Tel.: +716-845-7681, Fax: +716-845-1698
Role for transcription factor TFII-I in the suppression of SSeCKS/Gravin/Akap12 transcription by Src
Article first published online: 21 JUN 2010
Copyright © 2010 UICC
International Journal of Cancer
Volume 128, Issue 8, pages 1836–1842, 15 April 2011
How to Cite
Bu, Y., Gao, L. and Gelman, I. H. (2011), Role for transcription factor TFII-I in the suppression of SSeCKS/Gravin/Akap12 transcription by Src. Int. J. Cancer, 128: 1836–1842. doi: 10.1002/ijc.25524
- Issue published online: 22 FEB 2011
- Article first published online: 21 JUN 2010
- Manuscript Accepted: 10 JUN 2010
- Manuscript Received: 7 DEC 2009
- NIH. Grant Number: CA94108
- DOD. Grant Numbers: PC040256, PC061246
- NIH/NCI Cancer Center. Grant Number: 2P30 CA016056
Vol. 130, Issue 2, 490, Article first published online: 13 DEC 2011
- mass spectrometry
The SSeCKS/Gravin/AKAP12 gene, encoding a kinase scaffolding protein with metastasis-suppressing activity, is transcriptionally downregulated in Src-transformed cells through the recruitment of HDAC1 to a Src-responsive proximal promoter site charged with Sp1, Sp3 and USF1. However, the ectopic expression of these proteins formed a suppressive complex in Src-transformed but not in parental NIH3T3 cells, suggesting the involvement of additional repressor factors. Transcription factor II-I (TFII-I) [general transcription factor 2i (Gtf2i)] was identified by mass spectrometry as being associated with the SSeCKS promoter complex in NIH3T3/Src cells, and moreover, the Src-induced tyrosine phosphorylation of TFII-I significantly increased its binding to the SSeCKS proximal promoter. siRNA-mediated knockdown of TFII-I or the expression of TFII-IY248/249F caused the derepression of SSeCKS in NIH3T3/Src cells. Taken with previous data showing that the tyrosine phosphorylation of TFII-I facilitates its nuclear translocation, these data suggest that Src-family kinase-mediated phosphorylation converts a portion of TFII-I into a transcriptional repressor.
The Src oncogene induces oncogenic transformation by several means, including altering signaling and cytoskeletal pathways through the direct phosphorylation of substrates via its intrinsic protein-tyrosine kinase activity.1, 2 A less studied mechanism involves the transcriptional regulation of genes that either promote or suppress oncogenic transformation.3 A small number of studies have identified gene expression signatures regulated by oncogenically activated Src,4, 5 and these have been used to predict cancer clinical outcomes and subtypes.6–9
A handful of Src-induced or suppressed genes have been studied in detail, and the Src-dependent transcriptional mechanisms governing the differential expression of these genes often involve seemingly opposite effects by the same transcription factor. For example, Src upregulates transcription of the urokinase receptor (uPAR), matrilysin, CD11c and prostaglandin synthase-2 genes via increased binding of Sp1 and AP-1 to proximal promoter sites.10–13 In contrast, downregulation of the SSeCKS/Gravin/AKAP12 and QR1 genes by Src requires Sp1 and AP1 promoter binding events, respectively.14–16 Indeed, Sp1 and AP1 can be influenced to either induce or repress gene transcription by cell- or cancer-specific transcription factors and/or post-translational modifications.17, 18 Thus, it is likely that Src alters their activity by controlling the expression of co-activators or co-repressors, or by directly modifying transcription factor activity through tyrosine phosphorylation.
Transcription factor II-I (TFII-I), also known as general transcription factor 2i (encoded by the GTF2I gene), is a ubiquitously expressed protein first identified as a positive regulator of mitogen-induced c-fos transcription (reviewed in Ref.19). Its ability to function in transcriptional regulation was subsequently shown to require tyrosine phosphorylation in response to growth factor signaling20, 21 or to B- or T-cell receptor activation.22, 23 The direct phosphorylation of TFII-I by mitogen-activated Src on Y248 facilitates its nuclear translocation, thereby facilitating its association with c-fos promoter sites and the subsequent induction of c-fos transcriptional activity.24 Several TFII-I splice variant products exist that exhibit nonoverlapping cellular distribution, suggesting varying functions.25 For example, TFII-Iβ is found in the nucleus constitutively, whereas mitogens induce the nuclear translocation of TFII-IΔ and the egress of TFII-Iβ, and thus, TFII-IΔ is mainly responsible for growth factor induced c-Fos transcription.26 Interestingly, the TFII-IΔ isoform also binds ERK1/2 via its poY248 domain and then shuttles it to the nucleus in response to mitogens, thereby transducing growth factor-induced cell cycle progression.26 Although the TFII-I-related protein family member, BEN, can repress transcription of TFII-I-induced genes such as GSC (encoding goosecoid),27 TFII-I has mainly been viewed as a transcriptional activator, in part, by its ability to bind to histone deacetylases (HDAC), thereby relieving their repressive activity at promoter sites.28 However, TFII-I can function as a transcriptional repressor.29
In analyzing how oncogenic Src induced the transcriptional downregulation of the SSeCKS/Gravin/AKAP12 (“SSeCKS”) gene, we identified a minimal Src-responsive proximal promoter sequence (−47/−106) requiring the binding of USF1, Sp1, Sp3 and HDAC1.14 Paradoxically, the binding of Sp1 and Sp3 to this promoter domain was increased in Src-transformed compared with NIH3T3 cells, although the forced overexpression of either or both of these proteins in NIH3T3 cells induced, rather than suppressed SSeCKS expression. This suggests that Src converts Sp1/Sp3 into repressive factors through modification or through the expression of co-repressors. Here, we identify TFII-I by mass spectrometry as a factor with increased binding to the Src-responsive SSeCKS proximal promoter/protein complex, and demonstrate that the downregulation of TFII-I or the expression of TFII-IY248F is sufficient to derepress SSeCKS expression in Src-transformed NIH3T3. Our data provide the first example of TFII-I repressor activity mediated by Src-induced phosphorylation, and they elucidate the mechanism by which Src promotes oncogenic progression through the transcriptional downregulation of the SSeCKS tumor suppressor.
Material and Methods
Cell culture, plasmids and nuclear lysate preparation
The growth of NIH3T3 and Src/3T3 cells and the production of nuclear lysates were described previously.14 The mouse TFII-I-specific siRNA (sense: 5′-UCAGCUCCAUGAGGAG GAUCU-3′; antisense: 5′-AUCCUCCUCAUGGAGCUGAUU-3′) and scrambled siRNA control (sense: 5′-UACCACUGAA GUUACUCAGCC-3′) described previously26 were synthesized by IDT (Coralville, IA), and the silencing of transfected cells was for 10 days. Plasmids expressing GST-tagged WT- or Y248/249F-TFII-IΔ were gifts of Ananda Roy (Tufts University School of Medicine) and were described previously.24 Stable cell lines were produced by co-transfecting the TFII-I plasmids along with pEGFP/neo (Clontech/Takara, Mountainview, CA) and then selecting for growth in media supplemented with 500 μg/mL G418 (Invitrogen, Carlsbad, CA).
Oligonucleotide pulldown and mass spectrometry
Pulldown of proteins associated with biotinylated double-stranded oligonucleotides representing the αSSeCKS proximal promoter (Fig. 1a) was described previously.14 Proteins that were associated with the SSeCKS promoter oligonucleotides after three washes were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and after the gels were stained with Deep Purple (GE Healthcare; http://www.gehealthcare.com), protein bands were scanned using a Typhoon 9410 Imager (GE Healthcare). Protein bands of interest were excised either manually or with an Ettan SpotPicker (GE Healthcare) and placed in microtubes [0.6 mL, Axygen, prewashed with 18 MΩ/cm−1 water (Mili-Q) and MeOH]. In-gel digestion with trypsin was performed according to standard procedures routinely used in the RPCI proteomics facility. Briefly, the gel pieces were destained with 50% acetonitrile (ACN)/100 mM ammonium bicarbonate (NH4HCO3) (200 μL) for 30 min with constant mixing (MixMate, Eppendorf, Westbury, NY). After removal of the solution, the gel pieces were dehydrated with ACN (100 μL) for 15 min at room temperature (RT) and dried in a Speedvac concentrator (Thermo Scientific, Waltham, MA). The dried gel pieces were reduced with 10 mM dithiothreitol (DTT) (Sigma)/100 mM NH4HCO3 at RT for 45 min. The solution was removed, and the samples were alkylated with 50 mM iodoacetamide (Sigma, St. Louis, MO)/100 mM NH4HCO3 (200 μL) at RT in the dark for 30 min. After removal of the solution, the gel pieces were washed with of 100 mM NH4HCO3 (200 μL) and incubated with 50% ACN/100 mM NH4HCO3 (200 μL) at RT for 10 min. Gel pieces were dehydrated with ACN (100 μL), dried in a Speedvac concentrator and digested with trypsin (Promega, Madison, WI; 10 ng/μL in 10% ACN/40 mM NH4HCO3, 30–50 μL) at 37°C for 16 hr. The digests were extracted twice with 50% ACN/0.1% trifluoracetate (TFA) (100 μL) at RT for 60 min with constant mixing. The extracts were pooled and dried in a Speedvac concentrator, and each sample was then reconstituted with 8 μL of 2% formic acid (FA).
The tryptic digests (4 μL of each sample) were analyzed by LC nanoelectrospray-tandem mass spectrometry (LC-ESI-MS/MS) using a nanoACQUITY UPLC (Waters, Milford, MA) coupled through a nebulization-assisted nanospray ionization source to a Q-ToF Premier mass spectrometer (Waters/Micromass). The LC consisted of a trap column (Symmetry C18, 5 μ, 180 μ × 20 mm, Waters), followed by separation on an analytical column (Atlantis C18, 3 μ, 100 μ × 10 cm, Waters). Samples were loaded, trapped and washed at a flow rate of 4 μL/min with 98% solvent A (0.1% FA)/2% solvent B (ACN containing 0.1% FA) for 5 min. Peptides were eluted with a gradient of 99% A/1% B to 50% A/50% B for 70 min at 0.4 μL/min, 10% A/90% B for 20 min at 0.6 μL/min and then 10% A/90% B at 0.6 μL/min for 10 min. Throughout the gradient, the mass spectrometer was programmed [Data Dependent Acquisition experiment (DDA)] to monitor ions with m/z in the range of 320–1620, and ions with +2 to +4 charges only were selected for MS/MS experiments using the preset DDA collision energy parameters.
Database search, peptide and protein identification
MS/MS spectra were processed and transformed to the PKL file format using Proteinlynx Global Server v2.2 (Waters/Micromass) and the default parameters of MaxEnt3 (Waters/Micromass). The PKL files were used to search the Rodentia subset of the Swiss-Prot database (release 53.2, containing 21,110 sequences) using a locally installed version of MASCOT (Matrix Science, Boston, MA, v2.2.0). The search parameters were as follows: trypsin as the proteolytic enzyme with two possible missed cleavages, carboxyamidomethylation of cysteine as a fixed modification, oxidation of methionine and phosphorylation of STY as a variable modification, the allowable mass error was 100 ppm for peptides and 100 mDa for fragment ions, peptide charge was set to 2+ and 3+ and the instrument was set to ESI-QUAD-TOF. The Mascot default significance threshold of p < 0.05 for assignments was used in the searches and a minimum of two unique peptides were used as a criteria for a match.
Chromatin immunoprecipitation (ChIP) assays were performed as described previously14 using TFII-I antibody (Cell Signaling, Beverly, MA; cat. #4562). The primers for PCR amplification of SSeCKS α proximal promoter sequence between −270 and +33 were 5′-TGCTGCTCCTGAACCTT CTG-3′ and 5′-GATCCTGCTGAGAACACACC-3′.
Immunoblotting, immunoprecipitation and RT-PCR
Immunoblotting (IB) and immunoprecipitation (IP) were performed as described previously14 using the following antibodies (Ab): rabbit anti-rat SSeCKS,30 TFII-I (Cell Signaling cat. #4562, 1:1000 dilution), GAPDH (Santa Cruz Biotechnology, Santa Cruz, CA; cat. #sc-257778, 1:1000 dilution) and antiphosphotyrosine MAb-4G10 (Millipore/Upstate, Billerica, MA). The semi-quantitative RT-PCR for SSeCKS α isoform and β-actin was described previously.14
Results and Discussion
We previously showed that a minimal Src-responsive element in the SSeCKS promoter (−47/−106) required USF1 binding to an upstream E-box and Sp1/Sp3 binding to a downstream GC-box, plus the association of HDAC1.14 Moreover, although the level of USF1 binding did not change between NIH3T3 and Src/3T3 cells, the binding of Sp1/Sp3 increased at least threefold in Src/3T3 nuclear lysates compared with those of NIH3T3, corresponding to an overall transcriptional suppression of the SSeCKS gene by oncogenic Src. In some systems, Sp3 can induce transcriptional suppression by competing with Sp1 for DNA binding sites, normally a transcriptional activator,31 from promoter sites.32 However, in our system, the ectopic expression of Sp1 and/or Sp3 led to SSeCKS upregulation. Alternatively, Sp1/Sp3 can be converted into repressors upon the recruitment of HDACs,33–35 and indeed, the forced co-expression of Sp1, Sp3 and HDAC1 induced SSeCKS transcriptional downregulation in untransformed NIH3T3 cells.14 However, the levels of total and nuclear Sp1, Sp3 or USF1 did not vary between NIH3T3 and Src/3T3 cells, strongly suggesting that Src either modifies one or more of these factors or induces the expression of repressor proteins that complex with the Sp1/Sp3/USF1/HDAC1 complex.
To address the latter possibility, nuclear lysates from NIH3T3 and Src/3T3 cells were bound to a biotin-labeled, double-stranded oligonucleotide spanning the minimal Src-responsive SSeCKS promoter (Fig. 1a; “oligo 8”). The DNA proteins complexes were precipitated by binding to streptavidin-agarose beads, and after washing, the associated proteins were identified by denaturing SDS-PAGE followed by staining with Deep Purple. Figure 1b shows five bands found more abundantly in the Src/3T3 sample, with “Src Ib” being the most prominent, as well as three bands with more abundance in the NIH3T3 sample. Given the increased binding of Sp1 and Sp3 to this SSeCKS promoter region in the Src/3T3 cells,14 we analyzed the Src Ia–c bands first. Protein bands were excised from the gels, digested to completion with trypsin and then subjected to LC-MS/MS analysis. The identities of the proteins in each band are shown in Figure 1c, with the most abundant peptides ordered from top to bottom.
We focused on general TFII-I because its apparent binding to the SSeCKS proximal promoter region was the most abundantly increased in the Src/3T3 cells. Moreover, TFII-I is known to interact directly with USF1 via E-box promoter sequences,36 factors required for Src-responsiveness by the SSeCKS proximal promoter. To confirm that TFII-I binds to the SSeCKS promoter, two biotin labeled, double-stranded oligonucleotides were used to precipitate proteins from nuclear lysates of NIH3T3 or Src/3T3 cells: “oligo 8,” used in the original pulldown assay described in Figure 1a, contains both the E- and GC-boxes and is sufficient to bind USF1 and Sp1/Sp3, respectively,14 “oligo 2” contains only the GC-box plus several downstream bases from oligo 8. Although the levels of total TFII-I [β and Δ isoforms] in NIH3T3 and Src/3T3 lysates are similar (see “input” lanes), there was a preferential binding of TFII-I to both oligonucleotides in the Src/3T3 lysates (Fig. 2a). The absence of detectable TFII-I binding to streptavidin beads (“no oligo”) argues against the possibility of nonspecific binding in this system. The enhanced recruitment of TFII-I to the αSSeCKS proximal promoter in Src/3T3 cells was confirmed by ChIP analysis (Fig. 2b). We then investigated whether TFII-I plays a direct role in suppressing SSeCKS expression in Src-transformed cells. Treatment of Src/3T3 cells with TFII-I-specific siRNA, shown previously to knockdown mouse both TFII-Iβ and Δ isoforms,26 induced a 2.5-fold decrease in TFII-I protein levels (Fig. 2c, upper panel) and a concomitant increase in αSSeCKS RNA levels as determined by semi-quantitative RT-PCR (Fig. 2c, lower panel). Treatment with a control siRNA (scrambled) affected neither TFII-I protein nor SSeCKS RNA levels. This indicates that TFII-I plays a role in the Src-mediated downregulation of SSeCKS.
In agreement with previous data,24 the relative tyrosine phosphorylation of TFII-I was increased 9.8-fold in the Src/3T3 cells compared with NIH3T3, even when normalized to the slightly higher (1.5-fold) TFII-I protein levels in the Src/3T3 cells (Fig. 2d). Because Src-induced transformation involved TFII-I tyrosine phosphorylation rather than changes in protein abundance, we addressed whether the expression of the TFII-IY248/249F mutant, incapable of being phosphorylated by Src, could derepress SSeCKS. Thus, stable Src/3T3 cells expressing WT or Y248/249F TFII-I (or empty vector) were produced. Representative clones were isolated that showed similar levels of these ectopic proteins (Fig. 2e; note that the exogenous GST-TFII-I proteins are 27 kDa larger than the endogenous forms). RT-PCR analysis indicates that expression of the Y F TFII-I mutant, but not WT TFII-I, resulted in increased αSSeCKS transcript levels (Fig. 2f). Taken with the above data, this finding strengthens the notion that Src-induced phosphorylation of TFII-I followed by its recruitment to the proximal SSeCKS promoter results in the repression of SSeCKS transcription. This further implies that for some genes, Src-induced tyrosine phosphorylation converts TFII-I into a transcriptional co-repressor. Importantly, the expression of WT or Y248/249F TFII-I, or TFII-I siRNA, was not sufficient to either induce or suppress morphological transformation in NIH3T3 or Src/3T3 cells, respectively, or to suppress the loss of contact-inhibited growth in Src/3T3 cells (Fig. 3).
Although we demonstrated that TFII-I is involved in the transcriptional repression of SSeCKS by Src, SSeCKS downregulation most likely requires additional factors (e.g., HDAC1, as we described previously14), as antagonism of TFII-I expression or phosphorylation only led to partial restoration of the SSeCKS expression levels found in untransformed NIH3T3 cells (Figs. 2c and 2f). Indeed, several of the minor binding proteins we identified in the MS analysis in Figure 1a could contribute to SSeCKS downregulation by Src. For example, the Myb binding protein 1a (in the “Src 1a” band) can function as a transcriptional co-repressor with NF-6B,37 and the UHRF1 E3 ubiquitin ligase (in the “Src 1c” band) is known to repress the transcription of the Tip60 gene through an interaction of HDAC1.38 It is also possible that the partial restoration of TFII-I was due to limited siRNA effects (Fig. 2c) or expression levels of exogenous TFII-IY248/249F (Fig. 2f).
We showed previously that the overexpression of USF1, Sp1 and Sp3 resulted in the induction of the αSSeCKS promoter, driven by cognate E- and GC-box cis-acting motifs in the −106/−47 proximal promoter14 (Fig. 4). EMSA and ChIP analyses indicated that the binding activity of Sp1/Sp3 for this proximal promoter increased in Src-transformed cells, and given that the downregulation of SSeCKS in these cells was sensitive to histone deacetylase inhibitors, we were able to show that the recruitment of HDAC1 helped convert USF1/Sp1/Sp3 to a repressor complex in the Src/3T3 cells. However, the fact that overexpression of all these proteins in untransformed cells only led to marginal downregulation of αSSeCKS promoter activity caused us to believe that additional repressive factors were critical for the greater (15-fold) SSeCKS downregulation by Src. The current data strongly suggest that recruitment of a Src-phosphorylated TFII-I enhances the transcriptional repression of SSeCKS (Fig. 4). Thus, the transcriptional downregulation of so-called Type II tumor suppressors such as SSeCKS/Gravin/AKAP12 can be facilitated by multiple, additive mechanisms during the process of oncogenic progression.
Our data suggest that the tyrosine phosphorylation of TFII-I by v-Src induces its association with a Sp1/Sp3-USF1-HDAC1 complex on the SSeCKS proximal promoter, resulting in the transcriptional repression of the SSeCKS/Gravin/AKAP12 metastasis suppressor gene (Fig. 4).