Alpha-Pal/NRF-1 regulates the promoter of the human integrin-associated protein/CD47 gene.

Integrin-associated protein (IAP or CD47) is expressed in a variety of tissues, including the nervous system and immune system. To understand how cells control the expression of the IAP gene, we cloned the 5'-proximal region of the human IAP gene and investigated IAP promoter activity by transient transfection. RT-PCR confirmed the expression of IAP transcripts in human neuroblastoma IMR-32 and hepatoma HepG2 cells. Deletion analysis identified a core promoter of the human IAP gene located between nucleotide positions -232 and -12 relative to the translation initiation codon in these two cell lines. Site-directed mutagenesis and gel electrophoretic mobility shift assay identified a alpha-Pal/NRF-1 binding element within the IAP core promoter. Supershift assays using the alpha-Pal/NRF-1 antiserum confirmed the binding of this transcription factor on the alpha-Pal/NRF-1 site. Overexpression of the DNA binding domain of alpha-Pal/NRF-1 in cells enhanced DNA-alpha-Pal/NRF-1 binding in vitro. Furthermore, overexpression of full-length alpha-Pal/NRF-1 significantly enhanced IAP promoter activity while overexpression of dominant-negative mutant reduced promoter activity both in the cultured human cell lines and primary mouse cortical cells. These results revealed that alpha-Pal/NRF-1 is an essential transcription factor in the regulation of human IAP gene expression.

Integrin-associated protein (IAP or CD47) is expressed in a variety of tissues, including the nervous system and immune system. To understand how cells control the expression of the IAP gene, we cloned the 5-proximal region of the human IAP gene and investigated IAP promoter activity by transient transfection. RT-PCR confirmed the expression of IAP transcripts in human neuroblastoma IMR-32 and hepatoma HepG2 cells. Deletion analysis identified a core promoter of the human IAP gene located between nucleotide positions ؊232 and ؊12 relative to the translation initiation codon in these two cell lines. Site-directed mutagenesis and gel electrophoretic mobility shift assay identified a ␣-Pal/ NRF-1 binding element within the IAP core promoter. Supershift assays using the ␣-Pal/NRF-1 antiserum confirmed the binding of this transcription factor on the ␣-Pal/NRF-1 site. Overexpression of the DNA binding domain of ␣-Pal/NRF-1 in cells enhanced DNA-␣-Pal/ NRF-1 binding in vitro. Furthermore, overexpression of full-length ␣-Pal/NRF-1 significantly enhanced IAP promoter activity while overexpression of dominant-negative mutant reduced promoter activity both in the cultured human cell lines and primary mouse cortical cells. These results revealed that ␣-Pal/NRF-1 is an essential transcription factor in the regulation of human IAP gene expression.
Integrin-associated protein (IAP), 1 also designated as CD47, is a multifunctional membrane protein that is expressed widely in the nervous system, immune system and many other tissues (1,2). In the adult rat central nervous system, IAP was related to memory formation of an aversive learning task (3). In good memory rats trained in the inhibitory avoidance learning paradigm, the mRNA level of IAP in the hippocampus was increased. Injection of antisense oligonucleotides or the monoclonal antibody of IAP into the rat hippocampus impaired memory formation (3,4). IAP-deficient mice showed deficits in memory retention in a similar behavioral paradigm (5).
In the peripheral tissues, IAP was first discovered as a cell surface protein associated with integrin ␣ v ␤ 3 and it was involved in the enhancement of neutrophil adhesion, chemotaxis, and phagocytosis triggered by an extracellular matrix (1,6,7). IAP is also a functional component of several processes, including the transepithelial migration of neutrophils (8), chemotaxis of endothelial cells and smooth muscle cells (9,10), spreading and aggregation of platelets (11), and modulation of T-cell activation (12,13). Moreover, IAP functions as a selfcheck marker on red blood cells to prevent their clearance by macrophages (14). IAP mRNA and protein sequences are conserved among humans, mice, and rats (3,15), suggesting that it is evolutionally important to biological functions. IAP mRNA has five alternative splicing forms. These forms are also conserved in evolution (3,(15)(16)(17). In humans and mice, different forms of IAP mRNA were expressed at varying levels in different tissues; macrophage and endothelial cells expressed predominantly form 2 mRNA and brain tissues expressed predominantly form 4. Why different tissues express different levels and forms of IAP mRNA and how the IAP gene is regulated are, however, unknown. To answer these questions, we investigated the regulation of IAP gene promoter in human neuroblastoma and hepatoma cell lines by using luciferase reporter and gel electrophoretic mobility shift assays. We found that ␣-Pal/NRF-1 in the core promoter region is a positive regulator of the human IAP gene.

EXPERIMENTAL PROCEDURES
Cell Culture-Human neuroblastoma IMR-32 (CCRC 60014) and hepatoma HepG2 (CCRC 60048) cell lines were purchased from Culture Collection and Research Center, Food Industry and Development Institute, Hsinchu, Taiwan. Cells were grown in minimum essential medium Eagle with Earle's salt base (Sigma) supplemented with 10% fetal bovine serum (HyClone, Logan, UT) in a humidified atmosphere containing 5% CO 2 at 37°C.
RNA Isolation and Reverse Transcription (RT)-PCR-Total RNA was isolated from the cultured cells using TRIzol reagent (Invitrogen). RT-PCR was performed as described previously (3). Briefly, total RNA (2 g) was reverse-transcribed into cDNA in 20 l of 1ϫ first strand buffer containing 0.5 g of oligo(dT) as a primer, 500 M dNTP, and 200 units of SuperScript II (Invitrogen). PCR was performed in 20 l of 1ϫ PCR buffer containing 2 l of RT products, 1 unit of AmpliTaq DNA polymerase (Roche Applied Science), 200 M dNTP, 1.5 mM MgCl 2, 0.5 M [ 35 S]dATP (Amersham Biosciences), and 0.4 M primer pair. We used the primer pair that can distinguish the alternative splicing forms of IAP mRNA, Hiap14: 5Ј-TAA CCT CCT TCG TCA TTG CC and Hiap15: 5Ј-CGT AAG GGT CTC ATA GGT G. The PCR parameters were 94°C for 30 s, 53°C for 30 s, and 72°C for 30 s for 30 cycles, followed by a final elongation at 72°C for 7 min. PCR products were analyzed on a 6% polyacrylamide-urea gel (acrylamide/bisacrylamide 19:1, 8 M urea in 1 ϫ Tris borate-EDTA buffer). The gel was finally dried and analyzed by autoradiography. The image of cDNA bands was scanned by the Scan-Jet 4C scanner (Hewlett Packard). The optical densities of cDNA bands were quantified with the one-dimensional advanced Universal Software (American Applied Biotechnology, Fullerton, CA).
Plasmids-pGL3-Basic and pRL-TK luciferase reporter vectors (Pro-mega, Madison, WI) were used for IAP promoter reporter assays. A 7.7-kb human genomic clone (pIAP38) containing exon 1 of the IAP gene and 5Ј-upstream region was kindly provided by Dr. F. P. Lindberg. A plasmid, pBSII-445, was generated by inserting the 445-bp SacII fragment from pIAP38 into the pBlueScript II (SKϪ) vector (Stratagene, La Jolla, CA). The SacI/XhoI fragment from pBSII-445 was then inserted into the pCRII vector (Invitrogen) to create pHIAP445 for subsequent constructions.
The reporter construct pGL3-272 was generated by inserting the XmaI/XhoI fragment from pHIAP445 into the pGL3-Basic vector. Different restriction enzyme-digested fragments from pIAP38 were then ligated into pGL3-272 to create a series of human IAP promoter constructs, including pGL3-1554, pGL3-730, and pGL3-456. Another series of human IAP promoter constructs containing shorter fragments than the insert in pGL3-272, including pGL3-232, pGL3-218, pGL3-209, pGL3-198, pGL3-191, pGL3-159, and pGL3-92, were similarly made by using differential forward primers on the IAP promoter with the KpnI site at the 5Ј-end and a common reverse primer (GLprimer2) on pGL3-Basic in the backbone of pGL3-272 for PCR to obtain the shorter fragments, which included differentially truncated IAP promoter regions and a common vector sequence containing the HindIII site. These fragments were then digested by KpnI and HindIII and inserted into pGL3-Basic.
Promoter constructs containing nucleotide substitutions in the sequence motifs of Sp1 and ␣-Pal/NRF-1 were individually generated by PCR amplification with primer pairs spanning the mutant nucleotides according to the protocol of site-directed mutagenesis by overlap extension (18). The plasmids pGL3-232m1, pGL3-232m2, and pGL3-232m3 were constructed in the backbone of pGL3-232 using primer pairs containing the introduced mutations. The Sp1 site GGGGCGGGGC was mutated into GTTGCTTGGC in the plasmid of pGL3-232m1. The ␣-Pal/NRF-1 element TGCGCGTGCGCG was mutated into TTTGCGT-GCGCG, and TGCGCGTTTGCG in the plasmid of pGL3-232m2 and pGL3-232m3, respectively.
Transfection and Dual-luciferase Assay-IMR-32 and HepG2 cells (1.5 ϫ 10 5 ) were plated in each well of six-well plates. Transient transfection was carried out by the calcium phosphate precipitation method (19). The plasmid pRL-TK was cotransfected to normalize the transfection efficiency. After 12 h of transfection, the medium was changed, and the cells were incubated at 37°C for 24 or 48 h. The cells were washed in phosphate-buffered saline (137 mM sodium chloride, 2.7 mM potassium chloride, 10 mM dibasic sodium phosphate, and 2 mM monobasic potassium phosphate) and the lysates were prepared by scraping the cells from plates in the presence of 1ϫ passive lysis buffer (Promega). Luciferase assays were performed by using Dual-Luciferase Assay System (Promega) and a Sirius luminometer (Berthold Detection System, Pforzheim, Germany).
Preparation of Nuclear Extracts-IMR-32 and HepG2 cells were plated onto 6-or 10-cm cultured dishes and incubated for 2 days. The cells were washed with 2 ml of phosphate-buffered saline and collected in 1 ml of phosphate-buffered saline. The cells were centrifuged at 2,000 ϫ g for 2 min, and the supernatant was discarded. The cell pellet was incubated in 400 l of buffer A (10 mM HEPES (pH 7.9), 1.5 mM magnesium chloride, 10 mM potassium chloride, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, 2 g/ml leupeptin, 10 g/ml aprotinin, 50 mM sodium fluoride, and 1 mM sodium orthovanadate) on ice for 10 min and then gently shaken for 10 s. The pellet of the crude nuclei was collected by centrifugation at 12,000 ϫ g for 10 s. The pellet was resuspended in 100 l of buffer C (20 mM HEPES (pH 7.9), 25% glycerol, 420 mM sodium chloride, 1.5 mM magnesium chloride, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, 2 g/ml leupeptin, 10 g/ml aprotinin, 50 mM sodium fluoride, and 1 mM sodium orthovanadate) by vortex for 15 s, and then incubated on ice for 20 min. After centrifugation at 12,000 ϫ g for 2 min, the supernatant containing the nuclear proteins was collected, quantified with BCA Protein Assay Reagent (Pierce), and stored at Ϫ70°C in aliquots. CCA AAC TTG-3Ј; and consensus E2F-1 (r), 3Ј-GTT AAA GCG CGG TTT GAA C-5Ј. 30 pmol of each of the forward and reverse oligonucleotides placed in a volume of 23 l of 1ϫ Klenow (DNA polymerase) buffer were heated at 94°C for 2 min and annealed at room temperature for 30 min. The annealed double-stranded oligonucleotides were end-labeled by a fill-in reaction using DNA polymerase (Klenow) (Promega). One unit of the DNA polymerase (Klenow) and 40 Ci of [␣-32 P]dCTP (PerkinElmer Life Sciences) were added into the annealed oligonucleotides and the mixture was incubated at 30°C for 15 min. The labeled oligonucleotides were purified by Sephadex G-50 columns (Amersham Biosciences). Cold double-stranded oligonucleotides were used as competitors. The DNA binding reaction was conducted at 4°C for 30 min in a mixture containing 3 g of nuclear extract, 10 mM Tris-Cl (pH 7.5), 50 mM sodium chloride, 0.5 mM dithiothreitol, 0.5 mM EDTA, 1 mM magnesium chloride, 4% glycerol, 0.05 g poly(dI-dC)⅐poly(dI-dC) (Amersham Biosciences) and 2 ϫ 10 4 cpm of 32 P-labeled double-stranded oligonucleotides. In supershift assays, antibodies were incubated with the reaction mixture at 4°C for 30 min before the addition of the IAP ␣-Pal/NRF-1 probes. The anti-NRF-1 goat polyclonal antiserum was kindly provided by Dr. Richard C. Scarpulla. The Sp1 and E2F antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-Myc antibody was from Invitrogen. The normal goat serum was from Vector Laboratories (Burlingame, CA). Samples were analyzed on a 4% polyacrylamide gel (acrylamide/bisacrylamide 29:1 in 0.5 ϫ Tris borate-EDTA buffer) at 10 V/cm for 2.5 h. The gel was dried and analyzed by autoradiography.
Construction and Overexpression of ␣-Pal/NRF-1-The cDNAs encoding the full-length and dominant-negative mutant of ␣-Pal/NRF-1 were constructed. The dominant-negative mutant consisted of the first N-terminal 304 residues of ␣-Pal/NRF-1, which contained the proposed DNA binding and nuclear localization domains and lacking the activation domain (20,21). Four ␣-Pal/NRF-1 cDNA fragments were obtained by RT-PCR. The primer pair 〈N-5 (5Ј-TTAAGCTT GCG CAG CCG CTC TGA GGA A) and 〈N-7 (5Ј-GACTCGAG CAC TGT TCC AAT GTC ACC AC) and primer pair 〈N-5 and 〈N-11 (5Ј-GACTCGAG TCA CTG TGA TGG TAC AAG ATG AGC) were used for the untagged full-length and dominant-negative ␣-Pal/NRF-1, respectively. Primer pair 〈N-5 and 〈N-6 (5Ј-GACTCGAG TCA CTG TTC CAA TGT CAC CA) and primer pair AN-5 and AN-10 (5Ј-GACTCGAG GTC TGT GAT GGT ACA AGA TGA G) were used for the Myc-tagged full-length and dominant-negative ␣-Pal/NRF-1, respectively. The underlined region indicates the restriction enzyme sites. These fragments were digested by HindIII and XhoI and inserted into pcDNA3.1 (Bϩ) vector (Invitrogen). HepG2 cells (8 ϫ 10 5 ) were placed onto 6-cm dishes for overexpression of ␣-Pal/ NRF-1 using the same procedure for transient transfection of the reporter plasmids. After transient transfection, the cells were incubated for 48 h and harvested for nuclear protein extraction.
Primary Cortical Culture-Primary cortical cells were prepared according to the protocol from Dichter (22). In brief, pregnant ICR mice, 15-days postconception, were anesthetized with pentobarbital, and embryos were removed. The cortices were removed and collected in MEM. The tissue was triturated three times with a fire-polished Pasteur pipette. Dissociated cells were plated onto 6-well plates coated with 0.09 mg/ml poly-L-lysine and grown in MEM supplemented with 10% fetal bovine serum (MEM 10), 50 units/ml penicillin, and streptomycin. After 24 h, plating media were replaced with MEM 10 and cells were treated with 3 M of cytosine arabinoside. Twenty-four hours later, the medium was replaced with MEM 10 again, and cells were cultured for another 48 h for the transfection experiments.
Statistics-The relative activity of different reporter constructs was compared. Statistical analysis was performed by unpaired Student's t test for pairwise comparisons. A p value Ͻ0.05 was regarded as significant.

Expression of Integrin-associated Protein Gene in Human IMR-32 and HepG2
Cells-To use IMR-32 and HepG2 cells to study the IAP gene promoter, we examined firstly the expression of IAP transcripts in these cells by RT-PCR. We used primers that can detect alternative splicing forms of IAP mRNAs. As shown in Fig. 1A, both form 1 and form 2 IAP mRNAs were expressed in these two cell lines at a similar level. The major form of IAP mRNA in IMR-32 was form 4. But form 4 is not expressed in HepG2 cells at all. Total IAP transcripts were quantified. The level of total IAP transcripts expressed in IMR-32 cells was ϳ3-fold of that expressed in HepG2 cells (Fig.  1B). These results confirmed the expression of the human IAP gene in these two cell lines.
Determination of IAP Promoter Activity in IMR-32 and HepG2 Cells-To define the boundaries of a minimal IAP promoter region and identify cis elements that regulate the expression of IAP, we generated a series of 5Ј-IAP promoter deletion constructs and transfected them into IMR-32 and HepG2 cells. All plasmid constructs were defined relative to the translation initiation codon ( Fig. 2A). The reporter constructs were cotransfected into IMR-32 and HepG2 cells with an internal control Renilla luciferase vector. The firefly luciferase activity of each reporter was normalized with the internal control to correct transfection efficiency. Results were represented as a fold-increase in activity with respect to that of the pGL3-Basic vector (Fig. 2, B and C) or the relative activity compared with that of the Ϫ232 construct (Fig. 4).
In IMR-32 cells, the shortest reporter that still retained the basal promoter activity was the Ϫ232 construct, whereas deletion for another 41 bp (the Ϫ191 construct), 73 bp (the Ϫ159 construct), or 140 bp (the Ϫ92 construct) resulted in markedly loss of reporter activity. Addition of 40 bp to the Ϫ232 construct generated the Ϫ272 construct and stimulated the reporter activity by ϳ25%. In the construct containing additional 184 bp (the Ϫ456 construct), the promoter activity was not increased. Interestingly, the reporter activity decreased to about the basal proximal promoter activity in the construct containing additional 274 bp (the Ϫ730 construct) and in the longest construct (the Ϫ1554 construct) (Fig. 2B). A similar pattern of promoter activity was observed when the reporter constructs were transfected into HepG2 cells, but with three exceptions (Fig. 2C). First, overall promoter strength relative to pGL3-Basic was much lower in HepG2 cells, about one-sixth to one-fourth compared with IMR-32 cells. Second, the constructs of Ϫ272 and Ϫ456 produced maximal activity in IMR-32 cells; in HepG2 cells, however, the construct of Ϫ730 produced maximal activity. Third, there were negative regulators located between Ϫ457 and Ϫ730 in IMR-32 cells; in HepG2 cells, however, there were negative regulators located between Ϫ730 and Ϫ1554. Results suggested a core promoter of the human IAP gene located between Ϫ232 and Ϫ12 upstream of the translation initiation codon in both IMR-32 and HepG2 cells.
Identification of cis-Elements in the Core Promoter of the IAP Gene-To determine more precisely the core promoter of the IAP gene, we generated several shorter or point mutation constructs. When 14 and 23 bp were deleted from the 5Ј-end of the Ϫ232 construct to generate the constructs of Ϫ218 and Ϫ209, respectively, the promoter activity of these two constructs was identical to that of the Ϫ232 construct in IMR-32 cells. This indicated that the sequence between Ϫ232 and Ϫ209 played no significant role in IAP gene expression under the current experimental conditions. When 34 bp were deleted from the 5Јend of the Ϫ232 construct to generate the Ϫ198 construct; however, the promoter activity was markedly reduced by 90% in IMR-32 cells, indicating that the sequence between Ϫ209 and Ϫ198 might be required for the IAP promoter activity (Fig.  4). As shown in Fig. 3, the region from Ϫ232 to Ϫ198 consists of a GC-rich sequence that includes the putative Sp1 and ␣-Pal/NRF-1 sites. Point mutations were introduced into these sites in IMR-32 cells to determine whether these sites were necessary for IAP promoter activity. When four bases of the

␣-Pal/NRF-1 Regulates the Human IAP Promoter
Sp1 site were substituted with four T residues to generate the Ϫ232m1 construct, no significant effect on the IAP promoter activity was observed (Fig. 4), indicating that Sp1 site in this region was not required for the IAP promoter activity under this condition. The putative ␣-Pal/NRF-1 site in the IAP pro-moter is a 12-base tandem-repeat sequence, TGCGCGT-GCGCG. When two bases in each of the repeat sequence were replaced by two T residues to generate the Ϫ232m2 and Ϫ232m3 constructs respectively, an 80% and a 90% drop in promoter activity was observed (Fig. 4). These results suggested that the consensus ␣-Pal/NRF-1 sequence, but not the consensus Sp1 sequence, is a functional regulatory element in the IAP promoter in IMR-32 cells.
␣-Pal/NRF-1 Is a Transcription Factor Regulating the IAP Promoter Activity-To demonstrate that the consensus ␣-Pal/ NRF-1 sequence was functional, i.e. that there were endogenous nuclear proteins binding to this region, we performed the EMSA experiment. Nuclear extracts from IMR-32 cells were combined with 32 P-fill-in-labeled double-stranded oligonucleotides in vitro. A major band of DNA-protein complex was found in all lanes when the nuclear extracts were incubated with the wild-type IAP ␣-Pal/NRF-1 probes (Fig. 5A, lane 3 and lanes 5-12), but not with the mutant IAP ␣-Pal/NRF-1 probes (Fig.  5A, lane 4). No band was found when nuclear extracts were not added into the probes (Fig. 5A, lanes 1 and 2). Competition analysis using a 10-or 60-fold molar excess of unlabeled probes was used to characterize the factor, which specifically binds to this sequence. As expected, the addition of a 10-or 60-fold molar excess of published wild-type consensus ␣-Pal/NRF-1 element reduced the intensity of these complexes (Fig. 5A,  lanes 5 and 6), whereas the addition of the mutant consensus ␣-Pal/NRF-1 sequence did not (Fig. 5A, lanes 7 and 8). Results suggested that ␣-Pal/NRF-1 proteins might bind to the IAP ␣-Pal/NRF-1 element. However, the ␣-Pal/NRF-1 site is GCrich and might therefore interact with factors other than ␣-Pal/ NRF-1, such as Sp1 and E2F. We therefore used the unlabeled consensus Sp1 and E2F sequence for the competition experiment. The intensity of the migrating bands was not significantly reduced (Fig. 5A, lanes 9 -12). Supershift assays using the anti-␣-Pal/NRF-1 antiserum were used to further confirm the binding of the ␣-Pal/NRF-1 on its DNA element. The migrating bands were weakened when increasing amounts of ␣-Pal/NRF-1 antiserum were added and supershifted bands appeared (Fig. 5B, lanes 3-5). However, the Sp1 or E2F antibody did not generate any supershifted band (Fig. 5B, lanes 6  and 7), neither did the normal goat serum (Fig. 5B, lane 8). The EMSA experiments also revealed that the DNA binding activity of ␣-Pal/NRF-1 in IMR-32 cells (Fig. 5C, lanes 1-3) was much higher than that in HepG2 cells (Fig. 5C, lanes 4 -6). The oligonucleotide probes and competitors used in the EMSA ex- The sequence spanning from Ϫ272 to the translation initiation codon ATG is shown. This region contains putative binding sites for transcription factors such as AP-2, Maz, CREB, Sp1, E2F, and ␣-Pal/ NRF-1. However, two Sp1 (TGCGC) and one E2F sites share partial sequences with the ␣-Pal/NRF-1 site (23). The positions of these sites are indicated by underlines. The plasmids for the luciferase assay were constructed by inserting fragments spanning from nucleotides (indicated by arrows and numbers) to nt Ϫ12, relative to ATG, into the firefly luciferase expression vector (pGL3-Basic).
FIG. 4. Identification of functional cis elements in the IAP core promoter. Shorter promoter fragments were made by PCR (the Ϫ218, Ϫ209, and Ϫ198 constructs). Mutant constructs were made by sitedirected mutagenesis (the Ϫ232m1, m2, and m3 constructs). The putative Sp1 and ␣-Pal/NRF-1 sites are indicated. Independent constructs and pRL-TK plasmids were cotransfected into IMR-32 cells using the same protocol as shown in Fig. 2. The promoter activity was expressed with respect to the Ϫ232 construct. *, p Ͻ 0.05; ***, p Ͻ 0.001; unpaired Student's t test.
To further confirm that ␣-Pal/NRF-1 is the major transcription factor that binds to the IAP ␣-Pal/NRF-1 element, plasmids encoding the full-length or dominant-negative mutant of ␣-Pal/NRF-1 with or without a Myc tag were transiently transfected into HepG2 cells. The addition of a Myc tag in the C-terminal of ␣-Pal/NRF-1 is useful for the supershift assay in the EMSA experiments. In HepG2 cells, overexpression of the full-length ␣-Pal/NRF-1 (Fig. 6A, lanes 4 -6) and Myc-tagged ␣-Pal/NRF-1 (Fig. 6B, lanes 3-5) enhanced the binding of DNAprotein complex in a dose-dependent manner as compared with the mock controls (Fig. 6A, lane 3 and 6B, lane 2). Overexpression of the dominant-negative mutant of ␣-Pal/NRF-1, which contains only the N-terminal DNA binding domain, did not affect endogenous DNA-protein binding in HepG2 cells but generated an additional band of DNA-protein complex with a smaller molecular weight. The DNA binding activity was strongly enhanced by the overexpression of the dominant-negative mutant (Fig. 6, A and B, lanes 7-9). This discrepancy may be related to the higher transfection efficiency of the plasmid ␣-Pal/NRF-1 Regulates the Human IAP Promoter that contained the dominant-negative mutant. In supershift assays, the band of DNA-protein complex containing the ␣-Pal/ NRF-1-myc fusion protein was supershifted when monoclonal anti-Myc antibody was used (Fig. 6B, lane 6). The band of DNA-protein complex containing the dominant-negative mutant of ␣-Pal/NRF-1 fused with the Myc protein fragment was also completely supershifted by the monoclonal anti-Myc antibody (Fig. 6B, lane 10). These data strongly suggested that ␣-Pal/NRF-1 binds to this region (Ϫ204 to Ϫ193) of the human IAP gene promoter in vitro.
If the ␣-Pal/NRF-1 protein binds to the ␣-Pal/NRF-1 element in the IAP promoter in vitro, it should functionally regulate the promoter activity of the IAP gene in vivo. To test this possibility, the full-length or dominant-negative ␣-Pal/NRF-1 constructs were cotransfected with the reporter construct Ϫ232 into IMR-32 and HepG2 cells, and the luciferase activity was measured. As shown in Fig. 7A, overexpression of dominantnegative ␣-Pal/NRF-1 significantly reduced the IAP promoter activity in a dose-dependent manner in both cell lines, as compared with the mock controls. At the highest dose (50 ng) the promoter activity of the reporter construct was reduced markedly up to 50 and 92% in IMR-32 and HepG2 cells, respectively. In contrast, overexpression of full-length ␣-Pal/NRF-1 significantly increased the IAP promoter activity in a dose-dependent manner (Fig. 7B). At highest dose (2 g) the promoter activity of the reporter construct increased up to 3.9-and 5.3-fold at the highest dose (2 g) in IMR-32 and HepG2 cells, respectively. In Fig. 7B, the relative activity was represented as the percentage of firefly luciferase activity of mock control but not normalized with the Renilla luciferase activity, because the full-length ␣-Pal/NRF-1 could enhance the activity of the Renilla luciferase by unknown effects. These results confirm that ␣-Pal/ NRF-1 interacts with IAP promoter and regulates its downstream gene expression in vivo.
Regulation of the IAP Promoter Activity by ␣-Pal/NRF-1 in Primary Cells-To investigate if ␣-Pal/NRF-1 regulates IAP gene promoter under physiological conditions, we transfected the truncated or mutant human IAP promoter constructs into the mouse primary cortical cells. Results are shown in Fig. 8. Similar to the results observed in IMR-32 and HepG2 cells (Fig.  2), the Ϫ232 construct retained the basal promoter activity, but the activity of shorter constructs, Ϫ198 and Ϫ195 constructs, was reduced 76 and 62%, respectively. The promoter activity of point-mutation constructs, Ϫ232m2 and m3, was significantly decreased about 40%, but not the Ϫ232m1 construct (Fig. 8A). Overexpression of dominant-negative ␣-Pal/NRF-1 in the primary cortical cells decreased the IAP gene promoter activity in a dose-dependent manner (Fig. 8B). Overexpression the wildtype ␣-Pal/NRF-1 in the primary cortical cells significantly enhanced the IAP gene promoter activity (Fig. 8C). These results indicate that the function of ␣-Pal/NRF-1 in primary cells is similar to that in cell lines. DISCUSSION We have identified in the core promoter of the human IAP gene a cis-acting element that contributes to the control of IAP gene expression. This element is a 12-base pair direct-repeat sequence, TGCGCGTGCGCG, located from Ϫ204 to Ϫ193 nu- cleotides upstream of the IAP translation initiation codon, which is a consensus sequence for the transcription factor ␣-Pal/NRF-1. ␣-Pal/NRF-1 belongs to a new class of transcription factors that contain a unique putative basic leucine zipper (bZip) DNA binding domain (24). The consensus sequence recognized by ␣-Pal/NRF-1 is (T/C)GCGCA(T/C)GCGC(A/G) (20,21,24). The reverse complement sequence of the ␣-Pal/NRF-1 site in the human IAP promoter is CGCGCACGCGCA, conforming well to the consensus sequence. This DNA element is critical to the expression of the IAP gene not only in the human neuroblastoma and hepatoma cell lines but also in the mouse primary cortical cells.
␣-Pal was first discovered as a key transcription factor in the eukaryotic initiation factor 2␣ gene, which is a target of the post-translation mechanisms when eukaryotic cells respond to growth, metabolic, and other signals (24,25). It was designated because it binds to a palindromic sequence TGCGCATGCGCA (25). NRF-1 was discovered independently as a nuclear transcription factor that is important for the regulation of mitochondrial genes responsible for modulation of energy transduction (26). Later studies found that the ␣-Pal/NRF-1 recognition sequence can be identified in many other genes involved in energy transduction, translation/protein turnover, DNA synthesis/repair, and cellular proliferation (24,27). The ␣-Pal/ NRF-1 shows strong homology with two invertebrate genes, sea urchin P3A2 and Drosophila erect wing gene (ewg). Sequence comparison revealed that the half N-terminal region has been evolutionarily conserved. This region harbors the DNA binding, dimerization, and nuclear localization signal domains. The C-terminal halves comprise the bipartite hydrophobic activation domain and have only conserved short patches of homology between the vertebrate and invertebrate members of the family (20,24). These two genes have been implicated in embryonic or larval development (28,29). On the other hand, ␣-Pal/NRF-1 shows 91% identity to its homologue in zebrafish, not really finished (nrf) (30). Both ewg and nrf have After 48-h incubation, the firefly luciferase activity was measured and normalized to the Renilla luciferase activity. The relative activity was represented as percentages of the activity of pGL3-232. All groups were compared with the pGL3-232 by unpaired Student's t test. B, overexpression of dominant-negative ␣-Pal/NRF-1 decreased the promoter activity in primary mouse cortical cells. Various amounts (1-50 ng) of the dominant-negative ␣-Pal/NRF-1 construct or 50 ng of the empty vector were transiently transfected with 50 ng of pGL3-232 and 10 ng of pRL-TK into cells. After transfection, primary cortical cells were cultured for 48 h. The firefly luciferase activity was measured and normalized with Renilla luciferase activity. The relative activity was represented as percentages of the mock control. C, overexpression of full-length ␣-Pal/NRF-1 increased the promoter activity in primary mouse cortical cells. Various amounts of the full-length ␣-Pal/NRF-1 construct or 2 g of the empty vector were transiently transfected with 50 ng of pGL3-232 into cells and cells were cultured for 48 h. The relative activity was plotted as percentages of the firefly luciferase activity of the mock control. In B and C, all groups were compared with mock control. Values are presented as the mean Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; unpaired Student's t test.
been associated with the development of the central nervous system (29,30).
In the nervous system, we originally found that the IAP/ CD47 mRNA level is up-regulated in the rat hippocampus during the processes of memory formation (3). As a critical positive regulator of IAP/CD47 gene, the binding activity of ␣-Pal/NRF-1 may be increased after learning. Experiments are underway to determine the level and DNA binding activity of ␣-Pal/NRF-1 in the rat brain during the processes of memory formation. Several important neuronal genes, such as gene encoding GluR2, FMR-1, or synapsin I, also has the ␣-Pal/ NRF-1 element in their promoter region (20,31,32). GluR2 is one of the subunits of the AMPA subtype of glutamate receptors that mediate a large fraction of the postsynaptic response at most fast excitatory synapses in the brain. FMR-1 is the gene that is involved in the fragile X mental retardation syndrome. Synapsin I is one of the most abundant proteins in the presynaptic terminal and is involved in memory function (33,34). The involvement of ␣-Pal/NRF-1 in memory functions remains to be determined.
Similar to the ubiquitous expression of IAP across tissues, ␣-Pal/NRF-1 is also expressed in a variety of tissues (35). One will wonder how IAP would be regulated from such a ubiquitously expressed transcription factor. Our data suggested that the DNA binding activity instead of the expression level of ␣-Pal/NRF-1 is the determinant of the expression of the IAP gene. In IMR-32 cells, the total IAP mRNA level is about 3-fold of that in HepG2 cells (Fig. 1B). The DNA binding activity of ␣-Pal/NRF-1 in IMR-32 cells is also about several folds of that in HepG2 cells (Fig. 5C). The higher IAP mRNA level in the neuroblastoma cells is correlated with the higher DNA binding activity of ␣-Pal/NRF-1 in these cells. However, mutation or truncation of the ␣-Pal/NRF-1 binding site not only reduced the promoter activity of IAP gene in neuroblastoma cells but also in hepatoma cells. This suggests that ␣-Pal/NRF-1 is not a transcription factor that specifically regulates the expression of IAP mRNA in neuronal cells. The question will then be what is the function of ␣-Pal/NRF-1 in neurons? We recently found that overexpression of ␣-Pal/NRF-1 in the neuroblastoma cells significantly enhances neurite outgrowth. 2 It seems that the regulation of IAP gene by ␣-Pal/NRF-1 plays a role in neuronal differentiation.
The other question is that if the preferential expression of form 4 IAP in IMR-32 cells or the nervous systems is correlated with the activity of ␣-Pal/NRF-1. The ␣-Pal/NRF-1 recognition site is located in the core promoter of the IAP gene. Our results suggested that ␣-Pal/NRF-1 is responsible for not only the expression of form 4 IAP in IMR-32 cells, but also the expression of form 1 and form 2 IAP in HepG2 cells. Form 4 as well as form 1 and form 2 IAP mRNA are alternative splicing forms of the IAP gene. Other activators or repressors for alternative splicing will be the determinants for the preferential expression of form 4 IAP in the nervous system. We compared the core promoter sequence of the human IAP gene with the core promoters of mouse and rat IAP genes and found that the sequence of the ␣-Pal/NRF-1 site is identical in all three species (Fig. 9). This suggested that this site is evolutionarily important for the control of IAP gene expression. Upstream of the ␣-Pal/NRF-1 site, there are several other conserved regions in these three species. The first is the Sp1 site next to the ␣-Pal/NRF-1 site. The three sequences here are also identical. The core promoter sequence of the human IAP gene is TATA-less. According to previous reports, Sp1 plays an important role in TATA-less promoters (23,36). Although mutation or truncation of this Sp1 site did not affect the activity of human IAP core promoter (Figs. 4 and 8), we did find in EMSA experiments that Sp1 factor binds to this site (data not shown). This implied that Sp1 has the potential to regulate the IAP gene under some conditions or in other cells. The next is the region from Ϫ236 to Ϫ225 of the human IAP promoter. The human sequence, with only one base missing, is almost identical to those of the mouse and rat. This sequence contains a putative CREB site. Whether CREB binds to this site remains to be established. The third is the region from Ϫ245 to Ϫ272. There is 67% homology between human and mouse but 94% homology between mouse and rat. This region contains a putative binding site for AP-2 and Maz. But the evidence that AP-2 or Maz will bind to this region is still not available yet. The promoter activity of the Ϫ272 construct, which contains this region, is about 25% higher than that of the Ϫ232 construct only in IMR-32 cells but not in HepG2 cells. This suggested that this region contains cell-specific positive regulators for the IAP gene expressed in IMR-32 cells. The reporter constructs also revealed that there are positive regulators between Ϫ272 and Ϫ456 specific for HepG2 cells, because the promoter activity of the Ϫ456 construct is significantly higher than that of the Ϫ272 construct in this cell line. On the other hand, there are cell-specific negative regulators in the human IAP gene promoter. In IMR-32 cells, there are cell-specific negative reg-ulators located between Ϫ272 and Ϫ456; in HepG2 cells, however, there are cell-specific negative regulators located between Ϫ730 and Ϫ1554. What the negative and positive regulators in the IAP gene promoter are and whether they interact with ␣-Pal/NRF-1 to control the expression of the IAP gene remain to be studied.
We conclude that the core promoter of the human IAP gene is located between Ϫ232 and Ϫ12 nucleotides upstream of the translation initiation codon, and that ␣-Pal/NRF-1 is a critical transcription factor in the regulation of human IAP gene expression both in cell lines and primary cells. The DNA binding activity of ␣-Pal/NRF-1 in different cells is an important determinant of the promoter activity of the IAP gene. Understanding the upstream signaling cascades that affect the activation of ␣-Pal/NRF-1 will help to clarify the biological functions of the IAP gene.