Two consanguineous families from Pakistan, comprising six male patients, were enrolled to investigate the underlying genetic cause of male infertility. Detailed pedigrees were constructed according to the provided information by the parents of the patients. An initial physical and andrological examination displayed that all patients have normal body mass index and no primary ciliary dyskinesia (PCD) symptoms. Other associated diseases, such as hypogonadotropic hypogonadism, cryptorchidism, varicocele, seminal ductal obstruction, testicular trauma, and andrological tumor, were also not observed. Two repeated semen analyses were performed according to the WHO manual of 2010, and all the parameters were recorded ( Table 1 ). We also recruited the fertile siblings of patients and their parents to serve as a positive control for genetic analysis. The study was approved by a bioethical committee of Pakistan and the ethical review board of the University of Science and Technology of China. Informed consent was obtained from all participants of both families.

Genomic DNA was extracted from all available members of both families by using QIAamp Blood DNA Mini Kit (QIAGEN) as per the protocol of the manufacturer. WES was conducted on III:2, IV:1, III:3, V:1, V:2, V:3, and V:5 from family members of PK-INF-15 and PK-INF-319, respectively, previously reported in (28). Briefly, WES data from the selected members were enriched by the Agilent SureSelect XT Human All Exon Kit. The Illumina HiSeq XTEN platform accompanied next-generation sequencing. The obtained raw reads were aligned to the human genome reference assembly (GRCh37/hg19) using the Burrows-Wheeler Aligner (29). Next, the Picard software was engaged to remove polymerase chain reaction (PCR) duplicates and evaluate the quality of variants. DNA sequence variants were analyzed by the best practice genome analysis kit (30). VCF files were used for parametric linkage analysis, as described previously (31), and four linkage regions were identified with a logarithm of odds (LOD) score of more than 0.5. Further screening was opted for the variants that were located on linkage regions and following Mendelian inheritance pattern.

Only variants that have depth >×20, genotype quality >90, and 0.5-cM intervals between each other were designated as markers. MERLIN software was employed on genotyped single-nucleotide polymorphism for linkage analysis with these parameters: due to consanguinity in the families, recessive mode of inheritance was adopted, with a disease allele frequency of 0.001 and 100% penetrance. A number of five peaks with a LOD score of more than 0.5 were considered as linkage regions. The variants that resided in the linkage regions were further annotated by ANNOVAR using the NCBI RefSeq gene annotation. We adopted the following filtration strategy to select potential candidate disease-causing variants in patients: (i) variants that were heterozygous in the father and control brothers and homozygous in patients were retained; (ii) variants with minor allele frequency of more than 0.01 in public databases such as 1000 Genome project (32), ESP6500 (33), or ExAC database (34), and variants homozygous in our in-house WES variants call set generated from 578 fertile male samples (41 Pakistanis, 254 Chinese, and 283 Europeans) were excluded; (iii) variants that affect protein sequence and predicted to be deleterious by around 10 of 13 in silico employed tools were kept; and (iv) variants which have a predominant expression in the testes and their inactivation affects male fertility were retained based on spermatogenesis online database (35) and were checked by Sanger sequencing to verify Mendel inheritance pattern. All these strategies are more clearly explained in a flow chart diagram in Supplementary Figure S2 . Finally, only DNAH1 variants followed the inheritance pattern and thus were considered the potential candidates causing MMAF phenotype in subjects under investigation. The list of primers used for Sanger sequencing is shown in Supplementary Table S1 .

Immunofluorescence on the spermatozoa of patients was performed, as we have previously reported (36). Briefly, sperm from patients were smeared on the clean slide and fixed with 4% paraformaldehyde followed by three times of washing in phosphate-buffered saline (PBS). The slides were permeabilized with 0.5% Triton X-100 for 30 min and blocked with 1% BSA. Next, the slides were incubated with primary antibodies, including α-tubulin (Sigma, F2168), DNAH1 (Abcam, ab122367), and SPAG6 (Proteintech, 12462-1-AP), overnight at 4°C. On the next day, the slides were washed with PBST (PBS containing Triton X-100) and incubated with secondary antibodies DAR555 (Molecular Probes, A31572) and GAM488 (Molecular Probes, A21121) for 1 h at 37°C. Finally, the slides were washed again and sealed with Hoechst and Vectashield. Images were captured by using a laser scanning confocal microscope (Olympus).

Sperm samples from patients were processed in 0.1 M phosphate buffer (pH 7.4) containing 4% paraformaldehyde, 8% glutaraldehyde, and 0.2% picric acid at 4°C. Scanning and transmission electron microscopy (SEM and TEM) were performed as previously described, with minor modifications (36). Briefly, spermatozoa were washed in 0.1 M PBS four times, followed by fixation in 1% OsO₄ and dehydration. Then, samples were infiltrated for acetone and epon resin mixture and embedded in sputter-coated (SCD 500, Bal-Tel). Thinly sectioned (70 mm) samples were stained with uranyl acetate and lead citrate. The morphology of spermatozoa was examined by Hitachi S-4800 field emission scanning electron microscope under an accelerating voltage of 15 kV. The ultrastructure of the samples was analyzed and captured by Tecnai 10 or 12Microscope (Philips) at 100 or 120 kV or by H-7650 microscope (Hitachi) at 100 kV.

Dnah1^-/- and Dnah1^{△iso1/△iso1} mice were generated by using CRISPR/Cas9 genome editing technology as previously reported (37). Briefly, single guide RNAs (sgRNAs) designed to target exon 2 and 49 of Dnah1 to generate KO and mutant mice, respectively, were transcribed in vitro (Addgene, 5132). Single-strand oligodeoxynucleotides (ssODNs), with a mutation similar to family A2 patients and a synonymous mutation at the protospacer adjacent motif, were created by the Sangon Biotech system. The designed sgRNAs and ssODNs were microinjected with Cas9 mRNAs into the zygotes of B6D2F1 (C57BL/6×DBA/2J) mice, following a previously published methodology (38). The genotyping of newborn pups was determined by Sanger sequencing, and heterozygous Dnah1+/- and Dnah1+/△iso1 mice were bred to obtain the desired homozygous results of Dnah1^-/- and Dnah1^{△iso1/△iso1} mice for future experiments. The sequence of guide RNAs and all the primers used for genotyping is shown in Supplementary Table S1 .

Testes from 8-week-old Dnah1^-/- and Dnah1^{△iso1/△iso1} mice were detached and processed in radio-immunoprecipitation assay buffer (25 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, and 0.1% sodium dodecyl sulfate) complemented with protease inhibitors cocktail using a Tissuelyzer and then cleared by centrifugation. Western blotting was performed as previously reported (39). Nitrocellulose membranes were incubated with primary antibodies against DNAH1 (this study) and GAPDH overnight at 4°C. On the next day, membranes were washed in PBST and incubated with secondary antibodies for 1 h at room temperature. The blots were developed with chemiluminescence (Image Quant LAS 4000, GE Healthcare).

According to the protocol of the manufacturer, total RNA from mice and control human testes was extracted by using Trizol (Invitrogen) kit. RT-PCR was conducted as we have previously described (40). The primer used for RT-PCR is listed in Supplementary Table S1 .

The fertility status of Dnah1^{△iso1/△iso1} mice was assessed by caging each mutant mouse with two wild-type females (C57/BL) for 3 months. A total of five Dnah1^{△iso1/△iso1} mice were tested, and none of them produce a single pup ( Table 2 ). Testes morphology, sperm count, H&E, sperm morphology, and sperm motility were performed as previously described (36, 39).

In the present study, we recruited two infertile consanguineous Pakistani families (PK-INF-15 and PK-INF-319) consisting of six patients that were suffering from primary infertility ( Figure 1A ). The semen analysis of four patients showed a sufficient sperm concentration, while all sperms were immotile ( Table 1 ). SEM and light microscopy investigation revealed abnormalities in sperm tail, displaying different defects of sperm flagella, including absent, short, coiled, bent, and irregular caliber, suggesting MMAF ( Supplementary Figures S1A, B ). Similarly, the SEM of sperm from patients also confirmed the MMAF phenotype ( Figure 1B ). Altogether these results indicated that MMAF accompanies the underlying cause of infertility in these families.

In order to explore the underlying genetic cause of infertility, WES was performed on four patients, the control brother and their parents (III:2, IV:1, III:3, V:1, V:2, V:3, and V:4). Considering the consanguinity within the families, we screened the WES data and only retained the variants that displayed recessive inherence patterns. Furthermore, we filter the variants based on minor allele frequency, testis-specific expression, and review of literature about the genes that have an association with the MMAF phenotype. In short, we identified two novel mutations in DNAH1: one was the insertion of C (c.7646_7647InsC, p.N2549Qfs*61), causing a truncation of protein in the family A1, and the other was a missense mutation (c.6212T>G, p.C1789Y) in family A2 ( Figure 1D ). Sanger sequencing on DNA from all available members of both families indicated that the mutations are recessively segregating in all patients with infertility phenotype ( Figures 1Ca, b ). To check the effect of these mutations on protein function, we performed immunofluorescence (IF) of DNAH1 and alpha-tubulin on sperm smear slides from patients IV:1, V:1, and V:2. The IF result showed the absence of DNAH1 signals on the sperm tail of patients, while positive signals of DNAH1 were observed on the control sample ( Figure 1E ). Thus, these results indicated that the identified mutations are loss-of-function mutations and caused transcript decay by non-sense mRNA-mediated decay.

In order to investigate the axonemal structures, we performed TEM on patient (IV:1) spermatozoa. We statistically analyzed good-quality cross-sections of microtubule doublets from the patient and control spermatozoa samples and found that the central singlet of microtubules was missing in 66% of these sections ( Figures 2A, B ). Furthermore, the fibrous sheath was disorganized in 10% of the sections, and missing microtubule doublets were observed in 6% of analyzed sections. To further confirm the disruption of the central pair, we stained the sperm smear slides of the patients with SPAG6 antibody, which is a well-known marker generally used to check the integration of microtubules of the central pair. No signals of SPAG6 were observed on the sperm flagella of the patients, while sperm from the control individual displayed well-decorated signals on the entire tail of the flagella ( Figure 2C ). Overall, these results intimidated that the identified mutations have a deleterious effect on the axonemal structures of sperm flagella.

In order to study the in vivo effects of mutations, we generated Dnah1^-/- mice by deletion of the Dnah1 transcript and Dnah1^{△iso/△iso1} mutant mice similar to a mutation in family A2 (c.7646_7647InsC) by using CRISPR/Cas9 genome editing technology ( Figure 3A ). Sanger sequencing of the PCR products confirmed the genotype of Dnah1^-/- and Dnah1^{△iso/△iso1} mice ( Supplementary Figure S3A ). Furthermore, western blotting confirmed the absence of full-length DNAH1 protein in both mice models, while intact full-length DNAH1 (herein referred to as DNAH1 isoform1) protein was observed in DNAH1^+/△iso1 and DNAH1^+/- mice testes ( Figure 3B ). Interestingly, we observed the presence of a short isoform of DNAH1 (referred to as isoform2) in the testes of Dnah1^{△iso1/△iso1} mice, while no such isoform was found in Dnah1^-/- mice ( Figure 3C ). Thus, it can be deduced that the Dnah1^{△iso1/△iso1} mice still harbor the short-length DNAH1 protein. Similarly, reverse transcriptase PCR with a specific primer corresponding to isoform2 of DNAH1 also verified the presence of a small-length transcript in Dnah1^{△iso1/△iso1} and wild-type (WT) mice ( Supplementary Figures S2A, B ). Moreover, we performed RT-PCR on control human spermatozoa to check the presence of isoform2. However, no isoform2 was detected in control spermatozoa, while a clear band corresponding to isoform1 was observed ( Supplementary Figure S2C ).

After confirming the deficiency of full-length DNAH1 protein in Dnah1^{△iso1/△iso1} mice, we first scrutinized the fertility of mutant mice by caging one mutant mouse with two WT females for 3 months. During this duration, the female mice display vaginal plugs, demonstrating normal mating efficiency. However, no pregnancy was conceived, indicating that the Dnah1^{△iso1/△iso1} mice were completely infertile ( Table 2 ). Subsequently, testes morphology ( Supplementary Figure S3A ) and testes-to-body-weight ratio were comparable in mutant and WT mice ( Table 2 ). On the other hand, reduced sperm count was evident in Dnah1^-/- and Dnah1^{△iso1/△iso1} mice ( Table 2 ). Further examination of testis and epididymis section from Dnah1^{△iso1/△iso1} mouse displayed an intact seminiferous tubule structure and the presence of all types of germ cells ranging from spermatogonia to spermatozoa ( Supplementary Figure S2C ). Similarly, an ample number of mature spermatozoa were present in the lumen of cauda in Dnah1^{△iso1/△iso1} mouse, indicating that sperm production is not severely affected on the grass root level ( Supplementary Figure S3D ). Next, we analyzed sperm motility by CASA system and observed that progressive motility is severely compromised in Dnah1^{△iso1/△iso1} mouse, while no progressive motile sperm was observed in Dnah1^-/- mouse ( Table 2 ). To explore the reason of reduced motility, we performed a comprehensive inspection of sperm morphology of Dnah1^{△iso1/△iso1} mouse and Dnah1^-/- mouse. Interestingly, we found that most of the spermatozoa (53/49%) lack full-length flagella structure, only possessing an intact head structure ( Figure 3D ), while few other abnormalities were observed ( Table 2 ). Interestingly, there are still some (34/45%) sperm with normal flagella in Dnah1^{△iso1/△iso,} and Dnah1^-/- mouse, and the flagella without sperm head also have a normal morphology. On the other hand, ultrastructural investigation of sperm flagella from Dnah1^{△iso1/△iso1} and Dnah1^-/- mice displayed no significant variation in the architecture of axoneme ( Figures 3E, F ). However, statistical analysis indicated more abnormal sections in Dnah1^-/- mouse than Dnah1^{△iso1/△iso1} mouse, indicating that isoform2 may have a role in mediating the assembly and stabilization of axonemal structures. Altogether our study provided clues about the phenotype differences in Dnah1 KO and DNAH1 mutant human spermatozoa ultrastructure for the first time.