Novel biallelic FSIP2 variants cause male infertility with multiple morphological abnormalities of sperm flagella in humans

Abstract

Biallelic variants in fibrous sheath-interacting protein 2 (FSIP2) gene are a known cause of multiple morphological abnormalities of the sperm flagella (MMAF). This study aimed to identify novel FSIP2 variants and evaluate their impact on sperm ultrastructure and intracytoplasmic sperm injection (ICSI) outcomes. Whole-exome sequencing (WES) was employed to screen a cohort of 92 MMAF patients, with candidate variants validated via Sanger sequencing and third-generation sequencing. We identified one homozygous variant in a proband from a consanguineous family and two pairs of compound heterozygous variants in two unrelated, non-consanguineous families. Routine semen analysis demonstrated markedly reduced motility across all probands. Detailed morphological and ultrastructural assessments using Papanicolaou staining, scanning electron microscopy (SEM), and transmission electron microscopy (TEM) demonstrated that approximately 80.0% of spermatozoa exhibited pathological elongation of the mitochondrial sheath in the midpiece. Furthermore, 50.0%–70.0% of spermatozoa displayed fibrous sheath dysplasia or loss in the principal piece. Immunofluorescence assays and Western blotting confirmed that FSIP2 protein localization was disrupted, and the expression of key axonemal assembly factors was dysregulated. Notably, successful pregnancies were achieved via ICSI in the partners of two probands. This study expands the mutational spectrum of FSIP2 in both consanguineous and non-consanguineous populations. Ultrastructural abnormalities, such as mitochondrial sheath elongation and fibrous sheath disassembly, highlight FSIP2’s critical role in flagellar assembly. Clinical results further support ICSI as an effective therapeutic intervention for affected individuals.

INTRODUCTION

Male infertility affects approximately 10.0%–20.0% of couples worldwide,1 with genetic factors accounting for at least 15.0% of cases.2 Multiple morphological abnormalities of the sperm flagella (MMAF) represent a critical yet understudied etiology of male infertility. This condition is defined by five specific flagellar defects, absent, short, bent, coiled, and irregular-width flagella. Importantly, these abnormalities occur without any other symptoms associated with primary ciliary dyskinesia (PCD).3 MMAF accounts for 10.0%–15.0% of severe asthenoteratozoospermia (AT) cases and is strongly associated with genetic defects in flagellar structural components.4 The increasing prevalence of idiopathic AT underscores the urgency of identifying genetic variants linked to MMAF, as such discoveries could revolutionize diagnostic and therapeutic strategies.

The flagellum is a sophisticated protrusion of the apical cell membrane essential and is characterized by its compartmentalized architecture. The midpiece of sperm flagellum, extending from the neck to the annulus, contains the conserved “9+2” axoneme which composes of nine peripheral doublet microtubules and two central pair microtubules (CP) and is surrounded by nine outer dense fibers (ODFs).4,5 Encasing these ODFs is a tightly wound mitochondrial sheath (MS), typically comprising 10–12 helical gyres in humans.6,7 In contrast, the principal piece retains the axonemal core but replaces mitochondrial gyres with the fibrous sheath (FS), a scaffold composed of dorsal and ventral longitudinal columns (LCs) and circumferential ribs (CR).8 These LCs originate from the third and eighth doublet microtubules (DMTs) of the axoneme, enforcing planar bending by restricting flagellar movement to a single plane, a mechanism crucial for efficient propulsion.9 Beyond its structural role, the FS harbors glycolytic enzymes such as glyceraldehyde-3-phosphate dehydrogenase (GAPDH), enabling localized adenosine triphosphate (ATP) production during female tract transit.10,11 Disruption of this intricate architecture, particularly the FS, leads to impaired motility and is a hallmark of MMAF.

Fibrous sheath-interacting protein 2 (FSIP2) is located on chromosome 2q32.1 and is approximated 115.4 kb in length and consists of 25 exons. Homo sapiens FSIP2 is transcribed into a 20 879-nucleotide mRNA (NM_173651.4) that encodes a 6907-amino acid protein containing four conserved FSIP2 domains and nine intrinsically disordered regions. Recent studies have established FSIP2 as a pivotal regulator of sperm flagellar integrity;5 it localizes to the FS and directly interacts with A-kinase anchoring protein 4 (AKAP4), the major structural component of the sheath.12 Although FS provides structural support and contributes to signal transduction and glycolysis,13 no direct association between FSIP2 and glycolytic enzymes has yet been established. Furthermore, despite the established role of biallelic FSIP2 variants in MMAF, significant phenotypic heterogeneity exists across studies. Regarding sperm count, reports vary widely: while 62.9% (22/35) of patients presented with normal sperm counts,14,15,16,17,18,19,20,21,22,23,24 37.1% (13/35) exhibited oligozoospermia.16,25,26,27,28 Similarly, progressive motility ranges from near-normal28 to severe impairment,14,15,16,17,19,20,22,24,27 likely influenced by specific variant pathogenicity. Ultrastructural findings are also inconsistent; for instance, Martinez et al.16 observed truncated mitochondrial sheaths, whereas Lv et al.17 reported elongated sheaths. Assisted reproductive technology (ART) outcomes remain equally contentious. Although intracytoplasmic sperm injection (ICSI) has achieved high pregnancy rates in confirmed biallelic carriers,14,17,20,22 one unvalidated case reported success via in vitro fertilization (IVF) in a patient with unverified compound heterozygous variants.28 Conversely, Hou et al.14 reported IVF failure followed by ICSI success, potentially linking FSIP2 defects to acrosomal dysfunction. Finally, head morphology remains debated: while some studies reported globozoospermia and amorphous heads,19,20 others observed normal head morphology.14,17 This variability highlights the need for further investigation into the functional consequences of FSIP2 variants to optimize genetic counseling and ART strategies.

In the present study, we employed whole-exome sequencing (WES), Sanger sequencing, and third-generation sequencing to identify one novel homozygous and two compound heterozygous FSIP2 variants in three unrelated MMAF cases predominantly characterized by fibrous sheath dysplasia (FSD). We performed functional analyses to determine the impact of these variants on sperm motility, morphology, and flagellar ultrastructure. Furthermore, we evaluated the ICSI outcomes of two FSIP2-affected couples. Our findings suggest that these novel FSIP2 variants disrupt the ultrastructural organization of the sperm flagellum, ultimately leading to MMAF and male infertility.

PARTICIPANTS AND METHODS

Study participants and ethical compliance

Between August 2019 and December 2024, a total of 92 MMAF patients and 3777 non-MMAF infertile patients were recruited from the Reproductive Center of Northwest Women’s and Children’s Hospital (Xi’an, China) for this study. Within the non-MMAF infertile cohort, phenotypes included oligozoospermia (sperm concentration <15 × 10⁶ ml⁻¹; n = 275), asthenozoospermia (progressive motility <32%; n = 1017), teratozoospermia (normal morphology <4%; n = 1971), and idiopathic infertility with normal semen parameters (n = 1723). Patients exhibiting multiple defects were categorized into all applicable groups. A total of 495 patients who had undergone prepregnancy semen examination from January 2022 to December 2022 in Northwest Women’s and Children’s Hospital and whose partners were pregnant within 1 year were recruited as normal controls. All participants provided written informed consent for the collection of the blood and semen samples, as well as for the publication of their clinical data. The study protocol was reviewed and approved by the Institutional Review Board of Northwest Women’s and Children’s Hospital (Ethics Approval No. 2022020).

The inclusion criteria for infertile Chinese males were as follows: (1) a history of infertility of no less than 1 year; (2) age between 22 years and 45 years, matched to the age distribution of the fertile control group; and (3) agreement to participate in genetic carrier screening based on WES technology. The exclusion criteria were (1) a diagnosis of azoospermia or severe oligozoospermia (sperm concentration <1 × 10⁶ ml⁻¹); (2) an abnormal chromosome karyotype; (3) microdeletions in the azoospermia factor (AZF) region of the Y chromosome; (4) a physical examination revealing congenital absence of the vas deferens; (5) an ultrasound revealing an absence of seminal vesicles or the vas deferens; (6) a physical examination revealing a testicular volume <15 ml; or (7) a refusal to provide detailed follow-up information. The inclusion criteria for normal controls were as follows: (1) had undergone prepregnancy semen examination at the Reproductive Center of Northwest Women’s and Children’s Hospital; (2) were between 22 years old and 45 years old; (3) had partners who were pregnant within 1 year; (4) had normal semen concentration, progressive motility, and morphology according to the 5^th or 6^th editions of the World Health Organization (WHO) manual;29,30 (5) had no history of reproductive disorders; (6) had no family history of genetic diseases; and (7) had no hormonal therapy or exposure to reproductive toxins within 3 months before prepregnancy semen examination. The exclusion criteria were (1) a physical examination revealing testicular volume <15 ml; (2) a physical examination revealing congenital absence of the vas deferens; or (3) a refusal to provide detailed follow-up information. Notably, total sperm immotility was not used as an exclusion criterion for either cohort. Detailed semen characteristics for all participants are provided in Supplementary Table 1.

Supplementary Table 1.

Semen characteristics of infertile men and normal controls

Semen parameters	Non-MMAF cohort	FSIP2 heterozygous cohort	Control	Reference
Semen volume (ml)	3.7±1.8	3.9±1.5	3.9±1.5	≥1.5
Sperm concentration (106/ml)	55.2±33.0	60.4±18.8	70.1±32.8	≥15
Progressive motility (%)	44.8±18.9	48.6±14.3	52.5±8.0	≥32
Normal morphology (%)	3.8±2.6	4.0±2.2	6.3±2.4	≥4

Routine semen parameters were evaluated according to the WHO-2010 guidelines; sperm morphology parameters were evaluated according to the WHO-2021 guidelines; FSIP2: fibrous sheath-interacting protein 2; MMAF: multiple morphological abnormalities of the sperm flagella; WHO: World Health Organization

WES and variant analysis

The specific details of the variant filtering strategy are described in our previous study.31 Briefly, genomic DNA was isolated from peripheral blood samples using the QIAamp DNA Blood Mini Kit (catalog No. 69504; QIAGEN, Hilden, Germany), according to the manufacturer’s protocol. The extracted DNA was subsequently subjected to WES for exonic variant screening. Library preparation, exome sequencing, and data analysis were performed by the AmCare Genomics Laboratory (Guangzhou, China). Briefly, raw fast alignment search tool-quality score (FASTQ) reads were processed using fastp software (https://github.com/OpenGene/fastp, last accessed on December 19, 2024) to remove adapter sequences, polyN/polyA tracts, and low-quality reads (Q score <20), generating clean reads for downstream analysis. Sequencing reads were aligned to the reference human reference genome GRCh37/hg19 using Burrows-Wheeler Aligner (version 0.7.15; https://bio-bwa.sourceforge.net/, last accessed on December 30, 2024). The clinical significance of variants was assessed using publicly available databases: gnomAD (https://gnomad.broadinstitute.org/, last accessed on January 08, 2025), ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/, last accessed on January 08, 2025), and Online Mendelian Inheritance in Man (OMIM; https://www.omim.org/, last accessed on January 08, 2025), and peer-reviewed literature. Pathogenicity predictions were performed using rare exome variant ensemble learner (REVEL), Polyphen2, MutationTaster, and combined annotation dependent depletion (CADD) to evaluate evolutionary conservation and structural/functional impacts. Variants were classified according to the American College of Medical Genetics and Genomics (ACMG)/Association for Molecular Pathology (AMP) guidelines.32 In addition, ACMG/AMP evidence was translated into a Bayesian point-based framework as recommended by the Clinical Genome Resource (ClinGen) to quantitatively refine variant classification.33 Candidate pathogenic variants were validated by Sanger sequencing of polymerase chain reaction (PCR) products (primers in Supplementary Table 2) using an ABI 3730XL sequencer (Applied Biosystems, Foster City, CA, USA) and analyzed with Mutation Surveyor version 4.0.5 software (Softgenetics, State College, PA, USA). The WES datasets supporting the current study have not been deposited in a public repository because of privacy and ethical restrictions, but are available from corresponding authors on request.

Supplementary Table 2.

Primers for amplification and sequencing analysis of fibrous sheath-interacting protein 2 variants

Premiers name	Sequences
FSIP2-M1-F	5’-TGTTACCTTCTGTCTCCAAAACA-3’
FSIP2-M1-R	5’-ACAGGGTTCTGCCTGAAATG-3’
FSIP2-M2-F	5’-GGCCATTCCCTCTCTCTCTT-3’
FSIP2-M2-R	5’-TTCTAAAATGGCATGCACCA-3’
FSIP2-M3-F	5’-ACTGAGCCAAGCCAGAGTGT-3’
FSIP2-M3-R	5’-TGAGAACCAGAGGTTTGCAC-3’
FSIP2-M4-F	5’-TCTCTGCTTACCTCCTTTCCA-3’
FSIP2-M4-R	5’-AGGTGTGGTTGGTCTCCTTA-3’
FSIP2-M5-F	5’-CCAACCAAGGCAGACATCAG-3’
FSIP2-M5-R	5’-TGGCAACCATTTTCTGTGCA-3’
FSIP2-P2-F	5’- CGTCAGTAAGGACACCAACCACACC-3’
FSIP2-P2-R	5’-TGACTGCTCTGCACTGGTGTTCAAT-3’

M: mutation; P: proband; F: forward; R: reverse; FSIP2: fibrous sheath-interacting protein 2

In silico structural modeling

To model the structural impact of identified variants, the wild-type FSIP2 sequence (UniProt ID: Q8N0X7) was modified with the specific mutations and processed via AlphaFold3 (http://alphafoldserver.com/, last accessed on June 30, 2025).34 Predicted three-dimensional (3D) structures (PDB format) were analyzed for conformational changes using PyMOL (version 2.5.2; Schrödinger, New York, NY, USA).

Routine semen analysis

Semen samples collected after 2–7 days of abstinence were analyzed using a computer-assisted sperm analyzer (HamiltonTOX IVOS; Hamilton Thorne Inc., Beverly, MA, USA) in compliance with the 5^th edition of the WHO manual.29 Motility and kinetic parameters were quantified.

Assessment of sperm morphology

Sperm morphology parameters were evaluated according to the 6^th edition of the WHO manual.30 Briefly, sperm smears were fixed in 95% ethanol, stained with modified Papanicolaou reagents (Cariad Medical Technology, Zhuhai, China), and evaluated under a Leica TCS SP8 microscope (Leica, Wetzlar, Germany).

Electron microscopy evaluation

For scanning electron microscopy (SEM), glutaraldehyde-fixed samples were sputter-coated (ACE200; Leica) and imaged on a Nova Nano SEM 450 (Thermo Fisher, Waltham, MA, USA). For transmission electron microscopy (TEM), resin-embedded ultrathin sections (70 nm) were stained with uranyl acetate/lead citrate and were examined using a Thermo Fisher Tecnai 10 (Thermo Fisher).

Immunofluorescence (IF) staining

The sperm were fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, Darmstadt, Germany), and blocked with 3% bovine serum albumin (Servicebio, Wuhan, China). Primary antibodies against FSIP2 (1:100; PA5-112962; Invitrogen, Carlsbad, CA, USA), β-tubulin (1:500; GB15140-100; Servicebio), A-kinase anchoring protein 4 (AKAP4; 1:100; 24986-1-AP; Proteintech, Wuhan, China), sperm-associated antigen 6 (SPAG6; 1:100; 12462-1-AP; Proteintech), dynein axonemal heavy chain 17 (DNAH17; 1:100; 24488-1-AP; Proteintech), dynein axonemal light intermediate chain 1 (DNALI1; 1:100; 17601-1-AP; Proteintech), translocase of outer mitochondrial membrane 20 (TOM20; 1:100; 11802-1-AP; Proteintech), and intraflagellar transport 20 (IFT20; 1:100; 20536-1-AP; Proteintech) were incubated overnight at 4°C. Alexa Fluor 488-conjugated peanut agglutinin (PNA; 40 μg ml⁻¹, L21409; Invitrogen), fluorescently-conjugated secondary antibodies (goat anti-rabbit IgG Alexa Fluor 488, 1:500, GB25303, Servicebio; goat anti-mouse IgG Alexa Fluor 594, 1:500, GB28303, Servicebio; CoraLite647-conjugated F[ab’]2 fragment donkey anti-mouse IgG [H+L], 1:500, SA00014-8, Proteintech), and 4’,6-diamidino-2-phenylindole (DAPI; G1012; Servicebio) counterstaining were applied. Images were captured using a laser scanning confocal microscope (Leica).

Western blotting

Sperm samples were lysed with radioimmunoprecipitation (RIPA) lysis buffer supplemented with protease inhibitors (R0010; Solarbio, Beijing, China) and centrifuged at 13 800g for 10 min at 4°C (Centrifuge 5424R; Eppendorf, Hamburg, Germany). Then, 20 μg of protein was loaded on a 10% sodium dodecyl sulfate (SDS)-gel, resolved using SDS-polyacrylamide gel electrophoresis (PAGE), and transferred to a polyvinylidene fluoride (PVDF) membrane. The membranes were incubated with primary antibodies against TOM20 (1:500; Proteintech), DNAH17 (1:500; Proteintech), AKAP4 (1:500; Proteintech), IFT20 (1:500; Proteintech), SPAG6 (1:500; Proteintech), and DNALI1 (1:500; Proteintech), followed by incubation with a horseradish peroxidase (HRP)-goat anti-rabbit recombinant secondary antibody (1:5000; RGAR001; Proteintech) for 2 h at room temperature. The bands were visualized using enhanced chemiluminescence (G2161; Servicebio). β-actin (1:2000; GB15003; Servicebio) and β-tubulin antibody (1:2000; GB15140; Servicebio) served as an internal control.

ART procedures

Controlled ovarian stimulation was performed using either a gonadotropin-releasing hormone (GnRH) antagonist protocol or a long GnRH agonist protocol, selected according to the female partner’s ovarian reserve, age, and clinical characteristics. Follicular development was monitored by transvaginal ultrasonography using a Voluson E8 system (GE Healthcare, Chicago, IL, USA) and serum hormone measurements. Final oocyte maturation was triggered by urinary-derived human chorionic gonadotropin (hCG; 4000 IU; Livzon Pharmaceutical Group, Zhuhai, China) or recombinant hCG (Ovidrel, 250 μg; Merck Serono, Darmstadt, Germany), and oocyte retrieval was carried out 34–36 h later via transvaginal ultrasound-guided follicular aspiration.

Retrieved oocytes were cultured in commercially available fertilization and embryo culture media (G-Series; Vitrolife, Göteborg, Sweden) under sterile mineral oil (Vitrolife) at 37.0°C in a humidified atmosphere containing 6.0% CO₂ using a tri-gas incubator (K-MINC-1000; Cook Medical, Brisbane, Australia). Metaphase II (MII) oocytes were identified prior to ICSI.

Semen samples were processed by density gradient centrifugation using a two-layer gradient (PureSperm; Nidacon, Mölndal, Sweden), followed by a swim-up procedure in sperm preparation medium (G-MOPS; Vitrolife) to enrich motile spermatozoa.

For ICSI, spermatozoa were selected under high magnification using an inverted microscope equipped with micromanipulators (IX73 microscope with Narishige micromanipulation system; Olympus, Tokyo, Japan) by experienced embryologists according to the routine clinical embryology practice and the 6^th edition of the WHO manual.30 Preference was given to motile spermatozoa exhibiting a morphologically normal head, an intact acrosomal region, and a visually continuous flagellum, despite overall severe flagellar abnormalities. Sperm selection was performed immediately prior to injection to ensure optimal viability.

Fertilization was assessed 16–18 h after injection by the presence of two pronuclei and two polar bodies. Embryos were cultured to the blastocyst stage (day 5) and graded according to established morphological criteria.35 A freeze-all strategy was applied, and viable blastocysts were cryopreserved by vitrification using a commercial vitrification system (Cryotop; Kitazato Biopharma, Shizuoka, Japan). Frozen embryo transfer (FET) was subsequently performed in a prepared cycle with the transfer of a single thawed blastocyst using a soft embryo transfer catheter (Cook Medical).

Statistical analyses

Statistical analysis was performed using GraphPad PRISM 8.0 (GraphPad Software, San Diego, CA, USA). For statistical comparisons of the probands to the control group, the entire dataset was analyzed via Mann–Whitney U test (two groups) or Brown–Forsythe analysis of variance (ANOVA) test (three or more groups). P < 0.05 was considered statistically significant.

RESULTS

Identification of FSIP2 variants in male infertility

In the set of genetic investigation screening by WES of infertile males (3869 cases), 92 (2.4%, 92/3869) cases exhibited MMAF. Among the MMAF infertility cohort, we identified biallelic FSIP2 variants in three probands (3.3%, 3/92): one homozygous variant in a proband from a consanguineous family (Proband 1) and two pairs of compound heterozygous variants in two unrelated probands (Probands 2 and 3). Additionally, 23 heterozygous FSIP2 variants were identified in 29 unrelated non-MMAF infertile individuals, accounting for 0.8% (29/3777) of the non-MMAF cohort (Supplementary Table 3). Following the identification of biallelic FSIP2 variants, comprehensive analysis of WES data systematically excluded pathogenic variants in other known MMAF-associated genes, confirming FSIP2 as the sole candidate causative gene. Sanger and third-generation sequencing validated the segregation patterns in the three representative families. Proband 1 from consanguineous family carried a homozygous c.1581delG: p.Leu527Phefs*61 variant in FSIP2 (Figure 1a and 1b). The parents of Proband 1 declined to participate in genetic testing for personal reasons. Thus, we failed to obtain DNA samples from the patient’s parents or perform genotyping. Proband 2 from a non-consanguineous family carried compound heterozygous c.2409dupA: p.Pro804Thrfs*15 and c.13555C>T: p.Gln4519* variants in FSIP2 (Figure 1c and 1d). Although the parental DNA of Proband 2 was also unavailable, third-generation sequencing indicated that these alleles mainly resided on different chromosomes (Figure 1e). Proband 3 from a non-consanguineous family carried compound heterozygous c.1619dup: p.Asn540Lysfs*2 and c.10517T>C: p.Val3506Ala variants in FSIP2, inherited from unaffected heterozygous parents (Figure 1f and 1g).

Supplementary Table 3.

Overview of the disease causing fibrous sheath-interacting protein 2 variants in 3869 infertile male cohort

Variants number	Patients number	cDNA change	Protein change	Exon/Intron	Mutation type	Mutation zygosity	Allele frequency		In silico bioinformatics prediction				Novel or known
							gnomAD	gnomAD-EAS	PolyPhen2	REVEL	Mutation taster	CADD
1	P1	c.1581delG	P.Leu527Phefs*61	Exon 16	Frameshift	Homozygous	ND	ND	NA	NA	Disease causing	NA	Novel
2	P2	c.2409dupA	p.Pro804Thrfs*15	Exon 17	Frameshift	Compound heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
3	P2	c.13555C>T	p.Gln4519*	Exon 16	Stopgain	Compound heterozygous	ND	ND	NA	NA	Disease causing	Damaging	Novel
4	P3	c.1619dup	p.Asn540Lysfs*2	Exon 16	Frameshift	compound heterozygous	0.000007332	0.0001	NA	NA	Disease causing	NA	Novel
5	P3	c.10517T>C	p.Val3506	Exon 17	Missense	Compound heterozygous	ND	ND	NA	Tolerable	Polymorphism	Damaging	Novel
6	P4 and 5	c.5125C>T	p.Arg1709*	Exon 16	Stopgain	Heterozygous	0.000007323	0.00009549	NA	NA	Disease causing	Damaging	Novel
7	P6 and 7	c.18595_18598del	p.Arg6201Glnfs*4	Exon 17	Frameshift	Heterozygous	0.00001714	0.00005878	NA	NA	Disease causing	NA	Novel
8	P8–11	c.17323C>T	p.Arg5775*	Exon 17	Stopgain	Heterozygous	0.000008311	0.000	NA	NA	Disease causing	Damaging	Known
9	P12 and 13	c.16747_16749delinsA	p.Tyr5583Asnfs*8	Exon 17	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
10	P14	c.6306T>G	p.Tyr2102*	Exon 16	Stopgain	Heterozygous	ND	ND	NA	NA	Disease causing	Damaging	Novel
11	P15	c.4729A>T	p.Lys1577*	Exon 16	Stopgain	Heterozygous	ND	ND	NA	NA	Disease causing	Damaging	Novel
12	P16	c.10133delA	p.Ser3380Valfs*34	Exon 16	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
13	P17	c.11304C>A	p.Cys3768*	Exon 17	Stopgain	Heterozygous	ND	ND	NA	NA	Disease causing	Damaging	Novel
14	P18	c.16068dupT	p.Glu5357*	Exon 17	Stopgain	Heterozygous	ND	ND	NA	NA	Disease causing	Damaging	Novel
15	P19	c.10240dupA	p.Glu3417Glyfs*12	Exon 16	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
16	P20	c.278delT	p.Leu93Trpfs*13	Exon 3	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
17	P21	c.11601delT	p.Phe3868Leufs*6	Exon 17	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
18	P22	c.18448G>A	p.Val6150Ile	Exon 17	Missense	Heterozygous	0.00007243	0.0004111	NA	Tolerable	Polymorphism	Tolerable	Known
19	P23	c.11689delC	p.Gln3897Argfs*6	Exon 17	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
20	P24	c.20473+2T>C	NA	Intron 20	Splice-site loss	Heterozygous	0.000007718	0.00005514	NA	NA	Disease causing	NA	Novel
21	P25	c.4905_4906delTC	p.Leu1636Valfs*39	Exon 16	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
22	P26	c.12695dupA	p.Ser4234Lysfs*11	Exon 17	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
23	P27	c.11779C>T	p.Gln3927*	Exon 17	Stopgain	Heterozygous	ND	ND	NA	NA	Disease causing	Damaging	Novel
24	P28	c.8003C>A	p.Ser2668*	Exon 16	Stopgain	Heterozygous	0.0000238	0.000	NA	NA	Disease causing	Damaging	Known
25	P29	c.14438_14442dup	p.Asp4816Serfs*9	Exon 17	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
26	P30	c.3111dupA	p.Asn1040Lysfs*5	Exon 16	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
27	P31	c.7487delA	p.Arg2497Glufs*2	Exon 16	Frameshift	Heterozygous	ND	ND	NA	NA	Disease causing	NA	Novel
28	P32	c.15051C>G	p.Tyr5017*	Exon 17	Stopgain	Heterozygous	0.000009257	0.000	NA	NA	Disease causing	Damaging	Novel

The transcript used in this study was NM_173651.4. gnomAD: Genome Aggregation Database; EAS: East Asian; PolyPhen2: polymorphism phenotyping v2; REVEL: rare exome variant ensemble learner; CADD: combined annotation dependent depletion; ND: not found; NA: not available; FSIP2: fibrous sheath-interacting protein 2

Figure 1.

Identification of novel biallelic FSIP2 variants in males with MMAF. (a) Pedigree of P1 with FSIP2 M1. Filled symbols denote affected males; arrows indicate P1. (b) Sanger sequencing chromatograms validate FSIP2 M1. (c) Pedigree of P2 with FSIP2 M2 and M3. Filled symbols denote affected males; arrows indicate P2. (d) Sanger sequencing chromatograms validate FSIP2 M2 and M3. (e) Third-generation sequencing validated that M2 and M3 mainly resided on opposite alleles. (f) Pedigree of P2 with FSIP2 M4 and M5. (g) Sanger sequencing chromatograms validate FSIP2 M4 and M5 in P3 and his parents. Pro: proband; P1: Proband 1; P2: Proband 2; P3: Proband 3; M: mutation; M1: mutation 1; M2: mutation 2; M3: mutation 3; M4: mutation 4; M5: mutation 5; WT: wild-type; Ref: reference; FSIP2: fibrous sheath-interacting protein 2; MMAF: multiple morphological abnormalities of the sperm flagella; K: lysine; P: proline; S: serine; Q: glutamine; H: histidine.

Structural consequences of FSIP2 variants

We identified three frameshift variants (c.1581delG: p.Leu527Phefs*61; c.2409dupA: p.Pro804Thrfs*15; and c.1619dup: p.Asn540Lysfs*2) and one nonsense variant (c.13555C>T: p.Gln4519*), all of which mapped to exons 16–17 of FSIP2 (Supplementary Figure 1a ^{(475KB, tif)} and 1b). These truncating variants resulted in premature termination of the FSIP2 protein, leading to complete or partial loss of all four conserved FSIP2 domains. The remaining missense variant (c.10517T>C: p.Val3506Ala), located in a phylogenetically conserved residue (Supplementary Figure 1c ^{(475KB, tif)} ), was classified as damaging by CADD. AlphaFold3 structural modeling revealed that these variants cause significant conformational disruptions. The truncating variants (Leu527Phefs*61, Pro804Thrfs*15, Asn540Lysfs*2, and Gln4519*) lead to premature termination, which abolishes the C-terminal FSIP2 domains critical for FS assembly (Supplementary Figure 1d ^{(475KB, tif)} ). Meanwhile, the Val3506Ala substitution increases the inter-residue distance between Val³⁵⁰⁶ and Leu³⁵¹⁰ (2.9 Å to 3.0 Å), thereby destabilizing the local α-helix (Supplementary Figure 1d ^{(475KB, tif)} ). These findings underscore the significant structural consequences of FSIP2 variants.

According to the ACMG/AMP guideline32 and a Bayesian point-based interpretation,33 the missense variant c.10517T>C: p.Val3506Ala in FSIP2 was classified as pathogenic. Functional evidence confirmed a damaging effect on the gene product (pathogenic strong 3 [PS3]), while the variant’s absence from population databases further supported its pathogenicity (pathogenic moderate 2 [PM2]). Genetic and clinical evidence included the variant being detected in trans with a known pathogenic truncating variant (pathogenic moderate 3 [PM3]), a highly specific phenotype (pathogenic supporting 4 [PP4]), and evolutionary conservation (pathogenic supporting 3 [PP3]).

In addition, four truncating variants, c.1581delG: p.Leu527Phefs61, c.2409dupA: p.Pro804Thrfs15, c.13555C>T: p.Gln4519, and c.1619dup: p.Asn540Lysfs2, were identified and classified as pathogenic according to Bayesian point-based interpretation of the ACMG/AMP guidelines.32,33 The classification was primarily supported by the predicted loss-of-function (LOF) mechanism (pathogenic very strong 1 [PVS1]), their rarity in population databases (PM2), and the phenotypic consistency with autosomal recessive MMAF (PP4).

FSIP2 variants disrupt sperm flagellar integrity and motility

Although ultrastructural analyses, including IF, SEM, and TEM, were not available for Proband 1 due to limited semen samples, indirect phenotypic evidence supports a consistent genotype–phenotype correlation. Comparative analysis of routine semen parameters revealed highly concordant clinical characteristics between Proband 1 and Proband 2, including normal sperm concentration, severely reduced progressive motility, and a predominance of short flagella under light microscopy, which are hallmarks of MMAF (Supplementary Table 4). Given the shared genetic architecture and highly concordant clinical manifestations, Proband 2 serves as a representative model for the ultrastructural consequences of biallelic LOF FSIP2 variants.

Supplementary Table 4.

Semen characteristics of the patient carrying fibrous sheath-interacting protein 2 variants in three times analyses

Semen parameters	P1	P2	P3	Control	P	Reference
Semen volume (ml)	4.8	3.2±0.70	2.15±0.2	3.9±1.5	0.21	≥1.5
Sperm concentration (106/ml)	31.4	81.4±1.1	99.9±10.2	70.1±32.8	0.41	≥15
Total sperm count (106/ejaculate)	150.9	261.0±61.0	213.6±0.9	262.5±142.3	0.68	≥39
Total motile sperm count (106/ejaculate)	22	8.7±3.6	11.4±2.0	140.4±82.5	<0.01	-
Grade A sperm (%)	4	0.0±0.0	0.5±0.7	13.6±7.6	<0.01	-
Grade B sperm (%)	11	3.5±2.1	5.0±1.1	38.8±7.4	<0.01	-
Grade C sperm (%)	7	1.5±0.7	7.0±4.2	4.1±2.0	0.86	-
Grade D sperm (%)	78	95.0±1.4	87.5±3.5	43.5±8.4	<0.01	-
Progressive motility (A + B%)	15	3.5±2.1	5.5±0.7	52.5±8.0	<0.01	≥32
Vitality (%)	59	60.5±3.5	63.5±2.1	66.9±5.9	0.06	≥58
Sperm flagella morphology
Normal flagella (%)	25.6	0.3±0.5	9.1±1.3	92.3±2.1	<0.01	-
Absent flagella (%)	1.0	20.0±3.0	0.6±0.9	0.0±0.0	0.10	-
Short flagella (%)	59.4	73.0±6.3	50.3±5.4	0.7±0.5	<0.01	-
Coiled flagella (%)	12.0	3.2±0.9	35.5±4.5	5.3±1.2	<0.01	-
Mid-piece morphology
Asymmetrical (%)	14.5	0.3±0.5	7.6±2.7	2.67±0.6	0.78	-
Lengthened and thickened (%)	77.3	75.6±2.2	80.4±3.4	8.7±3.9	<0.01	-
ERC (%)	2.4	3.2±0.9	4.4±4.2	1.4±0.5	0.22	-
Head morphology
Normal head (%)	39.1	34.8±3.7	42.4±1.3	41.3±4.6	0.38	-
Non-oval shape (%)	44.4	36.9±7.6	42.5±4.5	38.3±4.1	0.52	-
Vacuolated (%)	15.9	31.2±4.2	33.6±4.2	20.7±1.7	0.11	-
Abnormal acrosome (%)	22.7	8.3±1.5	25.9±4.7	9.0±0.8	0.38	-
Sperm locomotion parameters
VCL (µm/s)	8.3	1.3±0.6	2.8±1.0	21.9±6.9	<0.01	-
VSL (µm/s)	4.5	0.7±0.3	1.2±0.5	10.6±3.2	<0.01	-
VAP (µm/s)	5.4	0.9±0.4	1.7±0.6	14.4±4.2	<0.01	-
MAD (°)	11.3	1.7±0.0	5.2±3.0	25.1±4.4	<0.01	-
ALH displacement (µm)	0.9	0.2±0.1	0.3±0.1	1.7±0.5	<0.01	-
BCF (Hz)	2.6	0.4±0.2	1.2±0.6	5.8±1.5	<0.01	-
LIN (%)	11.4	2.3±0.4	3.6±2.0	20.6±5.7	<0.01	-
WOB (%)	13.5	2.4±0.6	4.5±2.2	27.2±6.6	<0.01	-
STR (%)	16.4	3.2±0.3	5.6±3.5	30.3±7.9	<0.01	-

Routine semen parameters were evaluated according to the WHO-2010 guidelines, sperm morphology parameters were evaluated according to the WHO-2021 guidelines. : no reference range; P1: proband 1; P2: proband 2; P3: proband 3; WHO: World Health Organization; LIN: linearity; WOB: wobble; STR: straightness; BCF: beat-cross frequency; ALH: amplitude of lateral head; MAD: mean angular displacement; VCL: curvilinear velocity; VSL: straight-line velocity; VAP: average path velocity

IF revealed FSIP2 localization in the acrosome and along the entire flagella in control sperm, whereas complete absence of FSIP2 was observed in the acrosome and flagella region of spermatozoa from Proband 2 which had compound heterozygous LOF variants (Figure 2a). In Proband 3 with compound heterozygous frameshift/missense variants, staining could not be detected in the acrosome and tail region but remained mainly in the upper portion of the principal piece of the sperm flagella, although it was significantly weaker than that in the control (Figure 2a). A semen analysis revealed normal sperm counts (P = 0.68), but severely impaired motility in all three probands (P < 0.01; Supplementary Table 4, and Supplementary Video 1 and 2). The kinematic parameters, including straight-line velocity (VSL), curve line velocity (VCL), amplitude of lateral head displacement (ALH), average pathway velocity (VAP), beat cross frequency (BCF), mean angular displacement (MAD), linearity, straightness, and wobble, were also markedly reduced compared with those of controls (all P < 0.01; Supplementary Table 4), indicating global flagellar dysfunction. The 29 infertile individuals carrying heterozygous FSIP2 variants exhibited semen parameters comparable to controls (Supplementary Table 1 and 3), confirming that heterozygosity does not cause the MMAF phenotype.

Figure 2.

Sperm morphology and ultrastructural defects in FSIP2-mutant sperm. (a) Immunofluorescence staining of FSIP2 (green) in control versus FSIP2-mutant sperm. Nuclei were counterstained with DAPI (blue). (b) Papanicolaou-stained sperm showing coiled/short flagella and elongated midpieces in FSIP2-mutant sperm. (c) Scanning electron microscopy (SEM) and quantitative analysis revealing flagellar truncation and midpiece elongation in FSIP2-mutant sperm. ^**P<0.01. (d) Longitudinal TEM sections of sperm flagella. FSIP2-mutant sperm exhibit abnormally elongated mitochondrial sheaths (brackets). Blue triangle: normal mitochondrial; red triangle: hypertrophic mitochondria. (e) Transverse-sectional TEM image of the ultrastructure of the sperm flagella. ^**P < 0.01 vs control. P2: Proband 2; P3: Proband 3; FSIP2: fibrous sheath-interacting protein 2; SEM: scanning electron microscopy; TEM: transmission electron microscopy; IDA: inner dynein arm; ODA: outer dynein arm; CP: central pair microtubules; FS: fibrous sheath; MS: mitochondrial sheath; LC: longitudinal columns; CR: circumferential ribs; ODF: outer dense fibers; RS: radial spoke; DMT: peripheral doublet microtubules; DAPI: 4’,6-diamidino-2-phenylindole.

Papanicolaou staining and SEM revealed short or thin flagella with elongated mitochondrial sheaths in the midpiece of nearly 75.6%–80.4% of the spermatozoa from Probands 2 and 3 (Figure 2b and 2c, and Supplementary Table 4). Spermatozoa longitudinal sections were observed by TEM. Longitudinal TEM revealed profound abnormalities in FSIP2-mutant spermatozoa flagella. In the midpiece of spermatozoa from Probands 2 and 3, the mitochondrial sheaths exhibited pathological elongation, without evidence of fibrous sheath extension into this region (Figure 2d). Spermatozoa transverse sections were observed by TEM. Transverse-sectional TEM revealed that individual mitochondria displayed hypertrophic morphology (Figure 2e). The principal piece showed complete absence of the FS in nearly 70.0% of the spermatozoa from Proband 2 (compound LOF variants), leaving nine ODFs exposed, which was contrasted with the normal seven ODFs flanked by LCs and CRs of FS in the controls (Figure 2e). Nearly 50.0% of the spermatozoa from Proband 3 (frameshift/missense variants) exhibited dysplastic FS with fragmented LCs/CRs and disorganized ODFs (Figure 2e). Notably, nearly 50.0% of the principal and end piece cross-sections in Proband 2 lacked CP. In the end piece, both Probands 2 and 3 showed mild ultrastructural abnormalities, mainly manifested as disorganization and sporadic discontinuities of axonemal doublet microtubules (Figure 2e). These ultrastructural anomalies, spanning mitochondrial sheath disorganization, FS defects, and axonemal component loss, mechanistically explain the observed MMAF phenotypes.

FSIP2 regulates flagellar protein expression and mitochondrial function

WB and IF analyses revealed the severe disruption of flagellar ultrastructural components in FSIP2-mutant sperm. FS integrity was profoundly compromised, as evidenced by complete loss of AKAP4 (FS marker)36 in Proband 2 and significantly reduced AKAP4 signaling in Proband 3 (Figure 3a and 3b, and Supplementary Figure 2a ^{(113.9KB, tif)} and 2b). More specifically, double-label IF staining for FSIP2 and AKAP4 indicated that FSIP2 and AKAP4 were colocalized in the principal piece in control sperm flagella. Furthermore, AKAP4 signals were confined to the principal piece and were absent from the midpiece in spermatozoa from individuals harboring biallelic FSIP2 variants (Figure 3a and 3b, and Supplementary Figure 2b ^{(113.9KB, tif)} ); these findings indicate that fibrous sheath structures did not extend into the midpiece in FSIP2-mutant spermatozoa. However, in the acrosome, MS region, and the distal end piece of the sperm flagella, FSIP2 was expressed, whereas AKAP4 was absent (Supplementary Figure 2b ^{(113.9KB, tif)} ). In the sperm from Proband 3, the expression levels of both FSIP2 and AKAP4 were significantly reduced, but the colocalization pattern of FSIP2 and AKAP4 remained consistent with that observed in control sperm (Supplementary Figure 2b ^{(113.9KB, tif)} ). Conversely, although the expression level of the mitochondrial marker TOM20 remained unchanged, its localization was abnormally elongated (Figure 3a and 3b, and Supplementary Figure 2a ^{(113.9KB, tif)} ). In addition, the signals of the CP marker SPAG6,36 the ODA marker DNAH17,37 the inner dynein arm (IDA) marker DNALI1,38 and the intraflagellar transport (IFT) marker IFT2039 were also decreased in the sperm flagella of Probands 2 and 3 compared with control sperm (Figure 3a and 3b, and Supplementary Figure 2a ^{(113.9KB, tif)} ). Interestingly, although recent publication demonstrated that acrosomal markers actin-like 7A (ACTL7A) and PNA were present in fertile controls but absent in sperm from FSIP2-affected patients,27 we detected the acrosome marker PNA signals in the majority of the sperm from Probands 2 and 3, suggesting the presence of intact acrosomal structures in these patients (Supplementary Figure 2c ^{(113.9KB, tif)} ). These findings highlight genotype-dependent variability in axonemal protein assembly, which is correlated with observed differences in sperm motility and ultrastructural defects among probands.

Figure 3.

Disrupted expression of flagellar structural proteins in FSIP2-mutant spermatozoa. (a) Immunolocalization of key flagellar proteins (AKAP4, TOM20, SPAG6, DNAH17, DNALI1, and IFT20) in control sperm. (b) FSIP2-mutant sperm exhibit complete loss of the fibrous sheath marker AKAP4 (Proband 2) or severely reduced AKAP4 expression (Proband 3), along with abnormally elongated mitochondrial marker TOM20 signals in the mitochondrial sheath, and downregulated axonemal protein (SPAG6, DNAH17, DNALI1, and IFT20) levels. FSIP2: fibrous sheath-interacting protein 2; P2: Proband 2; P3: Proband 3; AKAP4: A-kinase anchoring protein 4; TOM20: translocase of outer mitochondrial membrane 20; SPAG6: sperm-associated antigen 6; DNAH17: dynein axonemal heavy chain 17; DNALI1: dynein axonemal light intermediate chain 1; IFT20: intraflagellar transport 20; β-TUB: β-tubulin; DAPI: 4’,6-diamidino-2-phenylindole.

ICSI outcomes in FSIP2-associated infertility

Following genetic counseling, we confirmed the absence of FSIP2 variants in the spouses of Probands 1 and 2. Sperm with progressive motility and normal head morphology were then isolated using swim-up and density gradient centrifugation. These selected sperm were subsequently used for ICSI treatment. After transferable blastocysts were obtained, we applied a freeze-all approach and cryopreserved all the blastocysts. The single blastocyst with the highest score was transferred on day 5 in a natural cycle. For Proband 1, 19 MII oocytes were retrieved from his wife. Following ICSI, 17 oocytes were successfully fertilized, resulting in 16 cleavage-stage embryos on day 3 and 8 blastocysts on day 5. Frozen embryo transfer of a single 4AA-grade blastocyst resulted in a full-term healthy male delivery. For Proband 2, 5 MII oocytes were retrieved from his wife. Following ICSI, 4 oocytes were successfully fertilized and developed into 4 cleavage-stage embryos on day 3, ultimately resulting in one blastocyst on day 5. Frozen embryo transfer of the 4BB-graded blastocyst similarly led to the birth of a healthy term male infant (Figure 4a and 4b, and Supplementary Table 5). These outcomes demonstrate that ICSI effectively circumvents FSIP2-related flagellar dysfunction, resulting in favorable embryological parameters (fertilization rate: 80.0%–89.5%; blastocyst rate: 25.0%–50.0%) and live births despite severe sperm ultrastructural defects. Although we identified the genetic cause of the sperm motility defects and morphological abnormalities in Proband 3, the ICSI procedure was delayed because his wife tested positive for tuberculosis during the assisted reproduction process.

Figure 4.

Reproductive outcomes and mechanistic model of FSIP2-associated infertility. (a) Clinical ART outcomes using ICSI in FSIP2-affected couples. (b) Images of in vitro early embryonic development from FSIP2-affected couples. (c) Schematic of WT and FSIP2-mutant spermatozoa flagella. A cross-section of WT sperm displays the typical axoneme structure with a normal MS and intact FS. FSIP2-mutant spermatozoa exhibit elongated and hypertrophic MS in the midpiece, FS defects and axonemal component loss in the principal and end piece of the flagella. FSIP2: fibrous sheath-interacting protein 2; P1: Proband 1; P2: Proband 2; ART: assisted reproductive technology; ICSI: intracytoplasmic sperm injection; MII: metaphase II; PN: pronuclei; WT: wild type; FS: fibrous sheath; MS: mitochondrial sheath; CP: central pair microtubules; LC: longitudinal columns.

Supplementary Table 5.

The clinical outcomes of assisted reproductive technology with intracytoplasmic sperm injection in the patients affected by fibrous sheath-interacting protein 2 variants

Subject	P1	P2	Reference/average value
Female age (years)	39	29	-
BMI	23.9	25.6	18.5–24.0
FSH (mIU/ml)	4.5	6.6	≤10.0#
LH (mIU/ml)	4.6	2.4	≤12.2#
E2 (pg/ml)	62.7	27.5	15.2–127.8#
Prog (ng/ml)	0.8	0.81	≤1.5#
Protocol	Antagonist protocol	Long GnRH Agonist protocol	-
E2 level on the trigger day (pg/ml)	4141.0	3759.7	-
Oocyte trigger	Chorionic gonadotropin, 4000U	Recombinant human chorionic gonadotropin α, 250 µg
Number of follicles ≥14 mm on the trigger day	15	11	-
Number of follicles ≥18 mm on the trigger day	7	5	-
Number of oocytes retrieved	20	6	-
Number of MII oocytes (%)	19 (95.0)	5 (83.3)	-
Number of 2PN (%)	17 (89.5)	4 (80.0)	-
Number of usable embryos (%)	16 (94.1)	4 (100.0)	-
Number of blastocysts (%)	8 (50.0)	1 (25.0)	-
Transfer cycles	1 (FET)	1 (FET)	-
Transferred blastocysts	1 (4AA)	1 (4BB)	-
Number of gestational sacs	1	1	-
Number of clinical pregnancies (%)	1 (100)	1 (100)	-
Number of live birth delivery (%)	1 (100)	1 (100)
Number of neonatal birth (%)	1 (100)	1 (100)

^#These data refer to the value in follicular phase. : no reference range; PN: pronucleus; MII: metaphaseII; 2PN: two pronuclei; FSH: follicle-stimulating hormone; LH: luteinizing hormone; E2: estrogen; Prog: progestogen; FET: frozen embryo transfer; P1: proband 1; P2: proband 2; ICSI: intracytoplasmic sperm injection; GnRH: gonadotropin-releasing hormone; BMI: body mass index

DISCUSSION

Our study identified one homozygous variant in a proband from a consanguineous family and two pairs of compound heterozygous variants in two unrelated, non-consanguineous families within a cohort of 92 MMAF patients. WES, Sanger, and third-generation sequencing results revealed that these probands harbored biallelic FSIP2 variants and lacked other known pathogenic gene variants associated with MMAF. Using Papanicolaou staining and SEM, we observed a significant shortening of the length of the sperm flagella and elongated mitochondrial sheaths. Furthermore, elongated mitochondrial sheaths, FS defects, and axonemal component loss in the sperm flagella were further validated by TEM. Preliminary mechanistic exploration revealed that these novel biallelic FSIP2 variants resulted in the dysregulation of established flagella ultrastructure regulators. Collectively, our findings implicate biallelic FSIP2 variants as a causative genetic factor driving the MMAF phenotype and asthenozoospermia.

FSIP2 plays a critical role in sperm flagellar development,17,19 contributing to the proper formation and motility of the sperm tail. Pathogenic variants in FSIP2 have been shown to cause a range of sperm flagellar defects, including bent, short, coiled, or absent flagella, which are hallmark features of MMAF.14,15,19 These structural abnormalities are often accompanied by disorganization of the FS,17,22 a specialized structure surrounding the axoneme that is essential for flagellar stability and function.9 However, recent studies have revealed that different FSIP2 variants lead to diverse sperm phenotypes, highlighting the genetic heterogeneity of this gene and its complex role in spermatogenesis. In this study, we identified 28 distinct pathogenic FSIP2 variants, predominantly LOF mutations, in 32 infertile males. Pedigree analysis confirmed that two patients carried compound heterozygous variants, one carried a homozygous variant, and the remaining 29 patients were heterozygous carriers. To date, 55 FSIP2 variants have been reported in 35 patients, including 12 homozygous and 23 verified or unverified compound heterozygous cases (Supplementary Table 6).14,15,16,17,18,19,20,21,22,23,24,25,26,27,28 Among patients carrying two different variants, 12 were confirmed to have compound heterozygous variants,14,16,17,19,20,24,27 whereas 11 lacked pedigree verification,21,25,26,28 making their allelic configuration uncertain. Notably, three patients with two FSIP2 variants exhibited near-normal sperm motility and morphology despite >65.0% flagellar defects typical of MMAF,28 although detailed morphological staining results were not provided. Family cosegregation analysis is essential for evaluating pathogenic variants, particularly for autosomal recessive genes such as FSIP2. Without such analysis, the pathogenicity of variants cannot be rigorously confirmed, potentially leading to misdiagnosis. Therefore, genetic studies on MMAF should prioritize pedigree verification to ensure accurate interpretation of inheritance patterns.

Supplementary Table 6.

Compound heterozygous or homozygous variants of fibrous sheath-interacting protein 2 identified in 35 patients with multiple morphological abnormalities of the sperm flagella in previous literatures

Patient number	Variant number	cDNA variant	Amino acid variant	Exon	Effect	Phenotype	Clinical pregnancy	Allelic status	Origin	Reference
1	1 2	c.1494C>A c.11020_11024del	p.Cys498* p.Tyr3675Cysfs*3	15 17	Stopgain Frameshift	Asthenoteratozoospermia	Success	Compound heterozygous	Chinese	1
2	3	c.8368_8369insC	p.Leu2790Serfs*22	16	Frameshift	Asthenoteratozoospermia	Not reported	Homozygous	Chinese	2
3	4	c.910delC	p.Gln304Leufs*13	8	Frameshift	Oligoasthenoteratozoospermia	Not reported	Homozygous	Tunisians	3
4	5	c.16389_16392delAATA	p.Glu5463Glufs*7	17	Frameshift	Oligoasthenoteratozoospermia	Not reported	Homozygous
5	6	c.2282dupA	p.Asn761Ilefs*4	16	Frameshift	Oligoasthenoteratozoospermia	Not reported	Homozygous
6	7	c.1606_1607insTGT; 1607_1616delAAAGATTGCA	p.Lys536Metfs*1	16	Frameshift	Asthenoteratozoospermia	Not reported	Homozygous
7	8 9	c.8104dup c.4574C>A	p.Leu2702Profs*44 p.Ser1525*	16	Frameshift Stopgain	Asthenoteratozoospermia	Success	Compound heterozygous	Chinese	4
8	10 11	c.8038dup	p.Thr2680gAsnfs*9	16	Frameshift	Asthenoteratozoospermia	Success	Compound heterozygous
		c.9234dup	p.Leu3079Thrfs*6		Frameshift
9	12 13	c.10247dup c.9043G>A	p.Glu3417Glyfs*12 p.Glu3015Lys	16	Frameshift Missense	Asthenospermia	Not reported	Compound heterozygous
10	14	c. 1907C>A	p.Ser636*	16	Stopgain	Asthenoteratozoospermia	Success	Homozygous	Chinese	5
11	15	c.8030_8031insA	p.Thr2680Asnfs*9	16	Frameshift	Asthenoteratozoospermia	Not reported	Homozygous
12	16	c.C19981T	p.Arg6661*	18	Stopgain	Asthenoteratozoospermia	Not reported	Homozygous	Chinese	6
13	17 18	c.G18448A c.5238-5240del	p.Val6150Ile p.Glu1746del	17 16	Missense Deletion	Asthenoteratozoospermia	Not reported	Compound heterozygous
14	19 20	c.A5480T c.T9056C	p.Asp1827Val p.Ile3019Thr	16 16	Missense Missense	Asthenoteratozoospermia	Not reported	Compound heterozygous
15	21 22	c.T10823C c.C17798T	p.Ser5933Phe p.Asp1827Val	17 17	Missense Missense	Asthenoteratozoospermia	Not reported	Compound heterozygous
16	23	c.1750T>A	p.Cys584Ser	16	Missense	Asthenoteratozoospermia	Success	Compound heterozygous	Chinese	7
17	24	c.13600A>G	p.Ile4534Val	17	Missense
18	25	c.16246_16247insCCCAAATATCACC	p.Thr5416Alafs*7	17	Frameshift	Asthenoteratozoospermia	Failure	Compound heterozygous	Chinese	8
	26	c.17323C>T	p.Gln5774*	17	Stopgain
19	1 27	c.1494C>A c.11280_11284delTAAGT	p.Cys498* p.Ile3761fs	15 17	Stopgain Frameshift	Asthenoteratozoospermia	Not reported	Compound heterozygous	Chinese	9
20	28 29	c.17G>A c.3365C>G	p.Gly6Asp p. Ser1122Cys	1 16	Missense Missense	Oliogasthenozoospermia	Not reported	Compound heterozygous	Chinese	10
21	30	c.1347 + 1G>A			Splice-site loss	Oliogasthenozoospermia	Not reported	Compound heterozygous
	31	c.5235del	p.Glu1747del	16	Deletion
22	32	c.2508delT	p.Ser836fs	16	Frameshift	Asthenoteratozoospermia	Not reported	Homozygous	Chinese	11
23	33 34	c.C19148T c.T17084C	p.Ser6383Leu p.Ile5695Thr	17 17	Missense Missense	Oligoasthenoteratozoospermia	Not reported	Compound heterozygous	Indian	12
24	35 36	c.8909C>T c.11171G>T	p.Pro2970Leu p.Arg3724Ile	16 17	Missense Missense	Oligoasthenozoospermia	Failure	Compound heterozygous	Chinese	13
25	37	c.9234dupA	p.Leu3079Thrfs*6	16	Frameshift	Asthenozoospermia	Success	Homozygous
26	38 39	c.5539G>T c.19925G>C	p.Val1847Phe p.Arg6642Pro	16 17	Missense Missense	Normal	Gemellary pregnancy	Compound heterozygous
27	40 41	c.302A>G c.14107delA	p.His101Arg p.Ser4703Alafs*18	3 17	Missense Frameshift	Asthenozoospermia	Gemellary pregnancy	Compound heterozygous
28	42 43	c.232C>T c.12329C>A	p.Arg78* p.Pro4110His	3 17	Stopgain Missense	Oligoasthenoteratozoospermia	Success	Compound heterozygous
29	44 45	c.16346C>T c.20146A>C	p.Thr5449Ile p.S6716R	17 17	Missense Missense	Oligoasthenozoospermia	Success	Compound heterozygous
30	46 47	c.3351G>A c.4345G>A	p.Met1117Ile p.Gly1449Arg	16 16	Missense Missense	Oligoasthenozoospermia	Success	Compound heterozygous
31	48 49	c.488C>G c.10726G>C	p.Ala163Gly p.Ala3576Pro	5 17	Missense Missense	Oligoasthenoteratozoospermia	Success	Compound heterozygous
32	50	c.2137C>T	p.Arg713*	16	Stopgain	Asthenozoospermia	Success	Homozygous
33	51 52	c.272_275delinsAGGTTTTTAA c.16788_16791del	p.Leu92Valfs*74 p.Ser5597Glyfs*6	3 17	Frameshift Frameshift	Asthenoteratozoospermia	Not reported	Compound heterozygous	Chinese	14
34	53 54	c.262C>A c.10948_10951del	p.Pro88Thr p.Asn3653Asnfs*22	3 17	Missense Frameshift	Oligoasthenoteratozoospermia	Not reported	Compound heterozygous	Pakistani	15
35	55	c.15982_15982del	p.Ile5328Leufa*333	17	Frameshift	Oligoasthenoteratozoospermia	Not reported	Homozygous

MMAF: multiple morphological abnormalities of the sperm flagella; Del: deletion; Ins: insertion; Delins: deletion and insertion

The localization of the FSIP2 protein in sperm remains unclear across studies. While four studies detected FSIP2 exclusively in the flagella,14,17,18,20 five reported its presence in both the flagella and the sperm head.19,22,24,27,28 Variants also differentially affect protein expression: compound heterozygous LOF variants often result in undetectable FSIP2 in flagella,14,17,24,27 whereas missense variants show variable expression patterns.19,20,28 Interestingly, some studies had reported that FSIP2 is absent in flagella but is present in the sperm neck of FSIP2-mutant spermatozoa, even in LOF cases.22,28 Truncating variants may undergo nonsense-mediated decay (NMD),40 but residual neck expression in that study remains unexplained. In our study, FSIP2 was detected in both the flagella and the acrosomal region of the sperm head. Patients with compound heterozygous LOF variants showed no FSIP2 expression, whereas those with compound heterozygous frameshift and missense variants exhibited weak expression throughout the sperm flagella. Interestingly, Zheng et al.19 and Yuan et al.20 reported that FSIP2 variants could cause acrosomal defects, leading to globozoospermia or other head abnormalities. However these findings contrast with our findings and those of others reporting isolated MMAF phenotypes without globozoospermia.14,17 A recent publication demonstrated that the acrosomal markers ACTL7A and PNA were present in fertile controls but absent in sperm from FSIP2-affected patients,27 we detected PNA signals in the majority of sperm from Probands 2 and 3, suggesting the presence of intact acrosomal structures in these patients. Notably, one study reported that sperm with compound heterozygous mutations in FSIP2 exhibited a high rate of two pronuclei following IVF fertilization, a process that requires intact and functionally competent acrosomes.28 These findings complicate the interpretation of the impact of FSIP2 on acrosomal function. Our study also revealed that FSIP2 and AKAP4, which are the major components of the FS, colocalize in the principal piece of the sperm flagella. This colocalization supports the previously reported interaction between these two proteins.12 However, in the acrosome, MS region, and distal end piece of the sperm flagella, FSIP2 was expressed whereas AKAP4 was absent, suggesting that FSIP2 may play a role beyond FS formation in the principal and end piece, possibly contributing to mitochondrial and acrosomal function or other structural roles.

FSIP2-related infertility primarily manifests as flagellar defects rather than impaired sperm production. Among the 38 reported patients with homozygous, verified or unverified compound heterozygous FSIP2 variants in previous studies and our current study, 65.8% (25/38) had normal sperm counts,14,15,16,17,18,19,20,21,23,24,28 whereas only 34.2% (13/38) had moderate-to-severe oligozoospermia (Supplementary Table 6 and 7).16,25,26,27,28 These findings suggest that FSIP2 dysfunction does not intrinsically disrupt spermatogenesis. However, the etiology of oligozoospermia in some patients with unverified FSIP2 heterozygous variants remains unclear because of insufficient information on endocrine and obstructive markers.25,26,27,28 The impact of FSIP2 variants on progressive motility varies markedly: 63.2% (24/38) of patients exhibited severe-to-extreme asthenozoospermia (progressive motility ≤10%),15,16,17,18,20,22,24,27,28 whereas 7.9% (3/38) maintained near-normal motility (progressive motility ≥20.0%).15,28 Notably, discrepancies between sperm viability and motility may indicate genetic flagellar dysfunction. In accordance with criteria of the 6^th edition of the WHO manual, more than 25% immotile-but-viable sperm strongly suggests this diagnosis.30 However, only 33.3% (5/15) of the relevant studies provided viability data,15,16,18,19 hindering comprehensive assessment. The abnormally high progressive motility observed in some patients could stem from unverified compound heterozygosity. Without pedigree validation, certain variants might be located on the same chromosome, leaving one functional allele to produce residual FSIP2. Such residual expression could partially maintain flagellar structure and function, especially in cases involving missense variants. However, the lack of viability data in these studies limits our ability to assess the proportion of immotile but viable sperm and to diagnose genetic flagellar dysfunction.

Supplementary Table 7.

Semen characteristics in multiple morphological abnormalities of the sperm flagella patients affected by compound heterozygous or homozygous variants of fibrous sheath-interacting protein 2 variants in previous literatures

Patient number	Volume	Concentration	Total motility	Progressive motility	Viability	Normal sperm morphology	MMAF phenotype	Reference
1	3.0	45	18	16	ND	2	Yes	1
2–1	3.5	44.9	38.2	8.6	45	4	Yes	2
2–2	3.0	143.8	10.8	1.5	48	2	Yes
2–3	2.5	47.5	8.9	1.2	45	1	Yes
3	2.5	6.7	1	NA	47	0	Yes	3
4	1.7	49.5	5	NA	52	0	Yes
5	4.7	1.2	0	NA	7	0	Yes
6	3.4	0.7	0	NA	30	0	Yes
7	1.0	47.1	6.0	3.4	ND	0	Yes	4
8	3.7	63.1	25	5.4	ND	0	Yes
9	2.1	57.3	13	4.4	ND	0	Yes
10	1.8	43.6	21.9	6.5	ND	1	Yes	5
11	3.4	43.1	17.3	9.8	ND	2	Ye
12	4.6	38.5	NA	12	74	2.5	Yes	6
13	4.0	20.3	NA	15	77	1.4	Ye
14	1.6	64.8	NA	10	69	0.5	Yes
15	5.2	44.7		21	70	0	Yes
16	2.3	20.0	5.9	3.7	ND	1	Yes	7
17	2.8	16.0	7.1	6.1	ND	1	Yes
18	3.4±0.5	22.0±3.0	6.5±1.3	NA	48.7±5.5	1.0±0.2	Yes	8
19	NA	2.9	1.7	1.69	ND	4	Yes	9
20	NA	108.3	2.1	0.52	ND	5	Yes
21	NA	30.7	38.7	33.5	ND	4	Yes
22	NA	32.4	29.1	24.3	ND	6	Yes
23	NA	9.8	0	0	ND	3	Yes
24	NA	7.8	28.9	27.7	ND	8	Yes
25	NA	1.5	0	0	ND	6	Yes
26	NA	2.6	0	0	ND	3	Yes
27	NA	64.7	0.7	0	ND	6	Yes
28	2.0±0.7	5.0±0.0	NA	10.0±0	ND	ND	Yes	10
29	2.0±0	3.5±0.7	NA	5.0±0	ND	ND	Yes
30–1	4	111.7	NA	6.8	ND	1	Yes	11
30–2	3.5	107.6	NA	9.8	ND	1	Yes
30–3	4	111.3	NA	5.0	ND	1	Yes

MMAF: multiple morphological abnormalities of the sperm flagella; ND: not detected; NA: not available

Although multiple studies have confirmed that biallelic FSIP2 variants cause MMAF, their specific effects on sperm morphology remain controversial. Martinez et al.16 first reported that homozygous LOF FSIP2 variants lead to MMAF, characterized by short, absent, or irregular flagella, severe FS disorganization, and the absence of a mitochondrial sheath in the midpiece. TEM revealed 9 ODFs in the principal piece, with ODFs 3 and 8 not replaced by longitudinal columns.16 Fang et al.,15 using a hypomorphic mouse model, reported shortened flagella, missing FS, and exposed axonemes in FSIP2-mutant sperm. Liu et al.22 reported a thickened midpiece with excessive residual cytoplasm wrapping, and Lv et al.17 reported “super-length” mitochondrial sheaths in the midpiece of sperm flagella, a finding that contradicts that of Martinez et al.16 Furthermore, they confirmed that the lengthening of the mitochondrial sheath could be due to mislocalization or absence of the FS.17 Our findings are consistent with those of Lv et al.17 and provide further evidence that FSIP2 mutations lead to elongation of the mitochondrial sheath rather than its absence. Notably, mitochondrial sheath elongation in FSIP2-deficient spermatozoa should not be interpreted as fibrous sheath extension into the midpiece. Consistent with our immunofluorescence findings using AKAP4, no fibrous sheath components were detected in the midpiece. This interpretation is also supported by Lv et al.,17 who demonstrated that mitochondrial sheath lengthening in FSIP2-deficient individuals results from dislocation or absence of FS rather than annulus ring. Our study contributes to the growing body of evidence that the structural integrity of the flagellum is compromised in FSIP2-related infertility, with specific abnormalities in the mitochondrial sheath length. However, both our study and the study of Lv et al.17 reported 9 ODFs in the principal piece, which is consistent with the findings of Martinez et al.16 Furthermore, we noticed that FS loss was more prominent in the principal piece of the sperm, which was also consistent with the findings of the study by Lv et al.17 These observations suggest that FSIP2 plays a differential role in the organization of the FS across different regions of the sperm tail. The principal piece appears to be more strongly dependent on FSIP2 for maintaining the structural integrity of the FS.

Another notable finding was that infertile male patients carrying heterozygous LOF FSIP2 variants exhibited normal concentrations, progressive motility and no MMAF phenotypes. These findings suggest that heterozygous LOF FSIP2 variants do not adversely affect axonemal assembly. The carrier frequency of LOF FSIP2 variants in the non-MMAF infertility cohort was 0.8% (29/3777), while the carrier frequency of FSIP2 pathogenic variants in MMAF patients was 3.3% (3/92). The collective frequency of 1072 LOF variants of FSIP2 recorded in gnomAD version 4.1.0 is 0.3%. Thus, the carrier frequency of FSIP2 pathogenic variants in both the non-MMAF infertility cohort and MMAF patients was higher than the collective frequency of LOF FSIP2 variants in gnomAD version 4.1.0. It is highly important to screen for carriers of FSIP2 pathogenic variants in the infertile population, especially in the MMAF population.

To date, 6 patients with compound heterozygous FSIP2 variants14,17,20 and 1 with a homozygous LOF variant22 have achieved successful pregnancies through ICSI (Supplementary Table 8). However, 1 patient with compound heterozygous LOF variants failed to achieve pregnancy after ICSI, with no success in fresh or frozen embryo transfers.18 Liu et al.18 proposed that FSIP2 might play a role in embryonic development. In our study, 1 patient with a homozygous LOF variant and 1 with compound heterozygous LOF variants achieved successful pregnancies through ICSI, further supporting ICSI as an effective treatment for FSIP2-related infertility. While one IVF success was claimed in unvalidated compound heterozygotes,28 the lack of pedigree analysis precludes definitive conclusions. Given FSIP2’s acrosomal expression and the report of IVF failure followed by ICSI success,14 we advocate ICSI as a first-line therapy for confirmed biallelic carriers. Notably, the favorable ICSI outcomes observed in our patients may, at least in part, be attributed to stringent sperm selection based on standard the 6^th edition of the WHO manual recommended criteria,30 prioritizing motile spermatozoa with preserved head and acrosomal morphology despite severe flagellar abnormalities.30,41

Supplementary Table 8.

The clinical outcomes of assisted reproductive technology with intracytoplasmic sperm injection in the patients affected by fibrous sheath-interacting protein 2 variants in previous literatures

Patient number	Male age	Female age	Number of oocytes retrieved	Number of MII oocytes	Number of 2PN	Insemination	Number of usable embryos	Number of blastocysts	Transferred cycle	Transferred embryos	Clinical pregnancies	Reference
1	NA	NA	NA	NA	NA	IVF/ICSI	NA	NA	2	NA	Yes	1
7	30	29	12	24	19	ICSI	12	7	2	4 BL	Yes	2
8	27	27	12	11	5	ICSI	4	3	1	1 BL	Yes
10	28	27	19	14	11	ICSI	11	8	1	1 BL	Yes	3
16	NA	NA	NA	5	3	ICSI	3	NA	1	1 CL	Yes	4
17	NA	NA	NA	6	5	ICSI	5	NA	1	1 CL	Yes
18	31	27	12	10	8	ICSI	5	2	2	2 BL+1 CL	No	5
19-1 19-2	34	36	3 9	3 9	2 6	ICSI ICSI	2 3	NA NA	1 1	2 CL 2 CL	No No	6
20	22	23	18	18	18	ICSI	13	NA	1	1 CL	Yes
21	36	31	11	7	6	IVF	4	NA	2	2 CL	Yes
22-1 22-2	22	22	25 23	20 20	18 19	IVF ICSI	15 18	NA NA	1 2	2 CL 2 CL	Yes Yes
23	29	29	15	12	8	ICSI	7	NA	2	4 CL	Yes
24	34	34	7	6	3	ICSI	3	NA	1	1 CL	Yes
25	29	29	15	9	7	ICSI	6	NA	2	4 CL	Yes
26	35	34	13	8	5	ICSI	3	NA	1	2 CL	Yes
27	39	34	19	19	16	ICSI	6	NA	4	6 CL	Yes

2PN: two pronuclei; ICSI: intracytoplasmic sperm injection; AOA: assisted oocyte activation; CL: cleavage embryo; BL: blastocyst; NA: not available; IVF: in vitro fertilization

Although ultrastructural analyses including IF, SEM, and TEM were unavailable for Proband 1 due to limited semen samples, several lines of indirect evidence support a consistent genotype–phenotype correlation between Proband 1 and Proband 2. Both probands harbor biallelic LOF FSIP2 variants and exhibit remarkably similar clinical features, including normal sperm concentration, severely reduced progressive motility, and typical MMAF-associated flagellar abnormalities under light microscopy. Moreover, Proband 2, who carries compound heterozygous truncating variants, demonstrated complete loss of FSIP2 protein expression and profound FS defects, providing a representative ultrastructural phenotype for biallelic LOF FSIP2 variants. Given the shared genetic architecture and highly concordant clinical manifestations, it is reasonable to infer that Proband 1 likely exhibits comparable flagellar ultrastructural abnormalities. Nevertheless, we acknowledge the absence of direct ultrastructural data for Proband 1 as a limitation of this study.

In conclusion, our results validate one novel homozygous and two compound heterozygous FSIP2 variants in patients with male infertility and MMAF phenotypes. Patients carrying FSIP2 variants exhibit short sperm flagella, decreased sperm motility and altered kinetic parameters, elongated mitochondrial sheaths in the midpiece, and absent or dysplastic FS in the principal piece (Figure 4c), which collectively improve our understanding of the genetic spectrum and clinical phenotypes of patients carrying FSIP2 variants. Finally, our findings suggest that ICSI could be recommended for patients with infertility caused by homozygous or compound heterozygous FSIP2 variants.

COMPETING INTERESTS

All authors declare no competing interests.

AUTHOR CONTRIBUTIONS

SJS and JY recruited subjects and sorted clinical information. MY performed WES and Sanger sequencing. SJS performed protein structural simulation and wrote the manuscript. JY and TTH collected the samples, performed SEM, TEM, and Papanicolaou staining, analyzed the results thoroughly, and created the schematic diagram. YNG, QC, and YX performed the Western blotting and immunostaining assays. SJS and SQ designed the study. SQ polished the manuscript. All authors read and approved the final manuscript.

Supplementary Figure 1

Bioinformatic analysis of biallelic FSIP2 variants in infertile males with MMAF. (a) Genomic architecture of FSIP2 (chr2q32.1) with novel variants mapped to exons 16-17. (b) Domain organization of FSIP2 protein (UniProt: Q5CZC0), highlighting mutated residues within conserved functional regions. (c) Cross-species alignment demonstrating evolutionary conservation of altered amino acid residues. (d) AlphaFold-predicted 3D structures of truncated FSIP2 proteins of LOF variants and mutation-induced conformational changes of missense variant. FSIP2: fibrous sheath-interacting protein 2; LOF: loss-of-function; MMAF: multiple morphological abnormalities of the sperm flagella.

Supplementary Figure 2

Expression of acrosome marker and flagellar structural proteins in spermatozoa from FSIP2-mutant patients compared with those from normal controls. (a) Expression of key flagellar proteins in spermatozoa from patients who carried biallelic FSIP2 variants quantified by Western blotting. MS marker: TOM20; FS marker: AKAP4; CP components: SPAG6; ODA components: DNAH17; IDA marker: DNALI1; IFT marker: IFT20. (b) Representative images of spermatozoa from P2, P3 and a fertile control, co-stained with FSIP2 and AKAP4 antibodies. Scale bars = 10 µm; (c). Representative images of spermatozoa from P2, P3 and a fertile control, co-stained with FSIP2 antibody and PNA. Scale bars = 10 µm; FSIP2: fibrous sheath-interacting protein 2; P2: proband 2; P3: proband 3; MS: mitochondrial sheath; FS: fibrous sheath; CP: central pair microtubules; ODA: outer dynein arm; IDA: inner dynein arm; TOM20: translocase of outer mitochondrial membrane 20; AKAK4: A-kinase anchoring protein 4; SPAG6: sperm-associated antigen 6; DNAH17: dynein axonemal heavy chain 17; DNALI1: dynein axonemal light intermediate chain 1; IFT20: intraflagellar transport 20; PNA: peanut agglutinin.

Videos Available on: www.ajajournal.com

Supplementary Video 1

Dynamic image of semen routine examination for proband 2

Supplementary Video 2

Dynamic image of semen routine examination for proband 3.