Stable transfection of the renal epithelial cell line MDCK with a constitutively active MEK1 construct resulted in a broad range of morphologic phenotypes. Observation of individual clonal populations by phase contrast microscopy indicated that the stable transfectants exhibited morphologic features extending from a conserved epithelial phenotype to a fully transitioned phenotype characterized by an extended and migratory fibroblastic morphology (, panel I). Notably, a number of clones were observed with intermediate features common to both epithelial cells and fibroblasts when observed by phase contrast microscopy. For the purposes of this study, we isolated a panel of clonal populations with morphologic features extending from typical epithelial (clone B) through intermediate phenotypes (clones C and D) to fully fibroblastic phenotypes (clones E and F).
E-cadherin staining and organization were used as qualitative markers of the epithelial phenotype, whereas staining and organization of vimentin was used as a marker for the degree of epithelial–mesenchymal transition. Control MDCK cells (, panel II, A) and the epithelial clone (B) showed maintenance of cell surface staining characteristic of intact E-cadherin junctional complexes. The intermediate clones showed progressive dissolution of the junctional complexes with increased cytoplasmic localization of E-cadherin staining (C and D). E-cadherin distribution was primarily cytoplasmic in the fully fibroblastic clones (E and F).
Staining for vimentin revealed an inverted pattern of expression as the MDCK cells transitioned from the epithelial to mesenchymal phenotypes. Vimentin staining in the controls and the epithelial cell clones (, panel III, A and B) was limited to a delicate filamentous pattern in a subcortical distribution. Vimentin staining progressively increased in intensity as the cells transitioned to the mesenchymal phenotype, with a diffuse cytoplasmic distribution (C and D). In the fully mesenchymal phenotypes, vimentin staining was intense and organized in dense filamentous structures with a predominantly perinuclear concentration (E and F). Vimentin exists as a 54 kDa non-phosphorylated form and as a 57 kDa phosphorylated form (22
). The 57 kDa phosphorylated form regulates intermediate filament assembly and cellular migration (22
). Western blots of the respective clones demonstrated a graded increase in both forms of vimentin as a function of the extent of epithelial–mesenchymal transformation (, panel IV).
The level of measured MEK1 activity, as determined by the rates of phosphorylation of recombinant MAPK1 protein, showed a direct relationship with the extent of epithelial–mesenchymal transformation (, Panel I). Control MDCK cells were assigned a relative MEK1 activity of 100%. Cells with an intermediate phenotype (clones C and D) showed relative MEK1 levels of 160 ± 14% and 210 ± 23%, respectively (P < 0.05 as compared with controls). MDCK clones with fully transformed mesenchymal phenotypes expressed relative MEK1 activities of 260 ± 22% and 280 ± 18%, respectively (P < 0.05 as compared with controls). Thus, relatively small, but sustained and graded increases in MEK1 activity are sufficient to induce graded degrees of epithelial–mesenchymal transformation.
Fig. 2. Quantitative analyses of MEK1 activity, MT1-MMP transcription rates, protein synthesis, MT1-MMP enzymatic activity and invasion. Panel I: MEK1 enzymatic activity was measured in the respective MDCK clones, with a value of 100% assigned to the control (more ...)
There was a similar relationship between the degree of epithelial–mesenchymal transformation and the levels of MT1-MMP transcription (Panel I) and protein synthesis (Panel II). Transcription rates for MT1-MMP as assessed with a luciferase reporter construct driven by the MT1-MMP promoter increased nearly 4-fold in the most transformed clones while the relative levels of MT1-MMP protein also increased by ~4-fold. The dominant 62 kDa MT1-MMP band detected on the western blots conforms with the pro- or inactive form of MT1-MMP. The levels of enzymatically active MT1-MMP are highly regulated by a complex process of proenzyme secretion, membrane complex formation, catalytic activation, internalization, degradation or recycling (24
). We therefore directly quantified the amounts of active MT1-MMP enzyme present in respective MDCK clones. As summarized in , panel III, the epithelial clones (A and B) and intermediate clone (D) had similar levels of MT1-MMP enzyme activity (~1.0 ng/100 μg cellular protein). The fully mesenchymal MDCK clone F had a higher MT1-MMP activity level of ~1.7 ng/100 μg cellular protein. Thus, while MT1-MMP proenzyme protein progressively increases as a function of MEK1 activity, MT1-MMP enzymatic activity does not and is only elevated in the fully invasive MDCK clone.
Acquisition of an invasive phenotype is an important determinant of tumor behavior and ultimately of prognosis in renal cell carcinoma. Although the prior figures describe a gradient of mesenchymal morphology, kinase activity and MT1-MMP transcription/translation across the MDCK/MEK clonal populations, this was not observed with a quantitative in vitro invasion assay. As summarized in , panel IV, clones with intermediate (or mixed epithelial/mesenchymal) phenotypes (C and D) had the same levels of invasive activity observed in the fully epithelial cell clones (A and B). The fully mesenchymal clone F demonstrated a nearly 4-fold increase in invasive activity as compared with the other clones. Thus, acquisition of an invasive phenotype is a feature of MEK/MT1-MMP-dependent epithelial transformation seen only in those cells expressing the highest levels of MEK and higher levels of active MT1-MMP.
Both MEK1 and MT1-MMP enzymatic activity were required for the maintenance of MDCK epithelial–mesenchymal transformation. The fully mesenchymal MDCK clone F, at the 10 passage, was incubated for 48 h in the presence or absence of the selective MEK1 inhibitor, PD98059 (30 μM) and examined by immunofluorescence staining for the MEK1 protein and with Nomarksi optics to define cellular morphology. As shown in , panel I, A, MDCK cells from clone F demonstrate an extended migratory morphology with prominent MEK1 protein staining. Inhibition of MEK1 activity reverts the cellular morphology from mesenchymal to fully epithelial (B). We also quantified the effects of selective MEK1 inhibition by PD98059 on MT1-MMP transcription rates. Incubation for 48 h with PD98059 decreased MT1-MMP relative transcriptional activity to 56 ± 24 relative luciferase units as compared with the control value of 245 ± 58 relative luciferase units (P < 0.05).
Fig. 3. MEK1 and MT1-MMP enzymatic activity are required to maintain MEK1-induced epithelial–mesenchymal transition. Panel I: Mesenchymal clone F was incubated with the selective MEK1 inhibitor PD98059 (30 μM) for 48 h, followed by immunofluoresence (more ...)
As detailed in , panel II, incubation of the fully mesenchymal clone F with a monoclonal antibody directed against the catalytic active site of the MT1-MMP protein induced a reversion to a fully epithelial cell phenotype, as demonstrated by morphology and immunohistochemical staining for E-cadherin and vimentin. Western blot analyses of MT1-MMP antibody-treated cells (panel III) show a progressive decrease in vimentin staining over a 96 h period while E-cadherin increased significantly at the same time point. Incubation with a control monoclonal IgG had no effect on vimentin or E-cadherin expression levels. Thus, sustained expression of both active MEK1 and active MT1-MMP are required for the maintenance of the mesenchymal phenotype. These findings confirm an absolute requirement for the integrity of the MEK1/MT1–MMP axis in maintaining MDCK epithelial–mesenchymal transformation.
We assessed the tumor-forming properties of the MDCK-MEK clones in nu/nu mice. For these studies, mice received a subcutaneous flank injection of MDCK clones with an epithelial phenotype (clone B), an intermediate phenotype (clone D) and the fully mesenchymal invasive clone F. Representative sections are shown in , Panel I. Tumors derived from the epithelial phenotype MDCK clone (panels A–C) were characterized as well-differentiated adenocarcinoma with developed tubulocystic structures and abundant intraluminal mucin. These tumors were relatively acellular within a dense stroma and surrounded by a very dense fibrous capsule. There was no evidence for local invasion or neoangiogenesis. The Fuhrman nuclear grade score for these tumors was 1.2 ± 0.2 (n = 50 scored nuclei).
Fig. 4. Characterization of in vivo tumors generated by MDCK-MEK1 clones. Panel I: Cultures from the epithelial clone B, the intermediate clone D and the fully mesenchymal clone F were suspended in Matrigel and injected subcutaneously into the flanks of athymic (more ...)
Tumors derived from the MDCK clone with an intermediate mesenchymal phenotype (clone D) remained encased in a relatively dense capsule but were considerably more cellular with a Fuhrman nuclear score of 1.8 ± 0.5 (panels D–F, n = 50 scored nuclei). In contrast, tumors derived from the fully mesenchymal MDCK clone F were highly cellular, lacked capsule formation and demonstrated local invasion into the surrounding soft tissues and muscle (panels G–H). The cellular morphology was anaplastic in nature with a Fuhrman nuclear score of 3.6 ± 0.3 (n = 50 scored nuclei).
Immunohistochemical staining of the tumors for MT1-MMP expression was performed (, panel II). There was little to no detectable MT1-MMP cellular expression in the tumors derived from the epithelial MDCK clones (B), whereas cells lining cystic structures were noted to express MT1-MMP in the tumors derived from the intermediate MDCK clone (C). There was intense cellular staining in the tumors derived from the fully mesenchymal MDCK clones, with prominent staining of columns of cells invading muscle and adipose tissue (D).
We next examined the relationship between VHL status, EGFR and MET signaling with rates of MT1-MMP transcription and synthesis using VHL+/+ Caki-1 clear cell carcinoma cells and VHL−/− 786-0 clear cell carcinoma cells. These studies are summarized in . MT1-MMP transcription rates were approximately six times greater in the VHL−/− 786-O cells as compared with the VHL+/+ Caki-1 cells (, Panel 1).
Fig. 5. Relationship between VHL status and MT1-MMP transcription and synthesis—effects of EGFR and MET inhibition. Panel I: VHL+/+ Caki-1 and VHL−/− 786-O clear cell carcinoma cells were transiently transfected with the MT1-MMP luciferase (more ...)
The effects of two signaling inhibitors on MT1-MMP transcription rates were also assessed. In the concentration used (10 nM), K252a is a potent and selective inhibitor of the receptor tyrosine kinase activity of c-MET (27
). The compound 4557w [4-(4-benzyloxyanilino)-6,7-dimethoxyquinazoline] is a potent and selective inhibitor of EGFR tyrosine kinase activity (28
). MT1-MMP transcription rates in both cell types were significantly reduced by either the EGFR or MET chemical inhibitors, indicating that basal transcription of MT1-MMP is primarily mediated by a constitutively active receptor tyrosine kinase-coupled Ras/Raf/MEK/MAPK signaling cascade.
Similar findings were observed in terms of MT1-MMP protein synthesis (, Panel II). VHL+/+ Caki-1 cells synthesized approximately one-fifth the amount of MT1-MMP as compared with the VHL−/− 786-0 cells. MT1-MMP protein synthesis was inhibited by inclusion of the EGFR or MET inhibitors in the culture media. Thus, these experiments link VHL status with Ras/Raf/MEK/MAPK signaling and resultant MT1-MMP transcription and synthesis.
Tissue microarrays of controls and 49 specimens of renal cell carcinoma (42 clear cell type, 2 papillary type, 2 collecting duct type and 3 granular type) were stained for phosphorylated MEK1 and MT1-MMP. Digitized images of each specimen were then quantitatively assessed for the levels of phosphorylated MEK1 and MT1-MMP staining using the ImageJ software package and correlated with the corresponding Fuhrman nuclear grade. Representative sections from the tissue arrays are shown in and show a progressive increase in staining for phosphorylated MEK1 (A–C) and MT1-MMP (D–F) as a function of increasing Fuhrman nuclear grade. summarizes the quantitative assessment of phosphorylated MEK1 and MT1-MMP expression. Spearman correlation coefficients were calculated to examine the association of tumor grade with phosphorylated MEK1 and MT1-MMP. Tumor grade is significantly associated with both phosphorylated MEK1 (r = 0.44, P = 0.002) and MT1-MMP (r = 0.56, P < 0.0001). The non-parametric Kruskal–Wallis test was used to test for differences in phosphorylated MEK1 and MT1-MMP levels across the four tumor grades. This test also revealed a significant difference in protein expression across tumor grades, both for phosphorylated MEK1 (P = 0.005) and MT1-MMP (P = 0.002).
Fig. 6. Representative sections of sections from renal carcinoma tissue microarray stained for phospho-MEK1 and MT1-MMP and quantitative assessment of phosphor-MEK1 and MT1-MMP expression as a function of nuclear grade. Panel I: Tissue microarrays were stained (more ...)
Montesano et al.
) first reported a relationship between high level MEK1 expression, elevated MT1-MMP synthesis and acquisition of an invasive phenotype in three-dimensional culture. In the current study, we have attempted to build on this initial observation and to provide a mechanistic linkage between graded activation of the MAPK/MT1–MMP axis and renal cell carcinoma phenotypic features directly associated with clinical outcomes. The levels of relative MEK1 activity in the MDCK clonal populations were in the same range as reported for human renal cell carcinoma samples (29
), indicating that the observed phenotypes are unlikely to be the result of gross MEK1 overexpression. The functional linkage between both components of the MEK/MT1–MMP axis for the determination of the final cellular phenotype is underscored by the observation that expression of MT1-MMP alone in MDCK cells generates tumor cells that maintain a well-differentiated, fully epithelial non-invasive phenotype (30
Sustained activation of the MEK signaling module undoubtedly alters the expression of numerous genes in addition to MT1-MMP. It is intriguing, however, that inhibition of MT1-MMP enzymatic activity with an antibody directed against the catalytic site is sufficient to revert fibroblastic, fully mesenchymal MDCK cells to a differentiated epithelial phenotype. Thus, both sustained MEK1 activity and MT1-MMP enzymatic activity are required for the development of the fully mesenchymal phenotype, but this phenotype cannot be maintained in the absence of MT1-MMP enzymatic activity.
The relationship between MEK1 and MT1-MMP is bidirectional. A recent study by Sounni et al.
) demonstrated that binding of TIMP2 to cell surface MT1-MMP-stimulated cellular migration via activation of MEK1/2 phosphorylation, a process that occurs independently of MT1-MMP proteolytic activity (32
). Thus, elevated expression of the MT1-MMP proenzyme is sufficient, via MEK1/2 phosphorylation, to induce proliferation and migration, whereas expression of the active enzyme is required for cellular invasion of extracellular matrices. Distinct roles of the catalytic and hemopexin domains of MT1-MMP have been defined in the epithelial–mesenchymal transformation of prostate cancer cells (33
). This underscores the conclusion that MT1-MMP effects on cellular behavior are multilayered and involve both proteolytic and non-proteolytic activities that are intricately linked to activation of the MEK1/ERK signaling cascade.
MT1-MMP plays a critical role in the ability of tumor cells to invade three-dimensional extracellular matrices (14
). In addition, MT1-MMP has been shown to induce aneuploidy and chromosomal instability in model epithelial cells systems (35
). This process may provide an explanation for the association of higher tumor nuclear grade and anaplastic morphology with higher levels of MT1-MMP expression observed in this study.
Induction of MT1-MMP transcription by MEK1 signaling provides at least a partial mechanistic explanation for the efficacy of protein kinase inhibitors for the treatment of renal cell carcinoma (29
). Furthermore, MT1-MMP protein synthesis is regulated by the mammalian target of rapamycin (38
) and the positive treatment results observed in some patients with renal cell carcinoma treated with the mammalian target of rapamycin inhibitor everolimus may be, at least in part, a result of inhibition of MT1-MMP synthesis.
There is considerable interest in the identification and validation of biomarkers for renal cell carcinoma that are either associated with tumor behavior or response to treatment. The level of insulin-like growth factor-1 receptor expression correlates with Fuhrman nuclear grade and the membrane-associated metalloproteinase ADAM has been associated with renal cell cancer progression (40
). Our current findings suggest that expression of phosphorylated MEK1 and MT1-MMP may also provide new biomarkers that are mechanistically linked and represent potential treatment targets.