We enrolled 10 healthy pregnant women with unintended pregnancy as the control group and 12 early RM women as the experimental group, with early RM defined as at least two consecutive unexplained miscarriages with regular menstrual cycles. Both groups were enrolled at the Department of Family Planning at the Chongqing Health Centre for Women and Children (Chongqing, China) from December 2016 to September 2017. The mean ages of the control group and the experimental group were 30.5 ± 3.1 and 30.8 ± 3.1 years, respectively, with mean gestational ages of 7.50 ± 0.58 and 7.75 ± 0.50 months, respectively. Patients with infections, endocrine/metabolic disorders, anatomical abnormalities, autoimmune diseases, or parental chromosomal abnormalities were excluded. All participants provided informed consent to have their clinical samples used in this study. All patients included in this study provided written consent before surgery, and the Institutional Review Board for Ethics of Chongqing Medical University and Chongqing Health Centre for Women and Children approved the study protocol. Decidua tissues were separated from the products of conception immediately, thoroughly washed with sterile normal saline, and stored in liquid nitrogen for future use.

Ishikawa cells were obtained from the American Type Culture Collection (ATCC) and cultured in RPMI-1640 with 5% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Beyotime, Shanghai, China). The medium was replenished every 48 h. The human endometrial stromal cell lines (CRL-4003) were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/Ham’s F-12 medium (Sigma-Aldrich, Saint Louis, MO, United States) supplemented with 10% charcoal-stripped FBS (Biological Industries, Israel) and 1% penicillin/streptomycin (Beyotime, Shanghai, China). All cells were cultured in a 5% CO₂ incubator at 37°C. The circ-CYP24A1 siRNA, miR-224 mimic, miR-224 inhibitor, and PRLR siRNA were transfected using Lipofectamine 2000 (Lip 2000; Invitrogen, Carlsbad, CA, United States). The cells were cultured for 2 h in a penicillin/streptomycin-free medium before transfection. SiRNA and Lip 2000 were diluted separately in 50 μl RPMI-1640 or DMEM/Ham’s F-12 medium and mixed for 15 min at 25°C away from light. Equal amounts of components were then added to orifice plates, and the penicillin/streptomycin medium was added to make the total volume of 500 μl. The circ-CYP24A1, PRLR, and negative control (NC) siRNAs were purchased from Sangon Biotech (Shanghai, China). The miR-224 mimic, miR-224 inhibitor, and NC were purchased from RuiBo Biotechnology (Guangzhou, China).

In vitro artificially induced decidualization of human endometrial stromal cell lines was achieved as described earlier (Gellersen and Brosens, 2014; Salsano et al., 2017). Cells were cultured in 2.5% complete medium containing medroxyprogesterone 17-acetate (MPA; 1 mM) and the stable cyclic adenosine 3′:5′ monophosphate (cAMP) analog 8-bromo-cAMP (0.5 mM). The induction for in vitro decidualization took 7 days, and the culture medium was changed every 2 days. The circ-CYP24A1 siRNA and miR-224 inhibitor were transfected after artificially induced decidualization in vitro.

Cell counting kit-8 cell (CCK-8) counting kit (Beyotime Biotechnology, Nanjing, China) was used to detect cell proliferation rates. The Ishikawa cells were plated in 96-well plates at a density of 1 × 10⁴ per well and cultured for 24 h. As mentioned earlier, siRNA and inhibitor experiments were conducted. The effect of SiRNA and inhibitor on the proliferation of Ishikawa cells was assessed every 24 h for 72 h after transfection. A 10 μl CCK-8 solution was added to each well at the specified time points, following incubation for 1 h at 37°C. The absorbance at 450 nm was calculated using a microplate reader (Multiskan FC, Thermo, Waltham, MA, United States). All experiments were performed in triplicate.

According to the bioinformatics software (circinteractome, miRanda, and TargetScan), we predicted the binding sties between circ-CYP24A1 and miR-224, as well as the potential target genes of miR-224. Predicted binding sites between circ-CYP24A1 and miR-224, miR-224 and PRLR, and a mutant sequence that lack the binding site were cloned into GV272, which was purchased from JiKai Biotechnology (Shanghai, China). These constructs and the miR-224 mimic were co-transfected into HEK293T cells cultured in a 96-well plate. After 48 h of incubation, Firefly and Renilla luciferase activities were measured using the Promega Dual-Luciferase system (Shanghai Genechem Co., Ltd., Shanghai, China), according to the manufacturer’s instructions. To lyse cells, 100 μl of Passive Lysis Buffer was added to 96-well plate, and the Lysis Solution was collected. Then, Firefly luciferase was calculated by adding 100 μl Luciferase Assay Reagent II and 20 μl collected cell Lysis Solution into a 96-well plate. The Renilla fluorescence was calculated by adding 100 μl Luciferase Assay Reagent to the 96-well plate. Relative luciferase activity was calculated as the ratio of Firefly fluorescence to Renilla fluorescence. According to the bioinformatics software (CircInteractome, miRanda, and TargetScan), we predicted the binding sties between circ-CYP24A1 and miR-224, as well as the potential target genes of miR-224. Predicted binding sites between circ-CYP24A1 and miR-224, miR-224 and PRLR, and a mutant sequence lacking the binding site were cloned into GV272, which was purchased from JiKai Biotechnology (Shanghai, China). These constructs and the miR-224 mimic were co-transfected into HEK293T cells cultured in a 96-well plate. After 48 h of incubation, Firefly and Renilla luciferase activities were measured using the Promega Dual-Luciferase system (Shanghai Genechem Co., Ltd., Shanghai, China), according to the manufacturer’s instructions. HEK293T cells were lysed with Passive Lysis Buffer, and then Luciferase Assay Reagent II was added to the cell Lysis Solution to obtain and record the Firefly luciferase. Then Stop & Glo^® Reagent was added to the plate to get the Renilla fluorescence. Relative luciferase activity was calculated as the ratio of Firefly fluorescence to Renilla fluorescence.

As per the manufacturer’s protocol, total RNA was extracted using the RNAiso Plus reagent (TaKaRa, Kyoto, Japan) and reverse transcribed using the PrimeScript™ RT kit (TaKaRa, Kyoto, Japan). The RNA expression was calculated on the Bio-Rad CFX Manager Detection system (United States) using the SYBR Premix Ex Taq™ kit (TaKaRa, Kyoto, Japan). The RNA expression was calculated using the 2^{–ΔΔ CT} method. GAPDH was used as an internal control for mRNA and circRNA. U6 was used as an internal control for miRNA. The primers for circ-CYP24A1 were designed by Guangzhou Jisai Co. Ltd. (Guangzhou, China) and synthesized by Sangon Biotech (Shanghai, China). The primer for miR-224 was designed and synthesized by Sangon Biotech (Shanghai, China). PCR primers (Sangon Biotech, Shanghai, China) are shown in Table 1.

Digoxygenin-labeled probes specific for circ-CYP24A1 were designed and synthesized by Guangzhou Jisai Co. Ltd. (Guangzhou, China) As described in the previous study of our team (Liu S. et al., Reprod Sci., 2014, 21:686–695), ISH was performed as follows. Frozen tissue sections (10 μm) were prepared and fixed on glass slides pre-treated with poly-lysine. After being fixed in 4% paraformaldehyde and treated with 1% Triton X-100, sections were incubated with the digoxin-labeled probes at 42°C for 16 h for hybridization, using an anti-digoxygenin antibody conjugated to alkaline phosphatase (1:5,000, Roche, Basel, Swiss Confederation). The NC was incubated with hybridization fluid instead of probes. Nitro-blue-tetrazolium and 5-bromo-4-chloro-3-indolyl-phosphate (Beyotime Biotechnology, Nanjing, China) were then used for staining. The nuclei were counterstained with 1% methyl green. Finally, sections were photographed using a photomicroscope (Olympus, Tokyo, Japan).

As described in a previous study (Su et al., 2020), IHC was performed as follows. Uterine tissue samples were fixed in a 4% paraformaldehyde solution for 4–6 h, dehydrated with increasing concentrations of ethanol (75, 85, 95, and 100%), and embedded in paraffin. Sections of 5 μm were cut for further detection. Antigen retrieval was carried out in pH 8.0 EDTA buffer (Zhongshan Biosciences, Beijing, China) for 15 min at 95°C, followed by cooling naturally to room temperature. H₂O₂ (3%) was used to eliminate the endogenous peroxidase interference. To block non-specific binding, sections were incubated with 10% goat serum and then with the mouse monoclonal anti-PRLR (1:500; Abcam, Cambridge, United Kingdom) overnight at 4°C. The NC was incubated with isotype IgG (1:500; Bioss, Beijing, China) instead of primary antibody. Isotype IgG is used at the same concentration as the primary antibody. The sections were then incubated with secondary antibodies at 37°C for 1 h, followed by an avidin-biotinylated peroxidase complex system (Zhongshan Biosciences, Beijing, China) for 30 min. The chromogenic reaction was initiated by incubating with diaminobenzidine (Zhongshan Biosciences, Beijing, China) for 3–5 min and terminated with water. Nuclei were counterstained with hematoxylin. Finally, sections were photographed using a photomicroscope (Olympus, Tokyo, Japan).

The Ishikawa cells were cultured in a 48-well plate with coverslips. When the density of adherent cells reached about 80%, the cell slides were fixed with ice methanol for 15 min, and PBS was washed three times for 3 min each time. The cells were perforated by incubating in a 0.5% Triton X-100 solution for 20 min at room temperature. After washing for three times, the cell slides were incubated with the mouse monoclonal anti-Ki67 (1:500; 9449T; Cell Signaling Technology, Boston, MA, United States) overnight at 4°C. The slides were washed and incubated with a CY3-labeled fluorescent mouse antibody for 1 h at 37°C. The plate was incubated with DAPI for 10 min and sealed with anti-fluorescence quenching tablets. Finally, the image was captured using a microscope and quantified using ImageJ software.

A one-step terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeling (TUNEL) assay kit (Beyotime Biotechnology, Nanjing, China) was used to detect apoptosis in Ishikawa cells. The Ishikawa cells were plated in 48-well plates at a density of 1 × 10⁴ per well and cultured for 24 h. siRNA and miR-224 inhibitor experiments were conducted as mentioned earlier. The cells were fixed with 4% paraformaldehyde and permeabilized with 0.3% Triton X-100. The cells were then incubated at 37°C for 60 min in the dark with TUNEL reaction buffer according to the manufacturer’s protocol. Subsequently, fluorescence microscopy was used to observe the apoptotic cells.

Total proteins were extracted using a protein extraction kit purchased from Beyotime (Shanghai, China). The extracts were fractionated on 10% sodium dodecyl sulfate-polyacrylamide gels and then transferred to polyvinylidene difluoride membranes (Millipore, Billerica, MA, United States). To block non-specific binding, membranes were incubated in 5% BSA and then incubated overnight with primary antibodies. The primary antibodies used were as follows: mouse monoclonal anti-PRLR (1:500; ab2773, Abcam, Cambridge, United Kingdom), mouse monoclonal anti-cyclinD3 (1:200; 2936 s, Cell Signaling Technology, Boston, MA, United States), mouse monoclonal anti-PCNA (1:500; 2586 s; Cell Signaling Technology, Boston, MA, United States), rabbit polyclonal anti-Bax (1:300; 2772 s; Cell Signaling Technology, Boston, MA, United States), and mouse monoclonal anti-Bcl-2 (1:300; 3498 s; Cell Signaling Technology, Boston, MA, United States). After incubating with secondary antibodies, antibody binding was quantified using an ECL reagent (Millipore, Darmstadt, Germany), and bands were visualized using a ChemiDoc™ XRS + (Bio-Rad).

All data are shown as means ± SEM. All experiments were repeated at least three times. Data were analyzed using the SPSS 22.0 package (SPSS China, Shanghai, China). Comparisons between two groups were calculated using t-tests. Comparisons among three or more groups were calculated using one-way ANOVA. The p-value < 0.05 was considered statistically significant.

Circular CYP24A1 was derived from the segment spanning exons 3–10 of the CYP24A1 gene of human chromosome 20. Circ-CYP24A1 is located at chr20:52773707-52789638 with a length of 1297 nt (Figure 1A). Salzman et al. (2013) had reported the relative expression of Circ-CYP24A1 in A549, AG04450, and HeLa cells. However, the function of Circ-CYP24A1 is still not clear. CircRNAs often function as ceRNAs to adsorbed miRNAs, thus regulating mRNA and protein synthesis. In our previous study, using bioinformatics predictions from miRanda and the TargetScan database, we found that circ-CYP24A1 may act as a sponge for miR-224 and regulate the expression of the PRLR gene. We had demonstrated high expression of circ-CYP24A1 and PRLR and decreased expression of miR-224 in the decidua of patients with early RM (Li C. et al., 2020). In this study, we used ISH to verify the expression of circ-CYP24A1 in the decidua of patients with early RM and unintended pregnant women. As shown in Figure 1B, the high expression of circ-CYP24A1 in the decidua of patients with an early RM was observed, which is consistent with our previous data. To analyze the binding of miR-224 and circ-CYP24A1, a luciferase reporter assay was conducted. As shown in Figure 1C, a fragment of circ-CYP24A1, including the predicted target site or a mutated target site, was inserted into the downstream part of the Firefly luciferase gene (pmirGLO-circ-CYP24A1-wt and pmirGLO-circ-CYP24A1-mut). The plasmids were co-transfected with miR-224 mimic or mimic NC into HEK293T cells. Overexpression of miR-224 significantly reduced luciferase activity in cells transfected with the vector containing the complete circ-CYP24A1 sequence but did not change the luciferase activity with the vector containing the mutated miR-224 binding site (Figure 1D). The result suggests that circ-CYP24A1 could act as a molecular sponge for miR-224. To verify the competing endogenous function of circ-CYP24A1, the expression of miR-224 was examined after circ-CYP24A1 silencing. The transfection efficiency was verified using Real-time PCR (RT-PCR) (Figure 1E). As shown in Figure 1F, relative to the NC, the level of miR-224 increased significantly in Ishikawa cells transfected with circ-CYP24A1 siRNA.

Circular RNAs act as miRNA sponges to regulate their target genes (Hsiao et al., 2017). In this study, we verified whether PRLR was the downstream target of the circ-CYP24A1-miR-224 axis. The 3′ UTR fragments of PRLR mRNA containing miR-224 binding sites and the same fragments with mutated binding sites were cloned into the pGL3-basic luciferase reporter vectors (Figure 2A). As shown in Figure 2B, the co-transfection of pmirGLO-PRLR-wt and miR-224 mimic into HEK293T cells reduced luciferase activity compared with the co-transfection of pmirGLO-PRLR-mt and mimic NC. While no significant difference was observed in group co-transfection of pmirGLO-PRLR-mut and miR-224 mimic and group co-transfection of pmirGLO-PRLR-mut and mimic NC, this assay confirmed the direct binding of miR-224 and PRLR mRNA 3′ UTR. The regulatory effect of miR-224 on PRLR was further confirmed using the miR-224 mimic to simulate the high expression of miR-224 in Ishikawa cells. The stimulatory effect of miR-224 mimic was confirmed using RT-PCR (Figure 2C). The expression of PRLR was significantly decreased after miR-224 mimic treatment (Figures 2D,E). Besides, a rescue experiment was further conducted in Ishikawa cells. Ishikawa cells were transfected with miR-224 inhibitor or co-transfected with miR-224 inhibitor and PRLR siRNA. The transfection efficiency was verified using RT-PCR (Figure 2F). The expression of PRLR was significantly increased after miR-224 inhibitor treatment. While the activation of miR-224 on PRLR expression was significantly inhibited by PRLR siRNA (Figures 2G,H), considering that decidual cells are more representative cells model of decidua tissue, we conducted the rescue experiment in decidual cells. First, the human endometrial stromal cell line was induced to transform into the decidual cell using medroxyprogesterone 17-acetate (MPA; 1 mM) and stable cyclic adenosine 3′:5′ monophosphate (cAMP) analog 8-bromo-cAMP (0.5 mM) for 7 days. High expression of IGFBP1 mRNA indicated the decidualization of the stromal cell (Supplementary Figure 1B). The decidual cells were then transfected with miR-224 inhibitor or co-transfected with miR-224 inhibitor and PRLR siRNA. The results were similar to that in Ishikawa cells (Supplementary Figures 1C,D). Next, the expression of PRLR protein after circ-CYP24A1 silencing was examined using Western blotting. As shown in Figures 2I,J, compared with the NC (si-NC), PRLR expression was significantly reduced in the Ishikawa cell treated with circ-CYP24A siRNA. To confirm that circ-CYP24A regulates PRLR expression through sponging miR-224, a rescue experiment was conducted in both Ishikawa cells and decidual cells. As shown in Figures 2K,L, and Supplementary Figures 1E,F, the expression of PRLR reduced significantly after circ-CYP24A1 silencing, and the miR-224 inhibitor reversed the suppression of circ-CYP24A1 siRNA on PRLR expression. Overall, these results indicated that circ-CYP24A1 elevates PRLR expression by acting as a sponge for miR-224.

Circular RNAs often act as miRNA sponges, regulating mRNAs that function in the cell cycle (Du et al., 2016), apoptosis (Du et al., 2017a), and senescence (Du et al., 2017b). Defects in the proliferation and apoptosis of the decidual cells are known as two important issues associated with RM. IHC was performed to verify the levels and locations of PRLR protein expression. As shown in Figure 3A, high expression was observed in the decidua of patients with early RM. PRLR has been reported to promote cell proliferation and inhibit apoptosis in breast cancer (Leehy et al., 2018). In this study, the proliferation and apoptosis of Ishikawa cells were detected, even after PRLR silencing. As shown in Figures 3B–F, knocking out PRLR significantly decreased the expression of the proliferation-related proteins Cyclin D3 and PCNA. Simultaneously, enhanced expression of Bcl-2 and reduced expression of Bax were detected after PRLR silencing, reflecting the activation of apoptosis. The above results demonstrated the role of PRLR in promoting proliferation and inhibiting apoptosis.

As reported earlier, miR-224 plays an important role in regulating the proliferation and apoptosis of many diseases (Liu W. et al., 2016; Liu et al., 2020; Xie et al., 2020). In this study, CCK-8 assay and IF staining were used to verify the role of miR-224 in regulating proliferation in Ishikawa cells. Ishikawa cells were treated with miR-224 inhibitor or co-transfected with miR-224 inhibitor and PRLR siRNA. Cell viability was then detected at different time points. As shown in Figure 4A, miR-224 inhibitor showed a significant increase in cell viability compared with control (inhibitor-NC). The increase in cell viability was inhibited by PRLR siRNA. The IF staining of Ki67 yielded similar results, the miR-224 inhibitor significantly increased the expression of Ki67 in Ishikawa cells, and PRLR siRNA suppressed the activation of Ki67 (Figures 4B,C). Then TUNEL assay was used to verify the regulating function of miR-224 in apoptosis in Ishikawa cells. TUNEL-positive cells were observed in the inhibitor-NC, miR-224 inhibitor, and miR-224 inhibitor + si-PRLR groups (Figures 4D,E). Apoptosis of Ishikawa cells was decreased in the miR-224 inhibitor group compared with the control group. However, cell apoptosis in the miR-224 inhibitor + si-PRLR group was increased compared with the miR-224 inhibitor group (Figures 4D,E). These results demonstrated that miR-224 promotes the proliferation and inhibited apoptosis of Ishikawa cells by downregulating PRLR.

Next, we investigated the effects of circ-CYP24A1 on proliferation and apoptosis in Ishikawa cells. Western blotting showed that silencing circ-CYP24A1 dramatically reduced the expression of Cyclin D3 and PCNA. Moreover, after circ-CYP24A1 knockout, the expression of Bax was reduced while the expression of Bcl-2 increased significantly (Figures 5A–E). These results indicate that silencing circ-CYP24A1 also increased apoptosis and reduced cell proliferation. Next, we investigated whether circ-CYP24A1 regulates Ishikawa cell proliferation and apoptosis via the miR-224-PRLR axis. Ishikawa cells were transfected with circ-CYP24A1 siRNA or co-transfected with circ-CYP24A1 siRNA and miR-224 inhibitor. CCK-8 assay showed that silencing circ-CYP24A1 dramatically reduced cell viability compared with si-NC, while miR-224 inhibitor rescued the reduction of cell viability (Figure 5F). Besides, the expression of Ki67 in Ishikawa cells decreased significantly after silencing circ-CYP24A1, but the reduction was rescued by a miR-224 inhibitor (Figures 5G,H). The level of apoptosis in Ishikawa cells was examined using a TUNEL assay. More TUNEL-positive cells were observed after circ-CYP24A1 knockout. However, compared with the si-circ-CYP24A1 group, the apoptosis was decreased after co-transfection of circ-CYP24A1 siRNA and miR-224 inhibitor (Figures 5I,J). The results above demonstrated that circ-CYP24A1 regulates Ishikawa cell proliferation and apoptosis via the miR-224-PRLR axis.