Obligatory Role of Cyclic Adenosine Monophosphate Response Element in Cyclooxygenase-2 Promoter Induction and Feedback Regulation by Inflammatory Mediators
Background— Cyclooxygenase-2 (COX-2) plays a key role in human inflammatory disorders such as vascular inflammation. COX-2 promoter activity is induced by proinflammatory mediators, but the role of cyclic adenosine monophosphate response element (CRE) in promoter stimulation remains unclear.
Methods and Results— Transient transfection of a 0.9-kb COX-2 promoter fragment bearing CRE mutation abrogated COX-2 promoter activity induced by proinflammatory mediators in human endothelial cells and fibroblasts. Dual mutations of CRE and an upstream CCAAT/enhancer binding protein (C/EBP) site did not have an additional effect. Binding of CREB-2, ATF-2, USF-2, and c-Jun transactivators to a wild-type and CRE-mutated oligonucleotide was analyzed by a novel DNA-binding assay. CREB-2 and ATF-2 in nuclear extracts of unstimulated endothelial cells bound to CRE, whereas USF-2 and c-Jun or c-Fos bound to non-CRE sites. CREB-2 and c-Fos binding was increased by phorbol 12-myristate 13-acetate but not tumor necrosis factor-α. The binding assay and chromatin immunoprecipitation revealed binding of P300 coactivator to the COX-2 promoter region.
Conclusions— CRE plays an obligatory role in COX-2 promoter activation by diverse stimuli. CREB-2 and ATF-2 bound to CRE serve as an anchor for P300 interaction with upstream transactivators and downstream transcription machinery.
Received January 16, 2002; revision received March 28, 2002; accepted March 28, 2002.
Cyclooxygenase-2 (COX-2) is the inducible COX isoform that catalyzes the formation of prostaglandins in response to mitogens and proinflammatory cytokines.1,2⇓ Robust COX-2 protein expression by these factors contributes to the synthesis of abundant prostaglandins that mediate diverse pathophysiological processes such as inflammation, tissue injury, and tumor growth.3–7⇓⇓⇓⇓ COX-2 expression induced by diverse stimuli occurs mostly at the transcriptional level and also posttranscriptionally. COX-2 is encoded by a single copy of 7.5-kb genomic DNA comprising 10 exons.8 The genomic structure of human COX-2 is highly homologous to that of murine COX-2.9 The 5′-flanking promoter regions of murine and human COX-2 have a canonical TATA-box and multiple regulatory elements, but there exist differences in the cis-acting regulatory elements. For example, the proximal promoter region of murine COX-2 has a single nuclear factor-κB (NF-κB) site,9 whereas the human gene has 2 putative NF-κB–binding sites.10 On the other hand, murine promoter has 2 CCAAT/enhancer binding protein (C/EBP) sites, whereas the human gene has only 1 C/EBP site.9,10⇓ These differences are likely to influence transcriptional regulation of COX-2 in response to diverse stimuli. Mouse COX-2 promoter function is more extensively characterized, and several regulatory elements in the 5′-flanking region, notably C/EBP sites, have been shown to be involved in COX-2 induction.11 In a few recent studies on human COX-2 promoter, NF-κB sites were shown to be involved in hypoxia- and cytokine-induced expression, whereas C/EBP sites were involved in phorbol 12-myristate 13-acetate (PMA)–induced and interleukin-1 (IL-1)–induced COX-2 expression.12–14⇓⇓ A conserved cyclic adenosine monophosphate (cAMP) response element (CRE) site has been shown to be required for COX-2 expression in mouse cells induced by viral oncogenes and mitogenic factors15 and in rabbit chondrocytes induced by IL-1β16 but not in rat ovarian cells induced by hormones.17 These results are consistent with species- and stimuli-specific COX-2 transcriptional regulation and underscore the importance of characterizing human COX-2 promoter function.
We are interested in elucidating the mechanisms by which human COX-2 transcription is regulated by proinflammatory mediators. Although recent studies from our laboratories and others have shed light on the role of C/EBP and NF-κB in COX-2 promoter regulation,14 except for a study showing that CRE is crucial for COX-2 expression in PMA-induced differentiation of U937 cells, 18 the role of CRE in human COX-2 promoter function is largely unknown. In this study, we determined the role of CRE and its binding proteins in human COX-2 expression induced by diverse proinflammatory mediators. Our results revealed that CRE mutation rendered COX-2 promoter unresponsive to IL-1β, tumor necrosis factor-α (TNFα), PMA, and prostaglandins. Using a novel DNA-binding assay, we found that CRE-binding protein-2 (CREB-2) bound to CRE at basal state, which was upregulated by PMA but not TNFα. ATF-2 bound constitutively, and its binding was not altered by stimuli. Binding assay and chromatin immunoprecipitation confirmed the interaction of P300 coactivator with CREB-2 and other transactivators on the CRE region.
Human umbilical vein endothelial cells (HUVECs) were cultured as described previously19 in medium 199 (Gibco-BRL) containing 20% fetal bovine serum (FBS), 12.5 μg/mL endothelium cell mitogen (ICN Biomedical Research Products), 100 μg/mL heparin, 100 U/mL penicillin, and 100 μg/mL streptomycin. Only second passage cells were used for transfection experiments. EA.hy926 cells (EA), a transformed HUVEC, were cultured as previously described.19 Human foreskin fibroblasts (HFFs) were cultured as previously described14 in DMEM, supplemented with 10% FBS, penicillin, and streptomycin. All reagents for cell cultures were obtained from Sigma.
Construction of COX-2 Promoter Region Plasmids
We have previously shown that the 5′-flanking sequence −891 to +9 of human COX-2 promoter comprises enhancers and core promoter unit.10 This fragment was constructed into a promoterless luciferase expression plasmid, pGL3 basic (Promega). The 5′-deletion mutants of this fragment were prepared by a method previously described.14 Site-directed mutations of the CRE and C/EBP sites were performed by an oligonucleotide-directed mutagenesis procedure as previously described.20 The CRE site of −891/+9 fragment was mutated from −59TTCGTCA −53 (CRE-WT) to TTgagCt (CRE-M), and the C/EBP site was mutated from −132TTACGCAAT −124 to gcgatagCt (C/EBP-M). Single or dual mutants were constructed into luciferase expression vector pGL3.
Transfection and Luciferase Assays
Transfection of luciferase expression vectors was achieved by a method previously described.14 Briefly, 8 μg of lipofectin (Gibco-BRL) and 2 μg of the promoter constructs were mixed in 1.2 mL of Optimem (Gibco-BRL). The mixture was slowly added to cultured cells in a 6-well plate and incubated for 5 hours. Lipofectin and DNA plasmids were subsequently removed and replaced with complete medium overnight. This medium was changed to serum-free medium for another 24 hours. All cell types, including HUVECs, were morphologically unaltered in serum-free medium for 24 hours. Cells were washed once with serum-free medium and stimulated for 6 hours by PMA (100 nmol/L), IL-1β (1 ng/mL), TNFα (1 ng/mL), forskolin (1 μmol/L), dibutyryl-cAMP (dbcAMP) (1 mmol/L), PGE1 (1 μmol/L), PGE2 (1 μmol/L), or iloprost (100 nmol/L). The medium was removed and the cells were washed twice with ice-cold PBS, lysed by addition of 200 μL/well of reporter-lysis buffer (Promega), scraped off the well, and frozen at −70°C. Luciferase assays were performed in a liminometer. Luciferase activity was related to the protein content of each sample, which was determined by the BCA protein assay kit (Pierce) using BSA as a standard.
DNA Binding Assay
To obtain semiquantitative data on binding of nuclear extract proteins to CRE-containing oligonucleotide sequence, we performed the binding assay by a newly developed method as described in a recent publication.21 In brief, 6 μg of a 5′-biotinated 20-mer corresponding exactly to the CRE region of COX-2 promoter (5′-CAGTCATTTCGTCACATGGG-3′) or its CRE mutant (5′-CAGTCATcgaGTCACATGGG-3′) was incubated with 60 μL of 4% streptavidin-conjugated agarose beads (Sigma) and 600 μg nuclear extract proteins prepared from EA hy926 cells at room temperature for 1 hour. Beads were pelleted and washed. Proteins bound to the DNA probe were eluted, separated by SDA-PAGE, and identified with specific antibodies on Western blots. All the specific antibodies were obtained from Santa Cruz.
Chromatin immunoprecipitation (ChIP) assays were done as described with minor modifications.22 Eighty percent to 90% confluent EA.hy926 endothelial cells were serum starved for 24 hours and then treated with an agonist at 37°C for 4 hours. Formaldehyde 1% was added. After incubation for 20 minutes at 37°C, cells were washed twice, scraped, and lysed in lysis buffer (1% SDS, 10 mmol/L Tris-HCl, pH 8.0, with 1 mmol/L PMSF, pepstatin A, and aprotinin) for 10 minutes at 4°C. The lysates were sonicated 5 times for 10 seconds each, and the debris was removed by centrifugation. One third of the lysate was used as DNA input control. The remaining two thirds were incubated with anti-P300 antibody or a nonimmune rabbit IgG (Santa Cruz) overnight at 4°C. Immunoprecipitated complexes were collected by using protein A Sepharose beads and extensively washed. Cross-linking of protein-DNA complexes was reversed at 65°C for 5 hours, and DNA was digested with 100 μg/mL proteinase K for 3 hours at 50°C. DNA was extracted 3 times and subjected to polymerase chain reaction (PCR) amplification using the following primers specific for the COX-2 promoter: the 5′ primer −709 CTGTTGAAAGCAACTTAGCT-690 and the 3′ primer −32 AGACTGAAAACCAAGC CCAT-51. The resulting product (678 bp) was separated by agarose gel electrophoresis.
The data are mean±SEM of n independent experiments. Data were analyzed by Student’s t test. P<0.05 was considered significant.
Obligatory Role of CRE (−59/−53) in COX-2 Promoter Activity Induced by PMA, IL-1β, and TNFα
We investigated the role of the CRE region (−59/−53) in COX-2 promoter activity in human endothelial cells and fibroblasts by transfecting CRE-M into HUVEC, EA, or HFF. These transfected cells were treated with or without PMA, IL-1β, or TNFα. The CRE-M reduced the basal promoter activity and abrogated the activity stimulated by PMA (Figure 1A), IL-1β, or TNFα in endothelial cells and fibroblasts (Figure 1B). We next compared the promoter activity conferred by CRE-M versus dual CRE-M and C/EBP-M. The double mutant did not additionally dampen promoter activity (Figure 2), supporting the dominant role of CRE in COX-2 promoter activation.
CRE Was Essential for COX-2 Promoter Upregulation by PGs and cAMP Analogs
It has been shown in murine MC3T3-E1 cells that PGs induce COX-2 expression.23 The transcription mechanism has not been elucidated. In this study, we evaluated the effects of PGE2, PGE1, and iloprost on COX-2 promoter activity in HUVEC and HFF. They stimulated COX-2 promoter activity, and 5′-deletion mutation analysis shows that the promoter activity was markedly reduced in −53/+9 and moderately reduced in −96/+9 (Figure 3A). Forskolin and dbcAMP increased COX-2 promoter activity in a manner similar to prostaglandins, and 5′-deletion mutation analysis revealed a similar pattern (Figures 3B and 3C). Furthermore, CRE-M abrogated promoter stimulation by PGE2, PGE1, iloprost, and dbcAMP (Figures 4A and 4B). Pretreatment with H89 reduced COX-2 promoter activity by forskolin and dbcAMP to the basal level (data not shown). These results support the concept that CRE plays a major role in positive feedback regulation of COX-2 promoter activity by prostaglandins.
Binding of Multiple Transactivators to the CRE Region
CREB, ATF-2, AP-1, and USF-2 have been reported to bind to the CRE/E-box region in COX-2 promoter.15–17⇓⇓ However, it was unclear whether they bound concurrently. By using a novel method to quantify transactivator binding to a 20-bp CRE/E-box–containing probe, we were able to compare binding of transactivators to CRE site at basal and stimulated states and determine binding of P300 coactivator to the complex. CREB-2, ATF-2, USF-2, and c-Jun but not c-Fos bound constitutively to wild-type sequence, and PMA increased CREB-2 and c-Fos binding (Figure 5A). To determine whether these transactivators bound specifically to CRE, we carried out identical binding experiments using a CRE mutant probe. Neither CREB-2 nor ATF-2 bound to the mutant, consistent with their specific binding to CRE (Figure 5A). USF-2 bound to CRE-M as well as CRE-WT, consistent with binding to E-box as previously reported.17 c-Jun and c-Fos binding was not influenced by CRE mutation, suggesting that AP-1 also binds to a non-CRE site. TNFα treatment for 4 hours did not alter binding of CREB-2 or ATF-2 to CRE but reduced c-Jun and USF-2 binding to the non-CRE site (Figure 5B).
Interaction of P300 With CRE
P300 and CREB binding protein (CBP) are homologous proteins with similar coactivator properties.24 They bind to several DNA-bound transactivators, including CREB, and thereby recruit the transcription machinery to the TATA promoter region.24 We determined P300 binding to the CRE/E-box complex. P300 bound to the CRE-WT in unstimulated cells, which was not altered by CRE mutation (Figure 6). PMA and TNFα did not significantly influence P300 binding (Figure 6).
Binding of CREB-2 and P300 to Chromatin
Binding of CREB and P300 to CRE was additionally evaluated by ChIP. Sonicated chromatin was precipitated with antibodies against CREB-2 or P300. A nonimmune IgG was used as control. The region of COX-2 promoter pulled down by immunoprecipitation was identified by PCR using 2 COX-2–specific primers. Nonimmune IgG failed to precipitate CRE-containing COX-2 promoter sequence, whereas anti–CREB-2 and anti-P300 precipitated the CRE-containing sequence (Figure 7). There was no significant difference between unstimulated and PMA- or TNFα-treated endothelial cells.
Results from the present study provide important information about the obligatory role of CRE at −59/−53 in COX-2 induction and feedback regulation by major classes of proinflammatory mediators including IL-1β, TNFα, PMA, and PGE2. This site is bound at the basal state by CREB-2, ATF-2, and P300. Results from ChIP assays confirm binding of CREB-2 and P300 to this region at the basal state. Together with the transfection data, the results demonstrate the essential nature of CRE for basal promoter activity. The transcription machinery is activated by basal binding of CREB-2 and ATF-2 to this site and the recruitment of P300 to these DNA-bound transactivators. This site is also essential for COX-2 promoter stimulation by agonists, but the stimulated activity uses additional transcription activators including potential changes in binding of c-Fos and c-Jun to this region and binding of C/EBPβ and NF-κB to upstream enhancer elements. Recruitment of P300 by CRE-bound CREB/ATF facilitates the transcriptional stimulation, because P300 is capable of interaction with multiple transcription factors including c-Jun, C/EBPβ, and NF-κB.
Our results show a differential regulation of CREB-2 and AP-1 binding to the CRE region by PMA versus TNFα. PMA-induced c-Fos and enhanced CREB-2 binding, whereas TNFα had no effect. Increased AP-1 and CREB-2 bindings may contribute to PMA-induced promoter activation. However, results from the ChIP assay did not confirm an increased CREB-2 binding by PMA. One possible explanation for this discrepancy is that the ChIP assay is not quantitative and may not be suitable for detecting a difference in binding. Additional studies using more advanced quantitative assays will be needed to resolve this issue. AP-1 binding to the CRE region has been previously reported to regulate COX-2 promoter activity.25,26⇓ However, the exact regulatory element that binds AP-1 remains controversial. It was reported that PMA induced AP-1 binding to a probe containing the CRE and E-box region in human COX-2 promoter25 and shear stress induced AP-1 binding to a site that was not influenced by E-box mutation.26 Our results show that AP-1 binding to this region was not altered by CRE mutation. Because there is no putative AP-1 binding site in this region in human sequence, it is possible that AP-1 binds to a novel site. The DNA-bound AP-1 can also recruit P300, thereby increasing the transcriptional activity. P300 and CBP have been shown to be involved in COX-2 promoter activation.16,25⇓ In this study, we provide direct evidence for P300 binding to transactivator-DNA complex. Because P300 binds equally well to CRE-WT and CRE-M, it is likely that P300 interacts with AP-1 and USF-2 in addition to CREB/ATF. USF-2 is known to bind E-box and involve in basal gene expression. Its role in COX-2 promoter regulation by stimuli is less clear. It has been reported that E-box mutation does not alter COX-2 promoter activity induced by mitogenic factors in murine and human cells but severely reduced COX-2 induction by hormones in rat ovarian cells.15,17,25⇓⇓
Gene expression is highly controlled by chromatin structure.27 We carried out ChIP to determine binding of P300 to COX-2 promoter in chromatin structure. Out results confirm the binding of CREB-2 and P300 to a chromatin region containing COX-2 promoter sequence in basal state as well as in PMA-stimulated or TNF-α–stimulated state. These results suggest that the CRE region of COX-2 promoter is accessible to transactivators. These data additionally support the importance of this region in COX-2 expression and regulation in the cell.
PGE2 and PGI2, major products of COX-2, stimulate adenylyl cyclase and increase cAMP formation. A series of studies in murine MC3T3-E1 cells have shown that PGE2 and cAMP analogs increase COX-2 expression.23 In this study, we confirm that PGE2, PGE1, and iloprost, a stable analog of PGI2, are all capable of stimulating COX-2 promoter activity. Forskolin and dbcAMP also stimulate COX-2 promoter activity. Because mutation of CRE abrogates COX-2 promoter activation by prostaglandins and cAMP and H89 abrogates COX-2 promoter stimulation, it is likely that PGE2 and PGI2 amplify COX-2 promoter activity by increasing CREB binding to the CRE site. This CRE-dependent positive feedback regulation of COX-2 promoter activity additionally underscores the importance of CRE in COX-2 promoter function. It should be noted that PGE2, PGI2, and cAMP per se do not increase COX-2 protein levels in most primary human cells, including endothelial cells and fibroblasts. However, they may play an important role in amplifying and solidifying COX-2 stimulation by inflammatory mediators.
In summary, our data demonstrate the essential role that CRE plays in COX-2 promoter activity. Activation of the transcription machinery including P300 recruitment is mediated by basal CREB/ATF binding to this site, and the stimulated activity requires binding of additional transcription activators including AP-1 binding to this region and C/EBPβ and NF-κB binding to upstream enhancer elements. Agonists such as PMA and TNFα use different transactivators to stimulate the promoter activity. PMA uses AP-1 and C/EBPβ, which binds to the −132/−124 site to enhance transactivation. C/EBPβ has been shown to interact with P300.28 On the other hand, TNFα uses 2 NF-κBs that bind to −213/−222 and −447/−438, respectively, to enhance transactivation. Both stimuli require the basal binding of CREB/ATF to the CRE site. It is likely that the basal CREB/ATF binding to the CRE site is essential for COX-2 promoter activation by other stimuli.
This work was supported by grants from the National Institutes of Health (P50-NS 23327 and R01-HL 50675) of the United States Public Health Service.
↵*Drs Schroer and Zhu contributed equally to this study.
Guest editor for this article was Joseph Loscalzo, MD, PhD, Boston University School of Medicine, Boston, Mass.
- ↵Wu KK. Inducible cyclooxygenase and nitric oxide synthase. Adv Pharmacol. 1995; 33: 179–207.
- ↵Smith WL, Garavito RM, DeWitt DL. Prostaglandin endoperoxide H synthases (cyclooxygenases)-1 and -2. J Biol Chem. 1996; 271: 33157–33160.
- ↵Vane JR, Mitchell JA, Appleton I, et al. Inducible isoforms of cyclooxygenase and nitric-oxide synthase in inflammation. Proc Natl Acad Sci U S A. 1994; 91: 2046–2050.
- ↵Seibert K, Zhang Y, Leahy K, et al. Pharmacological and biochemical demonstration of the role of cyclooxygenase 2 in inflammation and pain. Proc Natl Acad Sci U S A. 1994; 91: 12013–12017.
- ↵Wu KK. Cyclooxygenase-2 induction in congestive heart failure: friend or foe? Circulation. 1998; 98: 95–96.
- ↵Fletcher BS, Kujubu DA, Perrin DM, et al. Structure of the mitogen-inducible TIS10 gene and demonstration that the TIS10-encoded protein is a functional prostaglandin G/H synthase. J Biol Chem. 1992; 267: 4338–4344.
- ↵Schmedtje JF Jr, Ji YS, Liu WL, et al. Hypoxia induces cyclooxygenase-2 via the NF-κB p65 transcription factor in human vascular endothelial cells. J Biol Chem. 1997; 272: 601–608.
- ↵Saunders MA, Sansores-Garcia L, Gilroy DW, et al. Selective suppression of C/EBPβ binding and cyclooxygenase-2 promoter activity by sodium salicylate in quiescent human fibroblasts. J Biol Chem. 2001; 276: 18897–18904.
- ↵Xie W, Fletcher BS, Anderson RD, et al. V-src induction of TIS 10/PGS 2 prostaglandin synthase gene is mediated by an ATF/CRE transcription response element. Mol Cell Biol. 1994; 14: 6531–6539.
- ↵Morris JK, Richards JS. An E-box region within the prostaglandin endoperoxide synthase-2 (PGS-2) promoter is required for transcription in rat ovarian granulose cells. J Biol Chem. 1996; 271: 16633–16643.
- ↵Xu X-M, Sansores-Garcia L, Chen X, et al. Suppression of inducible cyclooxygenase-2 gene transcription by aspirin and sodium salicylate. Proc Natl Acad Sci U S A. 1999; 96: 5292–5297.
- ↵Chen P-F, Tsai A-L, Wu KK. Cysteine-184 of endothelial nitric oxide synthase is involved in heme coordination and catalytic activity. J Biol Chem. 1994; 269: 25062–25066.
- ↵Zhu Y, Saunders MA, Yeh H, e al. Dynamic regulation of cyclooxygenase-2 promoter activity by isoforms of CCAAT/enhancer-binding proteins. J Biol Chem. 2002; 277: 6923–6928.
- ↵Subbaramaiah K, Lin DT, Hart JC, et al. Peroxisome proliferator-activated receptor γ ligands suppress the transcriptional activation of cyclooxygenase-2. J Biol Chem. 2001; 276: 12440–12448.
- ↵Ogasawara A, Arakawa T, Kaneda T, et al. Fluid shear stress-induced COX-2 expression is mediated by C/EBPβ, CREB and AP-1 in osteoblastic MC3T3-E1 cells. J Biol Chem. 2001; 276: 7048–7054.
- ↵Mink S, Haenig B, Klempnauer K-H. Interactivation and functional collaboration of P300 and C/EBPβ. Mol Cell Biol. 1997; 17: 6609–6617.