The present study is based on a multi-center cohort of 43 patients who received surgical treatment for primary epithelial ovarian cancer between 2003 and 2009. Tissues were sampled during primary surgery at the Departments of Obstetrics and Gynecology of the university Medical Center Freiburg, Charité Campus Virchow-Klinikum Berlin, Medical Center Bayreuth, university Medical Center Greifswald, university Medical Center Vienna and the Oncological Institute of Moldavia and stored at the Tumor Bank Ovarian Cancer (TOC). Specimens were freshly frozen in liquid nitrogen and immediately stored at −80°C until further analysis. Clinicopathological parameters were documented in the TOC database including age, date of initial diagnosis and surgery, FIGO and TNM stage, grade, histological subtype, tissue sample localization, post-surgical therapies and follow-up data. Forty-five samples from 43 patients were collected and analyzed. Usually one tumor specimen was obtained from each patient (41 patients). In two patients 2 tumor samples of different origin (omentum majus/left ovary and omentum majus/uterus) were collected. In addition, 4 non-pathological ovarian tissue specimens were obtained at surgery performed at the Department of Obstetrics and Gynecology of the university Medical Center Freiburg and were histologically classified at the Institute of Pathology. All patients provided written consent and the study was approved by the Ethics Review Committee of the university Medical Center Freiburg, Germany, in accordance with human rights for patients in research.

The patient cohort (N=43) was subgrouped for statistical analysis in regard to age, FIGO stage, TNM stage, histological subtype, grade, localization of tumor tissue sampling and resistance to platinum-derived chemotherapy (Table II). Statistical analysis was performed using SPSS software (version 20.0; SPSS, Inc., Chicago, IL, USA). Correlations were tested with the Spearman’s correlation coefficient (r_S) and the Mann-Whitney U test. A P-value <0.05 was considered to indicate a statistically significant result.

Total RNA was isolated by TRIzol^® reagent (Invitrogen™, Life Technologies™ GmbH, Thermo Fischer Scientific, Darmstadt, Germany) method according to the manufacturer’s protocol after homogenization of tissue specimens by Ultra-Turrax^® (IKA Werke, Staufen, Germany). RNA quality and quantity was determined by densitometry. Reverse transcription of 2 μg purified RNA into cDNA was performed using MMLV-RT (Promega GmbH, Mannheim, Germany), RiboLock RNase inhibitor (Fermentas GmbH, Thermo Fischer Scientific, St. Leon-Roth, Germany), and random hexamer primers (New England BioLabs GmbH, Frankfurt, Germany) followed by either real-time PCR using QuantiFast SYBR-Green PCR Master Mix (Qiagen, Hilden, Germany) or conventional RT-PCR Taq polymerase (Genaxxon Bioscience GmbH, Ulm, Germany) and gene-specific primers. For best possible precision only singular PCR was performed. Isoforms with known oncologic potential, such as isoforms with exon 5 and 6 skip (potential RONΔ165) and exon 11 skip (potential RONΔ160) were investigated. A PCR spanning exons 4 to 12 was not performed since estimated amplicon sizes (>1,000 bp) were too big for accurate separation. Expression levels of housekeeping gene RPS18 served as the comparative value. The sequence of primers and expected amplicon sizes were: for real-time PCR: RPS18-S, 5′-TAC TCA ACA CCA ACA TCG ATG GGC-3′ and -AS, 5′-GCT TTC CTC AAC ACC ACA TGA GCA-3′; HIF-1α-S, 5′-CGT TCC TTC GAT CAG TTG TC-3′ and -AS, 5′-TCA GTG GTG GCA GTG GTA GT-3′; RON standard-S, 5′-TGA GGT CAA GGA TGT GCT GA-3′ and -AS, 5′-GTG ACT TGA TGG CAC ATT GG-3′; SRp75-S, 5′-TAA GGG CTA CGG GAA GAT CC-3′ and -AS: 5′-CCA CAA AGG TCT TTG CCA TT-3′; SRp55-S, 5′-GAC GGC TAC AGC TAC GGA AG-3′ and -AS, 5′-CCA ACT GCA CCG ACT AGA AA-3′; SRp40-S, 5′-TCC AAG GGA TGC AGA TGA TGC TGT-3′ and -AS, 5′-TAT CGT CCT CTA CCT CTT CCA CCT-3′; SRp20-S 5′-GAA GAC TCA TCG GAG CGT GT-3′ and -AS, 5′-TGT TGC CAT TGT TTC CAA GA-3′; htra2-β1-S, 5′-ATG ACC AGC AGT CTA GGC GTT CAA-3′ and -AS, 5′-ATC CTA CGC CCA TCA AGC TCC ATT-3′ and YB-1-S, 5′-TGG TTC AAT GTA AGG AAC GGA-3′ and -AS, 5′-ACT GCG AAG GTA CTT CCT GG-3′; ASF/SF2-S, 5′-CGT GTT CTA CAA ATA CGG CG-3′ and -AS, 5′-ACC CAT CGT AAT CAT AGC CG-3′; for RT-PCR: RONΔ165 (exon 11 skipping) S 5′-ACC TAG TTC CAC TGA AGC CT-3′ and -AS, 5′-ACC AGT AGC TGA AGA CCA GT-3′ (339 and 147 bp); RONΔ160 (exon 5–6 skipping) S, 5′-CAC TGC CCA CCT AAG CTT AC-3′ and -AS, 5′-CTG GTG CCT ACA GAC AGA CT-3′ (445, 285, 279 and 119 bp). RT-PCR products were subsequently analyzed by gel electrophoresis.

Protein isolation was performed in parallel to RNA extraction by TRIzol® reagent method according to the manufacturer’s protocol after homogenization of tissue specimens by Ultra-Turrax®. Protein concentrations were determined with the BCA™ protein assay kit (Pierce, Thermo Fischer Scientific, Rockford, IL, USA). Protein samples were separated by 10% SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF) membranes (Carl Roth GmbH & Co. KG, Karlsruhe, Germany) by using a Bio-Rad Trans-Blot Cell (Bio-Rad Laboratories GmbH, Munich, Germany). Membranes were soaked in blocking solution [phosphate-buffered saline (PBS) with 3% dry non-fat milk, 0.1% Tween-20] for 1 h at room temperature and incubated with the primary polyclonal rabbit IgG antibody anti-RON β (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA; dilution 1:500) in blocking buffer overnight at 4°C. After washing, the membranes were further incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (Jackson ImmunoResearch Europe Ltd., Suffolk, UK; dilution 1:7,500) in blocking buffer for 2 h at room temperature. Finally the immunocomplexes were made visible by an enhanced chemiluminescence western blot detection kit (Pierce) and exposed to X-ray film (Fujifilm Europe GmbH, Düsseldorf, Germany).

Forty-five tumor samples of primary ovarian cancer and 4 non-pathological ovarian tissue samples were analyzed in regard to the expression profiles of total RON (all isoforms). In addition, the expression of the alternatively spliced RON isoforms with exon skip 5–6 (potential RONΔ160 and RONΔ155) and exon skip 11 (potential RONΔ165) bearing a known oncogenic potential was investigated in particular.

High levels of RON expression were detected in all 45 ovarian tumor tissues with statistical significance (105.94±6.46 vs. 0.00, P=0.001). The origin of the tumor specimens did not influence the expression levels of RON; tissue samples of primary tumors (N=25) as well as metastases (N=20) showed high levels of RON expression (105.29±5.45 vs. 106.70±7.46, P=0.337). In contrast, all non-pathological (physiological) ovarian tissue samples (N=4) were characterized by no detectable RON expression (Fig. 1 and Table III). Alternative splicing of RON exon 5/6 or exon 11 was noted in 39 of the 45 tumor samples (86.67%) whereas 6 tumor samples showed no alternative splicing of RON (13.33%). Notably, the non-pathological ovarian tissues did not express any RON mRNA variants.

RT-PCR analysis employing RON exon 4 to 7 spanning primers resulted in the detection of three different amplicons. Full length RON (445 bp) expression was found in most malignant ovarian tissues (39 of 45; 86.67%). The expected potential RONΔ160 or RONΔ155 mRNA (119 bp), characterized by exclusion of exon cassette 5-6, was observed in a fraction of tumor samples only (11 of 45; 24.40%). The third RT-RCR amplicon with estimated 280 bp was detected at a higher frequency in the ovarian cancer tissue specimens (29 of 45; 64.44%). This RON mRNA could have been either characterized by solitary exon 5 (284 bp) or exon 6 exclusion (297 bp), whereas subsequent sequence analysis clearly identified the exon 5 skipping mRNA variant (284 bp) (Fig. 2A). In the metastatic tissues, expression of RON mRNA variants with exclusion of exon cassette 5-6 was more frequent than that in the primary tumors (35 vs. 16%), whereas in the primary tumor samples, expression of the exon 5 skipping mRNA variant (285 bp) was observed more often (76 vs. 50%).

Expression analysis of RON variants with exon 11 skipping (potential RONΔ165) via RT-PCR was performed using RON exon 10 to 12 spanning primers and resulted in the detection of conventionally spliced RON including exon 11 (339 bp) as well as an exon 11-skipping RON variant (192 bp) with varying intensity in the majority of the ovarian tumor tissues (37 of 45; 82.22%). In the metastatic tissues, expression of RON variants with exclusion of exon 11 was more frequent than that in the primary tumor tissues (90 vs. 76%). Additionally, another amplicon (509 bp) occurred in almost all analyzed malignant tissues (44 of 45; 97.78%). This mRNA originated from intron 10–11 and intron 11–12 retention. Sequencing analysis revealed that this PCR product does not encode for a biologically active protein, since premature termination codons (PTCs) are introduced (Fig. 2b). PTCs most likely constitute RNA degradation via nonsense-mediated decay.

Quantitative RON protein expression analysis (western blot analysis) employed a specific antibody directed against the C-terminus of RON. Non-pathological ovarian tissues showed expression of two RON protein isoforms only, in fact unprocessed pre-RON (190 kDa) and ‘short form RON’ (RONΔ55 with 55 kDa). Ovarian tumor specimens also expressed pre-RON and RONΔ55, plus a variety of additional protein isoforms, e.g. RONΔ155 (125 kDa β-chain), RONΔ160 (135 kDa, β-chain), RONΔ165 (165 kDa, single chain) (data not shown). Only slightly differing molecular weights of the different isoforms made a precise quantification impossible.

In the ovarian tumor samples, a significant correlation between total RON splicing and the expression of splicing factors ASF/SFRS1 (P=0.035), SRp40 (P=0.015), SRp55 (P=0.015) and SRp75 (P=0.039) was detected. More detailed analysis comparing tumor samples with alternative splicing of RON exon 5/6 or exon 11 (N=39) to tumor samples without alternative splicing of RON exon 5/6 or exon 11 (N=6) revealed different correlations to splicing factor expression. While tumor samples with alternative splicing events of RON were characterized by significant correlations of RON expression and the expression of certain splicing factors (SRp40, P=0.047; SRp55, P=0.005; SRp75, P=0.019), no significant correlations in tumor samples lacking RON alternative splicing were found, except for SRp40 (P=0.005). Notably, tumor tissues with alternative splicing of exon 5/6 (potential RONΔ16 0, N=11) showed no correlation of RON expression to splicing factors. In contrast, in the tumor samples with alternative splicing of exon 11 (potential RONΔ165, N=37), significant correlations of RON expression to SRp40 (P=0.033), SRp55 (P=0.05) and SRp75 (P=0.028) were determined (Table IV).

A significant correlation of FIGO stage and grade (P=0.014) as well as of FIGO stage and platin-resistance (P=0.034) was observed in the tumor samples in general. Corresponding results were observed in the tumor samples with RON alternative splicing events (FIGO stage/grading, P=0.031; FIGO-stage/platinum-resistance, P=0.028). Tumor samples without alternative splicing of RON showed no correlations to the clinicopathological parameters.

Significant correlations of RON expression and clinicopathological parameters such as age, FIGO stage, grade, histological subtype or resistance to platinum-derived chemotherapy were not detected in the analyzed tumor samples in general as well as in any subgroup (e.g. tumor samples with RON alternative splicing) (Table V). An exceptional significant correlation of RON expression with grade (P=0.043) was found in the metastatic issues.

To elucidate the potential role of intrinsic splicing factors within malignant cells, we investigated the expression patterns of those factors, known to be involved in the regulation of RON mRNA processing.

Expression levels of splicing factors ASF/SFRS1, htra2-β1, YB-1, SRp20, SRp40, SRp55 and SRp75 were evaluated in the ovarian tumor samples (N=45) and non-pathological ovarian tissue specimens (N=4) by real-time-RT-PCR analysis. In comparison to the non-pathological tissues, the expression of all investigated splicing factors showed a significant upregulation in the ovarian tumor samples (Table VI).

Except for SRp20 (118.00±7.86 vs. 113.75±6.72, P=0.027) no difference in expression levels of the splicing factors in the primary tumor samples (N=25) vs. metastases (N=20) were evident.

In tumor samples with alternative RON splicing various significant splicing factor interactions were observed (e.g. SRp55/SRp75; P<0.001). Notably, this phenomenon was not detected in the tumor samples lacking RON alternative splicing (Table VII).

Statistical correlation analysis in regard to splicing factor expression and clinicopathological features showed a significant correlation of YB-1 (P<0.001), SRp40 (P=0.046), SRp55 (P=0.002) and SRp75 (P=0.002) with FIGO stage I and II (N=12) compared to FIGO stage III and IV (N=30). YB-1 also showed a significant correlation to platinum resistance (P=0.042). Other significant correlations of splicing factor expression and clinicopathological parameters such as age, grade or histologic subtype were not observed.