Cfap97d1 is important for flagellar axoneme maintenance and male mouse fertility

The flagellum is essential for sperm motility and fertilization in vivo. The axoneme is the main component of the flagella, extending through its entire length. An axoneme is comprised of two central microtubules surrounded by nine doublets, the nexin-dynein regulatory complex, radial spokes, and dynein arms. Failure to properly assemble components of the axoneme in a sperm flagellum, leads to fertility alterations. To understand this process in detail, we have defined the function of an uncharacterized gene, Cfap97 domain containing 1 (Cfap97d1). This gene is evolutionarily conserved in mammals and multiple other species, including Chlamydomonas. We have used two independently generated Cfap97d1 knockout mouse models to study the gene function in vivo. Cfap97d1 is exclusively expressed in testes starting from post-natal day 20 and continuing throughout adulthood. Deletion of the Cfap97d1 gene in both mouse models leads to sperm motility defects (asthenozoospermia) and male subfertility. In vitro fertilization (IVF) of cumulus-intact oocytes with Cfap97d1 deficient sperm yielded few embryos whereas IVF with zona pellucida-free oocytes resulted in embryo numbers comparable to that of the control. Knockout spermatozoa showed abnormal motility characterized by frequent stalling in the anti-hook position. Uniquely, Cfap97d1 loss caused a phenotype associated with axonemal doublet heterogeneity linked with frequent loss of the fourth doublet in the sperm stored in the epididymis. This study demonstrates that Cfap97d1 is required for sperm flagellum ultra-structure maintenance, thereby playing a critical role in sperm function and male fertility in mice.


Introduction
Mammalian sperm, like most other vertebrate sperm, carry the haploid genome in the head and use the flagellum for motility [1]. The flagellum consists of the midpiece, the principal piece, and the end piece, with the axoneme extending through all three parts. In addition, the midpiece contains mitochondria, along with the outer dense fibers (ODFs) and the fibrous sheath extending to the principal piece, while the endpiece is devoid of peri-axonemal structures [1,2]. The axoneme is composed of two central singlet microtubules cylindrically surrounded by nine doublet microtubules, which is referred to as the 9 + 2 structure. The central pair of singlet microtubules are called C1 and C2. They are connected by periodic bridges and surrounded by a fibrous structure-the inner sheath, also referred to as the central pair projection (CPP) [3]. Each outer microtubule doublet (OMtD) consists of A and B tubules. The complete A tubule is fused with the incomplete B microtubule. OMtDs are associated with inner dynein arms (IDA), outer dynein arms (ODA), and the nexin-dynein regulatory complex (N-DRC) [4]. Radial spokes (RS) extend from each A tubule of the outer doublets towards the central singlets.
Sperm motility is generated by controlled sliding of OMtDs. The inner-and outer-arms of dyneins are identified as the main motor proteins, promoting sliding of microtubules along each other and resulting in flagellar bending. Bending force is generated by dyneins bound to A-tubules that are sliding on the associated B-tubules along the entire axoneme and in that way generate bending force [5]. Dynein activity is regulated by the radial spokes and central pair of microtubules [4,6]. The (a-) symmetry of the flagellar beat controls the swimming path of the sperm cell: a symmetrical flagellar beat leads to a straight swimming path of the sperm cell, whereas asymmetries in the beat pattern lead to a curved or even spiral swimming path [7,8]. Morphological or functional flagellar defects impair sperm motility (asthenozoospermia) and fertility [9,10]. Mutations in several genes have been associated with asthenozoospermia (Reviewed in [11]) including Tekt4 [12], Tecte1 [13] and others. However, the mechanism of flagellar beat regulation is not well understood.
Although various studies show that 1000-2000 genes are expressed abundantly in testis [14][15][16], only a fraction of these genes have been well characterized. This knowledge gap inspired our in silico database screens for testis-enriched genes. Wherefrom, the uncharacterized Cfap97 domain containing 1 (Cfap97d1) gene emerged as a candidate. It belongs to the cilia and flagellum associated 97 (Cfap97) gene family, which contains three members: Cfap97, Cfap97d1, and Cfap97d2. Human orthologues exist for all three members. Cfap97 and Cfap97d1 contain one coiled-coil domain that is absent in Cfap97d2. The entire protein family supported by grants from the Deutsche Forschungsgemeinschaft (DFG): SPP1926: WA3382/2-1 (to DW), SPP1726: grant WA3382/3-1 (to DW), FOR2743 (to DW), SFB/TRR83 (to DW), and under Germany's Excellence Strategy -EXC2151 -390873048 (to DW). Boehringer Ingelheim Fonds (J.N.H.). This work was also supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT)/the Japan Society for the Promotion of Science (JSPS) [KAKENHI grants JP19J21619 (to SO), JP18K14715 and 20K15804 (to JC), JP17H04987 (to HM)]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
is poorly characterized. Human and mouse CFAP97 is an ortholog to Chlamydomonas reinhardtii FAP97, which has been identified in protein extracts of demembranated axonemes [17]. A study using the proximity-dependent biotinylation assay, has identified CFAP97 in a complex with human centrosome-cilium interface proteins but did not provide further characterization [18]. Hemingway (Hmw) is suggested to be the Drosophila orthologue of Cfap97 and Cfap97d1, and was shown to be required for motile cilia function in sperm flagellum and auditory sensory neurons [19]. Out of these three genes, Cfap97d1 was exclusively expressed in testes and highly conserved among mammals, indicating its potential involvement in male fertility.
Based on these findings, we have chosen Cfap97d1 as a candidate gene having a putative function in male fertility. To analyze gene function, we have used two Cfap97d1 knockout mouse models, which were independently generated in two laboratories to study the function of Cfap97d1 in vivo. The results demonstrated that loss of the Cfap97d1 gene in mice leads to sperm motility alterations (asthenozoospermia) associated with axoneme structural instability, and cause male fertility defects.

Cfap97d1 is a testis-enriched gene
Phylogenetic analysis showed a relationship between three members of the Cfap97 gene family: Cfap97, Cfap97d1, and Cfap97d2 (S1A Fig). Amino acid sequence alignment demonstrated that CFAP97D1 was highly conserved among mammals (S1B Fig). To determine the actual expression profile of Cfap97, Cfap97d1, and Cfap97d2, we performed multi-tissue RT-PCR from adult mice. The results showed that Cfap97d1 cDNA was detected only in mouse testes ( Fig 1A, also see S1C Fig. for more tissues) whereas Cfap97 and Cfap97d2 were expressed in several mouse tissues (Fig 1A). Similarly, in human multi-tissue RT-PCR, CFAP97D1 was only expressed in testes ( Fig 1B). Then, we performed RT-PCR using postnatal testes of various ages to identify the time point when Cfap97d1 gene expression starts during spermatogenesis. Data show that Cfap97d1 expression begins around postnatal day (PND) 20, which corresponds to the late diplotene diakinesis stage and round spermatid occurrence. It is then continuously expressed from PND 25 onwards (Fig 1C).
The mouse Cfap97d1 gene is located on chromosome 11, whereas the human orthologue CFAP97D1 is located on chromosome 17p21. Both contain six exons, five of them are coding and one is non-coding. The mouse CFAP97D1 protein is composed of 164 amino-acid residues (S1B Fig). Based on secondary structure predictions (i.e http://www.compbio.dundee.ac. uk/) CFAP97D1 is likely to contain 4 helices (S1B Fig., grey boxes) one of which is forming a coiled-coil region (S1B Fig., blue dash box).

Cfap97d1 knockout male mice have severe fertility defects
To examine the function of Cfap97d1 in vivo, we obtained and analyzed two knockout mouse models: Cfap97d1 (Cfap97d1 tm1a(KOMP)Wtsi , referred to as Cfap97d1 KOMP ) mice from the Wellcome Trust Sanger Institute Knockout Mouse Project and Cfap97d1 knockout mice generated using CRISPR/Cas9 (referred to as Cfap97d1 em1 ). The cassette used to generate Cfap97d1 KOMP mice has FRT site followed by a LacZ reporter inserted between exons 1 and 2 ( Fig 1D). Cfap97d1 expression was hindered by splicing to the LacZ trapping element. The deletion was verified by genotyping ( Fig 1E) and RT-PCR ( Fig 1F) using specific primers (S1 Table). The Cfap97d1 em1 knockout mice were generated using CRISPR/Cas9. To ensure gene disruption and to avoid an effect on the expression of the Dusp3 gene, which is located in the vicinity of Cfap97d1, we designed two crRNAs targeting exon 3 and intron 5 of Cfap97d1 (S2A Fig). We microinjected or electroporated two crRNA/tracrRNA/Cas9 ribonucleoproteins (RNPs) into zygotes [20] and obtained 20% and 33% mutants respectively (S2B We did not observe gross defects in development, behavior, and survival rate in homozygous mutant mice of either strain. Next, we assessed the fertility of Cfap97d1 KOMP-/and Cfap97d1 em1/em1 knockout mice. As heterozygous males produced normal number of pups ( Fig 1G and S2E Fig), we used littermate heterozygous males as controls throughout the study. The control Cfap97d1 KOMPwt/males sired 7.0 ± 3.0 (SD, n = 9) pups per litter, whereas knockout males sired on average 0.6 ± 1.3 (SD, n = 10, Fig 1G). Half (50%) of Cfap97d1 KOMP-/males did not sire pups at all. Similarly, the control Cfap97d1 wt/em1 males sired 8.2 ± 2.8 (SD, n = 3) pups per litter, whereas Cfap97d1 em1/em1 knockout males sired 0.9 ± 1.9 pups per litter (SD, n = 3, (S2E Fig)). On the other hand, we did not observe changes in female fertility. Thus, Cfap97d1 is required for normal fertility in male mice.
Given that Cfap97d1 knockout males occasionally sired pups, we analyzed whether this could be attributed to changed expression of other Cfap97 family genes, namely Cfap97 and Cfap97d2. We performed qRT-PCR using cDNA prepared from Cfap97d1 KOMPwt/and Cfap97d1 KOMP-/testes total RNA. Cfap97d1 deletion did not affect the expression levels of Cfap97 and Cfap97d2 (Fig 1H) mRNA. However, this does not rule out the possibility of a change at the protein level, which could not be assessed due to lack of functional antibodies for the proteins of the Cfap97 family. Furthermore, meiotic progression and acrosome formation did not reveal significant differences between the Cfap97d1 KOMP-/and control mice as shown by synaptonemal complex protein 3 (SYCP3), expressed during meiotic prophase, and WGA lectin, labeling acrosomes ( Fig  2G and 2H). The β-galactosidase staining, that indicates Cfap97d1 expression, was detected in elongating spermatids in Cfap97d1 KOMPwt/testes (Fig 2I and 2J).

Cfap97d1 knockout sperm show an impaired ability to penetrate the zona pellucida
To examine sperm function, we performed an in vitro fertilization (IVF) assay. Both, Cfap97d1 KOMP-/- (Fig 2K) and Cfap97d1 em1/em1 (S4G Fig) derived sperm displayed a reduced capability to fertilize cumulus-intact oocytes [Cfap97d1 KOMP line: 1.6% in knockouts and 51.2% in heterozygous controls ( Fig 2K); Cfap97d1 em1 line: 27.0% in knockouts and 88.7% in heterozygous controls (S4G Fig)]. However, the fertilization rate of Cfap97d1 knockout spermatozoa of both strains with zona pellucida-free oocytes was comparable to that of the control (Fig 2K, S4G Fig). These results demonstrate that a lack of Cfap97d1 affects sperm ability to penetrate the zona pellucida, resulting in severe fertility defects in vivo.
Finally, to determine whether the Cfap97d1 deficient spermatozoa's genome is intact and can contribute to the next generation, we performed intracytoplasmic sperm injection (ICSI) and IVF in zona-loosening conditions using glutathione containing medium [21,22] using the Cfap97d1 em1 mouse line. As a result, egg activation ability of Cfap97d1 knockout sperm heads (ICSI, (S4H Fig)) and intact sperm (zona-loosened IVF, (S4I Fig)) was similar to heterozygous controls. The number of delivered pups was comparable between the two genotypes (S4J and S4K Fig). These results demonstrate that nuclei of knockout spermatozoa have the ability to produce viable pups by using assisted reproduction techniques.

Cfap97d1 determines flagellar bending and frequency
Next, we examined sperm motility as zona penetration defects often appear when spermatozoa show reduced motility [23,24]. The percentage of motile and progressive sperm were not significantly different between control and knockout mice of both strains (S3E and S3F Fig). We next analyzed sperm motility parameters defining the speed of sperm motion in knockout and control mice. Curvilinear velocity (VCL, the average velocity of the sperm head through its real path) and average path velocity (VAP, average velocity of the sperm head through its average trajectory) were reduced in knockout sperm compared to controls (Figs 3A, 3B, S5A and S5B), whereas straight-line velocity (VSL, average velocity of the sperm head through the straight line connecting the first position with the last track) (Fig 3C, S5C Fig) was not changed.
Next, we analyzed the flagellar beat in greater detail by recording tethered sperm tail motion at 200 frames per second (fps) with a high-speed camera (S1 and S2 Movies). The analysis revealed that the flagellar beat in Cfap97d1 KOMPwt/control sperm was symmetrical with respect to a line through the midpiece (Fig 3D and 3F), whereas Cfap97d1 KOMP-/sperm had a more asymmetrical waveform pattern (Fig 3E and 3G). We have quantified the difference in flagellar bending between control and Cfap97d1 KOMP-/null sperm by calculating the asymmetry index. A symmetrical flagellar beat is indicated by an asymmetry index of 0. The asymmetry index of Cfap97d1 KOMP-/knockout sperm was higher compared to control sperm ( Fig  3H). Quantification showed that 59.3% of Cfap97d1 KOMP-/sperm cells were predisposed to be in the anti-hook conformation in comparison to control 29.6% of control Cfap97d1 KOMPwt/mice. The asymmetrical flagellar waveform was also recorded in Cfap97d1 em1/em1 sperm before (S5D and S5E Fig Additionally, the average flagellar beating frequency was reduced in the midpiece of Cfap97d1 KOMP-/sperm at 5-30μm ( �� P � 0.01) and 52-64 μm ( � P � 0.05 (Fig 3I)) as measured from the head. The amplitude of the flagellar beat with respect to the head-midpiece axis in the Y-direction was also decreased in Cfap97d1 KOMP-/sperm ( Fig  3J, red line, significant differences observed from 29-56 μm ( �� P � 0.01) and 57-67 μm, ( � P � 0.05), total analyzed cells n = 27 from three mice of each genotype).
To analyze whether Cfap97d1 sperm can undergo capacitation we analyzed tyrosine phosphorylation. The protein kinase A-dependent activation of tyrosine kinase results in tyrosine phosphorylation during capacitation, which can be visualized by Western blot [25,26]. The analysis did not show any difference in tyrosine phosphorylation between sperm derived from control Cfap97d1 wt/em1 and knockout Cfap97d1 em1/em1 mice (S6A Fig).

Cfap97d1 knockout sperm flagellum frequently lack microtubule doublet 4
To reveal whether the defect in sperm motility in Cfap97d1 deficient sperm is due to defects in the axoneme, we analyzed the flagellar ultrastructure of epididymal sperm of the Cfap97d1-KOMP line by transmission electron microscopy (TEM; Fig 4A-4F). The flagellar cross-section showed that mitochondrial sheaths in the midpiece were comparable in control and Cfap97d1-KOMP-/sperm (Fig 4A and 4B). Similarly, the outer dense fiber (ODF) layer was mostly intact in controls and knockouts (Fig 4) except for the very few cases when ODF counts were not complete in Cfap97d1 KOMP-/- (Fig 4B). Central microtubule singlets (CMtS) were centrally located and indistinguishable in control and Cfap97d1 KOMP-/mice ( Fig 4A' and 4B'). Outer microtubule doublets (OMtD) had outer-(yellow arrowhead) and inner dynein arms (blue arrowhead) attached to them as well as prominent radial spokes (yellow star in Fig 4A' and 4B') in control and Cfap97d1 KOMP-/mice. However, the number of OMtD were irregular in 45% of flagellum in Cfap97d1 KOMP-/sperm ( Fig 4G) while only 7% of OMtDs were defective in control (Fig 4G). When the axoneme is viewed from the sperm head towards the tail, the outer doublets can be numbered 1 through 9 [27]. Number 1 is the doublet situated on a plane perpendicular to that bisecting the microtubules of the central pair, the next doublet clockwise is number 2 and so on (Fig 4A-4F). The following quantification using this system showed that in 63% of the cases axonemes with abnormal counts in Cfap97d1 KOMP-/sperm were missing the fourth doublet ( Fig 4H). The 7 th doublet was missing less often, in 14.8%, whereas 9.8% axonemes had 10 OMtDs. We have also quantified the cases where multiple doublets were missing but they were relatively rare (4 th and 7 th in 1.6%; 4, 5, and 7 th in 3.7%; 4-6 th in 1.23%; 4-7 th in 1.23%). All parts of the flagellum, mid-, principal and end pieces had missing doublets. Additionally, the radial spoke associated with the missing OMtD was also gone (Fig 4B', 4D compare to 4A' and 4C).
We were interested if observed abnormal OMtD counts were associated with flagellum biogenesis or OMtD destabilization. To address this question, we have done TEM analyses of the sperm flagella in testes seminiferous tubules. From these experiments, we have observed that both control and Cfap97d1 knockout flagellum had normal (9+2) OMtD counts (S6B and S6C  Fig). This indicates that Cfap97d1 is likely not controlling the flagellum biogenesis but rather plays a role in axoneme integrity maintenance.
Taken together our data demonstrate that Cfap97d1 is required for axonemal doublet stabilization and sperm flagellum structural integrity (Fig 5).

Discussion
The prerequisite for successful fertilization in mammals is sperm motility within the female reproductive tract in order to reach and fertilize the egg [39,40]. Herein, we report Cfap97d1 as an essential gene for axoneme integrity maintenance, which is required for sperm motility and fertilization.  We have identified that Cfap97d1 is a testis-enriched gene in humans and mice. In the mouse testes, Cfap97d1 starts to be expressed at PND 20. This correlates with the diplotene spermatocyte stage. The expression becomes prominent from PND 25 onwards, the stage when post-meiotic round spermatids and flagella begin to form [41,42].
We have used two independent Cfap97d1 knockout mouse models to characterize the gene function in vivo. About half of the Cfap97d1 knockout male mice were infertile, while remaining males were severely sub-fertile due to asthenozoospermia, i.e. reduced sperm motility, and could sire some pups. The two independently generated mouse models presented in this article had similar sub-fertility phenotypes, excluding the possibility that sub-fertility could have been attributed to inefficient gene deletion. Even though mRNA levels of the genes from the same family (Cfap97 and Cfap97d2) were not changed the possible compensatory effect on the Cfap97d1 knockout phenotype is yet to be determined by more detailed studies. The comprehensive studies by assisted reproduction techniques (IVF and ICSI) confirmed that nuclei of Cfap97d1 knockout spermatozoa have the ability to produce viable pups.
Sperm motility is one of the key factors determining fertility. The examination of Cfap97d1 knockout sperm motility, by an automated CASA system, has identified a motility defect. Indepth analysis using high speed cameras has revealed that the flagellar waveform was highly asymmetric and was prone to stay in the anti-hook conformation. Correspondingly, swimming velocity was significantly reduced in mutant sperm. Additionally, flagellar beat frequency and amplitude of sperm flagella were reduced in Cfap97d1 KOMP knockout mice.
Flagellum formation starts in round spermatids, when the axoneme is formed, and later in elongating spermatids the accessory structures are organized. [41,42]. Cfap97d1 is necessary for maintenance of the flagellar ultra-structure; loss of Cfap97d1 resulted in reduced counts of outer doublets and specifically doublet 4 was frequently lost in the sperm stored in the cauda epididymis. It has been proposed that doublets 4-7 facilitate microtubule sliding during motility and doublet loss was associated with motility defects [43,44]. Failure to maintain intact doublet structure was shown to be associated with infertility, e.g. the deletion of Ttll9 in mice caused doublet 7 shortening in the distal portion of the principal piece in sperm flagellum, along with reduction of doublet 5 polyglutamylation, leading to biased anti-hook bending and male infertility, similar to Cfap97d1 knockout males [45]. Similarly, deletion of Vdac3 [44], Pla2g3 [46], and DNAH17 [47] caused instability of sperm microtubule doublets 4-7, associated with sperm motility defects and male infertility. Additionally, in a DNAH17 missense variant, spermatozoa were disorganized during storage in cauda epididymis but not in testes, consistent with our studies. Given this, our and other laboratory studies show that doublet 4-7 stability contributes to a subtle control mechanism of microtubule sliding. Therefore, further studies will be needed to gain a thorough understanding of how this delicate system functions.
Interestingly, in Chlamydomonas, Pazour and colleagues have identified FAP97 (Cfap97d1 orthologue in Chlamydomonas reinhardtii) in the axoneme fraction extracted by the KCl method [17]. The KCl extraction releases various axonemal proteins, including those of the inner dynein arms and the C2 central microtubule [17]. It is tempting to consider that Cfap97d1 expression could be associated with C2 axonemal central pair apparatus proteins that are in close proximity to doublet 4 ( Fig 5). However, it should be noted that our TEM analysis reveals that dynein arms and the C2 microtubule still exist in the Cfap97d1 knockout mice. Mammalian homologs of other members of C1d or C2b have been identified. Mutations in the underlying genes, CFAP54 (C1d) and HYDIN (C2b), lead to symptoms associated with primary ciliary dyskinesia including sperm motility defects [48][49][50].
Multiple morphological abnormalities of the sperm flagella (MMAF) is a rare syndrome that causes primary infertility. So far there are several genes associated with MMAF: CFAP43, CFAP44, CFAP69, AKAP4, DNAH1, DNAH17, CCDC39, and QRICH2 [47,[51][52][53][54][55][56]. However, the genetic causes are unknown in approximately half of MMAF cases. Generally, in MMAF syndrome, sperm heads are normal and the sperm defects occur during the last stages of spermatogenesis in spermatids, when the flagellum is being assembled and undergoes elongation [57]. Thus, Cfap97d1 could be considered as a possible candidate gene for MMAF syndrome as Cfap97d1 deletion causes ultrastructural (but not gross-morphological) flagellar defects characteristic in MMAF. It remains to be seen how the phenotype will be affected by functional alterations of CFAP97D1 in other species.
This study provides insight into the function of a previously uncharacterized gene and demonstrates that Cfap97d1 is an important gene controlling structural integrity of the flagellum, sperm motility, and fertilization properties in mice.

Transgenic animals, genotyping, and breeding
Cfap97d1 tm1a(KOMP)Wtsi/+ (referred to as Cfap97d1 KOMPwt/-) LacZ-tagged 'knockout-first' conditional allele mice were obtained from the KOMP consortium and maintained in a C57BL6J (C57BL/6N)/129SvEv background. The null allele in these mice is generated through splicing to a LacZ trapping element present in the targeting cassette [58]. Primers used for genotyping are presented in S1 Table. The Cfap97d1 wt/em1 mice were generated using the CRISPR-Cas9 technology (description follows) and maintained in B6D2 background. All

Egg collection for genome editing
To prepare eggs for knockout mouse production, female mice were superovulated using injection of CARD HyperOva (0.1 mL, Kyudo, Saga, Japan) into the abdominal cavity of B6D2F1 females, followed by injection of human chorionic gonadotropin (hCG) (7.5 units, ASKA Pharmaceutical, Tokyo, Japan). Natural mating was done with B6D2F1 males 46~48 h after CARD HyperOva injection. After 19-21 h, cumulus-intact eggs were collected and treated with 0.33 mg/mL hyaluronidase (Wako, Osaka, Japan) for 5 min to remove cumulus cells for genome editing. Obtained eggs were cultured in KSOM medium [59] at 37˚C under 5% CO 2 until subsequent treatments.

RNA isolation, Reverse Transcription-and quantitative Real Time-Polymerase Chain Reaction
Mouse cDNA was prepared from multiple adult tissues of C57BL6J/129SvEv hybrid mice and testes. Briefly, tissues were dissected and snap frozen in liquid nitrogen. RNA was extracted using RNeasy Protect Mini kit (Qiagen). RNA template (0.5-1 μg/uL) was transcribed to cDNA using First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) or qSCRIPT cDNA supermix (Quanta) following the manufacturer's conditions. The generated cDNA was used to perform Reverse Transcription-(RT-PCR) and quantitative Real Time-Polymerase Chain Reaction (qRT-PCR). The primers used are listed in S1 Table. QRT-PCR was performed as described earlier [60]. Briefly, cDNA was diluted 1:10 and 1 μL was used for qRT-PCR in a total volume of 10 μL. The qRT-PCR program consisted of 40 cycles at 95˚C for 30 s and at 60˚C for 1 min in a CFX96 Real-Time System (BioRad) thermocycler. Gapdh was used for normalization by the ΔΔCT method [61].

Histology, beta-galactosidase staining and immunostaining
Observation of testes and sperm morphology. Testes and epididymides were fixed in Bouin's solution at 4˚C for 8 h, followed by dehydration in increasing ethanol concentrations and embedding in paraffin. Five micrometer thick paraffin sections were hydrated in decreasing ethanol concentrations, stained with periodic acid-Schiff (PAS, Wako), counterstained with Mayer's Hematoxylin solution (Wako), dehydrated in increasing ethanol concentrations, and finally mounted with Permount. The sections were observed using BX53 (Olympus, Tokyo, Japan) and DM LB2 (Leica, Germany) microscopes. Whole testis and cauda-epididymal sperm were observed using BX50 and BX53 microscopes with phase contrast (Olympus).

Immunoblot analysis and isolation of sperm proteins for tyrosine phosphorylation analysis
Protein lysates were resolved by SDS/PAGE under reducing condition (with 5% 2-mercaptoethanol) and transferred to PVDF membranes. After blocking with 10% skim milk, blots were incubated with primary antibodies overnight at 4˚C and then incubated with secondary antibodies conjugated to horseradish peroxidase for 1 h at room temperature.
For tyrosine phosphorylation analysis spermatozoa were collected from the cauda epididymis and incubated under capacitating conditions in TYH medium [63] for 10 min. or 120 min. Spermatozoa were then collected in PBS and centrifugated at 500 x g for 3 min at room temperature. The collected spermatozoa were resuspended in sample buffer and boiled for 3 min. The cell debris was removed by centrifugation (15,000 x g, 5 min.) and the supernatant was subjected to immunoblot analysis as described above using 5% BSA instead of 10% skim milk for blocking.

Fertility testing
Sexually mature Cfap97d1 KOMPwt/and mutant Cfap97d1 KOMP-/male mice were housed with wild-type females for at least three months, copulation was confirmed by checking for vaginal plugs and the number of pups in each cage was recorded.

In vitro fertilization and genome integrity analysis by intracytoplasmic sperm injection (ICSI) and zona-loosen IVF
In vitro fertilization (IVF) was performed as described previously [65] with slight modifications. Briefly, cumulus-intact eggs collected from super-ovulated females 14-16 h after hCG injection (62-64 h after CARD HyperOva or PSMG injection) were placed in TYH medium [63]. To prepare zona-free eggs, eggs were treated with 1 mg/mL collagenase (Wako). Cauda sperm were collected from sexually mature males and incubated in TYH medium [63] for 2 h for capacitation. Capacitated spermatozoa were added to the drop containing eggs at a final concentration of 2 x 10 5 sperm/mL. ICSI was performed as previously described with some modifications [50,66]. Briefly, mature oocytes were collected from super-ovulated B6D2F1 mice. After treatment with hyaluronidase to remove cumulus cells, oocytes were placed in KSOM medium at 37˚C under 5% CO 2 until ICSI. Sperm heads were separated from tails by applying a few piezo pulses, then injected into the MII oocyte using a piezo manipulator (Prime Tech, Ibaraki, Japan).
Zona-loosened IVF was performed using CARD MEDIUM (KYUDO company, Saga, Japan) as described in the instruction manual. From vial B, 15 μL usually used for IVF with frozen-thawed sperm was used for CARD MEDIUM.
Two-cell embryos obtained by ICSI and zona-loosened IVF were transferred to pseudopregnant females the following day. Pups were genotyped at birth.

Sperm motility analysis
Cauda sperm were extracted from control (HET) and knockout Cfap97d1 littermates and incubated in HTF (capacitating conditions) media (Millipore) supplemented with fetal bovine serum (10%) for Cfap97d1 KOMP or TYH medium [63] for Cfap97d1 em1 mice at 37˚C. Sperm samples were diluted and analyzed using Hamilton Thorne's CEROSII sperm analysis system (software version 1.5.2; Hamilton Throne Biosciences, Beverly, MA).
Motility of Cfap97d1 KOMPwt/and Cfap97d1 KOMP-/sperm was additionally recorded after incubation in HTF media (Millipore) supplemented with bovine serum albumin (0.3 mg/mL) using a Zeiss Axio Observer microscope mounted with a high-speed CCD camera (Hamamatsu ORCA-Flash 4.0 V2) at 200 frames per second (fps), Biocenter Oulu Tissue Imaging Center). Sperm motility was analyzed using SpermQ Analysis and Evaluator programs [67]. Briefly, the average flagellar beat frequency, amplitude, and curvature angles were calculated at each arc-length position from all frames in the recording (between 0 and 400 frames, 20.6 ± 7.4 full beat cycles). To calculate the average flagellar beat asymmetry index, the median curvature angle values were used [68]. To analyze sperm bending (pro-or anti-hook conformation), one beat cycle was observed manually to see whether the cell opened clockwise or counterclockwise into the pro-hook or anti-hook conformation. If the sperm flagellum opened clockwise into pro-hook, the median curvature angle values remained the same, but if the flagellum opened clockwise to anti-hook, the values were multiplied by -1 for normalization. A total of twenty-seven sperm cells were analyzed in capacitating conditions from three mice of each genotype, Cfap97d1 KOMPwt/-(n = 27 cells) and Cfap97d1 KOMP-/-(n = 27 cells).
Sperm motility analysis of Cfap97d1 wt/em1 and Cfap97d1 em1/em1 was performed as previously described [23] with slight modifications. Briefly, cauda-epididymal spermatozoa from mice used for IVF were suspended and incubated in TYH medium (capacitating condition) [63]. Sperm motility was analyzed with an Olympus BX-53 microscope equipped with a highspeed camera (HAS-L1, Ditect, Tokyo, Japan). The sperm motility was recorded at 200 frames per second. Obtained images were analyzed for beat frequencies and waveforms using sperm motion analyzing software (BohBohsoft, Tokyo, Japan).

Transmission electron microscopy
Tissues were fixed in 1% glutaraldehyde and 4% formaldehyde in 0.1 M phosphate buffer, pH 7.4, post-fixed in 1% OsO4, dehydrated in acetone and embedded in Epon LX 112 (Ladd Research Indus-tries, VT, USA) at Biocenter Oulu Tissue Imaging Center or alternatively fixed in 2% PFA + 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4 and post-fixed in 1% OsO4 in 0.1 M cacodylate buffer + 0.2% potassium ferricyanide, following embedding in Spurr's Low Viscosity resin at Integrated Microscopy Core at Baylor College of Medicine. Semi-thin sections (1 μm) were cut and stained with toluidine blue for light microscopic inspection and selection of regions of interest. Thereafter, thin sections (80 nm) were cut and post-stained in uranyl acetate and lead citrate. Specimens were examined using the Tecnai GS Spirit microscope (FEI Europe, Edinhoven, Netherlands) and images were acquired with a Quemesa CCD camera controlled by the iTEM software (Olympus Soft Imaging Solutions GmbH, Munster, Germany) at Biocenter Oulu Tissue Imaging Center.

Statistical analysis
Statistical analysis was performed using a two-tailed student's t-test ( � P � 0.05, �� P � 0.01 and ��� P � 0.001) by Microsoft Excel, GraphPad Prism 6 (GraphPad, San Diego, CA, USA) or R. Data represent the means ± standard deviation (±SD). At least three mice were used in each experimental group.  /em1 (B). PAS staining of epididymis sections of Cfap97d1 wt/em1 (C) and Cfap97d1 em1/em1 (D) illustrates sperm presence in the tubules (arrowheads). Spermatozoa collected from cauda epididymis of control Cfap97d1 wt/em1 (E) and knockout Cfap97d1 em1/em1 (F) does not show gross morphological changes. (G) IVF with cumulus-intact oocytes indicates significantly reduced fertilization ability of Cfap97d1 em1/em1 sperm, whereas IVF with zona pellucida-free oocytes with Cfap97d1 em1/em1 deficient sperm is comparable with control. Males (n = 3) each for Cfap7d1 wt/em1 and Cfap97-d1 em1/em1 were examined (sperm concentration: 2.0x10 5 sperm/mL). Error bar indicates unbiased standard deviation of fertilization rate per male. for assisting with high-speed camera imaging at Biocenter Oulu and Dr. Lari Lehtiö for valuable feedback on this manuscript.