RAS proteins (comprising Ha-RAS, Ki-RAS and N-RAS) are very important molecular switches for a wide variety of signal pathways that control proliferation, cell adhesion, apoptosis and cell migration. RAS proteins are often deregulated in cancer, leading to increased invasion and metastasis and decreased apoptosis. Mutations in the RAS family of proto-oncogenes are very common, found in 30% of all human tumors and in 50% of colon tumors in particular.1 The most common mutations are found on residues 12 and 61. The glycine to valine mutation on residue 12 renders RAS insensitive to inactivation. RAS protein functions within a signal transducing cascade of reactions. Among them, the mitogen activated protein (MAP) kinases that transmit signals downstream to other protein kinases and gene regulatory proteins, are of leading importance.2
Epithelial to mesenchymal transition (EMT) is a highly conserved and fundamental process that not only governs morphogenesis but also cancer invasion and metastasis in multicellular organisms. There is good evidence that EMT gives rise to the dissemination of single carcinoma cells from the site of primary tumors. In addition, increasing evidence suggests that EMT could play a specific role in the migration of cells from a primary tumor into the blood circulation. EMT can require the cooperation of oncogenic RAS or tyrosine kinase receptor, with endogenous signaling molecules. It involves the transition from an epithelial to a fibroblastic or mesenchymal cell phenotype, accompanied by a large number of changes in gene expression.3 EMT appears as a loss of epithelial polarity accompanied by an increase of cell motility, repression of the epithelial markers E-cadherin and cytokeratins and activation of the mesenchymal marker vimentin. Moreover, reorganization of actin-based cytoskeleton and increase of nuclear β-catenin has also been observed. The above characteristics define the so-called “complete” EMT, and this is correlated with invasion and metastasis. However, many subtypes of EMT exist; they induce a mesenchymal phenotype to different extents by enhancing or repressing the cell's epithelial gene expression program.
E-cadherin, an invasion suppressor protein, has emerged as a critical protein driving EMT. A connection between E-cadherin, activating protein-1 (AP-1) and EMT has been reported. In particular, its absence has been associated with increased amounts of the AP-1 components c-JUN and FRA-1.4 Therefore, increasing evidence suggests that EMT could play a specific role in the migration of cells from a primary tumor into the blood circulation.
A hallmark of EMT is the upregulation of vimentin, a developmentally regulated intermediate filament protein (IFP).5 Although vimentin does not seem to be essential for growth, cell division or development, it does contribute to the metastatic potential of melanoma and mammary tumors.6 Cells in the centre of a colorectal cancer maintain an epithelial phenotype, whereas cells at the invasive front exhibit a mesenchymal phenotype characterized by a loss of cell–cell contacts and increased expression of fibronectin and vimentin.7 Reprogramming of intermediate filament expression leading to the production of vimentin, either alone or in combination with specific keratins, promotes tumor cell invasion.8 Upregulation of the vimentin gene is a marker of malignant progression in breast cancers and it has been shown to be regulated by the dual AP-1 site within its promoter.9
AP-1 transcription factors consist of homodimers and heterodimers of JUN (c-JUN, JUNB, JUND), FOS (c-FOS, FOSB, FRA-1, FRA-2) and the related activating transcription factor (ATF2, ATF3/LRF1, B-ATF) subfamilies.10 Studies have shown that AP-1 activity is increased in transformed cell lines. Oncogenic RAS proteins regulate AP-1 activity at the transcriptional and posttranslational level.11 Multiple approaches such as overexpression or knockout of specific components, dominant negative derivatives or antisense expression vectors and generation of knockout or transgenic mice have established an oncogenic role for AP-1.12, 13 AP-1 can also promote invasion and the transition of tumor cells from an epithelial to a mesenchymal morphology, which is one of the early steps in tumor metastasis.14 Components of AP-1 have also been shown to stimulate EMT.15 These oncogenic properties of AP-1 are primarily dictated by the dimer composition of the AP-1 family proteins and their posttranscriptional and posttranslational modifications.
The mechanism by which oncogenic RAS alone contributes to EMT in colon cancer is largely unknown. To this end, we have previously developed a model system where the colorectal adenocarcinoma cell line Caco-2 has been stably transfected with the Ha-RASV12 or Ki-RASV12 mutated oncogene. Ha-RasV12 overexpression gave rise to an in vitro human colon cell system for studying EMT.16 In our study, we demonstrated that vimentin is strongly related to EMT and that during this process, vimentin expression is specifically regulated by the AP-1 transcription factor FRA-1. Therefore, a potential mechanism for Ha-RAS-dependent EMT in colon cells was reported in our study, involving changes in RAS-induced gene expression important for this process.
Material and methods
Cell lines and culture conditions
The Caco-2 cells constitutively overexpressing active RAS proteins were described in Ref.16. All cells, including SW620, HCT116, DLD-1 and HT29, were cultured in a humidified atmosphere at 37°C with 5% CO2. Cell lines were grown routinely in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Invitrogen Life Technologies, Carlsbad, CA), 1% penicillin/streptomycin (Invitrogen Life Technologies, Carlsbad, CA) and non essential amino acids.
For the inhibitor studies, the MEK inhibitor UO126 (LC Laboratories) and the PI3K inhibitor Wortmannin (Sigma) were added directly to the cells at a concentration and for a time period mentioned in the corresponding figures.
Nuclear and total cell extract preparation
For preparation of total cell extracts, cells were grown in 10-cm Petri dishes, washed 3 times in ice cold phosphate-buffered saline (PBS), scraped in ice cold lysis buffer (50 mM Tris–HCl pH 7.4, 250 mM sucrose, 1 mM EDTA, 10 mM NaF, 1 mM EGTA, 1% Triton X-100 plus protease inhibitors) and left in ice for 30 min. After centrifugation the supernatant was collected and aliquoted.
Nuclear extracts for electrophoresis mobility shift assays (EMSA) and western blot analysis were prepared from 15-cm Petri dishes. Cells were lysed by swelling in hypotonic buffer (10 mM Hepes pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA) as described previously.17 Nuclear proteins were finally solubilized in a buffer comprising 20 mM Hepes pH 7.9, 0.4M NaCl, 1 mM EDTA, 1 mM EGTA. Protein concentration was determined by the Bradford method (Bio-Rad Laboratories, Hercules, CA) using bovine serum albumin as a standard.
For alkaline phosphatase treatment, nuclear extracts were incubated for 2 hr at 37°C in the presence or absence of 20 U of the enzyme and were immunoblotted with an appropriate antibody.
Western blot analysis
Western blot analysis was performed using conventional methods. Extracts were resolved on SDS-PAGE (10% w/v acrylamide), transferred to nitrocellulose membranes (BioTrace NT; Pall LifeSciences, NY) followed by immunoblotting with the following antibodies:
c-JUN (sc-1694), JUNB (sc-8051), JUND (sc-74), c-FOS (sc-7202), FRA-1 (sc-605), FRA-2 (sc-171), phosphor ERK 1/2 (sc-7383), p-c-JUN (sc-822), vimentin (sc-6260), E-cadherin (sc-7870), Ha-RAS (sc-29) are from Santa Cruz Biotechnology, Santa Cruz, CA; phosphor c-JUN N-terminal kinase (JNK) (Thr 183/Tyr 185;#9251), phosphor-Akt (Ser473; #9271) are from Cell Signalling Technology, Danvers, MA. The bands were visualized with horseradish peroxidase conjugated anti-rabbit IgG or an anti-mouse IgG (Jackson Immunoresearch Laboratories, West Grove, PA) using the ECL (enhanced chemiluminescence) detection system (Amersham Biosciences, Buckinghamshire, UK).
Reverse transcriptase PCR
Total RNA was isolated by cell lysis in TRIzol Reagent (Invitrogen Life Technologies, Carlsbad, CA). First-strand cDNA synthesis was performed using the Superscript RNase H− reverse transcriptase (Invitrogen Life Technologies, Carlsbad, CA), following the manufacturer's instructions. Briefly, RT reactions (1 hr) were performed at 42°C using 3.5 μg RNA, oligo (15)-dT primers (1.25 μg) and 500 units Superscript RNase H− reverse transcriptase in a total volume of 50 μl. PCRs were also performed in a total volume of 50 μl, comprising 5 μl cDNA product, 0.24 mM deoxynucleoside triphosphate, 20 pmol of each primer, 2.5 mM MgCl2+ and 1 unit Taq DNA polymerase (Promega Life Science, Madison, WI). A 612 bp cDNA fragment from the human vimentin cDNA was amplified by using the oligonucleotide primers (annealing temperature: 67°C for 1 min, 28 cycles):
The sequence of the primers used to amplify the human FRA-1 gene (annealing temperature: 57°C for 1 min, 28 cycles) was the following:
As an internal control, a 126-bp cDNA fragment of the human glyceraldehyde phosphate dehydrogenase (GAPDH) cDNA was amplified (annealing temperature: 57°C for 1 min, 28 cycles) by using the primers
PCR products were separated on 1.5% agarose gels containing ethidium bromide (0.5 μg/ml) and quantified by digital imaging (ImageQuant, Molecular Dynamics, Sunnyvale, CA).
The assay was carried out as described previously.12 The oligonucleotides used was the following:
The first pair of primers corresponds to the AP-1 consensus sequence from the human Collagenase promoter (position −73 to −65),18 whereas the second one recognizes the 2 AP-1 sites of the human vimentin promoter.19, 20 Hybridization of complementary oligonucleotides (equimolar ratio) was performed by heating the mixture at 85°C for 5 min and allowing it to slowly cool down to room temperature. The annealed oligonucleotides were end labeled with [γ-32P] ATP (Amersham Biosciences, Buckinghamshire, UK) using T4 DNA polynucleotide kinase (New England Biolabs, Ipswich, MA) and were purified using the Bio-Spin 6 chromatography columns (Bio-Rad Laboratories, Hercules, CA).
Binding reactions were performed in a final volume of 20 μl containing 20% glycerol, 50 mM Hepes pH 8.0, 500 mM NaCl, 1 mM EDTA, 2.5 mM dithiothreitol, 0.5 mg/ml bovine serum albumin, 2 μg of poly di-dc and 10 μg of nuclear extract. The probe (10 μCi/μl) was then added and the mixture was incubated at 4°C for 30 min. For competition, 1 μl of non labeled probe was incubated with nuclear extracts for 30 min at room temperature before the addition of the labeled probe.
For supershift assays, 1 μg of the corresponding antibody was added and further incubated for 30 min at room temperature. The complexes were resolved on 6% non-denaturing polyacrylamide gels, dried and analyzed using Phosphoimager Storm 820 and ImageQuant Software (Amersham Biosciences, Buckinghamshire, UK).
Oligonucleotides were obtained from Dharmacon (Chicago, IL). Twenty-four hours before transfection, cells were plated in a 6-well plate for mRNA and total protein extraction. Transfection of siRNA duplexes, in a concentration of 10 pmol per well for FRA-121 and 20 pmol for vimentin, was carried out using the calcium phosphate method with BBS transfection buffer as standard protocols state. FRA-1 si-RNA-treated cells were harvested 48 hr after transfection whereas vimentin si-RNA-treated cells were harvested after an incubation time of 72 hr, and both were used for subsequent analysis.
si-RNA targeting vimentin:
Vimentin 1—sense sequence: UCACGAUGACCUUGAAUAAUU
Vimentin 2—sense sequence: GAGGGAAACUAAUCUGGAUUU
Chromatin immunoprecipitation assay
Subconfluent cells were cross-linked with 1% formaldehyde for 15 min at room temperature, washed and scraped in ice cold PBS containing proteinase inhibitors. After lysis, the DNA was sheared into 200–1000-bp fragments by sonication (11 pulses of 10 sec each). Complexes were immunoprecipitated with specific antibodies (4 μg) and then assessed for binding to the AP-1 sites of the vimentin promoter by PCR (annealing temperature: 62°C for 45 sec, 30 cycles). Primers had the following sequence:
Vimentin AP-1 (Fw): 5′-CCGCTAGGAGCCCTCAATC-3′
Vimentin AP-1 (Rv): 5′-TTCGGACGGCGGGAGTTG-3′
The PCR products were analyzed on 1.5% agarose gels in the presence of ethidium bromide.
Cells cultured as monolayers on 48-well plates were fixed with methanol/acetone (4:1) for 10 min at −20°C. After intermediate washes in PBS, cells were blocked with 5% fetal calf serum in PBS for 1 hr at room temperature. Fixed cells were then incubated with either a vimentin monoclonal antibody (1:200 dilution; Santa Cruz Biotechnologies, sc-6260) or a FRA-1 polyclonal antibody (1:200 dilution; Santa Cruz Biotechnologies, sc-605) overnight at 4°C in a humidified environment. After several PBS washes, cells were incubated with appropriate fluorescent secondary antibodies (Jackson Immunoresearch Laboratories, West Grove, PA) at room temperature for 1 hr. After antibody incubation, nuclei were labeled with Hoechst for 30 min. Cells were washed in PBS and observed under a microscope.
Cell migration assay
Cell migration assays were carried out using 24-well transwell plates containing 8-μm pore size polycarbonate filters (Corning Costar, Cambridge, MA). Equal numbers of cells (1 × 104), suspended in 1% serum medium, were plated on the top chamber and incubated for 28 hr at 37°C. Using a serum gradient as a chemo attractant cells were allowed to migrate through the pores towards the bottom chamber for 28 hr. Cells that did not migrate through the pores were removed using a cotton swab. Migrated cells were fixed with methanol for 5 min, stained with 1% crystal violet for 20 min and counted by bright-field microscopy (10–13 random fields/membrane) by 2 independent investigators.
Caco-H1 cells plated on 12-well plates were treated with FRA-1 si-RNA. After 48 hr, cells were fixed with methanol, stained with 0.5% crystal violet, washed with PBS and the remaining crystal violet was extracted by using 30% acetic acid. Absorbance was measured at 595 nm. The results are presented as an average of 3 experiments and a standard deviation function was used for error bar generation.
Oncogenic Ha-RASV12-induced EMT by increasing vimentin expression through the MAPK pathway
To study the effect of different RAS isoforms during colon carcinogenesis, the colorectal adenocarcinoma Caco-2 cell line was stably transfected with a mutated, constantly activated Ha-RAS or Ki-RAS oncogene.16 A set of 2 transformed Ha-RasV12 clones [designated as Caco-H1 (H1) or Caco-H2 (H2)] and 2 Ki-RasV12 clones [Caco-K6 (K6) or Caco-K15 (K15)] was selected for further analysis. The selection of the 2 couples of clones was based on their comparable Ras activity.16 Immunofluoresence analysis showed that vimentin was uniformly expressed throughout the cell only in Caco-H1 and -H2 cell lines, but not in Caco-K15. E-cadherin, a protein mainly associated with epithelial cells, followed an opposite pattern of expression with higher amounts in Caco-K15 cell lines (Fig. 1ai). In addition, vimentin mRNA (Fig. 1aii) and protein levels (Fig. 1aiii) were strongly enhanced in Caco-H1 and Caco-H2 cells as opposed to parental cells. Low vimentin expression was observed in Caco-K15 cell line (Fig. 1a). The increase of vimentin protein levels was confirmed in Caco-2 clones overexpressing low levels of mutated Ha-Ras protein. Notably, in cells containing comparably low amounts of mutated Ha-Ras or Ki-Ras protein, only the presence of the Harvey isoform resulted in an enhancement of vimentin levels (data not shown). E-cadherin protein levels followed an opposite pattern of regulation, with higher amounts observed in control Caco-2 cells as compared to Caco-H1 and -H2 cell lines. Thus, vimentin and E-cadherin protein levels confirmed that the particular phenotype, observed in Caco-H1 and -H2 cell lines, was due to an EMT.
Interestingly enough, quite a large pool of vimentin appeared in the nucleus of Caco-H1 and -H2 cell lines, despite the fact that it is usually referred to as a cytoplasmic protein (Fig. 1ai). Western blots with cytoplasmic extracts of RAS clones showed similar increase of vimentin in Caco-H1 and Caco-H2 cell lines and the amount was comparable to that in the nucleus (Fig. 1aiii). To exclude the possibility of contamination of nuclear with cytoplasmic extracts, nitrocellulose membranes were blotted with the cytoplasmic protein Ha-RAS. Although this protein was seen in the cytoplasmic fraction, none was observed in the nucleus (Fig. 1aiii).
The chemical inhibitors of MEK (UO126) and PI3K (wortmannin) were used to determine whether MAPK pathways were involved in the Ha-RAS mediated induction of vimentin. The specificity of the MAPK inhibitors used was confirmed by western blot analysis. Vimentin nuclear protein and mRNA decreased to half upon treatment of cells with UO126 compared to cells treated only with DMSO (Fig. 1b), indicating a MEK-related transcriptional regulation mechanism. The PI3K had a smaller effect on vimentin expression with a reduction of only 20% compared to Caco-2 cells (Fig. 1b). In addition, c-FOS protein, which is well documented to be regulated by this pathway, was reduced by almost 40% upon inhibitor treatment.
It is apparent that determining what controls vimentin expression in these cells could lead to an understanding of what contributes to changes in gene expression that occur during transformation to the malignant state.
EMT induction by increased FRA-1 and JUNB binding to the vimentin AP-1 TRE upon Ha-RASV12 overexpression
The in vitro protein binding to the vimentin-promoter AP-1 sites, known to be involved in vimentin expression, was studied. For this, we used one 32P-labeled probe (vimentin) containing the vimentin promoter TRE (TRE; TPA-responsive element) composed of 2 AP-1 sites and a second probe (TRE Col II) containing the TRE of the human collagenase II promoter. When nuclear extracts were added, a major protein complex (marked with an arrow) was formed around the vimentin probe. The binding was shown to be specific, because it disappeared in a competition experiment performed in the presence of an excess of vimentin non-labeled probe (cold vimentin). In agreement, the TRE Col II probe containing an AP-1 site competed with vimentin probe for protein binding, whereas a non-specific probe (cold Ets 1.3) did not affect the intensity of the signal (Figs. 2a and 2b).
EMSA with vimentin tandem 32P-labeled AP-1 elements displayed considerably more binding with nuclear extracts from Caco-H1 and Caco-H2 transformed cells compared to Caco-2 control or Caco-K15 and Caco-K6 cells (Fig. 2a, left panel). A similar pattern of binding was observed when the TRE Col II probe was used (Fig. 2b, left panel).
To study the composition of the complex formed around the AP-1 site of vimentin in Caco-H1 clones, we performed gel supershift analysis using antibodies recognizing individual members of the JUN and FOS protein families. Supershifted complex was observed with JUNB and FRA-1 specific antibodies, whereas a weaker supershift was observed with the JUND antibody. Similarly, the complex formed around the collagenase II TRE in Caco-H1 cells, contained both FRA-1 and JUND proteins (Fig. 2b, right panel). No binding of JUNB was detected on this probe. Together, these results suggest that JUNB and FRA-1 are the major components of the AP-1 complex of the vimentin promoter in Caco-H1 cells (Fig. 2a, right panel).
To verify the interaction of specific AP-1 components within the native environment, we performed chromatin immunoprecipitation (ChIP) of in vivo cross-linked extracts from Caco-H1, Caco-H2 and Caco-K15 cell lines (H1, H2 and K15, respectively) followed by PCR with primers encompassing the vimentin TRE region (Fig. 3a). As a positive control, the particular genomic region was amplified by PCR using DNA from DLD-1 and HT-29 adenocarcinoma cell lines (data not shown). The results of the ChIP when using the FRA-1 and JUND antibodies showed a considerable amount of amplified fragment in the Caco-H1 and Caco-H2 chromatin immunoprecipitates, compared to Caco-K15. Fold changes relative to the parental Caco-2 cells are depicted below each corresponding picture. The binding of other transcription factors on the TRE promoter of vimentin, also presented small changes when compared between the different RAS isoforms. (Fig. 3b). Finally, the FRA-2 transcription factor was shown not to bind to the TRE site of the vimentin promoter.
To investigate the activation status of chromatin within the analyzed region, we also performed ChIP with the antibody recognizing the acetylated isoform of histone H3 (α-AcH3; Upstate). Chromatin immunoprecipitation with anti-acetyl histone H3 antibody showed that Lys-9 and/or Lys-14 of histone H3 in nucleosomes were hyperacetylated at the vimentin promoter after Ha-RASV12 overexpression but not in the case of Ki-RASV12 overexpression, where a decrease can be observed (Fig. 3b).
FRA-1 and JUN protein levels increased upon Ha-RASV12 transformation of adenocarcinoma cells
To determine whether MAPK pathways were activated during Ha-RAS-induced EMT, Caco-H1, -H2 and Caco-K6, -K15 cells were analyzed and MAPK activation was assessed by Western blot analysis using phospho-specific antibodies. The results demonstrated that ERK, JNK and PI3K were activated by Ha-RAS. All were strongly enhanced in Caco-H1 and Caco-H2 cell lines as opposed to the adenocarcinoma Caco-2 cells. In contrast, Ki-RASV12 overexpression was less effective in activating the pathways (Fig. 4a).
As the transcription factor AP-1 is a known primary target of MAPK pathways, we determined the protein levels of the individual members composing the AP-1 factor, upon overexpression of different RAS isoforms. As Figure 4a shows, all JUN and FOS family members were upregulated in Caco-H1 and Caco-H2 cell lines compared to Caco-2. No detectable change was observed in Caco-K15 cells. Among the JUN family members, JUNB exhibited the highest and c-JUN the lowest fold change. The protein levels of the FOS family member FRA-1, showed the highest upregulation among all the members of the AP-1 transcription factor upon Ha-RASV12 transformation (16.1-fold and 16.8-fold in Caco-H1 and Caco-H2 cell lines, respectively) whereas no change was observed in Caco-K15 cells. A different pattern from all the factors mentioned above was observed with c-FOS protein. Upregulation of c-FOS protein (˜6-fold) was observed in both cells stably transfected with the Ha-RASV12 or the Ki-RASV12 isoforms (Fig. 4a).
To further investigate the role of FRA-1 and the reason of its high upregulation upon Ha-RASV12 transformation, we examined the protein levels of FRA-1 and vimentin in some well characterized colon carcinoma cell lines. Interestingly enough, vimentin and FRA-1 showed a similar expression pattern (Fig. 4b). Both vimentin and FRA-1 were highly overexpressed in HCT116 and SW620 cells, whereas their expression levels were significantly lower in HT29 or DLD-1 cells. It is well known that SW620 is a cell line that has gained EMT characteristics.
Because FRA-1 was mainly involved in binding to the AP-1 site of the vimentin promoter and its protein levels were highly induced upon Ha-RASV12 overexpression, we focused on this factor. FRA-1 mRNA levels were increased by 2.6- and 5-fold in Caco-H1 and Caco-H2 cell lines respectively, as compared to Caco-2. The pattern of FRA-1 regulation was similar to the one observed for vimentin, even though Caco-2 cells have a basal amount of this factor (Fig. 4c & Fig. 1a(ii)).
The signaling pathways regulating FRA-1 were common to those involved in vimentin regulation. Specifically, the MAPK–ERK pathway exerted a strong effect on FRA-1 expression. A complete disappearance of its mRNA and reduction to half of its protein levels upon cell treatment with UO126 (a MEK inhibitor) was observed. The PI3K had a smaller effect as inhibitor studies showed. A 30% reduction was also observed for JUND protein upon UO126 treatment but no effect at all was to be observed using the PI3K inhibitor (Figs. 4c and 4d).
The signal of FRA-1 in western blot analysis of Caco-H1 and –H2 cell extracts appeared as 3 distinct bands. Treatment of nuclear extracts with alkaline phosphatase resulted in the elimination of the slower migrating FRA-1 band and the appearance of a faster migrating band on the blot for both DLD-1 (an adenocarcinoma cell line containing a RAS mutation) and Caco-H1 transformed cells (Fig. 4e). No effect was observed in Caco-K15 transformed cells (Fig. 4e middle panel). Therefore in our system, consistent with the findings from other laboratories,22 FRA-1 is phosphorylated.
Silencing of FRA-1 by siRNA in Ha-RASV12 transformed cells reduced vimentin expression and attenuated cell migration
To investigate the potential role of FRA-1 on vimentin, we silenced the expression of FRA-1. To selectively knock down the function of FRA-1 in Caco-H1 cells, we took advantage of RNA interference using a siRNA duplex derived from the endogenous mRNA. The siCONTROL RISC-Free siRNA (Dharmacon, Chicago, IL) was used as a control.
mRNA and total protein levels of FRA-1 were decreased ˜80%, 48 hr after transfection of Caco-H1 cells with the siRNA as opposed to control. Inhibition of FRA-1 resulted in reduction of both vimentin mRNA (40%) and protein (60%) levels, compared to the control siRNA. Expression of c-FOS and Actin, which are not modulated by FRA-1, were unaffected demonstrating the specificity of the above-described result. Moreover, the protein levels of 2 known targets of FRA-1 were impaired to different extent upon its inhibition. Matrix Metalloproteinase-9 (MMP-9) expression, which is highly dependent on FRA-1, was reduced by 55%. On the other hand, transfection of FRA-1 siRNA decreased cyclin D1 protein levels by 30% (Fig. 5a).
The effect of FRA-1 knockdown on cell migration was evaluated in transwell migration assays. Transfection of Caco-H1 cells with the siCONTROL siRNA produced insignificant effect on cell migration (Fig. 5b). However, reduction of FRA-1 expression levels resulted in cells migrating slower than cells transfected with the control siRNA (55% less migrating cells). To exclude any involvement of a decrease in cell proliferation on the reduction of cell migration after the FRA-1 siRNA treatment, we performed a proliferation assay. The results of this experiment showed that the proliferation rate of Caco-H1 cells was slightly affected (10% decrease in the number of cells treated with the siRNAControl and 14% decrease for FRA-1 siRNA treated cells) upon FRA-1 siRNA treatment (Fig. 5c). Therefore, enhanced expression of FRA-1 was critically important to maintain enhanced motility of Caco-H1 cells (Fig. 5b). To firmly link vimentin to cell migration, we downregulated vimentin expression by siRNA (Fig. 5e). Caco-H1 siRNA-treated cells exhibited a slower migration capacity (25%) in comparison to the siRNA control indicating that vimentin is indeed involved, to some extent, in cell migration. Light microscope images demonstrated the distinct morphologies presented by the cells upon inhibition of FRA-1. Caco-H1 cells transfected with the control siRNA, like the non-transfected ones, have a fibroblastic elongated appearance. Elimination of FRA-1 expression resulted in cells with a more flattened epithelial shape, which grew overlapping one another (Fig. 5d). Vimentin immunofluorescence analysis of Caco-H1 cells exhibited that vimentin was found around the nucleus, forming long protrusions towards the cytoplasm. FRA-1 siRNA resulted in the disappearance of these protrusions and the relocation of vimentin around the nucleus (Fig. 5d).
Ha-RASV12 strongly enhanced expression of the EMT marker vimentin by increasing FRA-1 binding on AP-1 sites of its promoter
Human tumors frequently exhibit constitutively activated RAS signaling, which contributes to the malignant phenotype. Evidence suggests unique roles of the RAS family members in normal and pathological conditions. In an effort to dissect distinct RAS isoform specific functions in malignant phenotypic changes we previously established Ha-RAS and Ki-RAS activated Caco-2 human colon epithelial cell lines.16
Ha-RasV12 overexpression in Caco-2 cells increased their vimentin expression both at the mRNA and protein levels and enhanced their migratory abilities in Boyden Chamber assays. E-cadherin expression followed an opposite path. These results do not present a definitive demonstration that EMT is occurring. However, our findings conclude that there is indeed a complete repression of E-cadherin, as observed via immunocytochemistry and expression studies. Also, vimentin immunostaining showed a significant increase in its expression. Elevated expression of vimentin is frequently associated with an epithelial to mesenchymal transition and the metastatic conversion of epithelial cells.23 The increased migration capacity of Caco-H1 cells further supported the mesenchymal phenotype exhibited by these cells. By contrast, no migration was evident in the parental cell line (Fig. 5c). These data biochemically support the idea that Ha-RAS specifically promotes EMT in colon cells.
In support of our findings, microarray analysis using the Atlas Cancer cDNA Array (Clontech, Palo Alto, CA) and the RZPD Oncochip nylon array (RZPD, Heidelberg, Germany) to identify RAS regulated genes, showed significant vimentin upregulation in Caco-H1 and -H2 cell lines relative to that in controls, and that these cells acquired a mesenchymal phenotype. At the same time, epithelial marker genes were downregulated. In these clones, mesenchymal changes in cell morphology, anchorage independent growth advantage and tumorigenicity in SCID mice are observed.16 It has been reported that in mammary epithelial cells another activated mutant member of the RAS subfamily, M-Ras, increased vimentin expression and resulted in EMT.24 It appears that Ha-RASV12 has a higher transforming potential as opposed to its isoform Ki-RASV12. The unique gene expression profiles generated by each of the 2 RAS isoforms probably are responsible for these differences.16
Vimentin is not just a simple EMT marker. Evidence exist that vimentin is increased in tumors from patients with colorectal cancer.25 For the first time, we provide evidence that vimentin is a protein differentially regulated between Ha-RAS and Ki-RAS, leading to Ha-RAS specific induction of migrative phenotype and eventually EMT. Vimentin regulation seems to be partly done through the MAPK–ERK pathway. An understanding of the mechanism of vimentin gene regulation could ultimately contribute to controlling the invasiveness of some tumor cells. Vimentin contains 2 tandem AP-1 sites within its promoter important for its serum inducibility.19 We examined the possibility that oncogenic Ha-RAS could exert its effect on vimentin through the AP-1 transcription factor. We show, both in vitro and in vivo, that Ha-RASV12 increases protein binding to these 2 sites (Figs. 2 and 3). FRA-1 and JUND are the main AP-1 components occupying the vimentin promoter at these sites in vivo as chromatin immunoprecipitation experiments showed. This binding may also be responsible for the increased levels of vimentin mRNA seen only in the presence of Ha-RASV12. In agreement with these findings, these 2 proteins are the main components of the AP-1 transcription factor more strongly upregulated upon Ha-RAS overexpression.
AP-1 has been shown to directly contribute to the differential regulation of the human vimentin promoter in invasive, vimentin positive cells, versus non-invasive, vimentin negative cells.26, 27 Along this line, it was shown previously that stable transformation of the MCF7 cell line with the AP-1 factor c-JUN enhanced the expression of vimentin gene20 and a potential mechanism has been suggested.9 However, in our system FRA-1 appeared to mainly contribute to vimentin expression. This is not surprising because it is well documented that the influence of an AP-1 protein on a promoter depends on the dimer partners, and functions in a tissue specific way.13, 28
Of special interest was the finding that considerable amount of vimentin, which belongs to the intermediate filament (IF) family, could be found in the nucleus of Caco-H1 and -H2 cells. The conventional textbook view of IF has been one of a static cytoskeleton that provides the cell with a mechanism for resisting mechanical stress and deformation.29 Expression of the Ha-RAS oncogene in NIH-3T3 fibroblasts resulted in the retraction of vimentin filaments to the perinuclear region.30 In addition, vimentin has been referred to as a nuclear pore complex associated protein31 and contains a nucleic acid binding region, which is located in its non-α-helical binding region.32 Recent studies, based on a variety of different cell types, have revealed that IFs are remarkably dynamic and exhibit a complex array of motile activities related to their subcellular assembly and organization.33 Vimentin interacts directly with several organelles, membrane or cytoskeleton associated proteins as well as cellular compartments for which linker proteins remain to be defined.5 Vimentin could also link extracellular and cytoplasmic events to the nucleus through interaction with a multitude of bindings partners.34, 35 Therefore, the assumption that vimentin could be transported into the nucleus is feasible. However, this hypothesis will be tested in the future.
FRA-1 inhibition reduced vimentin levels and cell migration
To assess a direct role of FRA-1 on vimentin regulation, we employed RNA interference to block the elevated expression of FRA-1 in Caco-H1 cells. By transfecting FRA-1 specific siRNAs, we inhibited expression of FRA-1. Interestingly, inhibition of FRA-1 expression by RNAi strategies in Caco-H1 cells decreased vimentin expression. Therefore, a direct effect of the AP-1 component FRA-1 in colon epithelial cells that over express Ha-RasV12 could be suggested. We have to mention here that vimentin expression was not totally attenuated despite the disappearance of FRA-1. This could indicate that other levels of vimentin regulation could exist. A simple explanation could be a different AP-1 dimer formation, capable of keeping basal levels of vimentin in the cells. Alternatively, vimentin promoter contains several other regulatory elements capable of stimulating its expression (Fig. 3a). A member of the ets transcription factor family (PEA3) has been shown to contribute to the differential regulation of the human vimentin promoter in invasive vimentin positive cells versus non-invasive, vimentin negative cells.36
The role of FRA-1 on vimentin expression could also be verified by indirect observations. First, we showed that FRA-1 is the AP-1 component more strongly enhanced upon Ha-RASV12 overexpression. Elevated FRA-1 levels upon RAS overexpression were observed in a wide variety of tissue cell types.11, 21, 37–39 Ha-RASV12 exerted its effect on FRA-1 mainly through the MAPK–ERK pathway, exactly as for vimentin, even though the PI3K pathway could also play a role (Fig. 4d). Accordingly, it was recently reported that Akt could induce transcription of FRA-1 through either phosphorylation or heterodimerization of Sp1 with the retinoblastoma protein in prostate cancer cells.40 FRA-1 accumulation could be achieved either by prevention of its proteosomal degradation21 or its transcriptional autoregulation and posttranslational stabilization.37
FRA-1 reduction appeared to reverse the mesenchymal phenotype observed in Caco-H1 and -H2 cells as it can be visualized by the morphological appearance of the cells and the elimination of their migrational ability after treatment with FRA-1 siRNA. Vimentin protein was restricted around the cell periphery, in a smooth appearance, with loss of the characteristic long protrusions upon FRA-1 elimination. Vimentin relocation may be a reason for the decreased motility of Caco-H1 cells. However, we cannot exclude the possibility that migration is also affected by one of the genes being regulated by FRA-1. Our results therefore implicate FRA-1 in the regulation of vimentin observed in the Ha-RAS induced EMT associated with colon tumor cell migration, a pathway that may contribute to the metastatic progression of colon cancer. The effect of FRA-1 on cell motility could be through RhoA downregulation as it has been described previously.41
EMT is a pathological mechanism that facilitates the process by which epithelial tumors (carcinomas) become invasive and progress towards aggressive disease. We have developed a tumor model of EMT consisting in highly differentiated colon carcinoma cells that undergo EMT in response to Ha-RAS. Importantly, this model mimics the progression to invasive colon carcinoma. This EMT model could represent a valuable tool for identifying clinically relevant markers, and also provide a compelling new biochemical framework for understanding how this process contributes to tumor progression and malignant evolution.
This work was supported by the EU Research Network Grants LSHG-CT-2004-502950 “TRANS-REG” and MRTN-CT-2004-50422 “TAF-Chromatin” to A.P.