The TEX44-CPT1B axis regulates mitochondrial sheath assembly and fatty acid oxidation in sperm

Abstract

Mitochondrial fatty acid β-oxidation (FAO) is essential for energy production and cellular homeostasis, yet its role in sperm function has remained unclear. Through whole-exome sequencing (WES) of 800 patients with asthenozoospermia, we identified biallelic Testis-Expressed Protein 44 (TEX44) variants in six individuals, all of whom exhibited defective mitochondrial sheath assembly and impaired sperm motility. Using Tex44 knockout mice, we show that TEX44 interacts with carnitine palmitoyltransferase 1B (CPT1B) to form a mitochondrial glue, anchoring adjacent mitochondria and facilitating the assembly of the sperm-specific mitochondrial sheath. In vitro, we show that purified TEX44 protein can modulate CPT1B enzymatic activity, limiting the conversion of long-chain fatty acids such as palmitic acid and myristic acid into acyl-carnitines, thereby reducing reactive oxygen species (ROS) production. Loss of TEX44 disrupts this regulatory mechanism, leading to unregulated FAO, excessive ROS generation, and severe oxidative damage to sperm DNA and flagellar structure. Additionally, germ cell-specific Cpt1b knockout mice exhibit phenotypes similar to TEX44 deficiency, including mitochondrial sheath defects and reduced sperm motility. These findings reveal a sperm-specific mechanism by which TEX44 regulates CPT1B activity to balance FAO and ROS generation, providing critical insights into energy metabolism, mitochondrial integrity, and male infertility.

Mitochondrial fatty acid β-oxidation is essential for cellular energy production. Here the authors show that mutations in Testis-Expressed Protein 44 (TEX44) are associated with misregulated fatty acid β-oxidation and male infertility.

Introduction

Mitochondrial FAO is a crucial pathway for energy production and cellular homeostasis^1–4. Fatty acids (FAs) obtained from the diet and adipose tissue are transported via the bloodstream, with short- and medium-chain FAs diffusing passively across the plasma membrane upon reaching target cells^3,5,6. In contrast, long-chain FAs require active transport by specific protein carriers such as FA transport proteins (FATPs), FA translocase (FAT), and FA-binding proteins (FABPs). Within the cytosol, long-chain FAs are converted into acyl-CoA esters, while short- and medium-chain FAs diffuse into the mitochondrial matrix and are converted into acyl-CoA esters^3,6,7. This process involves carnitine palmitoyltransferase 1 (CPT1), which catalyzes the formation of acylcarnitines (CARs) from acyl-CoA esters and free l-carnitine, mitochondrial carnitine–acylcarnitine translocase (CACT), responsible for the exchange of a CAR for a free l-carnitine molecule, and carnitine palmitoyltransferase 2 (CPT2), which re-esterifies CARs to acyl-CoA esters within the mitochondrial matrix^1,3–5,8. Once in the mitochondrial matrix, long-chain acyl-CoA undergoes FAO, producing acetyl-CoA, which then enters the tricarboxylic acid (TCA) cycle. NADH and FADH2 generated during FAO and the TCA cycle are utilized by the electron transport chain to produce adenosine triphosphate (ATP). Disorders involving specific FA transport proteins or enzymes in FA β-oxidation pathways are known as FA β-oxidation disorders (FAODs). Over 15 distinct FAODs have been described, although none have been linked to sperm abnormalities⁹.

Spermatozoa, the only cells in higher animals capable of functioning outside the body, rely heavily on ATP production for motility. While glycolysis has been suggested as the primary energy source for sperm motility^10–12, sperm from various species, including humans, can maintain motility in sugar-free media^13,14, indicating alternative energy sources. Mitochondrial FAO, utilizing exogenous fatty acids as the main energy source, is a key regulator of sperm motility in bovine and boar species. Inhibition of CPT1 by etomoxir significantly reduces sperm motility. Human sperm contain enzymes involved in lipid metabolism, including those associated with mitochondrial FAO, suggesting a role for FAO in sperm function^15–18. Disruption of FAO may contribute to sperm motility disorders and male infertility, although no gene variants have been reported to cause male infertility.

Infertility affects approximately 15% of couples worldwide^19–21, with male factor infertility accounting for half of cases^22–24. Asthenozoospermia, characterized by reduced sperm motility, affects about 19% of infertile males^25,26. Recent studies have identified several genes associated with asthenozoospermia, particularly those involved in sperm flagellar structure^27–29. However, these genes only account for a subset of asthenozoospermia cases. In 2023, a population-based survey identified microtubule inner protein (MIP) variants as the most common cause of asthenozoospermia without severe flagellar defects³⁰. However, these variants only account for 11.4% of asthenozoospermia cases, suggesting that other causes remain elusive, and the potential association of FAO-related genes with asthenozoospermia is unclear.

Here, we analyzed WES data from 800 infertile men with asthenozoospermia and identified 6 individuals with deleterious biallelic variants in the TEX44 gene, whose protein product was previously detected in human spermatozoa³¹; these individuals all exhibiting partial loss of sperm mitochondrial sheath. Using immunoprecipitation-mass spectrometry (IP-MS), we found that TEX44 interacts with CPT1B, a key enzyme in FAO³². Unexpectedly, TEX44 deficiency increased long-chain acyl-CoA generation, resulting in more severe reproductive phenotypes than those observed in conditional Cpt1b knockout mice. This study sheds light on a unique sperm survival strategy involving the restriction of long-chain fatty acid utilization to prolong sperm motility ex vivo.

Results

Biallelic TEX44 variants disrupt sperm mitochondrial sheath

Our investigation into the genetic underpinnings of asthenozoospermia involved a cohort of 800 infertile males using WES (Fig. 1A). The semen parameters, including sperm concentration, count, progressive motility, motility, and normal morphology, exhibited heterogeneity among the patients (Table S1). Based on these parameters, we further classified the patients into four categories: isolated asthenozoospermia (4.16%, n = 28), oligoasthenozoospermia (1.63%, n = 11), asthenoteratozoospermia (11.29%, n = 76), and oligoasthenoteratozoospermia (82.91%, n = 558) (Fig. 1B). This effort led to the identification of potentially harmful biallelic variants in the TEX44 (GenBank: NM_152614.3) gene in six of these subjects (0.75%, 6/800) (Fig. 1A, C, Table S2), and the other two candidate genes (Table S3) have no known relationship with male fertility. The TEX44 gene spans 41.60 kb on chromosome 2q37.1 in humans and encodes a 395 amino acid protein featuring an unknown function DUF4678 domain (Fig. S1). Analysis of GTEx and NCBI databases revealed TEX44 exhibits a testis-specific expression pattern.

Fig. 1. Asthenozoospermia and mitochondrial sheath defects in spermatozoa associated with human TEX44 variants.

A Flow chart of the identification of human TEX44 variants related to asthenozoospermia. Created in BioRender. f Zl, v}. (2025) https://BioRender.com/kbr0lu4. B Distribution of patient categories based on semen analysis. C Pedigree analysis of six families of individuals harboring biallelic TEX44 variants identified through whole-exome sequencing. Infertile males are denoted by filled black squares. Sanger sequencing results are shown beneath the pedigrees, with mutated residues highlighted by red arrows and boxes. D HE staining was used to assess the morphology of spermatozoa from a fertile control man and patients with biallelic TEX44 variants via light microscope. Scale bar: 10 μm. E Quantification of abnormal mitochondrial sheaths in fertile controls and patients with biallelic TEX44 variants (P = 1.45 × 10⁻⁹, n = 5). F Sperm tail length comparison between fertile controls and patients harboring biallelic TEX44 variants (ns not significant, n = 5). G Immunofluorescent staining for TEX44 in spermatozoa from fertile controls and males harboring biallelic TEX44 variants. Anti-TEX44 (green) and anti-TOMM20 (magenta) antibodies were used, with Hoechst (blue) for nuclei staining. Scale bar: 5 μm. H Western Blot analysis of TEX44 in spermatozoa from a fertile control and patients harboring biallelic TEX44 variant. TEX44 is missing in spermatozoa from patients, with β-Tubulin as a loading control. Uncropped blots are provided in Source data. I Transmission electron microscopy of longitudinal sperm flagellar midpiece sections from fertile controls and individuals with biallelic TEX44 variants. Red arrows indicate sloughed-off mitochondria; yellow arrowheads indicate the annulus. Scale bar: 500 nm. J Schematic diagram of mitochondrial sheath defect in spermatozoa from fertile controls and patient harboring biallelic TEX44 variant. K Immunofluorescence analysis showing the localization of TOMM20 (green), SEPT4 (magenta), and Hoechst (blue) in control and Family IV-1 spermatozoa. Scale bar: 10 μm. For (E, F), data are mean ± s.e.m. P values were determined using two-tailed Student’s t-tests. n values represent the number of biologically independent experiments. **** indicates statistical significance at P < 0.0001. H, I Experiments were performed on samples from three unrelated individuals with biallelic TEX44 variants. Similar results were observed across all biological replicates. K Images are representative of multiple spermatozoa from the same individual, with consistent staining patterns observed.

Among these variants, the homozygous TEX44 frameshift mutation [c.424_427del (p. Glu142fs)] was recurrently detected in one proband from a consanguineous family (family1) and four non-consanguineous families (families 2–5) within our cohort (Fig. 1C, Table S2). This mutation is anticipated to lead to premature translational termination, resulting in either a truncated, non-functional protein or complete absence of TEX44 protein (Table S2). According to population databases, the allele frequency of this variant is 0.008368 in gnomAD and 0.0124 in the 1000 Genomes Project (Table S2).

Furthermore, a novel missense mutation in TEX44 [c.236A>G (p. Gln79Arg)] was identified in one male from family 6. This mutation affects a highly conserved residue in the TEX44 coding sequence, with predictive tools including SIFT, PolyPhen-2, and MutationTaster suggesting its deleterious nature (Fig. 1C, Table S2). Its allele frequency is reported as 0.01392 in gnomAD and 0.0361422 in the 1000 Genomes Project (Table S2). Notably, these TEX44 variants were absent in 157 fertile men included in our study. We thus postulated that these biallelic TEX44 variants might contribute to the infertility-related phenotypes observed in affected males.

Subsequent semen parameter analyses of the probands carrying biallelic TEX44 variants revealed either normal or slightly reduced sperm concentrations (Table S4). All five patients with biallelic TEX44 frameshift mutations showed concurrent impairments in sperm concentration, motility, and morphology, and were therefore classified as having oligoasthenoteratozoospermia. Notably, impaired sperm motility was observed in all six individuals, with four probands displaying a progressive motility rate of 0% and one proband exhibiting a rate of just 1.0% (Table S4). Notably, one proband demonstrated 33.9% motile sperm (Table S4), suggesting some variation in motile sperm proportion among TEX44-deficient individuals. Morphological examination using Hematoxylin and eosin (H&E) staining revealed normal, elongated flagella in fertile control individuals, while probands from families 1–5 exhibited mitochondrial sheath defects (Fig. 1D, E), but flagellum length remained unchanged (Fig. 1F).

To elucidate the pathogenicity of TEX44 variants, we examined TEX44 protein localization using in-house-prepared specific antibodies for immunofluorescent staining. In sperm from fertile controls, TEX44 localized specifically to the flagellar midpiece (Fig. 1G). Conversely, proband Family1 IV-1 displayed no detectable TEX44 staining in the flagella, indicating potential nonsense-mediated mRNA decay due to premature translational termination (Fig. 1G). The absence of TEX44 in sperm from proband Family1 IV-1 was also confirmed by immunoblotting (Fig. 1H).

Transmission electron microscopy (TEM) of longitudinal flagellar sections revealed mitochondrial sheath abnormalities in individuals with TEX44 variants. Control individuals exhibited regularly arranged mitochondrial sheaths, in contrast to absent or shed structures observed in sperm from Family1 IV-1, Family3 II-2 and Family4 II-1, with the annulus structure remaining intact (Fig. 1I, J). SEPT4 immunofluorescence staining further confirmed the normal integrity of the annulus in these patients (Fig. 1K).

We utilized a CRISPR/Cas9 approach to induce a 4-bp deletion and associated frameshift mutation, generating Tex44^−/− mice (Fig. S2A). Sanger sequencing and western blotting confirmed the successful generation of Tex44^−/− animals (Fig. S2A, B). Male Tex44^+/+ and Tex44^−/− mice showed comparable body and testis weights (Fig. S2C), but male Tex44^−/− mice were subfertile (Fig. S2D). Although sperm concentrations were normal in male Tex44^−/− mice (Fig. S2E), sperm motility and progressive motility were significantly impaired (Fig. S2F–H). We also performed Eosin-Nigrosin staining in Tex44 knockout mice, which showed no increase in sperm death (Fig. S2I, J), further supporting that the impaired motility is not due to reduced sperm viability.

Microscopic evaluation revealed localized flagellar defects and mitochondrial sheath abnormalities in spermatozoa from mutant mice, without shortened tails (Fig. S3A–C). We have conducted a more detailed quantitative analysis of the distinct mitochondrial sheath defects observed in spermatozoa from the caput, corpus, and cauda regions of the epididymis. The abnormalities were categorized into three distinct types: (1) MP-PP disjunction, characterized by the separation between the midpiece and principal piece; (2) Partial Bent, defined by localized bending of the flagellum at angles less than 180°; and (3) Complete Bent, defined by a 180° curvature of the flagellum (Fig. S3D). This refined analysis reveals a significantly increased prevalence of each abnormality in Tex44^−/− mice compared to controls (Fig. S3E). Notably, in the Tex44^−/− mice, complete bending emerged as the most predominant abnormality in the cauda, indicating a progressive worsening of flagellar defects during the epididymal transit (Fig. S3E).

Immunofluorescence analysis demonstrated the absence of TEX44 in sperm flagella from Tex44^−/− mice (Fig. S3F). TOMM20 immunofluorescence of mitochondrial sheaths showed a marked reduction in sheath length in Tex44 knockout mice (Fig. S3F, G). We examined the morphology of mature spermatozoa from the Tex44^−/−cauda epididymis by scanning electron microscopy (SEM), in Tex44^−/− spermatozoa, mitochondria were absent near the annulus, resulting in a partially absence of mitochondrial sheath (Fig. S3H). TEM analyses also confirmed the partial absence of the mitochondrial sheath near the annulus in sperm from mutant mice, consistent with human patients harboring TEX44 variants (Fig. S3I). SEPT4 immunofluorescence revealed a normal localization pattern at the annulus in both Tex44^+/+ and Tex44^−/− mice (Fig. S3J). Additionally, co-immunostaining of TEX44 and SEPT4 in mouse spermatozoa indicated that TEX44 does not localize to the annulus (Fig. S3K, L). These results collectively confirm that sperm flagellar defects were due to the lack of the mitochondrial sheath rather than the annulus.

In vitro fertilization (IVF) using sperm from male Tex44^−/− mice resulted in significantly lower rates of zygote, 2-cell embryo, and blastocyst formation compared to Tex44^+/+ mice (Fig. 2A–E). However, intracytoplasmic sperm injection (ICSI) showed comparable rates of zygote, 2-cell embryo, and blastocyst formation between Tex44^−/−and Tex44^+/+ male mice (Fig. 2F–J). Successful pregnancies were achieved following ICSI treatment for two probands (Family1 IV-1 and Family3 II-2) harboring biallelic TEX44 variants identified in the study.

Fig. 2. IVF and ICSI outcomes with Tex44^−/− and Tex44^+/+ mice.

A Schematic diagram of in vitro fertilization (IVF). B–E Representative images of zygotes, 2-cell embryos, and blastocysts obtained from IVF in mice (B). The percentage of zygotes (C), 2-cell embryos (D), and blastocysts (E) in Tex44^+/+ and Tex44^−/− mice is illustrated (P = 0.0136 for (C), P = 0.0108 for (D), P = 0.0395 for (E); n = 3). Scale bars: 200 μm. F Schematic diagram of intracytoplasmic sperm injection (ICSI). G–J Representative images of zygotes, 2-cell stage, and blastocysts obtained from ICSI in mice (G). Comparable percentages of zygotes (H), 2-cell embryos (I), and blastocysts (J) were observed in Tex44^+/+ and Tex44^−/− mice (ns not significant, n = 3). Scale bar: 200 μm. For (C–E, H–J), data are presented as mean ± s.e.m. P values were determined using two-tailed Student’s t-tests. n values represent the number of biologically independent animals. * indicates statistical significance at P < 0.05.

TEX44–CPT1B links mitochondria and exhibits phase separation

To explore proteins potentially interacting with TEX44 during sperm mitochondrial sheath formation, we conducted an IP-MS study using mouse testicular protein extracts, identifying four candidate proteins (Table S5). Notably, we discovered an interaction between TEX44 and CPT1B, a mitochondrial outer membrane protein. Co-immunoprecipitation experiments confirmed the interaction between HA-TEX44 and FLAG-CPT1B when co-expressed in HEK293T cells (Fig. 3A), which was further validated by reverse IP (Fig. 3B). In HeLa cells, TEX44-EGFP and CPT1B-HA exhibited co-localization patterns on the mitochondria, forming a “mitochondrial glue”. Additionally, TEX44-EGFP signals were observed between adjacent mitochondria (Fig. 3C). Additionally, we performed co-immunoprecipitation experiments by co-expressing TEX44-FLAG and TEX44-HA in HEK293T cells. The results showed that TEX44-HA could be pulled down with TEX44-FLAG and vice versa, indicating that TEX44 is capable of forming homodimers (Fig. S4), which may account for the detection of a band approximately twice the expected molecular weight. TEX44 expression initiated during mouse spermatogenesis phase II-III, corresponding to step 14 spermatids (Fig. 3D), immunofluorescence analysis revealed co-localization of TEX44 and CPT1B in the mitochondrial sheath of both murine and human spermatozoa (Fig. 3E), implicating TEX44 and CPT1B in sperm mitochondrial sheath formation.

Fig. 3. TEX44 interacts with CPT1B to function as an inter-mitochondrial linker.

A, B Co-immunoprecipitation assays assessing interactions between FLAG-TEX44 and HA-CPT1B. TEX44-FLAG and CPT1B-HA were immunoprecipitated from HEK293T cell lysate transfected with the indicated vectors with anti-FLAG (A) and anti-HA (B) antibodies. Interactions between proteins were detected with antibodies to FLAG or HA. Uncropped blots are provided in Source data. C Immunofluorescent staining for TEX44-EGFP (green) or CPT1B-HA (magenta) with antibodies to EGFP or HA in HeLa cell lines. TOMM20 served as a mitochondrial marker (blue). The magnified region in the right panel shows TEX44 assemblies between mitochondria (marked by white arrows). Scale bars: 10 μm (left), 2 μm (right). D Immunofluorescent staining was performed to examine the expression of TEX44 and CPT1B during various stages of spermiogenesis (steps 1–16) in the testes of adult wild-type mice. TEX44 (magenta) and CPT1B (green) antibodies were used, along with PNA (white) to highlight acrosomal structures. Nuclei were stained with Hoechst (blue). Scale bars: 20 μm. E Immunofluorescent staining for CPT1B (magenta) and TEX44 (green) in human and mice sperm samples. Nuclei were stained with Hoechst (blue). Scale bars: 5 μm. F Graph showing disordered regions (IDRs) of TEX44 identified by IUPred3. A score of ≥0.5 indicates disordered regions. A schematic representation of TEX44 protein and truncated mutants is shown, with orange boxes indicating IDRs. The numbers represent amino acid residues. G Co-immunoprecipitation assay for full-length or truncated TEX44-FLAG and CPT1B-HA with antibodies to FALG or HA. Uncropped blots are provided in Source data. H Immunofluorescent staining for full-length or truncated TEX44-FLAG (green) and CPT1B-HA (magenta) in HeLa cells. Nuclei were stained with Hoechst (blue). Scale bars: 10 μm. A–E, G, H All experiments were independently repeated at least three times with consistent results. Representative data are shown.

We employed IUPred3 to predict the intrinsically disordered regions (IDRs) within TEX44, TEX44 1–384 amino acids contain 3 IDRs, we constructed plasmids to express the N-terminal (1–384aa) and C-terminal (385–530aa) truncated variants of TEX44(Fig. 3F). To delineate the interaction domain between TEX44 and CPT1B, we conducted truncated protein co-immunoprecipitation assays. Deletion of the C-terminal 385–530 amino acids abolished TEX44 binding to CPT1B (Fig. 3G), along with TEX44 mitochondrial localization (Fig. 3H). TEX44 385–530aa directly bound to CPT1B and localized to mitochondria (Fig. 3G, H). Interestingly, overexpression of full-length TEX44 (1–530aa) led to mitochondrial aggregation compared to TEX44 (385–530aa) overexpression (Fig. 3H), suggesting a role for TEX44 (1–384aa) in linking adjacent mitochondria.

SEC16A was initially identified as a putative TEX44-interacting protein in our IP-MS analysis. However, subsequent co-immunoprecipitation assays in HEK293T cells failed to confirm a direct interaction between TEX44 and SEC16A (Fig. S5A). We further validated the interaction between TEX44 and two proteins involved in adherens junctions: CDH2 (N-cadherin) and CTNNB1 (β-catenine), through co-immunoprecipitation (Fig. S5B, C). In the absence of CPT1B, TEX44 fails to localize to the mitochondria (Fig. S5D). In contrast, when TEX44 and CPT1B are co-expressed, both proteins co-localize at the outer mitochondrial membrane (Fig. S5D), suggesting that CPT1B facilitates TEX44 anchoring to mitochondria. CDH2, on the other hand, remains localized between mitochondria regardless of CPT1B expression (Fig. S5E). Notably, when CPT1B, TEX44, and CDH2 are co-expressed, CDH2 is localized between distant mitochondria (Fig. S5F). Similarly, CTNNB1 exhibits inter-mitochondrial localization upon co-expression with TEX44 and CPT1B (Fig. S5F), suggesting that adhesion proteins may facilitate mitochondrial sheath formation via TEX44 by bridging non-adjacent mitochondria. During spermatogenesis, CDH2 exhibits cytoplasmic localization in spermatocytes and round spermatids. Additionally, partial co-localization of CDH2 and TEX44 was observed in elongated spermatids during stages V–VIII (Fig. S5G), further supporting their potential role in mitochondrial organization.

Based on the presence of multiple IDRs in TEX44, we hypothesized that TEX44 might exhibit phase separation properties. Live-cell imaging of TEX44-EGFP fluorescence recovery after photobleaching (FRAP) revealed a rapid recovery of TEX44-EGFP fluorescence following photobleaching (Fig. S6). These findings suggest that TEX44 possesses liquid-liquid phase separation characteristics and exhibits dynamic mobility within the cell.

CPT1B loss disrupts TEX44 assembly and mitochondrial sheath

We generated a mouse model, Cpt1b gKO (Fig. 4A), with germ cell-specific deletion of Cpt1b driven by Stra8. Fertility tests revealed significantly reduced litter sizes in Cpt1b gKO male mice compared to controls (Fig. S7A). Optical microscopy revealed partial loss of the mitochondrial sheath in sperm from Cpt1b gKO mice (Fig. S7B), which was further quantified and found to be significantly increased compared to controls (Fig. S7C). TEM and SEM revealed that this partial loss of the mitochondrial sheath did not impair mitochondrial compaction (Fig. 4B, C), similar to mice lacking TEX44. Additionally, CPT1B deficiency leads to a progressive worsening of sperm abnormalities from the caput to the cauda region of the epididymis (Fig. S7D). While Cpt1b gKO mice exhibited normal sperm concentration (Fig. S7E), motility and progressive motility were significantly decreased (Fig. S7F, G). Eosin-Nigrosin staining further confirmed that the observed motility defects were not due to increased sperm death (Fig. S7H, I). These findings suggest that CPT1B plays a crucial role in mitochondrial sheath assembly.

Fig. 4. Loss of CPT1B in male germ cells leads to TEX44 assembly failure and abnormal mitochondrial sheath formation.

A Generation of Cpt1b gene knockout mice through Cre-LoxP mediated exon deletion. B Scanning electron microscopy of spermatozoa from the cauda epididymis of control and Cpt1b gKO mice. White arrows indicate regions lacking mitochondria. Scale bars: 10 μm (left), 2 μm(right). C Transmission electron microscopy of spermatozoa from the cauda epididymis of control and Cpt1b gKO mice. The black boxed areas in the upper panels of each group are shown at higher magnification in the corresponding lower panels. Scale bars: 2 μm (upper), 1 μm (lower). D Immunofluorescence analysis of TEX44 (green) in spermatozoa from control and Cpt1b gKO mice. The mitochondrial sheath was labeled with TOMM20 (magenta), and nuclei were stained with Hoechst (blue). Scale bars: 50 μm. E Immunofluorescence analysis of TEX44 (green) from control and Cpt1b gKO testis. The mitochondrial sheath, acrosome, and nuclei were stained with GPX4 (magenta), WGA (yellow), and Hoechst (white), respectively. The regions demarcated by white dashed boxes in the middle panels of each group are presented at higher magnification in the corresponding right panels. The lower panels display the fluorescence intensity profiles corresponding to the areas indicated by the light blue lines. Scale bars: 20 μm (left), 10 μm (right). F Immunofluorescence analysis of CPT1B (green) from Tex44^+/+ and Tex44^−/− testis. TOMM20 (magenta) were used to stain the mitochondrial sheath, Hoechst (white) were used to stain nucleus, wheat germ agglutinin (WGA; yellow) were used to determine the step of spermiogenesis. The regions demarcated by white dashed boxes in the middle panels of each group are presented at higher magnification in the corresponding right panels. The lower panels display the fluorescence intensity profiles corresponding to the areas indicated by the light blue lines. Scale bars: 20 μm (left), 10 μm (right). B–F All experiments were independently repeated at least three times with consistent results. Representative data are shown.

Through immunofluorescence, we observed TEX44 depletion in epididymal sperm tail of Cpt1b gKO mice (Fig. 4D). In testicular sections of Cpt1b gKO mice, TEX44 expression was detected but mislocalized within the elongated spermatid mitochondrial sheath (Fig. 4E). These findings indicate that TEX44 is initially expressed in Cpt1b^−/− testes but is not properly integrated into the sperm flagellum during spermiogenesis. Consistent with these findings, western blot analyses further confirmed that TEX44 is present in the testes but absent in spermatozoa of Cpt1b gKO mice (Fig. S8). Conversely, the mitochondrial sheath localization of CPT1B in elongated spermatids was unaffected in Tex44^−/− testicular sections (Fig. 4F). These results indicate that TEX44 anchors to the mitochondrial sheath through interaction with CPT1B, thereby participating in sperm mitochondrial sheath assembly.

TEX44 limits long-chain acylcarnitine production in sperm

CPT1B plays a critical role in the transport of long-chain acyl-CoAs into the mitochondria for β-oxidation. (Fig. 5A). To determine if TEX44 deficiency affects FAO, we compared the metabolite changes in Tex44^−/− sperm to control sperm through metabolomics. We found significant alterations in FA metabolism, particularly an unexpected increase in L-palmitoylcarnitine and Myristoyl-l-carnitine production in Tex44^−/− sperm (Figs. 5B, S9 and S10, Supplementary Data 1). Further examination of L-palmitoylcarnitine and Myristoyl-l-carnitine levels in epididymal sperm tail from Cpt1b gKO mice revealed significant downregulation (Figs. 5B and S9 and S10, Supplementary Data 1). These results suggest that CPT1B participates in the generation of L-palmitoylcarnitine and Myristoyl-l-carnitine in sperm, and without TEX44 restriction, their production levels increase significantly.

Fig. 5. TEX44–CPT1B limits the generation of L-palmitoylcarnitine and myristoyl-l-carnitine of CPT1B in sperm.

A Schematic diagram of fatty acid transport. CPT1B, located on the mitochondrial outer membrane, converts long-chain acyl-CoA into long-chain acyl-carnitine. Short-chain fatty acids freely diffuse across mitochondrial membrane. B Quantification of carnitine, carnitine-conjugated long-chain fatty acids (LCFAs), and carnitine-conjugated short-chain fatty acids (SCFAs) in spermatozoa from control (n = 16), Cpt1b gKO (n = 9), and Tex44^−/− (n = 9) mice. For L-palmitoylcarnitine, significant differences were observed between Cpt1b gKO and control (P = 3.64 × 10⁻⁵) and between Tex44^−/− and control (P = 0.0019). For myristoyl-L-carnitine, significant differences were also observed between Cpt1b gKO and control (P = 0.0002) and between Tex44^−/− and control (P = 0.0002). Asterisks indicate statistical significance: P < 0.01 (**), P < 0.001 (***), P < 0.0001 (****); ns not significant, P ≥ 0.05. C Immunofluorescence analysis of CPT1B (green) in spermatozoa from Tex44^+/+ and Tex44^−/− mice. TOMM20 (magenta) marks mitochondria, Hoechst (blue) stains nuclei. Scale bars: 50 μm. D Western blot analysis showing comparable CPT1B protein levels in sperm from Tex44^+/+ and Tex44^−/− mice, with β-tubulin as a loading control. Uncropped blots are provided in Source data. E Quantitative analysis of CPT1B protein level was performed and normalized with β-tubulin protein level (n = 3). No statistically significant difference was observed (ns, P ≥ 0.05). F Schematic workflow for protein expression, purification, and subsequent CPT1B enzyme activity assay. CPT1B enzyme activity was tested alone and in combination with TEX44-FL or TEX44-C. G Coomassie brilliant blue staining to assess protein purity (CPT1B, TEX44-FL, TEX44-C), with BSA used as a reference for protein quantification. H CPT1B enzyme activity at incremental concentrations showing a linear relationship (R² = 0.9884). Relative enzyme activity of CPT1B in the presence of TEX44-FL (I) or TEX44-C (J) at various TEX44-to-CPT1B ratios. TEX44-FL inhibits CPT1B activity in a dose-dependent manner, whereas TEX44-C enhances activity. For (B, E), data are presented as mean ± s.e.m. P values were determined using two-tailed unpaired Student’s t-tests. n values represent the number of biologically independent animals. C–E Experiments were performed using samples from three biologically independent mice, with consistent results observed. Representative images and blots are shown.

To investigate the mechanism by which loss of TEX44 in sperm promotes the accumulation of long-chain acyl-CoAs, we first assessed whether TEX44 deletion affects the expression or activity of CPT1B. Immunofluorescence analysis (Fig. 5C) showed that the loss of TEX44 does not affect the localization of CPT1B in the sperm mitochondrial sheath. Western blot analysis further revealed that CPT1B protein levels in sperm lysates from both wild-type and Tex44^−/− mice were comparable (Fig. 5D, E), indicating that TEX44 regulates CPT1B activity rather than its expression or localization. We then expressed and affinity-purified FLAG-tagged CPT1B, full-length TEX44 (TEX44-FL, 1–530aa), and its C-terminal fragment (TEX44-C, 385–530aa) in Expi293F cells (Fig. 5F, G). CPT1B activity was assessed using a DTNB-based colorimetric assay that measures the release of CoA-SH from palmitoyl-CoA, as indicated by absorbance at 412 nm. A linear relationship between CPT1B concentration and absorbance confirmed the validity of this assay (Fig. 5H). We then investigated whether TEX44 modulates CPT1B activity by co-incubating a fixed amount of CPT1B with increasing amounts of TEX44-FL or TEX44-C. TEX44-FL induced a concentration-dependent reduction in CPT1B activity (Fig. 5I), suggesting an inhibitory regulatory effect. In contrast, co-incubation with TEX44-C resulted in a modest enhancement of CPT1B activity (Fig. 5J), implying that the C-terminal region of TEX44 alone may not mediate full inhibition and may even facilitate enzymatic activity.

We utilized AlphaFold3 to predict the complex structure between TEX44 and CPT1B (Fig. 6). Consistent with sequence analysis, the N-terminal region of TEX44 was largely flexible and disordered when predicted using the full-length protein (Fig. 6A). By contract, TEX44 (385–530) exhibited a structured conformation and maintained close contact with CPT1B in the predicted structures, whether using full-length TEX44 or TEX44 (385–530), which supports our co-immunoprecipitation results (Fig. 6A–C). Unfortunately, the predicted local distance difference test (pLDDT) score of the predicted structures indicated high confidence for CPT1B but relatively low confidence for TEX44 (Fig. 6A–C). In the top five predicted results using full-length CPT1B and TEX44 (385–530), all CPT1B structures were similar, whereas TEX44 structures were more diverse (Fig. 6D). Despite of that, one segment of TEX44 (424–450) remained invariant in all predicted structures, which sits in a cavity of CPT1B and shows perfect electrostatic complementary with CPT1B (Fig. 6E). This suggests that this segment may contribute to the direct interaction with CPT1B. Based on this analysis, TEX44 may affect the enzymatic activity of CPT1B by interfering with substates entry and product release, thereby explaining our experimental observations.

Fig. 6. Predicted complex structures between CPT1B and TEX44 by AlphaFold3.

A Predicted structure of full-length CPT1B and TEX44 colored by chain (left) or pLDDT scores (right). The top hit was presented. B, C Predicted structure of full-length CPT1B and TEX44 (385–530). Two representative results were presented. The N-terminal of TEX44 insert through CPT1B in (B) but not in (C). D All five predicted results of the complex structure between CPT1B and TEX44 (385–530) show an invariant segment of TEX44 (424–450) highlighted in red. E Electrostatic surface of CPT1B (left) and TEX44 (424–450) (right). Positively and negatively charges are indicated by blue and red colors, respectively.

Tex44^−/− sperm show increased ROS damage in palmitoyl-CoA media

FAO is known to increase ROS generation, potentially damaging target cells. It remains unclear whether changes in FAO in sperm affect ROS levels. To address this, spermatozoa were treated with palmitoyl-CoA and L-carnitine in human tubal fluid (HTF) medium, and ROS levels, bent tail deformities, and DNA damage were subsequently assessed (Fig. 7A). We found that ROS production in sperm from Cpt1b gKO mice cultured in palmitoyl-CoA media decreased compared to the control group (Fig. 7B). Conversely, significantly elevated ROS damage was observed in Tex44^−/− sperm cultured in palmitoyl-CoA media (Figs. 7B and S11). These Tex44^−/− sperm also exhibited more severe bent tail deformities and increased DNA damage compared to Cpt1b gKO sperm (Fig. 7C–F). Moreover, the higher proportion of complete bent sperm observed in Tex44^−/− mice compared to Cpt1b gKO mice is presumably a consequence of unregulated fatty acid oxidation and excessive ROS generation caused by TEX44 deficiency (Figs. S3E and S7D). These results suggest that while CPT1B-mediated FAO supports sperm motility by providing energy, unregulated FAO leads to excessive ROS generation, causing severe oxidative damage in sperm.

Fig. 7. Tex44^−/− sperm exhibit elevated ROS levels, abnormal morphology, and DNA damage when cultured in a medium containing palmitoyl-CoA and carnitine.

A Schematic workflow for detecting ROS levels, morphological defects, and apoptosis in control, Tex44^−/−, and Cpt1b gKO spermatozoa after treatment with palmitoyl-CoA and carnitine. B Mean fluorescence intensity of ROS levels in sperm treated with increasing concentrations of palmitoyl coenzyme A (0, 10 nM, 100 nM, and 1 µM) in control, Tex44^−/−, and Cpt1b gKO mice (n = 9 per group). Compared to controls, Tex44^−/− sperm showed significantly higher ROS levels at 0 nM (P = 9.15 × 10⁻¹⁵), 10 nM (P = 2.38 × 10⁻¹⁷), 100 nM (P = 2.88 × 10⁻²⁰), and 1 µM (P = 2.81 × 10⁻¹⁹); Cpt1b gKO sperm showed significantly lower ROS levels at 0 nM (P = 4.88 × 10⁻¹⁸), 10 nM (P = 4.32 × 10⁻¹⁴), 100 nM (P = 6.94 × 10⁻²⁰), and 1 µM (P = 1.86 × 10⁻¹⁹). The concentration of L-carnitine in each group was 1 mM. C, D HE staining of spermatozoa from control, Tex44^−/− and Cpt1b gKO mice (C) before and after treatment with 1 µM palmitoyl-CoA and 1 mM carnitine (n = 3, each). Quantification of bent tail rates in spermatozoa after treatment with 1 µM palmitoyl-CoA and 1 mM carnitine. Tex44^−/− spermatozoa show a significant increase after treatment (P = 2.73 × 10⁻⁵), while a moderate increase is seen in controls (P = 0.0341). No significant change is observed in Cpt1b gKO spermatozoa (P = 0.1304). Scale bars: 50 μm. E, F TUNEL staining (red) of control, Tex44^−/− and Cpt1b gKO spermatozoa, before and after treatment with 1 µM palmitoyl-CoA and 1 mM carnitine (n = 3, each). The quantification of TUNEL-positive spermatozoa is comparable between the control (P = 0.2236) and Cpt1b gKO (P = 0.5059) mice after treatment with 1 µM palmitoyl-CoA and 1 mM carnitine, but a significant increase in the Tex44^−/− mice (P = 1.24 × 10⁻⁷). The nuclei were stained with Hoechst (blue). Scale bars: 100 μm. For (B, D, F), data are presented as mean ± s.e.m. P values were determined using two-tailed unpaired Student’s t-tests. n values represent the number of biologically independent animals. * indicates statistical significance at P < 0.05. **** indicates statistical significance at P < 0.0001. ns not significant.

Discussion

In this study, we identified six unrelated males with asthenozoospermia harboring biallelic variants in the TEX44 gene through a WES study of infertile patients from three hospitals (0.75%, 6/800). Among the identified mutations, the missense variant [c.236A>G (p.Gln79Arg)] was associated with a milder phenotype, whereas the frameshift variant [c.424_427del (p.Glu142fs)] led to severe oligozoospermia. While the precise functional consequences of the missense variant [c.236A>G (p.Gln79Arg)] remain unclear due to the lack of a mouse model specifically mimicking this variant, it is generally recognized that missense mutations may exert limited effects on protein function, particularly when the altered residue is not situated within a functional domain. In this case, this residue lies outside the TEX44–CPT1B interaction domain, making it unlikely to disrupt their binding. This may help explain the milder phenotype observed in patient 6, in contrast to the severe defects associated with the frameshift mutation [c.424_427del (p. Glu142fs)], which leads to a complete absence of TEX44 protein in spermatozoa.

By generating Tex44 knockout mice, we confirmed the pathogenicity of TEX44 defects. Interestingly, Tex44 knockout mice exhibited increased generation of L-palmitoylcarnitine and Myristoyl-l-carnitine, as well as elevated ROS levels. This raises questions about whether the overall damage level of sperm affects the outcomes of assisted reproductive therapy. For patients diagnosed with idiopathic asthenozoospermia, no effective empirical treatments have been shown to improve sperm parameters, with ICSI being the only reliable approach to achieving a pregnancy³³. Previous studies suggest that the clinical outcomes of asthenozoospermia patients differ based on the mutated genes, with certain gene variants associated with positive pregnancy outcomes while others are linked to pregnancy failure^34–37. When we analyzed ICSI outcomes using sperm from male Tex44 knockout and wild-type mice, we found no differences in rates of two-cell embryo or blastocyst formation, suggesting that ICSI can overcome infertility associated with TEX44 loss. Consistently, clinical pregnancies were achieved through ICSI-based assisted reproductive technology for two probands in this study. Thus, ICSI appears to be a viable option for overcoming infertility in asthenozoospermia patients with deleterious TEX44 variants.

The identification of pathogenic TEX44 variants in patients, together with phenotypes observed in knockout mouse models, underscores the essential role of TEX44 in mitochondrial sheath assembly. Notably, a recent study by Dupuis et al.³⁸ also reported mitochondrial sheath abnormalities and subfertility in Tex44 knockout mice³⁸, consistent with our findings. While that study focused on phenotypic characterization, we further investigated the molecular mechanism of TEX44 function. Through IP-MS analysis, we revealed putative interactions between TEX44 and CPT1B, SEC16 Homolog A (SEC16A), CDH2, and CTNNB1. CPT1B, an outer mitochondrial membrane protein, is a member of the carnitine palmitoyltransferase family involved in transporting cytoplasmic long-chain fatty acyl-CoA into mitochondria³⁹. Co-IP and functional assays confirmed the interaction between TEX44 and CPT1B, with both proteins localizing to the mitochondria in spermatozoa and HeLa cells. TEX44 was also found to connect adjacent mitochondria, and when cells were transfected with TEX44 alone, a decrease in the frequency of mitochondrial-localized TEX44 was observed. Furthermore, we generated Cpt1b germ cell knockout mice and found that these mice exhibited similar sperm deformities to Tex44 knockout mice. Although we did not quantify the exact number of mitochondria per spermatozoon, SEM analysis of mouse mutant sperm revealed that mitochondria remained orderly aligned along the midpiece. Mitochondrial defects were predominantly localized to the distal end of the sheath, near the annulus, where partial mitochondrial loss was consistently observed. This suggests that although mitochondrial size appears unaffected, fewer mitochondria are incorporated into the distal sheath near the annulus, indicating a defect in local assembly rather than global mitochondrial biogenesis. During spermiogenesis, TEX44 failed to localize to the mitochondrial sheath in Cpt1b knockout spermatozoa, further supporting the role of TEX44 in anchoring to the mitochondrial sheath through CPT1B. These results reveal a novel function of CPT1B beyond its role as a carnitine palmitoyltransferase.

This aligns with recent cryo-electron microscopy data showing that the mitochondrial outer membrane proteins Voltage Dependent Anion Channel 2/3 (VDAC2/3) can anchor mitochondria to the cytoskeleton in mammalian spermatozoa^40–43. Therefore, a series of outer mitochondrial membrane proteins may anchor different subcellular structures during mitochondrial sheath formation. In the mitochondrial sheath anchoring model, VDAC2/VDAC3 acts as a mitochondrial-cytoskeleton anchor point, participating in the mitochondrial sheath anchoring process mediated by proteins such as Glutathione Peroxidase 4 (GPX4) and Glycerol Kinase 2 (GK2)^44–47. Additionally, Armadillo Repeat Containing 12 (ARMC12) interacts with VDAC2/3 and is essential for proper mitochondrial elongation and coiling along the flagellum during mitochondrial sheath formation⁴². Unlike impaired mitochondrial sheath compaction due to Vdac3 knockout, the mitochondrial sheath abnormality due to TEX44 defects mainly manifests as abnormal mitochondrial distribution. Therefore, the mitochondrial sheath formation disorder caused by TEX44 deficiency is an independent mechanism mediated by binding with CPT1B. It is noteworthy that we observed incomplete mitochondrial sheath abnormalities, similar to those seen in Vdac3-deficient mice⁴³. Despite the partial loss of the mitochondrial sheath and reduced sperm motility, both Cpt1b germ cell knockout and Tex44 knockout male mice are capable of producing some offspring. This phenomenon is similar to another mouse model with partial mitochondrial sheath loss, Kastor knockout or Kastor/Polluks dKO⁴³. These results suggest that the mitochondrial sheath assembly process involves various complexes, and there may be compensatory mechanisms between them.

Metabolically, significant differences are observed between Cpt1b knockout and Tex44 knockout sperm. Metabolomic analysis revealed a significant increase in long-chain fatty acyl-CoA levels in Tex44 knockout sperm, whereas the opposite trend was observed in Cpt1b knockout sperm. Under culture conditions with palmitoyl-CoA and carnitine, Tex44 knockout sperm displayed higher ROS levels and more severe sperm damage. Although germ cell knockout of CPT1B also leads to decreased sperm motility, unlike Tex44 knockout mice, Cpt1b knockout does not significantly affect sperm ROS damage compared to the control group. Furthermore, studies have shown that the addition of long-chain fatty acids to healthy sperm improves their function, suggesting that sperm are primed to utilize lipid-based nutrients¹⁷. At the same time, sperm have evolved a unique mechanism to minimize ROS generation during this process by tightly regulating fatty acid utilization. Our findings reveal that TEX44 modulates the low-intensity enzymatic activity of CPT1B, uncovering a distinctive mechanism that controls fatty acid β-oxidation and ROS production, thereby ensuring the protection and functional integrity of sperm.

Both mutant mice and affected individuals display comparable motility defects and mitochondrial abnormalities, yet species-specific differences in phenotypic presentation can still be observed.

Specifically, while patients carrying biallelic TEX44 mutations exhibited reduced sperm counts, Tex44^−/− mice maintained normal sperm count. The underlying cause of this discrepancy remains unclear. In our study, we observed increased ROS production in Tex44^−/− mouse sperm. We cautiously speculate that human sperm may be exposed to more pathways for accumulating ROS damage compared to mouse sperm raised in a specific-pathogen-free (SPF) environment, potentially resulting in more severe effects on human sperm. Additionally, we observed significant individual variability in sperm concentration among human carriers of homozygous TEX44 variations. This variability suggests that changes in sperm concentration may not be solely attributed to genetic factors.

Another visible difference is the presence of bent tail morphology in Tex44^−/− mouse sperm, which was not prominent in human sperm carrying TEX44 mutations. Importantly, a partial absence of mitochondria was observed at the distal end of the midpiece near the annulus in both species. The bent tail morphology observed in the Tex44^−/− mouse sperm is therefore considered a secondary mechanical consequence of this mitochondrial discontinuity. We speculate that this apparent morphological difference arises from inherent anatomical and mechanical differences between mouse and human sperm. Notably, the mouse sperm flagellum, including the mitochondrial sheath, is approximately twice as long as that of humans. As a result, structural defects at the distal region of the mitochondrial sheath in mice may lead more readily to mechanical instability, which manifests as bent tail deformities. In contrast, the shorter human sperm flagellum may be less susceptible to such bending under similar structural defects. Therefore, we believe that the observed difference reflects species-specific mechanical properties rather than a divergence in the core phenotype.

Methods

Ethics statement

All experiments involving human participants and animals were conducted in accordance with relevant ethical regulations. Human studies were approved by the Ethics Committees of Shanghai General Hospital (2021-SQ-112), the Reproductive and Genetic Hospital of CITIC Xiangya (LL-SC-2017-025 and LL-SC-2019-034), and the First Affiliated Hospital of Nanjing Medical University (2019-SR-472), with written informed consent obtained from all participants. All mice experiments were approved by the Ethics Committee of Nanjing Medical University (Approval No. IACUC-1810020).

Study participants

In total, 800 males with asthenozoospermia diagnosed as per the 5th Edition WHO Laboratory Manual for the Examination and Processing of Human Semen were recruited from the Department of Andrology, Urologic Medical Center, Shanghai General Hospital, Shanghai Jiao Tong University (Shanghai, China), the Reproductive and Genetic Hospital of CITIC Xiangya (Changsha, Hunan, China) and the First Affiliated Hospital of Nanjing Medical University (Nanjing, Jiangsu, China). Only individuals with idiopathic asthenozoospermia were eligible for inclusion in the present study, whereas individuals with Klinefelter syndrome, AZF genomic deletions, or other potential causes of infertility, including testicular cancer, chemotherapy exposure, radiation exposure, orchitis, varicocele, or cryptorchidism, were excluded from our analyses. All patients exhibited normal findings upon physical examination, including an unremarkable mental state, body mass index (BMI), external genitalia, and distribution of pubic hair. In addition, 157 male Han Chinese individuals with normal semen parameters who had fathered one or more healthy children were recruited as matched healthy controls.

Whole-exome sequencing and data analysis

Whole-exome sequencing was performed on genomic DNA extracted from peripheral blood samples using the FastPure Blood DNA Isolation Kit (Vazyme, DC111-01). Exonic regions and flanking sequences were captured with the SureSelectXT Human All Exon V6 kit (Agilent Technologies, USA) and sequenced on a NovaSeq 6000 platform (Illumina) with 150-bp paired-end reads. The resultant reads were aligned to the GRCh37/hg19 human reference genome. Insertions/deletions (indels) and single-nucleotide variations (SNVs) were identified with the SAM, LoFreq, VarScan, and FreeBayes tools, with ANNOVAR being used for variant filtering and annotation. Those variants identified as essential splice-site, frameshift, nonsense, or potentially deleterious missense variants were retained for subsequent study.

Analyses of semen parameters and sperm morphology

Semen samples from identified individuals harboring TEX44 variants were obtained as per 5th Edition WHO guidelines via masturbation following abstinence for 3–7 days. Following a 30 min incubation at 37 °C to permit liquefaction, routine analyses of semen volume, sperm concentrations, and sperm motility were conducted. H&E staining was used to assess sperm morphology, with a minimum of 300 spermatozoa per patient being assessed to reliably establish the frequency of spermatozoa exhibiting morphological abnormalities.

For analyses of C57BL/6 mice, sperm were harvested via the dissection of cauda epididymis from 8-week-old male mice. Samples were then incubated for 30 min at 37 °C in modified HTF medium (FUJIFILM Irvine Scientific, 90126) supplemented with 10% FBS (Gibco, 16000-044) following dilution to a 1 mL volume, with a computer-assisted sperm analysis (CASA) system and H&E staining then being used to characterize these sperm. A minimum of 3 mice were assessed in each experimental group.

Electron microscopy

Briefly, sperm were routinely fixed, embedded with Epon, cut into ultrathin sections, stained using uranyl acetate and lead citrate, and Discoverers at an 80 kV accelerating voltage via Transmission Electron Microscope (TECNAI-10, Philips). Mitochondrial sheath abnormalities were quantified via TEM by analyzing a minimum of 50 flagellar cross-sections. For Scanning Electron Microscopy assay, the sperm specimens fixed by 2.5% glutaraldehyde for 2 h at room temperature. Wash samples with 0.1 M PB (pH 7.4) for 3 times, 15 min each. Then transfer samples into 1% OsO4 in 0.1 M PB (pH 7.4) for 1–2 h at room temperature. After that, wash samples in 0.1 M PB (pH 7.4) for 3 times, 15 min each. Samples were sequentially dehydrated using an ascending gradient of ethanol and then dried with Critical Point Dryer (K850, Quorum). Specimens are attached to metallic stubs using carbon stickers and sputter-coated with gold for 30 s. We observed and took images with scanning electron microscope(SU8100, Hitachi).

Immunofluorescent staining

For immunostaining of HeLa cells and spermatozoa, samples were fixed with 4% paraformaldehyde (PFA) and blocked with 1% bovine serum albumin (BSA) for 1 h at room temperature. Testis sections were deparaffinized using standard methods (xylene, absolute ethanol, 90%, 80%, and 70% ethanol, followed by sterile water). Heat-induced antigen retrieval was performed by boiling the slides in 10 mM citrate buffer (pH 6.0) in a microwave oven for 10 min. The slides were cooled to room temperature for 1 h, washed three times with Phosphate Buffered Saline (PBS), and blocked with 1% BSA for 2 h at room temperature. The samples (HeLa cells, spermatozoa, and testis sections) were incubated overnight at 4 °C with the following primary antibodies: mouse monoclonal anti-TOMM20 (Abcam, ab283317, 1:100), rabbit polyclonal anti-SEPT4 (Proteintech, 12476-1-AP, 1:400), rabbit polyclonal anti-TEX44 (prepared in-house, 1:200), monoclonal mouse anti-α-tubulin (Sigma, T6199, 1:200), rabbit polyclonal anti-CPT1B (Proteintech, 22170-1-AP, 1:200), monoclonal mouse anti-FLAG M2 (Sigma, F1804, 1:500), rabbit polyclonal anti-HA (Sigma, H6908, 1:500), and rabbit polyclonal anti-GFP (Abcam, ab13970, 1:500). After incubation with secondary antibodies (Thermo Fisher Scientific, A32766/A32790/A31572/A31570, 1:500), at room temperature for 2 h, samples were co-stained, when indicated, with either PNA (Vectorlabs, RL-1072, 1:1000) or WGA (Invitrogen, W32466, 1:1000) at 37 °C for 1 h. Nuclei were stained with Hoechst 33342 (Thermo Fisher Scientific, H3570) for 5 min, washed three times with PBS, and mounted with glycerol. Samples were analyzed using a TCS SP8X confocal microscope (Leica Microsystems).

Mice

Tex44 knockout mice were generated by Biogle Gene Technology (Jiangsu, China) using CRISPR/Cas9 with an sgRNA targeting the exon 1 (5′-GCGCTCCATGACCAGCCTGC-3′). This sgRNA and Cas9 mRNA were transcribed in vitro using T7 RNA polymerase, mixed, and microinjected into fertilized C57BL/6 mouse eggs. Following a brief in vitro culture period, the resultant blastocysts were implanted in pseudo-pregnant female mice. Sanger sequencing and PCR were used to validate the presence of Tex44 frameshift mutations in these animals.

Cpt1b^flox/flox mice were generated by Cyagen Biosciences Inc. In the targeted allele, exons 8 to 9 were flanked by loxP sites. Stra8-cre mice were from the Jackson Laboratory.

All mice used in this study were of the C57BL/6J background and housed in a specific-pathogen-free (SPF) facility under standard conditions, including ad libitum access to standard chow and hypochlorous weak-acidified water, a 12-h light/dark cycle, ambient temperature maintained at 24 ± 1 °C, and relative humidity at 50  ±  10%.

CASA analyses

Sperm motility was assessed using a CASA system (IVOS SpermAnalyzer, Hamilton-Thorne Research Co., Ltd., USA). Caudal epididymides were dissected from euthanized mice and immersed in 1 mL of mHTF. The tissue was minced with fine scissors and incubated for 30 min at 37 °C in 5% CO₂ to allow spermatozoa release. The sperm suspension was then placed into a disposable sperm analysis chamber (SC100-01-02-A-CE, Leja, The Netherlands) and analyzed using the IVOS system. CASA provided motility parameters including the percentage of motile spermatozoa (defined as those moving at speeds >5 μm/s), the percentage of progressively motile spermatozoa (progressive motility defined by a path velocity >50 μm/s and a straightness ratio >50%), and kinematic parameters: curvilinear velocity (VCL), straight-line velocity (VSL), average path velocity (VAP), linearity (LIN), straightness (STR), wobble (WOB), amplitude of lateral head displacement (ALH), and beat cross frequency (BCF). For motility analysis, 200–1000 spermatozoa were examined per experiment, and the experiments were independently repeated 7–10 times.

IVF and ICSI in Tex44^−/− mice

To conduct IVF and ICSI analyses of Tex44^+/+ and Tex44^−/− mice, 8-week-old animals were used as in prior reports⁴⁸. Briefly, WT females were superovulated via injection at 5 PM with 10 IU of pregnant mare serum gonadotropin (PMSG, Livzon) with a subsequent injection 48 h later with 10 IU of human chorionic gonadotropin (hCG, Livzon). Cumulus-intact oocyte harvesting was then performed after an additional 15–16 h.

IVF was performed by harvesting sperm from cauda epididymis in HTF and combining the harvested cumulus-intact oocytes into drops of fluid containing these sperm for 5 h. Another fertilization drop was then used to wash the resultant murine embryos, which were transferred to M16 media and incubated at 37 °C in a 5% CO₂ incubator. Rates of fertilization were assessed by counting numbers of two-cell embryos after 20 h and numbers of blastocysts after 96 h.

ICSI was performed by separating sperm heads from tails, with the heads then being injected with a piezo-driven pipette into oocytes collected from superovulated female mice. After injection, oocytes were cultured (37 °C, 5%CO₂) in M16 media. Fertilization and blastocysts rates were assessed at 20 h and 96 h, respectively.

Sperm viability assessment

Sperm viability in mice was assessed using the eosin-nigrosin staining kit (G2581, Solarbio, China). Caudal epididymides were dissected from euthanized mice and placed in pre-warmed HTF (MR-070, Sigma, USA) at 37 °C. The tissue was minced using fine scissors and incubated for 15 min at 37 °C to release spermatozoa. The sperm suspension was then adjusted to a concentration of 1 × 10⁶ sperm/mL with HTF. In a sterile microtube, 10 μL of the fresh sperm suspension was mixed with an equal volume of eosin Y solution and incubated for 15 s. Next, 20 μL of nigrosin solution was added, and the mixture was gently agitated and incubated for an additional 30 s. A 20 μL aliquot of the mixture was applied to a glass slide, air-dried, and immediately examined under an optical microscope (DM2500 LED, Leica). Viable spermatozoa were identified by their white heads, whereas non-viable spermatozoa displayed red-stained heads against a purple background. A minimum of 200 spermatozoa per sample were counted to calculate the viability rate.

Co-immunoprecipitation analyses and mass spectrometry

After plating in 100 mm culture dishes, HEK293T cells were co-transfected with appropriate constructs. Following a 36 h incubation, cells were harvested, rinsed using chilled PBS, and lysed using NP40 IP buffer. Lysates were then incubated overnight with anti-FLAG or anti-HA (as appropriate) at 4 °C on a roller shaker, after which antibody-protein complexes were precipitated with Protein G agarose beads. After washing beads three times with chilled PBS, proteins were suspended in 2× SDS-PAGE sample buffer assessed via Western blotting with appropriate antibodies.

A total of 6 samples were analyzed, comprising 3 wild-type (WT) mouse testis samples subjected to TEX44 antibody immunoprecipitation (IP) and 3 WT samples subjected to IgG control IP, with biological replicates (n = 3 per group). Testicular samples were lysed in RIPA buffer (Beyotime, P0013C) supplemented with proteinase inhibitor cocktail at 4 °C for 40 min, followed by centrifugation at 12,000 × g for 40 min. Supernatants were then incubated with Protein A/G magnetic beads (MedChemExpress, HY-K0202) for 1 h to pre-clear the lysates. Subsequently, the supernatants were incubated overnight at 4 °C with either anti-TEX44 antibody or control rabbit IgG (Cell Signaling Technology, 2729). The following day, 50 μL of Protein A/G magnetic beads were added, and the mixture was incubated at 4 °C for 3 h. The beads were then washed with PBST, and immunoprecipitated complexes were divided into two equal samples, with one being used for Western blotting following boiling in SDS loading buffer, and the other was processed for mass spectrometry. For proteomic analysis, 25 mM ammonium bicarbonate (NH₄HCO₃) was added to the IP eluates, and the pH was adjusted to 7–8. Then, 1 M dithiothreitol (DTT) was added to a final concentration of 5 mM, and the mixture was incubated at 56 °C for 25 min. Afterward, iodoacetamide (IAM) was added to a final concentration of 14 mM, and the reaction was allowed to proceed in the dark at room temperature for 30 min. Trypsin was added, and digestion was carried out at 37 °C for 16–18 h. After digestion, the peptide mixture was desalted and then vacuum-dried. The resulting peptide powder was resuspended in 0.1% (v/v) formic acid and analyzed using an EASY-nLC 1200 ultra-high-performance liquid chromatography (UHPLC) system. Mobile phase A was 0.1% (v/v) formic acid in water, and mobile phase B was 0.1% formic acid in acetonitrile. The elution gradient was as follows: 3% to 5% B for 5 s; 5% to 15% B for 23 min 55 s; 15% to 28% B for 21 min; 28% to 38% B for 7 min 30 s; 38% to 100% B for 5 s; and 100% B for 12 min 25 s. The flow rate was maintained at 300 nL/min throughout the gradient. After elution, the peptides were ionized and analyzed on a Thermo Q Exactive HF-X mass spectrometer. The ion source voltage was set to 2 kV. Both precursor ions and their corresponding fragment ions were detected using high-resolution Orbitrap analysis. The scan range for the full-scan MS spectrum was set to 350–1500 m/z, with a resolution of 60000, while the resolution for MS/MS scans was set to 15,000. Data acquisition was performed in data-dependent acquisition (DDA) mode, with the automatic gain control (AGC) for MS set to 3E6 and for MS/MS set to 1E5. A dynamic exclusion time of 30 s was applied for tandem MS scans. The tandem mass spectrometry (MS/MS) data were analyzed using the Proteome Discoverer (version 2.4). The database used for the search was the UniProt mouse proteome (UP000000589). As a digesting enzyme, Trypsin/P was selected with maximal 2 missed cleavages. Cysteine carbamidomethylation was set for fixed modifications, and oxidation of methionine and N-terminal acetylation were specified as variable modifications. A minimum length of 6 amino acids was tolerated. Unique and razor peptides were used for quantification. The precursor and fragment ion mass tolerances were set to 10 ppm and 0.02 Da, respectively. Peptide and protein identifications were filtered at a 1% false discovery rate (FDR).

Western blot

Proteins were extracted from HEK293T cells, spermatozoa, and mice testis using RIPA Lysis Buffer (Beyotime, P0013C) supplemented with a protease inhibitor cocktail (Sigma, 4693132001). SDS sample loading buffer was added to the lysates and boiled at 95 °C for 10 min to denature the proteins. The denatured proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto PVDF (polyvinylidene difluoride) membranes (Bio-Rad). The membranes were blocked with 5% nonfat milk in TBST buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.1% Tween-20) for 2 h at room temperature, followed by incubation overnight at 4 °C with the following primary antibodies: rabbit polyclonal anti-TEX44 (prepared in-house, 1:2000), mouse monoclonal anti-HA (MBL, m180-3, 1:5000), mouse monoclonal anti-FLAG (Sigma-Aldrich, F3165, 1:5000), rabbit polyclonal anti-CPT1B (abcam, ab134988, 1:2000), mouse monoclonal anti-β-Actin (Santa, sc-47778, 1:1000) or mouse monoclonal anti-β-Tubulin (ABclonal, AC021, 1:5000). After three washes with TBST (15 min each), the membranes were incubated with HRP-conjugated anti-rabbit or anti-mouse secondary antibodies (1:5000 dilution) for 2 h at room temperature. Immunoreactive bands were detected using a chemiluminescence imaging system (Tanon).

Plasmid construction

PCR was used to amplify full-length mouse Tex44 (GenBank: NM_027966.1), Tex44-N (1–384aa), Tex44-C (385–530aa), Cpt1b (GenBank: NM_009948.2), Cdh2 (GenBank: NM_007664.5), and Ctnnb1 (GenBank: NM_001165902.2) cDNA sequences. The resulting PCR products were cloned into the pCAG1.1 vector (Thermo Fisher). Full-length Tex44, Tex44-N, and Tex44-C were cloned with a C-terminal FLAG tag, while Cpt1b, Cdh2, and Ctnnb1 were cloned with a C-terminal HA tag for expression. Full-length Tex44 was also subcloned into pCAG1.1 to generate constructs with either a C-terminal HA tag or a C-terminal EGFP tag. A Tex44-C construct with an N-terminal Twin-Strep tag and a C-terminal FLAG tag was also generated for expression and purification in Expi293F cells.

Cell culture and transfection

HEK293T (American Type Culture Collection, CRL-3216) and HeLa (American Type Culture Collection, CCL-2) cells were cultured in high-glucose Dulbecco’s Modified Eagle Medium (DMEM; Thermo Fisher Scientific, C11995500BT) supplemented with 10% (v/v) fetal bovine serum (FBS; Biochannel, BC-SE-FBS01) and 100 U/mL penicillin and 100 µg/mL streptomycin at 37 °C in a humidified atmosphere with 5% CO₂. Transfections were performed at 60% confluency using Lipomaster 2000 Transfection Reagent (Vazyme, TL201) as the transfection reagent, with a plasmid: Lipomaster 2000 Transfection Reagent ratio of 1:2 (w/w). After 6 h, the transfection medium was replaced with complete medium, and the cells were cultured for further analysis by Western blot (WB) and immunofluorescence (IF).

Expi293F cells (National Collection of Authenticated Cell Cultures, SCSP-5207) were cultured in Union293 medium (Union Biotech, UP1003) at 37 °C in a humidified atmosphere with 5% CO₂ on a shaker at 120 rpm. Transfections were performed at a cell density of 2 × 10⁶ cells/mL using polyethylenimine (PEI; Sigma-Aldrich, 919012) as the transfection reagent, with a plasmid:PEI ratio of 1:2.5 (w/w).

Metabolomic analysis

Spermatozoa were washed with PBS at 37 °C and resuspended after the supernatant was removed. Metabolites were extracted using 800 μL of cold methanol/acetonitrile (1:1, v/v), which also served to precipitate proteins. The mixture was transferred to a new centrifuge tube and centrifuged at 14,000 × g for 5 min to collect the supernatant. The supernatant was dried under vacuum centrifugation, and the dried extracts were re-dissolved in 100 μL of acetonitrile/water (1:1, v/v) for LC-MS analysis.

Chromatographic separation was performed on a 2.1 × 100 mm ACQUITY UPLC BEH 1.7 μm column (Waters, Ireland) at 25 °C with a flow rate of 0.5 mL/min and an injection volume of 2 μL. The mobile phase consisted of A: water with 25 mM ammonium acetate and 25 mM ammonium hydroxide, and B: acetonitrile. The gradient was as follows: 0–0.5 min, 95% B; 0.5–7 min, linear decrease to 65% B; 7–8 min, to 40% B; 8–9 min, held at 40% B; 9–9.1 min, increased to 95% B; 9.1–12 min, held at 95% B. Samples were maintained at 4 °C in the autosampler and analyzed in random order. Quality control (QC) samples were inserted throughout the run to monitor instrument stability.

Samples from Tex44^+/+ and Tex44^−/− mice were performed using an Agilent 1290 Infinity LC system coupled with an AB SCIEX TripleTOF 6600 mass spectrometer. MS acquisition was carried out in both positive and negative ESI modes with the following parameters: Ion Source Gas1 and Gas2: 60; Curtain Gas: 30; source temperature: 600 °C; ISVF: ±5500 V. The TOF MS scan range was m/z 60–1000 Da, and the MS/MS scan range was 25–1000 Da. Accumulation times were 0.20 s (MS) and 0.05 s (MS/MS). MS/MS was acquired in information-dependent acquisition (IDA) mode with high sensitivity, DP ± 60 V, CE 35 ± 15 eV, excluding isotopes within 4 Da, and monitoring up to 10 candidate ions per cycle.

Samples from Cpt1b gKO and control mice were analyzed using a Vanquish UHPLC system coupled with an Orbitrap Exploris™ 480 mass spectrometer (Thermo Fisher). ESI was performed in both ion modes with the following settings: Ion Source Gas1: 50; Gas2: 2; source temperature: 350 °C; ISVF: 3500 V (positive), 2800 V (negative). Full-scan MS was acquired over m/z 70–1200 Da at a resolution of 60,000 with 100 ms accumulation time. MS/MS spectra were obtained in data-dependent acquisition (DDA) mode with the same resolution and scan time, and dynamic exclusion set to 4 s. The MS/MS scan range was also m/z 70–1200 Da.

Raw MS data were converted into MzXML format using ProteoWizard MSConvert and processed with XCMS. Peaks were detected using the centWave method (m/z = 10 ppm, peakwidth = c(10, 60), prefilter = c(10, 100)) and grouped with bw = 5, mzwid = 0.025, and minfrac = 0.5. Isotopes and adducts were annotated using CAMERA. Variables with more than 50% non-zero measurements in at least one group were retained. Metabolites were identified by comparing accurate m/z values (<10 ppm) and MS/MS spectra with an in-house database created using authentic standards. Missing data were imputed using the K-Nearest Neighbor (KNN) method, and outliers were removed. The total peak area was normalized to ensure comparability across samples and metabolites.

Protein expression and purification

The Expi293F cells transiently expressing CPT1B, TEX44 full-length (TEX44-FL), or the TEX44 C-terminal (TEX44-C, 385–530aa) were harvested 60 h post-transfection and resuspended in lysis buffer (25 mM HEPES, pH 7.0, 150 mM NaCl, 2 mM EDTA) supplemented with a protease inhibitor cocktail (Roche). For TEX44-FL and TEX44-C, both the lysis buffer and the wash buffer were further supplemented with 5 mM ATP and 10 mM MgCl₂. The cells were then lysed using a high-pressure homogenizer at 300 bar. After lysis, lauryl maltose neopentyl glycol (LMNG, Anatrace, NG310) and cholesteryl hemisuccinate tris salt (CHS, Anatrace, CH210) were added to the lysate to final concentrations of 1% (w/v) and 0.12% (w/v), respectively. The mixture was incubated on a rotating shaker at 4 °C for 2 h to ensure solubilization. The cell lysate was centrifuged at 12,000 × g for 30 min at 4 °C. The supernatant was incubated with anti-FLAG tag affinity resin (Smart-Lifescience) for 2 h at 4 °C, then the resin was washed with 15 column volumes of wash buffer (25 mM HEPES-Na, pH 7.0, 300 mM NaCl, 2 mM EDTA, 0.01% LMNG, and 0.0012% CHS). The protein was eluted with 10 column volumes of elution buffer containing 25 mM HEPES-Na, pH 7.0, 300 mM NaCl, 2 mM EDTA, 0.01% LMNG, and 0.0012% CHS and 250 μg/ml FLAG peptide. The eluates containing CPT1B and TEX44-FL were concentrated and filtered through a 30 kDa molecular weight cut-off centrifugal filter unit (Merck Millipore), while the TEX44-C eluate was concentrated and filtered through a 10 kDa unit (Merck Millipore). The eluates were further purified by a Superose 6 Increase 10/300 GL column (Cytiva) with wash buffer. The peak fractions were collected and used for the CPT1B activity assay.

CPT1B activity assay

The activity of CPT1B was determined spectrophotometrically by monitoring the release of CoA-SH from palmitoyl-CoA using DTNB [5,5′-dithio-bis-(2-nitrobenzoic acid)] (MedChemExpress, HY-15915) as the detection reagent. Purified proteins (CPT1B, TEX44-FL, and TEX44-C) were assayed individually or in combination under the following conditions. For each reaction, the total protein volume was standardized to 20 µl per well. When less than 20 µl of purified protein was used, the remaining volume was adjusted using wash buffer (25 mM HEPES-Na, pH 7.0, 150 mM NaCl, 2 mM EDTA, 0.01% LMNG, and 0.0012% CHS). The reaction buffer was prepared by supplementing the wash buffer with 0.5 mM DTNB. Protein solutions were mixed with 200 µl reaction buffer and incubated at room temperature in 96-well plates (BIOFIL) for 5 min to eliminate all reactive thiol groups pre-existing in the purified proteins. After incubation, the baseline absorbance was measured at 412 nm using a Spark microplate reader (Tecan). The enzymatic reaction was initiated by adding 100 μM palmitoyl-CoA (MedChemExpress, HY-134427) and 1 mM carnitine (MedChemExpress, HY-B0399) as the substrate and cofactor, respectively. Purified CPT1B (at different concentrations: 1.25 µg, 2.5 µg, 5 µg, 10 µg, 20 µg) was used to determine the dose-response relationship. Additionally, combinations of CPT1B (5 µg) with TEX44-FL (at 0 µg, 1.25 µg, 2.5 µg, 5 µg, 10 µg, and 15 µg) or TEX44-C (at 0 µg, 1.25 µg, 2.5 µg, 5 µg, 10 µg, and 15 µg) were tested to investigate potential interactions. After 30 min of incubation at room temperature, the release of CoA-SH from palmitoyl-CoA was spectrophotometrically determined at 412 nm using a Spark microplate reader (Tecan). The absorbance values (ΔA412) were recorded for each condition to evaluate CPT1B activity and its modulation by TEX44-FL and TEX44-C.

ROS detection

Spermatozoa in HTF medium were treated with various concentrations of palmitoyl-CoA (0, 10 nM, 100 nM, and 1 Μm), while the final concentration of L-carnitine was maintained at 1 mM across all samples. After treatment, sperm were stained with 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA, Beyotime, S0033S) to assess intracellular ROS levels. The staining was carried out at 37 °C for 20 min in a light-protected environment, and during the last 5 min, Hoechst was added to counterstain the nuclei. Flow cytometric analysis was performed using a FACSVerse instrument (BD Biosciences). Hoechst-positive spermatozoa were gated to identify the population of interest, and fluorescence from DCFH was measured using FITC settings (excitation: 488 nm, emission: 525 nm). Data acquisition and analysis were conducted with FlowJo V10 software.

Tunel assay

Spermatozoa were treated with 1 μM palmitoyl-CoA and 1 mM L-carnitine in HTF medium at 37 °C for 20 min in a light-protected environment. After treatment, the sperm were washed with 1× PBS and prepared for TUNEL staining. Samples were smeared onto glass slides, air-dried, and fixed with 4% PFA at room temperature for 20 min. After fixation, the slides were washed twice with 1× PBS to remove excess fixative. For the TUNEL assay, sperm smears were permeabilized with PBST (0.3% Triton X-100 in 1× PBS) for 20 min at room temperature. The slides were then washed twice with 1× PBS and equilibrated in 1× equilibration buffer (TUNEL BrightRed Apoptosis Detection Kit, Vazyme, A113) for 30 min. Subsequently, 50 μl of TUNEL reagent mix was applied to each slide, followed by incubation at 37 °C for 60 min in a humidified chamber, as per the manufacturer’s instructions. After TUNEL labeling, the slides were washed three times with PBST (0.1% Triton X-100 in 1× PBS) and counterstained with Hoechst 33342 (5 μg/ml) for 10 min at room temperature. Finally, the slides were mounted with 20% glycerol in 1× PBS and analyzed using a confocal microscope (Leica Microsystems, SP8).

Fluorescence recovery after photobleaching (FRAP) experiments

FRAP experiments were performed on a Zeiss LSM800 confocal microscope equipped with a 488 nm laser line. HeLa cells transiently transfected with a TEX44-EGFP plasmid were seeded on glass-bottom dishes (Corning) and imaged under live-cell conditions using a 63× oil immersion objective. A defined region of interest (ROI) was photobleached using 100% laser power for 1 ms with 1 repetition. Time-lapse images were acquired every 2 s for a total duration of 2 min. Image processing was carried out using FIJI, and fluorescence recovery curves were analyzed with GraphPad Prism 8.3.0.

Web resources

1000 Genomes Project, https://www.internationalgenome.org/; EMBL-EBI Expression Atlas, https://www.ebi.ac.uk/gxa/home; gnomAD, https://gnomad.broadinstitute.org/; GTEx, https://www.gtexportal.org/; Human Protein Atlas, https://www.proteinatlas.org; NCBI, https://www.ncbi.nlm.nih.gov/; OMIM, https://www.omim.org/; Picard, https://github.com/broadinstitute/picard, PolyPhen-2, http://genetics.bwh.harvard.edu/pph2/; SIFT, https://sift.bii.a-star.edu.sg; UCSC Genome Browser, http://genome.ucsc.edu; iProX, https://proteomecentral.proteomexchange.org; MetaboLights, https://www.ebi.ac.uk/metabolights.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

E.Z., H.B., C.R., Y.D., J.Z., Y.X. and C.T. contributed equally to performing most experiments and analyzing the data. M.L. and E.Z. wrote the original draft of the manuscript. M.L.and S.L. reviewed and revised the manuscript. Y.P. performed the sequencing data analysis. Y.X. and L.N.Z. contributed to the prediction and interpretation of complex structures. P.L., Y.H.Z., L.M., J.X., Y.X.Z., S.X., Z.J. and L.Y.Z. assisted in the collection and organization of clinical data. Y.T., M.J., Z.L. and C.Y. provided valuable suggestions for the study design and data interpretation. M.L., S.L., X.Y., J.W. and Y.C.Z. conceived, designed, and supervised the project. All authors reviewed and approved the final manuscript.

Peer review

Peer review information

Nature Communications thanks Aminata Touré, who co-reviewed with Maeva Drouault, Magalie Boguenet, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information files, or through public repositories. The raw sequence data of 800 males with asthenozoospermia reported in this paper have been deposited in the Genome Sequence Archive (GSA) in the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences, under accession numbers HRA012711 (project: PRJCA044120) and HRA012746 (project: PRJCA044190). These data are available under restricted access, as individual genomic sequencing data are protected owing to patient privacy and Regulations on the Management of Human Genetics Resources of China. The raw data can be requested via the GSA-Human System and can be authorized for downloading by the Data Access Committee for research and non-commercial use only. The mass spectrometry proteomics data generated in this study have been deposited in the ProteomeXchange Consortium via the iProX partner repository under accession code PXD060989 [https://www.iprox.cn/page/project.html?id=IPX0011141000]. Additionally, the metabolomics data generated in this study have been deposited in the MetaboLights database under accession codes MTBLS12397 and MTBLS12402. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-63280-x.