Following approval of a research protocol by the Institutional Review Board Committees of Tulane University and Louisiana State University in New Orleans, 66 archival formalin-fixed paraffin-embedded (FFPE) tissue blocks of ovarian tissue were obtained from the Departments of Pathology at the Tulane University Health Sciences Center and from the Interim LSU Hospital. Sixteen samples were representative of normal ovaries from women who had undergone gynecological surgeries for non-ovarian related causes. The rest of the 50 samples consisted of 3 benign ovarian tumors, 7 tumors of low malignancy potential and 40 cases of ovarian carcinoma. Each case of ovarian tumor was matched with corresponding benign tissue control. All FFPE tissue blocks were sectioned and stained with hematoxylin and eosin for histological assessment.

Tissue (1.5–2 mg) was excised from each normal and tumor FFPE tissue block using a sterile scalpel and placed in an autoclaved 1.5 ml centrifuge tube. Samples were deparafinized with 1 ml of xylene followed by vortexing at top speed for 2 min. The tissue was then centrifuged at 10,000 x g for 3 min using Microcentrifuge 16 from Beckman Coulter Inc. (Brea, CA, USA) and the supernatant was pipetted out. To remove any residual xylene and facilitate pelleting, 1 ml 100% ethanol was added to the tissue sample, followed by spinning at 10,000 × g. The supernatant was decanted and tissue pellets were allowed to air-dry at room temperature. Subsequently, pellets were subjected to protease digestion by 100 μl/ml proteinase K in Digestion Buffer [10 mM Tris-HCl (pH 8.0), 1 mM ethylenediaminetetraacetic acid (EDTA)] at 52°C for 16 h. Following the digestion, DNA was isolated using the Qiagen (Valencia, CA, USA) DNeasy for FFPE kit, following the manufacturer’s recommended protocol.

Ovarian surface epithelial cells from normal ovarian FFPE tissue samples were dissected using the PALM^® Robot Microbeam laser microdissection system (PALM GmbH, Bernried, Germany) at the Louisiana State University Morphology and Imaging core facility. DNA was then extracted from the epithelial cells using the proteinase K DNA extraction Solution and incubation at 65°C for 16 h, as suggested by the manufacturer (Arcturus^®, Applied Biosystems, Life Technologies Corporation, Carlsbad, CA, USA).

Both tumor and normal DNA samples were genotyped for SNPs: rs2279605, rs10120967, rs7851737 using Applied Biosystems TaqMan probes, with IQ power mix (Bio-Rad, Hercules, CA, USA) or Amplitaq Gold, 25 mM MgCl₂ and 10X PCR Gold buffer from Applied Biosystems and dNTPs from VWR International (Radnor, PA, USA). Applied Biosystems Taq Man probes are labeled with Fam and Vic dyes. A total of 20 μl PCR reactions were set up in a 96-well plate which was covered with Microseal ‘B’ film from Bio-Rad. Bio-Rad Thermal cycler IQ5 was used to run the real-time PCR and Image Quant 5 software from Bio-Rad used for plate read document and analysis of the real-time data post PCR. All genotyping assays were done in triplicates and when the three independent assays yielded ambiguous results, were repeated again.

DNA (0.5 μg) from each sample (tumor and normal from the same patient) was digested using 5 units of Afe1 enzyme (New England Bioscience, Ipswich, MA, USA) in a total reaction volume of 50 μl. The digestion was performed in 1X NEBuffer (New England Bioscience); 1X NE buffer contains: 20 mM Tris-acetate, 50 mM potassium acetate, 10 mM magnesium acetate, 1 mM dithiothreitol (pH 7.9 at 25°C). The samples were incubated with the enzyme for 1 h at 37°C to allow digestion of DNA, following which Afe1 was inactivated by incubating the samples at 65°C for 20 min. Alternatively, digestion was performed overnight.

All samples were simultaneously PCR-amplified for the promoter region of the β-actin gene and NR5A1 gene in 200 μl tubes. For each 50 μl reaction, 2 μl of DNA was used and reagent concentrations were optimized at: 5 mM for MgCl₂ from Applied Biosystems, 2 mM for each dNTP from VWR; 5 mM for primers (β-actin, forward primer: 5′-TGC AAA GAA CAC GGC TAA GTG TGC-3′, β-actin, reverse primer: 5′TCT AAG ACA GTG TTG TGG GTG TAG GTs-3′, NR5A1 gene, forward primer: 5′-AAC ACC AAC AAA GAA GGC GAG AGG-3′, NR5A1 gene, reverse primer: 5′-TCA CTT ACG AAG CGG AAG CAGC-3′) from IDT DNA (Coralville, IA, USA) in 10X PCR buffer II [final concentration: 50 mM potassium chloride and 10 mM Tris-HCl (pH 8.3 at room temperature) from Applied Biosystems] along with 1.25 units AmpliTaq^® DNA polymerase per 50 μl of reaction. PCR amplification was performed on a PTC-100™ programmable thermocycler from MJ Research Inc. (Quebec, Canada) allowing initial denaturation at 95°C for 20 min followed by 40 cycles of 95°C for 1 min, 62°C for 1 min, 72°C for 1 min and completing the terminal extension with 10 min at 72°C.

A total of 8 μl of PCR product was added to 2 μl of loading dye (2% xylene cyanol, 40% glycerol in DDi H₂O) from Boston Bioproducts (Ashland, MA, USA). For sizing 1 kb plus DNA ladder (Invitrogen, Life Technologies Corporation, Carlsbad, CA, USA) was loaded on a 2% agarose gel containing 1X Tris-acetate-EDTA buffer (40 mM Tris acetate and 1 mM EDTA) and 5 μg of ethidium bromide for staining. The gel was run on a horizontal system for Gel electrophoresis from Bethesda Research Laboratories Inc. (Gaithersburg, MD, USA) at 100 V for 60 min. Following the gel electrophoresis the amplified fragments were visualized on Molecular Imager^® Gel Doc™ using Image Lab™ software from Bio-Rad.

Electrophoretic images were analyzed for relative (NR5A1/β-actin) band intensity using Image Quant 5.1 software from GE Healthcare (Piscataway, NJ, USA). Relative intensities were categorized in quartiles as follows: −, 1st; + 2nd; ++ 3rd; +++ 4th quartile. All experiments were done at least twice and the relative intensities were averaged.

A retrospective chart review was performed gathering clinical data on the patients for whom we had malignant ovarian tissue. Characteristics examined were: age, race/ethnicity, date of diagnosis, years survived since diagnosis, stage of ovarian cancer, histologic grade, date of debulking surgery and treatment with chemotherapy or radiation.

For most statistical calculations, two-tailed p-values were obtained using Fisher’s exact test. The log-rank test was used for calculating p-values for potential differences in survival.

The clinical characteristics of the ovarian tumor samples examined are shown on Table I.

In the current study, we considered two molecular hypotheses of SF-1 protein loss in ovarian tumors: LOH and increased methylation. To probe for the prevalence of LOH at the NR5A1 locus, we genotyped matched ovarian tumor and normal FFPE tissue DNA samples from the same ovarian cancer patient, for three NR5A1 gene SNPs: rs2279605, rs10120967 and rs7851737. SNPs were selected based on the following criteria: i) high (>30%) frequency of heterozygosity in the racial/ethnic groups present in our study population (based on available data at dbSNP: http://www.ncbi.nlm.nih.gov/projects/SNP/); ii) availability of a preoptimized Taqman SNP Genotyping Assay for each SNP. Genotyping was performed by Taqman SNP Genotyping Assays using real-time PCR. Assays were performed in triplicates and repeated again, if the genotyping results were ambiguous. All three genotyped SNPs were in Hardy-Weinberg equilibrium in normal samples (data not shown).

The genotyping results for the ovarian tumors and the LOH data for each sample, are shown in Fig. 1A. These data show that out of the 36 ovarian tumor tissues that were heterozygous for at least one of the three NR5A1 gene SNPs, 16 (44%) had LOH (Figs. 1A and 2). The majority of the ovarian tumors had a single LOH event at the NR5A1 locus (out of maximum three possible), but 5 tumors (14%) showed multiple LOH events (Fig. 1A). Also, each SNP showed LOH in multiple tumors, with rs7851737 showing most losses (Fig. 1B). Thus, LOH occurs frequently at the NR5A1 locus in ovarian tumors.

With regards to the type of observed loss, LOH events at rs2297605 were equally distributed between the two alleles, while the other two SNPs showed bias in the LOH events towards one of the two alleles (Fig. 1B). The significance of this finding is unclear, since these are non-coding SNPs. Interestingly, the genotyping results (Fig. 1A) also uncovered somatic mutations other than LOH in the tumors, manifested as allele gains; base substitutions turning a homozygous genotype (normal DNA) into a heterozygote genotype in the tumor, hence called ‘gain of allele’; Fig. 1B. These somatic NR5A1 substitutions were present in 10% of ovarian tumors (Fig. 2). Thus the genotyping data show frequent genetic (LOH/substitution) events at the NR5A1 locus in ovarian tumors.

A methylation-sensitive restriction enzyme (Afe1) method (e.g. 19) was used to quantify site-specific methylation at −30 bp (compared to translation start) of the NR5A1 gene promoter in ovarian tumor tissue from patients with ovarian cancer and in matched normal tissue from the same patients (when available).

Afe1 cleaves genomic DNA at 5′-AGC/GCT-3′, but cleavage is blocked by methylation (http://www.neb.com/nebecomm/products/productr0652.asp). Since the Afe1 enzyme cleaves the un-methylated CpG’s, only methylated CpG’s can be amplified and quantified following gel electrophoresis. Complete DNA digestion was confirmed by prolonged (overnight) Afe1 digestion, which yielded similar results (data not shown). To control for differences in DNA level and/or quality between tumor samples, we also amplified β-actin as an internal control. The primers used for the β-actin gene were selected to amplify a region that lacks an Afe1 cleavage site. Therefore, β-actin is amplified regardless of methylation status, and relative band intensity (NR5A1/β-actin) was used as a function of NR5A1 gene methylation (see Materials and methods for details). This analysis indicated that 36 out of 46 (78%) ovarian tumors had appreciable NR5A1 methylation (2nd, 3rd and 4th quartile of methylation levels), and 17/46 (37%) had high levels of NR5A1 methylation (3rd and 4th quartile of methylation levels; Table II). Thus the NR5A1 gene is methylated in most ovarian tumors. Furthermore, we detected both a high level of NR5A1 gene methylation and LOH in 21% of the ovarian tumors that we analyzed (Fig. 1 and Table II). The cumulative data also demonstrate that 62% of the ovarian tumors have LOH, high level of methylation, or both (Fig. 2) at the NR5A1 locus.

As indicated by Table I, most ovarian tumors are diagnosed at an advanced stage, reducing the ability to obtain normal ovarian tissue from most patients. In the absence of an adequate number of matched normal ovaries available for study, 16 non-tumor ovaries were analyzed (from unrelated individuals) to evaluate the relative methylation of the promoter region of the NR5A1 gene in normal ovarian tissue, with the same methylation-sensitive restriction enzyme method used above. Since human ovarian tumors have epithelial origin (2,20), we obtained OSE cells from these ovarian tissues (by laser-capture microdissection) and analyzed them following DNA extraction. β-actin was again amplified from each sample as an internal control. These data show that only one out of the 11 (9%) normal ovaries that were successfully evaluated (i.e. that had β-actin amplification) showed appreciable NR5A1 methylation (Table III). This difference between the prevalence of NR5A1 methylation in tumor versus normal ovaries is statistically significant (p<0.0001). Thus, ovarian tumor tissues display significantly more frequent NR5A1 gene methylation than normal ovarian epithelial tissues.

Retrospective analysis of the clinical data suggest that presenting stage and histologic grade of ovarian tumors are not significantly affected by the presence of somatic NR5A1 gene alterations (Table I and data not shown). Furthermore, Kaplan-Meier survival curves were similar for both ovarian tumors with and without NR5A1 gene alteration (LOH/methylation; data not shown). Likewise, the presence of NR5A1 gene alteration did not correlate with race/ethnicity or treatment, such as radiation and chemotherapy (Table I and data not shown).