The human ovarian cancer cell lines studied were SK-OV-3, ES-2, PA-1, Caov-3, TOV-21G, TOV-112D, OV-90, OVCAR-3, and MDAH 2774, which were purchased from the American Type Culture Collection (Manassas, VA, USA), as well as A2780 and A2780cis, which were purchased from the European Collection of Cell Cultures (London, UK). All cell lines were initially cultured using the media and supplements recommended by the suppliers. Table S1 summarizes the components of culture media for individual cell lines (Table S1). All 11 cell lines were grown as monolayers and attached cells were fully disaggregated by trypsinization between passages. The cell lines were maintained in a 95% humidified and 5% CO₂ atmosphere at 37℃.

The cisplatin cytotoxicity of the 11 human ovarian cell lines (SK-OV-3, ES-2, PA-1, Caov-3, TOV-21G, TOV-112D, OV-90, OVCAR-3, MDAH 2774, A2780, and A2780cis) was determined using 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) assays. Briefly, 2×10⁴ cells were seeded onto 96-well plates and incubated at 37℃ overnight. The medium was exchanged with fresh medium containing different cisplatin concentrations (0–100 µM; Sigma-Aldrich; Merck KGaA). After incubation for 48 h, 20 µl of the 2.5 mg/ml MTT solution was added to each well and the plates were further incubated for 2 h at 37℃. One hundred microliters of dimethyl sulfoxide (DMSO; Sigma-Aldrich; Merck KGaA) was added to solubilize the MTT formazan product by oscillating for 10 min at 37℃. Absorbance at 540 nm was measured using a microplate reader (Molecular Devices LLC, Sunnyvale, CA, USA). Dose-response curves were plotted as the percentage of the control, which was obtained from the sample with no drug exposure. Half-maximal inhibitory concentration (IC₅₀) was calculated as the concentration of cisplatin that reduces cell growth by 50% under the experimental conditions using a nonlinear regression analysis with GraphPad Prism5 software (GraphPad Software, Inc., La Jolla, CA, USA). The 11 human ovarian cell lines were classified into three groups: sensitive, intermediate, and resistant.

Total RNA was extracted from eight human ovarian cancer cell lines (SK-OV-3, PA-1, Caov-3, TOV-21G, TOV-112D, OV-90, and OVCAR-3) using the RNeasy mini kit (Qiagen, Inc., Valencia, CA, USA) and amplified and labeled according to the Affymetrix GeneChip Whole Transcript Sense Target Labeling protocol (Thermo Fisher Scientific, Inc., Waltham, MA, USA). The resulting labeled cDNA was hybridized to Affymetrix Human Gene 1.0 ST Arrays (Thermo Fisher Scientific, Inc.). The scanned raw expression values were background-corrected, normalized, and summarized using the Robust Multiarray Averaging approach in the Bioconductor “affy” package (Bioconductor, http://www.bioconductor.org/). The resulting log2-transformed data were used for further analyses. To identify differentially expressed genes, we applied moderated t-statistics based on an empirical Bayesian approach [14]. Significantly up-regulated and down-regulated genes were defined as those with ≥1.5-fold difference in expression level between cisplatin-resistant and -sensitive groups after correction for multiple testing (Benjamini-Hochberg false-discovery rate [BH FDR]-adjusted p<0.01) [15].

Genomic DNA was extracted from eight human ovarian cell lines (SK-OV-3, PA-1, Caov-3, TOV-21G, TOV-112D, OV-90, and OVCAR-3) using the QIAmp mini kit (Qiagen, Inc.), according to the manufacturer's instructions. For genome-wide screening of DNA methylation, the Illumina HumanMethylation450 BeadChip (Illumina, Inc., San Diego, CA, USA) was used, which targets 450,000 specific CpG sites. DNA methylation values were described by β-values, which were determined by subtracting the background obtained from negative controls on the array and calculating the ratio of the methylated signal intensity to the sum of the methylated and unmethylated signals. β-values range from 0 (completely unmethylated) to 1 (fully methylated) on a continuous scale for each CpG site. To identify differentially methylated CpG sites, we calculated the difference in mean β-value (Δβ; mean β-value in resistant group – mean β-value in sensitive group). If the absolute difference in mean β-values (|Δβ|) was >0.3, the sites were defined as differentially methylated CpG sites.

One microgram of total RNA was converted to cDNA using Superscript II reverse transcriptase and oligo-(dT)_12-18 primers (both from Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. RT-qPCR was performed in a 20-µl reaction mixture containing 1 µl cDNA, 10 µl SYBR Premix EX Taq (Takara Bio, Inc., Otsu, Japan), 0.4 µl 50× Rox reference dye (Takara Bio, Inc.), and 200 nM primers for each gene. The primer sequences were: NAGA (forward), 5'-CCCAAGGGTGAACTACAGTCT-3'; NAGA (reverse), 5'-GCTCCACGAACCAATTCAGGAT-3'; GAPDH (forward), 5'-AATCCCATCACCATCTTCCA-3'; and GAPDH (reverse), 5'-TGGACTCCACGACGTACTCA-3'. The reactions were run on a 7500 Fast Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.) at 95℃ for 30 s, followed by 40 cycles of 95℃ for 3 s and 60℃ for 30 s, and a single dissociation cycle of 95℃ for 15 s, 60℃ for 60 s, and 95℃ for 15 s. All reactions were performed in triplicate, and the specificity of the reaction was determined using melting curve analysis at the dissociation stage. Comparative quantification of each target gene was performed based on the cycle threshold (C_T) normalized to GAPDH using the 2^−ΔΔCt method.

To demethylate methylated CpG sites, the eight human ovarian cell lines were treated with 10 µM 5-aza-dc (Sigma-Aldrich; Merck KGaA) for three days at 37℃. Each day, the medium was exchanged with fresh medium supplemented with 10 µM of 5-aza-dc.

To establish a transient expression system, cisplatin-sensitive TOV-112D and cisplatin-resistant SK-OV-3 cells were transfected with pCMV6-AC-NAGA (OriGene Technologies, Inc., Rockville, MD, USA) or pEGFP-N3 (Clontech, Mountain View, CA, USA) DNA constructs using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). Briefly, the cells were plated at a density of 6×10⁵ cells/well in 6-well plates containing antibiotic-free complete growth medium and allowed to grow overnight. Two micrograms of each plasmid DNA and 5 µl of Lipofectamine 2000 were diluted separately in Opti-MEM medium (Gibco; Thermo Fisher Scientific, Inc.) to a total volume of 250 µl. The diluted plasmid DNAs and Lipofectamine 2000 were mixed and incubated at room temperature for 20 min to generate the transfection mixtures. The transfection mixtures were added to each well of the 6-well plates, and incubated at 37℃ for 24 h in a 5% CO₂ incubator.

Pre-designed small interfering RNA (siRNA) against NAGA (siNAGA, CAT#ID L-011090-00-0005) and a non-targeting control (siNC, CAT#ID D-001206-13-05) were purchased from Thermo Fisher Scientific, Inc. To deplete endogenous NAGA expression, TOV-112D and SK-OV-3 cells were transfected with 100 nM siNAGA or siNC using DharmaFECT 1 transfection reagent (Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. Knockdown of NAGA was confirmed by RT-qPCR 24 h post-transfection.

The sensitivity of the transfected cells to cisplatin was determined using an MTT assay, described in the “Cisplatin cytotoxicity assay” section.

Proteins (30 µg) were resolved using denaturing 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were blocked in 5% skim milk in Tris-buffered saline with 0.1% Tween 20 (TBST) and subsequently incubated overnight at 4℃ with the following primary antibodies: rabbit anticaspase-3 polyclonal antibody (1:1,000, Cell signalling technology, Danvers, MA, USA and mouse anti-α-tublin monoclonal antibody (1:10,000, Sigma-Aldrich; Merck KGaA). After washing, the membranes were incubated with secondary antibodies conjugated to horseradish peroxidase for 1 h at room temperature. Chemiluminescence was detected using West Save Star (AbFrontier, Seoul, Korea) according to the manufacturer's protocol. Bands were visualized using an Image Quant LAS mini 4000 (General Electric life sciences, Chicago, IL, USA) and quantified using Image J software (http://rsb.info.nih.gov/ij/index.html).

Data are expressed as mean±standard deviation (SD) of ≥3 independent experiments. Statistical analyses were carried out using GraphPad Prism5 software (GraphPad Software, Inc.). An unpaired t-test was used to perform statistical analysis between cisplatin-sensitive and -resistant cell lines. p<0.05 was considered a statistically significant difference.

We determined the cytotoxicity of cisplatin in 11 human ovarian cancer cell lines (PA-1, TOV-21G, TOV-112D, Caov-3, A2780, A2780cis, MDAH2774, ES-2, OVCAR-3, OV-90, and SK-OV-3) using an MTT assay. Table 1 illustrates the cytotoxicity of cisplatin in the 11 human ovarian cancer cell lines in order of increasing cytotoxic response: PA-1, TOV-21G, TOV-112D, Caov-3, A2780, MDAH2774, A2780cis, ES-2, OVCAR-3, OV-90, and SK-OV-3. Based on the IC₅₀ values for cisplatin, we classified these cell lines into three groups: sensitive (PA-1, TOV-21G, TOV-112D, Caov-3, and A2780), intermediate (MDAH2774, A2780cis, and ES-2), and resistant (OVCAR-3, OV-90, and SK-OV-3).

We performed mRNA expression microarray analysis on five cisplatin-sensitive cell lines and three cisplatin-sensitive cell lines. We identified differentially expressed genes by performing moderated t-statistics based on an empirical Bayesian approach [14]. Differentially expressed genes were defined as genes whose levels were up-regulated or down-regulated by at least 1.5-fold between cisplatin-sensitive and -resistant groups after correction for multiple testing (BH FDR-adjusted p<0.01) [15]. NAGA mRNA expression was lower in cisplatin-resistant cell lines compared to levels in cisplatin-sensitive cell lines (Fig. 1A). The downregulation of NAGA mRNA expression was also confirmed by RT-qPCR, which showed significantly lower expression in all cisplatin-resistant cell lines (Fig. 1B).

To investigate the mechanism by which NAGA gene expression is regulated in the setting of cisplatin resistance, we performed global DNA methylation profiling of eight human ovarian cancer cell lines using the Illumina HumanMethylation450 BeadChip. To identify differentially methylated CpG sites between cisplatin-sensitive and -resistant cell lines, we applied two significant criteria, the t-test p-values and the difference in mean β-value (Δβ). CpG sites were considered differentially methylated if their t-test p-values were significant at a false-discovery rate threshold of 0.01 [15], and the absolute differences in mean β-values (|Δβ|) were >0.3 at a CpG site within the NAGA promoter region.

The Illumina HumanMethylation450 BeadChip contained 11 CpG sites within the promoter region of the NAGA gene, located between positions 42,071,248 and 42,070,341 of chromosome 22 (human GRCh38/hg38). The 11 CpG sites were at positions −656, −524, −498, −471, −299, −271, −260, −255, −157, +12, and +251 relative to the transcription start site (TSS), as shown in Fig. 2A. Among the 11 promoter CpGs, the CpG site located at +251 from the TSS was significantly hypermethylated in cisplatin-resistant cell lines compared with cisplatin-sensitive cell lines (Fig. 2B).

Next, we assessed whether NAGA expression is controlled by epigenetic regulatory mechanisms using a DNA methyltransferase inhibitor. Eight cisplatin-sensitive and -resistant cell lines were treated with 5-aza-2′-deoxycytidine, and NAGAmRNA expression was quantified by RT-qPCR. After treatment with 5-azadc, significantly increased NAGA expression was detected in the cisplatin-resistant cell lines, but not in the cisplatin-sensitive cell lines (Fig. 3A). NAGA expression level was restored by treatment of 5-aza-dc in all three cisplatin-resistant cell lines, but remained unaffected in untreated cells, indicating that NAGA expression is epigenetically silenced by DNA methylation in cisplatin-resistant cell lines (Fig. 3B).

We investigated whether NAGA overexpression could enhance sensitivity of ovarian cancer cells to cisplatin. Cisplatin-resistant SK-OV-3 cells and cisplatin-sensitive TOV-112D cells were transiently transfected with NAGA or EGFP expression plasmids. Following a 24-h transfection, the expression levels of NAGA were analyzed using RT-qPCR. As shown Figs. 4A and B, the expression of NAGA mRNA was increased ~258-fold in NAGA-transfected SK-OV-3 cells and ~317-fold in NAGA-transfected TOV-112D cells, compared with EGFP- mock transfected cells (Figs. 4A and B). The cytotoxicity to cisplatin was also determined in EGFP- or NAGA-transfected cells using an MTT assay. Overexpression of NAGA lead to a significant decrease (~13%) in the IC₅₀ of cisplatin in NAGA-transfected SK-OV-3 cells compared with EGFP-mock transfected cells (Fig. 4C). Similarly, the cisplatin IC₅₀ value dropped by ~49% in NAGA-transfected TOV-112D cells (Fig. 4D). These results indicate that overexpression of NAGA sensitizes human ovarian cancer cells to cisplatin.

We further explored how NAGA expression affects cisplatin resistance in ovarian cancer cells using siRNA-induced NAGA knockdown. After 24 h transfection, the knockdown of endogenous NAGA expression was confirmed by RT-qPCR. The expression of NAGA mRNA was reduced by ~89% in siNAGA-transfected SK-OV-3 cells and ~67% in siNAGA-transfected TOV-112D cells, compared with siNC- mock transfected cells (Figs. 5A and B). The cisplatin resistance of siRNA-transfected ovarian cancer cells was then evaluated using an MTT assay. Knockdown of NAGA expression induced a significant increase in the cisplatin IC₅₀ value by ~22% and 66% in SK-OV-3 cells and TOV-112D cells, respectively (Figs. 5C and D).

The cleavage status of caspase 3, an indicator of apoptosis was determined in the NAGA-overexpressed SK-OV-3 cells after treatment with cisplatin (30 µM for 10 h). Cisplatin-induced caspase 3 cleavage was increased by the overexpression of NAGA (Fig. 6), correlating with enhanced cytotoxicity observed in MTT assay (Fig. 4C). Decreased caspase 3 cleavage was also detected in the siNAGA-transfected SK-OV-3 cells compared with siNC-mock transfected cells (Fig. 6).

To determine whether the mitochondrial pathway was involved in cisplatin-induced apoptosis in SK-OV-3 cells, we examined changes in mitochondrial membrane potential. As shown in Fig. S1, the mitochondrial membrane potential did not significantly change in SK-OV-3 cells after 8 h of cisplatin treatment regardless of the level of NAGA expression (Fig. S1). These results indicate that cisplatin induces apoptotic cell death via mitochondrial-independent pathway in cisplatin-resistant SK-OV-3 cells.

Taken together, our data suggest that NAGA is involved in the regulation of cisplatin sensitivity by promoting apoptosis in human ovarian cancer cells.