HFM1 was predominantly expressed in the prenatal embryonic mouse ovary

Immunoblotting was used to examine the protein expression level of HFM1 in embryonic and neonatal mouse ovaries, exploring the relationship between HFM1 and the first meiosis of oocytes. The mouse ovaries of 14.5 days post coitum (dpc), 16.5 dpc, and 18.5 dpc prenatal, and 1 day post-partum (dpp) and 4 dpp were collected. As shown in Fig. 1A, B, the protein expression level of HFM1 was significantly higher in the prenatal than that in the postnatal mouse ovaries.

Fig. 1

HFM1 predominantly expressed in the prenatal embryonic mouse ovary. A Immunoblotting staining showed that HFM1 was relatively highly expressed in the ovaries of embryonic mice at 14.5 days post coitum (dpc), 16.5 dpc, and 18.5 dpc, while the expression was significantly lower at 1 day post-partum (dpp) and 4 dpp after birth. β-actin was used as a loading control. B Quantification of HFM1 gray value. n = 3 biologically independent experiments. Data represented as mean ± standard error of the mean and the different letters (a-c) indicate the difference between the groups was statistically significant (two-sided ANOVA test), P (a, b) = ns, P (a, c) < 0.05, P (b, c) < 0.01. C HFM1 was mainly expressed in the cytoplasm of oocytes in mouse ovaries. Embryonic and neonatal mouse ovaries were stained for HFM1 (green) and germ cell-specific marker DDX4 (red). The nucleus was dyed with DAPI (blue). Scale bars: 50 μm

To further investigate the function of HFM1, immunofluorescence was used to detect the localization of HFM1 in mouse ovaries. The co-staining of HFM1 (green) with the germ cell-specific marker DEAD-Box Helicase 4 (DDX4) (red) revealed that HFM1 was consistently highly expressed in the germ cells of mouse ovaries from 14.5 dpc to 18.5 dpc and primarily localized in the cytoplasm (Fig. 1C). During the postnatal period, the fluorescence intensity of HFM1 gradually weakened as the primordial follicle pool was established. The results showed that HFM1 had a comparatively high expression level in the prenatal period, suggesting that it may play a crucial role in the first meiotic prophase of oocytes as well as primordial follicle formation.

Deficiency of Hfm1 expression impeded oocyte meiotic process and cell survival in mouse ovaries

An adenovirus embedding Hfm1-RNA interference (AD-Hfm1i) was used to culture embryonic mouse ovaries of 14.5 dpc in vitro so as to understand the role of HFM1 in oocyte meiosis and primordial follicle formation. The ovaries transfected with adenovirus showed a strong green fluorescence after 4 days of in vitro culture, indicating successful transfection (Fig. S1A). Subsequent immunoblotting also confirmed that AD-Hfm1i significantly reduced the expression of HFM1 protein levels by half (Fig. S1B and S1C), implying AD-Hfm1i ovarian model was successfully constructed.

After successfully constructing the AD-Hfm1i ovarian model, the effect of HFM1 on germ cell numbers and oocyte meiotic prophase I were first investigated in mouse ovaries. Ovaries of 14.5 dpc were transfected with AD-Hfm1i or scrambled control and cultured for 4 days. The immunofluorescence and germ cell counts revealed the number of oocytes dramatically decreased after the treatment with AD-Hfm1i (Fig. S2A). The number of DDX4-positive oocytes in cultured ovaries was 8585 ± 1955 per ovary in the AD-Hfm1i group and 13,500 ± 1705 per ovary in the control group (P < 0.01; Fig. S2B), respectively.

Moreover, an increased TUNEL signal was observed in the 14.5 dpc AD-Hfm1i ovaries after 4-day cultivation with AD-Hfm1i (Fig. S2C) and the apoptotic rate was significantly higher in the AD-Hfm1i group (30.67%) than in the control group (7.25%) in ovarian sections (P < 0.05; Fig. S2D), suggesting a significant increase in the number of apoptotic cells. In addition, immunoblotting showed that the protein levels of cleaved-CASPASE3, P53, and P63 upregulated in HFM1-deficient embryonic ovaries (Fig. S2E), indicating increased apoptosis in the deficiency of HFM1. These results implied that the vital role of HFM1 in the survival of embryonic oocytes during meiotic prophase I and the deletion of HFM1 leads to extensive oocyte apoptosis.

Immunofluorescence was performed against Y box protein 2 (MSY2), which specifically presented in oocytes of the diplotene and afterward stages, to further evaluate the impact of HFM1 on the meiotic process of oocytes. The results showed that almost all the oocytes in the control group were positive for MSY2 on the fourth day of culture, while the number of MSY2-positive oocytes distinctly declined in the AD-Hfm1i group to about two thirds of that in the control group (Fig. S2F and S2G).

The oocyte-specific deletion of the Hfm1 mouse model was successfully established in a previous study [16]. To explore the function of HFM1 in oocytes before primordial follicle formation, a systemic knockout mouse model of HFM1 was established by mating and breeding the oocyte-specific deletion of the Hfm1 mouse model (Fig. 2A, B). The Hfm1−/−, GDF9-Cre+/ − female mice were regarded as Hfm1 KO mice, while Hfm1−/+, GDF9-Cre+/ − and Hfm1+/+, GDF9-Cre+/ − females were regarded as control mice. Immunoblotting validated that hardly any HFM1 was expressed in the KO mice (Fig. S3A and S3B).

Fig. 2

Deficiency of Hfm1 expression impeded oocyte meiotic process and cell survival in mouse ovaries. A Engineered a conditional floxed allele for Hfm1 and a Cre-mediated recombination to delete exons 6 and 7 of Hfm1 in mice. B Schematic of the specific mating and breeding method to obtain systemic Hfm1-KO mice. C Deletion of Hfm1 disrupted oocyte survival and early folliculogenesis in mice. Immunofluorescence staining showed ovaries of the Control and KO mice at the indicated developmental stages (14.5, 16.5, and 18.5 dpc and 1 dpp). Oocytes were stained with DDX4 (green). The nucleus was stained using DAPI (blue). Scale bars: 50 μm. D Apoptotic cells increased in KO ovaries compared with the Control ovaries with the development of oocytes. TUNEL signals (green) marked apoptotic cells, while the nucleus was stained using DAPI (blue). Scale bars: 50 μm. E, F Statistical analysis of total numbers of germ cells per ovary. (E) and the percentages of TUNEL+ cells per section (F) between Control and KO mice in the indicated developmental stages. *P < 0.05, **P < 0.01 (t test), n = 3. G First meiotic prophase in the Hfm1-KO mice at 1 dpp was arrested before the diplotene phase. The sections were stained with MSY2 (green) which specifically presented in oocytes of the diplotene and afterward stages and DDX4 (red). The nucleus was stained using DAPI (blue). The oocytes circled in the dashed line highlighted oocytes with no expression of MSY2. Scale bars: 50 μm. H Statistical analysis showed that the percentage of MSY2+ oocytes (number of cells both MSY2+ and DDX4+/number of cells DDX4+) per section decreased significantly following HFM1 deprivation. **P < 0.01 (t test), n = 3. I Chromatin spread of 18.5 dpc ovaries showed that the KO group had more oocytes in the pachytene stage and fewer in the diplotene stage than the Control group. *P < 0.05, **P < 0.01 (t test), n = 3

Proliferating cell nuclear antigen (PCNA) was assayed by immunofluorescence with the ovaries of 14.5 dpc, 16.5 dpc, 18.5 dpc, and 1 dpp in the control and KO mice (Fig. S4A). A considerable number of PCNA-positive germ cells were present in 14.5-dpc ovaries. As meiosis proceeded, the number of PCNA-positive germ cells gradually decreased until 1 dpp when nearly all germ cells showed no positive PCNA (Fig. S4B). Moreover, the apoptosis and survival of germ cells were also studied in Hfm1 KO mice. Similarly, immunofluorescence was performed on the ovaries of 14.5 dpc, 16.5 dpc, 18.5 dpc, and 1 dpp mice. The results revealed that as the meiotic process of oocytes advanced, the number of apoptotic cells in the KO mice increased significantly and the number of oocytes decreased significantly compared with those in the control mice (Fig. 2C, D). The number of oocytes in the KO mice decreased by half at 1 dpp compared with that in the control group (Fig. 2E). Similarly, the rate of cell apoptosis increased significantly in the KO mice (Fig. 2F).

Dual-color immunofluorescence of 1 dpp ovaries in KO and control mice with DDX4 and MSY2 showed results comparable to those in AD-Hfm1i ovaries (Fig. 2G). Most of the 1-dpp oocytes in the control demonstrated MSY2 positivity, while only 50% of oocytes showed MSY2 positivity in the Hfm1-KO group, exhibiting a statistically significant difference (Fig. 2H). We then performed chromatin spread and immunofluorescence of the axial element, synaptonemal complex protein 3 (SYCP3), to analyze the meiotic process in 18.5-dpc ovaries (Fig. S3C). Half of the oocytes in the control mice were in the diplotene stage (49.17%), whereas only one third of oocytes in the KO mice progressed to the diplotene stage Fig. 2I.

Depletion of Hfm1 expression damaged the repair of DNA double-strand breaks and synaptonemal complex formation in Hfm1-KO mice

Evidence supports that the homologous recombination of chromosomes in the meiotic prophase I originates from DNA double-strand breaks (DSBs) [21]. The immunofluorescence against DNA damage repair marker protein γ-H2AX and DNA break repair protein RAD51 was performed on the spread chromosome of 18.5-dpc mouse ovaries. The deletion of Hfm1 sustained γ-H2AX focus on the oocyte chromosomes, indicating the presence of unrepaired DSBs on the chromosomes (Fig. 3A). More chromosomes with abnormal γ-H2AX signals were observed in the KO mice compared with the control mice (Fig. 3B). Meanwhile, the RAD51 focus on chromosomes was significantly more in the HFM1-deficient oocytes (14.6 ± 7.6) than in the control oocytes (3.7 ± 3.4, Fig. 3C, D). Also, the fluorescence of SYCP3, a component of the synaptonemal complex (SC), in the KO mice seemed to diminish compared with that in the control mice. To figure out whether synaptonemal complexes is affected, key component proteins (SYCP1, SYCE1, REC8, and STAG3) of the synaptonemal complex were also examined by immunofluorescence. The fluorescence intensity of key proteins, such as SYCP1, REC8, SYCE1, and STAG3, was significantly reduced in HFM1-deficient oocytes (Fig. 3E). Immunoblotting also showed significantly reduced expression in REC8 and SYCE1 of knockdown group, which proponent of above result (Fig. 3F). The results suggested that synaptic defects were evident in the chromosomes of the KO mice and HFM1 deletion led to the disruption of synaptonemal complex formation and thus disorders of DNA break repair and chromosome synapsis.

Fig. 3

Depletion of Hfm1 expression damaged the repair of DNA double-strand breaks and synaptonemal complex formation. A, B Deletion of Hfm1 caused DSB repair deficiency in embryonic mouse ovaries. (A) Immunoblotting staining of the meiotic spread showed repaired or unrepaired DSBs in 18.5-dpc Control or KO ovaries. γ-H2AX (red) indicates unrepaired DSB sites. SYCP3 (green) demonstrates axial elements. Scale bars: 10 μm. (B) Statistical analysis showed that the mean fluorescence intensity of γ-H2AX on chromosomes per nucleus increased significantly following HFM1 deprivation. ***P < 0.001 (t test), WT: n = 31; KO: n = 41. C, D Deletion of Hfm1 resulted in the ectopic expression of RAD51. (C) Immunoblotting staining showed normal or ectopic RAD51 (DNA break repair protein) focus in 18.5-dpc Control or KO ovaries. Oocyte chromosomes were co-stained with RAD51 (red) and SYCP3 (green). Scale bars: 10 μm. (D) Statistical analysis showed that the number of RAD51 foci on chromosomes per slide increased significantly following HFM1 deprivation. ***P < 0.001, WT: n = 42; KO: n = 19. E Schematic demonstrating the synaptonemal complex during the first meiotic prophase (left). Deletion of Hfm1 impacted the formation of the synaptonemal complex. Immunoblotting staining of the meiotic spread showed abnormal expression of synaptonemal complex proteins SYCP1, SYCE1, REC8, and STAG3. Arrows demonstrated loose bivalent chromosome (right). Scale bars: 10 μm. F Immunoblotting showed significantly reduced expression in REC8 and SYCE1 of KO group (n = 3). β-Actin or Vinculin was used as a loading control

HFM1 may regulate the first meiotic prophase of the oocyte by interacting with FUS

The Co-IP following silver stain was performed on mouse ovaries at 17.5 dpc to identify the targets of HFM1, revealing that HFM1-IP-enriched proteins were concentrated at around 70 kDa (Fig. 4A). Totally, 445 proteins were identified by mass spectrometry (MS) (Table. S1). GO analysis suggested genes mainly enriched in “bind”, “cell part”, and “cellular process” (Fig. 4B). KOG analysis showed genes enriched in “translation, ribosomal structure and biogenesis”, “translational modification, protein turnover, chaperones” (Fig. 4C). We measured the mRNA level of the most abundant proteins, IGKC, FUS, IGHG3, CAPR1, EWS, and MYH10, among all the putative proteins. FUS was significant decreased after knockdown of Hfm1 in ovaries (Fig. 4D). FUS, identified as a putative HFM1-interacting protein by MS, has been reported to regulate DSB repair [22, 23] (Fig. 4E). Further Co-IP combined with immunoblotting was performed in embryonic mouse ovaries for validation the co-precipitation of endogenous HFM1 with FUS and vice versa (Fig. 4F). In addition, an in vitro binding assay demonstrated that exogenously expressed HFM1 interacted with FUS (Fig. 4G, H). Together, these data demonstrated that HFM1 could interact with FUS.

Fig. 4

HFM1 and FUS interacted with each other. A Co-IP and silver staining showed the proteins bound to HFM1. B, C Functional analysis of enriched genes by Co-IP. Gene Ontology (GO) analysis (B) and eukaryotic orthologous groups (KOGs) analysis (C) of enriched genes described the Molecular Function, Cellular Component and Biological Process of the enriched genes. D Real-time PCR showed that the interference of HFM1 expression decreased the expression of FUS, but not IGKC, IGHG3, CAPR1, EWS, and MYH10. **P < 0.01, n = 6. E Mass spectrometry (MS) analysis to determine which proteins bound to HFM1 showed FUS to be an interacting protein. F Endogenous protein interactions of HFM1 and FUS were assessed in embryonic mouse ovary lysates by immunoprecipitation with anti-HFM1 or anti-FUS and evaluated using immunoblotting with indicated antibodies. IgG was used as a negative control. G, H Exogenous protein interactions demonstrated in HEK 293 T cells. HEK 293 T cells transfected with indicated plasmid (Flag-tagged HFM1 plasmid, Myc-tagged FUS plasmid and plasmid vector, separately) and treated with proteasome inhibitor MG132 (10 μM). Cells were lysed with NP-40 and analyzed using Co-IP with Flag or Myc beads followed by immunoblotting

HFM1 inhibited the ubiquitination degradation of FUS and the cytoplasmic–cytosolic localization of FUS

As shown in Fig. 5A, the deletion of HFM1 led to a reduction in the FUS protein level. Further detection of the ubiquitination levels in FUS by adding the protease inhibitor MG132 revealed that the knockdown of HFM1 efficiently elevated the ubiquitination level of FUS (Fig. 5B). Then, the impact of HFM1 depletion on the stability of the FUS protein in the cultured ovaries treated with cycloheximide (CHX), a protein synthesis inhibitor, was examined. The half-life of FUS protein was markedly reduced in the HFM1 low-expression group compared with the control group, and the degradation level of the FUS protein was much higher at a same time point (Fig. 5C, D).

Fig. 5

HFM1 acted on the ubiquitination and degradation of FUS and the cytoplasmic–cytosolic localization of FUS. A HFM1 silencing led to a decrease in FUS protein expression (*P < 0.05, n = 3). GAPDH was used as a loading control. B Lysates from embryonic ovaries transfected with control or AD-Hfm1i, followed by treatment with MG132 before harvest, were immunoprecipitated and examined with indicated antibodies. Quantification of relative ubiquitin-FUS levels showed that the ubiquitination level of FUS increased after HFM1 knockdown. C, D Embryonic ovaries were cultured with control or AD-Hfm1i, treated with cycloheximide (CHX, 100 μg/mL), and collected for immunoblotting analysis at the indicated time points (C). Quantification of FUS band intensity was presented (D). *P < 0.05, **P < 0.01 (t test), n = 3. E Venn diagram showed that FBXW11 and MDM2 may be the potential ligating (E3) enzymes during the ubiquitination of FUS using UbiBrowser database (http://ubibrowser.bio-it.cn/ubibrowser/) and the Integrated Interactions Database (http://iid.ophid.utoronto.ca/search_by_proteins/). F Endogenous protein interactions of FBXW11 and FUS were assessed in embryonic mouse ovary lysates by immunoprecipitation with anti-FBXW11 or anti-FUS and evaluated using immunoblotting with indicated antibodies. IgG was used as a negative control. G Lysates from ovaries transfected with control or AD-Hfm1i were collected for Co-IP. The binding of FUS and FBXW11 increased with the knockdown of Hfm1. H, I HFM1 maintained the nuclear localization of FUS in embryonic mouse oocytes. 18.5-dpc embryonic mouse ovaries were stained for FUS (red) and germ cell-specific marker DDX4 (green), while the nucleus was stained using DAPI (blue). The area boxed by dotted line was the oocytes with aberrant localization of FUS (H). Scale bars: 50 μm. Statistical analysis showed the number of germ cells with aberrant-localized FUS per section (I). ***P < 0.001 (t test), n = 6

The ubiquitination degradation of protein requires the activation of ubiquitin (Ub) by an activating (E1) enzyme and transfer onto a conjugating (E2) enzyme. The ligating (E3) enzyme then attaches to the Ub bound to E2 to the substrate protein, and subsequently the proteasome specifically binds to and degrades the Ub-loaded protein [24]. We intersected eight potential E3 ligases of FUS from the UbiBrowser Database (http://ubibrowser.bio-it.cn/ubibrowser/) and 446 putative FUS interaction proteins from the Integrated Interactions Database (http://iid.ophid.utoronto.ca/search_by_proteins/) to identify ligases in the ubiquitination of FUS. Two possible E3 ligases, FBXW11 and MDM2, were obtained (Fig. 5E). Co-IP using fetal ovaries verified the mutual binding between FUS and E3 ligases, FBXW11 and MDM2 (Fig. 5F and Fig. S5A). Subsequently increased binding of FUS with FBXW11 was observed using Co-IP in the absence of HFM1, indicating an increased FBXW11-mediated ubiquitination degradation level of FUS (Fig. 5G) while binding of FUS and MDM2 did not changed significantly with the depletion of HFM1 (Fig. S5B).

In addition, the immunofluorescence showed that the localization of FUS was altered in part of embryonic mouse ovaries after knockout of Hfm1 at 18.5 dpc. FUS was widely expressed in the nuclei of oocytes in the control embryonic ovaries, but was aberrantly located in the cytoplasm of some oocytes in the ovaries of the KO mice (Fig. 5H, I). Interestingly, immunofluorescence of chromatin spread showed that HFM1 expressed not only in cytoplasm, but also in spots on chromosome axes (Fig. S5C). Consistent with that, HFM1 foci was found on chromosome axes in spermatocyte [25]. These results suggested that HFM1 contributed to maintaining the localization of FUS in the nucleus of oocytes during the meiotic prophase I.

BRCA1 might be the target of the HFM1–FUS axis

Two putative FUS-interacting proteins, cell cycle–associated protein 1 (CCPRIN1) and breast cancer susceptibility gene 1 (BRCA1), were obtained by the intersection of the FUS-interacting proteins retrieved from Genemania Database (http://genemania.org/), BioGrid Database (https://thebiogrid.org/), STING Database (https://string-db.org), and Integrated Interactions Database (http://iid.ophid.utoronto.ca/search_by_proteins/) (Fig. 6A). The FUS-interacting proteins are listed in Supplementary Table 5. Further, AD-Hfm1i and adenovirus Fus (AD-Fus) were used to transfect ovaries at 14.5 dpc. After 4 days of culture, the mRNA levels of key genes for meiosis and oocyte development were measured using real-time PCR. As shown, the mRNA level of Brca1 reduced, accompanied by the absence of HFM1, but increased after the overexpression of FUS, while ccprin1 showed little change (Fig. 6D, E). The important role of BRCA1 in DSB repair and recombination has been reported, and its dysfunction is associated with POI [26, 27]. Co-IP in embryonic ovaries verified the combination of FUS with BRCA1 (Fig. 6B). Additionally, the immunofluorescence showed that BRCA1 also co-localized with FUS in the nucleus of some oocytes (Fig. 6C). Furthermore, the mRNA levels of oocyte development–related genes Gdf9, Bmp15, Jag1, Figla, Nobox, Sohlh1, Amh, and Lhx8 (Fig. 6F) and meiosis-related genes Atm, Atr, Dazl, Msh4, Rad51, Rec8, and Smc3 (Fig. 6G) significantly decreased after the depletion of HFM1, but they were partially restored after the overexpression of FUS.

Fig. 6

BRCA1 might be a possible target of the HFM1-FUS axis. A Venn diagram showed that BRCA1 and CAPRIN1 were the binding proteins of FUS using BioGrid, IID, GeneMANIA, and STRING. B Lysates from embryonic ovaries were immunoprecipitated and examined with indicated antibodies to assess the endogenous protein interactions of BRCA1 and FUS. C BRCA1 co-localized with FUS in some oocytes of 18.5-dpc embryonic mouse ovaries. BRCA1 was stained with green, and FUS was stained with red. The nucleus was stained using DAPI (blue). The oocytes in which BRCA1 co-localized with FUS are highlighted in dashed boxes or pointed by arrows. Scale bars: 10 μm. D Real-time PCR showed that the interference of HFM1 expression decreased the expression of Brca1, and the overexpression of FUS restored the mRNA level of Brca1. E Real-time PCR showed interference of HFM1 expression would not change the expression of ccprin1. F, G HFM1 regulated the expression of oocyte development-related factors (F) and meiosis-related factors (G) by affecting FUS-BRCA1. Real-time PCR showed that the interference of HFM1 expression decreased the expression of those genes, while the overexpression of FUS restored the expression of these genes appropriately. Compared with the control group, *P < 0.05, **P < 0.01, ***P < 0.001; compared with the AD-Hfm1-RNAi group, #P < 0.05, ##P < 0.01,.###P < 0.001, n = 4