A DNAH17 missense variant causes flagella destabilization and asthenozoospermia

Using mice modelling patients’ variant, this study demonstrates that a homozygous DNAH17 missense variant causes asthenozoospermia and specifically destabilizes microtubule doublets 4–7 in flagella, which could be largely due to the storage of sperm in epididymis.


Introduction
According to the World Health Organization (WHO), men whose ejaculates have <32% progressively motile sperm are diagnosed with asthenozoospermia (WHO, 2010). Asthenozoospermia is one of the major causes of male infertility. Isolated asthenozoospermia accounts for 19% of all infertile men, and oligo-and/ or terato-asthenozoospermia could account for 63% of all infertile men (Curi et al., 2003). Numerous factors, like lifestyle, pollutants, prolonged sexual abstinence, partial blockage of seminal tract, varicocele, and infection, have been reported as causes of asthenozoospermia (Adams et al., 2014;Ortega et al., 2011;Salas-Huetos et al., 2017). Nonetheless, the genetic factors underlying asthenozoospermia remain largely unknown.
Axoneme is the core structure of sperm flagellum, presenting throughout the flagellar length. The axoneme is typically composed of 9+2 microtubules, where a central pair of microtubules is surrounded by nine peripheral microtubule doublets (MTDs) in the fixed order (Inaba, 2011). Axonemal dyneins are a pair of projecting "hooks," consisting of an inner and an outer dynein arm (IDA and ODA, respectively), which are attached to each of the nine MTDs (Kikkawa, 2013). IDAs and ODAs are structural subunits of axoneme and essential for generating beating forces of sperm flagella (Gibbons, 1963;Summers and Gibbons, 1971). Each dynein arm is composed of several light chain proteins, at least two intermediate chain proteins, and at least two heavy chain proteins that hydrolyze ATPs for microtubule sliding (Inaba, 2011;Roberts et al., 2013).
In this study, we recruited three primary infertile patients from Pakistan, born to a consanguineous union and suffering from asthenozoospermia with no MMAF-like phenotype or ciliary-related symptoms. Through whole-exome sequencing (WES) and Sanger sequencing, we identified a homozygous missense variant in DNAH17 recessively cosegregating with asthenozoospermia in this family. Further analyses of spermatozoa from patients and functional studies in mice carrying a Dnah17 mutation equivalent to that in patients collectively demonstrated that the DNAH17 variant specifically induces doublets 4-7 destabilization during sperm storage in epididymides and thus causes asthenozoospermia, signifying that DNAH17 is the first DNAH protein implicated in stabilizing flagellar structure.

Results
Three asthenozoospermic patients born to a consanguineous union This study was performed on a family with male infertility originating from Pakistan ( Fig. 1 A). The parents (III:1 and III:2) were first-degree cousins and gave birth to three daughters and four sons. Two sisters (IV:5, 42 yr old and IV:6, 27 yr old) had three and two children, respectively, and the youngest sister (IV:7, 25 yr old), who had normal menstrual cycles, was unmarried. Among the four brothers, one (IV:4, 28 yr old) was unmarried; the other three, IV:1 (43 yr old), IV:2 (41 yr old), and IV:3 (29 yr old), had been married for 20, 17, and 11 yrs, respectively, but all were infertile. They did not have any history of drinking, smoking, exposure to toxic chemicals, or any symptoms of ciliary-related diseases and were physically normal with respect to height, weight, external genitalia, and testicular size. Semen analyses of patients revealed that semen volumes, sperm concentrations, and percentages of morphologically normal sperm fell within the normal ranges (WHO, 2010). However, all three patients exhibited reduced sperm motility, with ≤25.0% of motile sperm and ≤17.5% progressively motile sperm. Hence, they were diagnosed with asthenozoospermia. Patients' clinical characteristics are summarized in Table 1.

Identification of a candidate pathogenic variant in DNAH17
To understand the genetic cause of asthenozoospermia in this family, we performed WES for patient IV:2 and his father. Following a pipeline of WES data analysis (Fig. S1), 12 variants in 12 genes were retained. Subsequent Sanger sequencing on genomic DNA from all the available family members (III:1, III:2, and IV: 1-7) verified four variants in four genes (DNAH17, GPS1, HID1, and USP36) recessively coinherited with asthenozoospermia ( Fig. 1 B and Fig. S1 B). GPS1, HID1, and USP36 are annotated ubiquitously expressed in various tissues (Zhang et al., 2013; DNAH17 mRNA exhibits testis-specific expression [Milisav and Affara, 1998]). Given that our patients did not show any other symptoms except asthenozoospermia, the homozygous missense variant (c.G5408A) in DNAH17 was favorably presumed to be responsible for the diminished sperm motility in the patients.
The DNAH17 c.G5408A occurred in exon 35 and caused a G-to-A substitution at cDNA (NCBI reference sequence no. NM_173628) nucleotide position 5408, predicted to replace cysteine (C) by tyrosine (Y) at amino acid position 1803 (p.C1803Y; Fig. 1 C). The altered amino acid is located in the N-terminal stem region of DNAH17, which is known to interact with other dynein components. Phylogenic analysis revealed that the altered amino acid was conserved from lower to higher organisms ( Fig. 1 D). All these findings suggest that the homozygous DNAH17 variant (c.G5408A) could be pathogenic for asthenozoospermia in this family.

Generation and validation of the anti-DNAH17 antibody
To determine the expression and localization of DNAH17, we generated an antibody recognizing an epitope of DNAH17 amino acids 3502-3801, which is highly conserved between mouse and human (Fig. S2). To test the specificity of this antibody, immunoblotting and immunofluorescence (IF) staining assays were performed with HEK293T cells overexpressing FLAG-tagged epitope-corresponding peptides from mouse and human DNAH17, DNAH9, and DNAH11, which are the mammalian homologues of Chlamydomonas reinhardtii ODA β-HCs (Pazour et al., 2006) and have high amino acid sequence similarity. The anti-DNAH17 antibody showed high affinity to the epitopecorresponding peptides of both mouse and human DNAH17 as expected. However, it also recognized the overexpressed epitope-corresponding peptides of mouse and human DNAH9, as well as mouse DNAH11, in immunoblotting assays ( Fig. S3 A), and mouse and human DNAH9 and DNAH11 in IF staining (Fig.  S3 B). To determine whether this anti-DNAH17 antibody recognizes the endogenous DNAH9 and DNAH11 full-length proteins in humans and mice, we performed IF staining of respiratory cilia where DNAH9 and DNAH11 have been reported to be expressed (Dougherty et al., 2016;Fliegauf et al., 2005). This anti-DNAH17 antibody yielded weak signals that were not distinguishable from those of rabbit control IgG (Fig. 2, A and B), indicating that DNAH17 is not expressed in respiratory cilia, and that this anti-DNAH17 antibody is not likely to recognize endogenous DNAH9 and DNAH11 in humans and mice.
The specificity of the antibody was further validated in Dnah17 −/− mice that were generated using the CRISPR/Cas9 technique (Fig. S3 C). Immunoblotting using the antibody against DNAH17 detected a specific band at the predicted size of DNAH17 protein in sperm lysates from Dnah17 +/− mice, but not in sperm lysates of Dnah17 −/− mice (Fig. S3 D). Similarly, IF staining of spermatozoa from Dnah17 +/− mice showed clear and specific signals of DNAH17 colocalizing with α-tubulin, one of the major constituents of microtubules (Amos and Klug, 1974), along sperm tails except the distal tip, but the signals of anti-DNAH17 antibody were completely absent in sperm flagella from Dnah17 −/− mice (Fig. S3 E). Noticeably, knockout of Dnah17 drastically reduced sperm count and resulted in morphologically abnormal spermatozoa with a typical human MMAF phenotype (Fig. S3, F-I). The 9+2 axonemal configuration was completely disrupted in flagella of Dnah17 −/− mice (Fig. S3 J).
These in vitro and in vivo studies collectively indicated that, although the anti-DNAH17 antibody could cross-react with the epitope-corresponding peptides of DNAH9 and DNAH11 overexpressed in cultured cells, it is not likely to recognize endogenous DNAH9 and DNAH11 proteins in mice and humans.

The expression and localization of DNAH17
In mice, Dnah17 mRNA was abundantly expressed in testes but not detected in lungs, tracheae, or oviducts ( Fig. 2 C). Using the antibody against DNAH17, we found that DNAH17 protein was indeed not detected in cilia from tracheae and was detected only in testes and epididymides of WT mice (Fig. 2, A, D, and E), but not in epididymides of Rpl10l −/− mice devoid of spermatozoa (Fig. 2 E;Jiang et al., 2017), indicating that DNAH17 expression is restricted to testes and spermatozoa. IF staining further showed that DNAH17 is localized predominantly in cytoplasm and The DNAH17 mutation occurs in exon 35 and causes a G-to-A substitution at cDNA (NCBI reference sequence no. NM_173628) nucleotide position 5408, replacing cysteine (C) with tyrosine (Y) at amino acid 1803 in the DNAH17 protein (UniProt accession no. Q9UFH2). (D) Sequence alignment shows conservation of the affected amino acid (cysteine) across different organisms. Arrowheads, the mutation site; WT, the wild-type allele; MT, the mutant allele; UTR, untranslated region. flagella of step 11-16 spermatids of adult mice (Fig. 2 F). Consistently, in humans, DNAH17 was detected only in testes, but not in various somatic cell lines ( Fig. 2 G and Fig. S3 K). Immunohistochemical staining on human testicular sections with normal spermatogenesis revealed that DNAH17 could be detected in cytoplasm and flagella of elongated spermatids (Fig. 2 H). IF staining on human semen smears showed localization of DNAH17 in sperm flagella (Fig. 2 I). Together, the expression and localization patterns of DNAH17 propose its potential role in spermatozoa.

Localization of the mutant DNAH17 in patients
To explore whether the identified variant affected expression and localization of DNAH17 in patients, we performed IF staining on semen smears using the anti-DNAH17 antibody. The signals of DNAH17 and α-tubulin were detected in sperm flagella of all three patients and were not distinguishable from those in the fertile controls (Fig. 3).

Morphological and ultrastructural analyses of sperm flagella from patients
Since mutations in DNAH1, another member of the DNAH family, have been reported to be associated with the MMAF phenotype (Coutton et al., 2018;Ben Khelifa et al., 2014;Sha et al., 2017;Tang et al., 2017;Wang et al., 2017), to understand whether this DNAH17 variant could also induce an MMAF-like phenotype, we conducted further examinations of sperm morphology for all three patients. Semen samples from a fertile man with normal spermogram were used as the control. There was no significant difference in the percentage of abnormalities in sperm head, tail, or head and tail between patients and control (Fig. S4, A and B). Moreover, spermatozoa were classified into six subtypes according to flagellar morphology, including normal, absent, bent, short, coiled, and irregular-caliber flagella (Ben Khelifa et al., 2014). The flagella of >80% of spermatozoa in the patients were morphologically normal, and the frequency of each type of flagella in the patients did not significantly differ from that in the control (Fig. S4 C). Hence, all three patients did not present with the MMAF phenotype.
We next examined whether the axonemal structure of sperm tails, the horsepower apparatus that drives sperm "swimming," was impaired by the DNAH17 variant (Inaba, 2011;Roberts et al., 2013). Transmission EM (TEM) analyses of sperm flagella were performed. In all three patients, cross sections of midpiece displayed a typical 9+2 axonemal configuration (Fig. 4 A). However, all three patients exhibited a frequent absence of MTD(s) 4-7, accounting for ≥39.7% and ≥60.3% of all cross sections at principal piece and end piece, respectively, which were significantly higher than those in controls (2.2% at principal piece and 1.0% at end piece; Fig. 4,A and B;and Fig. S4 D). Interestingly, the associated outer dense fibers (ODFs) were also absent in the cross sections with MTD(s) 4-7 missing in our patients. Further examination of the ultrastructural anomalies in the three patients revealed that the simultaneous loss of MTDs 4-7 was most frequently observed (≥32.6% and ≥46.9% of all cross sections at principal piece and end piece, respectively; Fig. 4, A, C, and D). Besides missing MTD(s) 4-7, other abnormalities, such as disorganization of MTDs, excess microtubules, missing almost all the microtubules, etc., were also observed in patients at low frequencies, which were not significantly different from those in controls (Fig. 4 E). Intriguingly, the ODA was clearly observed attached to each of the nine MTDs at midpiece and to each of the remaining MTDs at principal piece and end piece in all three patients ( Fig. 4 A). Taken together, these findings indicate that the impaired sperm motility in patients was likely due to the loss of MTD(s) 4-7 at principal piece and end piece of sperm flagella.
Diminished sperm motility in Dnah17 M/M mice Given the 91% identity in amino acid sequence (Fig. S2) and the same expression and localization patterns of DNAH17 between   (Table 2). Further examination showed that body weight, testis weight and their ratio, testicular histology, sperm count, and sperm morphology (particularly the frequencies of morphological normal flagella and each type of abnormal flagella) in Dnah17 M/M or Dnah17 +/M mice were all comparable to those in Dnah17 +/+ mice (Table 2 and Fig. S5, D and E). The percentages of motile and progressively motile sperm showed no significant difference between Dnah17 +/M mice and controls, but they were dramatically decreased in Dnah17 M/M mice (Table 2). Hence, consistent with the findings in the patients (Table 1), Dnah17 M/M mice displayed markedly diminished sperm motility, proving that the DNAH17 variant is indeed pathogenic for asthenozoospermia.
Ultrastructure of sperm flagella from Dnah17 M/M mice To understand whether the ultrastructure of sperm flagella in mice was also altered by the DNAH17 variant, TEM analyses of flagella of spermatozoa isolated from cauda epididymides were conducted. Cross sections of midpiece from both Dnah17 +/+ and Dnah17 M/M mice exhibited classical 9+2 axonemal configuration. Nonetheless, 42.9% and 87.2% of cross sections of principal piece and end piece, respectively, showed MTD(s) 4-7 missing with a concomitant loss of the associated ODF(s) in Dnah17 M/M mice, in sharp contrast to 0.6% and 1.7%, respectively, in controls (Fig. 5,A and B). Consistent with the observations in patients, the most frequent anomaly in Dnah17 M/M mice was the simultaneous absence of MTDs 4-7, accounting for 21.1% and 54.9% of all cross sections at principal piece and end piece, respectively (Fig. 5, A, C, and D). Moreover, the absence of MTD 4 or 7 was detected at lower frequencies (Fig. 5, C-E). Though some abnormalities other than the lack of MTD(s) 4-7 were also observed in flagellar cross sections of Dnah17 M/M and control mice, their percentages were low and showed no significant difference (Fig. 5 F). It is also worth noting that the ODA attached to each of the remaining MTDs was visible on all the cross sections examined for Dnah17 M/M mice (Fig. 5 A). Together, these TEM findings in Dnah17 M/M mice recapitulated the defects of patients' sperm tails, demonstrating that the DNAH17 variant was indeed responsible for the absence of MTD(s) 4-7 in our asthenozoospermic patients.

MTDs 4-7 were destabilized in cauda epididymis
The frequent absence of MTD(s) 4-7 observed in spermatozoa from both patients and Dnah17 M/M mice could be due to either defective flagellum biogenesis or MTD destabilization. TEM analyses of the sperm flagella in seminiferous tubules retrieved from testes were performed and revealed that abnormal axoneme structure was not observed in all cross sections examined for Dnah17 M/M mice (Fig. 6, A and B), indicating that the cause of MTD(s) 4-7 absence was not a defect in axonemal assembly.
To investigate whether the lack of MTD(s) 4-7 occurred during transition of spermatozoa, we performed TEM analyses of spermatozoa within caput, corpus, and cauda epididymides and found that axonemal structure abnormality was hardly detected in all cross sections examined for Dnah17 M/M caput and corpus epididymides (Fig. 6, C-F). However, flagella in Dnah17 M/M cauda epididymides displayed significantly higher frequencies of abnormal axoneme structure at principal piece and end piece than in controls (Fig. 6, G and H). Interestingly, in Dnah17 M/M cauda epididymides, 33.8% and 85.3% of cross sections at principal piece and end piece, respectively, presented with MTD(s) 4-7 missing, while only 1.7% and 0.9%, respectively, were detected with such abnormalities in controls (Fig. 6 I). The  levels of α-tubulin, β-tubulin, and DNAH17 in lysates of sperm from cauda epididymides were further evaluated and found to be comparable between Dnah17 M/M and control mice (Fig. 6 J), indicating that these proteins were not degraded and the lack of MTD(s) 4-7 could be due to destabilization. Hence, these findings revealed that the structural defects of sperm flagella observed in patients and Dnah17 M/M mice could result from the destabilization of MTDs 4-7 occurring specifically in cauda epididymis.
Destabilization of MTDs 4-7 is related to the storage of sperm in epididymides Since cauda epididymis is the place where spermatozoa are stored before ejaculation, we thus examined whether the storage time could predispose MTDs 4-7 to destabilization. We conducted epididymal duct ligation at the end of corpus adjacent to cauda for 2 d and 4 d, respectively (Fig. 7 A), simulating the different lengths of time that spermatozoa were stored in epididymides. 2 or 4 d after ligation, a large number of spermatozoa were accumulated in the corpus region in both WT and Dnah17 M/M mice (Fig. S5 F). TEM analyses of flagella of sperm in the corpus were subsequently performed. Compared with WT mice, significantly increased frequencies of MTD(s) 4-7 missing at the principal piece (16.5%) and end piece (59.1%) were observed in Dnah17 M/M mice 2 d after ligation (Fig. 7, B and C). 4 d after ligation, the frequencies were further increased to 24.2% at principal piece and 82.7% at end piece in Dnah17 M/M mice, while the MTDs with typical arrangement, numbers in red indicate the missing MTDs, and arrowheads highlight the ODAs. Scale bars represent 200 nm. (B) Quantification of flagella with loss of any combination of MTDs 4-7 at midpiece, principal piece, and end piece from controls and three patients. (C and D) The percentages of cross sections with MTD(s) 4, 5, 6, 7, 4+5, 4+7, 5+6, 5+7, 6+7, 4+5+6, 4+5+7, 4+6+7, 5+6+7, or 4+5+6+7 missing at principal piece (C) and end piece (D). (E) Quantification of cross sections with abnormalities other than the MTD(s) 4-7 missing at midpiece, principal piece, and end piece of sperm flagella from controls and three patients. Two independent experiments were performed. n, the number of axonemal cross sections analyzed. Data are presented as mean ± SEM. ***P < 0.001; one-way ANOVA test.  the frequencies in control mice remained very low (Fig. 7, B and C). These observations indicate that MTDs 4-7 destabilization is negatively associated with the length of ligation time. Thus, combined with the findings on flagellar structure of sperm in cauda epididymides (Fig. 6, G-I), we proposed that the destabilization of MTDs 4-7 in Dnah17 M/M mice is most probably due to storage of sperm in epididymides, regardless of storage in corpus or cauda.

Discussion
Our study identifies a homozygous missense variant (c.G5408A) in DNAH17, a functionally uncharacterized gene, from a consanguineous Pakistani family with three offspring suffering from asthenozoospermia (no MMAF-like phenotype) and provides genetic evidence that DNAH17 c.G5408A is pathogenic for asthenozoospermia using Dnah17 M/M mice modeling the patients' mutation. Extensive examinations of the spermatozoa from the three patients and Dnah17 M/M mice collectively elucidate that the DNAH17 variant causes frequent absence of MTD(s) 4-7 at principal piece and end piece during the sperm storage in epididymides. Thus, we demonstrate for the first time that DNAH17 is essential for sperm motility, and is the only known DNAH protein implicated in stabilizing flagellar structure, specifically MTDs 4-7. Eukaryotic cilia and sperm flagella share a highly conserved 9+2 axonemal structure, constituted of microtubules, motor dynein arms, and their associated structures. Thus, it is not surprising that a multitude of mutations in genes encoding axonemal machinery has been identified in humans that are commonly linked to PCD and asthenozoospermia . All the family members in our study declared not having any ciliary-related symptoms and thus refused to participate in any further related examination. One sister (IV:5), who is homozygous for the DNAH17 variant, had three children and no miscarriages, suggesting that DNAH17 may be dispensable for ciliary functions. Furthermore, we found that both human and mouse DNAH17 is highly expressed in testes and spermatozoa, but was not detected in respiratory cilia, indicating that DNAH17 is required only for flagella but not cilia. Hence, we conclude that the DNAH17 variant causes isolated asthenozoospermia without any other PCD-related symptoms.
DNAH17, along with DNAH9 and DNAH11, is a homologue of C. reinhardtii ODA β-HCs in mammals (Pazour et al., 2006). Different from the testis/spermatozoa-restricted expression pattern for DNAH17, DNAH9 and DNAH11 are localized to the distal and proximal regions of respiratory ciliary axonemes, respectively (Dougherty et al., 2016;Fliegauf et al., 2005). Besides, DNAH9 was also found localized through the sperm flagella except the distal tip (Fliegauf et al., 2005), which is similar to DNAH17 localization in sperm flagella. Biallelic mutations in DNAH9 cause PCD with a frequent loss of ODAs at the distal, but not the proximal, regions of cilia (Fassad et al., 2018;Loges et al., 2018), and the patient carrying two homozygous missense mutations in DNAH9 presented with oligoasthenoteratozoospermia (Fassad et al., 2018), yet the underlying ultrastructural anomalies and pathogenic mechanism remain unknown. Biallelic mutations in DNAH11 in patients are known to cause PCD without obvious defects in the ciliary ultrastructure; however, the sperm motility, flagellar ultrastructure, and fertility status of these patients were not mentioned (Bartoloni et al., 2002;Knowles et al., 2012;Lucas et al., 2012;Pifferi et al., 2010;Schwabe et al., 2008), with the exception of one man, who had one child without the use of medical assistance (Schwabe et al., 2008). The mouse model carrying a mutation (E2271K) in Dnah11 displayed reduced fertility and sperm motility, though the ultrastructure of sperm tails appeared normal (Lucas et al., 2012). Hence, it remains uncertain whether DNAH11 is required for male fertility. Future studies are needed to further explore the roles of DNAH11 and DNAH9 in sperm tails.
Disruptions in flagellar axoneme proteins have been reported to be associated with MMAF in humans . Thus, it is not surprising that knockout of Dnah17 induced an MMAF phenotype, which likely resulted from defective flagellar biogenesis. Noticeably, Dnah17 M/M mice displayed a normal flagellar biogenesis but destabilized MTDs 4-7 during epididymal sperm storage, which is a unique phenotype that has not been reported so far in human or animal mutants for any other DNAHs (Ben Khelifa et al., 2014;Lucas et al., 2012;Neesen et al., 2001;Sha et al., 2017;Wang et al., 2017;Zuccarello et al., 2008). It is implied that the Dnah17 missense mutation is less deleterious than Dnah17 knockout, thus allowing the identification of the indispensable role of DNAH17 in stabilizing MTDs 4-7 during sperm storage in epididymis. We have also tried to detect whether there was a frequent loss of MTD(s) 4-7 in Dnah17 −/− mice, but the flagellar axoneme structures were so disorganized that we were unable to determine the order of MTDs in Dnah17 −/− mice. Taken together, the phenotypic difference between Dnah17 −/− mice and our patients or Dnah17 M/M mice indicates that DNAH17 is required for both flagellar biogenesis during spermiogenesis and stabilizing MTDs 4-7 during sperm storage in epididymis. It would be interesting to know whether Immunoblotting with lysates of spermatozoa from cauda epididymides using anti-DNAH17, anti-α-tubulin, and anti-β-tubulin antibodies. Lamin B1 was used as the loading control. Three independent experiments were performed. N, the number of mice analyzed. n, the number of axonemal cross sections analyzed. Data are presented as mean ± SEM. ***P < 0.001; Student's t test.
other axoneme proteins are also implicated in stabilizing axonemal structures and their mutation frequency in asthenozoospermic patients with ultrastructure defects but normal morphology of flagella, which have not been paid enough attention in the past.
After leaving the testis, sperm entered the caput epididymis, progressing to corpus, and finally reached the cauda, where they were stored until ejaculation. Here, we found that, when sperm were kept in corpus for 2 or 4 d, MTDs 4-7 were also destabilized in Dnah17 M/M mice, showing a time-dependent manner. Noticeably, at the end piece, the frequency of MTD(s) 4-7 missing in the corpus of Dnah17 M/M mice with epididymal duct ligated for 4 d was similar to that in the cauda of unligated Dnah17 M/M mice; however, at the principal piece, the frequency in Dnah17 M/M corpus after ligation for 4 d remained significantly less than that in the cauda of unligated Dnah17 M/M mice. These findings not only indicate that MTDs 4-7 at the end piece could be more susceptible to destabilization than those at principal piece, but also imply that the destabilization likely occurs from the distal region of the flagellum. There are two possible explanations for why MTD(s) 4-7 at principal piece were more prone to destabilization in the cauda of unligated mice than in the corpus after ligation for 4 d. First, the destabilization of doublets at principal piece may need a longer time than those at end piece. In unligated mice, sperm could be stored in the cauda epididymis for >4 d, as the densities of sperm in the corpus lumen 4 d after ligation (Fig. S5 F) were obviously lower than those in the cauda from unligated mice (Fig. S5 E). Second, the relatively lower pH and ionic calcium concentration and higher osmotic pressure in the lumen of cauda than in corpus (Dacheux and Dacheux, 2013;Shum et al., 2009;Turner, 2008) may also contribute to the instability of MTD(s) 4-7. Consistently, previous studies have shown that the disruption of Pla2g3 (Sato et al., 2010), Ttll9 (Konno et al., 2016), or Vdac3 (Sampson et al., 2001) also induced instability of MTDs 4-7 or MTD 7 in mouse cauda epididymides. Together, these pieces of evidence led us to believe that MTDs 4-7 are different from the other MTDs, and must be armed with delicate machinery to cope with the prolonged storage and possible environmental challenges in epididymis. Therefore, future efforts should be attempted to interrogate the compositional, structural, and functional differences between MTDs 4-7 and the other MTDs. Moreover, the ultrastructural localization of DNAH17 in flagellar axoneme and interacting proteins of DNAH17 need to be elucidated in the future with antibodies suitable for use in immunoelectron microscopy and coimmunoprecipitation, to decipher the specific and interesting role of DNAH17 in stabilizing MTDs 4-7.
In conclusion, we demonstrated that a homozygous DNAH17 missense variant specifically induces MTDs 4-7 destabilization in cauda epididymis, resulting in asthenozoospermia. It would be fascinating to determine the frequencies of DNAH17 mutations in larger cohorts of infertile patients or patients from different ethnic backgrounds, which will have significant implications for genetic counseling, diagnosis, and ultimate treatment of asthenozoospermia, as well as uncovering novel and attractive targets for male contraceptive development.

Materials and methods
The participants and semen analyses Members of a consanguineous family with three brothers suffering from infertility were recruited from a rural area of Pakistan. All three patients had routine semen analysis performed twice according to the WHO guidelines (WHO, 2010). Written informed consent forms from all participants were obtained at the beginning of the study. This study was approved by the institutional ethics committee of the University of Science and Technology of China.

WES and linkage analysis
Genomic DNA was extracted from all the family members available, and WES of patient IV:2 and his father was performed as we previously described (Yin et al., 2019). Genome Analysis Toolkit best practice was adopted to generate variants from raw reads. Parametric linkage analysis was performed using a VCF file as previously described (Smith et al., 2011), and five regions were identified with logarithm of the odds scores >0 (Fig. S1 A). Variants within linkage regions and following Mendelian inheritance were kept for further screening.
Generation of the polyclonal anti-DNAH17 antibody DNAH17 polyclonal antibody was generated in rabbits using amino acids 3502-3801 of mouse DNAH17 (UniProt accession no. Q69Z23) as antigens by ABclonal Biotechnology. Briefly, the 900-bp cDNA encoding the epitope was cloned into pET-28a expression vector, and the His-tagged fusion protein was expressed in Escherichia coli. The purified recombinant protein was used to generate polyclonal antisera in female New Zealand rabbits. Sequences of the primers used are as follows: for 8xHis-DNAH17_C300-6xHis, forward 59-ACCATCATCACCATGCCA AAGAGTACCACCCCAGTTTCCGCCTGA-39 and reverse 59-TGG TGGTGGTGGTGCTCGAGTTCTTTGGGGAAGATCTCTTTCTCG-39; and for vector backbone, forward 59-CTCGAGCACCACCAC CACCA-39 and reverse 59-TTTGGCATGGTGATGATGGTG-39.

IF staining
Human respiratory epithelial cells were obtained from a healthy subject by transnasal brush biopsy using disposable cytology brushes (Olympus, BC-202D-3010) and then spread onto glass slides. Mouse respiratory epithelial cells were obtained from tracheal tissues. The tracheae were cut into pieces gently, followed by centrifugation. After discarding the supernatant, the cells were resuspended in PBS containing 50% FBS and then spread onto glass slides. Human semen smears were prepared following the guideline of WHO (WHO, 2010). Mouse sperm were obtained from cauda epididymides, washed in PBS twice, and spread onto glass slides. The slides were air dried, fixed with 4% paraformaldehyde, and stored at −80°C until use.
Immunohistochemistry and histological analyses of testicular and/or epididymal tissues Fresh testicular and epididymal tissues were fixed in Bouin's solution or in 4% paraformaldehyde at 4°C overnight, followed by paraffin embedding. Paraffin-embedded tissues were sectioned (5 µm). Immunohistochemistry and IF staining of testicular sections were performed as previously described (Jiang et al., 2015;Jiang et al., 2014). The antibodies used were normal rabbit IgG (CST, 2729S; 1:100), Alexa Fluor 555 donkey anti-rabbit IgG (Molecular Probes, A31572; 1:200), and the anti-DNAH17 antibody that was custom produced by ABclonal Biotechnology. H&E and periodic acid-Schiff staining were performed for histological analyses of epididymal and testicular sections, respectively. Images were captured using a microscope (Nikon Eclipse 80i) equipped with a digital camera (Nikon DS-Ri1).
TEM analysis TEM was performed as previously described, with minor modifications (Yuan et al., 2015). Briefly, spermatozoa or tissues were fixed in 0.1 M phosphate buffer (PB; pH 7.4) containing 4% paraformaldehyde, 8% glutaraldehyde, and 0.2% picric acid at 4°C for at least overnight. After four washes with 0.1 M PB, samples were post-fixed with 1% OsO 4 and dehydrated, followed by infiltration of acetone and epon resin mixture. Samples were embedded and ultrathin (70 nm) sectioned before staining with uranyl acetate and lead citrate. The ultrastructure of the samples was examined and captured by Tecnai 10 or 12 Microscope (Philips) at 100 kV or 120 kV, or by H-7650 Microscope (Hitachi) at 100 kV.

Fertility test
A fertility test was performed by mating one 10-wk-old Dnah17 M/M male with two 10-wk-old WT female mice (C57BL/6J) for 90 d. A total of five Dnah17 M/M , three Dnah17 +/M , and three WT male mice were tested. All the females were monitored for pregnancy. Dates of birth and numbers of pups were recorded for all the litters.
Analyses of mouse sperm count, morphology, and motility 10-wk-old mice were sacrificed by cervical dislocation. Sperm number and motility were analyzed as previously described (Castaneda et al., 2017;Jiang et al., 2017). For sperm morphology, slides were stained by Papanicolaou staining (Solarbio, G1612) according to the manufacturer's protocol. The percentages of morphologically normal spermatozoa were quantified according to the WHO guidelines (WHO, 2010), with at least 500 spermatozoa examined for each mouse.
Epididymal duct ligation Adult mice were anaesthetized by intraperitoneal administration of tribromoethanol. The epididymides of both sides were exposed through a median incision at the lower abdomen and ligated with a surgical suture at the end of the corpus adjacent to cauda. The mice were euthanized 2 or 4 d after the operation. The epididymides were removed, with one side fixed in Bouin's for histological analyses by H&E staining and the other side fixed in 0.1 M PB containing 4% paraformaldehyde, 8% glutaraldehyde, and 0.2% picric acid for TEM analyses.
Online supplemental material Fig. S1 presents WES data analysis. In Fig. S2, the alignment of human and mouse DNAH17 protein sequences show that 91% of amino acids are identical. Fig. S3 presents validation of the anti-DNAH17 antibody in transfected cells and Dnah17 knockout mice. Fig. S4 shows morphological and axoneme ultrastructural analyses of spermatozoa from patients. Fig. S5 shows generation of Dnah17 M/M mice modeling patients' mutation and histological examinations of testes and epididymides from Dnah17 +/+ , Dnah17 +/M , and Dnah17 M/M mice. constructive comments on the manuscript; L. Yuan and Y. Lu provided technique assistance for TEM analysis; X. Xu and Y. Wang provided assistance for experiments with human respiratory epithelial cells; H. Zhang, J. Gao, and J. Zhou performed the WES sequencing and WES data analysis. Q. Shi, H. Ma, and Y. Zhang conceived and supervised the study, designed and analyzed experiments, and wrote the manuscript.
Submitted: 20 December 2018 Revised: 10 June 2019 Accepted: 3 October 2019 Figure S2. Alignment of human and mouse DNAH17 protein sequences showing that 91% of amino acids are identical. Residues that are identical between human and mouse DNAH17 appear in red and as uppercase letters in the consensus line. Residues highly similar between human and mouse DNAH17 are indicated by red symbols (!, any one of I and V; $, any one of L and M; %, any one of F and Y; #, any one of N, D, Q, E, B, and Z). Unconserved residues are written in blue or as asterisks in the consensus line. Blue lines highlight the epitope (amino acids 3502-3801 for mouse DNAH17, corresponding to amino acids 3518-3817 in human DNAH17) for antibody generation. The alignment was performed using the online software MultAlin (http://multalin.toulouse.inra.fr/ multalin/multalin.html). Figure S4. Morphological and axoneme ultrastructural analyses of spermatozoa from patients. (A) Representative images of spermatozoa from the fertile controls and patients after Papanicolaou staining. Scale bars represent 10 µm. (B) Quantification of spermatozoa with normal morphology, or morphological defects in sperm head, tail, or head and tail. (C) Frequencies of sperm flagella that are morphologically normal, absent, short, coiled, bent, or of irregular caliber. Each spermatozoon was classified as only one type of flagellar morphology according to its major abnormality. For A and B, two independent experiments were performed with at least 500 spermatozoa examined per person each time. Data are presented as mean ± SEM. Compared with the fertile control, one-way ANOVA with Dunnett's multiple comparison test. Two independent experiments were performed. (D) Representative TEM micrographs showing cross sections of sperm flagella at low magnification from fertile men (controls) and the three patients. Red arrowheads, cross sections with MTD(s) 4-7 missing; blue arrowheads, cross sections with abnormalities other than the loss of MTD(s) 4-7. Scale bars represent 500 nm.