A total of 140 HGSOC tissue specimens were collected for the purposes of this study. The tissue microarray (TMA) included 120 HGSOC samples and these tissues were collected from female patients admitted to Qilu Hospital (Jinan, China) from May 2006 to July 2013. The HGSOC samples (n=20) used for western blot and reverse transcription-quantitative polymerase chain reaction (RT-qPCR) analyses were collected from patients admitted to Qilu hospital from September 2014 to July 2015. All HGSOC specimens were collected from patients with primary ovarian cancer that had not received chemotherapy prior to surgery. The fresh-frozen normal fallopian tube (FT, n=11) tissues were obtained from female patients with benign gynecological tumors that had received a hysterectomy with bilateral salpingo-oophorectomy at Qilu hospital from September 2014 to July 2015. The tumor diagnoses were verified by two gynecological pathologists. Ethical approval was obtained from Ethics Committee of Shandong University Qilu Hospital (Jinan, China). All patients provided written informed consent.

293T cells, and the human ovarian cancer cell lines, A2780 and HO8910, were purchased from the China Center for Type Culture Collection (Wuhan, China). Human ovarian cancer cell lines, SKOV3, OVCAR3, CAOV3 and the p53-null H1299 human lung cancer cell line, were purchased from the American Type Culture Collection (Manassas, VA, USA). A2780, HO8910 and H1299 cells were cultured in RPMI-1640 supplemented with 10% fetal bovine serum (FBS; both purchased from Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA); OVCAR3 cells were maintained in RPMI-1640 supplemented with 20% FBS and 0.01 mg/ml bovine insulin (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany); SKOV3 cells were cultured in McCoy’s 5A medium (Thermo Fisher Scientific, Inc.) supplemented with 10% FBS; CAOV3 cells were cultured in Dulbecco’s modified Eagle’s medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS. In addition, 100 U/ml penicillin and 100 μg/ml streptomycin (Sigma-Aldrich; Merck KGaA) were added to the complete culture medium for all cell lines. All cells were cultured at 37°C with 5% CO₂ in a humidified incubator.

Total RNA was extracted from fresh-frozen HGSOC and FT tissue samples, as well as OVCAR3 cells transfected with miR-30a-5p/miR-200a-5p mimics or inhibitors, using TRIzol reagent according to manufacturer’s instructions (Invitrogen; Thermo Fisher Scientific, Inc.). Total RNA (1,000 ng) was reverse transcribed to cDNA using the PrimeScript RT reagent kit or the Mir-X miRNA First-Strand Synthesis kit (both purchased from Takara Biotechnology, Co., Ltd., Dalian, China) according to manufacturer’s instructions. qPCR analysis was performed using the SYBR Premix Ex Taq (Takara Biotechnology, Co., Ltd.). The qPCR reaction was performed using the StepOnePlus™ Real-Time PCR System (Thermo Fisher Scientific, Inc.). The thermal cycling parameters were as follows: 30 sec at 95°C followed by 43 cycles at 95°C for 5 sec and 60°C for 30 sec. The primer sequences employed were as follows: FOXD1, forward, 5′-GATCTGTGAGTTCATCAGCGGC-3′ and reverse, 5′-TGACGAAGCAGTCGTTGAGCGA-3′; β-actin, forward, 5′-CATGTACGTTGCTATCCAGGC-3′ and reverse, 5′-CTCCTTAATGTCACGCACGAT-3′; miR-30a-5p, forward, 5′-TGTAAACATCCTCGACTGGAAG-3′ and reverse, mRQ 3′ Primer (provided in Mir-X miRNA First-Strand Synthesis kit; Takara Biotechnology, Co., Ltd.); miR-200a-5p, forward, 5′-CATCTTACCGGACAGTGCTGGA-3′ and reverse, mRQ 3′ Primer; U6, forward, 5′-GCTTCGGCAGCACATATACTAAAAT-3′ and reverse, mRQ 3′ Primer. The expression of β-actin and U6 was used as the internal control. Data analysis was performed using the 2^−ΔΔCq method (37).

Protein from fresh-frozen HGSOC and FT tissue samples was isolated using TRIzol reagent according to manufacturer’s instructions (Invitrogen; Thermo Fisher Scientific, Inc.). The HGSOC and FT tissue samples were divided into smaller sections (~0.1 g) and lysed with TRIzol for 10 min at room temperature. The lysate was centrifuged at 12,000 × g for 10 min at 4°C and the phenol-ethanol supernatant was incubated with 100% isopropanol for 10 min at room temperature. The mixture was then centrifuged again at 12,000 × g for 10 min at 4°C to pellet the protein. The protein pellet was washed twice with 0.3 M guanidine hydrochloride and once with 100% ethanol. Then the protein pellet was solubilized in 1% SDS. For isolating protein from cells, the cells were first washed three times with cold PBS and lysed using radioimmunoprecipitation assay buffer on ice for 30 min. The protein concentration was determined using a BCA assay (Merck KGaA). Protein samples (40 μg) were separated by 12% SDS-PAGE and electro-transferred onto polyvinylidene fluoride membranes using the semi-dry transfer method. The membranes were blocked for 1 h with 5% non-fat milk at room temperature, and then were incubated at 4°C overnight with the following primary antibodies: Anti-FOXD1 (dilution, 1:500; cat. no. ab179940; Abcam, Cambridge, USA), anti-p21 (dilution, 1:1,000; cat. no. 2947S; CST Biological Reagents Co., Ltd., Shanghai, China), anti-β-actin (dilution, 1:5,000; cat. no. ab8226; Abcam). The membranes were subsequently washed with Tris-buffered saline with 0.1% Tween-20, and then incubated with the secondary horseradish peroxidase (HRP)-conjugated anti-mouse IgG antibody (dilution, 1:8,000; cat. no. 074-1806) or anti-rabbit IgG antibody (dilution, 1:6,000; cat no. 5220-0336) (both from Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD, USA) for 2 h at room temperature. The signals were detected using an enhanced chemiluminescence system (GE Healthcare Life Sciences, Little Chalfont, UK). β-actin was used as an endogenous control.

For plasmid construction, the coding DNA sequence of FOXD1 was cloned into the pLenti-C-Myc-DDK-IRES-Puro tagged vector (PCMV; OriGene Technologies, Inc., Rockville, MD, USA). Short hairpin RNA (shRNA) targeting FOXD1 was purchased from Sigma-Aldrich; Merck KGaA, and cloned into the pLKO.1 vector (Addgene, Inc., Cambridge, MA, USA). A scrambled shRNA control (5′-CCTAAGGTTAAGTCGCCCTCGCTC GAGCGAGGGCGACTTAACCTTAGG-3′) was cloned into the pLKO.1 vector and used as a negative control. The lentiviral packaging and envelope plasmids used were pMD2.G and psPAX2 (Addgene, Inc.). For lentivirus production, 293T cells were seeded at a density of 4×10⁶ in 10 cm culture dishes and cultured for 16 h. The PCMV vector and lentiviral packaging and envelope plasmids were co-transfected into 293T cells using Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc.). For stable transfection, cells (A2780, HO8910, SKOV3, CAOV3, OVCAR3 and H1299) were infected with lentivirus at a multiplicity of infection of 50 for 24 h, and the cells were selected in medium containing puromycin (2 μg/ml) for 2 weeks.

OVCAR3 cells were seeded in 6 cm dishes at a density of 5×10⁵/dish at 24–36 h prior to transfection. Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) was employed to transfect the cells according to manufacturer’s instructions. The concentration of miRNA mimics and inhibitors used were as follows: 80 nM miR-30a-5p/miR-200a-5p mimics; 80 nM mimics control; 100 nM miR-30a-5p/miR-200a-5p inhibitors; and 100 nM inhibitors control. The medium was refreshed at 6 h following transfection. The sequences of miRNA mimics and inhibitors was as follows: miR-30a-5p mimics, 5′-UGUAAACAUCCUCGACUGGAAG-3′; miR-30a-5p inhibitors, 5′-CUUCCAGUCGAGGAUGUUUACA; miR-200a-5p mimics, 5′-CAUCUUACCGGACAGUGCUGGA-3′; miR-200a-5p inhibitors, 5′-UCCAGCACUGUCCGGUAAGAUG-3′; miRNA mimics negative control, 5′-UUCUCCGAACGUGUCACGUTT-3′; miRNA inhibitors negative control, 5′-CAGUACUUUUGUGUAGUACAA-3′. All the miRNA mimics and inhibitors were purchased from Guangzhou RiboBio Co., Ltd. (Guangzhou, China).

A2780 cells (2.5×10⁵) were plated in 6 cm dishes and transfected with p21 siRNA sequences using Lipofectamine 3000 according to the manufacturer’s protocol (Invitrogen; Thermo Fisher Scientific, Inc.). At 48 h following transfection, the cells were harvested for MTT and clonogenic assay analysis. The sequences of the p21 siRNAs were the same as described previously (38,39). The sequences of the p21 siRNA sequences were as follows: sip21-1, 5′-AGCGAUGGAACUUCGACUUTT-3′; sip21-2, 5′-AAUGGCGGGCUGCAUCCAGGATT-3′; siCtrl, 5′-UUCUCCGAACGUGUCACGUTT-3′ (all purchased from Shanghai GenePharma Co., Ltd., Shanghai, China).

Cells (n=800–1,000/well) stably expressing FOXD1 or transiently transfected with p21 siRNA and FOXD1 were plated in 6-well plates and cultured for 2–3 weeks. The colonies were fixed with 100% methanol for 15 min at room temperature and then stained with 1% crystal violet for 10 min at room temperature. The number of colonies was assessed using ImageQuant TL software (version 8.1; GE Healthcare Life Sciences). The data are presented as the mean ± standard error of three independent experiments.

Cell proliferation was assessed using an MTT assay. A2780 cells (800 cells/well) were first seeded in triplicate wells of a 96-well plate. MTT reagent (5 mg/ml; cat no. M2128; Sigma-Aldrich; Merck KGaA) was subsequently added to the wells and incubated for 4 h. DMSO was subsequently added to dissolve the formazan crystals and the absorbance at 490 nm was then read.

In order to investigate whether FOXD1 may be a direct target of miR-30a-5p and miR-200-5p, the 3′-UTR of the FOXD1 mRNA sequence containing the putative miR-30a-5p and miR-200a-5p binding sites were cloned into the pmirGLO Dual-Luciferase miRNA Target Expression Vector (Promega Corporation, Madison, WI, USA). The FOXD1 mutant sequences were constructed by Sangon Biotech, Co., Ltd. (Shanghai, China) and cloned into pmirGLO vector. 293T cells were co-transfected with miR-30a-5p or miR-200a-5p mimics and the pmirGLO vector containing either the wild-type or mutant FOXD1 sequence using Lipofectamine 3000 according to the manufacturer’s protocol (Invitrogen; Thermo Fisher Scientific, Inc.). At 48 h following transfection, luciferase activity was measured using the Dual-Glo Luciferase Assay System (Promega Corporation).

To investigate FOXD1 binding sites in the p21 promoter, three predicted wild-type binding sites of FOXD1 were separately cloned into the pGL4.26[luc2/minP/Hygro] vector (Promega Corporation), and the mutant binding sites were constructed by Sangon Biotech, Co., Ltd. and cloned into the pGL4.26 vector. The pGL4.74[hRluc/TK] vector (Promega Corporation) was used as an endogenous control. 293T cells (4×10⁴/well) were seeded in a 96-well plate and transfected with vectors using Lipofectamine 3000 according to the manufacturer’s protocol (Invitrogen; Thermo Fisher Scientific, Inc.). Cells were transfected with vectors at the following concentration ratio: PCMV (75 ng): pGL4.26 (25 ng): pGL4.74 (2.5 ng). At 48 h following transfection, luciferase activity was measured.

The TMA of HGSOC samples were incubated at 65°C for 30 min and deparaffinized in xylene immediately. The TMA was subsequently rehydrated in a graded ethanol series. Antigen retrieval was performed in citric acid buffer (pH 6.0) by heat treatment (98°C) for 15 min. This assay was performed using the SP-9000 IHC reagent kit (cat. no. SP-9000; OriGene Technologies, Inc.). The samples were blocked in 1.5% goat serum from the SP-9000 IHC reagent kit at 37°C for 20 min, and the TMA was then incubated with primary antibodies against FOXD1 (dilution, 1:1,000; cat. no. ab179940; Abcam) at 4°C for 14 h. The TMA was subsequently washed and incubated with a HRP-conjugated secondary antibody from the SP-9000 IHC reagent kit (OriGene Technologies, Inc.) for 30 min at 37°C. Staining was detected using a 3,3′-diaminobenzidine detection system. The Eclipse Ni-E microscope (Nikon Corporation, Tokyo, Japan) was used to visualize the stained tissue sections. As transcription factors modulate gene expression by binding to the promoter of target genes in the nucleus, the expression of FOXD1 in the nucleus was determined in the present study (40–42). The final score of each sample was decided by two pathologists and was dependent on the extent and intensity of staining. The intensity of staining was evaluated and scored as follows: 0, negative; 1, weak; 2, moderate; or 3, strong. The proportion of positively stained cells was divided into four levels and scored as follows: 1, 0–25%; 2, 26–50%; 3, 51–75%; and 4, 76–100%. Staining scores for FOXD1 were calculated by multiplying the score assigned to the intensity of staining (0–3) by the score assigned to the percentage of positively stained cells (1–4). According to the staining scores, the samples were divided into the low expression group (score <6) and high expression group (score ≥6). For the immunohistochemical staining of xenograft tumor samples from mice, the tumor tissues were fixed in 10% formalin for 2 days at room temperature and then embedded in paraffin. The samples were then sectioned (4-μm thickness), and then incubated at 65°C for 30 min before they were deparaffinized and rehydrated in a graded ethanol series. For antigen retrieval, the tissue sections were immersed in the citric acid buffer (pH 6.0) and incubated at 98°C for 15 min. The samples were blocked in 1.5% goat serum (SP-9000 IHC reagent kit; OriGene Technologies, Inc.) at 37°C for 20 min. The tissue sections were then incubated with primary antibodies against FOXD1 (dilution, 1:1,000; cat. no. ab179940; Abcam) and p21 (dilution, 1:1,000; cat. no. 2947S; CST Biological Reagents Co., Ltd.) at 4°C for 14 h. The sections were washed and incubated with a HRP-conjugated secondary antibody from the SP-9000 IHC reagent kit (OriGene Technologies, Inc.) for 30 min at 37°C. Staining was detected using a 3,3′-diaminobenzidine detection system.

Cells (A2780, HO8910, SKOV3, OVCAR3, CAOV3 and H1299) stably over-expressing FOXD1 were seeded in 6 cm dishes at a density of 1.2–2.0×10⁵ cells/well and cultured until they were 30–40% confluent. The cells were then synchronized using a double thymidine block as described previously (43) in order to synchronize cells in G1 phase. The cells were then released from this phase at different intervals (A2780, 3 h; HO8910, 6 h; SKOV3, 12 h; OVCAR3, 6 h; CAOV3, 12 h; H1299, 10 h) by replacing the medium with complete culture medium without thymidine. The cells were harvested for cell cycle analysis using the BD FACSCalibur Flow Cytometer (BD Biosciences, Franklin Lakes, NJ, USA) and the results were analyzed using ModFit LT software (version 2.0; Verity Software House, Topsham, ME, USA).

A ChIP assay was performed to enrich the specific regions of chromatin immunoprecipitated by a specific antibody via using the ChIP-IT Express Chromatin Immunoprecipitation kit (Active Motif, Carlsbad, CA, USA) according to the manufacturer’s protocol. A2780 cells (1×10⁷) stably overexpressing FOXD1 were first cross-linked and sheared into ~500-bp DNA fragments by sonication. The DNA fragments were then selectively immunoprecipitated from the cell debris by incubating with a DYKDDDDK Tag antibody (dilution, 1:50; cat. no. 14793S), or a rabbit IgG antibody (dilution, 1:50; cat. no. 2729) (both from CST Biological Reagents Co., Ltd.), which was used as negative control, for 16 h at 4°C. The purified DNA was amplified by PCR using the Phusion High-Fidelity PCR Master Mix with HF Buffer (cat no. M0531; New England Biolabs, Inc., Ipswich, MA, USA) and the following primers: Site B, forward, 5′-GTCGTGGTGGTGGTGA-3′ and reverse, 5′-CTGCTTTCAGGCATTTC-3′; site C, forward, 5′-ATGTCATCCTCCTGATCTTT-3′ and reverse, 5′-AGTCCCTCGCCTGCGTTGGT-3′. The thermal cycling parameters were as follows: 30 sec at 98°C followed by 40 cycles at 98°C for 10 sec, 60°C for 30 sec, 72°C for 15 sec, and then a final step at 72°C for 6 min. The PCR products were separated by 2% agarose gel electrophoresis and stained with ethidium bromide.

All experiments involving animals were approved by the Ethics Committee on Animal Experiments of Shandong University. A total of 20 BALB/c nu/nu female mice (age, 6–7 weeks; weight, 20.3±1.5 g) were maintained in a pathogen-free facility (temperature, 25°C; humidity, 50–60%; 12 h light/day cycles) and had free access to water and food. The mice were equally divided into A2780 and OVCAR3 groups, and each group was further subdivided into a control and FOXD1 overexpression group. For the in vivo xenograft experiments, A2780 and OVCAR3 cells transfected with PCMV or PCMV-FOXD1 vectors (6×10⁶ cells) were injected subcutaneously into the right flank region of the mice. Following 4 weeks (A2780 cells) or 5 weeks (OVCAR3 cells), the mice were sacrificed to determine tumor weight and volume. Tumor volume was calculated using the following formula: Volume = [(long diameter) × (short diameter)²] ×1/2. Immunohistochemical staining was employed for the detection of FOXD1 and p21 using the aforementioned methods.

The GSE9891 dataset (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE9891) contains gene expression data for 285 ovarian carcinoma samples (44). Following the removal of 7 samples without corresponding clinical survival information, and two samples derived from patients that succumbed due to unrelated causes, the remaining 276 samples were included in the current study. The inclusion criteria for HGSOC were as follows: Serous histological subtype and a pathological grade of 3. A total of 152 patients were identified and divided into high and low FOXD1 expression groups. The significance cut-off value was calculated using the X-Tile (version 3.6.1) bioinformatics tool (45). To identify predicted miRNA sequences that may regulate FOXD1, TargetScan (http://www.targetscan.org/) and miRanda (http://www.microrna.org) were employed (46,47). A total of 142 candidate miRNAs binding sites were identified by TargetScan (context, ++; score >90; version 7.1) and 146 candidate miRNA binding sites were identified by miRanda (version: August 2010 Release; mirSVR score ≤−0.1). To identify target genes that may be regulated by FOXD1, MatInspector (version 8.4; http://www.genomatix.de) and JASPAR (version 2016; http://jaspar.genereg.net) were employed (48,49). A region 2 kb upstream of target gene promoters was screened for FOXD1-binding sites.

Data are expressed as the mean ± standard deviation of three independent experiments. Overall survival and progression-free survival analysis was performed using Kaplan-Meier plots and a log-rank test. A Student’s t-test was selected to analyze significant differences between two groups. Analysis of variance and the Dunnett’s post hoc test was used to compare the means among >2 groups. The χ² test was used to compare clinical characteristics. P<0.05 was considered to indicate a statistically significant difference.

To determine the protein and mRNA expression levels of FOXD1 in HGSOC tissue samples and normal FT samples, western blot and RT-qPCR analyses were performed, respectively. At the protein level, the expression of FOXD1 was higher in FT (n=7) when compared with HGSOC (n=8) samples (Fig. 1A). At the mRNA level, the expression of FOXD1 was significantly higher in FT (n=11) compared with HGSOC (n=20) samples (Fig. 1B). To further confirm the role of FOXD1 in HGSOC, immunohistochemical staining was performed to measure its expression in the HGSOC TMA (n=120). As expected, the FOXD1 transcription factor was primarily located in the nucleus (Fig. 1C). The stained HGSOC tissue samples were scored and divided into high (n=58) and low (n=62) FOXD1 expression groups. The overall survival rate of the high expression group was significantly higher when compared with the low expression group (P=0.018; Fig. 1D). The association between FOXD1 expression and clinicopathological parameters was also analyzed; however, no statistically significant associations were identified (Table I). To investigate these observations further, the GSE9891 dataset was used to evaluate the prognostic role of FOXD1 at the mRNA level (44). The expression of FOXD1 in HGSOC samples alone was first determined. As demonstrated in Fig. 1E, the overall survival (P=0.030) and progression-free survival (P=0.003) of the low FOXD1 expression group was significantly reduced when compared with the high FOXD1 expression group. Samples of 276 in the GSE9891 cohort were then divided into high and low FOXD1 expression groups. As shown in Fig. 1F, the overall survival (P=0.010) and PFS (P=0.011) of the low FOXD1 expression group was significantly reduced when compared with the high FOXD1 expression group. These results provide evidence to suggest that reduced FOXD1 expression may predict poor prognosis in HGSOC patients.

To investigate the effect of FOXD1 expression on ovarian cancer cells, five ovarian cancer cell lines, including A2780, HO8910, SKOV3, OVCAR3, CAOV3 and one lung cancer cell line, H1299, were employed and transfected with a FOXD1 overexpression vector or an shRNA targeting FOXD1. As demonstrated in Fig. 2A, overexpression of FOXD1 significantly inhibited the colony-forming efficiency of all six cell lines when compared with empty vector controls. By contrast, downregulation of FOXD1 significantly increased their clonogenicity compared with an empty vector control (Fig. 2A). The cell cycle distribution among these transfected cell lines was then measured. For cell cycle analysis, the cells were synchronized in G1 phase using the double thymidine block method. Then the cells were released by replacing the medium with thymidine-free complete culture medium. As shown in Fig. 2B, upregulated FOXD1 expression induced cell cycle arrest at the G0/G1 phase and downregulation of FOXD1 reversed this effect. The results indicate that FOXD1 may suppress the proliferation of ovarian cancer cells by inducing cell cycle arrest in G1 phrase in vitro.

The Cancer Genome Atlas Research Network reported that HGSOC is characterized by TP53 mutations in ~96% of tumors (50). To further validate the role of FOXD1 in vivo, OVCAR3 (p53 mutant) and A2780 (p53 wild-type) cells with or without FOXD1 overexpression, were subcutaneously inoculated into nude mice (n=5 per group). Following 4–5 weeks, the mice were sacrificed, and the tumor weight and volume were measured. As demonstrated in Fig. 3A and B, the weight and volume of tumors from mice in the FOXD1 overexpression group was significantly lower when compared with the control group for both cell lines. Immunohistochemical staining of the xenograft tumor tissues was then performed to examine the expression of FOXD1 and p21. As shown in Fig. 3C, the expression of FOXD1 in nucleus was positively associated with p21 expression in the nucleus. The results indicate that FOXD1 may inhibit the proliferation of ovarian cancer cells in vivo.

From the results presented thus far, the authors hypothesized that FOXD1 may function as a tumor suppressor in ovarian cancer. To investigate the molecular mechanisms underlying the downregulation of FOXD1 expression in ovarian cancer, miRNA target prediction tools (miRanda and TargetScan) were used to identify putative miRNA sequences that may regulate FOXD1. Among all candidate miRNAs identified, miR-30a-5p and miR-200a-5p were selected for further investigation. According to previous studies, miR-30a-5p and miR-200a-5p have been observed to be upregulated in ovarian cancer (30–36). As demonstrated in Fig. 4A and B, the expression levels of miR-30a-5p and miR-200a-5p were determined in HGSOC and normal FT tissue samples. HGSOC samples were observed to exhibit significantly higher levels of miR-30a-5p and miR-200a-5p expression when compared with normal FT samples (Fig. 4A and B). A dual-luciferase reporter assay was subsequently employed to verify the binding sites for miR-30a-5p and miR-200a-5p in the FOXD1 3′-UTR region. Wild-type or mutant FOXD1 3′-UTR sequences were cloned into the luciferase vector, and then co-transfected with miR-30a-5p or miR-200a-5p mimics into 293T cells. Upregulation of miR-30a-5p or miR-200a-5p significantly decreased the luciferase activity of cells transfected with wild-type but not the mutant FOXD1 vector (Fig. 4C and D). The role of miR-30a-5p and miR-200a-5p in ovarian cancer cells was then investigated. OVCAR3 cells were transfected with miR-30a-5p and miR-200a-5p mimics or inhibitors. As shown in Fig. 4E and F, the expression of FOXD1 at mRNA and protein levels demonstrated a positive association with p21 expression. These results indicated that miR-30a-5p and miR-200a-5p may directly downregulate FOXD1 expression.

A region 2 kb upstream of the p21 promoter was screened for FOXD1-binding sites using online promoter prediction tools such as MatInspector and JASPAR. FOXD1 was identified as a putative regulator of p21 by JASPAR. A total of three candidate transcription factor binding sites (TFBSs) were also identified (Fig. 5A). A dual-luciferase reporter assay was employed to verify these TFBSs. As shown in Fig. 5B, two out of the three TFBSs in the FOXD1 overexpression group (sites B and C) demonstrated a significant increase in luciferase activity compared with the empty vector controls. When sites B and C were mutated, no significant difference in luciferase activity was observed between the FOXD1 over-expression group and control group (Fig. 5C). A ChIP assay and PCR were subsequently performed to further verify the TFBSs. As shown in Fig. 5D, enrichment of the p21 promoter fragments were confirmed by PCR using primers flanking FOXD1 binding sites (sites B and C). The expression of p21 in five ovarian cancer cell lines and one lung carcinoma cell line following overexpression or silencing of FOXD1 expression was then evaluated. As demonstrated in Fig. 5E, FOXD1 overexpression was associated with an increase in p21 protein expression, while a reduction in FOXD1 was associated with a decrease in p21 expression. These results, together with those obtained from the immunohistochemical analysis of p21 expression in ovarian tumor xenografts, suggest that FOXD1 may bind to the promoter of p21 directly to regulate its expression in vitro and in vivo.

To further investigate the role of FOXD1 in ovarian cancer cell proliferation via increasing p21 transcription, a rescue assay was performed by co-transfecting FOXD1 and p21-specific siRNA sequences in A2780 cells. The knockdown efficiency of the two p21-siRNAs is demonstrated in Fig. 6A. As indicated in Fig. 6B and C, the viability and clonogenicity of cells transfected with p21 siRNA and FOXD2 overexpression vectors was restored when compared with FOXD1 overexpressing cells transfected with siRNA controls. The results suggest that FOXD1 inhibits the proliferation of ovarian cancer cells via the direct targeting of p21.