YY1 Binds Five cis-Elements and Trans-activates the Myeloid Cell-restricted gp91 phox Promoter*

Four transcriptional activatingcis-elements within the gp91 phox promoter bind a protein complex of similar mobility and binding specificity, denoted BID (binding increased during differentiation). The intensity of BID complexes increases upon myeloid cell differentiation, coincident with induction of gp91 phox expression, and BID competes with the transcriptional repressor CDP for binding to each of these promoter elements. To determine the identity of BID, an expression library was ligand screened with the BID-binding site that surrounds the −145-base pair (bp) region of the gp91 phox promoter. One recovered factor that exhibits the expected binding specificity is YY1, a ubiquitous multifunctional transcription factor. BID complexes that form with the four binding sites within the gp91 phox promoter are disrupted by YY1 antiserum, and a fifth YY1-binding site was detected in the −412-bp promoter region. Overexpression of YY1 in transient co-transfection assays trans-activates a minimal promoter containing two copies of the −145-bp binding site from the gp91 phox promoter. Neither the level of YY1 protein nor DNA binding activity increases during myeloid cell differentiation. These studies identify a target gene of YY1 function in mature myeloid cells, and demonstrate that YY1 function can be controlled during myeloid development by the modulation of a competing DNA-binding factor.

Phagocytic blood cells such as monocyte/macrophages and neutrophils use the NADPH oxidase to produce a respiratory burst to kill microbes. Several subunits comprise the NADPH oxidase, including p67 phox , p47 phox , p22 phox , and gp91 phox (1). The gp91 phox gene is transcriptionally controlled and is expressed nearly exclusively in mature myeloid cells (2). Previously we demonstrated that the Ϫ450 to ϩ12 base pair (bp) 1 region of the gp91 phox promoter is capable of directing reporter gene expression in a subset of monocyte/macrophages in trans-genic mice (3), and responds to interferon (IFN)-␥ stimulation in transfected myeloid cell lines (4). An enhancer region located 50 kilobases upstream of the proximal promoter is additionally required to direct appropriate expression of gp91 phox in the full spectrum of mature myeloid cells (5).
Previously we reported that BID binds to four sites within the Ϫ450 to ϩ12 bp gp91 phox promoter (8,9). These conclusions were based on eletrophoretic mobility shift assay (EMSA) analysis that demonstrated complexes of similar mobility and binding specificity with each of four promoter probes (8,9). Mutations of the putative BID-binding sites surrounding the Ϫ355, Ϫ225, and Ϫ145 bp regions of the gp91 phox promoter result in decreased promoter activity in PLB-985 myeloid cells in response to IFN-␥ stimulation (9). Truncation of the gp91 phox promoter to Ϫ102 to ϩ12 bp removes four CDP-binding sites and reveals a promiscuous promoter that is active in some cells not expressing the endogenous gp91 phox gene (7). Specific ablation of a BID-binding site at Ϫ90 bp decreases this promiscuous promoter activity by 50% in HEL cells (8). These data indicate that BID functions as a transcriptional activator. Data base searching with the four putative BID-binding sites revealed no common consensus binding sites for known transcription factors. Eklund and Kakar (13) reported the cloning of a novel component of the BID complex (denoted TF1 phox in that report). However, data from Yamit-Hezi et al. (14) demonstrate that this clone is a bacterial contaminant present in some commercially available libraries. Hence, the identity of the BID complex remained to be identified.
We undertook a molecular cloning approach to identify BID. Ligand screening of a gt11 HeLa cell cDNA expression library was performed using as a probe the Ϫ145-bp BID-binding site derived from the gp91 phox promoter. One sequence specific DNA-binding factor obtained is YY1, a ubiquitously expressed multifunctional member of the GLI Krü ppel-related family of zinc finger transcription factors. Further examination of the Ϫ450 to ϩ12 bp gp91 phox promoter revealed a fifth potential YY1-binding site. YY1 is present in all five identified BID complexes within the gp91 phox promoter, and transient transfection studies confirm that YY1 functions as a transcriptional activator of the gp91 phox promoter.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-A ␤-globin TATA box minimal promoter fragment was excised from a human growth hormone gene reporter vector (gift of Ellis Neufeld, Harvard) with HindIII and BamHI and cloned into the luciferase vector, pXP2 (15), that was digested with HindIII and BglII. This construct was digested with BamHI, de-phosphorylated with shrimp alkaline phosphatase (Amersham Pharmacia Biotech/U. S. Biochemical, Cleveland, OH), and ligated to phosphorylated Ϫ145 Core or Ϫ145 mut oligonucleotides (see below). The nucleotide sequence was determined for constructs containing a dimer of each oligonucleotide, and those with a dimer in the forward orientation were prepared by cesium chloride ultracentrifugation to be used in transient transfections (see below).
Cell Culture and Transient Transfection Assays-HeLa human cervical choriocarcinoma cells, K562 human chronic myelogenous leukemia cells, and HEL human erythroleukemia cells were obtained from the American Type Culture Collection (Rockville, MD). The PLB-985 human myelomonoblastic cell line (16) was a gift of Thomas Rado (Birmingham, AL). Suspension cells (HEL, K562, and PLB-985) were grown in RPMI 1640 medium with 10% fetal bovine serum or Fetal Clone III (Bovine Serum Product, HyClone, Logan, UT) and 0.2 mM glutamine, 50 units/ml penicillin, and 50 g/ml streptomycin at 37°C and 5% CO 2 . Adherent cells (HeLa) were grown in similarly supplemented Dulbecco's modified Eagle's media.
Cells were diluted in media to a concentration of 10 7 cells/300 l and placed into electroporation cuvettes (0.4-cm gap). One microgram of luciferase test plasmid and 5 g of CB6 ϩ CMV driven expression plasmid or CB6 ϩ -YY1 (17) (gifts of Kenneth Walsh, Tufts) were added to each cuvette. Each sample also contained 0.25 g of cytomegalovirus promoter/␤-galactosidase (CMV/␤-gal) plasmid that serves as an internal control for transfection efficiency. Samples were electroporated using a Bio-Rad Gene Pulser at 960 microfarads and 220 V, re-suspended in 10 ml of media and incubated at 37°C for 24 h. Cells were harvested, washed with PBS, resuspended in 100 l of lysis buffer (Promega Inc., Madison, WI), and incubated at room temperature for 15 min. Twenty microliters of cell lysate was assayed for luciferase activity as described by the manufacturer (Promega, Inc.) using a Lumat LB 9501 (Berthold, Gaithersburg, MD) luminometer. ␤-Gal activity was detected as described (18) and used to adjust luciferase values to correct for differences in transfection efficiency. Each sample was transfected in duplicate, and two independent plasmid preparations of each construct were used in five independent experiments.
In Vitro DNA Binding Protein Assays-Nuclear extracts were prepared by the method of Dignam et al. (19). PLB-985 cells were treated with agents that induce gp91 phox expression as described previously (9). To prepare fractionated extract, six liters of K562 cells were grown to log phase and nuclear extract was prepared as described above. Nuclear extract was then fractionated by heparin-agarose (Sigma) chromatography using a stepwise gradient of KCl in Dignam Buffer D supplemented with protease inhibitors. Fractions exhibiting abundant BID DNA binding activity (0.2-0.3 M KCl) were pooled and dialyzed to 0.1 M KCl. For mini-nuclear extracts, 10 7 HEL cells were transiently transfected with 20 g of CB6 ϩ or CB6-YY1 expression plasmids as described above, incubated for 12 h, and mini-nuclear extracts prepared as described (20).
Complimentary oligonucleotides were annealed and radiolabeled using T4 polynucleotide kinase and [␥-32 P]ATP. Radiolabeled probes were purified as described previously (8). EMSA was performed as described previously (6), using 4 -9 g of nuclear extract and 0.1-0.5 g of poly(dI-dC). Non-radioactive competitor oligonucleotides were added to designated samples and incubated on ice for 12 min. One to two microliters of YY1 antiserum (catalog number sc-1703x, Santa Cruz Biotech, Santa Cruz, CA) or Ets-2 antiserum (catalog number sc-351x, Santa Cruz Biotech) was added to designated samples and incubated on ice for 40 min. Radiolabeled oligonucleotide probe (20,000 cpm) was added (except where otherwise indicated) and samples were incubated on ice for an additional 30 min. The reactions were loaded onto native 6% polyacrylamide gels (except where otherwise indicated) and electrophoresis was performed at 25 mA for 1.25 h at 4°C in 0.5 ϫ TBE. Gels were then dried and exposed to x-ray film at Ϫ70°C.
Ligand Screening a gt11 cDNA Expression Library-A gt11 cDNA expression library derived from HeLa cells (CLONTECH, Palo Alto, CA) was a generous gift of Dr. Saw Yin Oh (Indianapolis, IN). Y1090 cells were incubated with bacteriophage for 15 min at 37°C and plated at a density of 50,000 plaques/150-mm plate, incubated at 42°C for 5 h, and nitrocellulose filters impregnated with 10 mM isopropyl-␤-D-thiogalactoside were placed on the plates overnight at 37°C. Denaturation/ renaturation has been shown to increase the DNA binding affinity of some factors (22). Filters were twice immersed in 6 M guanidine hydrochloride (Roche Molecular Biochemicals, Indianapolis, IN) for 5 min at 4°C, then in successive 2-fold dilutions of denaturant to slowly renature the fusion proteins. The filters were then blocked overnight at 4°C in blocking buffer (2.5% dried milk, 25 mM HEPES (pH 8.0), 1 mM dithiothreitol, 10% glycerol, 50 mM NaCl, 0.05% lauryldimethylamine oxide (Calbiochem), 1 mM EGTA). Filters were rinsed briefly in TNE-50 (10 mM Tris (pH 7.5), 50 mM NaCl, 1 mM EGTA, 1 mM dithiothreitol).
The oligonucleotide containing a high affinity BID-binding site (Ϫ145 Core) was phosphorylated, concatenated using T4 DNA ligase, and radiolabeled with [␣-32 P]dCTP using random priming. Probe was added to 10 6 cpm/ml of 1 ϫ binding buffer (25 mM HEPES (pH 7.9), 3 mM MgCl 2 , and 40 mM KCl). Herring sperm DNA was added as a nonspecific competitor to a final concentration of 4 g/ml. Filters were incubated with probe at 4°C overnight, washed briefly three times in TNE-50 at room temperature, then exposed to x-ray film overnight at Ϫ70°C. A total of 2.5 ϫ 10 6 plaques were analyzed. Positive plaques were picked and eluted for a second round of ligand screening to test for binding specificity, for which filters were cut in half and probed with either the wild type Ϫ145 Core or Ϫ145 mut-concatenated probes. Clones exhibiting an appropriate binding specificity were purified by four successive rounds of ligand screening. DNA was prepared from purified plaques using the Wizard preparation (Promega, Inc., Madison, WI). Isolated DNA was digested with EcoRI and electrophoresis was performed on a 1% agarose gel. Insert fragments were recovered and cloned into Bluescript (Stratagene, La Jolla, CA). The nucleotide FIG. 1. Transcriptional regulators that interact with the proximal gp91 phox promoter. The transcriptional repressor CDP interacts with multiple promoter elements prior to terminal myeloid differentiation (6 -8). Upon terminal differentiation, the DNA binding activity of CDP is down-regulated, thus permitting the binding of transcriptional activators to overlapping binding sites (4, 6 -12). The transcription initiation site is located at ϩ1. sequence of subcloned fragments was determined by the dideoxy chain termination method using M13 and T3 primers. Obtained sequences were used to search the GenBank data base.
Western Analysis-Nuclear extracts were quantitated using the method of Bradford (23). SDS loading dye was added to 40 g of nuclear extract and the samples were boiled for 10 min then loaded onto a 4 -20% Tris glycine gel (Novex, San Diego, CA) and electrophoresis was performed at 180 V for 1.5 h at 4°C. The proteins were transferred to polyvinylidene difluoride membrane (Bio-Rad) for 2 h at 80 V, and the membrane was blocked in PBS ϩ 0.1% Tween 20 (PBS-T) containing 5% low fat milk. Following blocking, the membranes were washed three times with PBS-T and incubated with an antiserum raised against the full-length YY1 protein (Santa Cruz Biotech., Inc) diluted 1:20,000 in blocking buffer for 1 h at 25°C. The membrane was washed three times with PBS-T and secondary antibody conjugated to horseradish peroxidase was diluted 1:20,000 in PBS-T containing 5% low fat milk and incubated with the membrane for 1 h at room temperature. The membrane was washed five times in PBS-T, and chemiluminescent detection was performed according to the manufacturer's instructions (Amersham Pharmacia Biotech).

RESULTS
The BID DNA Binding Activity Is Not Induced during Myeloid Cell Differentiation-Numerous DNA-binding proteins interact with the proximal gp91 phox promoter to direct myeloid cell-restricted expression (Fig. 1). The transcriptional repressor CDP binds to multiple sites in the proximal promoter in undifferentiated myeloid cells, thus excluding the binding of transcriptional activators to overlapping binding sites. Myeloid precursor cells treated with agents that induce gp91 phox expression, such as phorbol ester (PMA), dimethylformamide, or IFN-␥ exhibit decreased CDP DNA binding activity and a concomitant increase in the intensity of BID complexes (6,7,24,25). CDP and BID bind to the gp91 phox promoter in a mutually exclusive manner (7). To facilitate study of BID, a truncated oligonucleotide was created which lacks several of the bases necessary for CDP binding to the Ϫ145 bp region of the gp91 phox promoter. This oligonucleotide (Ϫ145 Core) was

FIG. 2. BID is ubiquitously expressed and is not induced during phagocyte maturation.
A, BID is not a myeloid specific complex. EMSA analysis using the Ϫ145 Core oligonucleotide as probe was performed as described under "Experimental Procedures" using nuclear extracts derived from PLB-985, HeLa, K562, or HEL cells. Competitor Ϫ145 Core or Ϫ145 mut oligonucleotides (60 ng) were added where indicated. Arrows denote complexes that exhibit the binding specificity of BID. B, BID binding is not increased by agents that induce gp91 phox expression. EMSA analysis was performed as described under "Experimental Procedures" using 4 g of nuclear extract derived from either PLB-985 cells, or PLB-985 cells treated with PMA or IFN-␥. The Ϫ145 Core BID-binding site was used as a probe. The oligonucleotides Ϫ145 Core or Ϫ145 mut (60 ng) were added as competitors where indicated. Relative intensities of nonspecific EMSA complexes serve as internal controls for protein loading and integrity of the nuclear extracts. An arrow indicates the position of the BID complex.
FIG. 3. Isolation of a sequence-specific DNA-binding clone. A gt11 HeLa expression library was screened as described under "Experimental Procedures." The upper half of each filter was probed with a concatomer of the Ϫ145 Core-binding site, while the lower half of each filter was probed with the Ϫ145 mut concatomer that no longer binds the BID protein. A representative clone exhibiting sequence-specific DNA binding activity is shown on the right, and a nonspecific clone is presented on the left.
FIG. 4. The BID complex that binds the ؊145 Core site contains YY1. A, YY1 binds the Ϫ145 Core site. EMSA analysis was performed as described under "Experimental Procedures" using the Ϫ145 Core oligonucleotide as probe and nuclear extract derived from HEL cells. The Ϫ145 Core or Ϫ145 mut competitor oligonucleotides (60 ng) were added where indicated. YY1 (␣YY1) or Ets-2 (␣Ets2) antiserum (1-2 l) was added where indicated. An arrow denotes the position of the BID complex. B, complexes of similar mobility and binding specificity form with the Ϫ145 Core and YY1 consensus binding sites. EMSA analysis was performed as described under "Experimental Procedures" using Ϫ145 Core or YY1 consensus binding site probes and nuclear extract derived from HEL cells. The Ϫ145 Core, Ϫ145 mut, YY1, or YY1 mut oligonucleotides (60 ng) were added as competitors where indicated. An arrow denotes the position of the BID (YY1) complex. C, overexpression of YY1 leads to an increase in the intensity of the BID complex. EMSA analysis using mini-nuclear extract and the Ϫ145 Core probe was performed as described under "Experimental Procedures." Five micrograms of mini-nuclear extracts derived from HEL cells transfected with the CB6 ϩ parent vector or the CB6-YY1 expression vector were added where indicated. The Ϫ145 Core or Ϫ145 mut oligonucleotide competitor (60 ng) or YY1 or Ets-2 antiserum (1-2 l) was added where indicated. An arrow denotes the BID (YY1) complex. designed to include the binding site of BID, as deduced from methylation interference analysis (9). A similar strategy was used to design the Ϫ225 Core and Ϫ355 Core oligonucleotides (see below). The Ϫ145 Core oligonucleotide is capable of binding the BID protein, but does not bind CDP (Fig. 2, and data not shown). The Ϫ145 mutated oligonucleotide (Ϫ145 mut) contains a 4-bp mutation that disrupts BID contact sites and no longer binds the BID protein, as it no longer disrupts the BID EMSA complex formed with the wild type Ϫ145 Core probe (Fig. 2).
The Ϫ145 Core oligonucleotide probe was used to assess the distribution of the BID binding activity in the absence of competing CDP complexes ( Fig. 2A). The BID complex is disrupted upon addition of homologous oligonucleotide competitor (lanes 3, 6, 9, and 12) but not by the Ϫ145 mut competitor which no longer binds the BID complex (lanes 4, 7, 10, and 13). The BID complex is present in nuclear extracts derived from PLB-985 (lanes 2-4), HeLa (lanes 5-7), K562 (lanes 8 -10), and HEL (lanes 11-13) cells. Thus, expression of BID is not lineage specific. There are also faster migrating complexes in both the HeLa extract (lanes 5-7) and the K562 extract (lanes 8 -10) which behave with the same binding specificity as the previously identified slow mobility BID complex. We speculate that these may correspond to products of partial proteolysis of BID.
EMSA was performed using the Ϫ145 Core binding site probe and nuclear extracts derived from PLB-985 cells treated with agents that induce gp91 phox expression and were previously found to induce increased BID complex intensity (9,25). The BID complex, which is disrupted with the Ϫ145 Core oligonucleotide competitor (Fig. 2B, lanes 3, 6, and 9) but not with the Ϫ145 mut oligonucleotide (lanes 4, 7, and 10), is as intense in nuclear extract derived from untreated PLB-985 cells (lanes 2-4) as nuclear extracts derived from PLB-985 cells treated with PMA ( lanes 5-7) or IFN-␥ (lanes 8 -10). Effective induction of the PLB-985 cell cultures was confirmed by demonstrating the down-regulation of CDP DNA binding activity by EMSA analysis (data not shown). Hence, increased intensity of BID complexes with composite BID/CDP probes upon induction of gp91 phox expression is the result of decreased binding of CDP to overlapping binding sites.
Cloning of the BID DNA Binding Activity-Searches of the Transcription Factor Sites data base (Genetics Computer Group Inc., Madison, WI) (26) with the Ϫ145 Core BID-binding site sequence produced no matches with consensus binding sites for known transcription factors. Hence, a molecular approach was taken to determine the identity of the BID protein.
Because BID is ubiquitously expressed (Fig. 2A), ligand screening was performed with a gt11 HeLa cDNA expression library and the concatenated Ϫ145 Core BID-binding site probe. The Ϫ145 Core site demonstrates the cleanest EMSA complex and the highest affinity for the BID complex. 2 The Ϫ145 mut oligonucleotide, which contains a 4-bp mutation and no longer binds the BID complex, was used as a probe to examine the sequence specificity of recovered DNA binding activities.
Approximately 2.5 ϫ 10 6 plaques were screened, and 43 reproducibly positive DNA-binding clones were examined for binding specificity. Three clones bind to the wild type Ϫ145 Core probe, but not to the Ϫ145 mut probe (Fig. 3). The other 40 clones encode nonspecific DNA-binding factors as they bind both the Ϫ145 Core and Ϫ145 mut probes, and were not pursued further. A search of the GenBank data base revealed that the determined nucleotide sequence of a sequence specific clone matches the cDNA sequence of YY1 (data not shown), a ubiquitous multifunctional transcription factor that is a member of the GLI Krü ppel zinc finger factor family (27). The recovered 2 B. M. Jacobsen, unpublished observations. FIG. 5. YY1 interacts with multiple sites within the gp91 phox promoter. A, YY1 binds to each of four BID-binding sites. EMSA analysis was performed as described under "Experimental Procedures." Each of the four BID-binding sites: Ϫ90, Ϫ145 Core, Ϫ225 Core, and Ϫ355 Core was used as probe without (lanes 1, 4, 7, and 10) or with heparin-fractionated nuclear extract derived from K562 cells (all other lanes). YY1 antiserum (1-2 l) was added where indicated. An arrow denotes the position of the YY1 complex. B, high resolution of YY1 binding to the Ϫ225-bp binding site. EMSA was performed as described under "Experimental Procedures" using the Ϫ225 Core-binding site probe and heparin-fractionated nuclear extract isolated from K562 cells. To clearly resolve the BID complex, 60 ng of an oligonucleotide competitor containing a mutation which no longer binds the BID complex (BID mut), but retains binding of IRF members (8) was added (lanes 2-4). YY1 or Ets-2 antiserum (1-2 l) was added where indicated. An arrow denotes the position of the EMSA complex containing YY1. cDNA encompasses amino acid 85 to the stop codon and a portion of the 3Ј-untranslated region, and includes the region of YY1 previously reported to contain the DNA-binding domain (28).
Antiserum directed against YY1 was tested in EMSA analysis using the Ϫ145 Core probe (Fig. 4A). The BID complex that forms upon addition of nuclear extract derived from HEL cells (lane 2) is disrupted by homologous oligonucleotide competition (lane 3), but not by the Ϫ145 mut oligonucleotide competitor (lane 4), and is also abolished upon addition of YY1 antiserum (lane 5). The other complexes visible in the lanes do not exhibit the appropriate binding site specificity for BID and are not affected by the YY1 antiserum. Also, the BID complex is not affected by an antiserum directed against Ets-2 (lane 6), thus demonstrating the specificity of the disruption by YY1 antiserum.
Additional EMSA studies were performed using a previously described YY1-binding site oligonucleotide to compare the behavior of BID and YY1. The Ϫ145 Core site probe (Fig. 4B,  lanes 1-6) produces the BID complex (lane 2) which is disrupted by homologous competitor oligonucleotide (lane 3) and not by the Ϫ145 mut competitor (lane 4). The BID complex is also disrupted by the YY1 consensus binding site competitor (lane 5), but is not disrupted by a mutated YY1-binding site competitor (lane 6). Additionally, EMSA using the YY1 consen-sus binding site as a probe (Fig. 4B, lanes 7-10) demonstrates a complex of similar mobility (lane 8) and binding specificity (lanes 9 and 10) as the BID complex formed with the Ϫ145 Core probe. EMSA studies with nuclear extract derived from HeLa cells and the Ϫ145 Core site probe demonstrate that the previously detected faster migrating complexes ( Fig. 2A) are also disrupted by the YY1 specific antibody (data not shown). A complex of slower mobility also binds to the Ϫ145 Core site (lane 2). However, this complex does not contain YY1 because it is disrupted by both homologous and mutant competitors ( lanes  3 and 4), and is not disrupted by either the YY1 consensus oligonucleotide (lane 5) or YY1 antiserum (Fig. 4A, lane 5).
Nuclear extracts prepared from HEL cells transiently transfected with the CB6-YY1 expression vector (or empty parental expression vector) were analyzed by EMSA using the Ϫ145 Core site probe (Fig. 4C). The BID complex is greatly enhanced in cells overexpressing YY1 (lane 3) compared with those transfected with the parental CB6 ϩ expression vector (lane 2). The intensified complex is disrupted by competition with homologous oligonucleotide (lane 4) and not disrupted by the Ϫ145 mut oligonucleotide competitor (lane 5). The BID (YY1) complex is also supershifted upon addition of YY1 antiserum (lane 6), but is unaffected by antiserum directed against Ets-2 (lane 7). These results provide additional evidence that YY1 is present in the BID complex, and demonstrate that YY1 is either the FIG. 6. Identification of an additional YY1-binding site within the gp91 phox promoter. A, alignment of the lower strand of the consensus YY1-binding site (29) with the sequences of the four YY1 gp91 phox -binding sites as well as a putative fifth YY1-binding site in the Ϫ412-bp region of the gp91 phox promoter. All sequences correspond to the upper strand of the gp91 phox promoter (6), except the Ϫ225-bp site, for which the lower strand is shown. Underlined bases match the consensus YY1 binding sequence. B, YY1 binds the Ϫ412-bp site. EMSA was performed as described under "Experimental Procedures" using the Ϫ412-bp element as probe and heparin-fractionated nuclear extract isolated from K562 cells. YY1 or Ets-2 antiserum (1-2 l) was added where indicated. sole protein component of the BID complex, or is the concentration-limiting component of a multimeric BID DNA binding activity.
YY1 Binds to Multiple Sites within the gp91 phox Promoter-In previous work to biochemically purify and clone BID, nuclear extract derived from K562 cells was fractionated by heparin chromatography for use in DNA-affinity chromatography (data not shown). This partially purified material exhibits abundant BID activity in EMSA analysis (data not shown). Previous EMSA studies using four sites of the gp91 phox promoter demonstrated that a complex of similar size and binding specificity forms with each element, suggesting that the same factor binds to each of the sites (8,9). EMSA analysis using YY1 antiserum and a YY1 consensus oligonucleotide competitor demonstrated that YY1 binds the Ϫ145 site (Fig. 4, A and  B). To determine if YY1 also binds the other three BID-binding sites, EMSA was performed with heparin-fractionated K562 cell nuclear extract and YY1 antiserum (Fig. 5A). Each of the BID sites derived from the gp91 phox promoter (Ϫ90, Ϫ145 Core, Ϫ225 Core, and Ϫ355 Core) was used as a probe. The BID complex is formed upon addition of fractionated K562 nuclear extract with each of the four probes; Ϫ90 (lane 2), Ϫ145 Core (lane 5), Ϫ225 Core (lane 8), and Ϫ355 Core (lane 11). Importantly, the BID complex formed with each of the four probes is ablated upon addition of the YY1 antiserum (lanes 3, 6, 9, and 12).
Both the Ϫ90 and Ϫ225 bp BID-binding sites within the gp91 phox promoter also contain IFN-stimulated response element motifs that bind members of the IRF family (8,9). Because the Ϫ225 Core site binds IRF proteins with a high affinity, 2 the IRF complex makes it difficult to detect the BID complex of similar mobility (Fig. 5A, lane 8). Previous studies with the Ϫ90-bp element identified a mutation (BID-mut) that specifically ablates the binding site for BID (8). The BID-mut competitor was used to disrupt the IRF complex that forms with the Ϫ225 Core probe, clearly revealing the BID complex (Fig. 5B, lane 2). Addition of YY1 antiserum abolishes the BID complex (lane 3), while Ets-2 antiserum does not affect the complex (lane 4). Thus YY1 is a component of the BID complex which forms with four distinct sites within the gp91 phox promoter.
Alignment of a consensus YY1-binding site (29) with each of the YY1-binding sites within the gp91 phox promoter reveals strong similarity (Fig. 6A). The Ϫ90 bp and Ϫ225 bp elements exhibit a 7 of 9 bp match, while the Ϫ145 bp and Ϫ355 bp sites exhibit a 9 of 9 bp match. Interestingly, the Ϫ90 bp and Ϫ225 bp sites, which also contain IFN-stimulated response element motifs, do not contain the 5Ј-ATG-3Ј core motif that is invariant in the YY1 consensus binding sequence (29). Further analysis of the Ϫ450 to ϩ12 bp gp91 phox promoter revealed a fifth putative YY1-binding site in the Ϫ412-bp region. Alignment of this element with the YY1 consensus site demonstrates an 8 of 9 bp match (Fig. 6A). EMSA using an oligonucleotide to this region as a probe and heparin fractionated nuclear extract derived from K562 cells (Fig. 6B) demonstrates the presence of a complex (lane 2) that is immunoreactive with YY1 antiserum (lane 3), but is not disrupted by Ets-2 antibody (lane 4). This complex also exhibits similar mobility and sequence specificity as the BID complex formed with the other four sites (data not shown).
To our knowledge this is the first report of YY1 regulating a myeloid cell-specific gene. To determine if the level of YY1 protein is regulated during myeloid differentiation, a Western blot was performed using nuclear extracts isolated from differentiated and undifferentiated myeloid cell lines, as well as non-phagocytic cell lines (Fig. 7). Nuclear extracts derived from  7) were analyzed. The blot was probed with YY1 antiserum, and one immunoreactive band of approximately 68 kDa is detected in each of the extracts. Although the predicted size of YY1 is 44 kDa, it has previously been reported that YY1 exhibits a mobility of 68 kDa on SDS-polyacrylamide gel electrophoresis (30), most likely because of its highly charged amino terminus. YY1 is present in all cell types examined, consistent with previous reports of the ubiquitous expression of YY1. No reproducible difference was detected in the level of YY1 protein present in cells that are induced to differentiate into mature myeloid cells (lanes 2 and 4) versus those that are not (lane 1). A subtle increase in YY1 is apparent in the extract treated with IFN-␥ (lane 3). However, this is likely due to an increased amount of protein loaded, because this extract demonstrates a similar increase in Sp1 levels (data not shown). Western blots performed with a different YY1 antibody and independent preparations of nuclear extracts demonstrated no reproducible differences in YY1 levels in differentiated myeloid cell extracts (data not shown).
Binding of YY1 to the gp91 phox Promoter Increases in Terminally Differentiated Myeloid Cells-EMSA studies were performed using the Ϫ182 to Ϫ112 bp region of the gp91 phox promoter, which contains overlapping binding sites for CDP and BID. Addition of HEL nuclear extract to a limiting amount of probe results in a strong CDP complex (Fig. 8A, lane 1). Addition of an oligonucleotide containing a high affinity CDPbinding site (E36) disrupts binding of CDP to the probe, and a faster migrating complex (BID) becomes apparent (lane 2). Addition of YY1 antiserum disrupts the BID complex (lane 3), confirming the presence of YY1 within the BID complex. Additional studies were performed to examine whether binding of YY1 to the gp91 phox promoter increases during myeloid differentiation. A strong CDP complex (but no YY1 complex) is produced by nuclear extract isolated from immature myeloid cells (Fig. 8B, lane 1). In contrast, nuclear extract isolated from terminally differentiated myeloid cells fails to produce a CDP complex, and a faster migrating complex is detected binding to the probe (lane 2). The faster migrating complex is disrupted upon addition of YY1 antiserum (lane 3). Thus, YY1 binding to the gp91 phox promoter is increased in differentiated myeloid cell extract where CDP DNA binding activity is down-regulated.
YY1 Trans-activates a Minimal Promoter via the Ϫ145 Core Element of the gp91 phox Promoter-Our previous studies demonstrated that deletion or mutation of the BID-binding sites in the gp91 phox promoter results in decreased promoter activity, suggesting that YY1 is a transcriptional activator of the gp91 phox promoter (8,9). Transient co-transfection studies were performed to directly determine if YY1 trans-activates the gp91 phox promoter. A dimer of the wild type Ϫ145 Core BIDbinding site, or a dimer of the Ϫ145 mut oligonucleotide, was cloned upstream of the ␤-globin minimal promoter linked to a luciferase reporter gene. The luciferase constructs were transiently transfected into HEL cells with either the pCB6 ϩ parental vector or the pCB6-YY1 expression vector. Luciferase activity expressed by the wild type Ϫ145 Core construct is stimulated nearly 10-fold upon overexpression of YY1 (Fig. 9). The Ϫ145 mut construct, which lacks the Ϫ145-bp YY1-binding site, is stimulated 3-fold, similar to the stimulation observed with the promoterless vector (data not shown). Overexpression of YY1 was confirmed by Western blot analysis of transfected cell pellets (data not shown) and was similar to the levels illustrated in Fig. 4C. DISCUSSION Restriction of gp91 phox promoter activity to mature myeloid cells is partially a consequence of a tightly controlled balance between a transcriptional repressor and transcriptional activators. The transcriptional repressor CDP binds to multiple sites within the proximal gp91 phox promoter and excludes the binding of transcriptional activators (such as BID) to overlapping binding sites (7). CDP DNA binding activity is down-regulated during terminal myeloid cell differentiation, thus allowing activators to interact with the promoter. Previous studies demonstrated the importance of these regulated interactions. For example, constitutive overexpression of cloned CDP in a myeloid cell line prevents induction of gp91 phox upon terminal differentiation (31). Furthermore, ablation of BID-binding sites within the Ϫ450 to ϩ12 bp gp91 phox promoter abrogates the ability of this promoter fragment to respond to INF-␥ stimulation (9).
Binding of BID to the gp91 phox promoter increases when nuclear extract derived from differentiated myeloid cells is analyzed by EMSA (9). However, because CDP also interacts with each of these promoter elements, increased BID binding could be due either to an increase in BID DNA binding activity, or to the decrease in CDP DNA binding activity that occurs during myeloid differentiation, or a combination of both. These studies were conducted to directly assess this question. In addition, the identification of BID permits an assessment of the FIG. 8. Binding of YY1 to the gp91 phox promoter increases in terminally differentiated myeloid cells. A, EMSA analysis was performed as described under "Experimental Procedures" using 3500 cpm of the Ϫ182 to Ϫ112 bp probe and 5 g of HEL nuclear extract. Five ng of E36 oligonucleotide which contains a high affinity CDP-binding site (21), or 1 l of YY1 antiserum were added where indicated. Samples were resolved on a 3.5% nondenaturing polyacrylamide gel. Arrows denote the positions of CDP and YY1. B, EMSA analysis was performed as described under "Experimental Procedures" using 3500 cpm of Ϫ182 to Ϫ112 probe and 8 g of PLB-985 immature myeloid cell or PMA-treated PLB-985 (macrophage) cell nuclear extract. One l of YY1 antiserum was added where indicated. Samples were resolved on a native 3.5% polyacrylamide gel. Arrows denote the positions of CDP and YY1. hypothesis that four distinct regions within the gp91 phox promoter serve as binding sites for a common DNA-binding factor. Previously, this speculation was based on the similar mobility and binding site specificity exhibited in EMSA by each of the four identified BID complexes (8,9).
Screening of an expression library with the Ϫ145 bp BIDbinding site derived from the gp91 phox promoter resulted in the recovery of a cDNA that encodes YY1. EMSA studies using YY1 antiserum demonstrates that each of the BID complexes contains YY1. Furthermore, an additional YY1-binding site was identified at Ϫ412 bp of the gp91 phox promoter. Consistent with previous data demonstrating that ablation of BID-binding sites reduces gp91 phox promoter activity (9), transient transfection studies demonstrate that YY1 trans-activates a minimal promoter containing the Ϫ145 bp YY1-binding site derived from the gp91 phox promoter.
YY1 regulates multiple viral and cellular genes, although to our knowledge gp91 phox represents the first lineage-restricted target of YY1 within the myeloid cell lineage. YY1 functions as a repressor, activator, or initiator depending on cell type, binding site, and interactions with other factors (see Ref. 32 for review). YY1 has also been reported to associate with the nuclear matrix (33). How YY1 activates transcription is not well understood. Interaction of YY1 with other proteins can affect the function of YY1 or its binding to DNA. YY1 has been shown to interact with several proteins including p300 (34), c-Myc (35), myeloid nuclear differentiation antigen (MNDA) (36), Sp1 (37), and E1A (38). The presence of ElA converts YY1 from a repressor to an activator of the AAV P5 promoter (30), and the interaction of YY1 and E1A is likely mediated by p300 (34). Although myeloid nuclear differentiation antigen has been reported to bind to YY1 and stimulate DNA binding activity (36), myeloid nuclear differentiation antigen antiserum failed to disrupt the BID complex that forms with the Ϫ145 Core-binding site probe (data not shown). Additionally, the BID complex exhibits a similar mobility and intensity in HeLa cells versus myeloid cells ( Fig. 2A), thus myeloid nuclear differentiation antigen is not a component of the BID complex.
Direct examination of YY1 levels by Western blot analysis reveals that YY1 protein is not induced during phagocyte differentiation. However, this does not rule out the possibility of altered YY1 function during myeloid cell development, because YY1 activity may be post-translationally regulated by phosphorylation and/or ADP-ribosylation (39,40). Importantly, formation of the BID complex with the Ϫ145 Core site probe, which lacks a CDP-binding site, is not increased in nuclear extracts derived from mature myeloid cells. In contrast, binding of YY1 to a promoter element containing a CDP-binding site increases in terminally differentiated myeloid cells that lack CDP DNA binding activity. This is consistent with previous data demonstrating latent BID activity in HeLa cells following the specific disruption of the CDP complex (7). This supports a model in which YY1 activity does not change during myeloid differentiation, but rather the ability of YY1 to interact with the gp91 phox promoter is affected by the level of CDP DNA binding activity in the cell. The studies reported here thus reveal a novel regulatory mechanism for YY1, in which the access of the transcriptional activator YY1 to DNA-binding sites is controlled by modulation of a competing repressor DNA-binding protein (CDP). These results reinforce the conclusion that CDP is a critical regulator of the gp91 phox promoter.
In addition to gp91 phox , other genes that are specifically expressed in mature myeloid cells, such as the secondary granule genes lactoferrin and neutrophil collagenase, are also inhibited when CDP is overexpressed in differentiating myeloid cells (41,42). This indicates that CDP regulates a panel of myeloid cell-restricted genes. Interestingly, inspection of the CDP-binding site within the lactoferrin promoter (42) reveals the presence of two putative YY1-binding sites (each exhibit an 8 of 9 bp match with the YY1 consensus binding sequence), which are separated by four nucleotides (data not shown). It will be of interest to determine if physical competition between CDP and YY1 is a recurring motif within myeloid cell-restricted gene promoters, and whether YY1 and CDP compete for binding to other promoter sites for which YY1 functions as an activator.
CDP-binding sites have also been identified in many other promoters and enhancers for which competing transcriptional activators have not been identified (41,42,49,(53)(54)(55)(56)(57). In some cases CDP may not directly compete with activators for binding, and CDP-mediated transcriptional repression may be mediated via a carboxyl-terminal repressor domain (58). It was recently demonstrated that the CDP repression domain associates with histone deacetylase (43) thereby providing a possi- FIG. 9. YY1 transactivates a minimal promoter via the ؊145 Core-binding site derived from the gp91 phox promoter. HEL cells were transiently transfected as described under "Experimental Procedures" with 1 g of the Ϫ145 Core dimer/␤-globin TATA minimal promoter or the Ϫ145 mut dimer/␤-globin TATA construct and 5 g of either the CB6 ϩ parent vector (solid bars) or the CB6-YY1 expression construct (striped bars). Each sample also contained 250 ng of CMV-␤gal plasmid to provide an internal control for transfection efficiency. Data represent five independent experiments with two different plasmid preparations and are presented as mean Ϯ S.E. ble mechanistic explanation for CDP-mediated repression distinct from physical displacement of activators.
Although CDP-mediated exclusion of YY1 from the gp91 phox promoter is an important aspect of the regulation of this gene, modulation of CDP DNA binding activity is not sufficient to direct myeloid cell-restricted gene expression. CDP DNA binding activity is down-regulated more generally upon terminal differentiation and exit from the cell cycle, and occurs in tissues that do not express the gp91 phox gene, such as kidney cells and myotubes (59,60). In the absence of CDP DNA binding activity, activation of the gp91 phox promoter may additionally require physical interaction or functional cooperation between YY1 and other DNA-binding factors. Members of the IRF family, the CCAAT-binding factor CP1, and the hematopoietic cell-restricted ets factors Elf-1 and PU.1 all interact with the Ϫ102 to ϩ12 gp91 phox promoter region (4, 8, 10 -12). In this context it is of interest that YY1 can induce DNA bending (61). Hence, YY1 may contribute to the formation of an appropriate promoter architecture that permits efficient expression of the gp91 phox promoter. Additional studies will be necessary to understand the interactions that occur between ubiquitous and lineagerestricted transcription factors to direct myeloid cell-restricted activity of the gp91 phox promoter.