TATA Box-Binding Protein–Associated Factor 12 Is Important for RAS-Induced Transformation Properties of Colorectal Cancer Cells Free

Colorectal cancer develops through a multistage process known as the adenoma-carcinoma sequence, which drives normal mucosa to invasive carcinoma through early and advanced adenomas. Each step of the sequence occurs in parallel with the stepwise accumulation of genetic alterations that gradually result in tumor development and progression (1).

RAS is a monomeric, membrane-localized G protein. In mammalian cells, there are three related but distinct RAS genes, encoding for Kirsten (Ki-RAS), neuroblastoma (N-RAS), and Harvey (Ha-RAS) RAS proteins. All three proteins have been related to specific types of cancer because activating mutations result in constitutive signaling (2). Activated RAS mediates its biological activity via Raf/mitogen-activated protein kinase/extracellular signal-regulated kinase (ERK) kinase (MEK)/ERK kinase pathways that in return regulate transcription factors, such as the AP-1 and ETS families, to subsequently induce genes such as c-FOS (3, 4).

Interestingly, the incidence of K-RAS mutations in colorectal cancer reaches 50%. Deletion of the activated RAS allele suppresses the malignant phenotype in the human colon cancer cell lines DLD-1 and HCT116 (5). Mutant Ha-RAS has a low incidence in colorectal tumors but shows the highest transformation potential among the RAS proteins in NIH3T3 (6) and colon epithelial cells (7). In Caco-2 cells, it induces an epithelial to mesenchymal transition (EMT), which results in enhanced tumorigenicity and changes in cell morphology (7). EMT induces dramatic changes to the cells, which switch from a polarized epithelial to a highly motile mesenchymal phenotype related to cancer progression, invasion, and metastasis (8, 9).

Transcription factor IID (TFIID) is a large multiprotein complex composed of the TATA box-binding protein (TBP) and several TBP-associated factors (TAF). TBP and TAFs are highly regulated (10, 11). Interestingly, rather than affecting transcription globally, modified forms of TFIID selectively alter specific transcription networks. For instance, in cooperation with c-Jun, TAF4b can drive tissue-specific programs of gene expression (12), whereas other TAFs have been involved in particular phases of the cell cycle (13). Notably, TBP is up-regulated by a RAS-stimulated signaling pathway (14) potentially via an ETS protein-binding site on TBP promoter (15); (16) and contributes to cellular transformation in colon carcinoma cell lines (17).

TAF4 and TAF12 form a heterodimer important in the TFIID assembly (18). The TAF4 family also comprises the tissue-specific TAF4b, a substoichiometric subunit of TFIID able to heterodimerize with TAF12. Human TAF4 interacts with transcription factors involved in cell signaling (19, 20). Inactivation of TAF4 in mouse fibroblasts activates the transforming growth factor β signaling pathway (21). On the other hand, TAF12 has been linked to transcriptional activation, particularly through a direct interaction with ATF7, known to heterodimerize with Jun/Fos transcription factors (22). However, little is known about the involvement of TAF12 in cell proliferation.

We investigated potential changes in the expression of TBP and TAFs during carcinogenesis via mutated RAS in human colon carcinoma cell lines. Herein, evidence is presented that TAF12 is transcriptionally up-regulated in cell lines expressing mutant RAS to actively participate in cellular transformation by affecting E-cadherin expression.

We began by testing whether the multistep colorectal carcinogenesis process, driving cells from adenoma to adenocarcinoma, could affect the levels of TFIID components. For that, nuclear extracts from several cell lines were prepared and analyzed by Western blot. Among the tested proteins, TBP, TAF4, TAF4b, and TAF12 levels varied compared with β-actin (Fig. 1A; data not shown). Interestingly, TAF12 expression was enhanced in cells containing an activating mutation of Ki-RAS (8-fold in HCT116 and DLD-1 cells and 6.5-fold in SW620 compared with Caco-2 cells). In parallel, TAF4 and TBP levels presented a smaller but significant increase in SW620 and HCT116, whereas TAF4b levels presented changes unrelated to RAS.

To test the hypothesis of a TAF12 RAS-dependent expression, we used Caco-2 cell lines stably overexpressing a mutated, constantly active form of Ki-RAS and Ha-RAS. Caco-2 cells carrying the Ki-RASV12 mutation are named Caco-K (clones 6 and 15) and those carrying the Ha-RASV12 are named Caco-H (clones 1 and 2). Enhanced levels of Ki-RAS and Ha-RAS as well as levels of phosphorylated ERK corresponding to a MEK activity were confirmed by Western blot (Fig. 1B). Studies on mRNA levels also confirmed the specific overexpression of the RAS isoforms in the corresponding cell line (data not shown). Interestingly, levels of TAF12 were increased by 3-fold in extracts from Caco-H cells compared with Caco-2 (Fig. 1B, lanes 1-3), whereas in Caco-K15 cells an increase of 2.7-fold was observed (lanes 4 and 5). Smaller variations were observed for TAF4 and TBP levels. In extracts from HKE3 cell line, derived from HCT116 cells after disruption of mutated Ki-RAS allele, TAF12 expression was decreased (Fig. 1C). Notably, TBP expression was also decreased in HKE3 cells. For the rest of the study, we focused on the RAS-dependent regulation of TAF12 in Caco-H cells because the up-regulation of TAF12 was the strongest. In parallel, we were particularly interested in a potential role of TAF12 in the EMT, observed specifically in the Caco-H cells.

A microarray analysis conducted using the RZPD Oncochip showed a 2.9-fold up-regulation of TAF12 mRNA in Caco-H cells, suggesting a mechanism of transcriptional regulation (7). To validate this, we used both standard reverse transcription-PCR and real-time PCR (Fig. 1D). Compared with Caco-2 cells, TAF12 mRNA levels were increased by 5-fold in Caco-H cells (lanes 2 and 3) and to a lesser extent in Caco-K15 cells (lane 5), whereas no significant up-regulation was observed for the Caco-K6 clone, possibly due to low MEK activity. In parallel, a 5-fold induction of TAF12 mRNA was found in the SW620 cell line (lane 6), which had also high levels of TAF12 protein (Fig. 1A, lane 2).

To identify the mechanism implicated in TAF12 regulation by RAS, we used the UO126 inhibitor to block the Raf/MEK/mitogen-activated protein kinase pathway. Treatment of Caco-2 and Caco-H1 cells with UO126 in a rich medium resulted in a partial inhibition of the MEK activity but had little effect on TAF12 expression (Fig. 1E, lanes 1 to 2 and lanes 3 to 4; 61% and 42% inhibition of phosphorylated ERK, respectively). After serum starvation, levels of phosphorylated ERK were low in Caco-2 but not in Caco-H cells. Under these conditions, the UO126 inhibitor did not further affect the MEK activity in Caco-2 cells (lanes 5 and 6) but had a strong inhibitory effect (75%) on Caco-H cells (lane 7). In these cells, TAF12 showed a reduction of 60% (lane 8), suggesting that its expression is regulated via a MEK-dependent mechanism.

Together, those results show that TAF12 levels vary between different colorectal adenocarcinoma cell lines. Notably, TAF12 is transcriptionally enhanced via a MEK-dependent mechanism in the presence of the mutated Ha-RAS protein and, in a lesser extent, Ki-RAS protein stably overexpressed in Caco-2 cells.

To dissect the mechanism by which RAS affects the transcription of TAF12, we analyzed the proximal promoter region of TAF12 for oncogene-responsive transcription factor binding sites using MatInspector and Transfac. Among several putative binding sites, a sequence matching the consensus recognition site for the ETS family DNA-binding proteins was identified from −21 to −13. Sequence alignment in Fig. 2A shows that both the TBP and TAF12 promoter ETS-binding sites were similar to the optimum consensus recognition site for the endogenous human ETS1 protein on the ETS1.3 promoter (23). Moreover, sequence comparison of the human, mouse, and rat TAF12 proximal promoter revealed a high degree of conservation (Fig. 2B), underlying an important role in TAF12 transcriptional regulation. To test the functionality of this site, we did gel shift assays using nuclear extracts from H1 cells and a set of two probes corresponding to the TAF12 promoter putative ETS-binding site (ETST12) and the consensus ETS1-binding site on the ETS1.3 promoter (ETS1.3). The presence of the ETST12 probe induced the specific binding of a protein complex (Fig. 2C, lane 5). The interaction was abolished by the addition of an excess of nonradiolabeled ETST12 probe (lane 6) but not by that of a nonspecific competitor probe containing the AP-1–binding site of the vimentin promoter (lane 8). The binding was specific because the ETST12mut probe bearing a point mutation in the central core of the TAF12 promoter ETS-binding site failed to form the complex (lane 10). Moreover, ETST12 and ETS1.3 probes were in competition for protein binding when present together (lane 7), suggesting that similar protein complexes could interact with both DNA sequences. To determine a potential role of the ETS-binding site in the RAS-dependent regulation of TAF12, we monitored its occupancy in the presence of different protein extracts. As shown in Fig. 2D, the DNA-binding activity detected in Caco-H1 was stronger than that observed in Caco-2 extracts, in agreement with increased TAF12 mRNA levels (Fig. 1D).

Those results show that an ETS-binding site found on the TAF12 promoter is specifically recognized by a protein complex. Its occupancy is increased in cells containing the mutated form of Ha-RAS, pointing to a specific role in TAF12 up-regulation.

To identify whether the ETS1 protein was a member of the protein complex binding the ETS site on TAF12 promoter, we first analyzed the levels of ETS1 protein in the RAS-transformed clones. As seen in Fig. 3A (lanes 2 and 3), expression of ETS1 was found to be enhanced in Caco-H cells, where the occupancy of the ETS site was the highest in electrophoretic mobility shift assay (Fig. 2D). In parallel, the phosphorylated form of the ternary complex factor member ELK1 was also increased. When electrophoretic mobility shift assay was done in the presence of an antibody directed against ETS1, a small decrease of the signal intensity and the appearance of a weak shifted protein complex were observed (Fig. 3B, compare lanes 1 and 2), suggesting a binding of the ETS1 protein on the ETS site of the TAF12 promoter. To verify this hypothesis in vivo, chromatin immunoprecipitation (ChIP) of cross-linked chromatin from Caco-H1 cells was done. Immunoprecipitation was followed by PCR with primers encompassing the ETS site of the TAF12 promoter. As seen in Fig. 3C, a, the promoter region containing the ETS site was efficiently amplified from chromatin precipitated with an antibody recognizing the ETS1 protein (lane 4), whereas an antibody recognizing the ELK1 protein failed to precipitate this fragment (lane 5). Dilutions of the input chromatin were used as a positive control (lanes 1-3), whereas as a negative control a reaction done in the absence of any antibody was done (lane 6). To further show an in vivo role of the ETS1 protein binding in the Ha-RAS–induced TAF12 up-regulation, ChIP was done in the presence of an anti-ETS1 antibody and extracts from Caco-2, Caco-H1, and Caco-K15 cell lines. As seen in Fig. 3C, b, more ETS1 binding on TAF12 promoter was observed in Caco-H1 (lane 5) compared with Caco-2 and Caco-K15 cells (lanes 4 and 6). As a positive control for ChIP, an antibody directed against acetylated histone H3 immunoprecipitated the ETS site on TAF12 promoter (lane 8). A second set of primers encompassing the ETS site of TAF12 promoter produced identical results (data not shown).

To further show the role of the ETS1 protein in TAF12 regulation, short interfering RNA (siRNA)-mediated knockdown of the ETS1 protein was done. As seen in Fig. 3D, a, 48 h after transfection of Caco-H1 cells with the ETS1 siRNA, ETS1 was efficiently knocked down (lane 3). No effect was observed for the mock control (lane 1) and the nontargeting control siRNA (lane 2). RNA extracted from transfected cells was reverse transcribed and used to analyze the expression of both ETS1 and TAF12. Real-time PCR in Fig. 3D, b confirmed an important, 5-fold decrease of the ETS1 mRNA in ETS1 siRNA-treated cells and a smaller but significant decrease of the TAF12 mRNA.

Together, those results show that an ETS-binding site upstream of the transcription initiation site of TAF12 binds specifically, in vitro and in vivo, the human ETS1 protein. Binding to this site is higher in Caco-H1 cells, suggesting a role in the RAS-mediated TAF12 promoter regulation. The knockdown of ETS1 is accompanied by a decrease of TAF12, pointing to a physiologic role of ETS1 in TAF12 regulation.

To assess the cellular role(s) of TAF12, siRNA-mediated knockdown of the protein was done. As seen in Fig. 4A, 48 h after transfection of the cells by a synthetic siRNA duplex targeting TAF12 mRNA (lane 3), a reduction of intracellular TAF12 protein levels was observed compared with the control conditions (lanes 1 and 2). As expected, knockdown of TAF12 was paralleled by a decrease in levels of other TAFs, probably corresponding to loss of protein stability (Fig. 4B, lanes 1 and 3; data not shown). For instance, TAF4b was slightly decreased; TBP exhibited a moderate effect, whereas TAF10 was highly affected. Despite the effects due to down-regulation of TAF12, cells containing reduced TAF12 grew normally and only exhibited an apoptotic-related decrease in cell number corresponding to 20% and 25% of the control cells (48 and 72 h after transfection, respectively). Notably, TAF12 siRNA but not the control cells exhibited a particular change in cell morphology resulting in the appearance of long processes emanating from the cell body (Fig. 4C, top, siTAF12, black arrows). Processes were visible both under a light microscope (top) and after cell staining with an antibody recognizing the β-actin protein (bottom).

The above experiments suggest that a reduction of TAF12 by siRNA is not lethal but induces a decrease in protein levels of other TAFs and modifies the phenotype of transformed cells.

To identify a specific role of TAF12 in the RAS-induced carcinogenesis, we analyzed cell properties that were known to be modified by activated Ha-RAS or during EMT, such as adhesion and migration (Fig. 5). Interestingly, treatment of cells with TAF12 siRNA significantly blocked migration by 40% compared with the control siRNA-treated cells and by 60% compared with the nontreated cells. As a control, only 1% of the parental cell line Caco-2 showed a migrating capacity compared with its Ha-RAS–transformed counterpart Caco-H1, which has an EMT. In parallel, in cell adhesion assays, the cellular capacity to form complexes between the cytoskeleton and an extracellular matrix component, such as fibronectin, was diminished by 30% in cells treated with the TAF12 siRNA (Fig. 5B). Again, the Caco-2 cell line showed a restricted adhesive capacity compared with Caco-H1.

Together, those results show that the reduction of TAF12 is accompanied by a reduction of the adhesion and migration properties of mutated Ha-RAS–transformed cells. Further comparison of Caco-2 and Caco-H shows that mutated Ha-RAS transformation and consequent EMT dramatically increases these properties.

In an attempt to identify TAF12 target genes related to RAS transformation and EMT, reverse transcription-PCR analysis was used to monitor changes in the expression of several genes in TAF12 knockdown cells (Fig. 6A; data not shown). The genes tested code for different families of molecules such as cell surface receptors (E-cadherin and Integrin A6), intermediate filaments (vimentin), or transcription factors (TBP). Interestingly, most of the genes tested showed no difference in their expression after TAF12 reduction. Notably, only the E-cadherin mRNA quantity was altered, exhibiting an increase paralleled to the decrease of TAF12 mRNA. To quantify the effect of TAF12 on E-cadherin expression, we analyzed the expression of both genes using real-time PCR 48 h after transfection of Caco-H cells with TAF12 siRNA. As shown in the graph represented in Fig. 6A, a 2.5-fold reduction of TAF12 mRNA resulted in a 2.7-fold increase of E-cadherin mRNA levels. This increase was also confirmed at the protein level (Fig. 6B, lane 3). To test whether the morphologic changes observed in TAF12 knockdown cells (Fig. 4C) were associated to the reexpression of E-cadherin, we did immunofluorescence experiments (Fig. 6C). In siRNA-transfected cells (bottom), E-cadherin–associated signal (red) was covering the entire cell surface as well as the characteristic cell extensions. On the contrary, in nontransfected cells (top), E-cadherin signal was weak and diffuse. To confirm the effect of TAF12 reduction on E-cadherin, we used TAF12 siRNA in the HCT116 cell line, which naturally has high levels of TAF12 (Fig. 1A). Reverse transcription of total RNA followed by real-time PCR showed that a 2.5-fold reduction of TAF12 was accompanied by a 1.3-fold increase of E-cadherin (Fig. 6D).

Those results show that TAF12 reduction has no effect on the overall cellular transcription. On the contrary, they underline a specific role of TAF12 down-regulation on E-cadherin up-regulation both at the mRNA and protein levels.

To examine the relation between TAF12 and E-cadherin in more physiologic conditions, we analyzed the expression of E-cadherin in the RAS-transformed cells (Fig. 7A). In those conditions, both mRNA (Fig. 7A, a) and protein levels (Fig. 7A, b) of E-cadherin were dramatically reduced in Caco-H cells where TAF12 levels were found to be enhanced.

We next examined the E-cadherin levels in total extracts from colorectal cell lines. The Western blot in Fig. 7B, a showed that each cell line contained different amounts of E-cadherin: Caco-2, HT-29, and DLD-1 cells (lanes 1-3) had an important amount of protein, HCT116 and SW620 cells had barely detectable amounts (lanes 4 and 5), whereas the protein was no more detectable in Caco-H cells (lane 6). Interestingly enough, TAF12 expression was inversely proportional to that of E-cadherin. In addition, TAF12 increase was accompanied by an enhancement of the ETS1 expression, in agreement with an ETS1-dependent TAF12 regulation. As E-cadherin is an EMT marker, we also examined the levels of the second EMT marker, vimentin. As seen in Fig. 7B, a, vimentin expression was inversely proportional to that of E-cadherin, pointing on the existence of distinct stages of EMT in the panel of colorectal cell lines examined. To verify the parallel change in the expression of E-cadherin and TAF12 in this transition, real-time PCR was done on cDNA synthesized from Caco-2 and HCT116 cells. As seen in Fig. 7B, b, a 2.2-fold up-regulation of TAF12 mRNA in HCT116 cells was accompanied by a 30-fold reduction of the E-cadherin gene expression.

Finally, we examined by real-time PCR the expression pattern of E-cadherin in the HKE3 cell line, created by knockout of the naturally mutated Ki-RAS allele in HCT116 cells, where TAF12 expression was found to be altered (Fig. 1C). Interestingly, E-cadherin mRNA increased by 1.3-fold in HKE3 cells, whereas TAF12 mRNA was comparably decreased (Fig. 7C).

Together, those results show that TAF12 and E-cadherin display a common pattern of expression in established colorectal cell lines. This agrees with a role of TAF12 on E-cadherin regulation in colorectal cancer and EMT.

We have examined the expression pattern of TBP and TAFs in human epithelial colorectal adenoma Caco-2 cells stably expressing mutant, constantly activated forms of Ha-RAS and Ki-RAS. Herein, we provide evidence that TAF12 is up-regulated via oncogenic RAS signaling, mediated in part by the MEK pathway. In our knowledge, TAF12 is the first example of a TAF regulated by the RAS transforming function. Nevertheless, the identification of TBP as a RAS target gene supports the idea that the general transcription machinery activity could control transcription initiation in carcinogenesis events (14, 17).

We have previously shown that mutated Ha-RAS and Ki-RAS modulate specifically or commonly the expression of a large panel of genes (7). In our hands, up-regulation of TAF12 was common to both RAS-mutated isoforms, suggesting a central role of TAF12 in transformation by RAS. Nevertheless, the up-regulation of TAF12 was significantly higher in Caco-H cells, which were clearly subjected to the EMT during the RAS transformation (7-24). Notably, as this particular cell alteration into mesenchyme is the critical event leading to invasiveness and metastasis in cancer, we have used the Caco-H cell line as a model for EMT and transition to final stages of colorectal cancer. In parallel, we have confirmed TAF12 up-regulation during partial or complete EMT in established colorectal cell lines such as HCT116.

As a further support for the importance of TAF12 up-regulation in the development of cancer, we show that in the early-stage adenoma cells the expression of TAF12 is low compared with more advanced adenocarcinoma. We also show that TAF12 is transcriptionally regulated via a RAS-related mechanism. Accordingly, our initial study about the identification of RAS target genes by microarrays showed an up-regulation of both TBP and TAF12, which were at the time discriminated by the threshold applied (7), but further confirmed at the protein level. Notably, more TAF12 is monitored in DLD-1, SW620, and HCT116 cell lines, which bear an activating mutation of RAS. Nevertheless, the carcinoma HT-29 cell line containing wild-type RAS but mutant BRAF exhibits increased levels of TAF12 compared with the adenoma Caco-2 cells, suggesting the existence of a mechanism more probably related to the activation of the mitogen-activated protein kinase pathway and/or the activation of the ETS1 protein. Interestingly, a first study done in human colorectal tumors shows a small up-regulation of TAF12 in a subset of the tissues tested, raising the possibility of TAF12 involvement in particular tumor characteristics (data not shown). Ongoing work on human tumor samples will potentially allow the identification of the in vivo role of TAF12.

The data obtained from the analysis of the TAF12 promoter show that a putative ETS-binding site is functional in our cells. In parallel, we show by ChIP that the ETS1 protein binds to this site and is in part responsible for the RAS-mediated increase of TAF12 mRNA. Moreover, knockdown of the ETS1 protein by siRNA is accompanied by a decrease in TAF12 mRNA levels. Interestingly, the family of ETS transcription factors contains a large number of members that control the expression of genes involved in cell proliferation, differentiation, and transformation (25). Notably, an ETS-binding site on the TBP promoter has been found to be involved in the RAS-mediated induction of the human promoter activity (14). Nevertheless, the results that we obtained in electrophoretic mobility shift assay done in the presence of an anti-ETS1 antibody suggest that the complex that binds the ETS site of TAF12 promoter contains only a small proportion of the actual ETS1 protein. Like for TBP, other ETS family members may also be involved in TAF12 regulation (26). In agreement, knockdown of the ETS1 protein affects only, in part, the expression of TAF12. This suggests that other ETS family members could replace ETS1 action on TAF12 promoter or that, in the absence of the ETS1 protein, an ETS1-independent mechanism could be recruited to maintain levels of TAF12. Accordingly, our results show that TAF12 levels are reduced only when the MEK inhibitor is used in a low serum environment, possibly pointing to a more efficient inhibition of the MEK pathway in the absence of growth signals. Alternatively, this result could reflect the presence of a serum-dependent/MEK-independent mechanism acting on TAF12 expression. This suggests that, similarly to TBP promoter, at least one mechanism of regulation of TAF12 is acting independently of the ETS1 protein and the MEK/ERK pathway (14). Accordingly, TBP expression is altered by MEK inhibition in a low serum environment (14).

E-cadherin, a molecule implicated in cell-cell adhesion, is silenced in Ha-RAS–transformed cells (Caco-H) with an EMT-related phenotype (27). We found that TAF12 reduction by siRNA produces a specific derepression of E-cadherin and subsequent inhibition of cell adhesion and migratory capacity both in Ha-RAS–transformed cells with EMT and in HCT116 cells with some EMT characteristics. Evidence from recent report suggests that TAF12 can act as a coactivator of the VDR activity (28). As E-cadherin expression is regulated by nuclear receptors and androgens as well as vitamin D itself (29), it is conceivable that TAF12 could have a role in this mechanism. Alternatively, TAF12 may interfere with the function of complexes that retain E-cadherin silenced, particularly those involved in histone modification and DNA methylation. This last hypothesis is supported by the fact that TAF12 is effective on E-cadherin expression preferentially in cells that have low basal (repressed) E-cadherin (data not shown). TAF12 regulation of E-cadherin can be critical for particular tumor properties because long-term repression of E-cadherin can induce EMT and its reexpression can suppress tumor invasion (30). This suggests that the full RAS cellular transformation needs particular changes in gene expression, potentially achieved by overexpressed TAF12.

Nevertheless, knockdown of TAF12 is not sufficient to reverse the Caco-H phenotype. A possible explanation is that cells have acquired nonreversible changes during the EMT. Alternatively, and as suggested by others (31), a transient modification in E-cadherin levels could be insufficient to produce a reversion. A third possibility is that TAF12 specifically directs the transcription of only a subset of genes implicated in EMT. The observations of this study support the third explanation because no change in expression of several other genes tested was observed, most notably of vimentin expression, a characteristic EMT marker.

We have examined several colorectal cell lines and observed that some have partial EMT characteristics, such as repression of E-cadherin and/or up-regulation of vimentin at different degrees. Interestingly, the SW620 cell line, which in our hands showed both the more repressed E-cadherin and enhanced TAF12 and vimentin, has been established from a lymph node metastasis, has EMT characteristics, and is considered as a model for studying molecular events in the late stages of colon cancer (32). An appealing hypothesis is that EMT is a key event in colorectal tumor progression. Indeed, the transition from early adenoma represented by the Caco-2 cells to the late adenoma with metastatic properties such as SW620 cells could be mediated by different degrees of EMT defined by their content in E-cadherin and vimentin. Further work in cell lines and tumor samples is needed to show the dispensable presence of EMT in tumor progression as well as a biological relevance of TAF12 in EMT and cancer.

It has been shown that loss of one central TAF can affect the overall TFIID structure, thereby inducing degradation of other TAFs and further leading to the impediment of transcription (33). Specifically, the reduction of TAF12 by RNA interference in Drosophila Schneider cells results in a disorganization of TFIID (34). The results of our studies using the TAF12 siRNA confirm a reduction in several TAFs. Nevertheless, our data suggest a regulatory role of TAF12 for specific gene (group) regulation, which can influence particular cell properties, such as cell migration. In perfect agreement with this, recent studies show that, in the absence of TAFs, single cells are capable of fulfilling transcription but respond inappropriately (33). Future studies will be necessary to reveal the potential role of other TFIID members in regulating E-cadherin expression. This may lead to the discovery of new strategies aimed at uncoupling the RAS mutation from tumor development and progression.

American Type Culture Collection cell lines as well as RAS clones were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), antibiotics, and nonessential amino acids, all from Invitrogen Corp. The HKE3 cell line was described elsewhere (5). Cell lines were regularly passaged. All experiments were done at 80% confluence, unless differently stated. RAS mutations have been verified by sequencing. For MEK inhibition, cells were grown in 10% or 0.5% FBS for 24 h before inhibitor or DMSO addition (40 μmol/L UO126; Alexis Biochemicals). Inhibitor was incubated for 15 h, replaced by fresh, and incubated for an additional 5 h.

Cells were transfected with human TAF12 individual siGENOME ON-TARGET duplex (D-012149-03), ETS1 ON-TARGETplus SMART pool composed of four different duplexes (L-003887-00), or the siCONTROL RISC-free siRNA (D-001210-03; Dharmacon) as described in the text by the calcium phosphate method done based on a previously published protocol (35). For one well of a 12-well plate, 40 μL of a solution containing 64 pmol of duplex siRNA and 0.25 mol/L CaCl₂ were mixed to 40 μL of 2× BES-buffered solution, incubated for 25 min at room temperature, and added to the cells in a total medium volume of 800 μL. Eight hours after transfection, cells were washed twice with medium and further incubated for the required period.

Nuclear and cytoplasmic extracts were prepared following the protocol described elsewhere (36). Buffers were supplemented with protease and phosphatase inhibitors before use. Whole-cell lysates were prepared as described (37). Protein concentrations were determined by the method of Bradford (Bio-Rad protein assay). Nuclear (20 μg) and total extracts (40 μg) were subjected to SDS-PAGE and transferred to a nitrocellulose membrane (Pall Corp.). Immunoblotting was carried out using the following antibodies: anti-β-actin (Abcam); anti-phosphorylated ERK, anti-ETS1, anti-HRAS, anti-KRAS, anti-SAP1a, and anti-vimentin (Santa Cruz Biotechnology); anti-phosphorylated ELK1 (Cell Signaling Technology, Inc.); and anti-TAF4 (32TA; ref. 20), anti-TAF12 (22TA; ref. 38), anti-TAF4b (21), anti-TAF7 (19TA; ref. 39), anti-TAF10 (6TA) and anti-TBP antibody (2C1; ref. 40), and peroxidase-conjugated anti-mouse/rabbit IgG (Jackson ImmunoResearch Laboratories). Signal was obtained with the enhanced chemiluminescence plus Western blotting detection system (Amersham Biosciences) after exposure to Kodak Super RX film. Values were measured using the ImageQuant software (Amersham Biosciences). All the experiments were repeated at least thrice and representative images are shown.

Total RNA isolation from cultured cells was done using the Trizol reagent (Invitrogen). Reverse transcription was carried out from 3.0 μg of purified RNA using the SuperScript reverse transcriptase (Invitrogen) and an oligo(dT)₁₅ in 20 μL following the manufacturer's instructions.

PCR was done in the presence of 1 to 3 μL of a 1:7 dilution of the reverse transcription reaction using the TaqDNA polymerase (Invitrogen) in a total volume of 25 μL containing 2.5 mmol/L MgCl²⁺, 10 pmol of each primer, and 0.2 mmol/L deoxynucleoside triphosphate. Products were separated on 1.5% agarose gels containing ethidium bromide and quantified by digital imaging (ImageQuant, Molecular Dynamics). Results were confirmed from three independent experiments. Primers used for PCR were the following (5′ to 3′): glyceraldehyde-3-phosphate dehydrogenase (GAPDH), GAAGGTGAAGGTCGGAGT (forward) and CATGGGTGGAATCATATTGGAA (reverse); TAF12, CAGCCCTAATCAACCTCTCC (forward) and GCTAAGACGACCTCCTGCC (reverse); E-cadherin, CGACCCAACCCAAGAATCTA (forward) and GCTGGCTCAAGTCAAAGTCC (reverse); TBP, CACGAACCACGGCACTGATT (forward) and TTTTCTTGCTGCCAGTCTGGAC (reverse). Primers corresponding to integrin α6 receptor and vimentin are described in ref. 7.

Real-time quantification was carried out in 96-well PCR plates using a Bio-Rad iCycler and the iQ5 Multicolor Real-Time PCR detection system (Bio-Rad). Cycling conditions included a denaturing step of 3 min at 95°C followed by 40 cycles at 95°C for 40 s and annealing/elongation at 62°C for 40 s. Each reaction (25 μL) contained 1× iQ SYBR Green Supermix (Bio-Rad) and 150 nmol/L of each primer. Each gene was tested in triplicate. Results were analyzed on the iCycler software. Each value was normalized to GAPDH and expressed relatively to a reference. Primers for GAPDH and TAF12 amplification were the same than those used for regular PCR (see sequence above). For real-time amplification of E-cadherin, primers had the following sequence (5′ to 3′): CTTACTGCCCCCAGAGGAT (forward) and GCTGGCTCAAGTCAAAGTCC (reverse), whereas those used for ETS1 were the following: ATACCTCGGATTACTTCATTAGC (forward) and GGATGGAGCGTCTGATAGG (reverse).

Standard gel shift assays were done based on the protocol described in ref. 23. Briefly, 8 μg of nuclear extracts were incubated for 60 min at room temperature with ³²P-labeled double-stranded probe ([γ-³²P]ATP; Amersham Biosciences) in the presence of 1.5 μg of poly(deoxyinosinic-deoxycytidylic acid) (Sigma-Aldrich). Reaction was loaded onto native 6% polyacrylamide gel containing 0.25× TBE [10× TBE: 89 mmol/L Tris base, 89 mmol/L boric acid, 20 mmol/L EDTA (pH 8.0)]. Gel was prerunned for 30 min at 250 V, runned at 125 V for 2 h, dried for 1.5 h, and exposed for 12 h to a storage phosphor screen (Amersham Biosciences). Scanning was done with Storm 860 scanner (Amersham Biosciences) and values were measured using ImageQuant software. For competition, 1 μL of nonlabeled probe (100 molar fold excess of the labeled probe) was incubated with nuclear extracts for 30 min before addition of the labeled probe. For supershift experiments, 3 μL of antibody were preincubated with extracts 1 h before addition of the probe. For double-stranded probe preparation, oligonucleotides were mixed in equimolar amounts, heated, and allowed to cool for 16 h. Top strand sequences of oligos (Operon Biotechnologies) are shown below (5′ to 3′) with the protein-binding sites indicated in bold and mutations underlined. The AP-1 oligonucleotide was derived from ref. 41. Double-stranded oligos were blunt ended.

• ETS1-3: GATCTCGAGCCGGAAGTTCGA

• ETST12wt: GCGGTGCCGGAAGTAGTCGGG

• ETST12mut: GCGGTGCCAGAAGTAGTCGGG

Cells were grown to 80% confluence in 100-mm dishes, cross-linked with 1% formaldehyde for 10 min at room temperature, collected in ice-cold PBS in the presence of phenylmethylsulfonyl fluoride (0.5 mmol/L), and resuspended in swelling buffer [25 mmol/L HEPES (pH 7.8), 1.5 mmol/L MgCl₂, 10 mmol/L KCl, 0.1% NP40, 1 mmol/L DTT, 0.5 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin]. After homogenization for 10 times in a Dounce homogenizer, nuclei were resuspended in sonication buffer [50 mmol/L HEPES (pH 7.9), 140 mmol/L NaCl, 1 mmol/L EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, 0.5 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin]. Chromatin was sheared in an average range of 500 to 1,000 bp by sonication in a water bath, with generation of high-power ultrasound (Bioruptor, Diagenode): 15 cycles of 30 s on, 30 s off (1 cycle/min) at maximum power. Protein A-Sepharose matrix CL-4B (Amersham Biosciences AB) used for immunoprecipitation was presaturated with salmon sperm DNA (Sigma-Aldrich) and rabbit serum. Chromatin was precleared in the presence of protein A matrix, salmon sperm DNA, and rabbit serum during 1 h at 4°C. Precleared chromatin was incubated overnight in rotation with 8 μg (Ets-1 and Elk-1 from Santa Cruz Biotechnology) or 4 μg (acetylated histone H3; Upstate) of antibodies. Immunoprecipitation with protein A was done for 2 h at 4°C. Matrix was washed twice (10 min per wash) with sonication buffer, twice with washing buffer A [50 mmol/L HEPES (pH 7.9), 500 mmol/L NaCl, 1 mmol/L EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, 0.5 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin], twice with washing buffer B [20 mmol/L Tris (pH 8.0), 1 mmol/L EDTA, 250 mmol/L LiCl, 0.5% NP40, 0.5% sodium deoxycholate, 0.5 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin], and twice with TE buffer. Elution was done with 400 μL elution buffer [50 mmol/L Tris (pH 8.0), 1 mmol/L EDTA, 1% SDS]. After decross-linking and treatment with proteinase K (10 mg/mL), phenol/chloroform DNA extraction was done, DNA was resuspended in 20 μL TE buffer, and 5 μL were analyzed by PCR. PCR was done as described above for 35 cycles and an annealing temperature of 62°C. Half of the PCR was analyzed on an 8% polyacrylamide gel stained with ethidium bromide.

Primers used for amplification of the TAF12 promoter ETS site were the following (5′-3′): GCGGTGCCGGAAGTAGTC (forward) and GGCAGCAGCGTCTATCTCC (reverse).

Twenty-four hours after transfection with siRNA, cells were trypsinized, washed thrice in medium supplemented with 1% FBS, and counted with a Z2 Coulter Counter (Beckman Coulter). Cells (1 × 10⁴) were plated into the top chamber of 8-μm-pore Transwell filter (Corning) mounted in a 24-well dish containing medium with 10% FBS. Before use, filters were precoated for 10 h at 4°C with fibronectin (10 μg/mL; Sigma-Aldrich) and washed thrice. Cells were allowed to migrate in 5% CO₂ for 30 h at 37°C, fixed with methanol for 10 min at room temperature, and stained with 0.1% crystal violet. The underside of the filters was examined with a 40× objective of a light microscope and number of migrating cells was determined for 10 randomly selected fields. Data were obtained from two independent experiments, each repeated twice. For cell adhesion assay, 50 μL of cell suspension (4 × 10⁵ cells/mL) were seeded onto fibronectin-coated 96-well dishes and allowed to adhere for 30 min at 37°C. After methanol fixation and crystal violet staining, cells were washed with water and dye was solubilized with 30% acetic acid. Attachment was quantified by measuring the absorbance at 600 nm. For this experiment, fibronectin was used at 20 μg/mL. Coating was followed by incubation in 1× PBS + 0.5% (w/v) bovine serum albumin [1× PBS: 0.14 mol/L NaCl, 0.0027 mol/L KCl, 0.01 mol/L phosphate buffer (pH 7.4)] and three washes using PBS + 0.1% bovine serum albumin. Data were obtained from two independent experiments each done six times. SD function was used for error bar generation.

For confocal laser scanning microscopy, cells were grown on glass coverslips, fixed in 4% paraformaldehyde (10 min; room temperature), and permeabilized with 0.1% Triton X-100 for 10 min. Cells were blocked with 5% FBS (v/v) for 1 h and stained using rabbit anti-E-cadherin antibody diluted 1:100 and mouse anti-TAF12 antibody diluted 1:200 in 1× PBS containing 2% rabbit serum for 16 h at 4°C. Samples were washed and stained with Alexa Fluor 488–conjugated goat anti-mouse and Alexa Fluor 555–conjugated goat anti-rabbit secondary antibodies (1:300; Molecular Probes, Invitrogen) in 1× PBS containing 2% rabbit serum. After washing, nuclei were stained with Hoechst 33258 (Sigma-Aldrich) for 10 min and coverslips were mounted on glass slides in Gelvatol medium + DABCO, sealed with nail varnish, and viewed with a Leica TCS SPE confocal laser scanning microscope (Leica Lasertechnik). The objective lens used was 40×/1.15. The LAS AF software was used for image acquisition. For immunofluorescence experiments, cells were grown in 48-well plates (Greiner Bio One), treated as described above, and stained with an actin antibody (1:100 in PBS + 1% FBS) for 16 h at 4°C. Alexa Fluor 488–conjugated goat anti-mouse secondary antibody was used (1:400). Cells were visualized with a Nikon Eclipse TE200 inverted fluorescence/phase-contrast microscope equipped with a Sony charge-coupled device camera.