to randomization of the left/right body axis.

Cilia and flagella are found throughout the plant and animal kingdom. The axoneme is the core structure of a cilium or flagellum, and consists of a set of nine doublet microtubules arranged cylindrically around a pair of singlet microtubules. Chlamydomonas reinhardtii, a unicellular alga with two flagella, contains an axonemal structure highly similar to that of human respiratory cilia and sperm tails. Several dysmotile strains of Chlamydomonas have been reported, exhibiting ultrastructural defects identical to that seen in human PCD. By analyses of these mutant strains, mutations have been identified in many different genes encoding axonemal dyneins, including light (8 to 55 kD), intermediate (45 to 110 kD), and heavy chains (400 to 500 kD), which cause PCD in Chlamydomonas (5). These studies suggest that mutations in several different genes might cause PCD in humans, rendering a possible explanation for the high degree of heterogeneity observed in PCD. The hypothesis that homologous genes to Chlamydomonas might be responsible for PCD was recently confirmed by the detection in a patient with isolated PCD of two loss-of-function mutations in the human gene DNAI1 related to C. reinhardtii dynein encoding the intermediate chain IC78 on chromosome 9q (6). In the same study, genetic heterogeneity was demonstrated by the absence of linkage in five other families with PCD.

To detect an additional gene responsible for PCD we applied a homozygosity mapping strategy. To avoid genetic heterogeneity we studied one large, consanguineous family of Arabic origin with PCD and absence of outer dynein arms on electron microscopy of respiratory cilia. In this study, we report the results of a whole-genome scan and the localization of the gene to a region of homozygosity by descent on chromosome 5p15-p14 with a parametric multipoint logarithm of odds ratio (LOD) score of maximum LOD score (Zmax) = 3.51. Within the critical disease interval, we localized the candidate gene DNAH5 encoding an axonemal heavy dynein chain of the outer arm. DNAH5 is an excellent functional candidate for human PCD because mutations in the homologous Chlamydomonas gene are supposed to cause the slow swimming oda2 mutant with ultrastructural abnormalities identical to those observed in the presented family with PCD (7).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects

In a large consanguineous Lebanese family with occurrence of PCD, blood samples were obtained on the basis of informed consent from 10 family members, including four affected individuals. Parents were first cousins. The lack of direct disease transmission, the high degree of consanguinity, and equal sex distribution of affected individuals indicated an autosomal recessive inheritance pattern (Figure 1). Individuals were classified as affected by PCD on the basis of (1) recurrent infections of the respiratory tract, (2) complete lack of ciliary motility by direct visualization of respiratory cilia, and (3) absence of outer dynein arms on electron microscopic examination. The only exception was individual III-8 where the affected status was based on recurrent infections of the respiratory tract and presence of complete situs inversus. In the affected individuals, respiratory symptoms consisted of hypoplasia/agenesis of frontal sinus (Figure 2A), chronic sinusitis, chronic otitis media, and recurrent respiratory infections. All affected individuals developed respiratory symptoms within the first years of life, leading to bronchiectasis in individuals III-2, III-3, and III-4. In individual III-2, bronchiectasis and severe destruction of pulmonary segments had led to a right middle lobectomy at the age of 14 yr (Figure 2B). Chest radiography showed normal cardiac and visceral situs in all but two affected individuals (III-4 and III-8). These individuals had complete situs inversus, a main feature of Kartagener syndrome (Figure 2C). Direct visualization of respiratory epithelial ciliary motility by light microscopy revealed in III-2, III-3, and III-4 a complete lack of ciliary motility. Electron microscopic examination of at least 20 cross-sections of axonemes in one biopsy sample of respiratory cilia showed absence of outer dynein arms in all examined cilia of these affected individuals (Figure 2D). Otherwise, the ultrastructural composition of ciliary structures appeared normal with nine outer doublet microtubules surrounding a central pair of singlet microtubules (Figure 2D). Siblings were considered as unaffected if direct visualization of respiratory epithelial ciliary motility by light microscopy was normal. Unaffected siblings showed a normal ultrastructural composition of axonemal anatomy by electron microscopy (Figure 2E).

View larger version (61K): [in this window] [in a new window]   Figure 1.   Haplotypes and recombinations in the Arabic kindred with PCD. Genotypes are shown in telomere-to-centromere order (top to bottom) for microsatellite markers D5S1957, D5S208, D5S2095, D5S630, D5S667, D5S1987, D5S1954, D5S2114, D5S2031, D5S2074, D5S648, D5S2113, D5S502, D5S419, D5S2061, D5S477, D5S674, D5S663, D5S426, and D5S2061 on chromosome 5p. The distance given adjacent to each marker represents the distance of each marker to the p terminal end of chromosome 5. Marker positions were taken from references 10 and 11. Inferred alleles are shown in parentheses. For genotypes represented by a question mark, no data were available. Haplotypes are assembled by minimizing recombinants, and the chromosomal region cosegregating with the disease locus is represented by a solid bar. Indeterminate haplotypes are shown by a thick line instead of a bar. The consanguineous relationship is depicted by a double line. Circles denote females; squares, males; solid symbols, affected individuals; and open symbols, unaffected individuals.

View larger version (94K): [in this window] [in a new window]   Figure 2.   Clinical findings of PCD in the Arabic inbred kindred. (A) Radiograph of the skull of proband III-3. Note the absence of frontal sinus at age 17 yr and the reduced radiolucency of the right paranasal sinus as a result of sinusitis. (B) Computed tomography scan of the thorax of proband III-3. There is paracardial atelectasis and bronchiectasis caused by recurrent pulmonary infections. The patient has a normal situs solitus. (C) Chest radiograph of proband III-8 at the age of 8 mon. There is a complete situs inversus. (D) Electron micrograph of cross-sections of respiratory cilia (original magnification: ×90,000) from affected proband II-3. Absence of outer dynein arms is observed on all the peripheral doublets. (E) Electron micrograph of cross-sections of respiratory cilia from healthy individual III-7 for comparison. Normal sites of outer dynein arms are indicated by arrows in the left axoneme.

Linkage Analysis

Genomic DNA was isolated by standard methods directly from blood samples (8) or from peripheral blood lymphocytes after Epstein-Barr virus transformation (9). Total genome linkage analysis was performed using 340 microsatellite markers from the Genethon microsatellite set with an average spacing of 11 centimorgans (cM) and with 20 additional microsatellites from the region of 5p for further fine mapping (10, 11). The total genome scan was performed in a subset of the kindred containing four affected (individuals III-2, III-3, III-4 and III-8) and three unaffected individuals (individuals II-1, II-2 and III-5). Markers were scored for homozygosity in the affected individuals. All genotypes were tested with the LINKRUN program for Mendelian segregation (T. Wienker, unpublished data). Two-point LOD score analyses between the disease locus and microsatellite markers were calculated using the programs ALLEGRO (12) and M-LINK from the LINKAGE package (13). Parametric multipoint LOD score calculation was performed using the programs ALLEGRO and GENEHUNTER (14). For the genetic model autosomal recessive inheritance, full penetrance, disease allele frequency of 0.001, and equal allele frequencies for each marker were used. Marker data were taken from Dib and coworkers (11). For graphical representation of pedigree and haplotype data, the program CYRILLIC version 2.0 was used (Cherwell Scientific, Oxford, UK).

Localization and Partial Sequencing of Human DNAH5

Partial sequences of several genes encoding heavy dynein chains have been identified by a polymerase chain reaction (PCR) approach using degenerated primers derived from the highly conserved P1 motif (15). One of these PCR products (HL1 [gene accession number: U61735]) hybridized to a murine chromosomal region on chromosome 15, which is syntenic to human chromosome 5p and 8q (15). During the search for candidate genes for the Cri du Chat syndrome on chromosome 5p, three dynein- related sequences (3h8 [gene accession number: B07656], 4a1 [gene accession number: B07655], and 4a2 [gene accession number: B07656]) were found by exon-trapping experiments on P1-related artificial clone (PAC) 474k3 (16). These data suggested that a gene encoding an axonemal heavy dynein chain potentially resides within the disease interval on chromosome 5p. By a sequence-tagged site (STS) content mapping strategy where we designed primers for HL1, 3h8, 4a1, and 4a2, we showed that these PCR products indeed are located on PAC-clone 474k3, which resides on chromosome 5p15 (data not shown). To test whether these four PCR products belong to one gene encoding an axonemal heavy dynein chain residing on chromosome 5p, we performed PCRs between these PCR products on a human testis complementary DNA (cDNA) library (Clonetech Laboratories, Palo Alto, CA). Oligonucleotide primers were prepared and used to amplify the human testis cDNA by PCR: (A) 5'-TGGCTCAAGCTCTGGGAATGAGCATG-3' (sense strand, bp 33 to 58 of GenBank accession no. U61735) and (B) 5'-GCGATACAAGTCTGGGAAAGACTCGGTG-3' (antisense strand, bp 51 to 79 of GenBank accession no. B07656). Using combinations of primers A and B, one detectable band on 1% agarose gel electrophoresis was obtained. The PCR product was agarose gel purified and sequenced using the big dye sequencing system (ABI-377; Perkin Elmer, Norwalk, CT).

Sequence Analysis of DNAH5

Homology searches were performed by use of BLAST (17). Deduced amino-acid sequences of the gene encoding Chlamydomonas -heavy dynein chain and DNAH5 were aligned with the use of the BLAST-2 program.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Linkage Analysis

The total genome scan was performed in a subset of the kindred containing four affected individuals. This screening procedure found only one informative marker (D5S630) located on the short arm of chromosome 5 for which all affected individuals were homozygous. By further fine mapping of a 34-cM region between markers D5S1957 and D5S2101, a region of homozygosity by descent in all affected individuals was identified (Figure 1). When haplotype analyses were evaluated, a proximal recombination event was detected for marker D5S2095 in affected individual III-4, and a distal recombination event was detected for marker D5S502 in unaffected individual III-7 (Figure 1). These two flanking markers define a critical interval of approximately 20 cM of sex-averaged genetic distance (18). To illustrate homozygosity by descent, we show the PCR products for marker D5S1954 for a subset of the examined pedigree (Figure 3). Markers D5S630, D5S667, D5S1987, D5S1954, D5S2114, D5S2031, and D5S2074 are cosegregating markers compatible with linkage in all affected individuals and all healthy family members of the kindred (Figure 1). Parametric two-point LOD score analysis calculated with the program ALLEGRO yielded the following maximum LOD scores for the cosegregating markers at the recombination fraction  = 0.00: D5S630, Zmax = 2.96; D5S667, Zmax = 3.26; D5S1987, Zmax = 3.21; D5S1954, Zmax = 3.14; D5S2114, Zmax = 3.25; D5S2031, Zmax = 1.18; D5S2074, Zmax = 2.98. Parametric two-point LOD score analysis with M-LINK of the LINKAGE package calculated similar results. Parametric multipoint analysis was performed with the program ALLEGRO, resulting in a maximum LOD score of Zmax = 3.51. Parametric multipoint analysis with GENEHUNTER calculated a maximum parametric LOD score of Zmax = 3.39 within the interval between markers D5S1987 and D5S1954 (Figure 4). One unaffected sibling (III-1) had to be excluded from this calculation due to memory constraints, which explains the slight difference to the calculation with ALLEGRO. Nonparametric maximum multipoint LOD score calculated with GENEHUNTER was Zmax = 10.22. LOD score calculation remained stable when lower allele frequencies (one to three alleles) were used. Thus, the parametric linkage data give significant evidence for a gene locus for PCD on chromosome 5p.

View larger version (61K): [in this window] [in a new window]   Figure 3.   Autoradiograph of PCR products of marker D5S1954 for a subset of individuals of the presented pedigree. Affected individuals (III-2, III-3, III-4, and III-8) are homozygous by descent for the disease-associated allele. Parents and unaffected siblings are heterozygous for this allele.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Parametric multipoint LOD score analysis performed with GENEHUNTER for the presented kindred with PCD. The maximal LOD score was Z = 3.39.

Localization and Partial Sequencing of DNAH5

PCR using primer combination A/B resulted in the amplification of an ~ 1.3-kb PCR product. Together with the remaining sequences of HL1 and 4a2, we obtained by the sequencing of these PCR products a 1.5-kb partial cDNA sequence of DNAH5. Sequence similarity studies with BLAST revealed a high degree of homology to the C. reinhardtii gene encoding the -axonemal heavy dynein chain of the outer dynein arm (Figure 5), thus demonstrating that DNAH5 indeed encodes an axonemal heavy dynein chain of the outer dynein arm. By STS content mapping we could show that DNAH5 is at least partially located on PAC 474k3. According to the maps described by Broman and colleagues (10) and Church and associates (16), DNAH5 is located between markers D5S667 and D5S1954 (http://www-hgc.lbl.gov/human-maps.html). Thus, we give evidence that DNAH5 is localized within the critical disease interval of the 5p locus for PCD.

View larger version (70K): [in this window] [in a new window]   Figure 5.   Partial deduced amino-acid sequence of human DNAH5 in comparison with the C. reinhardtii sequence for the -axonemal heavy dynein chain of the outer dynein arm. The P1 and P2 loops and the potential microtubule binding site (MTBS) are conserved and underlined. Amino-acid identities or conserved residues (+) are shown in the third line.

Sequence Analysis of DNAH5

Homology searches with BLAST found with the dynein gamma chain of the flagellar outer arm (GenBank accession no. Q39575) 297 of 497 (59%) identities and 369 of 497 (73%) positives with the deduced amino-acid sequence of DNAH5 (Figure 5).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Studies in C. reinhardtii mutants suggest genetic locus heterogeneity in PCD. This is supported by the failure of total genome scans that used multiple small PCD-families to establish linkage to a specific chromosomal site (19). Thus two approaches appear to be feasible in order to identify loci or genes involved in PCD. The first approach is a candidate gene approach, which recently was successfully applied, by identification of two loss-of-function mutations in one patient with PCD (6). This approach is feasible in PCD because there are good candidate genes that originate from studies with Chlamydomonas mutants. The second approach is the use of a homozygosity mapping strategy for gene localization (20). This method is a powerful tool because it avoids the use of multiple unrelated kindreds but is restricted by the availability of informative consanguineous families with PCD. We followed the second strategy and studied a large Arabic kindred with PCD containing four affected individuals. An assumption of identity by descent was made on the basis of pedigree information and the rarity of the disease in the general population. Based on data of total genome haplotyping, a search for homozygous regions shared by affected patients led to the identification of the responsible gene locus on chromosome 5p15-p14. Further evidence of linkage was demonstrated by extensive allele sharing among affected individuals. The critical region of PCD is restricted to a 20-cM interval flanked by D5S2095 and D5S502 because of a recombination event that occurred in affected proband III-4 and the unaffected individual III-7, respectively. An additional distal flanking marker is D5S419 because its heterozygosity in all affected children interrupts the region of homozygosity, thus indicating an ancestral recombination. A previous linkage study with 31 multiplex families with PCD found on chromosome 5p a maximum LOD score of 2.2, suggestive of linkage, thus supporting our data of a new gene locus for PCD on chromosome 5p (19). The gene locus on chromosome 5p responsible for PCD in the Arabic kindred defines a new disease variant that is clearly distinct from the known loci on chromosomes 19q13.3-qter and 9p21-p13 (6, 21). This confirms the expected genetic heterogeneity in PCD. Only for the locus on chromosome 9p21-p13 is a responsible gene known (DNAI1), which was found to be mutated in one patient with PCD (6).

The disease phenotype linked to the 5p locus and the disease phenotype linked to mutations in DNAI1 share the same ultrastructural changes seen on electron microscopy of respiratory cilia, which consist of the absence of outer dynein arms (Figure 2D) and is the most prevalent ultrastructural defect in human PCD. The understanding of genetic defects resulting in the absence of outer dynein arms is aided by the C. reinhardtii model. Chlamydomonas outer arm dyneins contain three types of heavy chains, , , and , two intermediate chains, and at least 10 light chains (22). In slow-swimming strains of C. reinhardtii that lack the entire outer dynein arm (oda mutants), mutations in at least 12 independent loci result in the absence of outer dynein arms in Chlamydomonas (23). There were no oda mutants lacking partial structures of the outer arm, suggesting that lack of a single component results in the failure of assembly of entire outer arms. The genes encoding for axonemal -, -, and -heavy dynein chains and the intermediate dynein chains IC78 and IC69 have been mapped to the genetic loci of oda mutants in C. reinhardtii (24). Subsequent mutational analysis of oda mutants identified mutations in most of these genes (7, 24, 27). Recently, in a patient with isolated PCD and absence of outer dynein arms, loss-of-function mutations in the human gene DNAI1 related to C. reinhardtii-dynein encoding the intermediate chain IC78 were identified (6). Thus, human homologues of C. reinhardtii related genes causing oda mutants appear to be good candidates for human PCD with the absence of outer dynein arms.

Partial sequences of several genes encoding heavy dynein chains have been identified by a PCR approach using degenerated primers for the highly conserved P1 motif to identify genes encoding heavy dynein chains (15). One of these PCR products, called HL1, hybridized to a murine chromosomal region, which is syntenic to human chromosome 5p and 8q (15). By a PCR-based approach, we showed that HL1 and the exon-trapped products 4a1, 4a2, and 3h8 described by Church and coworkers (16) map to chromosome 5p, and that HL1 and 4a2 belong to DNAH5, which encodes an axonemal heavy dynein chain (30). Sequence similarity studies with BLAST revealed a high degree of homology between DNAH5 and the C. reinhardtii gene encoding the -axonemal heavy dynein chain of the outer dynein arm (Figure 5). A specific feature of the heavy chain is the presence of multiple P loop elements involved in -phosphate binding and hydrolysis in adenosine triphosphatases and guanosine triphosphatases (31). The partial deduced amino-acid sequence of DNAH5 contains two of the four P loops that are observed in heavy dynein chains. The P1 loop is completely conserved and the P2 loop differs only in two positions from the C. reinhardtii gene encoding the -axonemal heavy dynein chain (Figure 5). Also the potential microtubule binding site is mostly conserved (25). DNAH5 appears to be an excellent positional and functional candidate gene for human PCD linked to chromosome 5p15-p14 because mutations in the homologous Chlamydomonas gene are supposed to cause the slow-swimming oda2 mutant with ultrastructural abnormalities identical to those observed in the presented family with PCD (7).

In summary, we have identified by a homozygosity mapping strategy a gene locus for PCD with absence of outer dynein arms in one large, inbred family of Arabic origin and localized the candidate gene DNAH5 encoding for a heavy dynein chain within the critical disease interval.