Transcriptional Regulation of the p67 Gene ROLE OF AP-1 IN CONCERT WITH MYELOID-SPECIFIC TRANSCRIPTION FACTORS*

We have investigated the myeloid-specific transcriptional regulation of p67, an essential component of phagocyte respiratory burst NADPH oxidase. Analysis was carried out on the p67 5 -flanking region from –3669 to –4 (relative to ATG), including the first exon and intron and part of the second exon. The construct extending from –985 to –4 produced the highest luciferase activity in myeloid HL-60 cells but was not active in HeLa or Jurkat cells, indicating myeloid-specific expression. Four active elements were identified: Sp1/Sp3 at –694, PU.1 at –289, AP-1 at –210, and PU.1/HAF1 at –182, the latter three being in the first intron. These cis elements bound their cognate transacting factors both in vitro and in vivo. Mutation of the Sp1, PU.1, or PU.1/ HAF1 site each decreased promoter activity by 35–50%. Mutations in all three sites reduced promoter activity by 90%. However, mutation of the AP-1 site alone nearly abolished promoter activity. The AP-1 site bound Jun and Fos proteins from HL-60 cell nuclear extract. Coexpression with Jun B in AP-1-deficient cells increased promoter activity by 3-fold. These data show that full p67 promoter activity requires cooperation between myeloid-specific and nonmyeloid transcription factors, with AP-1 being the most critical for function.

The NADPH oxidase of phagocytic cells is a key component of the host antimicrobial defense system. The oxidase is composed of the membrane-associated heterodimeric flavocytochrome b 558 (gp91 phox and p22 phox ) and the cytosolic factors p47 phox , p67 phox , p40 phox , and Rac1 or Rac2 (1,2). Following stimulation of resting phagocytes, the cytosolic factors translocate to the membrane to form a complex with the flavocytochrome, which generates superoxide by the transfer of electrons from NADPH to molecular oxygen. The reactive oxygen species derived from superoxide are mediators of microbial death, but they can also cause injury to host tissues during the inflammatory process.
Expression of the gp91 phox , p67 phox , p47 phox , and p40 phox components of NADPH oxidase shows tissue selectivity for cells of myeloid lineage, such as neutrophils, eosinophils, and monocyte/macrophages (3)(4)(5). In contrast, the Rac proteins and p22 phox show a much broader distribution (6). Cellular models of granulocyte differentiation indicate that gp91 phox , p67 phox , and p47 phox are not expressed in immature myeloid precursor cells but are induced during the differentiation/maturation process (4,(7)(8)(9)(10). The p40 phox protein, on the other hand, can be detected in undifferentiated myeloid cell lines, and its expression is increased during Me 2 SO-induced granulocyte differentiation (10 -12). Thus, the phox proteins are coordinately upregulated during terminal differentiation and maturation, suggesting shared mechanisms in the control of their expression. However, variations in the kinetics of induction indicate that regulatory factors unique to each gene may also be important.
Major contributions to the regulation of expression of the components of the phagocyte NADPH oxidase occur at the level of transcription. Studies of the gp91 phox promoter have demonstrated a complex regulatory mechanism involving both positive and negative transcription factors (13)(14)(15)(16). The upstream promoter contains a CCAAT-box motif that binds the transcription factor CP1, ISRE sequences that can bind IRF-1/2, and binding sites for a novel transacting factor termed BID. It is thought that in undifferentiated myeloid precursor cells, the positive promoter effect of these elements is masked by CDP, a CCAAT-displacement protein that binds the regions around these motifs, thereby suppressing transcription (13,15). Down-regulation of the repressor protein during terminal differentiation is required for the expression of gp91 phox . A critical regulatory element in the transactivation of gp91 phox is a consensus sequence for the ets family of transcription factors located in the proximal region (bp 1 -57 to -50) of the promoter. This sequence binds the myeloidspecific factor PU.1 and the hematopoietic-associated factor complex HAF-1 (16 -19). HAF-1 contains the interferon response factors IRF-1 and ICSBP and the ets family members Elf-1 or PU.1. The importance of this ets binding site is evidenced by the detection of single base pair mutations in this sequence in some kindreds with chronic granulomatous disease (17)(18)(19), a genetic disorder characterized by a nonfunctional phagocyte NADPH oxidase, impaired microbidical function, and recurrent severe infections.
Our recent studies indicate that ets factor cis elements also play critical roles in the transcriptional regulation of both the p47 phox and p40 phox genes. In p47 phox , a single consensus PU.1 binding site located on the noncoding strand from bp -40 to -45 (relative to the transcription start site) accounted for most of the myeloid-specific promoter activity found in the proximal 2151 base pairs of the 5Ј-flanking region of the gene (20). Mutation of this site essentially abolished p47 phox promoter/ reporter gene activity in myeloid cells. Furthermore, we showed that this site bound PU.1 with high avidity, and mutations of flanking sequences that reduced the avidity of PU.1 binding also resulted in a proportionate reduction in promoter activity (21). In p40 phox , we analyzed ϳ6000 bp of the gene and identified three PU.1 binding sites that directed gene transcription in a myeloid-specific manner. 2 Two sites are located in the proximal 120 bp of the 5Ј-flanking sequence of the gene, and the third is in the putative 5Ј untranslated region of the transcript. Mutation analysis indicated that each site contributed to a different extent to the overall promoter activity, and mutation of all three sites abolished promoter activity in myeloid cells. In accord with our mutation studies on the p47 phox PU.1 site, the promoter activity mediated by each p40 phox PU.1 site correlated with its binding avidity for PU.1 (21).
The ets factor PU.1 and the ets factor-containing transcriptional complex HAF-1 thus play critical roles in the transcriptional control of the gp91 phox , p47 phox , and p40 phox components of the phagocyte NADPH oxidase. In the current report, we describe our studies on the transacting factors and cis elements that regulate p67 phox transcription in myeloid cells. Although PU.1 sites were identified in the upstream 5Ј-flanking region of the p67 phox gene, we found that most of the functional activity in myeloid cells is directed by elements in the first intron, including binding sites for PU.1, AP-1, and, as recently described (19), PU.1/HAF-1. Unique among the phox genes, the promoter activity mediated by these sites and by a functional upstream Sp1 site was demonstrated to be entirely dependent on an intact AP-1 binding site. Mutation of this site abolished promoter activity in the myeloid cell line HL-60, even in the presence of intact PU.1 and HAF-1 binding sites.

EXPERIMENTAL PROCEDURES
Materials-RPMI 1640 medium was obtained from Life Technologies, Inc. Restriction enzymes, T4 polynucleotide kinase, and pGL3-Basic luciferase vector and the dual luciferase assay kit were from Promega (Madison, WI). [␥-32 P]ATP, 6000 Ci/mmol, was obtained from PerkinElmer Life Sciences. The TOPO-TA cloning kit (containing the pCRII vector) for cloning products of the polymerase chain reaction (PCR) was obtained from Invitrogen (San Diego, CA). Oligonucleotide synthesis and DNA sequencing were carried out by the Advanced DNA Technology Unit, University of Texas Health Science Center. Antibodies to transcription factors were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
Cloning and Sequencing of Human p67 phox Promoter-The p67 phox 5Ј-flanking regions were cloned using the PromoterFinder kit (CLON-TECH, Palo Alto, CA) according to the manufacturer's protocol. PCR was performed using the human genomic libraries provided as templates to amplify the desired sequences. The forward primer was complementary to the adaptor ligated to the genomic DNA fragments in each library. The reverse primer (5Ј-GAAACCCCACCTTCAAGCC-3Ј) corresponded to bp -501 to -520 relative to the translation initiation codon of the p67 phox gene. The amplified products were analyzed by agarose gel electrophoresis and then subjected to a second round of PCR using nested primers. The reverse primer (5Ј-GGCTGAAAGACTAGGC-TACTGGTCC-3Ј) was derived from bp -534 to -558 of the p67 phox gene. The final PCR products were cloned directly into the pCRII vector, and their identities were confirmed by sequencing both strands.
Luciferase Vector Construction and Mutagenesis-Two sets of reporter vectors, differing primarily in the location of their 3Ј ends (bp -534 versus -4), were constructed in the pGL3-Basic luciferase vector. The promoter regions were amplified using human genomic DNA as templates. For the first set of reporter constructs, the reverse primer was similar to the nested reverse primer indicated above, but with an added XhoI site (5Ј-CACTCGAGGGCTGAAAGACTAGGCTAC-3Ј). For the second set, the reverse primer (5Ј-CACTCGAGTAGGTAGAAAC-TAGGACC-3Ј) was derived from bp -4 to -21 upstream of the translation initiation code. The forward primers corresponded to the upstream sequences of desired promoter regions with an added 5Ј-flanking KpnI site. PCR products were digested with KpnI and XhoI and cloned into the promoterless luciferase reporter plasmid pGL3-Basic at the same restriction sites. Site-directed mutagenesis was carried out using the QuikChange kit (Stratagene, La Jolla, CA). The mutated nucleotides are shown in italics in Fig. 1. All constructs were confirmed by restriction mapping and sequencing.
Cell Culture-The human promyelocytic cell line HL-60 and the myeloid leukemia cell line PLB-985 were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum and 10 mM HEPES. The human cervical carcinoma epithelial cell line HeLa, the lymphoid leukemia cell line Jurkat, and the human colon epithelial cell lines LoVo and HCT-116 were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. All media contained penicillin and streptomycin.
Transient Transfections-For suspension cultures, transfection was carried out by electroporation when cells reached a density of about 5 ϫ 10 5 cells/ml. Cells were resuspended in medium containing 20 g of the luciferase reporter constructs and 0.5 g of a Renilla luciferase vector (pRL-null, Promega) as a transfection efficiency control. Electroporation was carried out at 960 F and 250 V. Unless otherwise indicated, at 8 h, the cells were washed twice in phosphate-buffered saline, pH 7.4, lysed in 100 l of 1ϫ reporter lysis buffer (Promega), and centrifuged at 12,000 rpm at ambient temperature, and 20-l aliquots of the supernatants were tested in the dual luciferase assay system (Promega) using a Turner Designs TD-20/20 luminometer. For adherent cells, transfections were done using LipofectAMINE Plus reagent (Life Technologies, Inc.) according to the manufacturer's protocol.
Nuclear Extracts-HL-60 cells were disrupted by cavitation using a technique described previously for polymorphonuclear leukocytes (23). The cells were washed twice in phosphate-buffered saline, resuspended in 10 ml of cold relaxation buffer (100 mM KCl, 3 mM NaCl, 3.5 mM MgCl 2 , 10 mM PIPES, pH 7.3), and 3.5 l diisopropyl fluorophosphate (Sigma) added. The cells were kept on ice for 10 min and then centrifuged at 400 ϫ g for 5 min. The cell pellet was resuspended in 10 ml of relaxation buffer, pressurized in N 2 at 350 p.s.i. for 20 min in a nitrogen bomb (Parr Instrument Co., Moline, IL), and released into 750 l of a solution containing 20 mM EGTA, 100 mM MgCl 2 , 20 mM dithiothreitol, 4 mM phenylmethylsulfonyl fluoride, and 2 mM sodium orthovanadate. The cavitated cells were centrifuged at 400 ϫ g for 10 min at 4°C, and the nucleus-enriched pellet was resuspended and further purified on a discontinuous gradient of sucrose (0.3/0.88 M). The nuclear fraction was extracted in ϳ100 l of urea extraction buffer (1.1 mM urea, 1% Nonidet P-40, 5% glycerol, 0.5 mM MgCl 2 , 5 mM KCl, 0.05 mM EDTA, 5 mM HEPES, pH 7.9) and microcentrifuged. The supernatant was collected and stored in aliquots at Ϫ70°C. The protein concentration was determined using the Bradford reagent (Bio-Rad, Hercules, CA).
DNase I Protection Assay-A fragment of DNA corresponding to the p67 phox promoter region between -731 and -635 was labeled on the noncoding strand by successive T4 polynucleotide kinase and polymerase chain reactions. Briefly, 10 fmol of the reverse-complemented oligonucleotide (5Ј-GCCTTTTAACCAATTGAATG-3Ј) was incubated at 37°C for 30 min in a 10-l volume containing 30 Ci of [␥-32 P]ATP, 1ϫ T4 polynucleotide kinase buffer, and 5 units of T4 polynucleotide kinase (Promega). The kinase was inactivated by heating at 65°C for 20 min, and the reaction was adjusted to 1ϫ PCR buffer; 2 mM MgCl 2 ; 125 M each dATP, dCTP, dGTP, and dTTP; 10 fmol of the forward primer 5Ј-gagctcAGTGCGGGTGAGTCTG-3Ј; 50 ng of the template DNA (pGL3-p67-330); and 1 unit of Taq polymerase (Promega) in a final volume of 50 l. Thirty cycles of PCR were carried out (94°C for 30 s, 54°C for 30 s, and 72°C for 30 s), and the ϳ100-bp product was purified by agarose gel electrophoresis. The activity of the DNA preparation was ϳ10 5 cpm/l. Nuclear extracts were prepared from HL-60 cells as noted above.
Electrophoretic Mobility Shift Assay (EMSA)-Complementary DNA oligonucleotides were annealed by heating in 1ϫ NET at 95°C for 5 min and cooling at ambient temperature. Probes were then labeled with [␥-32 P]ATP and T4 polynucleotide kinase. For gel shift assays, nuclear extract (6 g) was incubated for 20 min at ambient temperature with 5 ϫ 10 4 cpm of the labeled DNA probe in 20 l of binding buffer containing 10 mM Tris-HCl, pH 7.6, 50 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 5% glycerol, 1 g/l bovine serum albumin, and 2 g of poly(dI-dC). For supershift assays, 2 g of specific antibody were added, and the reaction was continued for 30 min. Samples were loaded on 5% nondenaturing polyacrylamide gels, and electrophoresis was carried out at 4°C at 200 V in 25 mM Tris, pH 8.5, with 190 mM glycerol and 1 mM EDTA. Competition assays were carried out in the same manner, except that the reaction mixture was preincubated with competitor DNA for 10 min at 4°C before addition of the labeled probe.
Chromatin Immunoprecipitation (ChIP) Assay-Formaldehyde cross-linking and immunoprecipitation of chromatin were carried out as described by Farnham and co-workers (24). Briefly, formaldehyde was added directly to cell cultures to a final concentration of 1%, the cells were incubated at 23°C for 10 min, and then glycine was added to stop the fixation. Cells were collected by centrifugation and washed with phosphate-buffered saline containing 0.5 mM phenylmethylsulfonyl fluoride. Cells were allowed to swell in 5 mM PIPES (pH 8.0), 85 mM KCl, 0.5% Nonidet P-40, 0.5 mM phenylmethylsulfonyl fluoride, and 100 ng/ml leupeptin and aprotinin and incubated on ice for 10 min. Nuclei were collected, resuspended in 50 mM Tris⅐HCl (pH 8.1), 1% SDS, 10 mM EDTA, plus phenylmethylsulfonyl fluoride, leupeptin, and aprotinin and then incubated on ice for 10 min. Samples were sonicated on ice to break chromatin DNA to an average length of ϳ600 bp and then precleared with formalin-fixed Staph A cells (Roche Molecular Biochemicals). Precleared chromatin from 2 ϫ 10 7 cells was incubated alone or with 1.5 g of affinity-purified rabbit polyclonal antibody (Santa Cruz Biotechnology) and rotated at 4°C for 12-16 h. Immunoprecipitation was carried out with Staph A cells. The supernatant from the reaction lacking primary antibody was saved as total input of chromatin and was processed in the same way as the eluted immunoprecipitates, beginning at the cross-link reversal step. Cross-links were reversed by incubation at 65°C for 5 h in the presence of 300 mM NaCl. During this process, RNA was digested by 0.01% RNase A. Samples were then precipitated, resusupended, and treated with proteinase K followed by extraction with phenol/chloroform/isoamyl alcohol and precipitation with sodium acetate and ethanol plus tRNA and glycogen as carriers. Pellets were collected by microcentrifugation, resuspended in 30 l of H 2 O, and analyzed by PCR. Total input samples were resuspended in 100 l of H 2 O and then diluted 1:100. PCRs were carried out in a total volume of 25 l containing 2-3 l of immunoprecipitate or diluted total input and 0.4 M of each primer. After 35 cycles of amplification, PCR products were separated by electrophoresis on a 1.5% agarose gel. The primer pairs used for p67 phox were as follows: forward, 5Ј-GAACTGATCCTTCAGGCAG-3Ј, reverse, 5Ј-TTCATTCCAGAGGC-TGATG-3Ј for AP-1, PU.1(1)/HAF1, and PU.1(2); forward, 5Ј-CAGAA-GGCGAGATCATGG-3Ј, reverse, 5Ј-CCCTGACACTCAGAACTC-3Ј for PU.1(3); forward, 5Ј-TAGACTGCCCCTGCATTG-3Ј, reverse, 5Ј-CCTT-CCATGGCTTTGCAC-3Ј for PU.1(4); and forward, 5Ј-TTCAGTGGAG-AGGGGATG-3Ј, reverse, 5Ј-CTAGGCAGATAGGGGCAG-3Ј for Sp1. The primers used for controls were as follows: forward, 5Ј-GCGCCAA-GGACTGACATC-3Ј, reverse, 5Ј-GAGCAGGTGGTGCGTCTC-3Ј for p40-PU.1; and forward, 5Ј-GCGCCAAGGACTGACATC-3Ј, reverse, 5Ј-GAGCAGGTGGTGCGTCTC-3Ј for CCR5.

RESULTS
A functional Sp1 Site Is Present in the Proximal 5Ј-Flanking Sequence of the Human p67 phox Gene-The promoter region of p67 phox was cloned using the PromoterFinder kit (CLON-TECH). The specific PCR product amplified from the genomic library appeared as a single band (ϳ1.5 kilobases) on agarose gel. This amplicon was cloned directly into the pCRII vector (Invitrogen), and its identity was confirmed by the overlapping sequence between the 3Ј end of the clone and the known 5Ј end of p67 phox cDNA (25,26). A clone was sequenced and later found to be identical to the sequence of the p67 phox gene deposited in the GenBank TM (accession number NM 005003 (51550 -533640)), except for a G to A change at bp -912 and the deletion of 3 of 18 tandem A residues between bp -1171 and -1169 (nucleotide positions relative to the ATG translation initiation codon). The sequenced clone was subcloned into the luciferase reporter vector pGL3-Basic. Based on this plasmid, a series of deletion constructs was made, all of which extended downstream to nucleotide -534 in exon 1 of the p67 phox gene (Fig. 1). The activity of these constructs was assayed in the HL-60 human myeloid cell line by transient transfection. As shown in Fig. 2, a ϳ30-fold increase in luciferase reporter gene activity was observed in lysates prepared from cells transfected with pGL3-p67-733 and pGL3-p67-862, compared with pGL3-Basic vector. Constructs extending further upstream gave progressively less luciferase activity, suggesting the presence of negative regulatory elements between bp -862 and -2022. However, a deletion to nucleotide -682 resulted in Ͼ80% reduction in activity, indicating that sequences important for p67 phox promoter activity are located between bp -733 and -682.
To identify the potential protein-binding sites in this active region, a DNase I protection assay was carried out using nuclear extracts from HL-60 cells and a 32 P-labeled fragment of DNA corresponding to the region of -733 to -637 of the p67 phox gene. A broad protected region corresponding to the 25 nucleotide sequence CTCAGAATAGGGGAGGGGCAGGACA located at -704 to -680 was identified (Fig. 3). The core of this sequence, GGGAGGG (-694 to -688), is very similar to the consensus binding sequence (GGGCGGG) for the Sp1 family of ubiquitously expressed transcription factors.
To determine whether the p67 phox sequence could bind Sp1 factors, EMSA was carried out using a labeled oligonucleotide corresponding to the protected region of the gene. This oligonucleotide formed a number of DNA-protein complexes with nuclear extract from HL-60 cells (Fig. 4A). A pattern characteristic of Sp1 factors binding to their cognate element was observed, and this was confirmed by competition with both Horizontal arrows indicate the exon/intron boundaries. The identified cis elements are boxed and labeled, and the core consensus sequences are underlined. Nucleotides in the core consensus sequences that were mutated in this study are shown in italics below the wild-type sequence. Oligonucleotides corresponding to the complete sequence shown in each box were used for EMSA and site-directed mutagenesis of reporter constructs. The bent arrows indicate the 3Ј ends of the two sets of reporter constructs that were used in this study. For clarity, the multiple transcriptional start sites described by others (19,25,26) were not included.
homologous probe and consensus Sp1 probe (data not shown). Specificity was further confirmed in supershift EMSA using antibodies to human Sp1 and Sp3 (Fig. 4A). These data indicate that Sp1 and Sp3 bind to the proximal promoter of the p67 phox gene.
To determine whether members of the Sp1 family of proteins bind in vivo to this cis element, we performed a ChIP analysis on the HL-60 cells. DNA-protein complexes were cross-linked in vivo by the addition of formaldehyde to the cultures, and the chromatin DNA was sheared and immunoprecipitated with anti-Sp3 antiserum. Cross-linking was reversed, and the released DNA was purified and used as the template for PCR amplification by genomic primers specific for this region of the p67 phox promoter. As a control for antibody specificity, we carried out the assay using an antibody to the transcription factor C/EBP␣, which is expressed in HL-60 cells but does not have a cognate binding site in this region of the p67 phox promoter. As a control for PCR specificity, we used p40 phox primers designed to amplify a region of the p40 phox gene that does not contain Sp1 or C/EBP␣ binding sites. The PCR primers specific for p67 phox produced an amplicon of predicted size (317 bp) from both the total input DNA and the anti-Sp3 immunoprecipitates but not from the anti-C/EBP␣ immunoprecipitates (Fig. 4B). Furthermore, the p40 phox primers produced an amplicon only from the total input DNA. These data indicate that endogenous Sp3 protein in HL-60 cells binds to the p67 phox promoter and that Sp3 or Sp1 factors may promote transcription of the p67 phox gene in vivo.
The First Intron Is Essential for Myeloid-specific Activity of the p67 phox Promoter-Although the reporter gene constructs described above showed promoter activity, it was not restricted to cells of myeloid lineage. These constructs were active in several cell types, including the nonmyeloid cell lines LoVo and HCT-116. To locate the DNA sequences that direct myeloidspecific expression of the p67 phox gene, we extended the promoter sequences in both directions. PCR primers were designed based on the sequence of the p67 phox gene deposited in GenBank TM (NM005003), and the genomic DNA libraries in the PromoterFinder kit were used as templates. Initially, we kept the 3Ј end of the p67 phox promoter fragment unchanged at bp -534 and extended the 5Ј end to bp -3669, producing the construct pGL3-p67-3669. It was reasonable that this construct might dictate myeloid-specific expression of the reporter luciferase gene, because it contained two strong binding sites for the myeloid-specific transcription factor PU.1, one at bp -3291 and the other at -2584, which were identified by sequence analysis and EMSA (data not shown; also refer to Fig.  6D). However, the pGL3-p67-3669 construct was 30 -50% less active than pGL3-p67-2022 and was functional in both myeloid (HL-60 and PLB-985) and nonmyeloid (LoVo and HCT-116) cell lines (data not shown).
We then analyzed the downstream sequences for cis elements that might mediate myeloid-restricted expression. A segment of the p67 phox gene extending from bp -4 to bp -3624 (relative to ATG) was obtained by PCR amplification. This DNA segment, composed of the 5Ј-flanking sequence, exon 1, intron 1, and 27 bp of exon 2 of the gene, was subcloned into the pGL3-Basic reporter vector as before. This reporter construct was designated pGL3-p67-3624. Two 5Ј deletion constructs that extended to bp -985 (291 bp 5Ј to the active Sp1 site) and bp -678 (16 bp 3Ј to the active Sp1 site) were also made for comparison. The highest luciferase activity was observed in HL-60 lysates prepared from cells transfected with construct pGL3-p67-985, compared with pGL3-Basic vector (Fig. 5). Further extension from bp -985 to -3624 (pGL3-p67-3624) led to a 65% decrease in the luciferase activity. Deletion construct pGL3-p67-678, which excluded the Sp1 site, showed a 55% reduction of the promoter activity compared with pGL3-p67-985. A similar reduction in promoter activity was obtained when the Sp1 site was mutated (Fig. 5). However, when this Sp1 mutant construct was tested in nonmyeloid HeLa or Jurkat cells, no significant changes of the promoter activity were observed. More importantly, the promoter activity of pGL3-p67-985 was about 35-fold higher in HL-60 cells than in either Jurkat or HeLa cells. Combined with the results obtained with the first set of reporter constructs (intron 1 excluded), these data suggest that 1) the first intron plays a critical role in myeloid-specific expression of the p67 phox gene, and 2) the Sp1 site can contribute to p67 phox expression in myeloid cells by cooperation with intron 1. The first finding is in good agreement with Eklund and Kakar's (19) report that cis elements present in the first intron of the p67 phox gene are responsible for IFN-␥ induction of p67 phox gene expression in the myeloid cell line U937. A sequence between -196 and -164 (AAAGGT-GGGGACATTTCCTGATGCATTTGCAAC) was found to be responsible for this induction and was identified as a HAF1 site, because it binds a trimolecular HAF1 complex (composed of PU.1 or Elf1, ICSBP, and IRF-1). We will refer it as the PU.1(1)/HAF1 site because this sequence also binds PU.1 protein independently (see below). Inspection of the intron 1 sequence revealed two additional well defined cis-acting elements, an AP-1 site (TGAGTCA) located at -210 and a second PU.1 site located at -289 ( Fig. 1). We further investigated the role of these binding sites in p67 phox gene transcription.
Transcriptional Activation of the p67 phox Promoter by PU.1 Is Influenced by the Location of the PU.1 Binding Site-To determine whether the PU.1 sites within intron 1 of the p67 phox gene bind PU.1 protein, we performed EMSA using HL-60 nuclear extract. A DNA-protein complex was formed between 32 P-labeled double-stranded oligonucleotide probe encompassing the PU.1(2) site and the nuclear extract (Fig. 6A). This complex was supershifted by PU.1 antibody but not by IgG, demonstrating that endogenous PU.1 protein binds to this site of the p67 phox promoter. However, the downstream PU.1 site, designated PU.1(1)/HAF1 in Fig. 1, exhibited more complexity. Oligonucleotide probes derived from this site formed several bands of DNA-protein complexes on EMSA (Fig. 6B), most of which could be considered specific, because their formation was inhibited by excess unlabeled probe. The two lower bands contained PU.1 or its degradation products because they were abrogated by PU.1 antibody. On the other hand, the two upper bands were supershifted by Elf-1 antibody, indicating that this alternative member of the ets family was a component of the complexes. The latter observation is reminiscent of reports that Elf-1, rather than PU.1, is present in complexes binding the HAF1 site in the gp91 phox promoter (18). Finally, the relative binding affinity of these PU.1 sites was compared with the previously characterized p47 phox promoter PU.1 site, using cold competition EMSA. As shown in Fig. 6C, p67-PU.1(2) and FIG. 4. Sp1 and Sp3 bind to the p67 phox promoter in vitro and in vivo. A, EMSA. A 32 P-labeled DNA probe (CTCAGAATAGGGGAGGG-GCAGGACA) corresponding to the putative Sp1/Sp3 site in the p67 phox promoter (bp -704 to -680) was incubated without (left lane) or with HL-60 nuclear extract (6 g). Two g of antibodies to Sp1 and/or Sp3 or control IgG were then added to the reaction as indicated. DNA-protein complexes were separated on a 5% polyacrylamide gel. Sp1/3 and Sp3 indicate the specific complexes, whereas SS indicates the supershifted complexes. B, Sp3 binding to the Sp1/3 site was confirmed in vivo by ChIP analysis. Formaldehyde-cross-linked chromatin prepared from 2 ϫ 10 7 HL-60 cells was incubated with antibodies to Sp3 or C/EBP␣ or in the absence of antibody (input). The cross-linking in the immunoprecipitates was reversed, and the DNA was purified and then analyzed by PCR using primers specific for either the p67 phox promoter or p40 phox promoter. PU.1(1)/HAF1 bound to PU.1 protein in HL-60 nuclear extracts only weakly, being 100-and 1000-fold less active, respectively, than the p47-PU.1site. The binding affinities of p67-PU.1 (2) and PU.1(1)/HAF1 were also compared with the apparently nonfunctional PU.1(3) and PU.1(4) sites described above. Competition EMSA showed that the upstream PU.1 sites possessed much stronger PU.1 binding affinity than those within the first intron (Fig. 6D). This result was not unexpected based on our previous study (21) showing that the (G/C)T residues flanking the 3Ј end and the tandem T or A at the 5Ј end of the PU.1 binding consensus sequence GAGGAA are required for optimal PU.1 binding ability.
Using ChIP analysis, we then investigated whether PU.1 protein binds to these PU.1 sites in vivo. The assay was carried out in HL-60 cells as before, except that antibody specific to PU.1 was used to immunoprecipitate the cross-linked chromatin DNA. As shown in Fig. 6E, PCR amplicons from the regions containing PU.1(4), PU.1(3), and both PU.1(2) and PU.1(1)/ HAF1 were obtained with the PU.1 immunoprecipitated chromatin, but not with the negative control C/EBP␣ immunoprecipitates. The PCR negative control, a DNA fragment of the CCR5 promoter that does not contain either PU.1 or C/EBP␣ binding sites, was amplified only from the input DNA. These data demonstrate that endogenous PU.1 protein binds to the p67 phox promoter at the indicated sites in vivo.
We next tested the role of these PU.1 sites in p67 phox promoter activity by transient transfection assays. Mutation of either the PU.1(2) or PU.1(1)/HAF1 site in the pGL3-p67-985 construct reduced the p67 phox promoter activity by ϳ50% in HL-60 cells (Fig. 7A), but produced no significant difference in promoter activity in HeLa or Jurkat cells, indicating the importance of these PU.1 sites and their binding to PU.1 protein for expression of the p67 phox gene in myeloid cells. On the other hand, mutations of the PU.1(3) and PU.1(4) sites in the pGL3-p67-3624 construct produced no reduction of promoter activity (Fig. 7B). We conclude that p67 phox transcriptional activation by PU.1 is dependent on the location of the PU.1 binding site.
AP-1 Is Critical for p67 phox Promoter Activity-A TF-SEARCH search analysis (27) identified an exact consensus AP-1 site, ATGAGTCAG, located between bp -211 and -203. When a labeled oligonucleotide probe derived from this site was used in EMSA, a specific band was observed with HL-60 nuclear extract. This complex could be inhibited with excess wildtype but not mutated (TGAGTCA to TGAGTTG) probes (Fig.  8A). The identity of the AP-1 proteins forming this complex was investigated by supershift assays using specific antibodies. A broadly reactive antibody to the Jun family of AP-1 proteins (c-Jun/AP-1) effectively abolished the formation of the DNAprotein complex, indicating that the Jun proteins were probably a component of all the dimeric complexes binding this sequence (Fig. 8A). Antibodies specific for the individual members of the Jun family showed that c-Jun, Jun B and Jun D formed complexes with the p67 phox AP-1 site, Jun D being the most abundant and c-Jun the least abundant. Antibody to the Fos family of proteins greatly reduced but did not abolish the specific complex, suggesting that the binding proteins in the HL-60 nuclear extract were made up predominantly of Jun/Fos heterodimers and a small amount of Jun/Jun family dimers. Antibody to CREB-1 and ATF-2, like the irrelevant anti-GATA-6 and the negative control IgG, had no effect on the EMSA reaction.
To investigate whether AP-1 factors bound this site in vivo, a ChIP assay was carried out with the antibodies broadly reactive with c-Fos or c-Jun family proteins. As a negative control, we included an immunoprecipitation reaction with the C/EBP␣ antibody. As shown in Fig. 8, B and C, the p67-intron 1 primers produced a 353-bp amplicon, as predicted, from the input as well as from the anti-c-Fos and anti-c-Jun immunoprecipitated DNA, but not from the anti-C/EBP␣ immunoprecipitates (Fig. 8, B and C). The PCR control p40-PU.1 primers produced a PCR product only from the input DNA. The p40-PU.1 primers were designed to amplify a region of the p40 phox promoter that does not contain consensus AP-1 or C/EBP␣ binding sites. These data indicate that Fos and Jun proteins in HL-60 cells bind to the p67 phox promoter, most abundantly as heterodimers, and thus may promote p67 phox gene transcription in vivo.
The influence of the AP-1 site on p67 phox promoter activity was investigated in the HL-60 cells. Mutation of the AP-1 site alone in the strongly active pGL3-p67-985 construct (ϳ700fold increase over empty vector) effectively abolished promoter activity (Fig. 9A). This result was surprising, because we had observed that the Sp1 and the PU.1(1)/HAF1 and PU.1(2) sites also contributed to the overall activity to the construct. To determine the contribution of the AP-1 site alone, the construct pGL3-p67-405-3 M, bearing mutations at the PU.1(1)/HAF1, PU.1 (2), and Sp1 sites, was investigated. The decrease in activity was greater than that caused by the sum of mutations of each site alone, but this reduction was not as complete as that seen with the AP-1 mutation alone. When all four cis elements  Fig. 1) and transfected into HL-60, HeLa, or Jurkat cells. Luciferase activity was assayed as described in Fig. 5. B, the pGL3-p67-3642/-4 construct was mutated at the PU.1(3) or PU.1(4) sites (see Fig. 1) and transfected into HL-60 cells. Luciferase activity was assayed and the results were expressed as in Fig. 5. The data shown (mean Ϯ S.E.) are from at least three independent experiments. were mutated, the promoter activity was completely lost, indicating that these four sites are responsible for all the transactivation activity displayed by this construct in HL-60 cells. Furthermore, in the construct pGL3-p67-678, in which the region containing the active Sp1 site was deleted, mutation of the AP-1 site again resulted in the abolition of transacting activity. This confirmed that the PU.1(1) and PU.1(2) sites in the intron are insufficient to drive promoter activity in myeloid cells and are completely dependent on the AP-1 site. These data strongly suggest that the AP-1 site is essential for the activity of the p67 phox gene promoter and that the factors that bind this site act cooperatively with those binding the other cis elements, with the AP-1 factors playing the dominant role.
To confirm that AP-1 can transactivate p67 phox promoter activity, co-transactivation experiments were performed in F9 embryonal carcinoma cells, which lack endogenous AP-1 activity and express only very low levels of Fos/Jun proteins. The promoterless luciferase reporter vector pGL3-Basic and the p67 phox luciferase constructs pGL3-p67-985, pGL3-p47-985-AP1mut and pGL3-p67-985-4m (all four identified sites mutated) were co-transfected into F9 cells with either the Jun B expression plasmid (CMV Jun B) or with the empty vector (Fig.  9B). Co-expression of Jun B increased the activity of the p67 phox promoters about 3-fold, whereas there was no such increase in the same construct with a mutated AP-1 site or four mutated sites. These results show that Jun proteins can transactivate the p67 phox promoter through the AP-1 site identified in the intron 1 region. The modest increase in transacting activity that was observed might reflect our earlier observations that the pGL3-p47-985 is only very weakly active in nonmyeloid cells. It could also be explained by the fact that only a very small pool of Fos proteins is available in the F9 cells. This would limit the formation of Fos/Jun heterodimers, which may be more active than Jun B/Jun B homodimers in activating the p67 phox promoter.

DISCUSSION
In previous studies, we demonstrated an essential role for a single ets family transcription factor, PU.1, in the transcriptional regulation of both the p40 phox and p47 phox genes (20). 2 In contrast, the studies presented here show that transcription of the p67 phox gene is regulated by a diverse array of factors, including AP-1, Sp1, PU.1, and the complex HAF-1. Moreover, our transfection studies indicate that it is those factors that bind the AP-1 site, rather than PU.1, that are essential for the functional activity of the promoter. Supershift analysis using specific antibodies and HL-60 nuclear extract identified members of the Jun and Fos family of transcription factors that bind this critical AP-1 site.
The region of the gene encoding the 5Ј-UTR of the p67 phox transcripts is interrupted by a 480-bp intronic sequence (intron 1) starting at bp -510 relative to the translation initiation codon. Previous studies had provided evidence that p67 phox transcription may be initiated from a number of sites, including at least one in the first intron (19,25,26). Our initial studies of the p67 phox promoter were focused on the 5Ј-flanking region of the gene and the first series of promoter deletion constructs all terminated downstream at bp -534, thus including a number of these putative transcription initiation sites, but excluding intron 1. The longest construct in this series extended from position -3669 and included two potential PU.1 sites (PU.1(3) and PU.1(4), Fig. 1), which were identified based on our previously defined criteria for optimal flanking sequences (21). In gel shift assays, both of these sites were shown to bind PU.1 in vitro, and ChIP analysis indicated that they were occupied by PU.1 in vivo. However, this construct demonstrated strong promoter activity in both myeloid and nonmyeloid cell lines, indicating that these distal PU.1 sites did not direct myeloidspecific promoter activity. Furthermore, mutation of these sites, which abolished PU.1 binding, did not affect the activity of the reporter gene constructs, suggesting that the distal PU.1 sites contributed minimally to overall promoter activity. Deletion studies mapped the most potent promoter activity in these constructs to a region between -733 and -682 bp, in which a strong binding site (-694 to -688) for the ubiquitous Sp1 family of factors was identified by both DNase I footprint and gel shift analysis. Deletion or mutation of this Sp1 site resulted in a nearly complete loss of activity in these initial reporter gene constructs.
While these studies were in progress, Eklund and Kakar (19) Cross-linked HL-60 chromatin was immunoprecipitated with antibodies to c-Fos, c-Jun/AP-1, or C/EBP␣ or in the absence of antibody (input). The cross-linking was reversed, and the DNA was purified and then analyzed by PCR using specific primers for the AP-1 binding sites of the p67 phox promoter (p67-intron 1) or for the p40 phox promoter (p40-PU.1) as a control.
reported that IFN-␥-induced expression of p67 phox in myelomonocytic U937 cells was mediated by a PU.1/HAF-1 binding site located in the first intron of the p67 phox gene (-180 bp relative to the ATG). The HAF-1 complex was reported to contain PU.1, IRF-1, and the macrophage-specific factor ICSBP. We examined the role of the first intron in myeloidrestricted function of the p67 phox promoter using a construct, pGL3-p67-985/-4, which extended from ϳ300 bp upstream of the active Sp1 site to 4 bp 5Ј of the ATG, and thus included 5Ј-flanking sequence, exon 1, intron 1, and 26 nucleotides of exon 2. This construct showed strong promoter activity when transfected into HL-60 cells but was inactive or very weak in both Jurkat and HeLa cells. Thus, it appears that the proximal 5Ј-flanking region, together with the sequence that extends to and includes the first intron and 26 nucleotides of exon 2, contains the information required to direct myeloid-specific transcription of the p67 phox gene.
Inspection of the DNA sequence of intron 1 identified potential AP-1 and PU.1 binding sites, in addition to the PU.1/HAF-1 site described previously (19). EMSA showed that PU.1 from either HL-60 nuclear extracts or in vitro translations bound the PU.1 and PU.1/HAF-1 sites and that c-Fos and the c-Jun, Jun B, and Jun D members of the Jun family present in HL-60 nuclear extract bound the AP-1 site. Combined mutations of the identified AP-1, Sp1, PU.1, and PU.1/HAF-1 sites in the pGL3-p67-985/-4 construct effectively abolished its activity in HL-60 cells, indicating that these sites are the major elements regulating transcription in this myeloid cell line. Mutation of either the Sp1, PU.1, or PU.1/HAF-1 sites alone each reduced promoter activity by 30 to 50%. However, mutation of the AP-1 site alone resulted in greater than 90% reduction in activity. Thus, the AP-1 site appears to be essential for the function of the promoter. However, because the activity mediated by the AP-1 site alone was weak when transfected into HL-60 cells, factors binding this site appear to be acting cooperatively or synergistically with those binding to other sites.
Regulation of transcription through cooperation between AP-1 proteins and members of the ets family of protooncogenes has been reported for several genes. It was originally shown that the ets factors c-ets-1 and c-ets-2 could cooperate with AP-1 factors to activate transcription through a combined AP-1/PEA3 site in the polyoma enhancer. (28) Similar cooperativity was described in the transcription of genes such as uPA (AP-1 and ets-2), interleukin 3 (AP-1 and Elf-1), and granulocyte-macrophage colony-stimulating factor (AP-1 and Elf-1), among others (29 -32). In these examples, cooperative transactivation is mediated by a combined AP-1/ets binding motif in which the AP-1 site is located im- FIG. 9. Among the four active promoter elements, the AP-1 site is essential for the p67 phox promoter activity in myeloid cells. A, mutation of the AP-1 site alone or in combination with mutations of the PU.1 and Sp1 sites reduces p67 phox promoter by Ͼ90%. The indicated mutations (also see Fig. 1) were introduced into the pGL-p67-678/-4 or pGL-p67-985/-4 constructs by site-directed mutagenesis. Transfection of HL-60 cells, determination of luciferase activity, and expression of results were as described in Fig. 5. B, coexpression of Jun B transactivates the p67 phox promoter in F9 cells. Two hundred ng of Jun B expression plasmid (CMV Jun B) or the empty vector (CMV vector) were co-transfected into F9 cells with 200 ng of the indicated wild-type, mutant AP-1 (AP-1Mt), or multiple site-mutated (AP-1/Sp1/PU.1(1)-(2)Mt) reporter constructs. Determination of luciferase activity and expression of results were as described in Fig. 5. mediately adjacent to or in very close proximity to the ets binding sequence. In contrast, the critical AP-1 site we have identified in the p67 phox gene is located 28 nucleotides upstream of the PU.1/HAF-1 site and 80 nucleotides downstream of the PU.1(2) site. Moreover, the nucleotides flanking the AP-1 site do not form a consensus sequence for the ets family of proteins. Nevertheless, the factors binding this site appear to cooperate with the factors binding the two PU.1 sites and with those binding the Sp1-binding motif located 487 bases upstream. A similar, but not identical case of cooperative interaction between nonadjacent PU.1, GC-box, and AP-1 sites has been described for the gene encoding the macrophage-specific transmembrane glycoprotein macrosialin, which is induced early in myelopoiesis and is further up-regulated in the monocytic, but not the granulocytic lineage (33). However, in the macrosialin promoter, the GC box (reported to bind factors distinct from Sp1 and Sp3) and PU.1 sites appear to be co-dominant with the AP-1 site, whereas our transfection data in undifferentiated HL-60 cells indicate that the PU.1, PU.1/HAF-1, and Sp1 binding sites in the p67 phox promoter are individually less dominant than the AP-1 site.
The mechanisms responsible for the cooperative effects among the functional cis elements in the p67 phox promoter remain to be determined. Recent studies have shown that activation of transcription by a number of DNA-binding factors requires the recruitment of co-activator complexes that contain CBP and p300 (19, 34 -41). These include CREB, NF-B, nuclear hormone receptors, p53, AP-1(fos/jun), and PU.1. In their recent report, Eklund and Kakar (19) showed that the HAF-1 complex that bound the PU.1/HAF-1 sites in both the p67 phox and gp91 phox promoters and mediated the induction of these genes in response to IFN-␥ also required CBP/p300 binding for transcriptional activity. Because CBP/p300 is present in limiting quantities in the cell, it is thought that ternary complexes of DNA-binding proteins that enhance CBP/p300 binding would function as a more effective transcriptional unit (22,42) Therefore, the cooperative interactions among the factors binding the multiple cis elements in the p67 phox promoter may, in part, be a function of the enhanced ability to bind co-activator proteins.
The pGL3-p67-985/-4 construct was preferentially active in HL-60 cells, indicating that it contains the elements necessary for myeloid-specific expression. Although this selectivity was most likely mediated by the myeloid-and B-cell-specific PU.1 protein, which we showed binds the intron 1 PU.1 sites both in vitro and in vivo, our competition EMSA studies indicated this binding was relatively weak compared with the binding of PU.1 to the active PU.1 site in the p47 phox promoter. In addition, the mutation of either of the PU.1 sites in the pGL3-p67-985/-4 construct resulted in only a moderate reduction in promoter activity, which is in keeping with our previous observations that the avidity with which PU.1 binds its cognate element in a promoter is predictive of the transactivation activity that results from the binding. These data indicate that PU.1 by itself is insufficient to direct the efficient transcription of the p67 phox gene in a myeloid-specific manner but that the other transcription factors are required. Interestingly, although a significant proportion of the transactivation activity of the pGL3-p67-985/-4 construct in HL-60 cells was mediated by the Sp1 and AP-1 sites, this construct was still relatively inactive in HeLa and Jurkat cells, which also express the ubiquitous factors that bind these sites. Together, these observations suggest that AP-1 and perhaps Sp-1 transcription factors bind and transactivate the p67 phox promoter in a configuration that is myeloid-specific, serving to reinforce the activity of the PU.1 factors. A similar but not identical situation has been described in the promoter of the myeloid integrin gene CD 11b, in which an essential Sp1 site is bound in vivo by its cognate factors in a myeloid cell line (U937) but not in HeLa cells, perhaps mediated by the prior binding of the PU.1 factor to a nonadjacent site (22).
These studies have delineated some of the major factors and cis elements that regulate p67 phox gene transcription in an undifferentiated myeloid cell line. Unique among the component genes of phagocyte NADPH oxidase, the transcription of p67 phox appears to be dependent on those ubiquitous factors that bind the AP-1 motif and involves the complex interaction of these factors with PU.1 and Sp-1, perhaps in a configuration that is restricted to cells of myeloid lineage. The level of p67 phox transcripts in HL-60 cells is relatively low, but it is rapidly increased in response to the induction of terminal differentiation with agents such as Me 2 SO or retinoic acid (10 -12). Important future directions will be to characterize the mechanism of AP-1 dependence of p67 phox gene transcription and to identify the essential pathways that modulate the increase in gene transcription during the process of cell differentiation.