FRG1 levels in prostate adenocarcinoma
FRG1 expression was analyzed in prostate cancer by immunohistochemistry in 20 needle core biopsies along with tissue array, consisting of 180 cores (including 90 paired tumor and uninvolved tissue). Out of 20 needle core biopsies, uninvolved prostate tissue was present in 10 biopsies. For prostate cancer samples, cohort information has been provided in (Additional file 2: Table S2). Figure 1a shows strong FRG1 staining in control tissue, compared to tumor tissue. The staining pattern revealed significant reduction of FRG1 expression levels in tumor cells, compared to uninvolved secretory ductal epithelial cells of prostate. Immunoreactive score (IRS), quantified for the staining pattern, revealed that 52 out of 100 cases (p value < 0.0005) had reduced FRG1 expression in tumor tissue (Fig. 1b). FRG1 staining was negative in 39% of tumor tissue compared to 14% of uninvolved tissue. Fisher’s exact test (2-sided, df = 1) showed significant (p < 0.0001) association of negative IRS score with tumor. Staining pattern of FRG1 was mostly moderate to weak in both tumor and uninvolved tissue, with 22% weak in tumor to 40% weak expression in uninvolved tissue. Moderate staining was distributed evenly with 37% tumor cases showing moderate staining, compared to 40% cases of uninvolved tissue. 6% uninvolved tissue had high FRG1 staining compared to 2% of tumor tissue (Fig. 1c). FRG1 expression (IRS) showed significant negative correlation (Spearman correlation, 2-tailed, r2 -0.285, p value < 0.005) with tumor grade (Gleason score) (Additional file 3).
Further, to understand the effect of FRG1 expression on tumor angiogenesis, correlation analysis was done for FRG1 IRS and MVD. No significant correlation (Spearman correlation, 2-tailed) could be derived between FRG1 protein expression levels and MVD (p value > 0.05, r2 0.105) (Additional file 3). Overall, patient IHC data revealed that FRG1 expression is reduced in tumor tissue but does not correlate with MVD count.
FRG1 expression doesn’t correlate with AR status in prostate cancer cell lines
To find out if there is any prostate cancer cell line specific expression pattern of FRG1, the endogenous FRG1 expression levels were determined in PC3, LNCaP, and DU145 cells. PC3 and LNCaP cells had higher FRG1 expression compared to DU145 (Fig. 1d). As PC3 and DU145 are androgen receptor negative cells and LNCaP is androgen receptor positive, FRG1 expression cannot be correlated with the presence of receptor.
Varying effect of FRG1 on proliferation of AR negative prostate cancer cells
To find out if FRG1 has any direct effect on proliferation, we prepared DU145 and PC3 cells with ectopic expression of FRG1 and with depletion of FRG1, along with their controls (Fig. 1e-h). Cell proliferation assay revealed that FRG1 over-expression had no significant effect on proliferation of DU145 cells (Fig. 2a) but FRG1 knockdown led to significant increase in cell proliferation (Fig. 2b). PC3 cells had significantly reduced proliferation in response to ectopic expression of FRG1 (Fig. 2c) but no significant effect was observed on cell proliferation of PC3 cells with FRG1 knockdown (Fig. 2d).
FRG1 affects motility and invasiveness in AR negative prostate cancer cells
Enhanced cell motility and invasiveness are important features of tumor progression. Therefore, to investigate the role of FRG1 in cell migration and invasion we performed scratch wound healing, transwell cell migration and matrigel invasion assays.
Wound healing was significantly (p value < 0.05) reduced in cells ectopically expressing FRG1, with 52% area healed in FRG1 expression set compared to 74% of empty vector, in DU145. In PC3 cell line, FRG1 expression led to 57% area being healed, compared to 90% of empty vector set (p value < 0.05) (Fig. 2e-f, i-j, respectively). To confirm the findings, scratch wound healing assay was done in FRG1 knockdown set. Increased wound healing in FRG1 knockdown set was observed, compared to scrambled control vector set (91% vs. 68% respectively, p value < 0.05) in DU145 (Fig. 2g-h). In PC3 cells, FRG1 knockdown led to 80% reduced wound area compared to 70% of scrambled vector set (p value < 0.005) (Fig. 2k-l).
Further support was provided by transwell cell migration data, which was decreased in cells ectopically expressing FRG1, compared to empty vector control, in both DU145 (p value < 0.01) and PC3 (p value < 0.005) cell lines (Figs. 3a-b and 4e-f, respectively). This observation was reversed when FRG1 expression was silenced, in both DU145 (p value < 0.05) and PC3 (p value < 0.01) (Figs. 3c-d and 4g-h, respectively). Ectopic expression of FRG1 led to significant reduction in cell invasion in both DU145 (p value < 0.05) and PC3 (p value < 0.05) cells (Figs. 3i-j and 4m-n, respectively). FRG1 knockdown had opposite effect on cell invasion, as FRG1 knockdown led to increase in cell invasion in both DU145 (p value < 0.05) and PC3 (p value < 0.05) cells (Figs. 3k-l and 4o-p, respectively). These results clearly indicate that FRG1 reduces cell migration and invasion in prostate cancer cells in vitro.
No effect of FRG1 expression on cell properties of AR positive cell line
To figure out if the effect of FRG1 expression on prostate cancer cells is irrespective of androgen receptor status, we used LNCaP cell lines with ectopic expression of FRG1 and depletion of FRG1 along with their controls (Fig. 4a), for cell based assays. We found that FRG1 ectopic expression didn’t change proliferation (Fig. 4b), migration (Fig. 4d-e) or invasion (Fig. 4h-i) properties of LNCaP cells significantly (p value > 0.05), compared to its control. Similarly, depletion of FRG1 expression didn’t change proliferation (Fig. 4c), migration (Fig. 4f-g) or invasion (Fig. 4j-k) properties of LNCaP cells (p value > 0.05). These results indicate that the effect of FRG1 expression is cell line specific and might be dependent upon androgen receptor status.
Collectively three cell lines show significant effect of FRG1 expression level on cell migration and invasion
To figure out the general effect of FRG1 expression of cell properties we combined the data of three cell lines and used GLM, univariate test and blocked by cell type and AR status. We found that FRG1 expression doesn’t affect cell proliferation significantly (p value > 0.05). However, ectopic expression affected the migration (p value < 0.0005) and invasion (p value < 0.05) of prostate cancer cell lines significantly. Likewise, knock down of FRG1 led to significant difference in rate of migration (p value < 0.0001) and invasion (p value < 0.0001). We didn’t find any interaction between migration (wound healing and transwell migration) and invasion (p value > 0.05).
FRG1 expression level dictates expression of various cytokines and MMP1
To identify the associated cytokines affected by FRG1 expression, q-RT PCR analysis for 13 cytokines and 7 Matrix metalloproteinases was done (see additional file 1 for primer information). These cytokines have previously been reported to affect cellular phenotypes, i.e. proliferation, migration, invasion and angiogenesis. Ectopic expression of FRG1 led to no significant change in expression of cytokine and matrix metalloproteinases, in both DU145 and PC3 cells (Additional file 4). On the other hand, FRG1 knockdown showed significant change in expression of certain targets in cell line specific manner. FRG1 knockdown in DU145 led to significant increase in expression of GM-CSF (fold change = 1.51, p value < 0.0005), PLGF (fold change = 2.04, p value < 0.0005) and MMP1 (fold change = 1.83, p value < 0.01) (Fig. 5a). By FRG1 silencing in PC3 cells, expression of GM-CSF (fold change = 3.022, p value < 0.0005), MMP1 (fold change = 1.56, p value < 0.01), PDGFA (fold change = 1.58, p value < 0.0005) and CXCL1 (fold Change = 1.67, p value < 0.0005) showed significant increase in expression (Fig. 5b). Here we can infer that FRG1 may affect proliferative, migratory and invasiveness properties of cells by modulating expression of above mentioned cytokines and MMPs.
FRG1 silencing enhances p38 MAPK activation
To identify effect of FRG1 expression on key signaling pathways we checked the activation levels of ERK and p38 MAPK. FRG1 knockdown showed enhanced phosphorylation of p38 MAPK, in both PC3 and DU145 cells (Fig. 6a-b). On the contrary, no significant difference was observed on ERK phosphorylation levels during FRG1 knockdown in DU145 but slightly elevated levels of phospho ERK were observed in PC3 cells (Fig. 6a-b). Since ERK phosphorylation was neither prominent nor consistent, it was not taken up for further study.
To further confirm the specificity of p38 activation by FRG1 knock down, we transfected FRG1 depleted DU145 and PC3 cells, with FRG1 expression vector. We observed that effect of FRG1 depletion on p38 activation, is rescued in both DU145 (Fig. 6c) and PC3 cells (Fig. 6d).
FRG1 regulates cytokine expression via p38 MAPK
To validate that FRG1 dictates change in expression of cytokines and MMPs (GM-CSF, PLGF and MMP1 in DU145 and, GM-CSF, PDGFA, MMP1 and CXCL1 in PC3 cells), through p38 MAPK, firstly p38 MAPK was activated in DU145 and PC3 cells and, change in expression of previously mentioned cytokines and MMPs was checked by qRT-PCR. In DU145 cells, activation of the p38 MAPK via Anisomycin treatment (Fig. 7a) increased the expression of GM-CSF and PLGF but MMP1 levels were not altered (Fig. 7c). Similarly, in the PC3 cells, p38 MAPK activation (Fig. 7b) increased the expression of GM-CSF, MMP1 and CXCL1 but not of PDGFA (Fig. 7d).
For further validation of FRG1’s effect on cytokines and MMPs expression, p38 MAPK was inhibited in DU145 and PC3 cells, with or without FRG1 knockdown. Inhibition of p38 levels in DUI45 cells with FRG1 knockdown, (Fig. 8a) led to the reduction of GM-CSF and PLGF levels but MMP1 levels remained unaltered, compared to DUI45 cells with FRG1 knockdown (Fig. 8c). Inhibition of p38 MAPK in PC3 cells with FRG1 knockdown (Fig. 8b), led to the reduced expression of GM-CSF, MMP1, CXCL1 and PDGFA, compared to PC3 cells with FRG1 knockdown (Fig. 8d). Overall, we found that FRG1 expression levels affect expression of GM-CSF and PLGF in DU145 cell via p38 MAPK. MMP1 expression alteration is not mediated via p38 MAPK. Data in PC3 suggests that FRG1 expression alters, expression of GM-CSF, MMP1 and, CXCL1 via p38 MAPK. Effect on PDGFA expression was not consistent.
These findings suggest that FRG1 might dictate cellular processes through specific cytokines and MMPs via modulation of p38 activity in cell specific manner.