Bi-allelic variants in DNAH3 cause male infertility with asthenoteratozoospermia in humans and mice

Abstract STUDY QUESTION Are there other pathogenic genes for asthenoteratozoospermia (AT)? SUMMARY ANSWER DNAH3 is a novel candidate gene for AT in humans and mice. WHAT IS KNOWN ALREADY AT is a major cause of male infertility. Several genes underlying AT have been reported; however, the genetic aetiology remains unknown in a majority of affected men. STUDY DESIGN, SIZE, DURATION A total of 432 patients with AT were recruited in this study. DNAH3 mutations were identified by whole-exome sequencing (WES). Dnah3 knockout mice were generated using the genome editing tool. The morphology and motility of sperm from Dnah3 knockout mice were investigated. The entire study was conducted over 3 years. PARTICIPANTS/MATERIALS, SETTING, METHODS WES was performed on 432 infertile patients with AT. In addition, two lines of Dnah3 knockout mice were generated. Haematoxylin and eosin (H&E) staining, transmission electron microscopy (TEM), immunostaining, and computer-aided sperm analysis (CASA) were performed to investigate the morphology and motility of the spermatozoa. ICSI was used to overcome the infertility of one patient and of the Dnah3 knockout mice. MAIN RESULTS AND THE ROLE OF CHANCE DNAH3 biallelic variants were identified in three patients from three unrelated families. H&E staining revealed various morphological abnormalities in the flagella of sperm from the patients, and TEM and immunostaining further showed the loss of the central pair of microtubules, a dislocated mitochondrial sheath and fibrous sheath, as well as a partial absence of the inner dynein arms. In addition, the two Dnah3 knockout mouse lines demonstrated AT. One patient and the Dnah3 knockout mice showed good treatment outcomes after ICSI. LARGE SCALE DATA N/A. LIMITATIONS, REASONS FOR CAUTION This is a preliminary report suggesting that defects in DNAH3 can lead to asthenoteratozoospermia in humans and mice. The pathogenic mechanism needs to be further examined in a future study. WIDER IMPLICATIONS OF THE FINDINGS Our findings show that DNAH3 is a novel candidate gene for AT in humans and mice and provide crucial insights into the biological underpinnings of this disorder. The findings may also be beneficial for counselling affected individuals. STUDY FUNDING/COMPETING INTEREST(S) This work was supported by grants from National Natural Science Foundation of China (82201773, 82101961, 82171608, 32322017, 82071697, and 81971447), National Key Research and Development Program of China (2022YFC2702604), Scientific Research Foundation of the Health Committee of Hunan Province (B202301039323, B202301039518), Hunan Provincial Natural Science Foundation (2023JJ30716), the Medical Innovation Project of Fujian Province (2020-CXB-051), the Science and Technology Project of Fujian Province (2023D017), China Postdoctoral Science Foundation (2022M711119), and Guilin technology project for people’s benefit (20180106-4-7). The authors declare no competing interests.


Introduction
Infertility has become a global health problem that affects an estimated 15% of couples (Sang et al., 2023), and infertility among men accounts for approximately half of all cases (Krausz and Riera-Escamilla, 2018).Asthenoteratozoospermia (AT) is a prevalent condition underlying male infertility characterized by a proportion of spermatozoa with low progressive motility (below 32%) and a low proportion with normal morphology (below 4%) (Coutton et al., 2015;Tu et al., 2020b).A multitude of causes and risk factors may contribute to this disorder, including genetic causes, environmental factors, socio-demographic risk factors, and behavioural/lifestyle risk factors (Tournaye et al., 2017;Agarwal et al., 2021;Okonofua et al., 2022).Genetic deficiency is an important causal factor (Sudhakar et al., 2021), and several genes have been reported to be responsible for AT, including those in the cilia and flagellum-associated protein (CFAP) family, the coiled-coil domain-containing family, and the dynein axonemal heavy chain (DNAH) family (Sha et al., 2019;Tu et al., 2019;Wang et al., 2019;Chen et al., 2021;Zhang et al., 2022;Zhuang et al., 2022).
The DNAH family genes encode axonemal dynein heavy chains associated with the assembly of the inner and outer dynein arms (IDAs and ODAs) of sperm flagella.Currently, 13 members of the DNAH gene family have been reported, including DNAH1 to DNAH3, DNAH5 to DNAH12, DNAH14, and DNAH17, among which 11 genes (DNAH1, DNAH2, DNAH5 to DNAH12, and DNAH17) are associated with male infertility (Yagi, 2009;Oud et al., 2021;Levkova et al., 2022).Variants in seven genes of the DNAH family (DNAH1 and DNAH2, DNAH6 to DNAH8, DNAH10, and DNAH17) have been reported to cause AT.However, whether DNAH3 is involved in male infertility remains unclear.
In this study, the genetic aetiology of AT was investigated.Biallelic DNAH3 variants were identified in three individuals with AT from three independent families using whole-exome sequencing (WES).Furthermore, two lines of Dnah3 knockout mice were generated, both of which demonstrated AT.Our findings suggest that DNAH3 is a novel candidate gene for AT in humans and mice.

Human subjects
A total of 432 AT-affected, infertile Chinese men were recruited from the Reproductive and Genetic Hospital of CITIC-Xiangya (Changsha, China) and the Women and Children's Hospital of Xiamen University (Fujian, China).Every individual was idiopathic with the normal number of chromosomes ( 46), and they presented the XY karyotype, which excluded Y-chromosome microdeletions and other potential causes of infertility, including iatrogenic injury, genital tract infection, testicular inflammation, and drug exposure.In addition, no abnormalities in height, weight, hair distribution, mental status, testicular size, or external genitalia were found upon physical examination.A total of 219 fertile Chinese men with normal semen parameters were recruited as controls.This study was approved by the Ethics Committees of the Reproductive and Genetic Hospital of CITIC-Xiangya (LL-SC-2019-034) and the Women and Children's Hospital of Xiamen University (KY-2019-060).All participants provided written informed consent.

WES and in silico bioinformatics analysis
Genomic DNA (gDNA) was extracted from peripheral blood samples using the QIAamp DNA Blood Midi Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions, and subsequently subjected to WES analysis using the Agilent SureSelect Human All Exon V6 Kit (Agilent Technologies, Santa Clara, CA, USA) and the Illumina HiSeq 2000 or HiSeq X-TEN platform (Illumina, San Diego, CA, USA), as described previously (Tu et al., 2021;Hu et al., 2023).The raw reads were aligned to the human genome assembly GRCh37/hg19 using the Burrows-Wheeler Aligner after removing adaptors (Li and Durbin, 2009).Candidate pathogenic variants were identified using methods previously described in detail (Tan et al., 2019).PCR and Sanger sequencing were performed to verify variants in the probands and their family members.The primers are listed in Supplementary Table S1.

WHAT DOES THIS MEAN FOR PATIENTS?
Asthenoteratozoospermia (AT) is a prevalent condition which affects the motility and morphology of spermatozoa.It is an underlying cause of male infertility and is usually attributed to genetic anomalies.Our study identified genetic variants in the DNAH3 gene in three patients with AT from three unrelated families.Two lines of mice with the DNAH3 gene knocked out demonstrated sperm characteristics similar to those in human patients.These findings indicate that DNAH3 is a novel candidate pathogenic gene for AT and enriches the genetic aetiology of this kind of disease.
Additionally, one patient, as well as the mice with the DNAH3 defects, showed good treatment outcomes with intracytoplasmic sperm injection (ICSI).This suggests that DNAH3-associated infertile males with AT have the choice of receiving clinically effective fertility therapy.Therefore, our results could provide insights for genetic counselling of individuals diagnosed with AT.

Semen parameter analysis and computer-aided sperm analysis
According to the guidelines of the WHO's laboratory manual for examining and processing human semen (fifth edition), semen samples from subjects P1, P2, and P3 in this study were collected through masturbation after 2-7 days of sexual abstinence and analysed after liquefaction at 37 � C for 30 min (Cooper et al., 2010).Routine semen tests were performed to analyse the semen volume and sperm concentration and motility.Sperm morphology was observed using haematoxylin and eosin (H&E) staining of at least 200 spermatozoa selected from patients to assess the percentage of morphologically abnormal spermatozoa.
Semen samples from two mouse models were collected from the cauda epididymis based on previously described methods and then diluted in 1 ml of capacitation solution for incubation at 37 � C for 30 min (He et al., 2020).The suspension was dropped into Sperm Counting Chambers (SAS Medical, Beijing, China), and the motility of the spermatozoa was assessed using computer-aided sperm analysis (CASA) systems (HTM-TOX IVOS version 12, Hamilton Thorne Research, Beverly, MA, USA).At least three WT and mutant mice (at least 8-week-old) were analysed in each group.

Haematoxylin-eosin staining
Haematoxylin-eosin staining of spermatozoa and testicular and epididymal tissues was performed according to a previous study (Wang et al., 2019(Wang et al., , 2021)).Briefly, semen samples from patients and mice were collected, diluted, and fixed in 4% paraformaldehyde (PFA) (Sigma-Aldrich, 158127, St Louis, MO, USA) on slides for 30 min to prepare sperm smears for subsequent experiments.Testicular and epididymal tissues from mice were fixed in picric acid, 4% PFA, or Bouin's solution (Sigma-Aldrich, HT10132) and then embedded, sectioned, processed, and cut into tissue sections.Sperm smears were hydrated by decreasing the concentration of graded ethanol, followed by staining with haematoxylin (Servicebio, G1004, Wuhan, China) and eosin (Servicebio, G1001), or haematoxylin-eosin (Solarbio Science & Technology, G1120, Beijing, China).The slides were then dehydrated with anhydrous ethanol at increasing concentrations from 50% to 100%, treated with xylene for clearing, and sealed with neutral resin for observation and imaging.The tissue sections from mice were first treated in an oven at 65 � C for 2 h and then dewaxed with turpentine oil for 20 min.The subsequent processing procedures of the tissue sections were consistent with those of the sperm smears.Brain and trachea were fixed in 4% PFA, embedded in paraffin, sectioned to a 3 lm thickness, and stained with haematoxylineosin as previously reported (Zhou et al., 2023b).

Transmission electron microscopy
Sperm samples from human subjects and mice were obtained for transmission electron microscopy (TEM) as described previously (Tan et al., 2019).Sperm samples were fixed with 2.5% glutaraldehyde and OsO 4 , treated with OsO 4 and sucrose, and dehydrated with graded concentrations of ethanol.Subsequently, the samples were embedded in Epon812, dodecenylsuccinic anhydride, methylnadic anhydride, and dimethylaminomethyl phenol.Ultrathin sections (70-90 nm) were stained with uranyl acetate and lead citrate.Images were captured using an HT7700 Hitachi electron microscope (Hitachi, Tokyo, Japan) equipped with a MegaView III digital camera (EMSIS GmbH, M€ unster, Germany) or a TECNAI-10 transmission electron microscope (Philips-FEI, Amsterdam, Netherlands).

Generation of Dnah3-knockout mouse models
Two lines of Dnah3-knockout (KO) mouse models were generated.The Dnah3 KO1 line was generated by Shanghai Model Organisms Centre, Inc. (Shanghai, China) using CRISPR-Cas9, while the KO2 line was constructed by inducing a nonsense mutation in mouse zygotes using ISTOP technology (Wang et al., 2023a).Briefly, sgRNA1 (target: 5 0 -ATTGTTGGAGATCCAATGGG-3 0 ) and sgRNA2 (target: 5 0 -TTTCCAGTAAGTATCCACAG-3 0 ) were designed against exons 35 and 22 of the mouse Dnah3 gene to generate the KO1 and KO2 lines, respectively.Gene editing components were prepared and injected into mouse zygotes to obtain the F0 generation, which were mated with C57BL/6 WT mice to obtain F1 generation mice.Homozygous Dnah3 KO mice (Dnah3 ko1/ko1 and Dnah3 ko2/ko2 ) were obtained by intercrossing heterozygous mice.Founder and offspring mice were genotyped by PCR and Sanger sequencing using the primers listed in Supplementary Table S2.Adult mice aged eight weeks or older were selected for this study.All animals were quarantined by the Department of Laboratory Animals.In addition, all animal experiments were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committees of Central South University and Fudan University.

Reverse-transcription PCR (RT-PCR)
RT-PCR was conducted to confirm the mRNA expression levels of Dnah3 in various tissues and testicular tissues from C57BL/6N mice at different ages.Total RNA was extracted from the heart, liver, spleen, lung, kidney, stomach, intestine, testis, epididymis, and brain of the mice using TRIzol reagent (Invitrogen, 15596026, Carlsbad, CA, USA) and then reverse transcribed into cDNA using the PrimeScript RT reagent Kit (Takara, RR047Q, Kyoto, Japan) or Hiscript III 1st Stand cDNA Synthesis Kit (Vazyme, R312-02, Nanjing, China) according to the manufacturer's instructions.mRNA expression levels of b-actin or Gapdh were used as internal controls.Primers used are listed in Supplementary Table S3.

Intracytoplasmic sperm injection analysis in human and mice
ICSI for human subjects was performed at the Reproductive and Genetic Hospital of CITIC-Xiangya, as described in our earlier study (Tan et al., 2022).ICSI with mouse sperm was performed as previously described (Liu et al., 2021;Cong et al., 2022).Briefly, cauda epididymal sperm were obtained and maintained in HTF medium (Millipore, MR-070-D, Billerica, MA, USA) (Wang et al., 2023b).Two-month-old B6D2F1 WT female mice were superovulated using 7.5 IU of pregnant mare serum gonadotropin, followed by 7.5 IU of hCG 48 h later, and then oocytes were collected.Sperm were injected into oocytes, and the ICSI embryos were then cultured in KSOMþAA WITH D-GLUCOSE medium (Millipore, MR-107-D) at 37 � C under 5% CO 2 .

Identification of bi-allelic DNAH3 variants carried in infertile men with AT
A cohort of 432 unrelated infertile men with AT was recruited to investigate the genetic aetiology through WES.After variant filtering and a functional analysis, �28% of patients had an identified genetic cause of AT, including gene mutations previously reported by our group (i.e.DNAH10 (Tu et al., 2021), DRC3 (Zhou et al., 2023a), DNHD1 (Tan et al., 2022), SPEF2 (Tu et al., 2020a), CFAP54 (Tian et al., 2023), TTC12 (Meng et al., 2023), and CFAP65 (Wang et al., 2021)), and other unreported mutations.This study focused on the novel disease-causing gene, DNAH3.Six variants of DNAH3 (GenBank: NM_017539.2) were identified in three individuals (P1, P2, and P3) from three non-consanguineous families (Fig. 1A).Segregation analysis was performed via PCR-Sanger sequencing of the three families using the primers listed in Supplementary Table S1.The six mentioned variants have been submitted to the ClinVar database.In family I, the patient (P1) harboured two compound heterozygous variants c.5143G>A (p.Gly1715Ser) and c.7477G>A (p.Asp2493Asn) of DNAH3; in family Ⅱ, the patient (P2) also carried two compound heterozygous variants c.6973T>C (p.Phe2325Leu) and c.8971C>T (p.Arg2991Cys) of DNAH3; and in family III, the patient (P3) carried two compound heterozygous variants c.10260 G>A (p.Trp3420X) and c.10439G>A (p.Arg3480Gln) of DNAH3 (Fig. 1A and Supplementary Table S4).All variants of DNAH3 were recorded as rare or absent and predicted to be deleterious via in silico bioinformatics databases (1000 Genomes Project and gnomAD) and tools (PolyPhen-2, Mutation Taster, and CADD) (Supplementary Table S4).Additionally, the parents of the three patients were all heterozygous carriers in accordance with the autosomal recessive mode of inheritance (Fig. 1A).
Human DNAH3 is predicted to encode a 4116-amino acid protein consisting of a tail, a microtubule binding domain (MTBD), and six AAA þ domains (Fig. 1B).These six altered amino acids were highly conserved among multiple species (Fig. 1B).Threedimensional modelling of the DNAH3 protein (NP_060009.1)showed that the six mutations disrupted hydrogen bonds or altered the distance between atoms of adjacent amino acids, which is likely to disrupt protein stability and function (Supplementary Fig. S1).Furthermore, RT-PCR showed that Dnah3 was abundantly expressed in the testes of mice and dramatically increased from postnatal day 21, which corresponds to the spermiogenesis stage (Supplementary Fig. S2).These findings suggest that the identified biallelic DNAH3 variants may represent the genetic aetiology of the three AT-affected men.

AT phenotypes in men with bi-allelic DNAH3 variants
Routine semen analysis showed that sperm motility in three patients harbouring the bi-allelic DNAH3 variants was dramatically decreased (Table 1).Semen samples from the three patients were collected for sperm morphology analysis using H&E staining and DNAH3 expression analysis by western blotting.Compared to spermatozoa with a normal length and smooth tail from the normal control, the spermatozoa from P1, P2, and P3 exhibited multiple morphological abnormalities of the flagella (MMAF), including short, coiled, absent, angulated, and irregular calibre flagella (Fig. 1C).In addition, the expression level of DNAH3 was lower in spermatozoa from P1 compared with that in sperm from a normal control (Fig. 1D and E).Taken together, these results demonstrate that biallelic DNAH3 variants may contribute to the AT phenotype in infertile men.

DNAH3 variants are associated with aberrant axoneme and axonemal accessory structures
The sperm flagellum contains an important core axoneme structure, a central pair (CP) of microtubules, nine peripheral doublets of microtubules, and other accessory structures, such as the mitochondrial sheath (MS) in the mid-piece and the fibrous sheath (FS) in the principal piece.Thus, we performed a TEM analysis to observe transverse and longitudinal sections of the sperm flagella from patient P1 harbouring bi-allelic DNAH3 variants.Compared with a normal typical "9 þ 2" axoneme and accessory structure of sperm flagella from the fertile control, the sperm flagella of the patient showed obviously disordered axonemal ultrastructures, including absence of the CP and disorganization of the MS and FS (Fig. 2A).Furthermore, a poorly assembled MS was also observed in longitudinal sections of the sperm flagella (Fig. 2A).Interestingly, the IDAs were partially absent in different parts of the sperm flagella (Fig. 2B).
To confirm the ultrastructural defects revealed by TEM, immunofluorescence assays were performed to detect the expression of several proteins, including TOMM20 (component of the MS), SPAG6 (component of the CPs), and AKAP4 (component of the FS).The results showed an absent SPAG6 signal and mislocated TOMM20 and AKAP4 signals in spermatozoa from P1 (Supplementary Fig. S3A-C).
As multiple morphologically abnormal sperm flagella were displayed in men harbouring bi-allelic DNAH3 variants, periaxoneme structures, such as the basal body and annulus, which are involved in the formation of the sperm tail and flagellar compartmentalization, were further investigated via immunofluorescence.Spermatozoa were stained for SEPTIN12 (component of Bi-allelic DNAH3 variants cause male infertility with AT | 5 annulus) and CENTRIN (component of basal body), which showed that a bright SEPTIN12 signal was clustered at the junction between the midpiece and principal piece of normal spermatozoa.However, both the localization and expression of the SEPTIN12 signal were anomalous in spermatozoa from P1 (Supplementary Fig. S3D and E).In addition, more than two CENTRIN signals were scattered in the spermatozoa of P1, whereas two adjacent signals located at the connecting piece of normal spermatozoa (Supplementary Fig. S3F and G).The proportion of abnormal signals in the annulus and centriole in P1 was substantially higher than that in the normal controls (Supplementary Fig. S3E and G).Collectively, these results suggest that DNAH3 is essential for the assembly of the axoneme and axonemal accessory structures in the sperm.

Dnah3 KO mice presented damaged male fertility with AT phenotypes
To further evaluate the role of DNAH3 in spermatogenesis and flagellogenesis in mice, we generated two Dnah3 KO mouse models (Dnah3 ko1/ko1 ) and (Dnah3 ko2/ko2 ) by inserting a base T in exon 35 of Dnah3 (c.5039_5040 ins T) and inducing a nonsense mutation (c.3227_3228 GG>AA) in exon 22 of Dnah3 via gene editing (Supplementary Fig. S4A and B).To further confirm the result, we analysed the potential off-target mutations in sites predicted in silico by CasOFFinder (Bae et al., 2014), and observed no off-target mutations in these sites (Supplementary Table S5).Western blots revealed that DNAH3 was undetectable in the testes of Dnah3 ko1/ko1 and Dnah3 ko2/ko2 mice (Supplementary Fig. S4C and  D).Fertility tests of Dnah3 þ/þ , Dnah3 ko1/ko1 , and Dnah3 ko2/ko2 male mice (8-11 weeks old) were performed by mating these mice with fertile Dnah3 þ/þ female mice (8 weeks old).The results showed that Dnah3 ko1/ko1 male mice were completely infertile, and Dnah3 ko2/ko2 presented severely disrupted male fertility (Fig. 3A and Supplementary Fig. S5A).Additionally, no significant differences in testicular size or testis/body weight ratio were observed between Dnah3 þ/þ male mice and Dnah3 ko1/ko1 and Dnah3 ko2/ko2 male mice (Fig. 3B and C, Supplementary Fig. S5B and C).H&E staining revealed the presence of spermatogenic cells at different stages in the testes of Dnah3 ko1/ko1 and Dnah3 ko2/ko2 mice (Fig. 3D and Supplementary Fig. S5D).
Sperm samples were collected from the mouse cauda epididymis to analyse sperm morphology and motility.H&E staining of sperm indicated that the sperm tail of Dnah3 ko1/ko1 and Dnah3 ko2/ko2 mice showed curved and irregular morphology, with a higher proportion of abnormal sperm flagella than that of Dnah3 þ/þ mice (Fig. 3E and F, Supplementary Fig. S5E and F).We also investigated the ultrastructure of the sperm axoneme by TEM.Dnah3 ko1/ko1 and Dnah3 ko2/ko2 male mice displayed partially absent IDAs but showed the typical "9 þ 2" axonemal structure and peri-axonemal structures (Fig. 3G, Supplementary Fig. S5G).The percentage of axonemes with missing IDAs was higher in Dnah3 knock-out mice than that in Dnah3 þ/þ mice (Fig. 3H).Western blotting was performed to validate the IDA phenotype.The levels of DNALI1, DNAH2, and DNAH6 (different components of IDA) were lower in Dnah3 ko1/ko1 mice than in Dnah3 þ/þ mice (Supplementary Fig. S6).
CASA analysis was performed to test the motility of epididymal sperm.Sperm motility, progressive motility, and swimming velocity in Dnah3 ko1/ko1 and Dnah3 ko2/ko2 mice were all lower than those in Dnah3 þ/þ mice (Supplementary Fig. S7).Furthermore, we conducted H&E staining to explore whether cilia from trachea or brain were impaired in Dnah3-KO mice.We found that like WT controls, the gross ciliary morphology of trachea and brain appear normal in Dnah3-KO mice (Supplementary Fig. S8).Collectively, these data demonstrate that DNAH3 defects contribute to male infertility associated with AT in both humans and mice.

ICSI treatment on the DNAH3-associated male infertility
To evaluate whether male infertility associated with AT caused by DNAH3 variants can be overcome with assisted reproductive technology, ICSI was performed using the spermatozoa of Dnah3 þ/þ and Dnah3 KO1 male mice.The rates of two-cell embryos and blastocysts in Dnah3 KO1 group was similar to those in the WT group (Fig. 4).Patient (P1) received ICSI treatment, and 12 embryos developed to the blastocyst stage.P1's wife accepted a single embryo transfer and eventually became pregnant.Overall, these results indicate that ICSI may be an effective strategy for infertile men harbouring DNAH3 variants with AT to father offspring.

Discussion
Thirteen members of the DNAH gene family are highly expressed in human testes.Eleven of them have been found to be involved in male infertility, whereas the role of DNAH3 and DNAH14 in male infertility has not been clarified.In this study, we identified biallelic variants of DNAH3 (p.Gly1715Ser and p.Asp2493Asn, p.Phe2325Leu and p.Arg2991Cys, and p.Trp3420X and p.Arg3480Gln) in three unrelated infertile men whose spermatozoa displayed multiple morphological abnormalities of the flagella and decreased motility.Subsequently, two lines of Dnah3 KO mice were generated by inducing a frameshift or nonsense mutation, which recapitulated the patients with AT.This is the first report showing that DNAH3 mutations are responsible for AT in humans and mice and providing new evidence for members of the DNAH gene family in spermatogenesis.
Dynein heavy chain contains eight functional domains, namely, six AAA domains with ATP hydrolysis sites, an MTBD that binds microtubules in the cleft between a-and b-tubulin, and an N-terminal tail domain that anchors to the A-microtubule of the axoneme (King, 2012;Redwine et al., 2012).In our study, biallelic variants of DNAH3 were identified in three infertile men and co-segregated with AT in the pedigree analysis.DNAH3 expression was observed in several axonemal-associated structures (lung, testis, and epididymis) and restricted to the spermiogenesis stage of the testis.Additionally, variants (p.Gly1715Ser and p.Asp2493Asn) in the patient from family I (P1) were both located The semen parameters and sperm flagellar morphology observed in infertile individuals were evaluated according to the World Health Organization guideline (Cooper et al., 2010).At least 200 spermatozoa were observed for the morphology analysis.
in the AAA domain of the DNAH3 protein, and variants (p.Phe2325Leu, p.Arg2991Cys, and p.Arg3480Gln, respectively) in patients from families II (P2) and III (P3) were located beside the AAA domain.These five variants were shown to be highly conserved across multiple species, and the nonsense variant (p.Trp3420X) resulted only in the retention of the AAA1-AAA4 domains.Patient P1 showed the most severe phenotype with 0% sperm motility, which is a different phenotype compared to other patients with low motility.This difference could be explained by the presence of two variants (M1 and M2) in the P1 patient, both of which were located in the AAA domain.Variants M3, 4, and 6 in the other two patients (P2, P3) were not located in the AAA domain.All of the mentioned DNAH3 variants were rare or absent in the databases and were predicted to be deleterious.Furthermore, Dnah3 knockout mice displayed AT, and the phenotype was similar to that observed in the human patients.These , and Dnah3 ko1/ko1 male mice at 2 months of age.The Dnah3 ko1/ko1 male mice showed complete sterility.���� P < 0.0001.(B) Gross morphology comparison between the testicular size of , and Dnah3 ko1/ko1 male mice.(C) Analysis of testis weight/body weight ratios between Dnah3 þ/þ , Dnah3 þ/ko1 , and Dnah3 ko1/ko1 male mice.ns indicates not significant.(D) H&E staining of seminiferous tubules and cauda epididymis from 2-month-old Dnah3 þ/ko1 and Dnah3 ko1/ko1 male mice.Scale bars, 100 lm.(E) H&E staining of epididymis spermatozoa from 2-month-old Dnah3 þ/ko1 and Dnah3 ko1/ko1 male mice.The aberrant curved flagellar morphology was found in spermatozoa of Dnah3 ko1/ko1 male mice.Scale bars, 10 lm.(F) Statistical data showed that the percentage of abnormal sperm flagella in Dnah3 ko1/ko1 male mice was significantly higher than that in Dnah3 þ/ko1 male mice.��� P < 0.001.(G) Transmission electron microscopy analysis of cross-sectional ultrastructure of cauda epididymal spermatozoa from Dnah3 þ/þ and Dnah3 ko1/ko1 male mice at 2 months of age.The typical '9 þ 2' axoneme structure and axonemal accessory structure, including central pairs (CPs; pink arrow), mitochondrial sheath (MS; yellow arrow), outer dense fibres (ODFs; green arrow), fibrous sheath (FS; red asterisk), and outer dynein arms (ODAs; blue triangle) were visible and intact in the spermatozoa from both Dnah3 þ/þ and Dnah3 ko1/ko1 male mice.However, there was partial loss of inner dynein arms (indicated by yellow triangles) in the Dnah3 ko1/ko1 male mice compared with the intact IDAs (green triangle) in the Dnah3 þ/þ male mice.The structures in the red dotted square are magnified proportionally on the right side.Scale bars, 200 nm.(H) Statistical data showing that the percentage of the axoneme with IDA missing in the sperm flagella of Dnah3 ko1/ko1 male mice was significantly higher than that of Dnah3 þ/þ male mice.��� P < 0.001.
results suggest that the identified novel DNAH3 variants represent the aetiology of the three infertile males.Dyneins are a class of motor proteins involved in intracellular cargo transport and ciliary/flagellar movement that are composed of four subunits: heavy chains, intermediate chains, light intermediate chains, and light chains (Antony et al., 2021;Wu et al., 2021).The DNAH family genes encode dynein axonemal heavy chains (King, 2016).Several members of the DNAH family are reportedly involved in AT (Levkova et al., 2022;Yogo, 2022).Defects in these genes can induce infertility in individuals with disrupted sperm flagellar structures and severely decreased spermatozoa motility.For example, patients carrying the biallelic DNAH17 mutation present with AT characterized by the absence of ODAs and a disorganized axoneme (Whitfield et al., 2019).In our study, defects in DNAH3 caused a partial loss of the IDAs, leading to AT in humans and mice.Our findings imply that the DNAH family genes are indispensable for the composition of sperm axonemal dynein arms.
Mouse models are useful tools for identifying novel disease genes and understanding the molecular mechanisms of human genetic diseases, because their genomes are similar to that of humans (Vandamme, 2014).However, mouse models may not completely mimic the disease state of patients in some cases, and different phenotypes between humans and mice have been described.Khan et al. (2021) reported that although DNAH1 defects cause male infertility in humans and mice, the phenotypes of humans and mice present subtle differences.Male patients with DNAH1 variants suffered from MMAF, whereas approximately half of the spermatozoa from Dnah1 knockout mice displayed normal flagella.This may be due to the species differences between mice and humans.In our study, two lines of Dnah3 KO mice were generated, both of which exhibited disrupted fertility.Morphological abnormalities of sperm from Dnah3 KO mice were milder than those from individuals with defects in DNAH3, but the levels of several dynein proteins were decreased in Dnah3 knockout mice.Therefore, we suggest that DNAH3 plays a major role in the assembly of sperm flagella in humans and mice.
The mechanism underlying AT caused by the DNAH family genes remains largely unknown.Patients with AT resulting from a deficiency in DNAH family genes usually present with an aberrant distribution of axoneme and peri-axoneme structures.For example, AT-affected men harbouring biallelic variants of DNAH1, DNAH2, DNAH6, or DNAH10 present with obviously absent CPs and IDAs and other structural defects in the sperm flagella (disorganized MS, outer dense fibres, and FS) (Li et al., 2019(Li et al., , 2022;;Tu et al., 2019;Hu et al., 2021).The mechanisms underlying this phenotype may be attributable to defects in intramanchette transport and intraflagellar transport (IFT), which are essential for sperm tail assembly (Engel et al., 2012;San Agustin et al., 2015;Lehti and Sironen, 2016).The proteins of the DNAH family may participate in the IFT pathway as cargo and/or motor proteins.Defective DNAH family proteins may obstruct the IFT pathway, which leads to abnormal assembly of the sperm axoneme and damage to the accessory structures (Tu et al., 2021).In our study, several dynein proteins (DNAH2, DNAH6, and DNALI1) were decreased in Dnah3 knockout mice.Additionally, the annulus of the sperm flagella was mislocalized in our patients, as has also been observed in Dnah2 knockout male mice (Hwang et al., 2021).These findings indicate that DNAH3 may participate in the assembly of sperm flagella through the IFT pathway and may play a role in the migration of the annulus in the sperm.
Although an increasing number of genes have been identified in the genetic aetiology of male infertility, there are no effective treatments to improve semen quality in infertile males with genetic defects.ICSI is an assisted reproductive technology that has been proposed as a therapeutic solution for infertile males with AT (Ferreux et al., 2021).Previous studies have revealed that patients with AT caused by mutations in the DNAH gene family, including DNAH1, DNAH2, DNAH7, DNAH8, and DNAH10, can achieve pregnancy following ICSI (Wambergue et al., 2016;Li et al., 2019;Tu et al., 2021;Wei et al., 2021;Weng et al., 2021).In our study, P1 carrying bi-allelic DNAH3 variants, as well as Dnah3 KO mice, achieved good ICSI outcomes.Therefore, our findings suggested that ICSI is beneficial for men with DNAH3-associated AT.
It has been reported that mutations in DNAH3 are associated with lung, colon, and breast cancer (Tanaka et al., 2008;Liu et al., 2012;Hamdi et al., 2018).For example, DNAH3 is a commonly mutated gene in lung tumours, including adenocarcinomas and squamous cell carcinomas (Liu et al., 2012).In addition, WES was performed on a Tunisian family with breast cancer, and the biological network construction and protein-protein interactions analyses showed that DNAH3 might be involved in breast cancer (Hamdi et al., 2018).In our study, the patients harbouring DNAH variants did not present with any cancers.However, we recommend that patients who carry variants in DNAH3 or other DNAH Bi-allelic DNAH3 variants cause male infertility with AT | 9 family genes should undergo regular check-ups and screening for early detection, and timely intervention if cancer is identified.
The main limitation of our study is that we did not establish a mouse model for each DNAH3 variant identified in our study.As all patients carried bi-allelic variants in DNAH3 (one allele carries a missense variant and the other allele carries a missense variant or nonsense variant), the production of different heterozygous knock-in lines was the most appropriate method to mimic the phenotype found in humans.However, because it requires a considerable time to produce six different mutant murine lines carrying M1 to M6, we only produced two lines of KO mice.Western blotting showed that the DNAH3 protein was significant decreased in the spermatozoa from patient (P1) and was undetectable in the two lines of knock-out mice.Although there was discrepancy regarding DNAH3 protein levels between patients and KO mice, the two lines of KO mice both displayed AT, and the phenotype was similar to that observed in the human patients.Therefore, combined with the characteristics of the six variants, we suggest that DNAH3 is a novel candidate gene for AT and that these DNAH3 biallelic variants are the cause of infertility in the three patients.
In conclusion, DNAH3 is a novel candidate causative gene for AT and is essential for sperm flagellar assembly and sperm motility in humans and mice.Our study provides a better understanding of the functions of DNAH family genes in male infertility and offers insights into the genetic aetiology and information for reproductive counselling of patients with AT.

Figure 1 .
Figure 1.Identification of bi-allelic DNAH3 variants in men with asthenoteratozoospermia. (A) Pedigree of three men affected by DNAH3 variants (M1-M6) from three unrelated families.The black-filled symbol represents the infertile men harbouring bi-allelic DNAH3 variants.Sanger sequencing chromatograms of the DNAH3 variants (M1-M6) in all infertile males are shown at the bottom.The red arrowhead represents positions of DNAH3 variants.(B) Locations and conservation analysis of DNAH3 variants (M1-M6) identified in the patients.The human DNAH3 gene encodes a 4116-amino acid protein that consists of an N-terminal tail, a stalk structure, and six AAAþ motor domains.(C) Haematoxylin-eosin (H&E) staining of the spermatozoa from the normal control (NC) and three patients with bi-allelic DNAH3 variants.Compared with the normal morphology of the NC sperm tail (a), the spermatozoa of patients showed multiple morphological abnormalities of the flagella, such as coiled tail (b, f, and j), absent tail (c, g, and k), angulated tail (d, h, and l), and short tail (e, i, and m).Scale bar, 5 lm.(D) Western blotting of DNAH3 expression in spermatozoa from normal control (NC) and the patient (P1).a-tubulin was used as the control.(E) Quantification of DNAH3 levels in spermatozoa from normal control (NC) and the patient (P1).��� P < 0.001.

Figure 2 .
Figure 2. DNAH3 variants are associated with aberrant assembly of axoneme and axonemal accessory structures.(A) Transmission electron microscope (TEM) analysis of the ultrastructure of spermatozoa from normal control (NC) and the patient (P1) harbouring bi-allelic DNAH3 variants.Cross-sections of the sperm flagella from NC showed a typical '9 þ 2' axoneme structure and axonemal accessory structure, including central pairs (CPs, bule arrow), outer dense fibres (ODFs, green arrow), mitochondrial sheath (MS, yellow arrow), and fibre sheath (FS, red asterisk).Cross-sections of sperm flagella from P1 revealed disordered axonemal ultrastructures, including the absence of CPs, and disorganization of MS and FS.Longitudinal sections of the sperm flagella from P1 showed poorly assembled MS and hyperplastic FS compared to the well-arranged MS and FS of sperm flagella from the NC.Scale bars, 200 nm.(B) TEM analysis on the axonemal dynein arms of the spermatozoa.The cross-sections of different pieces of sperm flagella from the NC showed intact inner dynein arms (IDAs; green triangle) and outer dynein arms (ODAs; blue triangle).However, the cross-sections of different pieces of sperm flagella from P1 showed partial absence of IDAs (indicated by yellow triangles).The structures in the red dotted square are magnified proportionally on the right side.Scale bars, 200 nm.

Figure 4 .
Figure 4. Dnah3-associated male infertility could be overcome by ICSI.(A) Representative two-cell embryos and blastocysts of Dnah3 þ/þ and Dnah3 ko1/ko1 male mice after intracytoplasmic sperm injection.Scale bar, 50 lm.(B) Rates of two-cell embryos and blastocysts in the Dnah3 ko1/ko1 group were comparable to those in the Dnah3 þ/þ group.ns indicates not significant.n ¼ 3 for biologically independent mice in the WT group, and n ¼ 3 for biologically independent mice in the Dnah3 knockout group.

Table 1 .
Semen characteristics and sperm morphology in the subjects harbouring DNAH3 bi-allelic variants.