Ciliary Ultrastructure Assessed by Transmission Electron Microscopy in Adults with Bronchiectasis and Suspected Primary Ciliary Dyskinesia but Inconclusive Genotype

Whole-exome sequencing has expedited the diagnostic work-up of primary ciliary dyskinesia (PCD), when used in addition to clinical phenotype and nasal nitric oxide. However, it reveals variants of uncertain significance (VUS) in established PCD genes or (likely) pathogenic variants in genes of uncertain significance in approximately 30% of tested individuals. We aimed to assess genotype–phenotype correlations in adults with bronchiectasis, clinical suspicion of PCD, and inconclusive whole-exome sequencing results using transmission electron microscopy (TEM) and ciliary image averaging by the PCD Detect software. We recruited 16 patients with VUS in CCDC39, CCDC40, CCDC103, DNAH5, DNAH5/CCDC40, DNAH8/HYDIN, DNAH11, and DNAI1 as well as variants in the PCD candidate genes DNAH1, DNAH7, NEK10, and NME5. We found normal ciliary ultrastructure in eight patients with VUS in CCDC39, DNAH1, DNAH7, DNAH8/HYDIN, DNAH11, and DNAI1. In six patients with VUS in CCDC40, CCDC103, DNAH5, and DNAI1, we identified a corresponding ultrastructural hallmark defect. In one patient with homozygous variant in NME5, we detected a central complex defect supporting clinical relevance. Using TEM as a targeted approach, we established important genotype–phenotype correlations and definite PCD in a considerable proportion of patients. Overall, the PCD Detect software proved feasible in support of TEM.


Introduction
Primary ciliary dyskinesia (PCD) is a rare and heterogeneous genetic multisystem disorder caused by the dysfunction of motile cilia [1].Its pulmonary manifestation is characterized by persistent retention of hyperconcentrated mucus in the airways, with subsequent chronic inflammation, recurrent upper and lower respiratory tract infections, and progressive lung damage resulting in bronchiectasis [2].Approximately half of the patients show laterality defects like situs inversus totalis, referred to as Kartagener syndrome.In addition, PCD is associated with extrapulmonary manifestations like male and female subfertility and, rarely, hydrocephalus [1].The global prevalence of PCD is approximately 1:7500 population [3], although PCD is probably underdiagnosed, in particular among people with bronchiectasis [4].
Unfortunately, there is no stand-alone diagnostic test for PCD and guidelines from the European Respiratory Society (ERS) and the American Thoracic Society (ATS) recommend a combination of tests including measurement of nasal nitric oxide (nNO), high-speed video microscopy (HSVM), immunofluorescence microscopy, transmission electron microscopy (TEM), and genetic analysis [5,6].However, only TEM and genetics are able to confirm definite PCD by detecting ultrastructural hallmark defects or unambiguous pathogenic or likely pathogenic variants in motile ciliopathy genes with established significance [7,8].To date, >50 genes have been described with well-known association to characteristic clinical symptoms and ultrastructural defects in TEM, also referred to as genotype-phenotype correlation [1,9].Genetic analysis by next generation sequencing (NGS) is an expeditious diagnostic that is increasingly used in many research settings [4], but also in the routine work-up of orphan diseases in Germany in the past few years [10].However, it reveals inconclusive results in approximately 30% of patients with a clinical phenotype compatible with PCD, due to either variants of uncertain significance (VUS; class 3 according to American College of Medical Genetics [ACMG]) in established PCD genes or variants in candidate genes of uncertain significance with assumed but unconfirmed association to motile ciliopathies [1,8].
In the present study, we aimed to assess genotype-phenotype correlations in adult patients with bronchiectasis and clinically suspected PCD, but inconclusive genetics in whole-exome sequencing (WES).In addition to conventional TEM, we assessed the feasibility of the PCD Detect software for evaluating ciliary ultrastructure with regard to the confirmation of outer and inner dynein arm defects by ciliary image averaging.

Patient Population
Between December 2020 and January 2022, we consecutively recruited adult patients with computed tomography-confirmed bronchiectasis from our dedicated clinic at Hannover Medical School, Hannover, Germany.At our center, we routinely use NGS, in terms of WES, for the confirmation of clinically suspected PCD since 2018, while HSVM, immunofluorescence, and TEM have not been readily available in the past due to technical complexity and limited staff resources.Thus, all included patients had clinically suspected PCD with unconfirmed diagnosis and were offered genetic testing by WES within the routine work-up for the etiology of bronchiectasis.We obtained informed and written consent before WES according to national legislation.After WES, all patients included in this study had inconclusive genetic results and did not meet the criteria of definite PCD according to ERS or ATS guidelines [5,6].They were included regardless of the expected TEM outcome, i.e., even if the gene, in which the respective variants were detected, was presumably associated with normal ultrastructure.Patients had either compound heterozygous or homozygous VUS in established PCD genes, e.g., DNAH5, or likely pathogenic compound heterozygous or homozygous variants in PCD candidate genes, e.g., NME5 [8].Candidate genes were genes of uncertain significance that at the time of patient recruitment were suspected to play a role in the pathogenesis of motile ciliopathies based on supporting data like genomic location or function, but which were still unconfirmed in human patients [8] or merely had few reported cases [11][12][13].We included one subject with confirmed PCD due to known pathogenic variants in DNAH5 and one healthy subject as positive and negative controls, respectively.All patients and control subjects provided informed and written consent for study participation before we conducted nasal brush biopsy for TEM.Finally, we grouped the etiology of bronchiectasis into the following categories: definite PCD; probable PCD; possible PCD; or other diagnoses, e.g., asthma.For the purpose of our study, definite PCD required the detection of an ultrastructural hallmark defect by TEM [7], while we diagnosed probable and possible PCD in patients with a suggestive clinical phenotype and at least one supporting abnormal test (or positive familial segregation analysis) or no clearly abnormal test, respectively [14].Cystic fibrosis (CF) and alpha-1 antitrypsin deficiency were ruled out in all patients as part of their clinical routine.We extracted clinical data, including medical history, signs and symptoms, and findings from routine diagnostics like nNO from our electronic clinical database (FileMaker Pro, Claris International, Cupertino, CA, USA).We evaluated all patients for laterality defects (situs inversus totalis, situs inversus thoracalis, situs inversus abdominalis, left/right isomerism) and excluded patients with current sinonasal or pulmonary exacerbation, epistaxis, hemorrhagic diathesis, and therapeutic anticoagulation.

Transmission Electron Microscopy and PCD Detect Software
Ciliated epithelial cells were obtained by nasal brush biopsy of the inferior nasal meatus with a cytological brush (Gynobrush, Herenz, Hamburg, Germany) at stable state, i.e., in the absence of acute airway infection or sinonasal and pulmonary exacerbation, respectively.Samples were fixed overnight in 150 mM Hepes solution buffer (pH 7.35) containing 1.5% glutaraldehyde and 1.5% formaldehyde.We processed samples to 60 nm ultra-thin sections for TEM as described in Mariani et al. [16].Ultra-thin sections were observed with a Morgagni 268 Transmission Electron Microscope (FEI, Eindhoven, Netherlands) and images of transverse orientated cilia (cross-sections) were taken using a side-mounted Veleta CCD camera (Olympus SIS, Muenster, Germany).We analyzed >50 ciliary crosssections from different cells per sample, only assessing cilia with clear structural features and intact membrane.We analyzed and classified the ciliary ultrastructure according to the international consensus guideline BEAT PCD TEM Criteria, with disease-defining class 1 (hallmark) defects, class 2 defects that indicate the diagnosis of PCD in combination with other supporting evidence, or normal ultrastructure [7].In addition, we used the software PCD Detect, 2020 (downloaded from: https://doi.org/10.5522/04/12327839,last accessed on 28 September 2023), to analyze outer and inner dynein arms on TEM images [17].PCD Detect is a cilia-averaging tool that generates an averaged image from multiple, manually chosen and layered cutouts, thus reducing background.Nine cutouts, each including a microtubular doublet with outer and inner dynein arms, were manually taken from each ciliary cross-section.At least 5 ciliary cross-sections from different cells, resulting in minimum 45 cutouts, were analyzed in each sample.We performed a reference free averaging using the same search settings as the developers of the software (angle search range: 360 • ; angle increment size: 1 • ; search X and Y: 10 pixels; and translational increment size: 1 pixel) [17].In the present study, we used the color contour map of best-matching features, with yellow and blue indicating low and high electron density, respectively.In a small subset of subjects we assessed the test-retest reliability of TEM analysis on separate samples from nasal brush biopsies of the same individual taken at different occasions.

Statistical Analysis and Ethical Approval
We assessed the patient cohort by descriptive statistics, using medians, interquartile ranges (IQR), and absolute ranges (SPSS, version 27.0, IBM Corp., Armonk, NY, USA).This study was conducted in accordance to the Declaration of Helsinki.The Institutional Review Board of Hannover Medical School approved our research (No. 9831_BO_S_2021).All patients and control subjects provided informed and written consent prior to study inclusion.

Patient Characteristics
We included 16 patients with inconclusive results in WES due to VUS in established PCD genes (n = 11) or variants in the candidate genes DNAH1 (n = 2), DNAH7, NEK10 and NME5 (n = 1, each; Table 1).Among those, seven patients (44%) were female and nine (56%) were male, with a median age of 41 years (IQR 23-53, range 18-63 years).The majority of patients descended from Germany (n = 7), followed by Turkey (n = 5), Russia, Tunisia, Syria, and Kenya (n = 1, each).At inclusion in our study, all patients had bronchiectasis without a definite diagnosis of PCD.Clinically, we graded patients as probable PCD (n = 10), possible PCD (n = 3), idiopathic bronchiectasis (n = 2), and asthma (n = 1).Nasal NO measurements were available from medical records for all but one subject with Mycobacterium abscessus infection.Eight patients showed levels below the diagnostic threshold of 77 nL/min.Overall, median nNO production rate was 46 nL/min (IQR 17-238, range 4-290 nL/min).Median FEV 1 was 54% of predicted (IQR 36-70, range 30-109; Table 1), indicating moderate to severe airflow limitation for the majority of patients.=" means that the entire protein coding region was analyzed and no variant was found that changes (or is predicted to change) the protein sequence; "*" means that the amino acid is changed to a stop codon; and "?" means that the consequence, at the protein level, of a variant affecting the translation initiation codon cannot be predicted, i.e. is unknown.

Transmission Electron Microscopy Analysis
Overall, we detected ultrastructural hallmark (class 1) defects in six patients with probable PCD and inconclusive WES results due to VUS in the established PCD genes CCDC40 (n = 3; Figure 1), CCDC103, DNAH5, and DNAI1 (n = 1, each; Figure 2).In addition, we observed a characteristic central complex (CC; class 2) defect in one subject with a likely pathogenic homozygous variant in the candidate gene NME5 (n = 1; Figure 3).In detail, TEM revealed disease-defining microtubular disorganization with inner dynein arm (MTD + IDA) defects in three patients with VUS in CCDC40 (Figure 1).Notably, the homozygous VUS in CCDC40 in Patient 13 were only discovered following the detection of the MTD + IDA defect with subsequent reevaluation of WES data, as this ultrastructural defect did not match the initially reported variants in DNAH5.Moreover, we found diseasedefining outer dynein arm (ODA) defects in three patients with VUS in CCDC103, DNAH5, and DNAI1 (n = 1, each; Figure 2).Overall, the findings from averaged images of the PCD Detect software showed perfect agreement with those of conventional TEM imaging (Figures 1 and 2).In the NME5 patient, TEM images showed ciliary cross-sections with normal 9 + 2 structure (in 42 of 119 evaluated cilia, 35%), but also deviant arrangements with the absence of the central pair (9 + 0 structure, 49/119, 41%; and 8 + 0 structure, 13/119, 11%), an additional central pair (9 + 4 structure, 3/119, 3%), and translocated microtubules (8 + 1 structure, 12/119, 10%; Figure 3).Nasal NO measurements showed nNO levels well below the diagnostic threshold of 77 nL/min in all patients with routine data available (Table 1).All patient had a history of early onset of chronic wet cough and chronic sinusitis with nasal polyps.Four patients, in whom we detected characteristic ultrastructural defects, recalled neonatal Respiratory Distress (NRD).One patient with VUS in CCDC40 and DNAI1 had Kartagener syndrome, respectively, and one patient with VUS in CCDC40 had a history of congenital heart but no laterality defect.One female with VUS in CCDC40 reported a history of infertility (Table 1).There was no history of hydrocephalus reported in these seven patients.

Transmission Electron Microscopy Analysis
Overall, we detected ultrastructural hallmark (class 1) defects in six patients with probable PCD and inconclusive WES results due to VUS in the established PCD genes CCDC40 (n = 3; Figure 1), CCDC103, DNAH5, and DNAI1 (n = 1, each; Figure 2).In addition, we observed a characteristic central complex (CC; class 2) defect in one subject with a likely pathogenic homozygous variant in the candidate gene NME5 (n = 1; Figure 3).In detail, TEM revealed disease-defining microtubular disorganization with inner dynein arm (MTD + IDA) defects in three patients with VUS in CCDC40 (Figure 1).Notably, the homozygous VUS in CCDC40 in Patient 13 were only discovered following the detection of the MTD + IDA defect with subsequent reevaluation of WES data, as this ultrastructural defect did not match the initially reported variants in DNAH5.Moreover, we found disease-defining outer dynein arm (ODA) defects in three patients with VUS in CCDC103, DNAH5, and DNAI1 (n = 1, each; Figure 2).Overall, the findings from averaged images of the PCD Detect software showed perfect agreement with those of conventional TEM imaging (Figures 1 and 2).In the NME5 patient, TEM images showed ciliary cross-sections with normal 9 + 2 structure (in 42 of 119 evaluated cilia, 35%), but also deviant arrangements with the absence of the central pair (9 + 0 structure, 49/119, 41%; and 8 + 0 structure, 13/119, 11%), an additional central pair (9 + 4 structure, 3/119, 3%), and translocated microtubules (8 + 1 structure, 12/119, 10%; Figure 3).Nasal NO measurements showed nNO levels well below the diagnostic threshold of 77 nL/min in all patients with routine data available (Table 1).All patient had a history of early onset of chronic wet cough and chronic sinusitis with nasal polyps.Four patients, in whom we detected characteristic ultrastructural defects, recalled neonatal Respiratory Distress (NRD).One patient with VUS in CCDC40 and DNAI1 had Kartagener syndrome, respectively, and one patient with VUS in CCDC40 had a history of congenital heart but no laterality defect.One female with VUS in CCDC40 reported a history of infertility (Table 1).There was no history of hydrocephalus reported in these seven patients.TEM showed normal ciliary ultrastructure in five patients with VUS in the established PCD genes DNAH11 (n = 2), CCDC39, DNAH8/HYDIN, and DNAI1 (n = 1, each).In addition, we observed normal ciliary ultrastructure in patients with variants in the PCD candidate genes DNAH1 (n = 2) and DNAH7 (n = 1; Figure 4).Again, the findings from averaged images of the PCD Detect software were well in line with those from conventional TEM imaging (Figure 4).Among patients with normal ultrastructure, two patients had probable PCD, three had possible PCD, and three had other diagnoses, including idiopathic bronchiectasis (n = 2) and asthma (n = 1), based on clinical signs and symptoms as well as diagnostic testing results (Table 1).Both patients with probable PCD had chronic upper airway disease, chronic wet cough, and low nNO levels, while one patient had NRD and chronic Pseudomonas aeruginosa infection, respectively.Among the three patients with possible PCD, all had chronic upper airway disease and chronic wet cough, but normal nNO levels.One of those patients had a positive family history and a history of parental consanguinity, each, while another patient recalled NRD and two had chronic P. aeruginosa Cells 2023, 12, 2651 9 of 14 infection.All of the three patients with other diagnoses had chronic wet cough and normal nNO levels, while two of them had chronic P. aeruginosa infection (Table 1).4).Again, the findings from averaged images of the PCD Detect software were well in line with those from conventional TEM imaging (Figure 4).Among patients with normal ultrastructure, two patients had probable PCD, three had possible PCD, and three had other diagnoses, including idiopathic bronchiectasis (n = 2) and asthma (n = 1), based on clinical signs and symptoms as well as diagnostic testing results (Table 1).Both patients with probable PCD had chronic upper airway disease, chronic wet cough, and low nNO levels, while one patient had NRD and chronic Pseudomonas aeruginosa infection, respectively.Among the three patients with possible PCD, all had chronic upper airway disease and chronic wet cough, but normal nNO levels.One of those patients had a positive family history and a history of parental consanguinity, each, while another patient recalled NRD and two had chronic P. aeruginosa infection.All of the three patients with other diagnoses had chronic wet cough and normal nNO levels, while two of them had chronic P. aeruginosa infection (Table 1).We were unable to obtain a robust sample for analysis by TEM and PCD Detect in one patient with a likely pathogenic homozygous variant in the PCD candidate gene NEK10 (Patient 16), even after repeat nasal brush biopsies on different occasions.Overall, we performed 24 TEM evaluations on separate samples from 16 patients, including repeats due to insufficient initial samples, evaluation of test-retest reliability (n = 3, each), and one positive and one negative control, respectively.Due to initially insufficient sample quality, we had to repeat nasal brush biopsies in three patients.In two of those, repeat sampling yielded robust findings (Patients 1 and 4), while we were unable to obtain sufficient samples in the patient with variants in NEK10.Furthermore, we assessed the testretest reliability of TEM evaluation in a subset of three patients.Of those, two patients with normal TEM had consistently normal results at repeat sampling and TEM evaluation.We were unable to obtain a robust sample for analysis by TEM and PCD Detect in one patient with a likely pathogenic homozygous variant in the PCD candidate gene NEK10 (Patient 16), even after repeat nasal brush biopsies on different occasions.Overall, we performed 24 TEM evaluations on separate samples from 16 patients, including repeats due to insufficient initial samples, evaluation of test-retest reliability (n = 3, each), and one positive and one negative control, respectively.Due to initially insufficient sample quality, we had to repeat nasal brush biopsies in three patients.In two of those, repeat sampling yielded robust findings (Patients 1 and 4), while we were unable to obtain sufficient samples in the patient with variants in NEK10.Furthermore, we assessed the test-retest reliability of TEM evaluation in a subset of three patients.Of those, two patients with normal TEM had consistently normal results at repeat sampling and TEM evaluation.However, in one patient with homozygous VUS in DNAI1 (Patient 8), in who we revealed an ODA defect on initial evaluation, the quality of the retest sample was insufficient and precluded repeat evaluation.In order to assure the validity of our TEM evaluation, we included a healthy person as negative control and a person with PCD and known biallelic pathogenic variants in DNAH5 as positive control, respectively.In both control subjects, TEM showed the expected findings (Figures 1 and 2).

Discussion
We analyzed 16 patients with inconclusive results from WES using conventional TEM and averaging of images by the PCD Detect software and, overall, revealed ultrastructural defects in 7 patients (44%).Considering the nine patients with probable PCD and available TEM, this rate was even higher (78%).In six patients the observed ultrastructural hallmark defects confirmed definite PCD.Moreover, we detected a CC (class 2) defect in another patient and, thus, provide supporting evidence for the relevance of genetic variants in NME5 as the cause of PCD.In summary, our findings highlight the use of TEM to validate the diagnosis of PCD in patients with inconclusive WES results and, in addition, demonstrate the feasibility of the PCD Detect software in support of TEM.
PCD is a heterogeneous genetic disorder, with reported prevalence rates relating to geographical variability, ethnicity as well as the expertise and availability of dedicated centers providing appropriate diagnostics [18,19].However, clinical presentation, disease manifestations, comorbidities, and the clinical course of disease may vary according to additional factors, including age (at diagnosis), the gene in which particular genetic variants occur (genotype), and others [20,21].All 16 patients included in our study had bronchiectasis, clinical features suggestive of PCD, and inconclusive WES results.Based on characteristic signs and symptoms as well as basic diagnostic testing like nNO, we grouped the etiology of bronchiectasis in probable PCD, possible PCD, and other diagnoses.Although the detection of low nNO levels is a sensitive and specific diagnostic test for PCD, other diseases like CF may also show low nNO values [1,22].However, seven out of nine patients (78%) with probable PCD and low nNO values had ultrastructural (class 1 and 2) defects in the present study.Notably, the two patients with probable PCD and normal ciliary ultrastructure in TEM had VUS in the PCD genes DNAH11 and DNAH8/HYDIN, which are known to have normal ciliary ultrastructure, although even subtle ultrastructural defects may be visualized by image averaging or electron tomography [17,23].DNAH8 had first been linked to male infertility, but was also described in a patient with PCD [24,25].Therefore, we did not expect to find ultrastructural defects in these patients.In contrast, in six out of eight patients with normal ciliary ultrastructure in TEM the diagnosis of PCD was a priori considered to be less likely.Of those, three patients had possible PCD and three patients had other diagnoses.All had normal nNO levels.This finding supports the measuring of nNO as a screening test along with characteristic history and clinical symptoms, as recommended by current diagnostic guidelines [5,6].However, it should be noted that normal nNO levels have been described in people with PCD associated with genetic variants in several specific genes [1], e.g., NEK10, as observed in our study (see below).
Beyond the clinical likelihood of PCD, TEM findings depended on the respective genetic variants in affected genes and their genotype-phenotype correlations.All seven patients with ciliary class 1 and 2 ultrastructural defects in TEM had genetic variants in established PCD genes or PCD candidate genes with known genotype-phenotype correlations, including specific ultrastructural defects.In agreement with previous reports, we reproduced ODA defects in three patients with VUS in the PCD genes CCDC103, DNAH5, and DNAI1, respectively [26][27][28].Moreover, in three patients with VUS in the PCD gene CCDC40 and another patient with a homozygous variant in the PCD candidate gene NME5, we were able to demonstrate MTD + IDA defects [29] and a CC defect [12], respectively.In addition, we observed characteristic genotype-phenotype correlations concerning clinical features of PCD.While respiratory symptoms virtually occur in all people with PCD, laterality defects, infertility, and congenital heart defects are associated to specific PCD genes [1].In this regard, we observed situs inversus totalis (Kartagener syndrome) in two patients with VUS in CCDC40 and DNAI1 and female infertility in another patient with VUS in CCDC40, as previously reported by others [26,29,30].
We revealed a CC defect of ciliary ultrastructure in a patient with a likely pathogenic homozygous variant in the PCD candidate gene NME5.To date, there is merely one patient reported from South Korea with homozygous nonsense variants in NME5, in whom the central pair was reported to be absent in TEM (9 + 0 structure) [12].In contrast, we also found ciliary cross-sections with normal ultrastructure (9 + 2 structure), but also crosssections with an additional central pair (9 + 4 structure) or a translocated microtubule (8 + 1 structure).In line with the patient reported by Cho and coworkers, our patient had chronic upper and lower respiratory symptoms and situs solitus [12].NME5 encodes for the RSPH23 protein from radial spokes [1,12].Other PCD genes encoding for radial spoke components like RSPH9 are also associated to CC defects [31].To the best of our knowledge, there is no report of laterality defects in patients with (likely) pathogenic variants in PCD genes encoding for radial spoke components, which implies that these do not influence the nodal cilia [32].Noteworthy, CC defects may also occur as secondary changes following infection.Accordingly, BEAT PCD TEM Criteria define them as class 2 defects [7], which only indicate PCD diagnosis and always require additional diagnostics to confirm disease.Taken together, the characteristic clinical features, the reduced nNO levels, abnormal HSVM, positive familial segregation analysis, and the CC defect observed in our patient, support the hypothesis that genetic variants in NME5 may cause PCD in humans.
Our findings have several important implications for the diagnostic work-up of PCD.On the one hand, TEM confirmed definite PCD in a considerable proportion of patients with inconclusive WES results in PCD genes with established genotype-phenotype associations.Remarkably, all subjects with definite PCD and available nNO had levels below the diagnostic threshold.Thus, PCD may be reasonably excluded in patients with bronchiectasis and repeatedly normal nNO levels and VUS in PCD genes with known associated ciliary ultrastructural defects as well as reduced nNO, if TEM unequivocally demonstrates normal findings.On the other hand, TEM cannot rule out or confirm PCD in patients with normal ciliary ultrastructure and VUS in PCD genes that are associated with such, e.g., DNAH11 or HYDIN.Following our diagnostic approach with early application of NGS in patients with suspected PCD, this may suggest to perform TEM as a targeted approach in patients with VUS in PCD genes, which are expected to show ultrastructural defects according to already known genotype-phenotype correlations.However, alternative diagnostics are needed in those patients with VUS in PCD genes expected to show normal ciliary ultrastructure, e.g., HSVM in addition to clinical features and nNO for DNAH11 and immunofluorescence microscopy for HYDIN [1].However, we observed a mutual interaction between genetic testing and TEM in one patient with Kartagener syndrome and VUS in DNAH5, in whom the observed ultrastructural defect (MTD + IDA) did not match the expected outcome and triggered genetic reanalysis and, subsequently, detection of homozygous VUS in CCDC40 that would otherwise have been missed.
Without doubt, we had to deal with the quality of nasal brush biopsies for use in TEM.This highlights the complexity of the procedure and may limit its use for broad application in routine care.Ultimately, this precluded TEM evaluation in the patient with a likely pathogenic homozygous variant in the PCD candidate gene NEK10.Notably, pathogenic variants in NEK10 have very recently been reported to be associated with situs solitus as well as normal nNO levels, while affecting ciliary growth, mucociliary transport, and thereby causing motile ciliopathy [33].In our patient, we performed additional familial segregation analysis of the homozygous variant in NEK10, including parents as well as affected and unaffected siblings, thus suggesting disease causation.Likewise, in the meantime, we produced additional evidence for the relevance of the detected likely pathogenic homozygous variant in NME5 [34], which is beyond the scope of the present work.Lastly, our findings confirm that the PCD Detect software is an easy-to-apply and useful tool, in particular when analyzing faintly contrasted ciliary components like the IDA [17].
Our study has limitations.We assumed that two heterozygous variants were compound heterozygous (in trans), while definite confirmation from the analysis of parents was available in merely two patients.In addition, the investigator who performed the visual evaluation of ciliary ultrastructure by TEM and PCD Detect (B.O.S.) was not blinded to the findings from WES.Moreover, the assessment of fertility in clinical routine did not follow a standardized approach.Therefore, we may have missed cases of subfertility.Finally, it should be noted that ultrastructural hallmark defects alone are insufficient to reclassify VUS as pathogenic variants, according to ACMG guidelines that require "well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or gene product" [8].In this respect, ultrastructural defects are considered to be unspecific for disease-causing variants in a unique PCD gene, as they may theoretically also be associated with other PCD genes, e.g., MTD + IDA defects are associated with CCDC40, but also occur in people with variants in CCDC39 [1].

Conclusions
In the present study, we established the definite diagnosis of PCD in a considerable proportion of patients with inconclusive results from WES by TEM, thus providing important information on genotype-phenotype correlations regarding VUS in CCDC40, CCDC103, DNAH5, and DNAI1 as well as a likely pathogenic variant in the PCD candidate gene NME5.This may suggest a targeted approach within the diagnostic work-up for PCD, requiring confirmation among a larger patient population from a multicenter study.However, the finding that we detected an ultrastructural hallmark defect in one patient, which reversely informed the interpretation of genetic analysis and resulted in the definite diagnosis of PCD associated with VUS in CCDC40, supports its broader application.Further, the PCD Detect software proved to be a feasible tool in addition to the visual analysis of conventional TEM.

Figure 1 .
Figure 1.Microtubular disorganization and inner dynein arm (MTD + IDA) defects in three patients with VUS in CCDC40.Top: TEM images (greyscale); below: averaged cutout images from the PCD Detect software (colored), representing one microtubule doublet corresponding to the pair oriented downward in TEM.Negative control shows normal ciliary ultrastructure with 9 + 2 arrangement and both dynein arms present (arrowheads).In contrast, arrangement of nine microtubule doublets

Figure 1 .
Figure 1.Microtubular disorganization and inner dynein arm (MTD + IDA) defects in three patients with VUS in CCDC40.Top: TEM images (greyscale); below: averaged cutout images from the PCD Detect software (colored), representing one microtubule doublet corresponding to the pair oriented

Figure 2 .
Figure 2. Outer dynein arm (ODA) defects in three patients with VUS in CCDC103, DNAH5, and DNAI1.Ciliary cross-sections in TEM images (greyscale images in upper row) and cutout images of microtubular doublets and ODA (if present) averaged with PCD Detect (colored images in lower row).ODA (white filled arrows) are absent in TEM images of three patients with VUS in CCDC103, DNAH5, and DNAI1, as well as the positive control (DNAH5).PCD Detect images of the positive control and Patient 8 show a partial loss of ODA with a remaining shortened ODA projection (black filled arrows), whereas Patients 1 and 4 show a complete loss of ODA (white filled arrows).Color scale: Black (TEM) and blue (PCD Detect) as well as white (TEM) and yellow (PCD Detect) indicate high and low electron-density, respectively.Scale bars: 100 nm in TEM images and 10 nm in PCD Detect images.Abbreviations: ODA = outer dynein arm; TEM = transmission electron microscopy; VUS = variant of unknown significance.

Figure 2 .
Figure 2. Outer dynein arm (ODA) defects in three patients with VUS in CCDC103, DNAH5, and DNAI1.Ciliary cross-sections in TEM images (greyscale images in upper row) and cutout images of microtubular doublets and ODA (if present) averaged with PCD Detect (colored images in lower row).ODA (white filled arrows) are absent in TEM images of three patients with VUS in CCDC103, DNAH5, and DNAI1, as well as the positive control (DNAH5).PCD Detect images of the positive control and Patient 8 show a partial loss of ODA with a remaining shortened ODA projection (black filled arrows), whereas Patients 1 and 4 show a complete loss of ODA (white filled arrows).Color scale: Black (TEM) and blue (PCD Detect) as well as white (TEM) and yellow (PCD Detect) indicate high and low electron-density, respectively.Scale bars: 100 nm in TEM images and 10 nm in PCD Detect images.Abbreviations: ODA = outer dynein arm; TEM = transmission electron microscopy; VUS = variant of unknown significance.

DNAI1.
Ciliary cross-sections in TEM images (greyscale images in upper row) and cutout images of microtubular doublets and ODA (if present) averaged with PCD Detect (colored images in lower row).ODA (white filled arrows) are absent in TEM images of three patients with VUS in CCDC103, DNAH5, and DNAI1, as well as the positive control (DNAH5).PCD Detect images of the positive control and Patient 8 show a partial loss of ODA with a remaining shortened ODA projection (black filled arrows), whereas Patients 1 and 4 show a complete loss of ODA (white filled arrows).Color scale: Black (TEM) and blue (PCD Detect) as well as white (TEM) and yellow (PCD Detect) indicate high and low electron-density, respectively.Scale bars: 100 nm in TEM images and 10 nm in PCD Detect images.Abbreviations: ODA = outer dynein arm; TEM = transmission electron microscopy; VUS = variant of unknown significance.

Figure 4 .
Figure 4. Normal ciliary ultrastructure in four patients with VUS in CCDC39, DNAH11, DNAI1, and DNAH7.TEM images of four patients with VUS show normal 9 + 2 structure and both dynein arms present (arrowheads) in conventional TEM imaging (greyscale images in upper row) and PCD Detect imaging (colored images in lower row).Color scale: Black (TEM) and blue (PCD Detect) as well as white (TEM) and yellow (PCD Detect) indicate high and low electron-density, respectively.Scale bars: 100 nm in TEM images and 10 nm in PCD Detect images.Abbreviations: TEM = transmission electron microscopy; VUS = variant of unknown significance.

Figure 4 .
Figure 4. Normal ciliary ultrastructure in four patients with VUS in CCDC39, DNAH11, DNAI1, and DNAH7.TEM images of four patients with VUS show normal 9 + 2 structure and both dynein arms present (arrowheads) in conventional TEM imaging (greyscale images in upper row) and PCD Detect imaging (colored images in lower row).Color scale: Black (TEM) and blue (PCD Detect) as well as white (TEM) and yellow (PCD Detect) indicate high and low electron-density, respectively.Scale bars: 100 nm in TEM images and 10 nm in PCD Detect images.Abbreviations: TEM = transmission electron microscopy; VUS = variant of unknown significance.