Role and expression of FRS2 and FRS3 in prostate cancer

Malignant and benign prostate cell lines were cultured in RPMI (Invitrogen) supplemented with 10% FBS. All cell lines used (LNCaP/DU145/PC3/PNT1A/PNT2) were purchased commercially (American Type Culture Collection or Health Protection Agency Culture Collections). 2 μg of myc-tag pCDNA3.1-FRS3 (received from Dr S Meakin) or pcDNA3.1 empty was transfected using Lipofectamine 2000 (Invitrogen). Transfected cells were placed under G418-sulphate selection for 14-20 days. Individual colonies were removed by trypsinization and expanded, and clones screened for FRS3 expression. For androgen induction experiments, LNCaP cells were initially grown for 24 hours in standard media and then maintained in charcoal stripped serum supplemented media for 24 hours. Following this, synthetic androgen (R1881) was added at a dose of 10 nM. Cells were harvested at the indicated time points and RNA extracted for real time PCR analysis of FRS2 and FRS3 and PSA expression.

Cells were grown for 24 hours prior to transfection. 33 nM of siRNA oligonucleotides: Silencer^(R) Select Pre-designed siRNA FRS2, s21262 (Applied Biosystems) and ON-TARGETplus FRS3, L-019038-00-0020 (Dharmacon) was used to transfect cells. Silencer^(R) Select Negative Control #1 siRNA, 4390844 (Applied Biosystems) was used in parallel as the scramble control. The results presented are the mean value of six separate set of experiments. Total RNA was extracted using RNeasy Mini Kit (Qiagen) and reverse transcribed using Transcriptor cDNA Synthesis Kit (Roche Diagnostics) according to the manufacturer's instructions. Real time sequence specific Taqman PCR primers for human FRS2 (Hs00183614_m1), FRS3 (Hs00183610_m1) and GAPDH (Hs02786624_g1) were purchased from Applied Biosystems. qPCR was preformed with Light Cycler 480 (Roche Diagnostics) and Taqman Universal PCR Master Mix (Applied Biosystems) and corrected for GAPDH expression. PCR was repeated in triplicate and performed three times with results expressed as the mean and standard deviations.

Cells were seeded at a density of 3000 (DU145, DUEV, DUFRS3) and 5000 (PC3, PNT2) cells in 96-well plates and allowed to grow for 36 h. Medium lacking serum was termed 'basal medium' (BM), while medium containing 10% FBS was termed as "full medium" (FM). Cells were then starved in BM for 16 h before stimulation with FM. Cell proliferation was assessed with WST-1 reagent (Roche Diagnostics). Experiments were repeated in triplicate and done three times.

Cells were lysed in Laemmli buffer and denatured. Samples were then separated using 10% Bis-Tris pre-cast gels (Invitrogen), followed by transfer to a PDVF membrane (GE Healthcare). Primary antibodies: FRS2-A5 (sc-17841), pERK-E4 (sc-7383), ERK-6G11 (sc-81458), c-Myc-9E10 (sc-40) were purchased from Santa Cruz Biotechnologies, Abcam (α-tubulin, ab4074-100) and BD Biosciences (PARP, 65196E). FRS3 antibody was received from Dr S Meakin and has been previously described [22]. Primary antibody complexes were detected using HRP-conjugated secondary antibodies (Dako). Protein bands were visualised using ECL (GE Healthcare).

Transfected cells were re-plated out into BD migration or invasion chambers (Scientific Laboratory Supplies) in basal medium (BM) at a concentration of 80, 000 cells per insert. FM or FGF1 (10 ng ml^-1), FGF2 (10 ng ml^-1) and FGF8 (10 ng ml^-1) prepared in basal media were used in the lower chamber as chemoattractant. Cells were allowed to migrate or invade for 24 hours and fixed in methanol for 20 minutes at -20°C, stained with haematoxylin, washed with dH₂O and allowed to dry before mounting them in a microscopic slide with DPX (Sigma Aldrich). Cells were counted using a bright field microscope at 20× magnification. Five different fields of view were used to obtain an average count per section. Results shown are the mean of three experiments and expressed as a fold increase over un-induced scramble controls (migration) or as invasion index (compared to and corrected for control inserts). Statistical analysis was performed using two-tailed Student's T-test. p < 0.05 being significant.

The tissue microarray (TMA) used in this study has been described [5]. The study cohort included 129 cancers and 36 benign biopsies. All material was used in accordance with approval granted by the local hospital ethical committee. Mouse monoclonal FRS2, (Santa Cruz) and rabbit polyclonal FRS3 antibody (received from Dr S Meakin) used in this study, have been previously validated for use in immunohistochemistry [18, 22]. Scoring was done by two independent observers blinded to the clinical detail and the scores collated and analysed by a third investigator (VJG). The score was carried out by assessing the intensity of stain for each core, where immunoreactivity signals was assessed as being absent or weak (0/+) and moderate or strong (++/+++). Data were analysed using the Kruskal-Wallis test. p < 0.05 was taken as being statistically significant.

A panel of cell lines was first analysed by real time PCR to define expression levels in prostate cells. FRS2 and FRS3 mRNA were ubiquitously expressed in normal epithelial prostate cell lines (PNT1A/PNT2) and prostate cancer cell lines (LNCaP, DU145 and PC3) (Figure 1A). FRS2 levels were generally higher in all cell lines compared to FRS3 regardless of androgen receptor expression status. A similar pattern of expression in these cell lines was seen in western analysis of FRS2 and FRS3 protein (Figure 1A). We also tested if FRS expression was inducible by androgens. In LNCaP cells neither FRS2 nor FRS3 mRNA expression levels were increased by the addition of androgens suggesting that they are not androgen regulated (Figure 1B). In contrast, simultaneous analysis of PSA expression as a control was significantly increased in LNCaP cells treated with androgens. The function of FRS2 has been well explored in the literature but less is known about FRS3. Given evidence of expression overlap in our cell lines, we tested functional overlap between FRS2 and FRS3 in prostate cancer. A stable myc-tag FRS3 over-expressing clone in DU145 prostate cancer cell lines was first generated (DUFRS3) (Figure 2A). DU145 were selected as an androgen independent cell line with relatively low FRS3 expression compared to FRS2 (Figure 1A). In proliferation assays FRS3 over-expression significantly enhanced cell proliferation in response to stimulation by FGF1 (p < 0.005), FGF2 (p < 0.001) and FGF8 (p < 0.001) compared to empty vector control (Figure 2B). FRS3 has previously been implicated as an inhibitor of EGF signalling [19, 20]. In these experiments however increasing FRS3 did not appear to alter (increase or inhibit) the magnitude of EGF stimulated proliferation (Figure 2B). We next asked if FRS3 over-expression could compensate for FRS2 down-regulation as has been previously described in embryogenesis [18]. In serum-starved DUEV cells, the addition of FGFs resulted in enhanced proliferation (p < 0.005) (Figure 2C) concomitant with rapid phosphorylation of ERK (Figure 2D, upper panels). Targeted FRS2 knock down reduced this induction regardless of the FGF ligand used (Figure 2C, D upper panels). The effect of silencing FRS2 in the DUFRS3 clone was next tested. Here, over-expression of FRS3 was able to compensate for the effect of targeted knock down of FRS2 when stimulated with different FGFs (Figure 2C). In parallel studies, FRS2 knock down in FRS3 over-expression clones had no discernable effect on ERK phosphorylation (Figure 2D, lower panels). These studies suggest functional redundancy between FRS2 and FRS3 in prostate cancer cells.

Our results suggest that both FRS2 and FRS3 are important in FGF mediated signalling in prostate cancer. We therefore investigated the effect of dual targeting of FRS2 and FRS3. The efficiency of dual silencing was confirmed at the protein level. Full media stimulation of scramble transfected DU145 and PC3 cells resulted in a significant increase in proliferation compared to un-stimulated cells. Dual silencing of FRS2 and FRS3 in PC3 and DU145 however reduced this induction in comparison to these scramble controls (p < 0.005) (Figure 3A and 3B). We did observe that the dual knock down effect was more noticeable in PC3 cells compared to DU145 particularly in the earlier phases of proliferation. This effect could be due to differences in transfection efficiency between the cell lines and/or inherent variability between the cell types in response to FGF stimulation and FRS utilisation. Immunoblot analysis did demonstrate comparable down-regulation of FRS2 and FRS3 protein in both cell lines following transfection. We next repeated the experiments in PNT2 benign prostate cells known to respond to FGF stimulation. In these cells we surprisingly observed that FRS2 and FRS3 suppression did not alter cell proliferation (Figure 3C). We next tested the effect of dual FRS2 and FRS3 silencing in the context of different FGF stimulation in migration and invasion experiments. Treatment of PC3 scramble control cells with FGF1, FGF2 and FGF8 resulted in a significant increase in migration compared to un-stimulated controls. FRS2 and FRS3 dual knock down in these cells however significantly reduced migration regardless of the ligand used in comparison to scramble transfected controls (Figure 4A). Similar results were seen in experiments with DU145 cells (Figure 4B). Invasion assays were next carried out in PC3 cells using FGFs as well as EGF as a chemo-attractant. Dual silencing of FRS2 and FRS3 in PC3 cells reduced the ability of cells to induce invasion upon FGF stimulation (p < 0.005) (Figure 4C). No significant change in the invasion index was noted when siRNA FRS2 and FRS3 PC3 cells were stimulated with EGF compared to EGF-stimulated scramble control cells. A similar finding was observed for siRNA FRS2 and FRS3 transfected DU145 cells in invasion studies (Figure 4D) (p < 0.05). The effect of dual knock down of FRS2 and FRS3 on FGF induced MAPK activation was next tested. In this experiment, the intensity and duration of ERK phosphorylation following PC3 cell stimulation with either FGF1 or FGF2 in siRNA treated FRS2 and FRS3 cells was dramatically reduced compared to scramble controls in keeping with the functional data observed above (Figure 4E).

We have observed that combined silencing of FRS2 and FRS3 resulted in a significant inhibition in FGF induced ERK activation in prostate cancer cell lines. MAPK signalling has been shown to play a major role in the cancer cell response and recovery to radiation therapy [23, 24]. We therefore tested if targeting FRS2 and FRS3 was a potential therapeutic target in combination with radiation. In these studies silencing FRS2 and FRS3 in PC3 cells significantly inhibited colony formation following irradiation compared to scramble transfected controls (p < 0.05) (Figure 5A). A similar finding was observed in siRNA FRS2 and FRS3 transfected DU145 cells (p < 0.002) (Figure 5B). These experiments were next repeated with benign PNT2 cells. In contrast to cancer cells, the clonogenic survival after radiation between scramble and siRNA FRS2 and FRS3 cells was not significantly different in these benign epithelial cell lines (Figure 5C). We next tested whether manipulation of FRS2 and FRS3 might have had a direct effect on inducing cell apoptosis. siRNA FRS2 and FRS3 or scramble transfected PC3 cells were assayed by immunoblotting for the extent of PARP cleavage following irradiation (Figure 5D). In this experiment we did not observe any difference in the levels of PARP cleavage in either scramble or siRNA FRS2 and FRS3 transfected samples. These results suggest that FRS2 and FRS3 suppression is not likely to have a direct effect in enhancing apoptosis. Rather, it is more likely that FRS2 and FRS3 suppression may function to inhibit pro-survival signalling which can enhance cell recovery and proliferation in the immediate period after a cytotoxic insult.

An important therapeutic question is if either FRS2 or FRS3 are over-expressed in cancer. We therefore tested expression of FRS2 and FRS3 transcript by quantitative real time PCR in a panel of archival clinical prostate cancers using methods previously developed in our group [25]. Benign (n = 5), Grade 3 (n = 4), Grade 4 (n = 6) and Grade 5 (n = 9) tumors each derived from separate individual patients were micro-dissected and RNA extracted and profiled for FRS2 and 3 mRNA expression. In this analysis we found that overall FRS2 transcript levels were higher in benign and malignant biopsies compared to FRS3 consistent with our findings in cell lines (Figure 6A). There was however no difference in expression of FRS2 or FRS3 comparing benign or malignant samples (p = 0.8 for both comparisons). We also tested if with increasing cancer grades there was any evidence of a progressive increase or decrease in expression and found no significant alterations (p = 0.23 for FRS2 and p = 0.87 for FRS3). To further investigate expression in clinical tissue, we interrogated a publicly available dataset. The MSKCC Prostate Oncogenome Project was interrogated for FRS2 and FRS3 transcript expression (n = 131 primary tumours with mRNA data available) using the cBio Cancer Genomics Portal [26]. In this resource (using a Z threshold value of 2.0) there was minimal alteration in FRS2 or FRS3 expression. Only 6% and 8% of tumours had any changes in FRS2 and FRS3 expression compared to normal controls respectively (Figure 6B). These findings corroborate our own data from clinical samples and show that both FRS2 and FRS3 mRNA are expressed in clinical samples suggesting expression redundancy. FRS2 appears to be the predominantly expressed transcript in prostate tissue but neither is significantly over-expressed or down-regulated in the transition to a malignant phenotype. Finally, we next asked if these observations were also true in terms of protein expression. Using a clinical prostate tissue microarray we tested for expression of FRS3 and FRS3 in cohorts of defined cohorts of benign (n = 34) and malignant (n = 129) prostate samples (Figure 7). Comparison of overall expression in these two groups demonstrated no overall significant difference between the groups at the protein level for either FRS2 or FRS3.