NF-IL6β regulates gene expression and plays function roles in many tissues. The EGF-regulated cyclooxygenase-2 ( cox-2 ) expression is mediated through p38 MAPK signaling pathway and positively correlates with NF-IL6β expression in A431 cells. NF-IL6β coordinated with c-Jun on cox-2 transcriptional activation by reporter and small interfering RNA assays. NF-IL6β could directly bind to CCAAT/enhancer-binding protein (C/EBP) and cyclic AMP-response element (CRE) sites of the cox-2 promoter by in vitro -DNA binding assay. The C/EBP site was important for basal and, to a lesser extent, for EGF-regulated cox-2 transcription, while the CRE site was a more specific response to EGF inducibility of cox-2 gene. SUMO1 expression attenuated EGF- and NF-IL6β-induced cox-2 promoter activities. NF-IL6β was found to be sumoylated by in vivo - and in vitro -sumoylation assays, and the SUMO1-NF-IL6β (suNF-IL6β) lost its ability to interact with p300 in in vitro -binding assay. NF-IL6β was also acetylated by p300, and acetylation of NF-IL6β enhanced the cox-2 promoter activity stimulated by NF-IL6β itself. In vivo -DNA binding assay demonstrated that EGF stimulated the recruitment of p300 and NF-IL6β to the cox-2 promoter, yet promoted the dissociation of SUMO1-modificated proteins from the promoter. These results indicated that NF-IL6β plays a pivotal role in the regulation of basal and EGF-induced cox-2 transcription.
Prostaglandins play important roles in many biological processes, including cell division, immune responses, blood pressure regulation, ovulation, bone development and wound healing. The cyclooxygenase (COX, prostaglandin endoperoxide synthase) is a key enzyme in prostaglandin, prostacyclin and thromboxane biosynthesis from arachidonic acid. Two COX isoforms were described ( 1 ). COX-1 is constitutively expressed in most tissues and cells in animal species. COX-2 is induced by a wide-range of stimulators, such as IL-1β ( 2 , 3 ), TNF-α ( 4 ), IL-18 ( 2 ), epidermal growth factor (EGF) ( 5 ) or LPS ( 6 ), in many distinct cell types ( 7 – 9 ) and is regulated mainly at the level of transcription. Human cox-2 promoter region contains a twin arginine translocation A and multiple regulatory elements, including two putative nuclear factor-κB (NF-κB) binding sites, one nuclear factor interleukin-6 (NF-IL6)/CCAAT/enhancer-binding protein (C/EBP) binding site and one cyclic AMP-response element (CRE) ( 10 ). Recent studies on human cox-2 promoter have shown that cox-2 transcription is regulated by different transcription factors, including NF-κB ( 11 ), NF-IL6/C/EBP ( 11 – 14 ), C/EBPδ ( 12 ), CREB ( 12 , 13 , 15 ) and activation protein 1 complex (AP-1) ( 5 , 11 ), supporting that regulation of cox-2 gene expression could involve complex interactions among diverse transcription factors. Thus, transcriptional mechanism of cox-2 induction relies on cell type-specific as well as combined interactions of several cis -acting regulatory elements, transcription factors and signal transduction pathways.
The C/EBP family contains three main activating members, C/EBPα, C/EBPβ and C/EBPδ, that recognize the same DNA sequence. These three members have a common structure: an N-terminal domain bearing the transaction domain, a basic DNA-binding domain and a C-terminal leucine zipper domain that allows the homo- or hetero-dimerization of these factors. C/EBPδ is involved in the ligand-stimulated transcriptional regulation of cox-2 gene ( 12 ). However, the detail molecular mechanism of human C/EBPδ, NF-IL6β, in the regulation of cox-2 gene transcription is unclear.
Post-translational modification of proteins by sumoylation is an important regulatory mechanism and has been found to be utilized in many cellular processes ( 16 – 18 ). SUMO modification of several transcription factors has been reported, including the androgen receptor ( 19 ), LEF1 ( 20 ), c-Myb ( 21 ), TEL ( 22 ), Sp3 ( 23 , 24 ), p53 ( 25 ), c-Jun ( 26 ) and C/EBPs ( 26 ). SUMO conjugation has been shown to regulate several different protein functions including protein stability, subcellular localization and transcriptional activation regulation ( 18 , 27 , 28 ). The consensus sequence, (I/L)KXE, for sumoylation has been defined ( 29 ). The C/EBP family belongs to the large family of basic leucine zipper (bZIP) transcription factors. The repression domain I of C/EBPε was demonstrated to be modified by SUMO1 ( 26 ), and this modification was proposed to be important for the inhibitory function of this domain. Kim et al . ( 25 ) also reported that conserved SUMO target sequences are present in C/EBPα, C/EBPβ and C/EBPδ, and that these isoforms can be conjugated to SUMO1 ( 26 ). However, the function of sumoylated C/EBPs is largely unknown, especially in the case of NF-IL6β.
Coactivator p300 and CREB-binding protein (CBP) serve as an integrator for gene transcription. Several reports have suggested involvement of p300 coactivator in cox-2 transcriptional regulation ( 4 , 5 , 30 ). p300 contains histone acetyltransferase (HAT) activity that modulates the acetylation of histones or transcription factors, thus affecting the DNA binding and transcriptional activation. p300 and CBP have been shown to participate in C/EBPs-mediated gene transcription ( 31 – 33 ). C/EBP family members trigger the phosphorylation of p300 and consequently increase p300-mediated transcriptional activation ( 34 ).
Several reports have shown C/EBPβ and C/EBPδ's involvement in cox-2 gene expression ( 35 , 36 ). However, the effects of C/EBPs on cox-2 transcription are dependent on cell type and stage of differentiation. Gain or loss of function of C/EBPδ and C/EBPβ regulate cox-2 promoter activity in various cell types ( 35 , 37 , 38 ). In our previous study, we found that induction of c-Jun is involved in EGF-induced cox-2 expression ( 5 ). In addition, we found that the level of NF-IL6β is also elevated by EGF treatment in human epidermoid carcinoma A431 cells ( 37 ). In this study, we extended our work to investigate the functional role of NF-IL6β in regulating cox-2 promoter activity in the basal and EGF-induced transcriptional state, and the effects that sumoylation and acetylation of NF-IL6β play function roles on the promoter. Our results indicated that NF-IL6β mediates the basal and EGF-induced cox-2 promoter state, and sumoylation of NF-IL6β attenuates the activation of cox-2 promoter, while p300 can acetylate NF-IL6β and participate in the NF-IL6β-enhanced cox-2 promoter regulation.
MATERIALS AND METHODS
Human EGF was purchased from Peprotech (Rocky Hill, NJ). SB203580 was obtained from Calbiochem (San Diego, CA). Antibodies against COX-2, NF-IL6β, SUMO1 and α-p300-conjugated agarose were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against acetyl-lysine were purchased from Upstate (Charlottesville, VA). Monoclonal α-HA antibody was purchased from BM (Boehringer, Mannheim, Germany). Lipofectamine 2000, Dulbeco's modified Eagle's medium (DMEM), SuperScriptTM III and Opti-MEM medium were obtained from Invitrogen (Carlsbad, CA). All oligonucleotides were synthesized by MDBio Inc. (Taipei, Taiwan). Fetal bovine serum (FBS) was from HyClone Laboratories (Logan, UT). Streptavidin–Sepharose beads were purchased from Amersham Biosciences (Buck, UK). In vitro transcription/translation kit was purchased from Promega (Madison, WI). Expression plasmid pcDNA3/HA was a gift of Dr. Hsin-Fang Yang-Yen (Institute of Molecular Biology, Academia Sinica, Tapei, Taiwan). p Sliencer ™ 3.0 vector was purchased from Ambion (Austin, TX). The sumoylation kit was purchased from LAE Biotechnology Co. (Taichung, Taiwan). The recombinant p300 protein was purchased from Active Motif (Carlsbad, CA). The cloning vector, yTA vector, was purchased from Yeastern Biotech. Co. (Taipei, Taiwan). DNA polymerase kit and BD Advantage GC™ PCR kit for PCR-cloning were purchased from BD Biosciences (Palo Alto, CA). Protein concentration column, Amicon ® Centriprep ® Filter Devices, was purchased from MILLIPORE (Billerica, MA). All other reagents used were of the highest purity obtainable. The expression vector TAM-67 encoding the truncated human c-Jun was the generous gift of Dr M. Birrer (NCI, National Institutes of Health, Rockville, MD). Small interfering RNA (siRNA) pool for c-Jun and a non-specific control siRNA were purchased from Darmacon (Lafayette, CO). pSUPERc-Jun siRNA was designed and constructed by KRII International Co. (Taipei, Taiwan).
Plasmid transfection and reporter gene assay
A431 and HeLa cells were maintained in DMEM supplemented with 10% FBS, 100 µg/ml streptomycin, and 100 U/ml penicillin. All EGF treatments of A431 cells were in the concentration of 50 ng/ml. Cells were transfected with plasmids by lipofection using Lipofectamine 2000 according to the manufacturer's instruction. Cells were replated 24 h before transfection at an optional density in 3 ml of fresh culture medium in a 3.5 cm plastic dish. For usage in transfection, 5 µl of Lipofectamine 2000 were incubated with reporter plasmid and the expression plasmids as indicated in each experiment, in 2 ml of Opti-MEM medium for 30 min at room temperature. Total DNA concentration for each experiment was matched with empty vector. Cells were transfected by changing the medium with 2 ml of Opti-MEM medium containing the plasmids and Lipofectamine 2000, unless otherwise stated. Cells were stimulated with EGF when necessary and incubated for 16 h. The luciferase activities in cell lysates were measured by the luciferase assay system and determined as described ( 37 ). Luciferase activity was normalized per microgram of extract protein.
Small interfering RNAs assay
Two oligonucleotides were synthesized according to the oligonucleotide design procedure described in the Ambion's manual. They were as follows: 5′-gATCCgCCAggAgATgCAgCAgAAgTTCAAgAgACTTCTgCTgCATCTCCTggTTTTTTggAAA-3′ and 5′-AgCTTTTCCAAAAAACCAggAgATgCAgCAgAAgTCTCTTgAACTTCTgCTgCATCTCCTggCg-3′. The 5′ ends of the two oligonucleotides were non-complementary and formed the BamHI and HindIII restriction site overhangs that facilitated efficient directional cloning into the p Sliencer ™ 3.0 vector. Cell transfection separately with NF-IL6β siRNA expression vectors (pSi-1) and p Sliencer ™ 3.0 negative control vectors (pSi-C), which encode a hairpin siRNA whose sequence is not found in the human genome databases was purchased from Ambion, were carried out by Lipofectamine 2000 or lipofectin according to the manufacturer's instruction. After 24 h recovery in complete medium, RT–PCR or western blotting assay was performed from the transfectants, which had been previously starved for 6 h and re-stimulated with EGF for 2 h.
DNA affinity precipitation assay
Nuclear extracts from A431 cells with or without EGF treatment were prepared, and DNA affinity precipitation assay was performed according to the method described previously ( 38 , 39 ). The 200 µg of lysates extracted from each group were incubated with 1 µg of biotinylated C/EBP or CRE oligonucleotides in the presence of DNA binding buffer containing 10 mM Tris–HCl pH 7.5, 50 mM NaCl, 1 mM DTT, 1 mM EDTA, 10% (v/v) glycerol, 10 mM NaF, 1 mM PMSF, 1 µg aprotinin/ml, 1 µg leupeptin/ml, 1 mM Na 3 VO 4 and 2 µg poly(dI–dC). After 1 h of incubation at 4°C, 40 µl of streptavidin–Sepharose were added to the reaction mixture and the incubation was continued for 1 h. The complexes were then precipitated by centrifugation and washed three times with DNA binding buffer before they were resolved by SDS–PAGE and subsequently analysed by immunoblotting with α-C/EBPδ antibodies.
Chromatin immunoprecipitation assay
The chromatin immunoprecipitation (ChIP) assay was carried out essentially as described by Saccani et al . ( 40 ). Briefly, A431 cells with or without prior stimulation with EGF were treated with 1% formaldehyde for 15 min. The cross-linked chromatin was then prepared and sonicated to an average size of 300–500 bp. The DNA fragments were immunoprecipitated with antibodies specific to p300, SUMO1 and NF-IL6β or control rabbit IgG at 4°C, overnight. After reversal of cross-linking, the immunoprecipitated chromatin was amplified by PCR amplification of specific regions of the cox-2 genomic locus. The primers were as follows: COX-2/F-186: 5′-CTGGGTTTCCGATTTTCTCA-3′, COX-2/R+49; 5′-GAGTTCCTGGACGTGCTCCT-3′, COX-2/F+800; 5′-CTAAGGCAGGTTAAAAAATTGTATTTCC-3′ and COX-2/R+1200: 5′-TCCCTTGAAGTGGGTAAGTATGTAGTG-3′. The amplified DNA products were resolved by agarose gel electrophoresis and confirmed by sequencing.
In vitro expression of NF-IL6β proteins
In vitro transcription/translation of NF-IL6β was performed using 1 µg of pcDNA3-HA/NF-IL6β and a wheat germ coupled transcription/translation system according to the instructions provided by the manufacturer. Recombinant His-tagged NF-IL6β (His/NF-IL6β) and NF-IL6βK120A (His/NF-IL6βK120A) were generated from the pET-28a (+) vector. The recombinant plasmids were transformed into BL21 (DE3) cells. Isopropyl-β-D-thiogalactopyranoside (IPTG) was used to induce recombinant protein expression in the transformants. His/NF-IL6β and His/NF-IL6βK120A were purified according to the instructions provided by the manufacturer and dialysed with dialysis buffer (50 mM HEPES pH7.4, 100 mM NaCl and 1 mM DTT). The dialysed proteins were concentrated by Amicon ® Centriprep ® Filter Devices.
In vitro - and in vivo -SUMO modification assays
In vitro -SUMO modification of NF-IL6β was performed using sumoylation kit. Briefly, assays were performed with 2 µl of SAEI and SAEII (7.5 µg/ml), 2 µl of UBC9 (50 µg/ml), 2 µl of 10× sumoylation reaction buffer (200 mM HEPES, pH 7.5, 50 mM MgCl 2 and 20 mM ATP), 2 µl of SUMO1 (50 µg/ml) and 2 µl of in vitro -translated HA/NF-IL6β. The reaction mixture was incubated at 37°C for 30 min and then quenched with SDS–PAGE sample buffer. The samples were subsequently analysed by SDS–PAGE and immunoblotting analysis with α-HA antibodies. In vivo -sumoylation assay was carried out in A431 cells. Cells were transfected with pcDNA3-HA/NF-IL6β expression vectors in the presence or absence of SUMO1-GG. Cell extracts were prepared in sample buffer (5% SDS, 0.15 M Tris–HCl pH 6.7 and 30% glycerol) and then diluted 1:3 with RIPA buffer (25 mM Tris–HCl pH 8.2, 50 mM NaCl, 0.5% NP-40, 0.5% deoxycholate and 0.1% SDS), containing 20 mM N -ethylmaleimide and 1 mM PMSF, 1 µg aprotinin/ml and 1 µg leupeptin/ml for subsequent immunoprecipitation assay.
In vitro - and in vivo -acetylation assay
Purified p300 protein (50 ng) and 0.5 µg of the indicated His/NF-IL6β protein or His/NF-IL6βK120 were incubated in a reaction mixture containing 50 mM Tris–HCl, pH 8.0, 10% glycerol, 0.1 mM EDTA, 1 mM DTT and 40 µM acetyl-coenzyme A for 1 h at 30°C. The reaction mixture was subjected to SDS–PAGE and analysed by western blotting using anti-α-acetyl-lysine antibodies. In vivo -acetylation assay was performed by transfecting cells with the pcDNA3-HA/NF-IL6β expression vectors in the presence or absence of p300 expression plasmid. Cells were lysed in sample buffer (50 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM EDTA and 1% Triton X-100) and then diluted 1:4 with TE buffer (10 mM Tris–HCl, pH 7.5 and 0.1 mM EDTA), containing 100 µM sodium butyrate, 1 mM PMSF, 1 µg aprotinin/ml and 1 µg of leupeptin/ml for immunoprecipitation assay.
Construction of reporter plasmids and expression vectors
The cox-2 promoter fragment from −207 to +49 bp (−207/+49wt) was obtained by PCR from the pXC918 reporter ( 5 ). The primers used were as follows: COX-2/KpnI-207: 5′-GGGGTACCTGCTCCCAAATTGGGGCAGC-3′, COX-2/HindIII+49: 5′-GGAAGCTTGAGTTCCTGGACGTGCTCC-3′. The PCR fragments were cloned into yTA vector and verified by sequencing. A KpnI/HindIII fragment was subcloned into the multi-cloning sites of the promoter-less vector pGL2-basic. Mutant reporter plasmids were derived from −207/+49wt by site-directed mutagenesis of each individual region as indicated. Plasmid pGL2-promoter/COX-2/1XCEBP (pGL2-promoter 1XC/EBP) was derived by inserting one copy of the DNA fragment containing the C/EBP motif sequence (5′-GGGCTTACGCAATTTTTTTAA-3′) into the SmaI site of the pGL2-promoter vector. NF-IL6β was generated from human liver cDNA library by PCR using BD Advantage GC™ PCR kit and using the following oligonucleotides: 5′-CGGGATCCAGCGCCGCGCTCTTCAGCCTG-3′ and 5′-GGCCTCGAGGCCGCGCGTTACCGGCAGTC-3′. The amplified fragment was digested with BamHI and XhoI and inserted into BamHI- and XhoI-digested pcDNA3-HA to produce HA-tagged NF-IL6β (HA/NF-IL6β). The cox-2 promoter plasmids pXC80 and pXC918 have been described previously ( 5 ).
Total RNA was isolated from A431 cells using the TRIzol RNA extraction kit. Of the isolated RNA 1 µg was subjected to reverse transcription with SuperScriptTM III. Specific primers for COX-2: 5′-CCCACTTCAAGGGATTTT-3′ and 5′-CCAGACCAAAGACCTCCT-3′, and for NF-IL6β: 5′-AGCGCAACAACATCGCCGTG-3′ and 5′-GTCGGGTCTGAGGTATGGGTC-3′, were used for analyses. The PCR products were separated by electrophoresis in 1% agarose gel and visualized with ethidium bromide staining.
p300 pull-down assay
In vitro -translated HA/NF-IL6β (2 µl) were incubated with purified p300 protein in a buffer composed of 50 mM Tris–HCl, pH 8.0, 10% glycerol, 0.1 mM EDTA and 1 mM DTT. After pulling down by α-p300-conjugated agarose (p300-AC), the immunoprecipitation pellets were washed and separated by SDS–PAGE for subsequent detection by immunoblotting with α-HA antibodies.
p38 MAPK inhibitor attenuates EGF-induced cox-2 transcription
Recent studies have shown that the known C/EBP family proteins can modulate cox-2 gene expression through interactions with each other or other transcription factors. Hence prior to this study, we have verified the expression of C/EBPs in the presence of EGF ( Figure 1A ). According to these data, the expression pattern of C/EBPα is not affected by EGF treatment, whereas C/EBPβ expression can be greatly increased by EGF treatment, but only after a 6 h delay. Otherwise, NF-IL6β pre-existed before EGF treatment and enhanced by EGF treatment. The induction pattern of NF-IL6β is similar with c-Jun and COX-2 expression patterns after EGF treatment.
We previously found that the p38 MAPK signaling pathway mediates the EGF-induced NF-IL6β transcription in A431 cells ( 37 ). To examine the possibility that NF-IL6β regulates cox-2 gene expression in A431 cells, we first studied the correlation between p38 MAPK activation and NF-IL6β/cox-2 gene transcription. RT–PCR assay was carried out for this study. The levels of EGF-induced NF-IL6β mRNA and cox-2 mRNA were attenuated by pretreatment with p38 MAPK inhibitor, SB203580, in a dose-dependent manner ( Figure 1B ). The results suggest a close relationship between EGF-induced NF-IL6β expression and cox-2 transcription through p38 MAPK signaling activation.
NF-IL6β plays a functional role in cox-2 promoter activity
To address and connect whether NF-IL6β would activate the cox-2 promoter, cells were cotransfected with an NF-IL6β expression vector and a reporter construct controlled by the cox-2 promoter. As shown in Figure 2A , overexpression of NF-IL6β enhanced cox-2 promoter activity in a dose-dependent manner. The effect could be enhanced by EGF treatment. The results suggest that NF-IL6β might play a functional role in the basal and EGF-induced expression of COX-2.
To investigate the effect of endogenous NF-IL6β on transcriptional activation of cox-2 gene, complementary specific oligonucleotides aimed to inhibit NF-IL6β gene expression by RNA interference were designed. Transfection of A431 cells with the NF-IL6β siRNA expression vector resulted in reduction of EGF-induced NF-IL6β mRNA and a concomitant decrease of cox-2 mRNA ( Figure 2B , upper panel). Consistent with the RT–PCR results, cells cotransfected with a luciferase reporter construct controlled by cox-2 promoter, pXC918 ( 5 ), and an NF-IL6β siRNA expression vectors, pSi-1, expressed significantly less luciferase activity than that cotransfection with negative control, pSi-C ( Figure 2B , lower panel). Cotransfection of HA/NF-IL6β expression vectors with pSi-C increased COX-2 expression, but the transfectant with pSi-1 decreased the COX-2 expression ( Figure 2C ). These results suggest that NF-IL6β plays a role in regulating COX-2 expression. Since c-Jun is involved in EGF-induced cox-2 transcriptional activation through the CRE binding site ( 5 , 41 ), we examined whether activation of cox-2 promoter, which bears a CRE binding site, by NF-IL6β requires the cooperation with c-Jun. As shown in Figure 2C , cells cotransfected with c-Jun siRNA expression vector reduced NF-IL6β-induced cox-2 promoter activity, indicating that the activation of cox-2 promoter by NF-IL6β is, at least in part, dependent on c-Jun. Furthermore, cotransfection of cells with c-Jun and NF-IL6β expression vectors resulted in a synergistic activation of the cox-2 promoter ( Figure 2D ). Taken together, these data indicate that NF-IL6β and c-Jun likely co-regulate EGF-dependent cox-2 expression.
C/EBP and CRE motifs are important for cox-2 gene activation in A431 cells
To further study whether C/EBP and CRE motifs are involved in the basal and EGF-induced cox-2 gene regulation, reporter expression controlled by the wild-type or mutant cox-2 gene promoter was assessed in transient transfection studies. The results are summarized in Figure 3B . A point mutation at the C/EBP motif resulted in 60–70% loss of the basal promoter activity, but only a 20–30% decrease in EGF-inducible increase of the promoter activity (compare −207/+49wt with −207/+49mCEBP). In contrast, a point mutation at the CRE motif lost ∼40% of the basal promoter activity and ∼45–55% of EGF-inducible activity (compare −207/+49wt with −207/+49mCRE). The stimulatory effect of EGF was abolished when both CRE and C/EBP sites were mutated (−207/+49mCE/C), suggesting that C/EBP and CRE motifs are essential for cox-2 promoter activity. These results demonstrate that CRE motif is more important in EGF response, while C/EBP site is more important in regulating the basal cox-2 expression. To evaluate the effect of NF-IL6β on C/EBP and CRE motifs, the NF-IL6β expression vector was cotransfected with various mutant cox-2 promoters in A431 cells. The results are summarized in Figure 3C . The mutation at C/EBP and CRE sites diminished 75 and 40%, respectively, of the stimulatory effect of NF-IL6β on the cox-2 promoter (lanes 2 and 3). While the NF-IL6β effect seems to be more prominent at the C/EBP motif, double mutation at both sites almost completely abolished the NF-IL6β response (lane 4). Since NF-IL6β played a functional role in the transcriptional activity of cox-2 promoter, we examined whether NF-IL6β binds to the C/EBP or CRE motifs of cox-2 promoter. To test the binding activity between CRE and C/EBP motifs, DNA affinity precipitation assay was performed with nuclear extracts prepared from control and EGF-treated A431 cells. As seen in Figure 3D (lanes 5 and 8), EGF apparently increased the amount of NF-IL6β bound to the C/EBP and CRE motifs of cox-2 promoter.
Sumoylation plays a negative regulatory role in cox-2 promoter activation
Base on these results, we have verified that NF-IL6β play a functional role in cox-2 transcription. We then tried to elucidate the mechanism in which NF-IL6β modulates the basal and EGF-induced transcriptional regulation of cox-2 gene. C/EBPδ was reported to be a SUMO1 substrate in vitro and in vivo , and a K120A mutant of the Gal4-C/EBPδ(1–142) fusion protein lost ∼60% of its repressive activity, as compared with the wild-type Gal4-C/EBPδ(1–142) protein ( 26 ). To investigate whether sumoylation acts on the transcription factors for the cox-2 promoter, transient reporter assay was performed. Cotransfection of the SUMO1 expression vector significantly repressed the NF-IL6β-enhanced pXC918 reporter activity ( Figure 4A , compare lanes 3 and 4 with lanes 5 and 6). To specifically examine whether the C/EBP-binding complex participated in the SUMO-mediated repressive effect, a heterologous reporter pGL2 promoter-C/EBP was used, and the NF-IL6β-enhanced pGL2 promoter-C/EBP reporter activity could be attenuated by exogenously expressed SUMO1 (compare lanes 9 and 10 with lanes 11 and 12). These results suggest that sumoylation might suppress cox-2 promoter activity under the EGF-deprived condition, and that the C/EBP motif might be the site of SUMO-mediated effect.
NF-IL6β is a SUMO substrate
Although the mouse C/EBPβ has been shown to be a target for SUMO1, sumoylation of human C/EBPδ (NF-IL6β) and its biological function have not been elucidated. By sequence comparison between mouse C/EBPδ and NF-IL6β, the consensus sequence for SUMO attachment, LKREP, in the regulatory domain motif (RDM) was conserved. To examine whether NF-IL6β is a SUMO substrate, in vitro -sumoylation assay was carried out with purified E1 (SAEI and SAEII), E2 (Ubc9) and in vitro -transcribed/translated HA/NF-IL6β. SUMO1 or SUMO3 could be covalently conjugated to HA/NF-IL6β in vitro ( Figure 5A , lanes 3 and 4). To map the site of sumoylation, we first examined the sequence and found a region between amino acid 110 and 151 in HA/NF-IL6β contained a potential sumoylation site (data not shown). The lysine 120 in this region has been reported to be a potential site of sumoylation ( 26 ). A K120A mutation was then generated by mutagenesis in HA/NF-IL6β to see whether sumoylation of the protein was affected. As shown in Figure 5B (lane 4), the K120A mutant protein could not be sumoylated by in vitro -sumoylation assay, suggesting that lysine 120 is the site of sumoylation. Since post-translational modifications of transcription factor often affect their DNA binding, the DNA-binding activity of NF-IL6β and suNF-IL6β was examined by gel-shift assay using in vitro -translated HA/NF-IL6β protein and labeled CRE or C/EBP probes. NF-IL6β bound to DNA as a homodimer (our observation, data not shown). The homodimerized HA/NF-IL6β could bind to the C/EBP motif at the cox-2 promoter ( Figure 5C , lane 2). Interestingly, it also bound to the CRE motif ( Figure 5C , lane 5), a result subsequently confirmed with purified His/NF-IL6β protein (data not shown). Using equal amount of the proteins, we found no appreciable differences between HA/NF-IL6β and sumoylated HA/NF-IL6β, in term of DNA-binding and homodimerization activities ( Figure 5C , compare lane 2 with lane 3, or lane 5 with lane 6). We next examined whether the turnover rates (half-life) of NF-IL6β and NF-IL6βK120A are different. Transfected cells were harvested at different time points post-cycloheximide treatment, and exogenously expressed HA/NF-IL6βwt or HA/NF-IL6βK120A in cell lysates were analysed by western blotting. We measured the different time points of remained HA/NF-IL6β proteins and separately normalize with the ‘0’ time point. As shown in Figure 5D , the turnover half-life of HA/NF-IL6βwt was ∼3.5–4 h, similar to that of HA/NF-IL6βK120A. There was also no appreciable difference in the patterns of nuclear localization between the wild-type and K120A mutant NF-IL6β proteins (data not shown). These results suggest that NF-IL6βK120A has the same DNA-binding activity, protein half-life and nuclear localization as NF-IL6βwt. We then moved on to address the issue of whether NF-IL6βK120A is involved in the transactivation activity of cox-2 transcription.
The regulatory domain motif of C/EBPs is critical for inhibitory domain function ( 26 ). To test whether the SUMO-conjugated NF-IL6β is important for silencing cox-2 transcription, we first examined whether NF-IL6β is a sumoylated protein in A431 cells. The in vivo -sumoylation assay was performed by transfecting cells with expression vectors of SUMO1 active form and HA/NF-IL6β or HA/NF-IL6βK120A. western blotting analysis indicated that the suHA/NF-IL6β was detectable in A431 cells. HA/NF-IL6β was endogenously sumoylated in cells ( Figure 6A , compare lane 1 with lane 3), and the sumoylated HA/NF-IL6β was increased when cells were cotransfected with SUMO1 ( Figure 6A , compare lane 3 with lane 4). However, the no SUMO-conjugated patterns were seen when HA/NF-IL6βK120A was cotransfected with the SUMO1-GG expression vector ( Figure 6A , compare lanes 3 and 4 with lanes 5 and 6). Sumoylation of proteins is agent-dependent. For example PMA can induce Elk-1, but not IκBα, sumoylation in COS7 cells ( 42 ). The constitutive sumoylation is also observed for STAT1 at the lysine703 residue ( 43 ). We therefore investigated whether EGF regulates SUMO1 conjugation to NF-IL6β in A431 cells. Using immunoprecipitation and western analysis, we found EGF attenuated the content of suNF-IL6β in A431 cells ( Figure 6B , compare lane 1 with lane 2, and lane 3 with lane 4). Finally, the transactivation activity of NF-IL6βwt and NF-IL6βK120A mutant was evaluated in co-transfection studies. The results indicate that the K120A mutant contributes higher reporter activities than the wild-type NF-IL6β on the −207/+49wt reporter ( Figure 6C , compare lanes 1 and 2 with lanes 5 and 6) suggesting that an intact lysine 120 of NF-IL6β is necessary for the repression in cox-2 transcription.
NF-IL6β is a HAT (p300) substrate
A previous study reported that PU.1 could enhance p300 mediated-C/EBPβ acetylation ( 44 ). Although C/EBPs were reported to interact with p300 ( 31 , 33 , 45 ), no evidence to date suggests that NF-IL6β is an acetylated protein including C/EBPδ. To address this issue and verify that EGF regulates cox-2 transcription through p300 ( 41 ), we investigated whether acetylated NF-IL6β (acNF-IL6β) is detectable in A431 cells. Using α-acetyl-lysine antibodies, we were able to analyse immunoprecipitated products, from the cell lysates containing endogenous p300. We found that EGF enhanced the acetylation of HA/NF-IL6β ( Figure 7A , compare lanes 3 with 4). Furthermore, exogenously expressed p300-enhanced NF-IL6β acetylation ( Figure 7B , compare lanes 3 and 4 with lanes 5 and 6). To confirm our in vivo data, an in vitro -acetylation assay was performed using purified p300 with His/NF-IL6β or His/NF-IL6βK120A. Despite removing the acetylation site at lysine 120, p300 is still able to acetylate NF-IL6βK120A at other unknown site ( Figure 7B , compare lanes 3 with 5). To examine whether NF-IL6β was involved in an increase in p300 transactivation activity, we perform a reporter assay by cotransfection of p300 and NF-IL6β expression vectors with or without knockdown expression vectors of NF-IL6β. The knockdown of exogenous expression of NF-IL6β inhibited p300/NF-IL6β-involved cox-2 reporter activity ( Figure 7C ). Additionally, to dissect whether K120A mutant participated in the p300 transaction activity, cox-2 reporter assay was carried out by combination of p300 with NF-IL6β or K120A expression vectors. Exogenous p300 proteins exhibited higher cox-2 promoter/reporter activity regardless whether wild-type or K120A mutant of NF-IL6β was used ( Figure 7D , compare lanes 1 with 3 and lanes 2 with 4). These results suggest that NF-IL6β is a HAT (p300) substrate and the position of lysine120 on NF-IL6β is not absolutely critical for p300 action.
SUMO1 represses p300-enhanced cox-2 promoter activity
Our results proposed that NF-IL6β acts as a bifunctional transcription factor on cox-2 gene suppression and activation by post-translational modification. To examine whether post-translated NF-IL6β could bind to the cox-2 promoter in vivo , a ChIP assay was performed for NF-IL6β, SUMO1 and p300. EGF treatment increased the level of binding of NF-IL6β to the cox-2 promoter. ( Figure 8A , compare lanes 3 with 4). The binding activity of SUMO1-modified proteins was abundant on cox-2 promoter but decreased after EGF treatment ( Figure 8A , compare lanes 5 with 6). Lastly, without EGF treatment, p300 is unable to associate with the cox-2 promoter ( Figure 8A , compare lanes 7 with 8). The results indicated that EGF increased the binding of p300 and NF-IL6β but reduced the binding of SUMO1-modified proteins on the A region of cox-2 promoter.
Since EGF decreased sumoylation and increased acetylation of NF-IL6β in A431 cells, we then studied whether suNF-IL6β lost its ability to interact with p300. Using western blotting assay, we demonstrated that p300 could not bind to the suNF-IL6β ( Figure 8B , compare lanes 3 and 4 with lanes 7 and 8). This result suggested that non-sumoylated or de-sumoylated NF-IL6β is available for the recruitment of p300 and support our hypothesis that suNF-IL6β mediates transcriptional inactivation, while p300 and NF-IL6β/acNF-IL6β could mediates transcriptional activation of cox-2 gene.
A number of recent reports described a role of MAPKs signaling in the induction of cox-2 gene in several cell types ( 46 – 49 ). In A431 cells, we demonstrated that U0126, a MEK1 inhibitor, and SP600125, a JNK inhibitor, repressed EGF-induced cox-2 transcription ( 5 ). Multiple cell signaling pathways including Src− focal adhesion kinase (FAK), PI3-K, p70S6 kinase, and MAPKs (p38 and ERK1/2) are involved in type I collagen-induced activation of C/EBP and CREB in serum-stimulated macrophages ( 12 ). Pharmacological inhibition of PI3-K, ERK1/2 activation, and p38 MAPK activity suppressed cox-2 induction by EGF in CaSki human cervical cancer cell line ( 50 ). Our previous study also indicated that PI3-K/p38 MAPK pathways mediate EGF-regulated NF-IL6β transcriptional activation ( 37 ). Thus, p38 MAPK contributes to the cox-2 transcriptional activation was verified, and we also provide a possible linkage of p38 MAPK /NF-IL6β pathway regulated cox-2 transcription in A431 cells.
CREB and AP-1 (c-Jun/c-Fos) have been reported to bind to the CRE cis -acting element in human cox-2 promoter ( 5 , 50 – 53 ). We previously demonstrated that c-Jun is involved in EGF-induced cox-2 transcription through transactivation of CRE binding site in A431 cells ( 5 , 41 ). AP-1 transcription factor mediates bombesin-stimulated cox-2 expression in intestinal epithelial cells ( 46 ), but has not been linked to the modulation of endogenous expression in epithelial cancer cells. In this study, we provided several lines of evidence suggesting that NF-IL6β is involved in the regulation of cox-2 by transactivating the CRE site on promoter. The C/EBPs binding consensus sequence is T (T/G)NNGNAA(T/G). Several reports ( 11 – 14 ) have shown that C/EBPβ or C/EBPδ could bind to the C/EBP motif, CTTACGCAATG, of human cox-2 promoter using in vitro -binding assay. However, the CRE motif, ATTTCGTCA CATG, is also a putative C/EBPs-like motif (the underlined sequence). By gel-shift assay, we found that purified or in vitro -translated NF-IL6β could directly bind to the C/EBP and CRE motif on cox-2 promoter ( Figure 5C ) and EGF treatment enhanced the binding of NF-IL6β to the CRE site ( Figure 3D ). The same phenomenon of C/EBPδ binding to the C/EBP and CRE sites was also observed in the LPS- and TPA-regulated cox-2 gene ( 54 ). Moreover, overexpression of NF-IL6β transactivated not only C/EBP motif but also CRE motif on cox-2 promoter ( Figure 3C ) and cotransfection of cells with c-Jun and NF-IL6β expression vectors resulted in a synergistic activation of the cox-2 promoter bearing CRE site ( Figure 2E ). These results support the notion that NF-IL6β-bound C/EBP motif coordinates with the CRE motif transactivated by NF-IL6β and c-Jun in the promoter activation of cox-2 gene upon by EGF treatment. EGF activated heterologous pGL2 promoter-C/EBP reporter activity about 4- to 5-folds ( Figure 4B ), suggesting that C/EBP motif is responsive to EGF stimulation. Activation of the basal transcription activity by overexpression of NF-IL6β was observed using pXC918 (−918/+49)( Figure 4A , lane 3), −207/+49wt (JM Wang et al ., unpublished data) and C/EBP-heterologous reporters ( Figure 4B , lane 9). However, EGF could not obviously enhance these cox-2 reporter activities, suggesting that the overexpressed NF-IL6β may have enough ability to directly recruit the cofactors and create a more intact transcription initiation complex, such as CBP or p300 ( 31 ), to mimic EGF stimulation. These results suggested that NF-IL6β plays a role at the switch control in cox-2 transcriptionalregulation.
In resting cells, a low-level of mRNA expression and SUMO-mediated suppression of the promoter have been reported ( 27 , 28 , 55 ). Sumoylation plays a role in cox-2 gene expression ( Figure 4 ). Several possible regulating proteins have been proposed for the transcription regulation of cox-2 gene, such as histone H4 ( 56 ), C/EBPs ( 5 ) and p300 ( 5 , 38 ). However, nobody has yet examined the transient change between basal level and ligand-stimulation of cox-2 gene expression. Our study focuses on a novel NF-IL6β-regulated pathway, involving post-translational modification of the protein to regulate cox-2 transcription. Recently, Kim et al . ( 25 ) reported the inhibitory function of C/EBPs by Gal4 fusion protein assay system and showed that lysine 120 was a potential SUMO-conjugated site on C/EBPδ. Therefore, we demonstrated NF-IL6β is a SUMO1 and HAT substrate and can regulate the cox-2 transcriptional regulation. We introduced the same point mutation into NF-IL6β and confirmed that lysine 120 was likely a SUMO-conjugated site ( Figure 5B and 6A ). The K120A mutant of NF-IL6β retained the same DNA binding activity, homodimerization activity, protein stability and nuclear localization as the wild-type NF-IL6β ( Figure 5C and D ). However, comparing with NF-IL6βwt, it enhanced the transcriptional activation ( Figure 6C ) suggesting that the post-translational modification might be involved in the modulation of critical protein–protein interaction. Although our reporter assay suggested that suNF-IL6β plays a repressive function role on cox-2 promoter ( Figure 6C ), we did not rule out the possibility that other SUMO-modified proteins might be involved in the maintenance of the state of repression. Several SUMO-regulated transcription factors could interact with various HDACs resulting in gene repression. For example, sumoylation of Elk-1 results in the recruitment of HDAC-2 and hence transcriptional repression at Elk-1 target genes ( 4 ). HDAC1 decreases LPS-induced cox-2 gene expression had been reported, although the mechanism of recruitment is still not clear ( 4 ). Our preliminary results from the DNA affinity precipitation assay demonstrated that suNF-IL6β could increase the interaction with some HDACs including HDAC1 (JM Wang et al ., unpublished data.) These results suggest that suNF-IL6β has the ability to recruit HDACs to regulate gene expression.
p300 acts as a coactivator for many transcription activators to modulate basal- and enhancer-regulated transcriptional activation ( 57 ). C/EBPβ binding and p300 recruitment are required for phorbol 12-myristate 13-acetate (PMA)-induced cox-2 transcription ( 38 ). We previously reported that p300 plays a functional role in EGF-induced cox-2 promoter activity ( 5 ). Transcriptional activation by C/EBPα and C/EBPβ coordinated the coactivators, CBP and p300, which promote transcription by acetylating histones and recruiting basal transcription factors ( 58 , 59 ). p300/CBP acetylates the histone tails of nucleosomes, thus favoring chromatin remodeling and activation of transcription ( 60 ). In this study, exogenously expressed p300 could increase the NF-IL6β-mediated cox-2 promoter activity ( Figure 7D ). We also demonstrated that p300 could directly acetylate NF-IL6β in vitro ( Figure 7B ) and EGF enhances acetylation of NF-IL6β in vivo ( Figure 7A ). The most interesting thing is lysine 120 of NF-IL6β can be a sumoylation or acetylation site. The same phenomenon occurs in Sp3 ( 61 ). The in vivo studies of something site capable of being both sumoylated and acetylated are difficult to carry out. Nevertheless, it can explain why the repression and activation effect of NF-IL6βK120A is not obvious. In addition, our experiments in this paper suggest that p300 and NF-IL6β/AcNF-IL6β were involved in the EGF-induced cox-2 transcription. This covalent modification might further enhance the architectural stability of the whole general basal transcription factors on cox-2 promoter. By ChIP assays, we showed that EGF enhanced cox-2 promoter binding activities of both p300 and NF-IL6β. ( Figure 8A ). The inability of suNF-IL6β to interact with p300 was also demonstrated by in vitro -binding assay ( Figure 8B ). Taken together, these results indicate that suNF-IL6β and acNF-IL6β are involved, respectively, in the silencing and activation of cox-2 transcription. The dynamic pattern of histone H4 acetylation has been demonstrated to associate with cox-2 transcription by bradykinin and IL-1β ( 62 ). Thus, different acetylation patterns of histones, in conjunction with acNF-IL6β or other modulators, may result in conformational changes of chromatin and selective association of transcription factor to the cox-2 promoter. We provided evidence to support that NF-IL6β is a sumoylated protein and also acetylated by EGF treatment in A431 cells. It brings us to deliberate the interplay between the post-translational modification and ligands-induction of NF-IL6β in downstream target genes. C/EBP proteins could be modified by SUMO-1 attachment within their RDM sequences ( 25 ). In vitro -acetylation assay shown NF-IL6β at least have two acetylation sites ( Figure 7B ). The RDM sequences of C/EBP proteins exist in the related region. Moreover, C/EBPβ was identified to be an acetylated protein ( 63 ). The similar phenomenon of acetylation site of C/EBPβ occurs nearby to its DNA binding domain. Align and compare the lysine residues in NF-IL6β−lysine 184, mouse C/EBPδ-lysine 184, human C/EBPβ-lysine 264 and mouse C/EBPβ-lysine 215, we found these lysine residues close to DNA binding domain are conserved. However, the exact acetylation site on NF-IL6β and whether the reciprocal action of suNF-IL6β and acNF-IL6β is involved in chromatin remodeling needs to be examined.
In addition to stabilization and de novo synthesis of transcription activators, a gene can also be regulated by post-translational modifications of pre-existing transcription factors. Here, we use the NF-IL6β-regulated cox-2 system to establish a model that links pre-existing NF-IL6β and its post-translationally modified form to regulate cox-2 transcription. The present study clearly indicated that NF-IL6β and its sumoylation and acetylation modifications play a functional role in the regulation of cox-2 promoter. NF-IL6β regulated the basal and EGF-induced cox-2 gene expression ( Figure 9 ). In resting cells, sumoylation of NF-IL6β attenuated the activation of cox-2 gene promoter. Upon EGF treatment, the recruitment of p300 and NF-IL6β to the cox-2 gene promoter is enhanced, while the sumoylated form of NF-IL6β on gene promoter is attenuated. It suggests that EGF treatment could result in a decrease in SUMO1-modified proteins or suNF-IL6β, an increase of NF-IL6β protein bound to the cox-2 promoter and the recruitment of p300 and involvement of NF-IL6β acetylation activate the promoter activity of cox-2 gene. These results demonstrated a possible interaction between p300 and the post-translational modification of NF-IL6β in controlling the cox-2 gene expression.
Thanks are due to Drs Wai-Ming Kan and Rong-Fong Shen for critical review of this manuscript. This work was supported by the Ministry of Education Program for Promoting Academic Excellent of University under the grant number 91-B-FA09-1-4 of Taiwan, Republic of China. Funding to pay the Open Access publication charges for this article was provided by the Ministry of Education.
Conflict of interest statement . None declared.