Vitamin D–Binding Protein Enhances Epithelial Ovarian Cancer Progression by Regulating the Insulin-like Growth Factor-1/Akt Pathway and Vitamin D Receptor Transcription

Epithelial ovarian cancer (EOC) is the leading cause of gynecologic malignancy–related deaths worldwide and is a substantial health threat to women (1, 2). Many patients eventually develop chemoresistant relapsed disease and die despite surgery and combination chemotherapy. Progress in improving the survival in EOC has been slow, despite significant advances in treatment over the past 25 years. EOC has an extraordinary biology and behavior at the clinical, cellular, and molecular levels. However, ascites can develop in any histologic subtype (3) and is a sign of advanced disease and poor prognosis (4). Both the fluid and cellular components of ascites provide opportunities for investigation of prognostic biomarkers, molecular profiling analysis (3), and evaluation of the role of ascites in chemoresistance and mechanisms of tumor progression (5).

Our preliminary proteomic study demonstrated that vitamin D–binding protein (DBP) is among the most upregulated proteins in EOC (Supplementary Fig. S1A). Significantly higher ascitic DBP levels were confirmed in patients with EOC compared with those with benign gynecologic disease or ovarian tumors with low malignant potential (Supplementary Fig. S1B).

DBP is a glycosylated α-globulin and the primary transporter for vitamin D and its metabolites. DBP is involved in actin and fatty acid binding, neutrophil chemotaxis, and macrophage activation (6). Vitamin D has a protective effect against carcinogenesis and cancer progression (7–13). However, its role in the development of human cancer is controversial (6, 14–18). To date, no investigation has focused on the association between DBP and EOC progression, or on the precise molecular mechanisms underlying the interaction between DBP and the vitamin D receptor (VDR).

To determine whether DBP expression in body fluids and malignant tissue could serve as a prognostic biomarker, correlations between DBP and clinicopathologic variables and their influence on long-term prognosis in patients with EOC were evaluated. The mechanisms underlying DBP involvement in the regulation of cancer progression were also explored. We also conducted animal and in vitro experiments, which generated results consistent with our clinical findings, and illustrated the importance of DBP in the regulation of EOC progression.

The research protocol was approved by National Cheng Kung University Hospital (NCKUH) Institutional Review Board and National Taiwan University Hospital (NTUH) Research Ethics Committee. Informed consent was obtained. Initially, ascitic and serum DBP analysis included samples and data from patients diagnosed with EOC between January 2002 and August 2012 at NCKUH in Southern Taiwan. To examine ascitic and serum DBP expression levels at the time of primary surgery and their prognostic significance in an expanded EOC population, 105 NTUH patients in Northern Taiwan diagnosed between 2002 and 2012 were included in the analysis, along with 115 NCKUH patients. The control group constituted patients diagnosed with borderline (n = 16) or benign tumors (n = 42) of the ovary at NCKUH. For DBP IHC, patients with EOC (n = 245) diagnosed at NCKUH between November 1993 and August 2012 were included. Patients diagnosed with EOC underwent comprehensive staging or cytoreductive surgery, with or without postoperative platinum-based chemotherapy. The staging was performed according to the criteria of the International Federation of Gynecology and Obstetrics (FIGO). Cancer progression was defined according to the objective Response Evaluation Criteria in Solid Tumors 1.1 or the Gynecologic Cancer Intergroup definition for CA 125 progression. Both progression-free survival (PFS) and overall survival (OS) were calculated from the diagnosis. The date of the last record retrieved was July 31, 2016. Medical records and pathology slides were reviewed for patient demographic data, clinical characteristics, progression-free interval (PFI), and treatment outcomes.

All participants provided serum and ascitic samples preoperatively. Samples were stored in the vapor phase of liquid nitrogen (-196°C). An ELISA Kit (Immundiagnostik AG; sensitivity 1.23 ng/mL) was used to measure DBP levels in duplicate. Measurements were repeated if the correlation coefficient between absorbance and the amounts in standards was <0.95. Any test with intra-assay coefficients of variation (CV) >10% was repeated. The average intra- and interassay CV was 5.0% and 12.7%, respectively.

Sections were prepared as described previously (19). Formalin-fixed, deparaffinized tissue sections were stained for DBP protein after microwave-enhanced epitope retrieval using a standard automated IHC procedure (XT autostainer; Ventana Medical Systems). The primary antibody (rabbit polyclonal anti-DBP, A0021; Dako Denmark A/S) was applied at a dilution of 1:200, and negative controls were treated with PBS. Intensely DBP-positive normal hepatocytes were used as positive controls. The investigator conducting these experiments (W.-L. Shen) was blinded to patient clinical outcome data. Staining intensity was categorized as follows: negative, weak, moderate, and strong staining (grade 0, 1, 2, and 3, respectively; Supplementary Fig. S2).

The human ovarian cancer cell lines ES-2, A2780, and A2780CP70; HAC-2, and OVCAR-3 and OVCAR-4 were obtained from the ATCC, the Japanese Collection of Research Bioresources (JCRB) Cell Bank, and the National Cancer Institute DTP tumor repository program, respectively. The ov2008 and ov2008CP20 cell lines were kindly provided by Dr. Macus T. Kuo (The University of Texas MD Anderson Cancer Center, Houston, TX), and cisplatin- (ES-2/CP) and paclitaxel (ES-2/TX)-resistant strains of ES-2 were developed in our laboratory as described previously (19). In vitro invasion selection to obtain highly invasive OVCAR-3 cells was performed as described previously (20). HAC-2, OVCAR-3, OVCAR-4, A2780, A2780CP70, ov2008, and ov2008CP20 cells were grown in RPMI1640. ES-2, ES-2/CP, and ES-2/TX cells were grown in Mycos 5A medium. Both media were supplemented with 10% FBS. Cells were grown at 37°C in a 5% CO₂ atmosphere, cultured, and stored according to the supplier's instructions. Cells were used between passages 5 and 20. Once resuscitated, cell lines were routinely authenticated (once every 6 months, with the last test in April 2017) through cell morphology monitoring, growth curve analysis, species verification by iso-enzymology and karyotyping, identity verification using short tandem repeat profiling analysis, and contamination checks.

Antibodies used were anti-DBP (PA5-29082, Thermo Fisher Scientific); anti-VDR, anti-calbindin-D9K, and anti-calbindin-D28K (sc-13133, sc-74462, and sc-365360, respectively; Santa Cruz Biotechnology); anti–IGF-1, anti–phospho-IGF-1, anti-Akt, anti–phospho-Akt, and anti–β-actin (3027, 3024, 4060, 9272, and 8480, respectively; Cell Signaling Technology); anti-IGFBP2 (ab109284, Abcam); and horseradish peroxidase–conjugated goat anti-mouse IgG and goat anti-rabbit IgG (sc-2005 and sc-2054, respectively; Santa Cruz Biotechnology). The NF-κB inhibitor was Bay 11-7082 (B5556; Sigma-Aldrich).

DBP cDNA (RG202051) and knockdown plasmids expressing short hairpin RNA (shRNA) targeting DBP (sc-41375-SH) and VDR (sc-45920-SH) were obtained from OriGene and Santa Cruz Biotechnology, respectively. To establish stable clones, DBP knockdown (shDBP) or control (shControl) plasmids were transfected into ES-2/CP cells using HyFect (Leadgene). After 24 hours, stable transfectants were selected in 500 μg/mL puromycin (Sigma-Aldrich). The selection medium was replaced every 3 days for 2 weeks, and clones of resistant cells were isolated and allowed to proliferate in a medium containing puromycin (500 μg/mL).

Cells were washed in PBS and lysed in lysis buffer (50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L MgCl₂, and 0.5% Triton X-100). Lysates were cleared by centrifugation (13,000 × g, 20 minutes, 4°C), separated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and probed with the indicated antibodies. The signals were detected by chemiluminescence.

Cells were cultured in a humidified incubator (95% air, 5% CO₂, 37°C) in 96-well flat-bottomed microtiter plates for 24, 48, 72, and 96 hours, and the number of viable cells at each time point was counted using the 3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich) assay by monitoring the absorbance at 570 nm. The doubling time was calculated by the formula in the technical information for working with animal cells in culture, provided by ATCC.

Cells were seeded at approximately 2 × 10⁵/well in transwell plates and assays performed as described previously (21).

All animal studies were approved by the Institutional Animal Care and Use Committee at National Cheng Kung University (Tainan, Taiwan). A lung metastasis model was established in 6-week-old female NOD/SCID mice (NOD.CB17-Prkdc Scid/NcrCrl) supplied by the Animal Center at National Cheng Kung University. Animals were acclimated for 1 to 2 weeks while being caged in groups of five under pathogen-free conditions. The mice were fed a diet of animal chow and water throughout the experiment. The mice were randomized into two groups that were administered either 10⁵ ES-2/CP/shControl or ES-2/CP/shDBP cells in 100 μL of Hank's Balanced Salt Solution via intrabursal injection (22). After 30 days, the mice were euthanized by anesthetic overdose, and all organs were examined for evidence of metastases. Lung samples were immersed in Bouin's solution, and all organs were removed, weighed, fixed in 10% formalin, embedded in paraffin, and sectioned (4 μm) for histopathology and IHC analyses. Paraffin sections of tumors were stained with hematoxylin and eosin. Ascites tumor cells were stained using Papanicolaou stain. One investigator (S.-L. Peng) was responsible for interpreting the extent of cancer involvement in each organ, the intensity and percentage of DBP staining, and the cellularity of ascitic cancer cells. An ELISA Kit (BlueGene Biotech; sensitivity, 1.0 ng/mL) was used to determine ascitic DBP levels. The average intra- and interassay CVs were 5.6% and 7.9%, respectively.

The region upstream and flanking the start site of the VDR gene (−480/+1; GenBank: AB002168.2) was PCR-amplified using the primers VDR forward 5′-GGTACCACAGGTTGCGACGGAGCCCG-3′ and reverse 5′-CTCGAGAGACAGCCCAGCACCTGGCC-3′, and cloned into the KpnI and XhoI sites of the pGL4 vector (Promega). The resulting construct was confirmed by DNA sequencing. A QuikChange Site-Directed Mutagenesis Kit (Stratagene) was used to generate the NF-κB and SP1 point-mutant constructs, using VDR (−480/+1) as a template. The primers used were as follows (only forward primers are listed): NF-κB–mutant primer: 5′-CGGGATGGACCATTCGTCGGAG-3′; and SP1-mutant primer, 5′-GGGCGGGGCGGGGCCGGGGC-3′.

Luciferase assays were conducted using a luciferase reporter assay system (Promega) 48 hours after transfection (21). Normalized luciferase activity is reported as luciferase activity/β-galactosidase activity.

Native protein–DNA complexes were crosslinked by treatment of cells with 1% formaldehyde for 15 minutes. Chromatin immunoprecipitation (ChIP) assays were performed as reported previously (21). Briefly, equal aliquots of isolated chromatin were subjected to immunoprecipitation with anti-p65 (sc-372, Santa Cruz Biotechnology) targeting the NF-κB p65 subunit, and IgG mAbs (sc-2005, Santa Cruz Biotechnology). The following primers were used: VDR, forward 5′-GGTACCACAGGTTGCGACGGAGCCCG-3′ and reverse 5′-GGGCTGACCAGGCCAGGACT-3′.

Total RNA (5 μg) was used as the template in cDNA synthesis reactions with random primers using Superscript III reverse transcriptase (Applied Biosystems). The resulting cDNAs were used (1: 20 dilution) to detect the level of endogenous DBP, VDR, calbindin-D9K, and calbindin-D28K mRNA expression by qPCR as described previously (21). The following primers were used: DBP (HS00167096), VDR (HS00172113), calbindin-D9K (HS0018785), calbindin-D28K (HS01077197), and GAPDH (HS99999905).

IGF-1 in culture supernatants was quantified with an ELISA Kit (ab100545; Abcam; sensitivity 2 ng/mL). Measurements were repeated if the correlation coefficient between absorbance and the amounts in standards was <0.95. Any test with intra-assay CV >10% was repeated.

Data were analyzed using SPSS (version 17.0; SPSS Inc.). Interval variables are presented as the mean ± SEM or median ± interquartile range. Animal studies involving measurement of tumor size and survival are generally performed with one control per experimental subject. The reason is that the sample size for the experimental and control groups needs to be sufficient to reject the null hypothesis that the population means of the experimental and control groups are equal with a power of 0.8 and type I error of 0.05. We estimated this using PS: Power and Sample Size Calculation (version 3.1.2; William D. Dupont and Walton D. Plummer Jr). We estimated the number of mice required to assess tumor volume as being at least four for each of the control and experimental groups. Differences between groups were evaluated using the Mann–Whitney U test. ROC curve–determined cut-off values of DBP levels were optimized for diagnostic sensitivity and specificity for the prediction of progression or death. To study the relationships between levels of DBP in serum or ascites and clinicopathologic characteristics at diagnosis, we divided all recruited patients into two subgroups according to clinical or pathologic variables. Frequency distributions between categorical variables were compared using the Pearson χ² and Fisher exact tests. Survival was estimated using the Kaplan–Meier method and compared using log-rank tests. Cox proportional hazards models were used to estimate HRs and confidence intervals (CIs). Possible confounders were included in multivariate analysis with the studied biomarkers. The independent effect of ascitic DBP level on survival and disease progression was analyzed in multivariate analysis. P < 0.05 (two-sided) was considered statistically significant.

Patients newly diagnosed with EOC (n = 115) aged 27 to 84 years (mean, 55 years), with a body mass index of 17.7 to 43.9 kg/m² (mean, 23.3 kg/m²), constituted the test set. Of these, 100 (87.0%) received adjuvant platinum-based chemotherapy. The median follow-up time was 29 months (range, 1–122 months). During follow-up, 51 patients (44.3%) developed progressive disease and 38 patients (33.0%) died.

The relationships between DBP expression and clinicopathologic factors are presented in Table 1. Women with high cancer antigen (CA) 125, advanced stage, serous histology, residual tumor size ≥1 cm, poor chemoresponse, PFI <12 months, and cancer progression had significantly higher ascitic DBP levels. No association was observed between serum DBP levels and these variables.

The influence of ascitic and serum DBP levels at diagnosis and other potential prognostic features on time to death and time to disease progression were estimated (Table 2). In the univariate analysis, high ascitic DBP levels, advanced stage, serous histology, and residual nodules ≥1 cm were significantly associated with high risk of death and disease progression. High CA 125 was related to the risk of cancer progression. In a multivariate-adjusted model, advanced stage was an independent predictor of death (HR, 2.8; 95% CI, 1.08–7.25; P = 0.034) and disease progression (HR, 2.64; 95% CI, 1.29–5.41; P = 0.008). Notably, high ascitic DBP levels were also an independent predictor of cancer progression (HR, 2.85; 95% CI, 1.35–6.01; P = 0.006).

Long-term OS and PFS curves for the 115 test set patients are presented in Supplementary Fig. S3. Patients with high ascitic DBP levels (>140.7 μg/mL) had significantly poorer OS and PFS than patients with low ascitic DBP levels (≤140.7 μg/mL; P = 0.02 and P < 0.001, respectively), whereas OS and PFS did not differ between patients with low and high serum DBP levels (≤280.1 and >280.1 μg/mL, respectively).

DBP expression and clinicopathologic factors are shown among a pool of patients with EOC in Northern and Southern Taiwan (Supplementary Tables S1 and S2). Women with high CA 125, advanced stage, serous histology, residual tumor size ≥1 cm, PFI <12 months, and cancer progression, and those who died had significantly higher ascitic DBP levels. In contrast, those who were premenopausal and who had cancer progression had significantly higher serum DBP levels. High ascitic DBP levels were significantly associated with an elevated risk of disease progression in univariate analysis and were also an independent prognostic factor of cancer progression (HR, 1.82; 95% CI, 1.17–2.84; P = 0.008) in multivariate analysis.

Samples and data from 245 patients diagnosed with EOC were included. During follow-up, 118 patients (48.2%) developed progressive disease and 117 patients (47.8%) died. Associations between DBP expression in tumor tissue at diagnosis and clinicopathologic factors were examined (Supplementary Table S3). Cellular DBP overexpression was significantly associated with PFI <6 months (P = 0.049); however, there was no correlation between cellular DBP expression and age, stage, tumor differentiation, residual tumor size, or cancer death. Moreover, no difference in overall survival was observed between patients with low and high cellular DBP expression (data not shown).

DBP expression in cancer cells and culture media from different ovarian cancer cell lines is illustrated in Supplementary Fig.S 4. To examine whether DBP modulates the invasion and growth of ovarian cancer cells, we generated an ES-2 cell line stably overexpressing DBP and an ES-2/CP cell line, where DBP was stably knocked down using shRNA. As expected, Western blotting showed that DBP expression increased in cells transfected with the DBP-expressing vector and decreased in DBP knockdown cells (Fig. 1A, top). The doubling time of DBP-overexpressing cells decreased compared with that of controls. Conversely, the doubling time of DBP-knockdown cells increased compared with that of controls (Fig. 1A, bottom, 33.28 ± 1.82 vs. 24.82 ± 1.34 hours for ES-2 cells; 29.19 ± 0.32 vs. 45.21 ± 0.89 hours for ES-2/CP cells). Transwell invasion assays indicated that invasion capacity was increased and attenuated in DBP-overexpressing and DBP knockdown cells, respectively (Fig. 1B).

We also used in vitro Boyden chamber invasion selection to obtain highly invasive OVCAR-3 cells to determine whether DBP regulates cell growth and invasiveness. Four cycles of selection yielded more invasive ovarian cancer sublines from OVCAR-3 parental cells. The highly invasive cells exhibited both higher DBP protein expression (Fig. 1C, top) and higher invasion capacity (Fig. 1D) than the parental cells. Cell doubling time was prolonged (Fig. 1C, bottom; 35.42 ± 1.91 hours for parental unselected OVCAR-3, 25.45 ± 2.00 hours for shControl of selected OVCAR-3, and 49.10 ± 1.38 hours for shDBP of selected OVCAR-3 cells), and the invasion capacity of the selected highly invasive cells was impaired by DBP silencing (Fig. 1D). Taken together, these results demonstrate that DBP expression is associated with cell invasion ability and growth of ovarian cancer cells.

To test whether manipulation of DBP alters ovarian cancer growth and spread in vivo, shControl- or shDBP-expressing ES-2/CP cells were inoculated into NOD/SCID mice (n = 7 and 6, respectively) intrabursally. An ES-2–derived cell line was selected for the animal study because, in the panel of the aforementioned ovarian cancer cell lines, ES-2 cells were the only cell type able to form ascites in mice. Western blotting confirmed that DBP expression was decreased in the ES-2/CP cell clone stably expressing shDBP compared with shControl cells (Fig. 2F). The body weight of animals inoculated with shControl or shDBP cells remained relatively unaltered, suggesting negligible toxicity (Fig. 2G). The mean survival rate of mice inoculated with shDBP cells was significantly higher than that of controls (P = 0.011; Fig. 2H). Ovarian tumor size was decreased in mice inoculated with shDBP cells (n = 4) compared with controls (n = 5; Fig. 2A, top and middle; 1355.10 ± 104.70 mm³, shControl vs. 592.40 ± 201.90 mm³, shDBP; P = 0.014). DBP protein levels were also decreased in cancerous tissues of mice inoculated with shDBP cells compared with that of controls (Fig. 2A, bottom).

Notably, the mean abdominal circumference of control inoculated mice was larger than that of mice inoculated with shDBP cells (Fig. 2I). The ascites amount (Fig. 2B, left; 3.7 ± 0.54 mL, shControl vs. 0.3 ± 0.12 mL, shDBP; P = 0.013) and the number of tumor cells in the ascites (Fig. 2B, right; 17368.90 ± 1914.90, shControl vs. 7036.30 ± 1804.60, shDBP; P = 0.01; Fig. 2J) were significantly reduced in mice inoculated with shDBP cells compared with that in controls. The DBP levels in mouse ascites were also significantly decreased in mice inoculated with shDBP cells compared with those in controls (Fig. 2C; 103.88 ± 9.54 ng/mL, shControl vs. 42.51 ± 8.93 ng/mL, shDBP; P = 0.003). Furthermore, rapid growth of the intraperitoneal tumor and tumor invasion in the liver, kidney, diaphragm, and intestine (Fig. 2D) were obvious in the control mice. In contrast, DBP knockdown abolished tumor formation inside the peritoneum (Fig. 2D). Interestingly, formation of lung tumor nodules was also observed, and the mean number of lung tumor nodules in control mice was greater than in mice inoculated with shDBP cells (Fig. 2D; 85.40 ± 8.59, shControl vs. 5.30 ± 2.06, shDBP; P = 0.013). These results suggest that DBP plays an important role in the development of ovarian tumors, malignant ascites, and cancer invasiveness.

The antiapoptotic effect of IGF-1/Akt has been shown to be antagonized by vitamin D signaling (23, 24). Thus, to test whether the VDR/IGF-1/Akt signaling pathway is involved in DBP-mediated cell invasion, the expression of VDR/IGF-1/Akt–related proteins was examined in cells with differing DBP status. We found that the expressions of VDR, phosphorylated IGF-1 (p-IGF-1), IGFBP2, and phosphorylated Akt (p-Akt) were increased in DBP-overexpressing cells and decreased in DBP knockdown cells (Fig. 3A). Similar results were observed in selected invasive OVCAR-3 cells (Fig. 3B). To further investigate the role of VDR in DBP-mediated cell invasion, DBP-overexpressing cells were transfected with shRNA targeting VDR. The upregulation of p-IGF-1 and p-Akt in DBP-overexpressing cells was inhibited by VDR knockdown (Fig. 3C). Moreover, DBP-enhanced cell invasion was also inhibited by VDR silencing (Fig. 3D). These results suggested that modulation of cell invasion by DBP is predominantly mediated through VDR activation of the IGF-1/Akt signaling pathway.

Calbindins are products of VDR-responsive genes and are regulated by the active form of vitamin D, 1, 25-dihydroxyvitamin D3 (25–27). To further examine the effect of DBP on activation of the vitamin D pathway, the expression of calbindin-D9k and -D28k was examined in cells with differing DBP status. The expression of calbindin-D9k and -D28k was decreased in DBP-overexpressing cells and increased in DBP knockdown cells (Fig. 3E). Similar results were observed in selected invasive OVCAR-3 cells (Fig. 3F). To determine whether DBP decreases calbindin-D9k and -D28k expression via VDR expression, DBP-overexpressing cells were transfected with VDR shRNA. The expression of calbindin-D9k and -D28k was decreased in DBP-overexpressing cells (Fig. 3G). These results suggest that DBP inhibits the activation of the vitamin D pathway in the presence of VDR expression.

To further explore the mechanism by which DBP regulates VDR levels, its effect on transcription from a fragment of the VDR promoter spanning positions −480 to +1 relative to the transcription start site cloned into a luciferase reporter plasmid, and then transiently transfected into cells with differing DBP status was determined. Activity from the VDR promoter was higher in DBP-overexpressing cells compared with that in control cells. Conversely, VDR promoter activity was lower in DBP knockdown cells than in control cells (Fig. 4A, top). Similar results were observed in selected invasive OVCAR-3 cells (Fig. 4A, bottom). These results are unlikely to have resulted with cytotoxicity induced by transfection because the luciferase constructs were only transiently expressed in this experiment. Together, these data indicated that the region between −480 and +1 of the VDR promoter is important for transcriptional regulation by DBP.

NF-κB has been identified as a transcriptional factor involved in ovarian cancer initiation and progression (28–31). TRANSFAC predicted putative NF-κB and SP1-binding sites in the region between −480 and +1 of the VDR promoter (Fig. 4B). To determine whether either of these transcription factors can regulate transcription through the VDR promoter, constructs with mutations in the NF-κB or SP1-binding sites were generated by site-directed mutagenesis and then transiently transfected into DBP-overexpressing cells and control cells. Luciferase activity from the VDR promoter reporter plasmid containing the SP1-binding site mutations was higher in DBP-overexpressing cells than in control cells, whereas that of the reporter plasmid carrying an NF-κB–binding site mutation was almost identical in both cases (Fig. 4B). Moreover, ChIP analysis showed that NF-κB p65 binding to the VDR promoter region was increased in DBP-overexpressing cells and inhibited in DBP knockdown cells (Fig. 4C, left). Consistent results were observed in selected invasive OVCAR-3 cells (Fig. 4C, right). Silencing NF-κB using an NF-κB inhibitor suppressed its binding to the VDR promoter in DBP-overexpressing cells (Fig. 4D). These results indicated that the NF-κB p65 binding site in the VDR promoter is important for transcriptional regulation of VDR by DBP.

We next evaluated the gene expression levels of calbindin-D9k and -D28k are involved in DBP-mediated VDR upregulation. The DBP and VDR mRNA expression levels were elevated by DBP overexpression but were reduced by DBP knockdown (Fig. 4E, top). The expressions of calbindin-D9k and -D28k were decreased in DBP-overexpressing cells and increased in DBP knockdown cells (Fig. 4E, top). Similar results were observed in selected invasive OVCAR-3 cells (Fig. 4E, bottom). The calbindin-D9k and -D28k in DBP-overexpressing ES-2 cells were increased by transfection of a shRNA against VDR (Fig. 4E, middle). In addition, the IGF-1 protein expression in the culture supernatant was also examined. The IGF-1 level was increased in DBP-overexpressing cells and decreased in DBP knockdown cells (Fig. 4F, top). Similar results were observed in selected invasive OVCAR-3 cells (Fig. 4F, bottom). The IGF-1 production in DBP-overexpressing ES-2 cells was increased by transfection of a shRNA against VDR (Fig. 4F, middle). The expressions of VDR, calbindin-D9k, and -D28k at the gene and protein levels regulated by DBP were consistent in serous and clear cell carcinoma cell lines. Furthermore, the alterations in cellular p-IGF-1 levels regulated by DBP or VDR are compatible with IGF-1 levels in culture supernatant in the same cell lines.

In the current study, patients with EOC with ascitic DBP overexpression had poorer chemoresponses after first-line platinum-based regimens, shorter PFI, increased cancer progression, and shorter PFS, indicating that elevated ascitic DBP expression is an unfavorable independent tumor progression biomarker of long-term survival. Cellular DBP overexpression was also linked to chemoresistance in the clinic. Furthermore, our findings in an orthotopic animal model reflect the important role of DBP in tumor formation, ascites formation, invasiveness, metastasis, and survival. These in vivo results were reinforced by in vitro findings in serous and clear cell carcinoma cell lines that DBP regulates VDR and thus promotes cell aggressiveness via activation of the IGF-1/IGFBP2/Akt axis and suppression of vitamin D–responsive genes (Fig. 4G). Notably, an NF-κB–binding site in the VDR promoter was identified as a major determinant of DBP-dependent VDR promoter activation.

Circulating DBP is associated with the development or outcome of various cancers. Patients with low circulating DBP concentrations have an elevated risk of pancreatic cancer (14), renal cell carcinoma (16), and colorectal cancer (17). However, other investigations have reported no association between circulating DBP levels and bladder cancer risk (15) or colorectal cancer (18). Turner and colleagues demonstrated that low serum DBP levels were related to cancer-specific death in lung cancer, whereas no correlation was found between vitamin D and survival (32). Few studies have addressed the role of DBP levels in ascites or tumors. Our study is the first to determine the prognostic value of ascitic DBP, rather than serum DBP, for prediction of poor outcomes in patients with EOC.

Vitamin D and its analogues inhibit cell growth and proliferation and promote apoptosis and differentiation of cancer cells (33). Chemoprevention with these agents has been tested in the clinic. Although 1,25(OH)₂D can induce apoptosis through the IGF1R–Pi3K–Akt–dependent pathway (34, 35) by downregulation of IGFBP2 and matrix metalloproteinases (36), there is a lack of information regarding the role of DBP in the IGF-1/IGFBP2 pathway. Here, we demonstrate that cellular IGFBP2 and p-IGF-1 are downregulated by DBP knockdown in chemoresistant cell lines and upregulated by DBP overexpression in their chemosensitive counterparts. We previously provided evidence that high serum IGFBP2 levels are correlated with advanced stage, serous histology, and an elevated risk of EOC-specific mortality (37). Our findings of IGF-1 levels in culture supernatant are also consistent with cellular p-IGF-1 expressions regulated by DBP. Collectively, these findings support the involvement of DBP overexpression in EOC progression through activation of the IGF-1/IGFBP2/Akt pathway.

In the presence of VDR expression, DBP decreased the protein and mRNA expression levels of vitamin D–dependent calcium-binding proteins (calbindin-D9k and D28k). Thus, DBP appears to have an important role in ovarian cancer progression via the elimination of 1,25(OH)₂D function. These findings support the “free hormone hypothesis” (38) that DBP binds to 1,25(OH)₂D to prevent bioavailable vitamin D from entering the nucleus, where it interacts with the VDR and stimulates downstream gene expression (e.g., calbindin-D28k).

Human VDR is a member of the steroid hormone receptor superfamily, which is stabilized until the activation of ligand-VDR-dependent transcriptional activity (39). VDR localized in the nucleus interacts with 1,25(OH)₂D to cause genomic effects, whereas in the caveolae of the cell membrane this interaction mediates nongenomic effects (7). Tissue-specific expression of VDR contributes to variations in vitamin D signaling (39). Tumors with the highest expression of VDR respond better to treatment with vitamin D analogues in colorectal, breast, and prostate cancers. Our study indicates that, in addition to the established antiproliferative effects of the VDR in the aforementioned cancers, VDR may also be required for promotion of EOC progression resulting from DBP overexpression. We provide evidence that the VDR is required to activate vitamin D signaling. However, the VDR is also required for DBP-regulated cell invasiveness, which is modulated by binding of NF-κB to the VDR promoter.

Although intracellular DBP overexpression was correlated with chemoresistance, it was not associated with patient survival in the clinic. Rather, ascitic DBP overexpression was predictive of patient prognosis. In contrast to observations in human patients, intracellular DBP expression was related to the formation of ascites and tumor growth in our animal study. It remains unknown which patient population is susceptible to elevated ascitic DBP levels, and an investigation of the underlying mechanism is warranted.

In conclusion, based on the findings of our proteomics study, we examined DBP in ascites as a novel predictor of prognosis for patients with EOC. The strengths of this study include the relatively large number of participants and the consistent observation of DBP simulation of malignant ascites production in mice, along with increased intraperitoneal spread and distant metastasis of EOC after intrabursal injection of ovarian cancer cells in the model animals. Finally, we exploited the relationship of DBP with the IGF-1/IGFBP2/Akt pathway in promoting tumor progression to reveal the role of regulation of VDR by DBP in inhibiting the antitumor activity of the vitamin D pathway in EOC.

No potential conflicts of interest were disclosed.

Conception and design: Y.-F. Huang, Y.-H. Wu, W.-L. Shen, C.-Y. Chou

Development of methodology: Y.-F. Huang, Y.-H. Wu, W.-F. Cheng, C.-Y. Chou

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y.-F. Huang, Y.-H. Wu, W.-F. Cheng, S.-L. Peng, W.-L. Shen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.-F. Huang, Y.-H. Wu, S.-L. Peng, W.-L. Shen

Writing, review, and/or revision of the manuscript: Y.-F. Huang, Y.-H. Wu, W.-L. Shen, C.-Y. Chou

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.-F. Huang, Y.-H. Wu, C.-Y. Chou

Study supervision: C.-Y. Chou

This work was supported by grants from the Ministry of Science and Technology (MOST: no. 103-2314-B-006-070 and 104-2314-B-006-088) and was partly supported by National Cheng Kung University Hospital (No. NCKUH-10605014). The principal investigator associated with the aforementioned grants is Y.-F. Huang.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.