MET oncogene overexpression has been described in a variety of human cancers including prostate [7
]. Aberrant c-Met activation has been shown to be strongly involved in prostate cancer aggressiveness and poorly clinical outcome [2
]. In the current study human metastatic prostate cancer PC-3 cells were found to overexpress not only c-Met but also HGF at the transcriptional level (Figure
A). Since a high basal level of phosphorylated c-Met is also observed in PC-3 cells, it was anticipated that an HGF/c-Met autocrine loop that induces constitutive c-Met activation exist in this cell line. However, the molecular weight of the secreted “HGF” by PC-3 cells was inconsistent with the recombinant HGF protein (Figure
B). Furthermore, c-Met-associated functions were not activated by CM from PC-3 cells (Figure
), suggesting that what was secreted by these cells was not functional HGF. This conclusion was subsequently supported by evidence indicating that PC-3 cells did not respond to the anti-HGF neutralizing antibody (Figures
and ); a finding that supports the conclusion that the constitutive c-Met activity in PC-3 cells is autocrine-independent.
Two questions arise from the results of the current study. Firstly, what is the “HGF” produced by PC-3 cells and what is its function? Mature HGF/SF is composed of an α-chain (69
kDa) and a β-chain (34
kDa) that are linked to form a heterodimer [33
]. Since the primers are designed to probe the α subunit of HGF mRNA (Figure
A) and a single band (~ 45
kDa) can be detected under non-reducing conditions (Figure
B, left), the secreted protein might be an isoform of HGF [34
]. Secondly, if an autocrine loop is not involved, then what accounts for the constitutive c-Met activation? To date MET gene abnormalities such as activating mutations or amplifications have not been reported in PC-3 cells nor prostate cancer in general [2
], suggesting alterations at the genetic level may not be involved. Since c-Met protein overexpression due to mRNA upregulation occurs predominantly in human cancers [17
], the basal level of phosphorylated c-Met in PC-3 cells may simply be a result of increased MET transcripts via unknown mechanisms. In addition, the cross-talk between c-Met and other signaling molecules post-transcriptionally could be a possibility given that c-Met is able to be transactivated by several other transmembrane proteins [35
]. In the PC-3 cell line, basal c-Met phosphorylation remained unaffected by exposure to either gefitinib or dasatinib (data not shown), suggesting that c-Met is not activated by epidermal growth factor receptor (EGFR) or c-Src, two kinases shown to be involved in c-Met transactivation in some studies [36
]. However other signaling molecules such as Ron, another Met receptor family member which is also overexpressed in PC-3 cells [38
], might transactivate c-Met. Finally, an HGF-mediated “intracellular” autocrine mechanism [25
], although rare, could be another possibility.
Despite the unresponsiveness of PC-3 cells to anti-HGF antibody, the Met kinase inhibitor BMS-777607 did significantly inhibit PC-3 cell proliferation, clonogenicity, migration and invasion (Figure
) as well as c-Met signaling pathways (Figure
). Coupled with our previous findings [15
], these results suggest that in the PC-3 tumor model, c-Met signaling plays a major role in the metastasis-related behavior irrespective of the HGF status. Consistent with the impact on cellular functions, BMS-777607 also significantly ablated molecular c-Met activity and downstream pathways including c-Src/FAK and Akt-mTOR, indicating that c-Src and Akt are two mediators of constitutive c-Met signaling. Interestingly, exogenous HGF cannot phosphorylate c-Src in PC-3 cells, suggesting that c-Src does not mediate HGF-induced c-Met activation [15
]. The discrepant role of c-Src in c-Met-mediated molecular events reveals the complex interplay between these signaling components.
PC-3 cells were originally isolated from a prostate cancer bone metastasis [39
]. Since HGF is enriched in the stroma of both the prostatic gland [40
] and bone marrow [41
] and is considered to be sufficient to trigger c-Met activation, acquisition of the c-Met activity in the absence of environmental HGF may facilitate tumor cells to survive and metastasize in a scenario where exogenous HGF is lacking. Anchorage independence is suggested as a factor in the survival of circulating tumor cells [42
], but our data indicate that c-Met is not essential for maintaining anchorage-independent cell survival (Figure
F). Thus while targeting c-Met kinase is unlikely to reduce viability of non-adherent tumor cells, small molecule Met kinase inhibitors may have significant therapeutic potential as agents that interfere with the metastatic phenotypes associated with c-Met.