LncRNA DANCR counteracts premature ovarian insufficiency by regulating the senescence process of granulosa cells through stabilizing the interaction between p53 and hNRNPC

To identify the expression of DANCR in ovary, human and mice primary granulosa cells were collected and carried with FISH, indicating the expression of DANCR in GCs (Fig. 1A). The function of ovarian granulosa cells is bound up with follicle development and POI process, knockdown of DANCR might influence granulosa cells, leading to follicle atresia. POI develops in parallel with characteristics of granulosa cells aging [27]. Based on our hypothesis mentioned above that DANCR may be implicated in the process of POI, we detected the DANCR level in granulosa cells of POI patients and normal women. The clinical basic characteristics of the two groups are not significantly different, except for anti-Müllerian hormone (AMH), antral follicle count (AFC) and retrieved oocytes count (Table 1). Remarkably, the expression of DANCR was significantly reduced in POI patients’ ovarian granulosa cells (Fig. 1B). Additionally, Dancr expression was distinctly decreased in old (32 weeks) C57BL/6 mice’s primary granulosa cells compared with the young (8 weeks) (Fig. 1C). DANCR in human granulosa tumor cell lines (KGN and COV434) was silenced via lentivirus transfection and the efficiency was confirmed by qPCR (Fig. 1D). The senescence-associated β-galactosidase (SA-β-Gal) staining found that DANCR-knockdown resulted in a significantly higher percentage of aging granulosa cell (KGN and COV434) in vitro (Fig. 1E and F). Altogether, these results indicated that DANCR was expressed in GCs and decreased DANCR was associated with POI and ovarian granulosa cells aging.

We further investigated the influence of DANCR to aging-related function changes in KGN and COV434 cell lines. The CCK-8 assays indicated cell viability of shDANCR group was visibly reduced at 48 h and 72 h compared with the control (Figs. 2A and 3A). Meanwhile, with knockdown of DANCR, the percentage of EdU-positive cells was significantly lower than the control (Figs. 2B and 3B). Both of the experiment results suggested DANCR is critical for granulosa cells proliferation. Interestingly, further experiments showed that DANCR-knockdown granulosa cells underwent G1 arrest (Figs. 2C and 3C) and more DNA damage (Figs. 2D and 3D), which both were one of the indications of cell aging. P53 functioned as a transcription factor, plays a central role in cellular senescence, governing the cell cycle arrest and senescence-associated secretory phenotype (SASP) independently. Also, p21, the p53 target gene, induces a senescent-like growth arrest [28‐30]. In our study, knockdown of DANCR significantly upregulated p53 and p21 expression (Figs. 2E and 3E), which demonstrated silencing DANCR accelerate granulosa cell aging in another point of view.

To investigate whether aberrant Dancr expression could lead to POI phenotypes and contribute to granulosa cell aging in vivo, Dancr conventional knockout mice were constructed by cre-loxp system. qPCR analysis identified the Dancr^−/− mice (Fig. 4A). Further, in situ hybridization (ISH) assays was performed in mice ovarian tissues with DANCR-specific probe. The results demonstrated that DANCR was expressed specifically in ovarian granulosa cells of Dancr^+/+ mice but not expressed in Dancr^−/− mice (Fig. 4B). Dancr^−/− mice produced significantly fewer offspring than Dancr^+/+ mice per litter, and some even lost their fertility (Fig. 4C). Considering the fertility decline of Dancr^−/− mice, we further assessed the ovarian function of Dancr^−/− mice by multiple methods. About 75% Dancr^−/− mice showed disturbance of estrous cycle, characterized by prolonged estrus, prolonged diestrus or prolonged estrus and diestrus, and the ratio of disordered estrous cycle was significantly higher than Dancr^+/+ mice (Fig. 4D). Hormones reflecting ovarian reserve in Dancr^−/− were disturbed compared with Dancr^+/+. Concretely, follicle stimulating hormone (FSH) was significantly higher while AMH and basic estradiol (E₂) was significantly lower in Dancr^−/− group (Fig. 4E). We performed HE-staining with sections of mice ovary tissues and counted the follicle number. It was discovered that the percentage of atresia follicles was sharply increasing in Dancr^−/− mice, but the percentage of primordial follicle and antral follicle was slightly declining (Fig. 4F), which is coincident with the pathological features of POI [4]. Further, we isolated and collected ovarian granulosa cells from sexually mature Dancr^−/− and Dancr^+/+ mice separately. SA-β-Gal staining and quantitative analysis revealed that aging GCs appeared more notably in Dancr^−/− mice than the controls (Fig. 4G). Moreover, the protein level of p53 and p21 were also elevated in granulosa cells of Dancr^−/−mice (Fig. 4H). In general, Dancr knockout in mice is more likely to induce the occurrence of POI and granulosa cells aging.

To elucidate the mechanisms involved in DANCR deficiency promoting granulosa cells aging and POI, HuProt™ human proteome microarray was applied to detect DANCR binding proteins [31, 32]. Volcano plot showed the binding proteins of DANCR sense and anti-sense (Suppl. Figure 1A). GO enrichment analysis of DANCR binding proteins revealed the top biological process items focused on RNA regulation process (Suppl. Figure 1B & C). 29 DANCR-binding proteins were screened by Z-Score ≥ 2.8, of which only 4 proteins (RAB2B, ELAVL4, hNRNPC, RBM41) specifically bind to DANCR sense chain, 9 proteins specifically bind to antisense chain, and 16 were non-specifically interacted (Table. S1). The candidate protein heterogeneous nuclear ribonucleoprotein C (hNRNPC) was found highly expressed in ovarian tissue based on the results of NCBI database query. Intriguingly, the protein–protein interaction (PPI) network constructed by STRING indicated that only hNRNPC interacted with p53 (Suppl. Figure 1D). Moreover, previous studies identified the interactions between hNRNPC and p53 protein, and suggested hNRNP family including hNRNPC participated in the p53 regulatory network by interacting with long noncoding RNAs [33, 34]. Thus, we postulated that DANCR interacts with hNRNPC to regulate p53 expression.

The binding of DANCR-hNRNPC and DANCR-P53 were validated in KGN cell lines by RNA pull-down assay and RNA-binding protein immunoprecipitation (Fig. 5A and B). The expression of P53 was upregulated apparently in KGN cells with DANCR knocked down, while the interaction between hNRNPC and P53 significantly decreased. Conversely, overexpression of DANCR could slightly enhance the binding between hNRNPC and P53 (Fig. 5C). In addition, hNRNPC and p53 proteins colocalization staining indicated that DANCR knockdown attenuate the binding of hNRNPC and P53 in the cell nucleus, and on the contrary DANCR overexpression enhanced the binding (Fig. 5D).

To identify whether downregulating DANCR promoted the granulosa cells aging by regulating the interaction of hNRNPC and P53, shDANCR and overexpressed hNRNPC (OEhNRNPC) were transfected into KGN cells. As the results show, when transfected with shDANCR alone, it upregulated the p53 expression and induced granulosa cells aging (Fig. 6A and B). Subsequently transfected with shDANCR + OEhNRNPC, the expression of p53 was decreased and the cell aging was notably alleviated compared with shDANCR group (Fig. 6A and B). In summary, DANCR knockdown reduces the binding of hNRNPC and P53 and increases the protein level of p53, which accelerates granulosa cell aging eventually.

In this study, we explored the function and mechanism of the lncRNA-DANCR in the pathogenesis of POI. The binding of hNRNPC and p53 was inhibited when DANCR knockdown, which contributed to the accumulation of p53 and granulosa cells aging. Conclusively, DANCR knockdown may induce accelerated senescence of granulosa cells, aggravate follicular atresia and hormone disorders and eventually cause POI (Fig. 6C).

This study was approved by the Ethics Committee of Changzheng Hospital Affiliated with Naval Medical University (No. 2018SLYS1). All patients enrolled were undergoing in vitro fertilization-embryo transfer (IVF) or intracytoplasmic sperm injection (ICSI) treatments in reproductive medicine center and signed informed consent. Our study included 8 patients diagnosed as POI according to the ESHRE guideline [46], and 10 infertile women because of tubal or male factor, with normal ovarian response, regular menses, normal basal hormone levels and BMI 19–24, as the control group. Because POI patients had been treated with hormone replacement therapy, the basic hormone level and menstrual cycle have recovered, and the low AMH level (AMH < 1.5 ng/ml) reflecting poor ovarian reserve was the key inclusion criteria for POI. The major anthropometric variables and endocrine parameters of the women are listed in Table 1. All the human ovarian GC samples were collected from the discarded products of transvaginal oocyte retrieval and isolation. The preparation of GC samples in the laboratory followed our previous procedure [47].

Dancr^fl/fl and CMV-Cre mice were constructed and purchased from Biocytogen Pharmaceuticals (Beijing) Co.,Ltd. Dancr gene is on the positive strand of chromosome 5, with a full length of 1.25 kb (NCBI ID: 70,036). The sgRNAs are designed to be approximately 2 kb upstream of Dancr and 500 kb downstream of Dancr in a non-conserved region. Dancr^fl/fl mice was screened with genotype verification. CMV-Cre mice express Cre recombinase under the control of a human cytomegalovirus (CMV) promoter, active in most cells and tissues. Dancr^fl/fl mice mated with CMV-Cre mice. Dancr gene knockout mice (Dancr^−/−) were obtained after three generations and Dancr^+/+ mice were a negative control for follow-up experiments. In other experiments, C57BL/6 mice, aged 8 weeks and 36 weeks, were purchased from SLAC Laboratory Animal Company. The following primers used for Dancr^−/− genotyping were 5′loxP-F: GACCTCTAGCAAGTAGACCAGGGGA; 5′loxP-R: GTCTGTTATCTGGATCCTGTAGTCCTGG;3′loxP-F:GACCTCTAGCAAGTAGACCAGGGGA and 3′loxP-R:TCGGGCTAGTGCAGGTGTTAGCAGGT [48]. Animal were maintained at a temperature of 25 °C and a humidity of 50% to 60%, with a 12:12 h light–dark cycle with ad libitum water and food. All experiments were approved by the Ethics Committee of Changzheng Hospital Affiliated with Naval Medical University.

8-week-old Dancr^−/− or Dancr^+/+ female mice were performed with vaginal smear once a day lasting for 20 days, and each group contained 12 mice. The slides were fixed with 90% ethanol for 5 min, Giemsa stained for 15 min, washed gently and dried. The types of cells in vaginal smears were evaluated by optical microscopy. With cytological evaluation, proportions of keratinized epithelial cell, nucleated epithelial cell and leukocytes as a criterion to determine the phase of estrous cycle [49]. There are three types of disordered estrous cycle in our study: a) prolonged estrous with normal diestrus; b) prolonged diestrus with normal estrous; c) prolonged estrous and diestrus.

Each mouse was weighed first, and there was no significant difference in mean weight between the Dancr^−/− and Dancr^+/+ groups. Each group contains 12 mice. 8-week-old female Dancr^−/− mice and Dancr^+/+ mice were 1:1 mated with 8-week-old male mice in the cage randomly. The next morning (day post coitum [DPC] 0.5), vaginal plug was detected. The female mice with vaginal plug were thought to be pregnant. If there’s not vaginal plug, the male mice were changed cages for further mating. The pregnant female mice were fed in the separate cages.

Both ovaries of sacrificed mice were carefully removed and cleaned with PBS gently. Place the ovaries into a dish under a stereo microscope for further operation. The antral follicles were punctured with a 1 ml disposable syringe and granulosa cells were extruded slowly. All of the granulosa cells were removed into EP tubes using a 10 µl pipette and digested with hyaluronidase at room temperature for 10 min. After washing with PBS for 3 times, the primary granulosa cells were prepared for subsequent experiments.

The human ovarian granulosa cell lines KGN and COV434 kept in our laboratory with STR (short tandem repeat) identification were cultured in DMEM (HyClone, USA) supplemented with 1% fetal bovine serum (Gibco, USA) and 100 IU/mL penicillin–streptomycin at 37 °C in a humidified atmosphere of 5% CO₂ and 95% air.

Mice whole blood was collected from the eye socket veins, and then balanced at 4 °C for 1 h, followed by centrifugation at 3000 rpm for 20 min. The upper serum was transferred and storage at -20 °C/-80 °C. The mice serum AMH, estradiol (E2), and FSH were measured with ELISA kits (Demeditec, Germany, DE1288/DE2693 & JianglaiBio, China, JL20476) according to manufacturer’s instructions.

The unilateral ovary tissues collected from 8-week-old Dancr^−/− and Dancr^+/+ mice were fixed in 4% paraformaldehyde (Sigma), dehydrated, and paraffin-embedded. Then the tissues were cut into continuous 4-μm-thick sections. One slide selected at interval of five, totally 20 sections per ovary tissue sample were obtained for HE staining. The follicles at each stage were counted separately and total follicles were counted.

ISH/FISH probes for DANCR (Human:5’-CY3-TGGCTTGTGCCTGTAGTTGTCAACCTGCGC-CY3-3’; Mouse:5’-DIG-AGTGGGACATGAAGAAGGGGTGGGGCA-DIG-3’) were synthesized by Wuhan Servicebio Company. ISH/FISH was performed according to the protocol of FISH assay kit. Representative images were captured under a fluorescence microscope.

Total RNA from cells or tissues were extracted with TRIzol reagent and RNeasy Mini Kit (Qiagen), and were subsequently reverse transcribed into cDNA by PrimerScript™ RT Master Mix Kit (Takara). RNA quantitation was performed with TB Green Premix EX Taq kit (Takara) on the ABI QuantStudio6 system according to manufacturer’s protocol. The primers used for amplification of DANCR are listed in Supplementary Table 2.

The lentiviral vector system pLKD-CMV-G&PR-U6 was performed to gene knockdown and overexpression in granulosa cell lines (KGN or COV434). The DANCR shRNA/overexpressed lentiviral vectors were constructed by OBiO Technology Corporation. (DANCR shRNA, 5′-GCAGCTGCCTCAGTTCTTA-3′; NC shRNA, 5′-TTCTCCGAACGTGTCACGT-3′). Cell infection was conducted as follows: the cultures were transfected with lentiviral particles according to OBiO Kit. After infection, the stable transfectants were selected using 2 mg/mL puromycin in the presence of 10% FBS for 72 h. The efficiency of DANCR downregulation in KGN and COV434 cells was confirmed by real-time qPCR. The qPCR primers were listed in the Table S2.

Cell viability and proliferation was detected by cell counting kit (CCK)-8 (Dojindo, Japan) and 5-Ethynyl-20-Deoxyuridine (EdU) assay (RiboBio, China). Cell cycle was detected by flow cytometry with cells incubated with DNA staining solution (PBS containing 20 μg/ml RNase A and 20 μg/ ml PI) at 37 °C for 30 min in dark place. For cell aging evaluation, senescent-associated β-gal (SA-β-Gal) and γ-H2AX staining were performed, with imagines captured under microscopy.

The sense and anti-sense lncRNA DANCR for protein microarray screening were synthesized labeling with Cy5 dye. The HuProt microarray from Wayen biotechnologies, comprised of more than ~ 20,000 full length human proteins with N-terminal GST-fusions, were performed according to the manufacturer’s instructions. Specifically, proteome microarrays were blocked for 1 h at room temperature. The lncRNA was placed in blocking buffer and incubated with the proteome microarray at room temperature for 1 h. The microarrays were washed three times for 5 min each time with TBST and then three times with Milli-Q water. Dried slides were scanned using GenePix 4000B, and images were analyzed using GenePix Pro 6.0. Results are presented in Supplementary File (Table S1).

Cells were lysed with radioimmunoprecipitation assay (RIPA) buffer (Beyotime, Shanghai, China) containing Protease/Phosphatase Inhibitor Cocktail (Epizyme, Shanghai, China) in ice-cold. The abundance of total protein was determined using a BCA Protein Assay Kit (Beyotime, Shanghai, China). Equal amount protein per lane was loaded onto an SDS–polyacrylamide gel, electrophoresed and transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, USA). After blocking with 5% nonfat milk for 1 h, the membranes were incubated with diluted primary antibodies γ-H2AX (Millipore, USA, 05–636-AF488); p53 (CST, USA, 2524); hNRNPC (Novusbio, USA, SN0652); P21(Affinity, USA, AF6290); β-actin (CST, USA, 4967/3700), GAPDH (CST, USA, 5174) overnight at 4 °C, then washed and incubated with indicated secondary antibodies (IRDye 680RD, IRDye800CW (Licor, USA, #c50408-02 #c50331-05). The Odyssey infrared scanner (Li-COR Biosciences, Nebraska, USA) was used to detect the protein bands and β-actin served as a loading control.

For Co-IP, the total protein was extracted as the WB protocol above. Antibodies for IP was incubated with magnetic beads (Thermo, USA, 88,802) for 40 min and washed with PBS for three times. Extracted protein and magnetic beads were incubated overnight at 4 °C. The magnetic beads were washed with PBS for 5 times, and 80 μl PBS was added to re-suspend the beads. After protein denaturation at 100 °C for 10 min, magnetic beads were removed by magnetic rack, and the protein sample was completely prepared for further WB detection.

RNA-binding protein immunoprecipitation assays were performed by a Magna RIP Kit (Millipore, USA) according to the manufacturer’s methods. Cells were lysed in RIP lysis buffer, and magnetic beads with HNRNPC (Novusbio, #SN0652), P53 (CST, #2524) or IgG antibody were prepared. The immunoprecipitation reactions were carried out by incubating the RIP lysate and the beads-antibody complex together with rotation overnight at 4 °C. Immunoprecipitated RNAs were purified and analyzed by qPCR and normalized to the input control.

DANCR plasmid was synthesized by OBiO Technology Corporation and extracted with plasmid extraction kit (Tiangen, China). SP6 and T7 enzymes were added overnight at 37 °C for enzyme digestion and transcription in vitro was performed with the kit. Biotin was labeled on the DANCR-RNA transcribed and incubated with Dynabeads (Invitrogen, USA, 65,001). Cell protein samples were prepared in advance and incubated with the Dynabeads-RNA complex at 4 °C overnight. The enriched protein was eluted and preserved for Western Blot to verify the protein lane.