The G Protein-Coupled Receptor RAI3 Is an Independent Prognostic Factor for Pancreatic Cancer Survival and Regulates Proliferation via STAT3 Phosphorylation

Pancreatic Ductal Adenocarcinoma (PDAC) is one of the deadliest tumors worldwide. Understanding the function of gene expression alterations is a prerequisite for developing new strategies in diagnostic and therapy. GPRC5A (RAI3), coding for a seven transmembrane G protein-coupled receptor is known to be overexpressed in pancreatic cancer and might be an interesting candidate for therapeutic intervention. Expression levels of RAI3 were compared using a tissue microarray of 435 resected patients with pancreatic cancer as well as 209 samples from chronic pancreatitis (CP), intra-ductal papillary mucinous neoplasm (IPMN) and normal pancreatic tissue. To elucidate the function of RAI3 overexpression, siRNA based knock-down was used and transfected cells were analyzed using proliferation and migration assays. Pancreatic cancer patients showed a statistically significant overexpression of RAI3 in comparison to normal and chronic pancreatitis tissue. Especially the loss of apical RAI3 expression represents an independent prognostic parameter for overall survival of patients with pancreatic cancer. Suppression of GPRC5a results in decreased cell growth, proliferation and migration in pancreatic cancer cell lines via a STAT3 modulated pathway, independent from ERK activation.


Introduction
Pancreatic cancer is one of the most perilous and lethal solid malignancies and the five-year relative survival rate, is one of the lowest for all cancers with 8% [1]. Typical alterations in the molecular signaling include mutations in KRAS in 95% of PDACs and in tumor suppressor's genes like p16/CDKN2A, p53 or SMAD4 [2]. Besides these, many other changes in different PLOS ONE | DOI: 10.1371/journal.pone.0170390 January 23, 2017 1 / 15 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 pathways have been described, which are caused by gene mutations as well as epigenetic mechanisms [2,3]. The G-protein-coupled receptor family C, member 5, group A (GPRC5A); or retinoic acidinducible 3 (RAI3) gene was firstly described in 1998 as a seven transmembrane helices protein with a molecular weight of 40 kDa and its expression is regulated by all-trans-retinoic acid (ATRA) [4]. GPRC5a is over expressed in pancreatic, breast, gastric and colon cancer, and it has been shown that overexpression of RAI3 predict poor prognosis in hepatocellular carcinoma [5][6][7][8][9][10][11]. Functional analysis of GPRC5a in different carcinomas indicated that RAI3 has pleiotropic effects and might confer resistance to Gemcitabine in pancreatic cancer, however the ligand of GPRC5a remains elusive [12], but GPRC5a overexpression might lead to a reduction of EGFR levels [13][14][15][16].
In this study, we investigated the expression and function of RAI3, in pancreatic cancer. We verified the reported RAI3 RNA over-expression using immunohistochemistry and we could show that apical RAI3 expression is an independent prognostic marker for overall survival. The siRNA mediated knock-down of RAI3 lead to proliferation and migration inhibition via STAT3 inactivation.

Tissue microarray (TMA): patient samples and immunohistochemistry
Under consideration of the appropriate guidelines, TMA were prepared and evaluated [17]. The tissue was collected from routine surgery dissections and the histological diagnosis was confirmed by an expert histopathologist. Samples of 598 patients were collected with approval by the ethics committee of the Technical University Dresden, University Jena and University Regensburg for biobanking purposes and written informed consent of each patient was obtained. Records of the informed consent are part of the biobank database. The use of the tissue for TMA construction and immunohistochemical staining was approved by the ethics committee.
Immunohistochemical staining was performed with a VENTANA BenchMark XT (Ventana, Roche Diagnostics, Freiburg) system. For deparaffinization and retrieval of antigen, the slides were heated and pretreated with CC1 (Ventana). Subsequently, peroxidase was blocked and the incubation with RAI3 antibody (1:400, Novus Biologicals, Wiesbaden, NBP1-89743) followed. The detection was performed using UltraView Universal DAB Detection Kit (Ventana). Finally, the slides were counterstained with haematoxilin and Bluing Reagent (Ventana). As a negative control rabbit IgG was used. The scoring of the TMA was performed by two scientists (EJ, DA), independent on clinical data. Three parameters were pre-defined: the intensity of the staining (ordinal scale, 0-3 points), the portion of stained ductal cells (metric scale, 0-100%, called "expression") and the portion of strictly apical stained cells as part of all stained ductal cells (metric scale, 0-100%, called "apical expression"). Patients were only included in the evaluation, when minimum one TMA core was interpretable. For a TMA core with an intensity of 0 points the expression is assessed with 0% and the apical expression is not assessable. The cut-off of >30% of apical stained cells were used in the univariate analysis.
Samples of 501 patients (83.8% of the samples) could be included in the analysis of RAI3-expression. Of these were 435 of pancreatic adenocarcinoma (tumor samples), 34 of Intraductal Papillary Mucinous Neoplasm (IPMN) and 32 of chronic pancreatitis. Normal ductal tissue adjacent to the tumor was used for comparison in 143 samples.
Analysis of clinico-pathological and survival data were performed for 376 patients with pancreatic adenocarcinoma and a postoperative survival time over 30 days.

Functional assays
As a assay of seeding efficiency after 72 hours of siRNA transfection cells were detached and 3.000 cells were seeded in 12-well-plates. Cells grow for seven days and stained with Coomassie Brilliant Blue staining solution (Brillant Blau G 250, Roth, Karlsruhe, Germany). Quantification was performed by Gene Tools Analysis software (Syngene, Cambridge, UK).
After 48 h of transfection cells were seeded in Fluoro Blok Individual Cell Inserts (Becton Dickinson, Heidelberg, Germany) with serum-free medium. Inserts were transferred in 24-Well-plates containing complete growth medium with 20% fetal bovine serum. The cells were allowed to migrate in a period of exactly 16 hours. The inserts were washed in PBS (Thermo Fisher Scientific), fixed with Methanol and stained with DAPI (Sigma-Aldrich). The distribution of migrated cells was assessed, four image sections per well photographed and counted with Image J (Wayne Rasband, National Institutes of Health).

Results
In PDAC and IPMN the percentage of RAI3 expressing cells is highly elevated (median = 90%) compared to normal (median = 40%) and CP tissue (median = 60%) and PDAC cells show mostly a strong expression. Apical expression was detected in 100% of stained normal and CP tissue as well as in 90% of the stained IPMNs. PDAC-tissue showed an obvious loss of apical expression (median = 30%) (Fig 1A-1F and Fig 2A and 2B) (Kruskal-Wallis-test p<0.001 for all parameters).
A diagnostic score to differentiate pancreatic cancer from CP or normal tissue, was calculated by addition of increasing RAI3-intensity and decreasing apical RAI3-expression displayed an AUC of 0.954 (p<0.001). Assuming that a value of minimum 4 (respective 5) of 6 points is typically for tumorous tissue, the diagnostic score exhibits a calculated sensitivity of 89.7% (respective 70.6%) and specificity of 94.3% (respective 98.3%) (Fig 3A and 3B).
Univariate analysis the clinic-pathological data of the 376 patients with PDAC analyzed displayed a significant difference (p<0.05) in survival concerning the pN-stage (regional lymph nodes affected, histology grade, resection margin and apical RAI3 expression (Table 1). Patients with an apical RAI3-Expression of 30% or lower showed a loss of 4 months in median  survival time (p = 0.017, Fig 3C). Cox-regression analysis revealed that apical RAI3-expression is a independent prognostic factor for the survival of the patients ( Table 2, RR: 1.37; p < 0.05).
To elucidate the role of RAI3 further we first compared expression levels in pancreatic cancer cell lines by Western blot and quantitative RT-PCR. The immortalized human ductal pancreatic epithelial (HDPE) cell line [21] showed only negligible expression of RAI3 protein or GPRC5A mRNA expression, whereas pancreatic cancer lines displayed highly variable but detectable over-expression (Fig 1G and 1H).
A central pathway that influences the cell proliferation and migration is the Ras/Raf/MEK/ ERK pathway. Analysis of the proteins ERK 1/2 and their activating phosphorylation sides phospho-ERK 1/2 (Thr202/Thr204) showed no alteration, but RAI3 mRNA suppression resulted in a reduced level of the phosphorylated form (Tyr705) of STAT3 in all cell lines analyzed (Fig 6).

Discussion
The results in TMA show clearly, that RAI3/GPRC5a is over-expressed in PDAC in comparison to normal pancreas or CP tissue. Interestingly, in IPMN, a cystic neoplasm that is considered to be a precursor lesion for pancreatic cancer, also high intensity and expression levels ofRAI3 were observed. Compared to other tissues the loss of apical staining in PDACs was revealed as an independent parameter for the overall survival of patients with pancreatic carcinoma. A possible background for this is the increasing loss of cell polarity and architecture with the dedifferentiation of the tumors. This could be detected earlier by immunohistochemistry than by conventional methods as seen with other marker [22]. The overexpression of RAI3 in pancreatic cancer could be confirmed in the comparison of mRNA and protein-levels in different cell lines. Thereby, RAI3 expression seems to seems to suggest a connection with the phenotypes respectively genotypes of the cell lines. The two cell lines HPAF and AsPc-1, which have a high RAI3 protein level, were isolated from ascites [23]. BxPc3 and PaCaDD 165, which have a low GPRC5A mRNA and protein level, show both a KRAS wild type [24] indicating that RAI3 expression increases with malignancy of a tumor. A previous described connection between the expression of RAI3 and an alteration of p53 in breast carcinoma cell lines [25] could not be confirmed, since cell lines with a high GPRC5A mRNA expression like PaCaDD 161 are p53 wild type and cell lines with a p53 mutation like BxPC3 display low GPRC5A level. However, the regulation of RAI3 might be dependent on the type of p53 mutation and might be co regulated by other factors like proteins of the retinoic acid receptor family [4,26].
Our results of RNA-interference experiments indicates, that GPRC5a has some function in promoting cell growth, proliferation and migration by reducing the crucial phosphorylation of STAT3 at the position Tyr705, whereas the KRAS/MAPK signaling remains mostly unchanged. This is consistent with the results of previous published experiments with the breast cancer cell lines T47D and MCF7, which showed a growth-inhibiting effect of GPRC5A-siRNA [14].
An alteration in proliferation ability might also influence the migration ability of the cells and could cause an ostensible effect on migration. However, in Panc1-cells, we demonstrated a high effect on migration ability, whereby the influences on proliferation were still low. Therefore, we assume a proliferation-independent effect of GPRC5A knock-down on migration ability of the PDAC-cells, which is dependent on the cellular context.
Nevertheless, we could demonstrate high overexpression of GPRC5a in PDACs. Our data lead to the suggestion, that RAI3 is an epigenetic regulated gene PDACs with oncogenic effects, which could be used for diagnostic and prognostic aspects. Since, GPRC5a is located in cell membrane; it represents a potential therapeutic target.

Fig 5. Effect of siRNA based knock-down on the colony formation and migration of pancreatic cancer cell lines.
A: Examples of the results of typical colony forming assays over 7 days after siRNA-treatment; B: Growth area compared to NC1 (statistically significant difference between RAI3 and NC2 is demonstrated by * for P<0.01 and ** for P<0.001 performed by t-test); C: Typical images of migrated cells in negative control (NC) and knockdown with GPRC5a siRNA (RAI3) in different cell lines; D: Number of migrated cells after 48 hour siRNA-treatment and subsequently 16 hours migration assay (* indicates statistically significant differences (P<0.05) in t-test between the negative control with non-sense-siRNA (NC) and the knock-down with GPRC5a-siRNA (RAI3)). (n = 3). doi:10.1371/journal.pone.0170390.g005