Early and consistent overexpression of ADRM1 in ovarian high-grade serous carcinoma

Background Ovarian carcinoma is highly dependent on the ubiquitin proteasome system (UPS), but its clinical response to treatment with the proteasome inhibitor bortezomib has been disappointing. This has driven exploration of alternate approaches to target the UPS in ovarian cancer. Recently, proteasome inhibitors targeting the 19S regulatory particle-associated RPN13 protein have been described, such as RA190. RPN13, which is encoded by ADRM1, facilitates the recognition by the proteasome of its polyubiquinated substrates. Inhibition of RPN13 produces a rapid, toxic accumulation of polyubiquitinated proteins in ovarian and other cancer cells, triggering apoptosis. Here, we sought to determine if RPN13 is available as a target in precursors of ovarian/fallopian tube cancer as well as all advanced cases, and the impact of increased ADRM1 gene copy number on sensitivity of ovarian cancer to RA190. Methods ADRM1 mRNA was quantified by RNAscope in situ hybridization and RPN13 protein detected by immunohistochemistry in high grade serous carcinoma (HGSC) of the ovary and serous tubal intraepithelial carcinoma (STIC). Amplification of ADRM1 and sensitivity to RA190 were determined in ovarian cancer cell lines. Results Here, we demonstrate that expression of ADRM1mRNA is significantly elevated in STIC and HGSC as compared to normal fallopian tube epithelium. ADRM1 mRNA and RPN13 were ubiquitously and robustly expressed in ovarian carcinoma tissue and cell lines. No correlation was found between ADRM1 amplification and sensitivity of ovarian cancer cell lines to RA190, but all were susceptible. Conclusions RPN13 can potentially be targeted by RA190 in both in situ and metastatic ovarian carcinoma. Ovarian cancer cell lines are sensitive to RA190 regardless of whether the ADRM1 gene is amplified. Electronic supplementary material The online version of this article (doi:10.1186/s13048-017-0347-y) contains supplementary material, which is available to authorized users.


Background
Missense mutation of TP53 is the dominant driver in ovarian/fallopian tube cancer [1][2][3][4]. Indeed TP53 is mutated in nearly all high grade serous carcinomas (HGSC), the histotype responsible for most deaths from ovarian cancer [5,6], and is also mutated in its precursor, serous tubal intraepithelial carcinoma (STIC) [7][8][9], suggesting this is a key and early event in carcinogenesis [10]. In addition to inactivating wild type TP53 function, these mutations frequently confer gain-of -function properties including redirection of Nrf2 to upregulate the proteasome [11]. The aberrant metabolism of ovarian cancer cells produces an excess of misfolded proteins that are polyubiquitinated for targeted degradation via the proteasome, and this is associated with proteasome upregulation [12]. Consequently, ovarian cancer cells are both highly dependent on proteasome function and especially sensitive to treatment with a proteasome inhibitor [12][13][14]. The FDA-approved drug bortezomib, approved for the treatment of multiple myeloma and mantle cell lymphoma, inhibits 20S core particle (CP) proteolytic function producing a rapid and toxic buildup of denatured protein aggregates [15]. The cells attempt to ameliorate this stress via the unfolded protein response (UPR) which promotes re-folding via chaperone upregulation as well as the rapid sequestration and degradation of the misfolded, ubiqutinated proteins. Inability to relieve proteotoxic stress and achieve homeostasis induces TP53independent cell death [16]. Indeed, treatment with bortezomib slowed the growth of ES-2 ovarian carcinoma xenograft, although alone it did not cure the mice [12].
Despite this promise, early studies with bortezomib in patients with ovarian cancer or other solid tumors have demonstrated minimal clinical benefit. Initial phase I studies suggested a dose of 1.3 mg/m 2 in combination with cisplatin/carboplatin, paclitaxel or pegylated liposomal doxorubicin had manageable neurotoxicity [17][18][19][20][21][22], although in combination with oxaliplatin a maximum dose of 1.0 mg/m 2 bortezomib was suggested because of neuropathy [23]. However, a phase II study of bortezomib alone a dose of 1.3 mg/m 2 in platinum-sensitive epithelial or primary peritoneal cancer showed minimal activity as a single agent [17]. Likewise, a phase II study of bortezomib and pegylated liposomal doxorubicin showed the combination was well tolerated, but the antitumor activity was considered insufficient to warrant further investigation [21]. Thus, despite bortezomib's efficacy against multiple myeloma, its toxicity, and limited activity in solid tumors has driven development of inhibitors of the ubiquitin-proteasome system (UPS) with alternative mechanisms to treat ovarian cancer.
RPN13, which is encoded by the gene ADRM1, is a ubiquitin receptor tethered to the 19S regulatory particle (RP) of the proteasome via interaction of its amino terminal Pru domain to RPN2 [24]. The Pru domain of RPN13 and RPN10 cooperate to capture K48-linked polyubiquitin chains bound to proteasome substrates. The carboxy terminal domain of RPN13 recruits and promotes the deubiquitinase activity of UCHL5/UCH37 which removes distal ubiquitin from the polyubiquitin chains linked to these substrates [25][26][27]. Deubiquitination of the substrate is followed by unfolding and transfer to the 20S CP for its proteasomal degradation.
Anchoori et al. recently developed a promising anticancer drug, RA190, which covalently binds to RPN13 C88 in its Pru domain [36,41]. RA190 has promising activity against ovarian cancer, cervical cancer and multiple myeloma (MM) in mouse models by triggering an unresolved unfolded protein response (UPR) and then TP53-independent apoptosis [41]. RPN13 expression is dramatically elevated in clinical samples of MM compared to normal plasma cells [42]. Furthermore, RPN13 knockdown by ADRM1 siRNA, or RA190 treatment causes loss of MM cell viability, including ex vivo patients' samples and bortezomib-resistant lines, via caspasedependent and UPR-related apoptosis [42]. Using CRISPR/Cas9 knockout of ADRM1, we show RA190 activity is dependent upon RPN13. Finally, RA190 treatment was well tolerated, inhibited MM xenograft tumor growth and extended survival [42]. Recently, another group developed KDT11, a reversible ligand of RPN13, which they found to be selectively toxic for MM cells. KDT11 provides further validation of RPN13 as an anticancer drug target, but does not have properties conducive to in vivo use [43]. Lu et al. identified a 38 aa peptide of RPN2 that binds to RPN13 Pru domain with 12 nM affinity. Overexpression of this peptide in 293T cells prevents RPN13 binding to the proteasome resulting in an increase in the cellular load of poly-ubiquitinated proteins [44].
These observations support RPN13 as a treatment target for ovarian cancer, but it is not clear whether it is expressed in all cases and if ADRM1 amplification impacts sensitivity to RA190. Recent findings suggest that the precursor lesion for HGSCs, STIC, arise at the fimbriated end of the fallopian tubes and shed cancer cells that implant on the surface of the ovary and peritoneum [45][46][47][48][49]. STIC typically exhibit high proportions of proliferating cells and missense TP53 mutations [48,50], but it is not known if they also express RPN13 (and to what level) and thus whether STIC are also potential treatable by RA190.

Results
Validation of a highly sensitive and specific Chromogenic In Situ Hybridization (CISH) assay for ADRM1 transcript levels To develop a sensitive ADRM1 mRNA CISH assay, we used a custom hybridization probe for the RNAscope® 2.0 assay from Advanced Cell Diagnostics (ACD) that is designed for detection at the single transcript level. To establish its ability to assess mRNA levels, ADRM1 levels were measured in 4 HGSC patients with matching frozen tissue samples and formalin-fixed, paraffin embedded (FFPE) tissue blocks by mRNA CISH or qRT-PCR, respectively (Additional file 1: Figure S1A, S1B). Levels of ADRM1 mRNA measured with qRT-PCR correlated with RNAscope® DAB dots (numbers of the dot-like signals) in RNAscope® demonstrating semi-quantitative nature of RNAscope® 2.0 assay.

Elevated ADRM1 mRNA expression in matched STICs and HGSCs
Formalin-fixed, paraffin embedded blocks of 11 ovarian cancer cases were selected by three gynecologic pathologists who independently assessed the presence of STIC and HGSC carcinoma using H&E, ki-67, and p53 staining to probe morphology, proliferation and, − surrogate of TP53 muational status, respectively. Of 11 patient samples, 7 FFPE blocks contained matched normal fallopian tube, STIC, and carcinoma ( Fig. 1, Rows 1-3). Tumor tissue samples from an additional 13 HGSC patients were arrayed on a TMA containing 5 cores per patient (Table 1, Fig. 1, Row 4).
The presence of detectable RNA in the tissue samples was verified by using positive control peptidyl prolyl isomerase B (PPIB) probe (Additional file 2: Figure S2). Next, we utilized the ADRM1 mRNA CISH assay to probe adjacent slides from the samples of 7 ovarian cancer patients with matched STICs, normal fallopian tube epithelium, and the additional samples from 13 HGSC patients. ADRM1 mRNA was detected in both normal fallopian tube epithelium as well as STIC (Fig. 2). Expression levels were assessed semi-quantitatively by gynecologic pathologist (DX) as higher in STIC tissue as compared to normal fallopian tube epithelium in all samples (n = 7). ADRM1 mRNA levels in matched invasive carcinoma on same slides were considered to be a similar to those in adjacent STIC. This implies that the elevation of ADRM1 mRNA levels is an early event that does not obviously increase further as disease progresses.
To confirm the semi-qualitative analysis of the gynecologic pathologist, automated quantification was used on the 7 STIC confirmed patients by ACD SpotStudio Software, which allows for direct quantification of target RNA molecules in single cells. After defining regions of interest (ROIs) including normal fallopian tube epithelium, STIC, and invasive high grade serous carcinoma, SpotStudio defines average cell size and quantifies DAB-stained (brown) dots per cell (Additional file 3: Figure S3). Overall, levels of ADRM1 mRNA were not significantly different between STIC and HGSC (P = 0.4) but each was significantly elevated as compared to matched normal epithelium (P = 0.009 and P = 0.003 respectively, paired T-test). STIC (~50%) and matched HGSC (~33%) showed a higher proportion of cells expressing high levels (>6) of RPN13 mRNA dots per cell (~23%) as compared to normal epithelium (Fig. 3, Additional file 4: Figure S4).
To determine whether RPN13 protein is expressed in these tissues, IHC was performed on adjacent serial sections with mouse monoclonal anti-RPN13 antibody (Fig. 2). As expected for a proteasome protein, RPN13 IHC showed diffuse nuclear and cytoplasmic staining. The specificity of the immunohistochemistry assay was confirmed by comparing the staining of HCT116 wild type cells and an ADRM1 CRISPR knockout HCT116 clone (Additional file 5: Figure S5). RPN13 expression was detected in STIC and HGSC as well as normal fallopian tube epithelium although there was no clear indication of change in protein level based on this semi-quantitative technique (Fig. 2).

Consistent expression of RPN13 in HGSC
All cases tested (n = 13), regardless of HGSC stage, maintained expression of both ADRM1 mRNA (determined by in situ hybridization) and RPN13 protein as visualized by immunohistochemistry (Fig. 4). Expression of RPN13 protein was further confirmed by Western blotting of same HGSC samples (n = 4) used for RNA-CISH verification (Additional file 1: Figure S1C). When grouped by similar PPIB mRNA levels, tissue exhibiting variable ADRM1 mRNA levels exhibited similar amounts of RPN13 (Additional file 6: Figure S6, Additional file 7: Figure S7).
Amplification of ADRM1 does not correlate with expression of RPN13 or sensitivity to RA190 The Cancer Genome Atlas (TCGA) Research Network collected tumor specimens and analyzed key genomic changes by mRNA-seq, whole exome DNA-seq, miRNAseq, methylation, copy number, in over 11,000 patient samples encompassing 33 different types of cancer [51,52]. From analysis of TCGA data it is apparent that ovarian cancer has a higher frequency of ADRM1 amplification (~8-15%) as compared to other cancers (Fig. 5a) and amplification is moderately correlated with elevated mRNA expression level (Pearson coefficient = 0.538, Fig. 5b). ADRM1 amplification and mRNA overexpression could lead to differences in sensitivity to RPN13 inhibitors, such as RA190. To test this hypothesis, we established the copy number variation status for a panel of n = 7 ovarian cancer cell lines in comparison to the SV40 large T-antigen immortalized normal fallopian tube cell line FT2021 using a TaqMan® Copy Number assay (Fig.  6a, left). Previously TCGA reported cell lines COV318 and OAW42 were amplified. Using the TaqMan® Copy Number assay and determination by the CopyCaller™ Software, this was confirmed. Additional cell lines such as OV2008 and PE014 were also identified by CopyCaller™ Software as amplified. A2780, Kuramochi, and the SV40 large T antigen immortalized normal fallopian tube cell line FT2021 had normal copy number of ADRM1, by this approach. However, OVCAR3 was also amplified by Copy-Caller™ Software, differing from TGCA results. Among these cell lines, ADRM1 mRNA levels (Fig. 6a, right) are not correlated to ADRM1 amplification, although this may reflect the relatively small sample size. Further, the panel of cell lines showed largely similar RPN13 protein levels by Western blot (Fig. 6b) regardless of amplification status, suggesting possible post-transcriptional regulation of RPN13 expression.
To address the hypothesis that variability in ADRM1 amplification and mRNA expression could potentially alter sensitivity to RA190, XTT cell viability assays were run on HGSC cell lines A2780, Kuramochi, PE014, OVCAR3, OAW42, and COV318, with differing ADRM1 copy number. All cell lines were sensitive to RA190, with IC50s ranging from 1.2-2.3 μM. There was no apparent correlation between ADRM1 copy number variation and sensitivity to RA190 in this limited panel of cell lines (Fig. 6c).

Discussion
The overabundance of misfolded proteins and consequent increased reliance on the ubiquitin proteasome system to relieve proteotoxic stress make ovarian carcinoma an attractive candidate for targeted proteasome inhibition. Recent inhibitors targeting the 19S ubiquitin receptor RPN13 have demonstrated therapeutic potential against cultured cancer cell lines as well as in vivo models and raised questions about RPN13 expression patterns and its functional role in cancer biology [41]. In this study, we have validated a chromogenic in situ hybridization assay to visualize ADRM1mRNA levels in normal fallopian tube epithelium, STICs, and HGSCs to determine their relationship to ovarian carcinogenesis. We show that overexpression of ADRM1 mRNA is an early event found in all probed STIC, the precursor lesions of HGSC. Further, we confirmed expression of RPN13 protein in normal fallopian tube epithelium, STIC and the ovarian tumors. The specificity of ADRM1 overexpression to developing tumors implies it may be reflect an early response to accommodate an overabundance of misfolded proteins found even in the cancer precursor. Additionally, the increased ADRM1 mRNA expression with respect to disease stage and recurrence further argues that RPN13 may contribute to disease progression. However, we did not see differences in expression between intraepithelial (STIC) and invasive (HGSC) carcinomas. [30]. Recent studies have shown that of the 214 known genes mapping to chromosome 20q13, overexpression of ADRM1 mRNA is the most highly correlated with this chromosomal amplification [30]. Further, in vitro overexpression of ADRM1 has led to increase in proliferation and migration of transfected ovarian cancer cell line, ES2, and has been correlated to shortened survival and more rapid relapse [53]. However, the specific pathways affected by ADRM1 overexpression have yet to be determined.
ADRM1 mRNA overexpression is associated with chromosome 20q13 amplification [30], but was not obviously associated with concurrent protein level changes in ovarian cancer cell lines. RPN13 protein was expressed in all normal fallopian tube epithelium, STICs, and HGSC samples tested. Patient tumor tissue samples as well as ovarian cancer cell lines show relatively consistent protein levels regardless of ADRM1 mRNA expression levels suggesting that RPN13 expression is heavily post-transcriptionally regulated.
Proteasome components exhibit coordinate regulation mediated by Nrf1 signaling in response to partial proteasome inhibition [54,55], and we have described their upregulation in ovarian cancer in association with ubiquitin-proteasome system (UPS) stress and increased metabolic burden [56]. Taken together, this implies the upregulation of ADRM1 mRNA in STIC and HGSC may reflect their increased UPS and metabolic burden when compared to normal fallopian tube epithelium.
The consistent RPN13 protein expression in HGSC and the lack of impact of ADRM1 amplification upon sensitivity to RA190 suggest that RPN13 may still be used as a therapeutic drug target without major dose adjustments for ADRM1 amplification status. Indeed, all of the ovarian cancer cell lines tested were sensitive to RA190, although the panel warrants expansion. Another factor may be a second target; RA190 appears to adduct to both RPN13 and UCH37 to inhibit proteasome function synergistically [57]. RA190 or similar compounds have promise for treatment of patients whose disease becomes resistant to proteasome inhibition by other proteasome licensed inhibitors, such as bortezomib. Initial studies have shown that RA190 is active against mouse models of ovarian cancer, ES2 and ID8, although they are also sensitive to bortezomib [41].

Conclusions
We show that ADRM1 mRNA overexpression is an early event in HGSC carcinogenesis. This is associated with TP53 mutation and increased burden of misfolded proteins in carcinomas that likely renders the cancer cells particularly sensitive to RPN13 inhibitors. However, regardless of ADRM1 amplification status the RPN13 protein levels are relatively consistent across ovarian cancer cell lines. Since ADRM1 amplification was also not correlated with sensitivity to RA190, this suggests

Tissue specimens and cell culture
Studies on archived formalin-fixed, paraffin-embedded ovarian carcinoma tissue specimens selected from cases in the Department of Pathology at The Johns Hopkins Hospital were approved by the Johns Hopkins University Institutional Review Board. A total of 20 specimens were selected with pathological diagnoses of high grade serous carcinoma, of which 7 also exhibited STIC. One pathologist (DX) reviewed all of the slides for all cases independently to assess the staining, and STIC diagnoses were confirmed by 3 pathologists (JDS, AVY, DX). OAW42, OV2008, COV318 cells were maintained in Dulbecco's modified Eagle medium (DMEM) with 10% fetal bovine serum, 100 U penicillin and streptomycin, 1 U non-essential amino acids, 1 mM sodium pyruvate. A2780, Kuramochi, OVCAR3 cell lines were maintained in RPMI medium supplemented with 10% fetal bovine serum, 100 U penicillin and streptomycin, 1 U nonessential amino acids, 1 mM sodium pyruvate (Gibco, Life Technologies, Grand Island NY).

Western blots
Prior to lysate preparation, adherent cells were released with 0.05% Trypsin-EDTA (Gibco, Life Technologies, Grand Island NY), then harvested by centrifugation at 2500 × g for 4 min. The cell pellets were washed 3× with 1 mL DPBS followed by centrifugation at 2500 × g for 4 min. Pellets were lysed using 1 mL of M-PER® Mammalian Protein Extraction Reagent for~100 μL of cells and processed as per manufacturer's instructions (Pierce, ThermoFisher Scientific, Rockford IL). The protein concentration of cell lysates was quantified using a BCA Assay Kit (Pierce, ThermoFisher Scientific, Rockford IL). Intensity of RPN13 and actin bands was analyzed using the NCI software ImageJ (http://rsb.info.nih.gov/ij/index.html) and their ratio determined for normalization.

Chromogenic in situ hybridization
Custom RNA in situ hybridization probes (Advanced Cell Diagnostics, Inc.) were prepared to detect the fulllength ADRM1 mRNA. RNAscope® assays were performed using the RNAscope® 2.0 FFPE Brown Reagent kit. Glass-mounted 4 μm sections of formalin-fixed, paraffin-embedded tissue (FFPE) were baked overnight at 65°C, then deparaffinized in three 10 min incubations in xylene, with the last incubation in fresh xylene. Endogenous peroxidase activity was blocked for 10 min (P1) followed by antigen retrieval by both heat (P2) and protease (P3). Protease antigen retrieval (P3) was diluted 1:5 in sterile DPBS (Gibco, Life Technologies, Grand Island NY) and incubated for 30 min. ADRM1 probe hybridization and amplifications were completed according to manufacturer's instructions. After washing, an HRP-based amplification system was then used to detect the target probes by 10 min incubation with manufacturer supplied DAB solutions. To ensure RNA integrity and assay procedure, adjacent sections were also hybridized with a probe to the endogenous housekeeping protein peptidyl prolyl isomerase B (PPIB). To control for non-specific staining, control probe against bacterial protein dihydrodipicolinate reductase (dapB) was used on adjacent slides. Quantification of slides was done using ACD supplied SpotStudio Software.

ADRM1 knockout cell line
Stable ADRM1 knockout cell line was established using CRISPR/Cas9-mediated gene targeting strategy in the HCT116 Cell line. A 0.9-kb fragment upstream of exon 4 as a 5′-homologous arm and a 1.0-kb fragment downstream of exon 4 as a 3′-homologous arm in ADRM1 gene (encoding RPN13) was amplified by PCR using genomic DNA from HCT116 cells as a template. The primers used were 5′-AACTCGAGTGAAGGGGACCA CCGTGACTCCG-3′ and 5′-TTGAATTCTTGGACCC TGCCTTGAACTTCAGC-3′ for the 0.9-kb fragment, Fig. 3 SpotStudio Quantification of RNA-CISH in fallopian tube epithelium, STICs, and HGSCs. Using ACD SpotStudio Software, single cell analysis for CISH were done on all 7 matched normal, STIC, and HGSC samples. STICs and HGSC showed a higher percentage of cells overexpressing ADRM1 mRNA as compared to fallopian tube epithelium. Distribution of cells containing higher number of ADRM1 mRNA spots increases FT progresses to STIC and carcinoma. Collectively, STIC and HGSC exhibited significantly different levels as compared to normal (P = 0.009, P = 0.003, paired T-test). One representative case (Patient A) is presented, and the remainder (Patients B-G) are presented in Additional file 4: Figure S4 and 5′-AAGGATCCATGCTGGCCCTGGTTCTAACG ATG-3′ and 5′-TTTGCGGCCGCTCCGAAGGCACTT AGCTGCTGC-3′ for the 1.0-kb fragment. A targeting vector was constructed by sequentially subcloning the 5′-arm, the puromycin resistance gene cassette, and the 3′-arm into pBluescript II in order to delete exon 4 of ADRM1 gene. sgRNA-encoding DNA oligos, 5′-AAACTGCGATTCGGTTAGGAACTTC -3′ and 5′-CACCGAAGTTCCTAACCGAATCGCA -3′, target 39-bp downstream and of exon 4. The pair of annealed oligos was subcloned into the BbsI site of pX330 (Addgene 42,230), which expresses sgRNA and Cas9 simultaneously. The targeting vector and the pX330 plasmid encoding each sgRNA were transfected into HCT116 cells. The cells were cultured in DMEM supplemented with 10% FBS and 4 μg/ml puromycin for 10 days. Colonies were then picked up, and   Similarly, purified mRNA was measured by qRT-PCR and found to have no correlation with amplification status. b As assayed by Western blot, RPN13 protein levels vary little despite ADRM1 gene amplification status. Ratio to loading control, ß-actin, was calculated by comparing pixel density as measured by ImageJ. C) XTT cell viability assay of various HGSC cell lines with known RPN13 CNV. All cell lines were shown to be sensitive to RPN13 inhibition by RA190 regardless of amplification status