Array-CGH analysis of 68 primary ovarian carcinomas shows DNA amplification (CGH > 0.7X) of ADRM1 in 15% of tumors. Among the amplified tumors was a stage 3, grade 4 Endometrioid Ovarian Carcinoma with a 262 kilobase amplification of 1.01× mapping from 60,071,478 kb within the TAF4 gene on chromosome 20q13 extending to 60,333,397 kb within the LAMA5 gene on 20q13. The amplicon includes ADRM1 and six additional genes whose expression levels do not correlate as significantly with amplification [1]. There are no gains either proximal or distal to this region on chromosome 20q in this tumor (Figure 1).

We speculated that the preferential amplification of ADRM1 gives overexpressing ovarian cancer cells a proliferative advantage. To determine whether or not this was the case, an ovarian cancer cell line that over-expresses ADRM1 (ES2A1) was created by stable transfection of an ADRM1 plasmid into ES2 cells. Cell counts in triplicate showed at least a 2.2-fold increase in cell growth on day 7 in ES2A1 cells compared to parental ES2 cells (Figure 2). In addition, naturally occurring ADRM1-amplified OAW42 cells treated with RNAi knock-down of ADRM1, resulted in a 1.4-fold growth inhibition compared to both parental ADRM1 untreated and non-targeting RNAi treated OAW42 cells by day 7 after transfection (Figure 3). Changes in cell cycle were not detected in the overexpressing and the knock-down cell line models.

ADRM1 overexpression correlates with recurrence and metastasis, and therefore, we hypothesized that expression levels effect the malignant tendency of the cells. In soft agar experiments, colony-forming efficiency in ADRM1-overexpressing ES2A1 cells was 1.9-fold higher than parental ES2 cells (Figure 4), and colonies are larger (Figure 5). Conversely, parental OAW42 cells showed a 2.3-fold increase and non-targeting RNAi-treated OAW42 cells showed a 1.8-fold increase in colony-formation efficiency, when compared to ADRM1 RNAi-treated OAW42 cells (Figure 6).

Additionally, a wound-healing assay was performed on ADRM1 knock-down OAW42 cells compared to both parental and non-targeting RNAi-treated OAW42 cells. A similar-sized wound was detected on the first day for each cell type. By day 3, the ADRM1-knock-down cells were only 50% healed, compared to 75% closing of the scratch wound in the parental and 88% in the non-targeting control cells (Figure 7).

Cell cycle and apoptosis experiments were performed to determine whether ADRM1 expression has an effect on cell cycle and whether blocking ADRM1 results in apoptosis. Cell cycle and apoptosis were examined by flow cytometry for ADRM1 over expressing clone ES2A1 and parental ES2, and no significant differences were detected (data not shown). Additionally, no significant differences in the cell cycle were detected between OAW42 cells with knock down of ADRM1 compared to parental untreated OAW42, nor dharmafect-treated OAW42 (nor OAW42 treated with either of two non-specific RNAi controls) (data not shown). However, there was a 20% increase in apoptosis, both early and late apoptosis (AN+/AN+PI+), and a 10% increase in cell death (PI+ cells) in ADRM1 RNAi-treated OAW42 cells compared to parental untreated OAW42, nor dharmafect-treated OAW42 and non-specific RNAi controls (Figure 8).

ADRM1 is a proteasomal ubiquitin receptor and therefore we predicted knock-down of ADRM1 might affect the regulation of other proteins. Therefore, we performed Difference Gel Electrophoresis (DiGE) analysis on OAW42 cells with RNAi knock-down of ADRM1 and compared protein levels to parental untreated and non-targeting RNAi-treated OAW42 cells. We identified 15 proteins that were down-regulated and 7 proteins that were upregulated when ADRM1 is knocked-down (Table 1). These proteins were excised from the gel and identified with mass spectrometry. Three of the proteins were confirmed to be differentially expressed by Western Blot Analysis on the original lysates, including MNAT1, HAX1, and ST1A3 (Figure 9).

Recently, we showed that among 225 ovarian samples, ADRM1 is a likely 20q13-amplification target in ovarian cancer. It was amplified and overexpressed in 23% of tumors, was significantly upregulated with respect to stage, recurrence and metastasis, and its overexpression correlates significantly with shorter time to recurrence and overall survival [1]. Consistent with these findings, using array-CGH and RNA expression analysis of ovarian cancer cell lines, we showed the amplification and overexpression of ADRM1 in 18% (7/39) of ovarian cancer cell lines [12]. Herein, CGH analysis of ADRM1 in a separate cohort of 68 primary ovarian carcinomas, including one with high level amplification of a 262 kilobase region containing ADRM1, provide additional evidence that ADRM1 is the driver of a 20q13 amplicon in ovarian cancer.

The role of this gene was explored using ADRM1-upregulated and ADRM1 down-regulated in vitro models. Evidence suggests that ADRM1 has traits of an oncogene as amplification leads to increased cell proliferation and increased colony-formation efficiency. Conversely, down-regulation of ADRM1 at the RNA and protein levels results in growth inhibition, decreased colony-formation efficiency, slower wound-healing, decreased cell viability, and significantly increased apoptosis. Knock-down of ADRM1 resulted in the dysregulation of multiple proteins already known to play a role in cancer. In particular, multiple HSP90 fragments were identified. A recent study showed HSP90 inhibition leads to simultaneous inactivation of multi-RTKs including EGFR, ERBB2, MET, and AXL and suppresses the downstream survival/proliferation signaling in multiple ovarian cancer cell lines [13] making it a promising target for ovarian cancer therapy. Drugs targeting heat shock proteins are currently of intense focus, and are involved in several clinical trials in ovarian and other cancers [14].

SULT1A1 (ST1A3) is up regulated with the knock-down of ADRM1 and encodes a protein that catalyzes the sulfate conjugation of many hormones. Variants in ST1A3 are linked to increased hormone-dependant ovarian cancer risk [15,16].

The gene HAX1 was shown in this study to be down regulated with ADRM1 and is known to associate with hematopoeitc cell-specific Lyn substrate 1, a substrate of the Src family of tyrosine kinases [17]. HAX1 promotes cell survival via inhibition of CASP9, responsible for apoptosis execution [18]. Therefore its down regulation may explain the significant increase in apoptosis in ADRM1 RNAi-treated ovarian cancer in this study. It also potentiates GNA13-mediated cell migration and invasion [19,20], and thus its down regulation may contribute to the decrease in wound healing in ADRM1 knock-down cells shown in this study. Thus one can hypothesize that the correlation between ADRM1 over expression and recurrence and metastasis shown previously in ovarian cancer may be through regulation of HAX1.

Finally, the most significantly dysregulated gene, MNAT1 (MAT1), stabilizes the cyclin H-CDK7 complex to form a functional CDK-activating kinase (CAK) enzymatic complex, which in turn activates the cyclin-associated kinases CDK1, CDK2, CDK4 and CDK6 by threonine phosphorylation [21]. Interestingly, in a mouse model for pancreatic cancer lymphatic metastasis, ADRM1 was one of 30 genes found to be upregulated in the metastatic cell line compared to the parental cell line [22]. In another study, direct RNAi silencing of MNAT1 suppressed the growth of pancreatic cancer cells in vitro, and significantly achieved an anti-tumor effect on the subcutaneously transplanted pancreatic tumor in vivo[23]. Thus, down regulation of MNAT1 in ADRM1-silenced ovarian cancer may explain the similar anti-tumor effects shown in this study. Additionally, MNAT1 has been shown to interact with Cyclin H, Estrogen receptor alpha, MTA1, and cyclin-dependent kinase 7, all linked previously to ovarian cancer [24–26].

Sixty-eight primary ovarian carcinomas were snap frozen in liquid nitrogen and prepared for DNA and RNA processing as described below. The samples consisted of 38 ovarian endometrioid carcinomas (19 stage I and II, 19 stage III and IV), 16 ovarian clear cell carcinomas (5 stage I and II, 11 stage III and IV), 11 mucinous ovarian tumors (10 stage I, 1 stage III), and 3 mixed ovarian epithelial carcinomas (1 stage I, 2 stage III and IV).

The cell line OAW42 was obtained from the European Collection of Cell Cultures (Salisbury, UK). The cell line ES2 was obtained from the American Type Culture Collection (ATCC, Rockville, MD, USA). OAW42 and cells was cultured in DMEM (ATCC) plus 10% FBS. ES2 was cultured in McCoy’S 5A medium (ATCC) plus 10% FBS.

Genomic DNA was extracted from frozen cell pellets using the DNeasy Blood and Tissue Kit (Qiagen) with the following modifications. Seventy percent ethanol was substituted for Buffer AW2 in the final wash, and DNA was eluted in 50 mL of sterile water (Invitrogen, Carlsbad, CA, USA). The concentration and quality of the DNA were measured with the NanoDrop Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) and by electrophoresis in 1% agarose.

Labeling and hybridization of Agilent 105K oligonucleotide CGH arrays was performed according to the manufacturer’s protocol for Human Genome CGH 105A Oligo Microarray Kit (version 5.0; Agilent Technologies, Santa Clara, CA, USA, 2008). Labeled tumor and reference DNAs were combined, annealed with COT-1 DNA (Invitrogen, Carlsbad, CA, USA) and 10× Blocking Agent (Agilent, Santa Clara, CA, USA) for 30 min at 37 °C after boiling, then hybridized to Agilent 105A arrays for 40 h at 65 °C according to the manufacturer’s instructions. After hybridization, arrays were washed according to Procedure B (which includes an ozone blocking wash), and scanned using an Agilent Scanner (G2565BA). Files were extracted using Agilent Feature Extraction software (version 9.5; Agilent, Santa Clara, CA, USA, 2007) with the default CGH protocol. Extracted arrays with a DRL Spread< 0.3 were included in the analysis.

CGH Analytics software (version 4.0; Agilent Technologies, Santa Clara, CA, USA, 2008) was used for copy number analysis, employing the ADM1 algorithm (Threshold 5), with Fuzzy Zero and Centralization corrections to minimize background noise. All map positions were based on the March 2006 NCBI 36/hg18 genome assembly. A minimum of 3 consecutive probes was required to define a region as amplified or deleted. The data was also filtered by requiring a minimum absolute average log2 ratio of 0.4. All data was inspected visually using the interactive view. Log2 ratios >0.4 (1.3-fold) were considered gain and log2 ratios >1 (2-fold) were considered amplified.

ES2 and ES2A1 cells were counted and seeded in triplicate on 12-well plates and were harvested by trypsinization on days 1, 3, and 7 and counted using a particle counter (Z1; Beckman Coulter, Inc., Brea, CA, USA). OAW42 cells were counted and seeded in triplicate on 6-well plates and treated with dharmafect only, dharmafect and ADRM1 siRNA, or dharmafect and si-negative control as explained in the RNA interference section. Cells were harvested by trypsinization on days 1, 2, 3, and 6 after transfection and counted using a particle counter (Z1; Beckman Coulter, Inc., Brea, CA, USA). Standard deviations were calculated from triplicate experiments and shown in bar graphs.

The effects of silencing ADRM1 on apoptosis and cell cycle were evaluated using Annexin V-FITC and Propidium Iodide (PI) staining and DAPI DNA staining respectively. Cells were plated evenly and allowed to grow to log phase before being transfected with siRNA to knockdown ADRM1 or control. Also, a vehicle for transfection Dharmafect was used as a control. For apoptosis analysis supernatant was collected, cells were washed with PBS, and trypsin was applied to release cells, which were then centrifuged at 2000 rpm for 5 min. Supernatant was aspirated and cells were then suspended in 200 and allowed to grow to log ained with 10 μL of Annexin V-FITC and 5 min. Supernatant was aspirated and cell (Medical & Biological Laboratories, Co., Woburn, MA, USA). The effects of siRNA ADRM1 on the cell cycle were assessed using Nim-DAPI staining (NPE Systems, Pembroke Pines, FL, USA). Flow cytometry was performed using a CellQuanta (Beckman Coulter, Brea, CA, USA) flow machine, and the results were analyzed using Flow Jo software (version 7.6.5; Tree Star Inc., Ashland, OR, USA, 2011).

ES2, ES2A1, OAW42, and siRNA-treated (ADRM1 siRNA and non-targeting control siRNA) OAW42 cells were counted and seeded in triplicate to 10,000 cells per well in a 24-well culture dish containing 0.4% low-melting agarose over a 0.6% agarose layer, both in culture medium, and incubated for 3–5 weeks at 37 °C. Colonies were stained with 10% Neutral Red (Fisher Scientific, Waltham, MA, USA) and counted. Standard deviations were calculated from triplicate experiments and shown in bar graphs. Experiment was repeated twice to confirm findings. Knock-down of ADRM1 in cells from siRNA experiments was confirmed by Western Blot with protein lysates isolated seven days after seeding.

Transfected and untransfected cells were isolated two days after transfection. Cells were washed in PBS and lysed at 4 °C in lysis buffer. Insoluble material was cleared by centrifugation at 10,000× g for 10 min. Protein was quantitated using BCA (Pierce Biochemicals, Rockford, IL, USA), resolved by SDS-PAGE, and transferred to nitrocellulose membranes (Invitrogen, Carlsbad, CA, USA). ADRM1, MNAT1, and ST1A3 were detected, respectively with antibodies to ADRM1 monoclonal 3C6, MNAT1 anti-MNAT antibody produced in rabbit, ST1A3 anti-SULT1A1 monoclonal 1F8 (Sigma-Aldrich, St. Louis, MO, USA). Monoclonal Anti-HAX1 antibody produced in mouse was purchased from BD Biosciences, San Jose, CA, USA. Detection was performed using chemifluorescent reagent (SuperSignal West Pico Chemiluminescent Substrate, Thermo Scientific, Rockford, IL, USA). Antialpha-tubulin mouse antibody (Calbiochem, San Diego, CA, USA) 0.1 μg/mL was used as a loading control.

Four interfering RNAs specific for ADRM1 were purchased from Dharmacon (ThermoFisher Scientific, Lafayette, CO, USA) as a Smartpool in addition to two non-targeting siRNA oligos to use as a controls. Transfection was performed using 0.4% Dharmafect transfection reagent and 10 nM siRNA according to the manufacturer (Dharmacon, Lafayette, CO, USA). OAW42 cells were seeded and transfected at 60% confluency in antibiotic-free media (DMEM (Cellgro, Lawrence KS, USA) supplemented with 1% l-glutamine, 0.91 μg/mL bovine insulin, 4.5 g/1 L glucose, and 3.7 g/L sodium bicarbonate). After 24 h at 37 °C, growth media with 10% fetal bovine serum and antibiotic was added. Knock-down of ADRM1 was confirmed by Western Blot at day 2, 3, 4, and 7 after transfection.

The ADRM1 Ultimate Human ORF Clone from IOH13967 was purchased from Invitrogen (Carlsbad, CA, USA) and cloning and transfection were performed following Invitrogen’s protocol for transfection of a mammalian expression clone. Kanamycin resistant IOH13967 clones were isolated. The plasmid DNA was subcloned into amp-resistant pcDNA 6.2/V5-DEST Mammalian expression vector. True expression clone was confirmed by testing ampicillin resistance and chloramphenicol sensitivity. Pure DNA was isolated with the Purelink HQ kit and the clone was sequenced to confirm it was in frame. Ovarian cell line ES2 was tested for Blasticidin sensitivity and a concentration of 0.015 mg/mL Blasticidin kills all ES2 parental cells. ES2 was transfected with the ADRM1 expression vector with lipofectamine. Blasticidin-resistant clones grown in 0.015 mg/mL Blasticidin were tested for increased ADRM1 expression by Western blot. The highest ADRM1 expressing clone ES2A1 was used in experiments in this study.

For the scratch assay, OAW42 cells treated with ADRM1-specific RNAi, non-targeting RNAi, and Dharmafect only (parental), were grown in complete growth medium until 90%–100% confluency was reached. A 3 mm cross-wound was introduced across the diameter of each plate. Cell migration was observed by microscopy and photographed at 24 h, 48 h, 72 h and 96 h later.

OAW42 ovarian cell lines were independently grown in eighteen 100 mm regular Petri dishes in complete medium for 24 h and then each of six OAW42 ovarian cell lines were treated either with 10 nM ADRM1-specific RNAi or non-targeting RNAi (scrambled RNAi) and the control cells with only 0.4% of Dharmafect transfection reagent (parental) in antibiotic-free medium. After 24 h, cells were incubated in complete medium for another 24 h. Then, cells were washed two times with low-serum medium at 4 °C, followed by 2 times with PBS at 4 °C, and directly lysed in 0.3 mL of ice cold labeling buffer (7 M urea, 2 M thiourea, 4% CHAPS and 20 mM Tris-HCl, pH 8.8) and then briefly sonicated in Bioruptor (Diagenade) on ice cold water bath for 30 s on and 1 min off for a total six times at high settings. Cell lysates were centrifuged at 13,500 rpm for 15 min at 4 °C and extracted proteins were stored at −80 °C.

DIGE labeling, analysis and mass spectrometry were performed as described previously [27]. Briefly, protein extracts used for the dige experiments were cleaned using the methanol/chloroform method and protein pellets were resuspended in the labeling buffer and centrifuged. For the dige analysis, protein samples were labeled with the N-hydroxysuccinimidyl ester derivates of Cy2, 3 and 5 Dyes according to the manufactures protocol (GE Healthcare, Piscataway, NJ, USA) for 30 min on ice water in the dark and then excess Cy Dyes were quenched by adding 10 mM Lysine solution and incubated 10 min. Briefly, 50 micrograms of protein from each experimental group repeated 6 times (ADRM1-specific RNAi, non-targeting RNAi treated, and control cells) were labeled with the 400 pmol of Cy3 and Cy5 Dyes, respectively, to avoid any Cy Dye labeling biases. Internal standard sample was prepared by pooling equal (25 micrograms) amount of proteins from all eighteen samples and labeled with the 400 pmol of Cy2. After the labeling, 50 micrograms of Cy2, Cy3 and Cy5 labeled samples were combined with the 300 micrograms of unlabeled protein mix from the all samples and the rehydration solution was added (7 M urea, 2 M thiourea, 4% CHAPS, 1% DTT. 0.5% pH 4–7 IPG buffer, 5% glycerol, 10% isopropanol). Samples were incubated for another 20 min, briefly centrifuged for 5 min at 11,000 rpm and then applied to 24 cm pH 4–7 IPG strips (GE Healthcare), passively rehydrated overnight at room temperature. Proteins were focused according with the following steps at 200 V (hold) for 2 h, 500 V (hold) for 2 h, 1500 V (gradient) for 2 h, 8000 V (gradient) for 4 h 30 min and 8000 V (hold) for 6 h 30 min for the total of 77,000 Vh. After the IEF, IPG strips were equilibrated in a 10 mL of reducing solution (50 mM Tris-HCl, pH 8.8, 6 M urea, 30% glycerol, 2% SDS, 1% DTT, 0.002% bromophenol blue) for 15 min and then a second equilibration step was followed with 4.5% (w/v) iodoacetamide in the same solution without DTT. After reduction and alkylation of proteins, IPG strips were placed on the top of the 12.5% SDS polyacrylamide gels and sealed with the 1% agarose solution. A second dimension SDS gel electrophoresis was run in Ettan DALTtwelve system (GE healthcare) at 50 V for 3 h at 24 °C and then continued at 1 W per gel at 24 °C for overnight until bromphenol blue runs out of the gels. The next day, gels were scanned using the Typhoon Trio Variable Mode Imager (GE Healthcare) at 100 micron resolution using 488 nm laser/520BP40 filter for Cy2, 532 nm laser/580BP30 nm filter for Cy3, 633 nm laser/670BP30 for filter Cy5. Gel images were cropped using Image Quant Software (GE Healthcare, Piscataway, NJ, USA). The Decyder 2D Differential Analysis software (version 6.5; GE Healthcare: Piscataway, NJ, USA) was used for the identification of statistically significantly differentially expressed proteins. After the analysis, proteins of interests were selected based on the fold changes and student’s t test. Gels were fixed and then stained by Sypro-Ruby. Sypro-Ruby stained images were re-matched to the dige images and protein spots were picked from three gels using the Ettan Spot Picker (GE Healthcare) and digested with modified trypsin (Trypsin Gold, Mass Spectrometry grade, Promega, (Madison, WI, USA) using the ProGest protein Digestion Robotic Station system (Genomic Solutions, Inc, (Marlborough, MA, USA). MALDI-TOF/TOF mass spectrometry analysis was performed on an Ultraflex TOF/TOF instrument (Bruker Daltonics, Bremen, Germany) [28].