A 12-year-old girl with situs inversus, recurrent respiratory infections and chronic rhinosinusitis, her parents, and non-PCD controls (6 individuals) were included in this study. Based on clinical symptoms, a decrease in ciliary beat frequency by high-speed video microscopy, ciliary structural defects detected by transmission electron microscopy, and exome sequencing, the patient was diagnosed with PCD. The six control subjects were excluded from the diagnosis of PCD and acute or chronic airway inflammation. The collection and use of peripheral blood, respiratory epithelial cells and bronchial biopsy specimens were approved by the Ethical Review Board of the Third Affiliated Hospital of Southern Medical University, and written consent was obtained. Clinical data were collected using standardized methodologies as reported previously (Shapiro et al., 2016; Zariwala et al., 2019; Guan et al., 2021).

The blood samples of the proband, her patrents and non-PCD controls (3 individuals) were used for DNA and RNA extraction.

Respiratory epithelial cells were obtained by segmental bronchus brushing of the proband and the controls (3 individuals) with a cytological soft sterile brush (AF -D1810WA, Alton, China). Each subject brushed once with a back and forth motion for a total of 8–10 strokes. Cells were suspended in PneumaCult-Ex Plus Medium (Stem cell, 05040) prewarmed to 37°C and used for high-speed video microscopy (HSVM) analysis. Then, bronchial biopsy specimens were collected from the segmental bronchus of the proband and the controls (3 individuals) using a disposable biopsy sampling forceps (AF -D1810BT, Alton, China). Three biopsies were taken from each subject. One was used to extract RNA for RNA analysis, one was fixed in 4% paraformaldehyde for immunofluorescence, and one was fixed in 2.5% glutaraldehyde for transmission electron microscopy.

Whole-exome sequencing was performed as described previously (Knowles et al., 2013; Onoufriadis et al., 2013). Genomic DNA was extracted from the peripheral blood and fragmented using Covaris S220 (Covaris, USA), with an average size of 150bp. The exonic DNA sequences were enriched using the GenCap V4.0 probe sequence capture array (Mygenostic, China), which captures the exons and the intron regions adjacent to the exons (50 bp in size). Sequencing was performed using the DNBSEQ-T7 (MGI Tech CO., Ltd, China). The final targeted region was 49.11 Mb and included 23,000 genes. The average sequencing depth was 185.59; 98.65% of target bases were covered at 10×, and 97.56% of target bases were covered at 20×. The raw data were assessed for quality using the base-calling method according to manufacturer’s instructions. The remaining sequence reads were aligned to the human reference genome (GRCh38/hg38 and GRCh37/hg19) using Burrows Wheeler Aligner (BWA; Version 0.7.16). Then the variants were further annotated using Variant Effect Predictor (http://asia.ensembl.org/info/docs/tools/vep/index.html) and ANNOVAR (http://www.openbioinformatics. org/annovar) and compared with 1,000 Genome Project, ESP6500SI, dbSNP, EXAC_ALL, HGMD and ClinVar. The pathogenicity of the candidate variants was evaluated based on the American College of Medical Genetics and Genomics guideline (ACMG). In addition, PCR amplification and Sanger sequencing were further performed to validate the candidate variants and segregation analyses were performed in family members. The primers used for PCR and Sanger sequencing are listed in Supplementary Tables S3, S4.

Nasal NO was measured with an EcoMedics CLD88 NO analyzer (Duernten, Switzerland). Because the individuals in this study were older than 5 years and were able to cooperate well, the nNO measurement was performed by exhalation against resistance maneuver as described previously (Leigh et al., 2013; Shapiro et al., 2020). The individuals had been in her baseline state of health for at least 2 weeks before the nNO measurement. The technician monitored the nNO curve on a computer screen and required an exhalation time of up to 30 s to achieve an acceptable plateau phase of at least 3 s that had less than 5% deviation. This might require several attempts until two successful trials are achieved in the first nostril. Then the procedure was repeated in the contralateral nostril, and two more plateau measurements were made in the same way. The plateau value in the same nostril should be less than 10%, and the plateau value between the right and left nostrils should be similar under ideal conditions. However, due to the difference in nasal airflow, the nNO values of the left and right nostril might differ by more than 10%. If the difference between the left and right nostrils was more than 10%, repeated the measurement at least twice in the same phase in each nostril. Even if there was such a difference, the maximum repeatable value of each nostril should be reported. The nNO production (nL/min) was calculated using the following equation: nNO (nL/min) = NO (ppb) × sampling rate (mL/min).

To evaluate the expected splicing results of the ODAD1 mutations, the SpliceAI model (v1.3.1) was applied to two sequences (reference allele and variant allele), calculating the probability of each nucleotide being used as an acceptor or donor site in a biological context. Prediction results are then visualized using SpliceAI-visual (https://mobidetails.iurc.montp.inserm.fr/MD/) as previously described (Agathe et al., 2023). The reference genome version was GRCh38/hg38.

To determine the effects of ODAD1 mutations on transcripts, reverse transcription polymerase chain reaction (RT-PCR) was performed using RNA extracted from the bronchial mucosa biopsy specimens of patient Ⅲ:1 and control individuals according to the manufacturer’s protocol. The primers used for PCR had the corresponding restriction enzyme sites (Supplementary Table S3). The amplification fragment was cloned into the pB-CAG-Amp vector to construct a recombinant plasmid containing ODAD1 cDNA fragment and was identified via Sanger sequencing (Figures 2A, C). The primers used for Sanger sequencing are listed in Supplementary Table S4.

To investigate the effects of ODAD1 mutations on the tertiary structure of ODAD1 protein, we predicted the tertiary structures of six ODAD1 variants: two (mutants 1, 2) identified in this study and four (mutants 4–7) reported in previous studies (Knowles et al., 2013; Onoufriadis et al., 2013; Ostrowski et al., 2022) (Supplementary Table S2). Briefly, the primary structures of proteins were predicted using SnapGene (version 3.2.1) based on the base sequences of ODAD1 mutant transcripts. Subsequently, three-dimensional models were constructed using SWISS-MODEL (https://swissmodel.expasy. org/interactive#structure, A0A6I8PTZ2) and optimized using the Pymol software.

Bronchial mucosa biopsy samples collected from patient Ⅲ:1 and control individuals were fixed, dehydrated, embedded, and cut into 10-μm-thick cryosections. Thereafter, the sections were incubated with mouse monoclonal anti-acetylated tubulin antibody (Sigma, T7451) and rabbit polyclonal anti-ODAD1 antibody (Novus, NBP1-93863) overnight at 4°C. The following day, the sections were incubated with Anti-mouse Alexa Fluor 488 antibody (Abcam, ab150105) and Dnk pAb to Rb IgG Alexa Fluor 594 antibody (Abcam, ab150064) and stained with DAPI (Yeasen, 36308ES11). Images were acquired using an Olympus FV3000 confocal microscope (Olympus, Japan).

Bronchial biopsy specimens were immersed in 2.5% glutaraldehyde immediately after collection. The following day, the specimens were fixed in 1% osmium tetroxide, dehydrated in a graded ethanol series, and embedded in epoxy resin. Subsequently, the specimens were cut into 60-nm-thick sections and stained with 2% methanolic uranyl acetate. Finally, the ultrastructure of cilia was analyzed using a transmission electron microscope (JEM -1,400; Jeol, Tokyo, Japan).

Respiratory epithelial cells were collected via bronchial brush biopsy and suspended in PneumaCult-Ex Plus Medium (Stem cell, 05040) preheated to 37°C. Ciliary activity was recorded at 120 frames/second using an inverted microscope (MI52-M; Mingmei, Guangzhou, China) as described previously (Knowles et al., 2013; Onoufriadis et al., 2013). To evaluate ciliary beat frequency (CBF), digital recordings were evaluated by three investigators in a blinded fashion. We used slow-motion playback of the video sequence to generate tracings of the ciliary beat. At least 3 visual fields per sample and 10 ciliary edges per visual field were traced manually.

Plasmids containing the wild-type sequence of ODAD1 and the mutant C-terminal fragments of ODAD1 (Mutant1, Mutant4 and Mutant6) was obtained from Azensta (Guangzhou, China). The cDNA of every mutant ODAD1 (mutants 1-7) with V5 tag was obtained with position-specific primers carrying corresponding restriction enzyme sites by PCR (Vazyme, P520- AA) and T4 ligase (Takara, 2011A).

To obtain the cDNA of outer dynein arm docking complex subunit 3 (ODAD3), RNA was extracted from bronchial mucosa biopsies using Trizol (Invitrogen, 10296010) and first strand complementary DNA was synthesized with Sensiscript RT kit (QIAGEN, 205213). PCR was performed with position-specific primers carrying corresponding restriction enzyme sites and 2×Phanta flash Master Mix (Vazyme, P520- AA) according to the manufacturer’s protocols.

To generate the expression vectors of pb-CAG-ODAD1 WT-V5-IRES-BSD-T2A-BFP, pb-CAG-ODAD1 Mutant-V5-IRES-BSD-T2A-BFP, V5-tagged wild-type ODAD1 and ODAD1 mutants 1-7 were cloned into a piggyBac transposable backbone carrying an IRES- BSD-T2A-BFP element. And meanwhile, ODAD3 was cloned into a piggyBac transposable backbone carrying a 3×HA- IRES- Puro-T2A-GFP to generate the pb-CAG-ODAD3-HA-IRES-Puro-T2A-GFP vector (Figure 5A).

All wild-type cDNA clones were confirmed by Sanger sequencing and matched RefSeq gene accession number NM_001364171 (ODAD1) or NM_145045 (ODAD3). The ODAD1 mutant clones were confirmed by sequence analysis and matched with the results of mutant transcripts we discovered (mutants 1, 2) and previously published (mutants 4-7). The primers used for cDNA cloning and Sanger sequencing are listed in Supplementary Tables S3, S4.

Given the advantages of 293FT cells, such as rapid proliferation, high transfection efficiency, and the structure of the expressed protein closest to its conformation in the human body, we chose 293FT cells to study the interaction between ODAD1 and ODAD3. The HEK293FT cell line was a generous gift from Xuefei Gao (Basic Medical College of Southern Medical University). The cells were cultured in DMEM supplemented with 10% FBS (Gibco, 10099141C) and 1% penicillin and streptomycin (Gibco, 15140122) at 37°C.

To examine the interaction between ODAD1 variants and ODAD3, HEK293FT cells were co-transfected with plasmids encoding V5-tagged wild-type or mutant ODAD1 or empty plasmids and plasmids encoding HA-tagged ODAD3 through electro-transfection (Lonza, Switzerland) according to the manufacturer’s protocol (Figure 5B).

To assess whether mutant ODAD1 affects the interaction between wild-type ODAD1 and ODAD3, vector encoding C-terminal V5-tagged mutant ODAD1 or empty plasmid was co-transfected with vector encoding C-terminal V5-tagged wild-type ODAD1 and HA-tagged ODAD3 in 293FT cells. The empty plasmid was used as a control (Figure 6A). The total amount of each plasmid used was 2 μg. The cells were screened in medium containing BSD (10 μg/mL) and Puro (1 μg/mL). The construction of expression vectors is described in Supplemental files.

Proteins extracted from HEK293FT cells were incubated with 4 µg of HA-tagged polyclonal antibody (Invitrogen, 71-5,500) and 4 µg of rabbit IgG (Cell Signaling Technology, 2729s) overnight at 4°C. The following day, each sample was incubated with 20 μL of DynabeadsTM Protein G (Invitrogen, 91216624) on a shaker for 2 at 4°C to capture immunoprecipitates. Bead complexes were washed thrice with IP lysis buffer (Beyotime, P0013), resuspended in a sample buffer (0.5% PMSF, 1% PI, 2% TCEP, and 25% SDS loading in IP lysis buffer), and heated for 10 min at 70°C.

Immunoblotting was performed as described previously (Zheng et al., 2022). The loading amount of Input was 20μg, and the volume of IgG and IP used was 12 μL. Membranes were incubated with V5-tag rabbit PolyAb (Proteintech, 14440-1-AP) and HA-tag rabbit mAb (Cell Signaling, C29F4) at 4°C overnight and with IPKine™ HRP Mouse Anti-Rabbit IgG LCS (Abbkine, A25022) at RT for 1 h. Images were digitally acquired using Amersham Imager 600 (Cytiva,USA). The expression of different proteins was quantified using the ImageJ software.

All statistical analyses were performed using the GraphPad Prism software (version 9.00). Data were expressed as the mean ± standard deviation. A t-test with two-tailed distribution was used to compare the data of two groups, whereas one-way analysis of variance (ANOVA) was used to compare the data of three or more groups. A p-value of<0.05 was considered significant.

The proband (patient Ⅲ:1) included in this study is the only patient in the Chinese family (Figure 1A; Supplementary Table S1). She developed a productive cough, chronic sinusitis, and otitis media as a young child. Besides, her nNO value was low (5.4 nL/min), and pulmonary function test revealed severe obstructive ventilator dysfunction (Supplementary Table S1). Computed tomography (CT) and magnetic resonance imaging (MRI) confirmed bronchiectasis, situs inversus totalis, and pansinusitis (Supplementary Figure S1).

Previously, all researchers except Ostrowski et al. (2022) have used NM_144577 in studies on ODAD1 (Knowles et al., 2013; Onoufriadis et al., 2013; Li et al., 2019; Chen et al., 2020; Guo et al., 2020; Guan et al., 2021; Hannah et al., 2022; Kos et al., 2022). In this study, NM_001364171 was detected in bronchial mucosa biopsy samples and peripheral blood cells collected from patient Ⅲ: 1 and control individuals (Supplementary Figure S2). This finding is consistent with that of a study by Ostrowski et al. (2022). (Ostrowski et al., 2022). Therefore, the transcript NM_001364171 was selected for further analysis. Structural similarities and differences between the two ODAD1 transcripts and the mutation sites of ODAD1 are shown in Figure 1B.

High-throughput exon-sequencing was used to identify candidate genes related to the pathogenesis of PCD. The results revealed two mutations in ODAD1, including c.71-2A>C and c.598-2A>C (Figure 1; Table 1). The results of Sanger sequencing demonstrated that c.71-2A>C and c.598-2A>C mutations were inherited from the mother and father, respectively (Figure 1C; Table 1; primer sequences are shown in Supplementary Tables S3, S4).

To analyze the effects of ODAD1 mutations on the resulting transcripts, silico predictions were performed. The results showed that c.71-2A>C weakened the canonical acceptor site in Intron 3 (acceptor loss of 0.68), and c.598-2A>C weakened the canonical acceptor site in Intron 7 (acceptor loss of 0.96) (Table 1; Supplementary Figure S3).

To determine the effects of ODAD1 mutations on the resulting transcripts, RT-PCR and Sanger sequencing were performed to detect ODAD1 cDNA in bronchial mucosal biopsy samples collected from patient Ⅲ:1 and three control individuals. The primer set (Exon7F + 9R) showed that the patient had two different amplification products (Figure 2B). Sanger sequencing of the smaller amplification product (indicated by the red arrow) showed that all 68 bp of exon 8 of this variant was deleted, causing a premature termination signal [Figure 2D (middle); Table 1]. Sanger sequencing of the larger amplification product (indicated by the black arrow) showed overlapping peaks, and the main peaks could not be determined. Therefore, the amplification fragment was cloned into the pB-CAG-Amp vector to construct a recombinant plasmid containing an ODAD1 cDNA fragment, which was identified by Sanger sequencing (Figure 2C). The results of the sequencing revealed two transcripts. One had a deletion of all 68 bp of exon8 and an insertion of 59 bp of intron8, leading to an in-frame deletion of 22 amino acids that were replaced by 19 mutant amino acids [Figure 2D (bottom); Table 1], whereas the other had the same sequence as the control individual. This may be due to the patient being a compound heterozygote or the mutation being leaky. The two aberrant transcripts (Mutant 1 and Mutant 2) were consistent with the high delta score (DS) for the variant sequence c.598-2A>C (acceptor loss: 0.96).

The c.71-2A>C was predicted by SpliceAI to result in an acceptor loss (AL: 0.68) in Intron 3. It has also been reported that the c.71-2A>C mutation in ODAD1 could lead to PCD (Guo et al., 2020). However, we and previous investigators were unable to verify how c.71-2A>C alters splicing by RT-PCR or RNA-sequencing (RNA-seq). In the future, if we can obtain samples from humans or animals carrying the c.71-2A>C mutation in ODAD1, we will perform in-depth studies to verify this result.

The template-based homology modeling approach was used to investigate the effects of ODAD1 splice-site mutations on the tertiary structure of the ODAD1 protein. We predicted the tertiary structures of six ODAD1 variants: two (mutants 1, 2) discovered in this study and four (mutants 4-7) reported in previous studies (Knowles et al., 2013; Onoufriadis et al., 2013; Ostrowski et al., 2022) (Figure 3; Supplementary Table S2). In Mutant2, 22 amino acids encoded by exon8 were replaced by 19 mutant amino acids encoded by intron8. However, all normal coiled-coil and disordered domains were retained (Figures 3B, G). Because each of the other six mutant transcripts had a premature translation termination signal, they all produced corresponding truncated proteins. The structure of the first five coiled-coil domains was completely conserved in Mutant6 and Mutant7 (Figures 3E, F). Mutant4 and Mutant5 retained the first three and first two coiled-coil domains, respectively (Figures 3C, D). Mutant1 retained the complete structure of the first two coiled-coil domains and had a frame-shift mutation in the third coiled-coil domain (Figure 3A). In the event that the complete tertiary structure of human ODAD1 wild-type and mutant proteins cannot be fully analyzed at present, modeling based on RNA transcripts can help us understand protein structure and function.

To further investigate the effects of ODAD1 mutations (c.71-2A>C or c.598-2A>C) on the expression of ODAD1, immunofluorescence microscopy was performed. ODAD1 mutations in patient Ⅲ:1 resulted in the expression of two mutant proteins (Mutant1 and Mutant2), and one protein retained all normal coiled-coil domains (Mutant2) (Figures 3A, B; Figures 4B, C). Therefore, an anti-ODAD1 antibody (NBP1-93863) was used to detect proteins that retained the first five coiled-coil domains (Figure 4A). ODAD1 localized throughout the entire length of ciliary axonemes in control individuals (Figure 4B); however, its expression was considerably downregulated in patient Ⅲ:1 (Figure 4C).

To investigate the effects of ODAD1 mutations (c.71-2A>C; c.598-2A>C) on the structure and function of the cilia, TEM and HSVA were performed. TEM of bronchial biopsy specimens from patient Ⅲ:1 demonstrated that most axonemes had no clear ODA, while several axonemes had two or more ODAs (Figure 4D; Supplementary Table S1). In addition, HSVA showed that CBF (0.78 ± 0.56 Hz) was significantly lower in patient Ⅲ: 1 than in control individuals (11.99 ± 0.70 Hz) (Figure 4E; Supplementary Videos S1–S3). However, contrary to what we expected and what had been found in previous studies in ODAD1 (Onoufriadis et al., 2013; Guo et al., 2020), the cilia of patient III: 1 were not completely immobile. Although the CBF of the patient III: 1 was significantly decreased, the cilia appeared to beat in a coordinated fashion and may provide some force for mucociliary clearance (Supplementary Video S3). These results indicated that protein variants resulting from the splice-site mutations in ODAD1 may retain partial function.

These results demonstrated that the compound mutation in ODAD1 (c.71-2A>C; c.598-2A>C) led to the absence of wild-type ODAD1 and the defects of the outer dynein arm in ciliary axonemes and caused a decrease in ciliary beat frequency.

To investigate the functional role of ODAD1 variants in the assembly of ODAs, we examined the interaction between ODAD1 variants and ODAD3, a subunit of ODA-DC that is responsible for axonemal microtubule attachment of the ODAs. The results revealed that wide-type ODAD1 could interact with ODAD3 (Figure 5C). Therefore, it was used as a positive control, and an empty plasmid as a negative control to investigate whether the proteins resulting from splice-site mutations in ODAD1 interact with ODAD3. All mutants that retained the first two coiled-coil domains were found to interact with ODAD3 (Figures 5C–J). These results indicate that the truncated proteins resulting from splice-site mutations in ODAD1, which retained the first two coiled-coil domains, could retain partial function.

Ostrowski et al. (2022) had found that truncated proteins resulting from the ODAD1 mutation (c.1502 + 5G>A) retain partial function but are insufficient to maintain normal ciliary function. Furthermore, (Chhin et al., 2009) restored cilia beating in human airway epithelial cells with DNAI1-mutaiton (c.48 + 2_48+3insT; c.1543G>A) using a lentiviral vector. However, the cilia beat in an uncoordinated manner. And the number of ODA per axoneme was significantly lower than normal, which may be due to some cells not being infected with the virus or to the endogenous mutant protein having a competitive inhibitory effect on the exogenous wild-type protein during dynein arm assembly. Based on the above results, we ask the following question: in gene therapy of PCD patients with ODAD1 mutations, do dysfunctional endogenous mutant proteins inhibit the function of exogenous wild-type proteins?

Therefore, a co-immunoprecipitation assay was performed to determine whether proteins resulting from splice-site mutations in ODAD1 affect the interaction between wild-type ODAD1 and ODAD3 in 293FT cells (Figure 6A). The results showed that the protein-binding ability (PBA) of Mutant1, which retained the first two coiled-coil domains, for ODAD3 was weaker than that of wild-type ODAD1 (Figures 6B, C). However, the PBA of other truncated ODAD1 proteins (Mutant4, Mutant6, and Mutant7), which retained the first three or more coiled-coil domains (Figures 6D–F), for ODAD3 was not different with that of wild-type ODAD1. Co-transfection of ODAD1 mutants resulting in proteins retaining the first three or more coiled-coil domains inhibited the interaction between wild-type ODAD1 and ODAD3 under the condition that no difference was observed in the expression of wild-type ODAD1 between groups (Figure 6G; Supplementary Figure S4). These results suggested that ODAD1 mutants retaining the first three or more coiled-coil domains inhibited the interaction between wild-type ODAD1 and ODAD3.