FBXO24 deletion causes abnormal accumulation of membraneless electron-dense granules in sperm flagella and male infertility

Ribonucleoprotein (RNP) granules are membraneless electron-dense structures rich in RNAs and proteins, and involved in various cellular processes. Two RNP granules in male germ cells, intermitochondrial cement and the chromatoid body (CB), are associated with PIWI-interacting RNAs (piRNAs) and are required for transposon silencing and spermatogenesis. Other RNP granules in male germ cells, the reticulated body and CB remnants, are also essential for spermiogenesis. In this study, we disrupted FBXO24, a testis-enriched F-box protein, in mice and found numerous membraneless electron-dense granules accumulated in sperm flagella. Fbxo24 knockout (KO) mice exhibited malformed flagellar structures, impaired sperm motility, and male infertility, likely due to the accumulation of abnormal granules. The amount and localization of known RNP granule-related proteins were not disrupted in Fbxo24 KO mice, suggesting that the accumulated granules were distinct from known RNP granules. Further studies revealed that RNAs and two importins, IPO5 and KPNB1, abnormally accumulated in Fbxo24 KO spermatozoa and that FBXO24 could ubiquitinate IPO5. In addition, IPO5 and KPNB1 were recruited to stress granules, RNP complexes, when cells were treated with oxidative stress or a proteasome inhibitor. These results suggest that FBXO24 is involved in the degradation of IPO5, disruption of which may lead to the accumulation of abnormal RNP granules in sperm flagella.

(A) Expression profile of mouse Fbxo24 in spermatogenic cells is shown.Expression profiling was obtained from the testis single cell RNA-seq data set (Hermann et al., 2018)       Spermatozoa were obtained Fbxo24 heterozygous or KO male mice.Immature spermatogenic cells or somatic cells were rarely observed.9

Figure S1 .
Figure S1.Characterization of mouse and human FBXO24.

Figure S2 .
Figure S2.Histological analyses of Fbxo24 KO testis and epididymis.(A) Gross morphology of testes obtained from Fbxo24 heterozygous and KO males.(B) Test weight (mg)/body weight (mg) of Fbxo24 heterozygous and KO mice.(C) PAS-hematoxylin staining of testes and cauda epididymis sections.(D) Spermatozoa obtained from the cauda epididymis were stained for IZUMO1 (magenta).Nuclei were stained with Hoechst 33342 (blue).

Figure S3 .
Figure S3.Lack of Fbxo24 impairs sperm viability and the acrosome reaction.(A) Viability of cauda epididymal spermatozoa was assessed with propidium iodide (PI) at 10 min and 120 min incubation in TYH medium.(B) The acrosome reaction rates of cauda epididymal spermatozoa were assessed using RBGS Tg mice after 10 min and 120 min of incubation in TYH medium.To induce the acrosome reaction, Ca 2+ ionophore A23187 was added to the medium after 120 min of incubation.(C) MS analyses of triton X-100 soluble proteins obtained from mature spermatozoa.Quantitative value of identified SNARE-related proteins and PLCD4 were listed.(D) Immunodetection of STX2 in Fbxo24 WT and KO spermatozoa.IZUMO1 was used as loading control.(E) Immunodetection of LY6K and ADAM3 in Fbxo24 WT and KO spermatozoa.SLC2A3 was used as loading control.

Figure S4 .
Figure S4.Sperm flagellum ultrastructure was disorganized in Fbxo24 KO mice.(A) Localization of the annulus in cauda epididymal spermatozoa.Anti-SEPT4 antibody was used to visualize the annulus (green).Nuclei were detected with Hoechst 33342 (blue).The acrosome was detected with anti-IZUMO1 antibody (magenta).(B) Transmission electron microscopy (TEM) observation of the principal piece.A white arrowhead indicates a disruption of the axoneme and outer dense fiber (ODF).(C) Percentages of morphologically abnormal fibrous sheaths (FS) observed with TEM.The number of flagellar sections analyzed is shown above each bar.(D) Percentages of morphologically abnormal axonemes (AX) observed with TEM.The number of flagellar sections analyzed are shown above each bar.(E) Percentages of morphologically abnormal ODF observed with TEM.The number of flagellar sections analyzed are shown above each bar.(F) TEM observation of the principal piece.A red arrowhead indicates electron-dense granules.(G) Percentages of electron-dense granules observed in principal piece cross sections.The number of flagellar sections analyzed are shown above each bar.

Figure S5 .
Figure S5.No clear differences were found in the amount and localization of RNP granulerelated proteins between control and Fbxo24 KO testes.(A) Immunodetection of ADAD1, ADAD2, MILI, MIWI, RNF-17, YTHDC2, TSKS and TSSK1 in Fbxo24 heterozygous and KO testes.β-actin was used as loading control.(B) Immunofluorescence observation of MIWI (green) in WT and Fbxo24 KO testes.Acrosomes were stained with PNA (red) and nuclei were stained with Hoechst 33342 (white).(C) Immunofluorescence observation of TSKS (green) in WT and Fbxo24 KO testes.Acrosomes were stained with PNA (red) and nuclei were stained with Hoechst 33342 (white).

Figure S6 .
Figure S6.Generation and analyses of Fbxo24-3×FLAG Tg mice.(A) Schematic of the Fbxo24-3×FLAG construct used to generate Tg mice.(B) Immunodetection of FBXO24-3×FLAG in the testis from the Tg mice.β-actin was used as a loading control.(C) Morphology of cauda epididymal spermatozoa.(D) Schematic for generating Fbxo24 KO #2 mice using the CRISPR/Cas9 system.White boxes indicate untranslated regions while black boxes indicate protein coding regions.The gRNAs used are shown.Fw and Rv indicate the forward and reverse primer used for genotyping, respectively.(E) Genotyping of obtained Fbxo24 mutant mice.Fw #3-Rv #3 primers in Fig. S6D were used.N.C.indicates negative control (water).(F) Predicted FBXO24 amino acid sequences of KO #2 mutant mice.The 205 bp deletion and 1 bp insertion resulted in E32D mutation with a premature stop codon introduced 49 amino acids later.(G) Morphology of mature spermatozoa obtained from cauda epididymis.(H) The list of identified proteins by IP-MS analysis.The top 15 identified proteins in Tg are shown.The QuantitativeValue (Normalized Total Spectra) was calculated using Scaffold proteome software.IgG-related proteins, which is likely from the antibody used for IP, were removed from the list.