Biliverdin reductase, a novel regulator for induction of activating transcription factor-2 and heme oxygenase-1.

Biliverdin IXalpha reductase (BVR) catalyzes reduction of the HO activity product, biliverdin, to bilirubin. hBVR is a serine/threonine kinase that contains a bZip domain. Presently, regulation of gene expression by hBVR was examined. 293A cells were infected with adenovirus-doxycycline (Ad-Dox)-inducible hBVR cDNA. High level expression of hBVR was determined at mRNA, protein, and activity levels 8 h after induction. Cell signal transduction microarray analysis of cells infected with expression or with the control Ad-inverted (INV)-hBVR vector identified ATF-2 among several up-regulated genes. ATF-2 is a bZip transcription factor for activation of cAMP response element (CRE) and a dimeric partner to c-jun in MAPK pathway that regulates the stress protein, HO-1, expression. Northern and Western blot analyses showed increases of approximately 10-fold in ATF-2 mRNA and protein at 16 and 24 h after Dox addition. Ad-INV-hBVR did not effect ATF-2 expression. In hBVR-infected cells, levels of HO-1 mRNA and protein were increased. In vitro translated hBVR and nuclear extract containing hBVR in gel mobility-shift assay bound to AP-1 sites in the ATF-2 promoter region and to an oligonucleotide containing the CRE site. Both bindings could be competed out by excess unlabeled probe; in the presence of hBVR antibody, they displayed shifted bands. Co-transfection of hBVR with ATF-2 or c-jun promoters caused a severalfold increase in luciferase activity. hBVR modulation of ATF-2 and HO-1 expression suggests it has a potential role in regulation of AP-1 and cAMP-regulated genes and a role in cell signaling. We propose that increased expression of the protein can be used to alter the gene expression profile in the cell.

Recent studies have uncovered features of the reductase that are unrelated to its reductase activity; BVR has been characterized as a serine/threonine kinase (17) that is activated by oxygen radicals and translocates into the nucleus in response to cGMP and oxidative stress (18,19). The presence of a bZip motif, Leu 129 (ϫ6)-Leu 136 (ϫ6)-Lys 143 (ϫ6)-Leu 150 (ϫ6)-Leu 157 , which is preceded by a conserved basic domain, together with mutation analyses identified the sequence as a dimerization domain. The DNA-binding ability of the protein was confirmed by demonstrating its binding to the consensus sequence of AP-1 sites in the HO-1 promoter (19). The crystal structure of rat BVR has been solved previously (20,21). Using coordinates for rat enzyme in the predicted three-dimensional structure of human reductase, the dimerization domain was found consistent with its ability to bind DNA. Collectively, these criteria were supportive of BVR being a member of the bZip DNA binding family of transcription factors. The members of the family, which include transcription factors Myc, GCN 4 , c-Jun, CREB, c-Fos, and sYAP and ATF-2 (22)(23)(24)(25)(26)(27)(28)(29), activate cell signaling pathways, including the MAPK pathway, for proliferation, differentiation, survival, and apoptosis. HO-1 is among those genes whose expression can be up-regulated by activation of the MAPK pathway. The general consensus associates upregulation of HO-1 with an enhanced defense mechanism against stress (30 -36), and recently the product of HO activity, biliverdin, has been shown to play an essential role in the earliest stages of embryogenesis to mandate dorsal axis formation in Xenopus embryo (37).
HO-1 stress response is mediated by AP-1 binding to multiple copies of consensus sequence TGACTCA (38). The AP-1 family of proteins forms homo-or heterodimers that include c-Jun/ATF-2 heterodimer, which binds to both AP-1 site and the CRE site (TGACNTCA). In addition to HO-1, AP-1 sites are found in many promoters of genes, including growth factors, chemokines, and cytokines. ATF-2 as a homodimer binds to CRE, whereas heterodimerization of ATF-2 with c-Jun increases its affinity for AP-1 by 4-fold (39) over that of the c-Jun/c-Fos heterodimer with an increase in its association time with DNA (40). HO-1 is also induced by cAMP and CRE activation (41).
The structural and activity profiles of the reductase in vitro are consistent with its having a regulatory role in cellular functions, in an effort to further understand whether criteria ascribed to BVR does in fact have biological significance, the present study was undertaken to examine whether increased expression of BVR in the cell effects regulation of gene expression. In this study, we show that in cells infected with Ad-hBVR, induction of hBVR gene expression results in increased levels of ATF-2 and HO-1 mRNA and protein in the cell.
The BVR amino acid sequence is highly conserved from humans to cyanobacteria, (4,6,7,42,43), with 84% amino acid residue identity between human and rat proteins. The evolutionarily conserved functional and structural features include the bZip motif, the kinase activity, and having dual pH/cofactor requirements (7,44). Therefore, findings with human enzyme predictably may be also applicable to other mammalian species. The findings of the present study together with previous reports identify BVR as a novel regulator of ATF-2 and HO-1 expression and suggest that increased expression of BVR is potentially a useful approach to change the gene expression profile in the cell.

EXPERIMENTAL PROCEDURES
Construction of Adenoviral Vector Expressing hBVR-Adenovirus recombinant DNAs with hBVR were constructed as follows. First, fulllength biliverdin reductase cDNA was amplified from the clone obtained earlier in the laboratory (6) using primers 724BVR (5Ј-GTC ACG AGA TCT CGA TTA TTA GGA CGA TGA CGA TAA GAT GAA TGC AGA GCC CGA GAG GAA GTT TGG CG) and 725BVR (5Ј-GTC ACG TCT AGA TTA CTT CCT TGA ACA GCA ATA TTT CTG GAT TTC TGC). Primer 724BVR allows the introduction of FLAG (DYKDDDDK) coding sequence just upstream from ATG codon of hBVR cDNA. The resulting fragment was digested with BglII and XbaI restriction endonucleases and cloned between the appropriate sites of vector plasmid pEGFP-3C. The cDNA sequence of hBVR was verified by sequencing. For transformation, it was used dam Ϫ strain GM119 (kindly provided by S. Hattman, Department of Biology, University of Rochester), because the XbaI site in the resulting construct named pGFP-hBVR was protected by Dam methylation when maintained in non-modified bacterial recipient. The NheI-XbaI fragment of pGFP-hBVR containing the fusion EGFP-FLAG-hBVR was subcloned between the sites NheI-XbaI of intermediate vector pTRE-Shuttle2. The fact that NheI and XbaI restriction endonucleases produce identical cohesive ends allowed us to obtained constructs with two alternative orientations of the fragment: one under control of the tetracycline-regulated plasmid promoter P minCMV (pTRE-hBVR), the other in the opposite direction (pTRE-INV-hBVR). After thorough sequence analysis, both constructs were used for subcloning into tetracycline-responsive pAdeno-X vector according to the manufacturer's instructions. Recombinant DNAs named, correspondingly, Ad-hBVR and AD-INV-hBVR were purified from an XL1-Blue Gold-recipient bacterial strain and analyzed with restriction endonucleases and PCR using gene-specific primers and primers provided by Clontech to check orientation of the insert. Finally, Ad-hBVR and AD-INV-hBVR were introduced into HEK 293A cells using the Lipo-fectAMINE 2000 protocol (Invitrogen, Carlsbad, CA). The viruses were isolated from cell culture by using the AdenoPure (Puresyn, Malvern, PA) purification kit according to the supplier's recommendations. The viral titer was determined by A 260 assay in accordance with the protocol of BD Biosciences Clontech.
Cell Culture and Transfection of hBVR into 293A Cells-293A cell line (a human embryonal kidney cell line) was obtained from ATCC (Rockville, MD). Cells (3 ϫ 10 6 for RNA analysis and 1 ϫ 10 6 for protein analysis) were grown in Dulbecco's modified Eagle's medium containing 10% tetracycline-free fetal bovine serum and 1% penicillin-G/streptomycin for 18 h. Then, virus was added at a multiplicity of infection of 5 pfu/cell for Adeno-X Tet-On and 10 pfu/cell for two recombinant constructs. This ratio was found to be optimal for overexpression of hBVR. For some analyses, upon the addition of the virus, cells were collected and used as controls. For most experiments, 2 h after the addition of the virus, Dox was added at a concentration of 5 g/ml. This time point was designated in figures as the 0 point. Samples were collected at time points indicated in figures.
cDNA Microarray Analysis-Total RNA (20 g) extracted from 5 ϫ 10 6 293A line cells infected with Ad-hBVR (10 pfu/cell) and Adeno-X Tet-On (5 pfu/cell) or Ad-INV-hBVR (10 pfu/cell) and Adeno-X Tet-On (5 pfu/cell) 24 h after induction with 5 g/l doxycycline (Dox) was used for cDNA microarray analysis by SuperArray (Frederick, MD). After cDNA synthesis and Biotin d-UTP labeling, the probes were hybridized with GEArray Q Series Human Signal Transduction PathwayFinder Gene Array HS-008 membrane containing cDNA fragments from 96 marker genes associated with 18 signal transduction pathways. Analysis of the arrays was performed with "AlphaEasy" software together with "Al-phaImager TM 3400" system (Alpha Innotech Corp., San Leandro, CA) to convert signal intensity to a numeric number. All signal intensities were normalized to housekeeping genes ␤-actin, glyceraldehyde-3-phosphate dehydrogenase, cyclophilin A, and ribosomal protein L13␣. Changes in genes expression were calculated as the ratio between intensities of signals after overexpressing wild type (Ad-hBVR) and inverted (Ad-INV-hBVR, control) adenoviral constructs during the microarray analysis giving a -fold increase/decrease: Ad-hBVR/Ad-INV-hBVR. Any value over 2-fold change was taken as a significant and is presented in Table II. Using the same criteria, any significantly suppressed gene was indicated as negative (Ϫ) -fold value.
ELISA-HO-1 protein was measured by using an ELISA kit developed by Stressgen Bioreagents (Victoria, British Columbia, Canada) according to the manufacturer's instructions.
In Vitro BVR Protein Translation, Nuclear Extraction, and Gel Mobility Shift Assay-Both in vitro translated BVR protein and nuclear extract from 293 cells transfected with Ad-hBVR were used in the gel shift assay experiments. In vitro BVR protein translation was performed using a TNT Quick Coupled Translation System from Promega (Madison, WI). Briefly, full-length hBVR cDNA was cloned into a pcDNA3 expression vector downstream from the T7 RNA polymerase promoter. 2.0 g of recombinant plasmid DNA obtained was used for protein translation with TNT Quick Master Mix in a 50-l reaction volume for 90 min at 30°C. Nuclear extract was isolated from 293 cells transfected with Ad-BVR. 10 7 cells were harvested at time points 0, 6, and 24 h after induction of hBVR expression; cells were washed with cold phosphate-buffered saline and then lysed with hypotonic buffer. Nuclei were collected by centrifugation at 800 ϫ g at 4°C for 10 min and resuspended in 200 l of buffer containing 20 mM HEPES, pH 7.9, 25% glycerol, 0.4 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 M phenylmethylsulfonyl fluoride, 10 mM KCl, and 0.5 mM dithiothreitol. The nuclei were extracted on ice for 30 min and followed by 30-min centrifugation at 4°C. Supernatant was then collected. The in vitro synthesized protein or nuclear extract were then analyzed in gel shift assay for their DNA-binding capability to a cAMP regulatory element (CRE) and an AP-1 consensus oligonucleotide (47). The sequences of oligonucleotides used in the present study are listed in Table I. The oligonucleotides were labeled with [␥-32 P]ATP by T4 kinase (Invitrogen) according to the manufacturer's instructions. For DNA binding assay, 3 l of in vitro translated protein or nuclear extract was preincubated with 2 l of binding buffer (20% glycerol, 5 mM MgCl 2 , 2.5 mM EDTA, 2.5 mM dithiothreitol, 250 mM NaCl, and 50 mM Tris-HCl, 0.25 mg/ml poly(dI: dC)) in 8 l reaction volume for 10 min at room temperature. Then, 2 l of labeled oligonucleotides were added, and samples were incubated for an additional 20 min at room temperature. The DNA-protein binding complexes were electrophoresed on a 6% non-denature polyacrylamide gel and processed for autoradiography. The unlabeled competitor DNA was used to determine the specificity of the binding. To identify proteins in the DNA-protein complexes, supershift experiments were performed with rabbit polyclonal anti-hBVR antibodies (45). For positive control, ATF-2 protein was in vitro translated using a pcDNA3 plasmid containing a 762-bp ATF-2 cDNA sequence as template (generous gift from Hicham Drissi, University of Rochester). The translated ATF-2 protein was then used in the CRE binding reaction. Each experiment was repeated at least twice to ascertain the reproducibility of results.
Luciferase Assays-Transfection of 293 cells in 24-well plates was performed by using LipofectAMINE (Invitrogen) with 0.4 g of the ATF-2 promoter (Ϫ612 to ϩ33) inserted into pGL3 reporter vector (pGL3/ATF-2), 0.4 g of c-jun promoter (Ϫ225 to ϩ150) (49), and recloned in PGL3 luciferase containing vector, 0.4 g of c-Jun promoter mutant (pGL3/c-Jun Ϫ ), containing a mutated AP-1 site (50) (both provided by Drs. McCance and Baglia, University of Rochester, Rochester, NY), 0.4 g of either empty pcDNA3 or pcDNA3 containing hBVR (pcDNA3/hBVR), 0.4 g of pCMV ␤-galactosidase plasmid, or 0.4 g of empty pGL3 vector (without the inserts). After 48 h of transfection, cells were lysed by Promega lysis buffer. The pellets were discarded by centrifuging cell lysates for 5 min at 4°C at maximum speed, and 5 l of obtained supernatant were applied for luciferase assays by using Promega kit (Madison, WI). Transfection efficiency was assessed by determining ␤-galactosidase activity, and luciferase activities were normalized against ␤-galactosidase activities.

Verification of the Adenoviral Expression Vector for BVR
Expression-To investigate the potential of hBVR to alter gene expression in the cell, a two-component adenovirus system provided by Clontech was used to develop an Ad-hBVR construct (Fig. 1). Functional hBVR cDNA was tagged with a FLAG sequence and cloned downstream in fusion with the EGFP gene taken from vector pEGFP-3C. The presence of FLAG and EGFP allowed for confirmation of BVR expression and cellular localization. The expressing cassette was subcloned into the shuttle vector pAdeno-X in two alternative orientations, one (named pAd-hBVR) under control of vector inducible promoter, the other (named pAD-INV-hBVR) in the opposite orientation. The latter construct was designed to use in experiments as a dominant-negative mutant of the overexpressing clone pAd-hBVR. The helper virus Adeno-X Tet-On, the second component of the system, provides reverse tetracycline-controlled transactivator, which in the presence of doxycycline (Dox) binds to Tet-responsive element (TRE) located upstream of the minimal immediate early promoter of cytomegalovirus (P minCMV ) and subsequently activates transcription of the gene cloned under control of the promoter. This approach allowed for the overexpression of specific products of pAd-hBVR using regulated induction and the detection of these products with anti-BVR or anti-FLAG antibodies or with EGFP fluorescence, which opens a broad range of opportunities such as study of hBVR trafficking (Fig. 2), among others. In preliminary experiments, it was found that Dox in a concentration of 5 g/ml provides the highest and most reproducible induction of recombinant protein (concentrations of 0.01, 0.1, 0.5, 1.0, and 2.0 g/ml were also studied).
Inducibility and expression of GFP-FLAG-hBVR mRNA was determined by Northern blot analysis (Fig. 2a). A prominent signal that corresponded to the size of fusion BVR mRNA plus GFP and FLAG messages was observed 24 h after the infection of 293 cells with wild-type construct. During the exposure period, exceedingly low levels of hBVR mRNA were detected in 293 cells, in cells infected with virus only, or in cells infected with INV Ad-hBVR construct. The finding suggested that the increase is not due to activation of hBVR transcription by the virus. Robust increase in mRNA message of BVR was accompanied by a marked increase in rate of biliverdin reduction (Fig. 2c).
The time course of hBVR expression was examined. As shown in Fig. 3, there was a time-dependent increase in protein (Fig. 3, a and b) and activity (Fig. 3c) in 293A cells transfected with hBVR. Prominent induction of hBVR was detected at the 8-h time point using an antibody raised against hBVR. Western blot analysis using anti-BVR or anti-FLAG antibodies show practically identical patterns (Fig. 3, a and b). It should be noted that endogenous BVR was detected with anti-BVR antibodies in non-induced 293 cells only when 20-fold higher   -AGA GAT TGC CTG ACG TCA GAG AGC TAG-3Ј  4ϫ CRE  5Ј-(AGC CTG ACG TCA GAG) x4-3Ј  1ϫ mut CRE  5Ј-AGA GAT TGC CAT GGC ATC GAG AGC TAG-3Ј  AP-1

nt106) AAT AGT GAC TAG TTT TGG GGT G A C A G T AGT (nt 120) ATA AG T T A T T C A ACT TAT G-3Ј
a Only the upper strands of double-stranded fragments are given. b The DNA fragment with three potential AP-1 binding sites as predicated by Malinspector software. The double underlined sequence has a similarity of 94% with AP-1 consensus sequence (T(G/T)AGTCA); the other two dash-underlined sequences have a similarity of 87% and 86%, respectively. The initiation ATG codon of ATF-2 original ORF is shown in boldface.
amounts of cell extract was used for the analysis. Non-induced cells showed low level reductase activity; upon addition of Dox it was gradually increased following a general pattern as that of expression of BVR protein. The peak activity was noted at 24 h after the Dox induction. Similar to non-transfected cells, cells infected with virus only or infected with INV hBVR construct showed low level reductase activity. Cellular localization of hBVR was detected from EGFP green fluorescence. As shown in Fig. 3d, hBVR does traffic in the cells as indicated by the impressive EGFP green fluorescence of the nucleus. BVR was recently thought to be exclusively a cytosolic protein; the observed nuclear translocation of the protein is consistent with its function as a regulator of gene expression. Fig. 3e shows cells visualized by visible light.
Overexpression of BVR in 293 Cells Results in an Increase of ATF-2 Protein Expression-To test the genes that were affected by BVR, total RNA was extracted from the 293A cells 24 h after infection with virus containing either wild type BVR or with inverted construct and subjected to gene micro array analysis of cell signaling genes. A number of genes were identified by the analysis. Only the genes with more than 2-fold changes in signal intensities are listed in the Table II; the brief description of their function is given in the table legend. The genes are representatives of such signal transduction pathways as follows: Wnt pathway (c-jun and WISP3), Hedgehog pathway (HHIP), TGF-b pathway (CDKN1B and CDKN2B), survival pathway (BCL2, BCL2L1, BIRC2, and c-jun), p53 pathway (GADD45), NF-B pathway (PECAM1), CREB pathway (CYP19A1), Jak-Stat pathway (CSN2), estrogen pathway (EGFR), calcium and protein kinase C pathway (PRKCA), phospholipase C pathway (PTGS2), and stress response pathway (ATF-2, HSF1, HSPB1, and HSPCA). Taking into account that hBVR is a potential transcription factor for AP-1 regulated genes (19), and the members of the list of genes identified by microarray analysis include AP-1 regulated genes, we examined whether their transcription could be directly effected by overexpressed hBVR protein. The ATF-2 gene, which contains

FIG. 2. Analyses of biliverdin reductase message and activity of 293A cells infected with adenovirus hBVR expression system.
Cells were infected with wild-type Ad-GFP-FLAG-hBVR construct, inverted hBVR construct, or the virus itself. 293 cells were used as the control. Northern blot analysis, using glyceraldehyde-3-phosphate dehydrogenase as the loading control, was carried out as described under "Experimental Procedures" 24 h after infection. 5 g of RNA was used in each lane. BVR activity was measured at pH 6.7 with NADH as the cofactor by measuring the rate of conversion of biliverdin to bilirubin (4).
To confirm whether BVR in fact regulates ATF-2 mRNA and protein expression, 293 cells were infected with Ad-hBVR construct. Cells infected with the INV-hBVR were used as controls. The transfectants were harvested at different time points after antibiotic induction and used for Northern blot analysis of ATF-2 mRNA and ATF-2 protein by Western blot analysis. As shown in Fig. 4, ATF-2 mRNA was minimally detected in cells prior to the addition of the antibody. ATF-2 mRNA was increased with tissue and peak levels were detected at 16 h. The signal for glyceraldehyde-3-phosphate dehydrogenase, the control for loading, was essentially constant over the duration of the experiment. Infection with hBVR also resulted in a significant increase of ATF-2 protein expression (Fig. 4c). The increase was detectable 4 h after induction and peaked at the 24-h time point. The DNA binding of ATF-2 is enhanced by N-terminal phosphorylation (48). To test whether the phosphorylated form of ATF-2 was also increased by overexpression of BVR in the 293 cells, the expressed protein was probed with antibody to phospho-Thr 69/71 ATF-2. As shown in Fig. 4d, an increase in the phosphorylated form of ATF-2 is detected 16 h after induction with antibiotic. The finding suggests that hBVR effects ATF-2 post-translational modification. At this time, it is not evident whether the increase in phosphorylation of ATF-2 is the result of direct interaction of hBVR with ATF-2 or reflects modulation of other kinases that phosphorylate ATF-2. As shown in Fig. 4e, ATF-2 levels were not affected when cells were transfected with the inverted construct.
hBVR Binds to the ATF-2 Promoter-Because overexpression of hBVR in 293A cells resulted in a significant increase in ATF-2 mRNA, we questioned whether hBVR interacts with consensus sequences in the ATF-2 promoter region. We previously showed that hBVR binds to the AP-1 sites in the HO-1 promoter region (19). Using Malinspector software, three potential AP-1 binding sites were predicted in the 1-kb ATF-2 promoter region. A DNA fragment containing the three potential binding sites was extracted from the ATF-2 promoter by HindIII and NcoI digestion. As shown in Table I, one binding site has a similarity of 94% with an AP-1 consensus sequence (TGTAGTCA), the other two have a similarity of 87 and 86%, respectively. An AP-1 binding assay was carried out using hBVR translated in vitro using a TNT protein translation system, and the ATF-2 promoter DNA fragment was labeled with [␥-32 P]ATP. As shown in Fig. 5a, in the gel mobility shift assay, a prominent protein plus DNA signal was detected. To verify the specificity of hBVR-AP-1 site binding, unlabeled DNA was used for competition analysis. As seen in lane 3, the intensity of the gel-shift band was nearly abolished when unlabeled DNA at a concentration 10ϫ excess of the labeled DNA was present in the binding assay. Specificity of binding was further substantiated by the observation that adding hBVR specific antibody to the reaction mixture (lane 6) resulted in a supershifted band.
To explore the potential significance of hBVR⅐DNA complex formation, as detected by the gel shift assay with in vitro translated hBVR, nuclear extract isolated from 239 cells transfected with Ad-BVR at times 0, 6, and 24 h after induction of hBVR expression was used for DNA binding assay. As shown in Fig. 5b, the gel shift assay revealed a time-dependent increase in protein⅐DNA complex formation. Addition of anti-hBVR antibody resulted in the appearance of a high mobility band. In  (a and b) and Western (c-e) blot analyses were carried out as described under "Experimental Procedures." Total RNA was isolated from cells transfected with Ad-hBVR and used for Northern blot analysis. a, Northern blot analysis of ATF-2 mRNA. b, GAPDH mRNA, which was used as a control for RNA integrity and loading. Extracts isolated from cells transfected with the indicated Ad-hBVR constructs were obtained and used for Western blot analysis. c, membrane was probed with anti-ATF-2 antibody. d, the same membrane was striped and re-probed with anti-phosphorylated ATF-2 antibody. e, lysate from the cells transfected with Ad-INV-hBVR was probed with anti-ATF-2 antibody.
FIG. 5. hBVR binds to ATF-2 promoter. AP-1 binding assay was carried out using either in vitro translated hBVR by a TNT protein translation system (a) or nuclear extract from 293 cells transfected with Ad-hBVR expression vector (b). A DNA fragment upstream of the ATF-2 ATG codon shown in Table I was used as a binding probe. A supershift assay was carried out using polyclonal antibody to hBVR. In a: Lanes: 1, translation system plus 32 P-labeled DNA probe without hBVR; 2, 4, and 5, translated hBVR plus labeled DNA; 3, translated hBVR plus labeled DNA plus 10-fold excess of unlabeled DNA; 6, translated hBVR plus labeled DNA plus antibody to hBVR. In b: Lanes 1-3, 32 P-labeled AP-1 DNA probe plus nuclear extracts isolated from 239 cells transfected with Ad-BVR at times 0, 6, and 24 h, respectively, after induction expression of hBVR. Lane 4, 32 P-labeled AP-1 DNA probe plus nuclear extracts isolated from 239 cells transfected with Ad-BVR at time 24 h after induction expression of hBVR plus antibody to human BVR. BVR and DNA complex and supershifted band by anti-BVR are indicated on the right. The protein doublet seen in a is believed to be a result of nonspecific binding of AP-1 oligonucleotide to certain protein components in the TNT Quick Coupled Translation System. Conditions of electrophoresis are described under "Experimental Procedures." addition, a slightly shifted but smeared band appeared. This observation may reflect the length of the AP-1 probe (300 bp). The findings suggest that hBVR may be among the various factors that could effect ATF-2 transcription.
BVR Binds to ATF/CRE Site-The consensus sequence of CRE, which is the binding site for dimeric ATF-2, differs from the AP-1 binding site by the presence of one added nucleotide to that of seven nucleotides AP-1 binding sites (TGACNTCA versus TGACTCA). Therefore, we questioned whether hBVR also binds to CRE. For this, gel mobility-shift assay was performed with in vitro translated hBVR protein and oligonucleotides containing CRE consensus sequence as listed in Table I. As shown in Fig. 6a, incubation of ␥-32 P-labeled CRE oligonucleotide with translated hBVR protein formed a complex that was detected by the gel shift assay. To verify the sequence specificity of CRE binding, unlabeled oligonucleotide competition assay was performed. As shown in the figure, the intensity of the gel shift band was decreased in the presence of unlabeled CRE oligonucleotide. In the presence of unlabeled DNA at concentrations 2ϫ, 5ϫ, and 10ϫ in excess of the labeled DNA signal for hBVR⅐CRE complex formation was reduced by about 30%, 60%, and 80%, respectively. The positive control for the experiment was in vitro translated ATF-2. As noted in the figure, an ATF-2⅐DNA complex was detected. Binding was not detected when DNA, with one copy of CRE or mutated CRE sequence (Table I), was used. Furthermore, the gel shift assay using nuclear extract isolated from 239 cells transfected with Ad-hBVR at times 0, 6, and 24 h after induction of hBVR expression shows a time-dependent increase of BVR⅐CRE binding and addition of anti-hBVR antibody to the DNA binding reaction mixture resulted in a supershifted band (Fig. 6b). These results indicate that hBVR protein is capable of binding to CRE site and suggest that hBVR binds to DNA sequences similar to those that CRE-binding proteins, such as ATF-2, bind.
BVR Up-regulates Transcription of ATF-2 and c-jun-Transient transfection assays were performed to examine the effect of hBVR expression on ATF-2 promoter directing CRE-mediated gene expression, and c-jun promoter directing AP-1-mediated gene expression (Fig. 7). The sequence from Ϫ612 to ϩ33 of ATF-2 promoter and Ϫ225 to ϩ150 of c-jun promoter, a region that is devoid of DNA binding sites except for 2 AP-1 sites (49), was inserted in the pGL3 vector containing luciferase reporter sequence, and activity of the promoters was measured using luciferase assay kit (Promega, Madison, WI). As expected, ATF-2 and c-jun promoters alone produced severalfold increase in luciferase activity. Whereas minimal luciferase activity was detected in cells transfected with pcDNA⅐hBVR alone or co-transfected with luciferase reporter empty vector and pcDNA3⅐hBVR. Co-transfection of the hBVR expression vector with either ATF-2 or c-jun promoter caused an additional ϳ6-fold increase in luciferase activity. Luciferase activity was similar to the basal promoter activity when cells were transfected with empty pcDNA3 vector. Furthermore, co-transfection of hBVR with reporter vector containing c-jun promoter reporter mutated at its TRRE site, which results in an unresponsive promoter (50), did not activate luciferase activity. Taken together, these data plus observations made by Northern blotting analysis and gel shift assays suggest that BVR is capable of activating both ATF-2 and c-jun transcription.
HO-1 Expression Is Increased in hBVR-overexpressing Cells-The consequence of increased expression of hBVR on HO-1 was examined. Previous studies using antisense BVR suggested the presence of BVR is required for HO-1 stress response (19), which requires AP-1 activation. As noted earlier, ATF-2 forms a heterodimer with c-Jun with high affinity for AP-1 sites. Presently, we examined whether increased expression of BVR would affect HO-1 expression. As shown in Fig. 8, HO-1 mRNA in 293 cells was increased at 48 h after addition of Dox (Fig. 8a). The observed increase in HO-1 mRNA did not result from differences in sample loading (Fig. 8c). Moreover, increase in HO-1 protein levels, as measured by ELISA, was consistent with an increase in HO-1 mRNA (Fig. 8c). The concentration of HO-1 in control cells was 46.7 ng/ml/mg of total cell proteins with that of Ad-hBVR-infected cells measuring 238.9 Ϯ 8% of the control cells. DISCUSSION We previously found the following features of BVR: the presence of the bZip motif in the primary structure of hBVR, together with kinase activity of the reductase and its nuclear localization in response to oxidative stress and cGMP (17)(18)(19), consistent with those of a protein that could potentially effect gene regulation. This concept was further reinforced by finding that hBVR could bind to a HO-1 promoter fragment containing AP-1 sites and flanking nucleotides (38). It remained, however, to be established that in the cell hBVR potentially plays a role in regulation of gene expression. To explore the potential targets for hBVR in the cell, the protein was overexpressed using an adenovirus construct, and altered gene expression was examined by microassay. This approach, namely overexpression FIG. 6. hBVR binds to ATF/CRE consensus sequence. The assay was carried out using hBVR translated in vitro with a TNT protein translation system and nuclear extracts from 293 cells transfected with Ad-BVR expression vector, ␥-32 P-labeled DNA fragments containing four or one ATF/CRE sites were used in the gel shift binding experiment (Table I). For competition analysis, unlabeled CRE containing oligonucleotide nucleotide was added at 2ϫ, 5ϫ, and 10ϫ excess that of labeled CRE oligonucleotide. a, left, lanes are: 1, translation system without hBVR; 2, translated hBVR plus labeled CRE oligonucleotide; 3-5, translated hBVR plus unlabeled CRE oligonucleotide nucleotide at 2ϫ, 5ϫ, and 10ϫ excess of that of labeled CRE, respectively; 6, translated ATF-2 plus labeled CRE oligonucleotide as a positive control for CRE binding. b, left, lanes 1-3, 32 P-labeled CRE DNA probe plus nuclear extracts isolated from 239 cells transfected with Ad-BVR at times 0, 6, and 24 h, respectively, after induction expression of hBVR; lane 4, 32 P-labeled AP-1 DNA probe plus nuclear extracts at time point 24 h after induction expression of hBVR plus antibody to human BVR. of the target gene(s) in adenovirus vectors and microarray analysis, is rather commonly used (51)(52)(53)(54).
The present investigation, in which used a cell culture system, supports this potential and defines hBVR as a regulator of ATF-2 (CREB-2) and HO-1 expression. The previously reported (19) AP-1 binding activity of hBVR was confirmed and was extended to the CRE site. CRE differs from the AP-1 site by one added base. Binding of in vitro translated hBVR and nuclear extract of cells overexpressing hBVR to the consensus AP-1 and CRE, however, does not indicate that hBVR directly regulates ATF-2 or HO-1 transcription; rather, these findings together with the results of mRNA and protein analyses support a role for hBVR in regulation of ATF-2 and HO-1 gene expression and, potentially, a component of the basal transcriptional machinery for their expression.
Increased levels of ATF-2 in cells overexpressing hBVR may affect a wide range of cellular functions. ATF-2 is a constitutive transcription factor whose expression, unlike that of c-Jun, which is an inducible factor, is not dependent on extracellular signals (47,48). Transcriptional factors, Fos (Fos-like), c-Jun, and ATF-2, like BVR, are "leucine zipper" type factors and bind DNA in homodimeric or heterodimeric forms. The availability of the dimeric partner determines their preference for DNA binding sites. In the case of ATF-2, it can form a heterodimer with c-Jun. And, when its levels are increased, it effectively competes with c-Fos, the usual dimer partner of c-Jun. The ATF-2/c-Jun heterodimer preferentially binds to the 7-base AP-1 sites (TGACTCA) rather than the usual site of ATF-2 , CRE (TGACNTCA) (Ca/cAMP response element) (48,55,56). The ATF-2/c-Jun dimer DNA complex is more stable than the c-Fos/c-Jun⅐DNA complex (39). Moreover, heterodimerization not only alters ATF-2 binding with remarkable variation in affinity among different CRE sites (57) but also gene regulation activity of the dimeric partner. The ability of hBVR to induce ATF-2, therefore, is likely to change the profile of gene expression in the cell. In the case of heterodimerizing with c-Jun to form a component of AP-1 complex, the association likely will result in a wide spectrum of changes in the cell. AP-1 sites are found in promoters of a variety genes that control transcription of growth factors, chemokines, and cytokines (58). AP-1 is activated by mitogens, oncoproteins, cytokines, and stress-inducing stimuli.
ATF-2, in addition to influencing the dimeric composition of c-Jun, plays an important role in induction of c-jun gene expression and its autoregulatory transactivation by c-Jun protein. ATF-2 also forms dimers with NF-B (59). NF-B is a transcriptional activator of a number of proinflammatory mediators, such as cytokines, growth factors, and adhesion mole- 293 cells were transfected with ATF-2 (pGL3/ATF-2, 0.4 g), or with AP-1-responsive c-jun luciferase reporters (pGL3/ c-jun, 0.4 g). The controls are noted in the figure. pCMV␣-galactosidase (0.4 g) vector was also included under all conditions for normalization of transfection. a, cells transfected with ATF-2 reporter were co-transfected with pcDNA3 vector expressing hBVR. b, cells transfected with c-jun promoter (pGL3/c-jun) were cotransfected with pcDNA3/hBVR. After 48 h, cells were lysed and assayed for luciferase and ␤-galactosidase activities. Luciferase activity, normalized for transfection efficiency against ␤-galactosidase activity, is presented as mean Ϯ S.D. of the fold activation when compared with that of the corresponding vector alone (pGL3/ATF-2 or vector pGL3/c-jun). The luciferase activities of vectors alone were assigned the value of 1. The results are representative of two experiments with quadruplicate wells.
cules (59,60). It is reasonable to suggest that binding to enhancer elements of genes target for two classes of NF-B and their family of homo-or heterodimeric forms would be affected by an increase in ATF-2 in the cell. Activation by hBVR of ATF-2, therefore, may be another mechanism by which NF-B activity is modulated. Because NF-B is a component of the signaling pathways that lead to conditions such as vascular inflammation and atherogenesis (59,60), the ability of BVR to increase ATF-2 gene expression may be of potential utility in therapeutic settings to regulate inflammatory processes.
It is reasonable to suspect a relationship between an increase in ATF-2 expression and increase in that of HO-1 expression. Activation of the MAPK signaling pathway and c-Jun/c-Fos⅐DNA binding is a key mechanism for the induction of HO-1. HO-1 is also responsive to gene activation by cAMP (61)(62)(63). Accordingly, it is reasonable to suspect that induction of ATF-2 has a direct effect on HO-1 expression. However, the possibility cannot be ruled out that the presently observed induction of HO-1 gene expression in cells overexpressing hBVR is independent of ATF-2. In this case, at the minimum, two mechanisms can be considered: one would involve the removal of the product, biliverdin, by hBVR functioning in a reductase capacity, and the other would involve activation of other genes that control HO-1 gene expression, for instance, that of c-jun. As noted in Table II, c-jun is among the list of genes identified by microarray analyses to be up-regulated in cells infected with Ad-hBVR. Of course, that all or a combination of the noted mechanisms are involved in increased expression of HO-1 is a distinct possibility. Considering the wide range of cellular functions that are modulated by HO activity and products of heme catalysis (30 -36), the finding that hBVR plays a role in regulation of HO-1 expression could be utilized as a method for modulating a wide spectrum of cellular processes. Also, modulation of hBVR expression in the cell may prove useful in regulation of those genes that are involved in proliferation, differentiation, survival, and apoptosis.