ABSTRACT
Retinoic acid (RA) inhibits matrix metalloproteinase 9 (MMP-9) expression due to AP-1 inhibition resulting from retinoic acid receptors (RARs) competing for limiting amounts of coactivator proteins. However, given the rapid kinetics of MMP-9 transcription, it seems unlikely that these interactions can be explained passively. Our previous studies indicated that coactivator and transcription factor phosphorylation may allow for rapid regulation of MMP-9 expression. In the present study we tested this hypothesis directly. CREB binding protein (CBP) and p300/CBP-associated factor (PCAF) were displaced from transcription factor binding sites on the MMP-9 promoter within minutes of RA treatment. The RAR interaction domains of CBP and PCAF were not required for this displacement. RA and epidermal growth factor had opposing effects on phosphorylation of CBP by extracellular signal-regulated kinase 1 that correlated with altered CBP occupancy of AP-1 sites and differential MMP-9 promoter activation. We identified a novel phosphorylation site in the CBP carboxyl terminus that mediated association with AP-1 sites in the MMP-9 promoter. Inhibition of c-jun phosphorylation displaced PCAF from AP-1 sites and reduced promoter activity. Phosphorylation deficient c-jun was less able to recruit PCAF to AP-1 sites. We also demonstrated novel interactions between coactivators and AP-1 proteins. We propose that extracellular signal-mediated coactivator exchange at AP-1 sites is mediated via protein kinase pathways.
Invading tumor cells must secrete proteolytic enzymes to degrade basement membranes (21). Activation of genes that regulate protease expression must be precisely coordinated for invasion to occur. The chemotherapeutic drug retinoic acid (RA) and its synthetic derivatives inhibit invasion in large part by decreasing matrix metalloproteinase (MMP) expression (13, 29, 32, 35, 36). Increased expression of MMPs has been demonstrated in many invasive tumors (for reviews, see references 9, 15, and 25). Members of the MMP family include MMP-1 (interstitial collagenase), MMP-2 and -9 (gelatinases A and B), MMP-3 (stromelysin), and MMP-12 (metalloelastase).
The expression of some MMP genes is regulated by AP-1 proteins. AP-1 is a sequence-specific transcription factor complex composed of members of the fos and jun families (for a review, see reference 18). These proteins belong to the bZIP DNA-binding proteins and associate to form homo- and heterodimers that bind to cognate sites in the promoters of target genes. AP-1 activity is induced by a variety of extracellular stimuli (for a review, see reference 2). These fos/jun complexes have similar DNA binding activities and specificities. RA is a potent inhibitor of AP-1 responsive gene expression in many cell types (31, 33). RA is believed to inhibit AP-1 activity by receptor competition for the coactivator protein CBP (17).
Coactivators such as CREB binding protein (CBP) and its close relative p300 interact with both nuclear hormone receptors and AP-1 family members (6, 17). CBP inactivation leads to tumor formation in transgenic mice and humans (19, 26). CBP was subsequently found to have histone acetyltransferase (HAT) activity, allowing for histone disassembly and activation of transcription (28). These studies led to the discovery of HAT activity in other coactivator proteins, such as the p300/CBP-associated factor (PCAF [37]). PCAF can associate with CBP and with nuclear receptors independently of CBP (5). p300/CBP has been shown to activate the collagenase type I gene via AP-1 sites (24). CBP also was required for the activation of RA responsive genes, and inhibition of AP-1 activity by RA was attributed to RAR competition for limiting amounts of CBP. However, given the rapid kinetics of altered gene transcription in these model systems, it seems unlikely that the interaction between coactivators, AP-1, and RAR proteins can be explained passively (see below).
Previous studies have shown that CBP recruitment was affected by phosphorylation of serine 133 in the amino terminus of the protein (7). CBP is phosphorylated by extracellular signal-regulated kinase 1 (ERK1) both in vivo and in vitro (1, 22). Phosphorylation of the amino or carboxyl termini affected both CBP recruitment and transactivation in some models (12, 38), but the required residues were not always defined. Alternatively, transcription factor phosphorylation has been shown to affect interaction with CBP. Mutation of serine residues 63 and 73 in c-jun reduced CBP binding and transactivation in vitro (3). Phosphorylation of a number of ets family members has been shown to increase CBP/p300 recruitment (10, 16, 20, 27, 30). We previously demonstrated that phosphorylation of the AP-1 protein c-jun is a potent activator of MMP-9 expression, suggesting a novel mechanism by which this modification may recruit coactivator proteins (8). These results suggest that phosphorylation of transcription factors and coactivators may provide a rapid means of regulating histone acetylation, chromatin unwinding, and induction of gene transcription. In addition, it is not known whether CBP and PCAF interact with AP-1 and ets transcription factors independently of each other or whether they are functionally interchangeable in transcriptional activation of the MMP-9 gene. We propose a novel mechanism of MMP-9 promoter regulation through extracellular signal-mediated coactivator exchange at AP-1 sites via the modulation of protein kinase pathways.
MATERIALS AND METHODS
Cell culture and stable transfections.The squamous cell carcinoma lines used in the present study were purchased from the American Type Culture Collection and cultured in Dulbecco modified Eagle medium, 10% fetal bovine serum, and 40 μg of gentamicin/ml in a humidified atmosphere of 5% CO2 at 37°C. Cultures were treated with 100 nM all trans-retinoic acid (Sigma) or vehicle for up to 24 h. Cultures were treated individually with 10 μM PD98059 (MEK inhibitor; Alexis Biochemicals) or 5 μM SP600125 (JNK inhibitor; Calbiochem) for up to 24 h. Other cultures were treated with 10 ng of epidermal growth factor (EGF)/ml for up to 24 h. SCC12 and SCC71 cells were transfected with 5 μg of CBP, phosphorylation-deficient CBP (Ser2079 and Ser2080 replaced by Ala), CBP lacking the nuclear receptor interaction domain (amino acid residues 1 to 101), PCAF, PCAF lacking the RAR interaction domain (amino acid residues 529 to 832), c-jun, or phosphorylation-deficient c-jun (Ser63 and Ser73 replaced by Ala) expression vectors. Cells were selected in 400 μg of G418/ml for 14 days. Resistant clones were picked for expansion and characterization.
Reverse transcription-PCR (RT-PCR).RNA was extracted from CBP and PCAF clones by using a commercially available kit (Qiagen) and reverse transcribed using SuperScript II reverse transcriptase according to the manufacturer's instructions (Invitrogen). cDNA was amplified using MMP-9 specific primers (5′-CCCTGCCAGTTTCCATTCATC-3′ and 5′-CCCACTTCTTGTCGCTGTCAAAG-3′) in 20 mM Tris-HCl (pH 8.3), 1.5 mM MgCl2, 63 mM KCl, 0.05% Tween 20, 1 mM EGTA, 50 μM concentrations of each deoxynucleoside triphosphate, and 2.5 U of Taq DNA polymerase (Roche Molecular Biochemicals). Amplification with β-actin cDNA using primers 5′-ACAGGAAGTCCCTTGCCATC-3′ and 5′-ACTGGTCTCAAGTCAGTGTACAGG-3′ as the internal control was carried out by real-time PCR (iCycler; Bio-Rad) using cycle parameters of 94°C for 25 s, 55°C for 1 min, and 72°C for 1 min.
Chromatin immunoprecipitation.SCC25 cells which expressed both CBP and PCAF proteins and SCC71 stable clones expressing wild-type or mutant constructs were treated with 100 nM RA or vehicle for up to 24 h. After a wash in phosphate-buffered saline (PBS), cells were fixed in 1% formaldehyde for 10 min at room temperature. Cells were washed in PBS and lysed in immunoprecipitation buffer containing protease inhibitors for 30 min at 4°C, sheared, and centrifuged at 10,000 × g for 10 min. Supernatants were cleared with 2 μg of sheared salmon sperm DNA, 20 μl of preimmune serum, and 20 μl of protein A/G-Sepharose beads for 2 h at 4°C. Aliquots of the supernatant were used as input DNA for normalization and amplified with β-actin PCR primers (5′-ACAGGAAGTCCCTTGCCATC-3′ and 5′-ACTGGTCTCAAGTCAGTGTACAGG-3′). Immunoprecipitation with anti-CBP, anti-PCAF, or control immunoglobulin G (IgG) antibodies (Santa Cruz Biotechnology) was performed overnight at 4°C. Preimmune IgG was used as the negative control antibody. Immunoprecipitates were washed extensively in immunoprecipitation buffer, resuspended in TE (10 mM Tris-HCl, 1 mM EDTA; pH 8), and incubated at 65°C for 6 h to reverse cross-links. The supernatants were extracted with phenol-chloroform and ethanol precipitated. After a wash in 70% ethanol, pellets were dried and suspended in 50 μl of TE. For PCR, 1 μl of template was amplified in buffer containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 2.5 mM MgCl2, 200 nM concentrations of each deoxynucleoside triphosphate, and 100 ng of each primer flanking the −79 (5′-CCCTCCCTTTCATACAGTTCCC-3′ and 5′-TGTGTGTGTGTGTGTGTGTGGC-3′) and −527/−540 (5′-AAGACATTTGCCCGAGGTCC-3′ and 5′-AAAGTGATGGAAGACTCCCTGAGAC-3′) regions of the MMP-9 promoter. The optimized cycle parameters were one cycle at 94°C for 3 min, followed by 25 cycles of 94°C for 25 s, 55°C for 60 s, and 72°C for 60 s, and then one final cycle at 72°C for 10 min. Amplified products were separated by agarose gel electrophoresis.
Immunoprecipitation and in vitro kinase assays.Cultures were lysed in 50 mM HEPES (pH 7.5), 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 1 mM dithiothreitol, 1% Nonidet P-40, 10% glycerol, 1 mM NaF, 0.1 mM sodium orthovanadate, and protease inhibitors for 30 min at 4°C. Lysates were centrifuged at 10,000 × g for 10 min, and anti-CBP antibody (Santa Cruz Biotechnology) was incubated with the supernatants for 1 h at 4°C. Preimmune IgG was used as a negative control antibody for immunoprecipitation. Antigen-antibody complexes were precipitated with anti-CBP, anti-PCAF, anti-RARα, anti-c-jun, or control IgG antibodies and protein A/G-agarose beads for 1 h at 4°C. Immunoprecipitated proteins were washed three times with 1 ml of lysis buffer. Samples were boiled in Laemmli buffer for 3 min, separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and blotted onto polyvinylidene difluoride membranes. Blots were incubated with anti-NCoR, anti-SMRT, anti-phosphoserine, anti-c-jun, anti-ets1, anti-PCAF, anti-c-fos, anti-Fra1, anti-JunD, and anti-JunB antibodies, followed by anti-CBP or anti-RARα antibodies, to ensure equal amounts of immunoprecipitated protein in each lane. Bands were quantitated by densitometry.
For in vitro kinase assays, ERK1 protein was immunoprecipitated from treated SCC71 cells and incubated in kinase buffer (50 mM HEPES [pH 7.5], 10 mM MgCl2, 1 mM dithiothreitol, 2.5 mM EGTA, 0.1 mM sodium orthovanadate, 1 mM NaF, 20 μM ATP, and 10 μCi of [γ-32P]ATP [3,000 Ci/mmol; Dupont]) for 30 min at 30°C with biotinylated CBP peptide corresponding to amino acid residues 2064 to 2097 encompassing the novel phosphorylation site or mutant peptides in which the serine residues were replaced by alanine. Samples were blotted as described above and exposed to Kodak XAR5 film at −80°C for 4 h, followed by incubation with anti-ERK1 antibody to ensure equal amounts of immunoprecipitated protein in each lane.
Western blot.A total of 75 μg of total cellular protein was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 10% resolving gels under denaturing and reducing conditions. Separated proteins were electroblotted to polyvinylidene difluoride membranes according to the manufacturer's recommendations (Roche Molecular Biochemicals). Blots were incubated with antibodies to CBP, Fra1, JunB, c-jun, NCoR, or SMRT for 16 h at 4°C. After a wash in Tris-buffered saline containing 0.1% Tween 20 (TBST [pH 7.4]), blots were incubated for 30 min at room temperature with anti-IgG secondary antibody conjugated to horseradish peroxidase. After extensive washing in TBST, the bands were visualized by the enhanced chemiluminescence method (Roche Molecular Biochemicals).
Transient-transfection and reporter gene analysis.Triplicate cultures of 50% confluent SCC71 cells were transiently transfected with 5 μg of the indicated human MMP-9-CAT promoter/reporter vectors along with 2 μg of CBP, PCAF, Fra1, JunB, c-jun, or blank expression plasmids using Lipofectamine according to the manufacturer's recommendations (Invitrogen). Four mutant promoter constructs containing double point mutations in the −79 and −533 AP-1 and −527 and −540 ets sites were transfected separately with expression vectors. In some experiments, cells were transfected with RARα or control siRNAs. Then, 1 μg of β-galactosidase expression plasmid was used to normalize for transfection efficiency. Cells were harvested and reporter gene activity determined by using a commercially available kit (Promega). Chloramphenicol acetyltransferase (CAT) activity was normalized to β-galactosidase levels for each sample.
RARα expression was inhibited by transient transfection of small interfering RNA (siRNA) according to the manufacturer's recommendations (Dharmacon). Mock and control siRNA-transfected cultures were used as controls. Cultures were treated with 100 nM RA or vehicle for 1 h prior to chromatin immunoprecipitation.
Yeast two-hybrid analysis.The pAS2-1 GAL4 DNA-binding domain yeast vector (Clontech) was used for subcloning the CBP and PCAF cDNAs. The PCAF cDNA was subcloned into the pACT2 GAL4 activation domain yeast vector. pAS2-1 carries the TRP1 selection marker, and the pACT2 vector carries the LEU2 selection marker. Plasmid pCL1, which encodes the full-length GAL4 transcription factor, was used as the positive control for activation of the yeast reporter genes. Plasmids pVA3 and pTD1, which encode fusion proteins of p53-GAL4 DNA-binding domain and simian virus 40 large T antigen-GAL4 activation domain, respectively, were used as the positive controls for protein interaction. pVA3 and pTD1 carry the TRP1 and LEU2 selection markers, respectively. Plasmid pLAM5′-1, which carries the TRP1 selection marker and encodes a lamin C-GAL4 DNA-binding domain fusion protein, was used with pTD1 as negative controls for the interaction. CBP/pAS2-1 and PCAF/pAS2-1 constructs were cotransformed individually with c-jun/pACT2, Fra1/pACT2, Fra2/pACT2, and FosB/pACT2 constructs into Saccharomyces cerevisiae strain Y190 according to the manufacturer's recommendations (Clontech). The positive and negative control plasmids also were transformed into this strain. Y190 strain contains the GAL4-driven reporter genes HIS3 and lacZ, which allow for detection of protein interactions through growth on histidine-deficient medium and measurement of the β-galactosidase activity. The transformation mixtures were plated onto amino acid-deficient medium (−leu/−trp for cotransformation and −leu/−trp/−his for protein interaction) to select transformants. Galactosidase activity in selected transformants was measured by luminometry.
RESULTS
To determine the effects of coactivator expression on MMP-9 gene transcription, we performed real-time RT-PCR on RA and vehicle-treated SCC25 cells (expressing CBP and PCAF) and the SCC71 line (lacking CBP and PCAF protein expression) (34). As shown in Fig. 1A, RA-mediated inhibition of MMP-9 expression was evident as early as 2 h after treatment in coactivator positive SCC25 cells. MMP-9 expression continued to decrease by 60% through 24 h after treatment. RA treatment had no effect on MMP-9 mRNA expression in coactivator negative SCC71 cells. These results indicate that coactivator expression may be required for RA-mediated inhibition of MMP-9 expression. Chromatin immunoprecipitation studies on RA-treated SCC25 cells showed that CBP and PCAF were displaced from the −79 AP-1 site and −527/−540 sites within 10 min of RA treatment and continued to decline throughout the 1-h time course (Fig. 1B). In contrast, RA-mediated repressor dissociation from RARα occurred with slower kinetics. RA-mediated reductions in SMRT and NCoR interaction with RARα were not observed until 30 min after RA addition (Fig. 1C). RA did not produce changes in CBP, NCoR, or SMRT protein expression (Fig. 1D). These results indicate that RA-mediated repressor displacement from RARα is likely not sufficient to induce coactivator recruitment from AP-1 sites in the MMP-9 promoter.
RA rapidly displaces CBP and PCAF coactivators from transcription factor binding sites on the MMP-9 promoter. (A) RA inhibits MMP-9 mRNA expression in CBP and PCAF expressing SCC25 but not coactivator-negative SCC71 cells. Cultures were treated with 100 nM RA for up to 24 h prior to RNA extraction for real-time RT-PCR analysis. Error bars indicate the standard error of the mean (SEM) of three independent experiments. (B) SCC25 cells were treated with 100 nM RA for up to 60 min prior to chromatin immunoprecipitation (ChIP) using anti-CBP, anti-PCAF, or control IgG antibodies. Immunoprecipitated DNA was amplified with primers to the −79 AP-1 site or −527/−540 region of the MMP-9 promoter. Genomic DNA prior to immunoprecipitation was amplified to determine relative amounts of input DNA in the samples. These experiments were performed three times with similar results. Representative gels are shown. (C) RA displaces repressor proteins from RARα and recruits coactivators with slower kinetics. SCC25 cells were treated with 100 nM RA for up to 2 h prior to immunoprecipitation with anti-RARα antibody (IP RARα). Repressor and coactivator interaction with RARα was determined by blotting of immunoprecipitated proteins with anti-NCoR, anti-SMRT, anti-CBP, and anti-PCAF antibodies. Blots were also incubated with anti-RARα antibody to determine relative amounts of immunoprecipitated protein in each lane. (D) CBP, NCoR, and SMRT protein expression is not changed by RA treatment. These experiments were performed three times with similar results. Representative blots are shown.
To determine whether the RAR interaction domains of CBP and PCAF were required for RA-mediated repression of MMP-9 promoter activity, we transiently transfected wild-type or mutant coactivator constructs with reporter vectors containing point mutations in specific transcription factor binding sites. As shown in Fig. 2A, both CBP and PCAF induced MMP-9 promoter activity by nearly threefold. RA treatment not only blocked coactivator induction but also reduced MMP-9 promoter activity by 70% compared to control levels. Point mutations of the AP-1 and ets sites inhibited basal activity of the MMP-9 promoter by two- to threefold. AP-1 site mutations blocked the ability of CBP and PCAF to induce MMP-9 promoter activity. Mutation of the −527 ets site inhibited promoter regulation by CBP and RA, but slight PCAF-mediated induction was noted. The −540 ets site point mutation inhibited basal promoter activity, but RA-mediated repression of CBP and PCAF induced activation was intact. Deletion of the RAR interaction domains of CBP and PCAF had no effect on RA-mediated repression or coactivator-mediated induction of MMP-9 promoter activity (Fig. 2B). In stable clones expressing CBP or PCAF lacking the RAR interaction domains, both mutant proteins occupied the −79 and −527/−540 sites of the MMP-9 promoter in genomic DNA (Fig. 2C). RA treatment displaced the mutant CBP and PCAF proteins from these sites beginning at 10 min and continuing throughout the 1-h time course. Inhibition of RARα expression by siRNA transfection did not affect RA-mediated displacement of coactivators from the binding sites or promoter activity (Fig. 2D and E). These results indicate that RA treatment can displace CBP and PCAF from transcription factor binding sites on the MMP-9 promoter in the absence of their ability to interact with RARs.
The CBP and PCAF nuclear receptor interaction domains are not required for RA-mediated displacement from the MMP-9 promoter. (A and B) The wild-type or mutant MMP-9-CAT promoter/reporter constructs containing point mutations in the −79 AP1, −527 ets, −533 AP1, or −540 ets were transiently transfected into SCC71 cells. CBP expression vectors with or without (mCBP) the nuclear receptor interaction domain were also cotransfected. In separate cultures, the reporter construct was transfected with PCAF expression vectors with or without (mPCAF) the nuclear receptor interaction domain. Neomycin resistance vector (vec) without the CBP or PCAF cDNAs was used as the control plasmid. Transfected cultures were treated with 100 nM RA or vehicle for 24 h prior to CAT assay as described in Materials and Methods. Error bars indicate the SEM of three independent experiments. (C) SCC71 cells stably transfected with CBP or PCAF expression vectors lacking the nuclear receptor interaction domains (mCBP or mPCAF) were treated with 100 nM RA for up to 60 min prior to chromatin immunoprecipitation (ChIP) with anti-CBP or anti-PCAF antibodies. Immunoprecipitated DNA was amplified with primers to the −79 AP-1 site or −527/−540 region of the MMP-9 promoter. Genomic DNA prior to immunoprecipitation was amplified to determine the relative amounts of input DNA in the samples. (D) CBP and PCAF are displaced from the MMP-9 promoter by RA when RARα expression is inhibited by siRNA. RARα expression is shown in control cells, cultures treated with control siRNA (i control, negative control), and those treated with RARα siRNA (iRARα, upper panel). CBP or PCAF was immunoprecipitated from these cultures treated with vehicle or 100 nM RA for 1 h prior to ChIP as described above. These experiments were performed three times with similar results. Representative gels are shown. (E) The wild-type or mutant MMP-9-CAT promoter/reporter constructs containing point mutations in the −79 AP1, −527 ets, −533 AP1, or −540 ets were transiently transfected into SCC71 cells with RARα or control siRNA. CBP or PCAF expression vectors also were cotransfected. Neomycin resistance vector (vec) without the CBP or PCAF cDNA was used as the control plasmid. Transfected cultures were treated with 100 nM RA or vehicle for 24 h prior to CAT assay as described in Materials and Methods. Error bars indicate the SEM of three independent experiments.
Given that our data indicated that RA-mediated repressor dissociation from RARα and coactivator displacement from AP-1 sites are rapid but not synchronous and that coactivator interaction domains are not required for this effect, we hypothesized that RA may regulate MMP-9 promoter activity through the phosphorylation of coactivators or AP-1 components. Treatment of cultured SCC25 cells with RA completely inhibited CBP phosphorylation within 10 to 20 min (Fig. 3A). RA and PD98059 treatment continued to inhibit CBP phosphorylation by 60% at 24 h (Fig. 3B). The jun N-terminal kinase inhibitor SP600125 had no effect on CBP phosphorylation, indicating that CBP was phosphorylated by ERK rather than JNK. In contrast, SP600125 treatment inhibited c-jun phosphorylation within 10 to 20 min (Fig. 3C). SP600125 treatment continued to inhibit c-jun phosphorylation by 70% after 24 h, while RA and PD98059 had little effect (Fig. 3D). In contrast, EGF treatment (which activates ERK1) increased CBP phosphorylation by eightfold beginning at 10 min, confirming that CBP was primarily phosphorylated via the ERK pathway (Fig. 3E). PD98059 treatment inhibited MMP-9 promoter activity by 40% compared to control levels in cells transiently transfected with the reporter construct (Fig. 3F). PD98059 treatment of cells transiently transfected with the reporter vector and CBP produced even greater repression (70%) of the MMP-9 promoter. PD98059 inhibition in PCAF-transfected cells was similar to vector-transfected control cultures, suggesting that ERK1-mediated CBP phosphorylation is critical for coactivator-mediated induction of the MMP-9 promoter. In contrast, treatment of transiently transfected SCC lines with SP600125 inhibited MMP-9 promoter activity by 30%, an effect that was independent of cotransfection with CBP or PCAF (Fig. 3G). These results indicate that ERK1-mediated phosphorylation of CBP regulates transcriptional activation or repression of this coactivator. However, JNK mediates repression of the MMP-9 promoter independently of coactivator overexpression.
CBP and AP-1 phosphorylation by extracellular signaling regulates coactivator recruitment to transcriptional complexes. SCC25 cells were treated with PD98059, SP600125, RA, EGF, or vehicle as described in Materials and Methods. CBP (A and B) or c-jun (C and D) proteins were immunoprecipitated from treated cellular lysates (IP CBP, IP c-jun). Immunoprecipitated proteins were blotted with anti-phosphoserine antibody (anti-pSer) prior to incubation with anti-CBP and anti-c-jun antibodies. The effects on CBP and c-jun phosphorylation of short-term treatment (up to 40 min) with RA and EGF (E) and longer-term exposure (24 h; B and D) to these factors or kinase inhibitors is shown. These experiments were performed three times with similar results. Representative blots are shown. (F and G) The wild-type or mutant MMP-9-CAT promoter/reporter constructs were transiently transfected into SCC71 cells. CBP or PCAF expression vectors were also cotransfected. Neomycin resistance vector (vec) without the CBP or PCAF cDNAs was used as the control plasmid. Transfected cultures were treated with PD98059, SP600125, or vehicle for 24 h prior to CAT assay as described in Materials and Methods. Error bars indicate the SEM of three independent experiments. (H) SCC25 cells were treated with PD98059 or SP600125 for up to 4 h prior to chromatin immunoprecipitation with anti-CBP or anti-PCAF antibodies. Immunoprecipitated DNA was amplified with primers to the −79 AP-1 site or −527/−540 region of the MMP-9 promoter. Genomic DNA prior to immunoprecipitation was amplified to determine relative amounts of input DNA in the samples. These experiments were performed three times with similar results. Representative gels are shown. (I) EGF induces MMP-9 mRNA expression in CBP and PCAF expressing SCC25 but not coactivator-negative SCC71 cells. Cultures were treated with 10 ng of EGF/ml for up to 24 h prior to RNA extraction for real-time RT-PCR analysis. Error bars indicate the SEM of three independent experiments. (J) EGF induces coactivator recruitment to the MMP-9 promoter. SCC25 cells were treated with 10 ng of EGF/ml for up to 180 min prior to chromatin immunoprecipitation (ChIP) with anti-CBP or anti-PCAF antibodies. Immunoprecipitated DNA was amplified with primers to the −79 AP-1 site or −527/−540 region of the MMP-9 promoter. Genomic DNA prior to immunoprecipitation was amplified to determine the relative amounts of input DNA in the samples. These experiments were performed three times with similar results. Representative gels are shown. (K) The wild-type or mutant MMP-9-CAT promoter/reporter constructs were transiently transfected into SCC71 cells. CBP or PCAF expression vectors were also cotransfected. Neomycin resistance vector (vec) without the CBP or PCAF cDNAs was used as the control plasmid. Transfected cultures were treated with 10 ng of EGF/ml or vehicle for 24 h prior to CAT assay as described in Materials and Methods. Error bars indicate the SEM of three independent experiments.
To further understand these findings, we performed CBP and PCAF chromatin immunoprecipitation on SCC25 cells treated with PD98059 or SP600125. As shown in Fig. 3H, PD98059 treatment displaced CBP but not PCAF from the −79 and the −527/−540 sites of the MMP-9 promoter. At later time points, when a fraction of CBP protein is phosphorylated, coactivator cycling back onto the promoter was observed at the −79 site. Conversely, SP600125 treatment displaced PCAF but not CBP from these sites. These changes were evident as early as 10 min after drug exposure (data not shown). These results indicate that CBP and PCAF can bind independently to AP-1 sites in the MMP-9 promoter, that CBP phosphorylation by ERK1 is an important mediator of this interaction, and that JNK activity is required for PCAF interaction with these sites.
Our data indicate that mitogen-activated protein kinase (MAPK)-dependent phosphorylation is important for CBP and PCAF binding to AP-1 sites in the MMP-9 promoter. To confirm these results, we treated coactivator-positive SCC25 cells with EGF, which activates ERK1 and induces MMP-9 expression in this line (Fig. 3I) (36). In contrast, MMP-9 expression was not induced by EGF in coactivator-negative SCC71 cells. As shown in Fig. 3J, EGF induced coactivator complex formation at the −79 and −527/−540 sites of the MMP-9 promoter within 15 min of exposure. Increased CBP and PCAF binding to these sites (up to 16-fold higher) was observed throughout the 3-h time course; greater recruitment was observed for the −79 site relative to the −527/−540 sites. EGF treatment of cultures transiently transfected with the MMP-9 reporter construct induced promoter activity by more than twofold (Fig. 3K). Cotransfection of CBP or PCAF also induced MMP-9 promoter activity, and EGF treatment produced an additive effect in cells expressing CBP (fivefold induction). This EGF-dependent effect was not observed in PCAF-transfected cultures. Mutation of either AP-1 or ets sites markedly inhibited coactivator and EGF induction of MMP-9 promoter activity. These results indicate that EGF-mediated phosphorylation of CBP recruits the coactivator to transcription factor binding sites in the MMP-9 promoter and is important for growth factor-mediated induction of this gene.
CBP is phosphorylated at both its amino and carboxyl termini (1, 12, 22, 38). To determine whether additional ERK phosphorylation sites existed in CBP, we performed computer-assisted searches of the coactivator amino acid sequence. We identified a previously unreported potential ERK phosphorylation site at Ser2079 and Ser2080 in the CBP carboxyl terminus. Phosphorylation of the Ser2080 site was inhibited by RA and PD98059 but induced by EGF treatment, as shown by in vitro kinase assays (Fig. 4A). We replaced the residues with Ala by site-directed mutagenesis. Expression of the mutant CBP and its interaction with RARα is shown in Fig. 4B and C. The phosphorylation mutant CBP was unable to induce MMP-9 promoter activity compared to the wild-type coactivator in transient-transfection assays (Fig. 4D). We stably transfected the mutant and wild-type CBP expression vector into SCC lines with the PCAF construct and performed chromatin immunoprecipitation with anti-CBP and anti-PCAF antibodies. The mutant CBP failed to interact with the −79 or −527/−540 sites of the MMP-9 promoter (Fig. 4E). Overexpression of wild-type CBP increased coactivator occupancy of these sites by twofold. Interestingly, the expression of both wild-type and mutant CBP increased complex formation with transfected PCAF protein at the −79 and −527/−540 sites of the MMP-9 promoter, suggesting that PCAF can bind to the CBP phosphorylation mutant as well as wild-type coactivator. These results indicate that a novel phosphorylation site is critical for CBP interaction with transcription factor binding sites in and activation of the MMP-9 promoter.
A phosphorylation-deficient CBP mutant fails to form transcriptional complexes on the MMP-9 promoter. (A) ERK1-mediated phosphorylation of the novel site (wt peptide) in response to kinase inhibitors, RA, and EGF is shown by in vitro kinase assay as described in Materials and Methods. Mutant peptides containing Ser-to-Ala substitutions (S2079A, S2080A, S2079/S2080) also were used as substrates in the kinase assay. (B and C) Expression of CBP and a phosphorylation-deficient CBP mutant protein (B), along with their interaction with RARα in SCC71 cells (C), is shown. (D) The wild-type or mutant MMP-9-CAT promoter/reporter constructs were transiently transfected into SCC71 cells. Wild-type CBP or phosphorylation-deficient CBP mutant (pmCBP) vectors were cotransfected as described in Materials and Methods. Neomycin resistance vector (vec) without the CBP cDNA was used as the control plasmid. Error bars indicate the SEM of three independent experiments. (E) SCC71 clones expressing either wild-type CBP or the phosphorylation-deficient mutant CBP (pmCBP) were subjected to chromatin immunoprecipitation (ChIP) with anti-CBP or anti-PCAF antibodies. Immunoprecipitated DNA was amplified with primers to the −79 AP-1 site or −527/−540 region of the MMP-9 promoter. Genomic DNA prior to immunoprecipitation was amplified to determine relative amounts of input DNA in the samples. These experiments were performed three times with similar results. Representative gels are shown.
It is possible that treatment with kinase inhibitors or coactivator overexpression may affect MMP-9 promoter activity through changes in AP-1 protein expression. To examine this possibility, we treated SCC lines with PD98059 or SP600125 and compared changes in AP-1 protein expression to levels in stable clones expressing CBP or PCAF. As shown in Fig. 5A, expression of both Fra1 and JunB proteins was suppressed by PD98059 (two-fold for Fra1, sevenfold for JunB) and SP600125 (threefold for Fra1, eightfold for JunB). Expression of these two AP-1 proteins was also suppressed in CBP (eightfold for Fra1, twofold for JunB) and PCAF (twofold for Fra1, sixfold for JunB) stable clones. Expression of other AP-1 proteins was not affected by kinase inhibitors and coactivators (e.g., c-jun in lower panel) or was not detected by Western blotting in these experiments (data not shown). To determine how these changes in AP-1 expression affected MMP-9 promoter activity, we transiently transfected Fra1 and JunB vectors with the reporter construct, followed by treatment with PD98059 or SP600125. As shown in Fig. 5B, Fra1 repressed MMP-9 promoter activity by 30% via AP-1 sites, while JunB overexpression had no detectable effect. These JunB results were consistent with our previous results showing lack of JunB interaction with the −79 site (8). Fra1 and JunB overexpression had no effect on PD98059-mediated repression of MMP-9 promoter activity. Similarly, Fra1 and JunB had no effect on SP600125-mediated repression of MMP-9 promoter activity (Fig. 5C). These results indicate that changes in AP-1 protein expression are not likely to induce the observed changes in MMP-9 promoter activity induced by coactivator expression or kinase inhibitor treatment.
Coactivator-mediated changes in AP-1 protein expression are not responsible MMP-9 promoter activation. (A) SCC71 clones expressing CBP and PCAF were treated with PD98059, SP600125, or vehicle (control) for 24 h prior to Western blotting with anti-AP-1 antibodies. Changes in Fra1 and JunB expression due to drug treatment were similar to those observed in stable clones expressing CBP and PCAF. AP-1 expression in these clones was compared to neomycin-resistant control cells (neo). c-jun protein levels that were not affected by drug treatment or coactivator expression were used as the gel loading control. (B and C) The wild-type or mutant MMP-9-CAT promoter/reporter constructs were transiently transfected into SCC71 cells. Fra1, JunB, or neomycin resistance (vec) vectors were cotransfected as described in Materials and Methods. Cultures were treated with either PD98059 or SP600125 for 24 h prior to CAT assay. Error bars indicate the SEM of three independent experiments.
Our results indicate that PCAF enhances MMP-9 promoter activity via AP-1 sites and that the JNK inhibitor SP600125 displaces PCAF from these promoter elements. To determine whether PCAF could interact directly with the AP-1 protein c-jun, we immunoprecipitated CBP and PCAF from cells treated with SP600125, PD98059, or vehicle. As shown in Fig. 6A, c-jun immunoprecipitated with both CBP and PCAF regardless of kinase inhibitor treatment. ets1, which is phosphorylated by ERK1, also immunoprecipitated with both coactivator proteins. Given that previous studies have shown that c-jun, ets1, and PCAF interact with CBP directly, it is possible that the coimmunoprecipitation results may be due to binding to CBP. To provide more direct evidence for PCAF interaction with phosphorylated c-jun, we transiently transfected PCAF, c-jun, or mutant c-jun containing point mutations at the Ser63/73 phosphorylation sites along with the MMP-9 reporter construct into SCC71 cells. As shown in Fig. 6B, c-jun cotransfection with PCAF induced MMP-9 promoter activity by twofold compared to PCAF and control vector. However, cotransfection with the phosphorylation mutant c-jun inhibited PCAF-mediated induction of the MMP-9 promoter. These effects were mediated via the AP-1 sites of the promoter. We performed chromatin immunoprecipitation using anti-CBP and PCAF antibodies with lysates from stable c-jun and phosphorylation mutant c-jun clones (Fig. 6C). Both CBP and PCAF interacted poorly with the −79 AP-1 site in mutant but not wild-type c-jun clones (eightfold decrease). This decrease was less apparent at the −527/−540 sites, which contains two ets recognition elements. These results indicate that inhibiting c-jun phosphorylation decreases the ability of PCAF to interact with and transactivate AP-1 sites in the MMP-9 promoter.
c-jun phosphorylation regulates coactivator interaction with AP-1 sites on the MMP-9 promoter. (A) SCC25 cells were treated with PD98059, SP600125, or vehicle (control) for 16 h prior to immunoprecipitation with anti-CBP or anti-PCAF antibodies. Immunoprecipitated complexes were blotted with anti-c-jun antibody to determine protein interactions. Anti-ets1 antibody was used as the positive control for coactivator interaction. These experiments were performed three times with similar results. Representative blots are shown. (B) The wild-type or mutant MMP-9-CAT promoter/reporter constructs were transiently transfected into SCC71 cells. PCAF, c-jun, a phosphorylation-deficient c-jun (pmc-jun), or neomycin resistance (vec) vectors were cotransfected as described in Materials and Methods prior to a CAT assay. Error bars indicate the SEM of three independent experiments. (C) SCC71 stable clones expressing c-jun, phosphorylation-deficient c-jun (pmc-jun), or neomycin resistance vector were subjected to chromatin immunoprecipitation (ChIP) with anti-CBP or anti-PCAF antibodies. Immunoprecipitated DNA was amplified with primers to the −79 AP-1 site or −527/−540 region of the MMP-9 promoter. Genomic DNA prior to immunoprecipitation was amplified to determine relative amounts of input DNA in the samples. These experiments were performed three times with similar results. Representative gels are shown. (D) c-jun interacts with both CBP and PCAF. CBP/jun and PCAF/jun interactions were determined by yeast two-hybrid assay as described in Materials and Methods. Blank vectors were used as the negative controls (con), and pCL-1 plasmid was used as the positive control for β-galactosidase activity (relative light units). Plasmids pVA3 and pTD1 were used as the positive interaction controls; pTD1 and pLAM were used as the negative interaction controls. Error bars indicate the SEM of three independent experiments. (E) CBP interacts with additional AP-1 proteins (Fra1, Fra2, FosB) as determined by yeast two-hybrid analysis. Fra1, Fra2, and FosB amino (N)- or carboxyl (C)-terminal protein interaction domains were mapped to the CBP CREB binding domain or zinc finger domain. The control plasmids are described above, and the β-galactosidase activity was measured in relative light units. Error bars indicate the SEM of three independent experiments.
We used the yeast two-hybrid system to examine these interactions in a different phosphorylation-permissive environment. As shown in Fig. 6D, c-jun interacted with both CBP and PCAF in yeast cells, while none of these constructs activated the endogenous reporter gene alone. Taken together with the chromatin immunoprecipitation and transactivation data, these results indicate that c-jun can regulate the interaction of PCAF with AP-1 sites in the MMP-9 promoter. We used the yeast two-hybrid system to examine novel interactions of AP-1 proteins with CBP. As shown in Fig. 6E, the AP-1 proteins Fra1, Fra2, and FosB primarily interacted with the third zinc finger domain of CBP, although Fra1 and the amino terminus of Fra2 showed weak interactions with the CREB binding domain of CBP. The Fra2 carboxyl terminus interacted more strongly with the CBP zinc finger domain than the amino terminus, suggesting that the amino-terminal interaction with both CBP domains may be nonspecific. The interaction of FosB with CBP was primarily through the amino terminus, as shown by the carboxyl-terminal truncation mutant. These results indicate that, in addition to c-fos, c-jun, and JunB, other AP-1 proteins (namely, Fra1, Fra2, and FosB) interact with CBP in the two-hybrid system.
DISCUSSION
One of the important findings of the present study is that disparate extracellular signals can induce rapid coactivator association and dissociation from transcription factor binding sites in the MMP-9 promoter (Fig. 7). Coactivators such as CBP were required for activation of RA-responsive genes, and inhibition of AP-1 activity by RA was attributed to RAR competition for limiting amounts of CBP (4, 11). In the case of retinoic acid signaling, the displacement of coactivator binding occurs with faster kinetics than repressor dissociation from RARα. These results are borne out by our results indicating that coactivator domains which interact with RARs are not required for coactivator displacement. These results indicate that additional mechanisms must induce coactivator dissociation from transcription factor binding sites. In the human stromelysin promoter, CBP was shown to interact with both ets-1 and ets-2 binding sites; this was stimulated by phosphorylation of ets-2 at threonine 72 (16). Activation of the MMP-9 gene by phorbol esters required intact transcription factor binding sites and assembly of coactivators in the distal promoter (23). Our results indicate that physiologic inhibition (via RA) and induction (by EGF) of MMP-9 gene transcription is regulated by rapid coactivator association and dissociation with AP-1 sites in the proximal promoter region.
Extracellular signal regulation of coactivator release from transcription factor binding sites. CBP is displaced from AP-1 sites via inhibition of ERK1-mediated phosphorylation at novel residues by RA or PD98059 treatment prior to recruitment to RAR. This mechanism provides a means for rapidly dissociating coactivators from one response element for recruitment to another. JNK1-mediated phosphorylation of c-jun also regulates interaction of PCAF with AP-1 sites.
Our data show that phosphorylation of both CBP and transcription factors (i.e., c-jun) regulates coactivator association and dissociation. We identified a novel CBP phosphorylation site at Ser2080 that regulates coactivator association with AP-1 sites in the MMP-9 promoter. CBP cycling back to AP-1 sites may be due to the return of coactivator phosphorylation at later time points. A previous study showed that CBP was phosphorylated by CaM kinase IV in the coactivator amino terminus (7). Growth factors also phosphorylated CBP via the MAPK pathway in the amino terminus of the coactivator (12, 22, 38). Recently, a separate p300 phosphorylation site at Ser1834 was shown to be a target of Akt induction of the ICAM-1 promoter (14). Transcription factor phosphorylation has also been shown to regulate coactivator binding. Mutation of the Ser 63/73 phosphorylation sites in c-jun reduced CBP binding and transactivation (3). However, interactions of mutant c-jun with CBP on chromatin may differ from those of soluble proteins in solution. Similarly, phosphorylation of the ets family of transcription factors has been shown to improve CBP/p300 association and transactivation (10, 20, 27, 30). Our data using the MMP-9 promoter as the target gene showed that CBP phosphorylation rather than that of its associated transcription factors was important for transcriptional regulation of the gene.
We showed that c-jun interacted with both CBP and PCAF in yeast and mammalian cells. However, the composition of AP-1 heterodimers bound to their promoters may change depending on cell context, thereby affecting interaction with coactivators. We used the yeast two-hybrid system to examine novel interactions of AP-1 proteins with CBP. Fra1, Fra2, and FosB primarily interacted with the third zinc finger domain of CBP and Fra1, and the amino terminus of Fra2 showed weak interactions with the CREB binding domain of CBP. The Fra2 carboxyl terminus interacted more strongly with the CBP zinc finger domain than the amino terminus. The interaction of FosB with CBP was primarily through the amino terminus, as shown by the carboxyl-terminal truncation mutant. These results indicate that in addition to c-fos, c-jun, and JunB, other AP-1 proteins interact with CBP in the two-hybrid system.
In summary, MMP-9 gene expression is regulated through extracellular signal-mediated coactivator exchange at AP-1 sites via modulation of protein kinase pathways. Future experiments will determine how altered phosphorylation of coactivator and AP-1 proteins regulate the availability and interaction of protein and DNA-binding domains, thus contributing to transcriptional complex formation and histone acetylation at the proximal MMP-9 promoter.
ACKNOWLEDGMENTS
We thank Ronald Evans (Salk Institute for Biological Studies, La Jolla, CA) for the CBP and PCAF expression vectors, Tom Curran (St. Jude Children's Research Hospital, Memphis, TN) for the c-jun expression plasmid, Dany Chalbos (INSERM, Montpellier, France) for the Fra1 plasmid, Yasuko Yamamura (Tokyo Medical University, Tokyo, Japan) for the FosB vector, Jane McHenry (Australian National University, Canberra, Australia) for the Fra2 constructs, and Hiroshi Sato (Kanazawa University, Kanazawa, Japan) for the MMP-9 promoter construct.
This study was supported by National Institutes of Health grants to D.L.C.
FOOTNOTES
- Received 18 August 2007.
- Returned for modification 27 December 2007.
- Accepted 16 April 2008.
↵▿ Published ahead of print on 28 April 2008.
- American Society for Microbiology
