Considerable variability in the clinical and ciliary features of primary ciliary dyskinesia in patients with DNAH5 mutations

Background: Primary ciliary dyskinesia (PCD) is a clinically and genetically heterogeneous disorder, but the relationship between genotype and phenotype is poorly established. We aimed to characterize the detailed clinical characteristics, ciliary phenotype and mutational spectrum of PCD patients with DNAH5 mutations. Methods: Whole exome sequencing followed by targeted copy number variation (CNV) analysis was used to screen 114 Chinese patients with highly suspected PCD. In cases with DNAH5 mutation, detailed demographic and clinical data, as well as ciliary beat pattern and ciliary ultrastructure, were comprehensively reviewed. Results: Thirty patients (median, 15.3 years; range, 0.6 to 43 years), who carried DNAH5 mutation, were identied. We characterized 38 novel and 7 known DNAH5 mutations (10 nonsense, 5 frameshift, 10 splicing, 16 missense mutations, 2 in-frame deletions and 2 large deletions). Despite bearing the same genotype, patients presented a highly variable phenotypic picture, including milder respiratory symptom, normal nasal nitric oxide levels (2/21, 9.5%) and “nontypical” ciliary beat pattern (4/21, 19.0%). Conclusions: DNAH5-associated PCD disease demonstrates dramatic phenotypic heterogeneity, contributing to the challenges of diagnosis. A composite work-up and cautious interpretation of the results are necessary during the PCD diagnosis process. in respiratory cilia that are also mostly immotile or only show ickering movements. Patients usually have the typical clinical phenotype of PCD. In this study, we systematically evaluated the phenotype and genetic characteristics of 30 PCD patients carrying DNAH5 mutations. Intriguingly, mild respiratory phenotype, normal nasal nitric oxide and diverse ciliary beat pattern were identied in our cohort group. For specied patients, IF was performed to investigate outer dynein arm (ODA) perturbations in nasal ciliated cells. It is hoped that the detailed phenotype information identied in our study will help to better delineate the genotype-phenotype relationship of DNAH5.


Introduction
Primary ciliary dyskinesia (PCD) is a rare, heterogeneous ciliopathy resulting in mucociliary clearance failure. Patients typically present with newborn respiratory distress, daily wet cough, chronic nasal congestion, and laterality defects. PCD diagnosis can be achieved by following the diagnostic algorithms that include nasal nitric oxide (nNO) measurements, high-speed video microscopy analysis (HSVA), immuno uorescence (IF), transmission electron microscopy (TEM), and molecular testing [1] . Currently, around 50 genes are known to be associated with PCD [2] .
The relationship between genotype and phenotype in PCD has been considered a matter of critical importance [3] , as better understanding of genotypephenotype correlations helps to promote more accurate prognostic assessments. It has been reported that patients with a CCDC39 or CCDC40 mutation have lower body mass indices and worse lung function that declines more rapidly compared with those with DNAH5 mutations [4] . Conversely, patients with RSPH1 and DNAH9 mutations appear to have a milder respiratory phenotype [5,6] .
The ciliary ultrastructural and beat patterns correlate with speci c genotypes. Speci cally, mutations in CCDC39 or CCDC40 cause axonemal disorganization and loss of inner dynein arms, conferring extremely stiff cilia with reduced beat amplitude [7] . Mutations in genes that encode cytoplasmic preassembly factors, such as DNAAF3, result in complete cilia immotility, with combined outer and inner dynein arm defects [8] . The consistency across ciliary beating anomalies and speci c genotypes makes HSVA a promising approach to assess the pathogenicity of VUS variants. However, extensive variability in the clinical and ciliary phenotype within a single mutation cohort has also been reported. In a study of CCDC103 p.His154Pro patients by Shoemark et al., the cilia were found to be static in half of the individuals, but completely or nearly normal in the others [10] . Variable diagnostic test results in nasal nitric oxide levels and ultrastructural images were also reported in patients with p.His154Pro mutation. Incomplete knowledge of phenotype could easily lead to underdiagnosis or false negatives. A larger number of patient subjects and detailed information was clearly needed to facilitate a better understanding of the subtle genotypephenotype correlations in PCD.
First reported in 2002 [11] , DNAH5 is the most frequent disease-causing gene in PCD patients. DNAH5 is an axonemal heavy chain dynein protein that functions as a force-generating protein to produce cilia bending. It has been reported that pathogenic variants in the DNAH5 gene are responsible for nearly 15% to 29% of all PCD cases [12] . Several previous studies focused on patients with DNAH5, revealing that mutations in DNAH5 lead to absent or shorter dynein arms in respiratory cilia that are also mostly immotile or only show ickering movements. Patients usually have the typical clinical phenotype of PCD. In this study, we systematically evaluated the phenotype and genetic characteristics of 30 PCD patients carrying DNAH5 mutations. Intriguingly, mild respiratory phenotype, normal nasal nitric oxide and diverse ciliary beat pattern were identi ed in our cohort group. For speci ed patients, IF was performed to investigate outer dynein arm (ODA) perturbations in nasal ciliated cells. It is hoped that the detailed phenotype information identi ed in our study will help to better delineate the genotype-phenotype relationship of DNAH5.

Patients and Clinical Materials
The project was approved by the ethics committee of the Ethics board of The Children's Hospital of Fudan University, and written informed consent was obtained from all family members participating in this study. Inclusion criteria were as follows: 1) at least 2 out of 4 key clinical features (laterality defects, unexplained neonatal respiratory distress, year-round daily cough, or nasal congestion) or siblings of con rmed patients; and 2) HSVA suggestive of PCD and/or low nNO level. nNO and HSVA were measured repeatedly in cases with equivocal results. Exclusion criteria were con rmed diagnosis other than PCD (e.g., immune de ciency or cystic brosis). Eventually, a total of 114 patients belonging to 104 unrelated nuclear families with symptoms suggestive of PCD and positive screening test (nNO and/or HSVA) results were enrolled between January 1, 2016 and June 1, 2021 at the Children's Hospital of Fudan University.
The clinical data, including lung function, high resolution CT (HRCT), and bacterial cultures were performed at enrollment. The diagnosis of PCD was based on the clinical ndings, nNO, HSVA, TEM, and genetic testing, in accordance with the guidelines of European Respiratory Society [1] .

Nasal Nitric Oxide Measurements
Measurements of nNO were performed with the use of an EcoMedics CLD88 chemiluminescence NO analyzer (Duernten, Switzerland); the aspiration-sampling rate of 330 nL/min was veri ed before and after each subject tested. The measurement of nNO in cooperative children was performed by breath hold maneuver for at least 10s to close their velum as described previously [13] . For children that were uncooperative (less than 5 years old), nasal sampling was performed for 60 seconds during tidal breathing. Results were reported in nL/min with the following equation: nNO (nL/min) = NO (ppb) × sampling rate (nL/min).

Electron Microscopy and High-Speed Video Microscopy Analysis
Nasal tissue was collected using a Rhino-Probe (Arlington Scienti c, Springville, UT). To avoid secondary ciliary dyskinesia, nasal samples were taken in patients who free from acute airway infection for at least 4 weeks. The tissue was suspended in L-15 medium (Invitrogen, CA) immediately for analysis using a Leica inverted microscope (Leica DMI300B, Solms, Germany) with a 63× oil objective under differential interference contrast optics. Cilia beats were recorded at 200 frames/sec at room temperature (25℃), using a 680 PROSILICA GE camera (Allied Vision, PA). At least ten videos were derived from each sample. The digital recordings were evaluated by two experienced but blinded investigators.
The ciliary ultrastructure was analyzed on nasal tissue, xed in 2.5% glutaraldehyde, as described previously [13] . For each specimen, at least 30 transverse ciliary sections of different cells were used to evaluate the internal axonemal structure. Ciliary abnormalities were de ned as the presence of defects in >50% of cilia.

Whole Exome Sequencing (WES) and Copy Number Variation (CNV) Analyses
Blood samples were obtained from proband and available family members. Genomic DNA was extracted using the Gene Blood DNA Rapid Extraction Kit (Qiagen, China). WES was performed by Gemple Biotech Co. Ltd (Shanghai, China). Whole exome libraries were constructed using KAPA platform and KAPA Hyper Prep kit (Roche KAPA, Switzerland). Exomes were captured using SeqCap EZ MedExome (Roche NimbleGen) and sequenced by a HiSeq X Ten instrument (Illumina, San Diego, CA). Raw data was evaluated using FASTQC (version 0.11.5), and the linker sequences were removed by Cutadapt (version 1.10). The software BWA (version 0.7.15) was used to align the reads to the human reference genome GRCh37/hg19(UCSC). For analyzing the phenotypeassociated mutation sites, the mutation data were annotated and categorized according to the American College of Medical Genetics and Genomics (ACMG) guidelines [14] . Sanger sequencing was performed to validate the candidate variants, and segregation analyses were performed in family members. The panelcn.MOPS software was introduced to detect CNVs in WES data, which is a new algorithm that builds upon copy number estimation by Mixture of Poissons (cn.MOPS) to detect CNVs with high sensitivity and speci city [15] .
The RefSeq accession numbers of the transcript and corresponding protein isoform we used for mutation nomenclature were NM_001369.2 and NP_001360.1.

Quantitative Real-Time Polymerase Chain Reaction (qPCR) Analysis
To con rm CNVs detected by WES, qPCR was performed using DNA from patients, their parents and controls. DNA was obtained as described previously.
qPCR was prepared using KAPA SYBR FAST qPCR Kit Master Mix(2X) Universal (Roche KAPA, Switzerland) on a CFX96™ Real-Time System instrument (BIO-RAD) and was performed on a LightCycler 480 System (Roche diagnosistic) in triplicate. The reaction conditions were set at 95 °C for 3 min, followed by 40 cycles of 95 °C for 5 sec and 55 °C for 30 sec. The primer pair sequences were designed by NCBI (https://www.ncbi.nlm.nih.gov/). The comparative ΔΔCt method was used to calculate relative expression with data normalized to the mean level of an internal standard.

Whole Genome Sequencing (WGS)
Genomic DNA was obtained and sheared to fragments of an average length of 350 bp to construct libraries using KAPA platform and KAPA Hyper Prep kit (Roche KAPA, Switzerland). The libraries were then subjected to sequencing on the Illumina Hiseq X platform and 150 bp paired-end reads were generated. The same bioinformatics pipeline as described above for the exome sequencing analysis was used to analyze DNAH5 mutations and precise breakpoints. The variants identi ed were con rmed by Sanger sequencing.

IF Analysis
Nasal tissue, obtained by nasal brushing biopsy, was suspended in cell culture medium and spread onto glass slides, then air dried. Samples were treated with 4% paraformaldehyde, 0.2% Triton-X 100 and 1% BSA before incubation with primary (overnight at 4 °C) and secondary antibodies (2 h at room temperature). Rabbit polyclonal anti-DNAH5 (HPA037470, Sigma, Sweden) and mouse monoclonal anti-alpha tubulin antibodies (ab24610, Abcam, UK) were used at a dilution of 1:1000. Highly cross-adsorbed secondary antibodies, goat anti-mouse Alexa Fluor 488 (1:1000) and goat anti-rabbit Alexa Fluor 647 (1:1000), were obtained from Abcam. Confocal images were taken using a Leica TCS SP8 confocal laser scanning microscope (Leica, Jena, Germany).

Clinical Phenotype of Individuals with DNAH5
In the present study, we identi ed 30 Chinese patients with DNAH5 from 27 unrelated families. The age of affected individuals ranged from 0.6 to 43 years old and 60.0% (18/30) were males. PCD classical symptoms in our cohort included: Year-round wet cough (96.7%, 29/30), neonatal respiratory distress (47.6%, 10/21), nasosinusitis (93.1%, 27/29), and chronic otitis media (95.2%, 20/21). Further examination revealed that 80.0% of patients (24/30) had situs inversus totalis and 7% of patients (1/30) had heterotaxy; median forced expiratory volume in 1 s (FEV1) was estimated to be 74.6%. In general, lung function, especially the forced expiratory ow at 25 and 75 % of the pulmonary volume (FEF 25-75%), gradually declined with age (Supplementary Figure 1). Surprisingly, as a 19-year-old boy, P40 had a mild clinical disease with normal lung function. Consistent with this, only small pieces of patchy exudation in his lower pulmonary lobes were identi ed by high resolution CT scans (Supplementary Figure 2). Patients P100 and P101 were from a Han Chinese consanguineous family ( Figure 1A). No situs inversus totalis and bronchiectasis were identi ed in their mother ( Figure 1B) and father ( Figure 1C). While situs inversus totalis, bronchiectasis and nasosinusitis were found in P100 ( Figure 1D-F), P101 ( Figure 1G-I) and their older brother (not available). Surprisingly, the two older brothers had much earlier onset and more frequent lung infections than P100. Progressively lung failure leaded to P101 death shortly after diagnosis. Another met with lung failure and death at the age of 38, according to their parents. HRCT indicated that P100 ( Figure 1E) had a milder destruction of lung than P101 ( Figure 1H) at enrollment. nNO testing of 19 individuals revealed a median nNO production rate of 22.8 nL/min (minimum, 5.6 nL/min; maximum, 99.0 nL/min). Interestingly, two siblings (P72 and P97) had nNO values that were above the disease-speci c cut off value (77 nL/min). The detailed clinical characterizations and nNO values of our cohort are summarized in Table 1 Ciliary Structural and Functional Analysis Ultrastructural analysis of 21 subjects with biallelic mutations in DNAH5 showed completely missing or shortened ODA. HSVA revealed that most DNAH5 mutant cilia were completely immotile or exhibited only occasional residual movement (Supplementary Video 1). However, four individuals (P24, P57, P69 and P97) with DNAH5 mutations exhibited diverse CBP. CBP of P24 was relatively stiff, with a decreased proximal bending (Supplementary Video 2). Cilia of P57 showed uncoordinated beating activity (Supplementary Video 3), while HSVA of P69 (Supplementary Video 4) and P97 (Supplementary Video 5) displayed a subtle reduction of the beating amplitude. As siblings, P72 and P97 carried the same mutations, but their ciliary beat patterns were distinctive. To further con rm the pathogenicity of the mutations, we studied P24, P57, P97 and P100 for DNAH5 localization in respiratory cells by IF, and found complete absence of DNAH5 in ciliary axonemes (Figure 2A-E). The details of ciliary structural and functional analysis are shown in Table 2.

Genetic Characteristics and cDNA Analysis
Collectively, 45 biallelic mutations were identi ed, of which 16 were missense variants, 5 were frameshift, 10 were nonsense variants, 2 were CNV, 2 were inframe deletions and 10 were splicing variants, while 37 variants (82.2%) were novel (Table 3). Interestingly, the previously reported mutation c.5563dupA was found in four families. To nd out whether the mutation was derived from a common ancestor (founder), we had sequenced 21 single-nucleotide polymorphisms (Supplementary Table 1) in the 200 kb DNAH5 locus to deduce the haplotypes. However, no identical haplotype was found in these families.
The WES results were routinely analyzed with an algorithm to detect CNVs. The positions of the CNV in the DNAH5 gene are shown in Figure 3A. A CNV causing heterozygous deletions corresponding to exons 71 to 72 of DNAH5 was suspected in P57 by analysis. Using qPCR, this deletion was recognized in P57 and her father ( Figure 3B). As WES only covers exonic regions, WGS was performed to specify the precise breakpoints. Eventually, chr5:13717909-13722659del was detected and con rmed ( Figure 3D). Two CNVs causing homozygous deletions spanning exon 69-71 and exon 77-79 were suspected in P100 and P101; qPCR veri ed that exon 69-71 and exon 77-79 were homozygous deletions in P100, P101 and heterozygous deletions in their parents ( Figure   3C). This indicated an autosomal recessive inheritance pattern conforming to the genetic disposition of PCD. To specify the precise breakpoints of the two CNVs, WGS was performed. Ultimately, chr5:13720087_13733030delinsGTTTTC and chr5:13649539_13707643del were detected and con rmed ( Figure 3E).

Discussion
In the present study, the utility of NGS screening with additional CNV analysis was evaluated in a large cohort of 114 patients with highly suspected PCD from China. Overall, DNAH5 was the most prevalent mutant gene, and 45 different allelic mutations were identi ed in affected individuals from 27 (24.7%) families.
It suggested that the genetic pro le in China was similar with those of the Europeans [16] . Multiple aspects of PCD characteristics, including clinical manifestations, ciliary phenotype, as well as genetic characterization were robustly obtained.
Several studies have systematically investigated associations between genotype and phenotype in PCD, which is a genetically heterogeneous disorder. It has been reported that lung function was signi cantly worse in individuals with CCDC39 and CCDC40 mutations and better in those with DNAH11 and DNAH5 mutations [17] . Consistent with the report of Shoemark et al, highly phenotypically diverse were also observed in our cohort [10] . For instance, P40 still retained nearly normal lung function in his adulthood with only a mild clinical phenotype. By contrast, P101 had severe clinical symptoms, and died of respiratory failure at middle age. Phenotypic diversity was in full display even among siblings. Despite frequent respiratory infections, P100 still had an apparently normal life at 38 years old with only mild to moderate impairment of pulmonary function. In contrast, his older brothers who carried identical mutations and were in the same living environment were not so fortunate; they all had severer bronchiectasis, and died from progressive lung failure at middle age. Similar heterogeneity was also observed in the pair of close-in-age sisters, P69 and P95. It is worth mentioning that data on mortality of PCD are extremely limited. To the best of our knowledge, the two respiratory deaths in our cohort are the rst reported in DNAH5-associated PCD.
It has been widely reported that cilia beat patterns characterized in PCD patients correlate highly with the implicated gene [9] . The ODA generate the main mechanical force to produce cilia bending. It is generally recognized that mutations in genes encoding ODA and ODA docking components (e.g. DNAH5) cause absent or de cient ODA in cilia, leading to immotile cilia. Intriguingly, as reported here, we observed substantial heterogeneity of ciliary beat pattern in our patients. During HSVA, the CBP differed markedly in the range of static, restricted to stiff movement. The ciliary beat functions of P24(c.13126-2A>G, c.7789A>G), P57(c.13724-1G>A, 4.7kB del), P69(c.5563dupA, c.4355+5G>A) and P97(c.8311C>T, c.6646_6657del) were largely preserved. Similar phenomena were also observed by Blanchon et al, who identi ed 50% of cilia with normal motion in a patient carrying a DNAH5 missense mutation (p.R2771C) [18] . It appears that the different mutations have diverse effects on the protein, and that the alleles with severe effects on the ciliary beating parameters generated drastic structural and functional alterations in the protein. However, it is di cult to explain the difference of beat patterns observed in siblings. Similarly, variable ciliary beat pattern of PCD patients who carried identical mutation is also reported by Shoemark et al [10] . Further research is needed to better understand the mechanisms and in uential factors of ciliary beat. It should be noted that HSVA provides important clues for the determination of pathogenicity of mutations. In recent years, genetic testing has rapidly advanced and become a viable diagnostic tool for PCD diagnosis [1,19] . In theory, a molecular genetic diagnosis of PCD can be established through the identi cation of biallelic pathogenic mutations in a known PCD-associated gene. In practice, however, frequent variants of unknown signi cance can make diagnosis equivocal, and in these situations, the concordance between genotype and phenotype becomes helpful. Overall, an insu cient knowledge of the ciliary phenotype for a speci c genotype may lead to misdiagnosis.
CNVs have been shown to be associated with complex diseases such as cancer, neuropsychological disorders, Crohn disease, autism spectrum disorders, or multiple congenital anomalies [20] . As technology developed, roles of CNVs in the pathogenesis of PCD were gradually decoded. Deletions in SPAG1, ARMC4, DYX1C1, LRRC50, ZYMD10 and DNAH5 were observed successively [21,22] , suggesting that CNV analysis should be used in the molecular evaluation of patients with PCD. Marshall reported that WES followed by targeted CNV analysis identi cation of 4 of 45 (8.8%) PCD patients harbored clinically signi cant CNVs [22] .
In a study by Fassad et al, CNVs accounted for 4% of variants overall in PCD patients [16] . In the present study, the CNVs were initially predicted by analysis with WES data plus panelcn.MOPS, a reliable algorithm to detect CNVs with high sensitivity and speci city, then con rmed by qPCR and Sanger sequencing. Through this process, we identi ed 2 of 27 (7.4%) families with clinically signi cant CNVs. To further evaluate the pathogenicity, we performed high-resolution IF analyses, and con rmed that DNAH5 were absent in the cilia. A considerable proportion of CNV in PCD suggests that routine CNV analysis in the molecular diagnosis of PCD is clinically bene cial.

Conclusions
In conclusion, by utilizing state-of-the-art diagnostic techniques, we have provided a detailed description of the high diversity in DNAH5-associated PCD, which included clinical manifestations, ciliary phenotypes, and genetic spectrum. We hope that the diversity of phenotypes and genotypes identi ed by this study will lead to a better diagnosis of PCD and recognize the heterogeneous nature of this orphan disease. The project was approved by the ethics committee of the Ethics board of The Children's Hospital of Fudan University, and written informed consent was obtained from all family members participating in this study.

Consent for publication
Parents or subjects gave informed written consent.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This work was supported by Shanghai Natural Science Foundation of Science and Technology Innovation Action Plan (Grant Number: 21ZR1409900 and 21ZR1410200).
Authors' contributions ZYG and WCC performed the research, analyzed and interpreted the data, and drafted the manuscript. MRL performed experiments and prepared Figures 2 and Table 3. LBW and LLQ conceived and designed the study, revised the manuscript. The authors read and approved the manuscript to be published.     Axoneme-speci c antibodies against acetylated α-tubulin (red) were used as the control. In respiratory epithelial cells from healthy probands (A), DNAH5 (green) localized predominantly along the entire length of the axonemes, as well as to the apical cytoplasm. In respiratory epithelial cells from individuals P24 (B), P57 (C), P97 (D) and P100 (E), DNAH5 was absent from the axoneme and markedly reduced in the apical cytoplasm.