INTRODUCTION

Mature phagocytes are characterized by the ability to generate superoxide and other toxic free radicals via the respiratory burst (1-3). The catalytic unit of the respiratory burst oxidase is a membrane-associated b cytochrome with heavy and light chain protein subunits that are referred to as gp91^phox (4) and p22^phox (5). Congenital absence of either subunit leads to chronic granulomatous disease, a disorder of host defense (6). gp91^phox is expressed nearly exclusively in phagocytic cells that have progressed beyond the promyelocyte stage of myelopoiesis (7). Therefore, expression of gp91^phox is both lineage- and differentiation state-specific. Transcription of the CYBB gene, which encodes gp91^phox, is augmented in mature phagocytes by inflammatory mediators including interferon- (IFN-)¹ (8), tumor necrosis factor-, and lipopolysaccharide (9), although transcription is decreased by IFN- (8). Regulation of gp91^phox expression is therefore important in modulation of the inflammatory response.

Previous investigations indicated that gp91^phox expression in immature myeloid cells was repressed by the CCAAT displacement protein (CDP) (10). CDP, a homeodomain protein, was purified from HeLa cells by affinity to the 76 to 138 base pair (bp) sequence of the CYBB promoter and cloned from an endothelial cell cDNA library (11). Overexpression of CDP in the myeloid leukemia cell line HL-60 blunted PMA induction of gp91^phox expression (12). Recombinant CDP bound in vitro to the promoters of a number of developmentally regulated genes and was postulated to function as a differentiation state-specific transcriptional repressor (13, 14).

A positive regulatory element in the CYBB promoter was identified by analysis of two chronic granulomatous disease kindreds (15) with promoter mutations at 55 or 57 bp (16, 17). Either of these mutations disrupted in vitro binding of a specific protein complex, referred to as HAF1 (ematopoiesis-ssociated actor ) (15). 450 bp of proximal promoter sequence directed IFN--inducible reporter gene expression in the promyelocytic leukemia cell line PLB985 (18), and this inducible expression was abolished by introduction of the 55 bp chronic granulomatous disease promoter mutation (15). HAF1 binding was necessary, but not sufficient, for IFN--stimulated gp91^phox transcription since PLB985 transfectants with constructs containing 100 bp of proximal promoter sequence did not respond to IFN-. These results suggested that elements between 100 and 450 bp must be involved in IFN--induced transcription.

Further evaluation of the 100 to 450 bp promoter region revealed the presence of three IFN- response elements (19). Successive promoter truncations located the IFN- response elements between 100 and 210 bp, 210 and 310 bp, and 310 and 450 bp (19). Electrophoretic mobility shift assay (EMSA) demonstrated sequence-specific binding of protein complexes to probes representing 114 to 184 bp, 213 to 262 bp, and 331 to 381 bp (19). Binding of the complexes was increased by IFN-, retinoic acid, or PMA treatment of PLB985 cells; therefore, the complexes were referred to as BID1, BID2, and BID3, respectively, because inding ncreased during ifferentiation (19). Treatment of the epithelial cell line HeLa with IFN- did not increase BID1, BID2, or BID3 binding (19).

The BID2 complex bound to 224 to 236 bp of the CYBB promoter (19). Although this sequence was homologous to both the interferon-stimulated response element (ISRE) (20) and the common binding site consensus sequence for the interferon response factors IRF-1 and IRF-2 (21), the BID2 complex was not recognized by antibodies to Stat1, IRF-1, IRF-2, ICSBP, or interferon-stimulated gene factor 3  (ISGF3) (19). IRF-2 bound to this region of the promoter in EMSA using nuclear proteins from undifferentiated PLB985 cells and was distinct from BID2 binding (19). However, IRF-2 binding was obscured by BID2 complex binding in nuclear proteins of differentiated PLB985 cells (19). Immediately 3 to the BID2-binding site was a homeodomain-binding consensus sequence (ATTA). This suggested the presence of a second CDP-binding site adjacent to the BID2 site since CDP preferentially bound ATTA-containing sequences in binding site selection assays (22, 23).

Introduction of a mutation into the 450-bp proximal promoter sequence that disrupted in vitro binding of the BID2 complex attenuated IFN--induced expression of a linked reporter gene in PLB985 transfectants (19). Therefore, the BID2 complex might represent a previously unidentified protein(s) that bound to a positive regulatory element, interacted with IRF-2 and/or CDP, and contributed to lineage-specific gp91^phox expression. The identity of BID2 component proteins was sought by molecular cloning techniques.

We previously reported cloning a cDNA by screening a human leukocyte library for proteins that bound a DNA probe representing the 214 to 262 bp sequence of the CYBB promoter (24). Although this promoter sequence included the BID2-binding site, the predicted amino acid sequence of the cloned cDNA was not homologous to previously described interferon response factors. However, the cloned cDNA contained a sequence homologous to the S1-RM domain of ribosomal protein S1 (24). The S1-RM domain has been postulated to bind single-stranded RNA (25). The cloned S1-like protein had high affinity for binding double- and single-stranded DNAs, but low affinity for binding single-stranded RNA (24). Since the S1-like protein was of low abundance and immunoreactive S1-like protein was not found in ribosomal preparations, we concluded that the cloned cDNA was unlikely to represent human ribosomal S1 protein (24). The S1-like protein was referred to as TF1^phox and is further characterized in this study.

EXPERIMENTAL PROCEDURES

Plasmids

The 1.5-kb sequence clone (includes the entire coding sequence) was excised from gt11 by digestion with EcoRI and subcloned into the plasmid vector pUC19 as described (24). A 5-clone sequence (+1 to +1866 bp of the cDNA) was similarly isolated from  DNA and subcloned into pUC19. The 2.69-kb sequence (+1 to +2690 bp, the entire cDNA) was constructed by ligation of the 5-clone and the original 1.5-kb clone sequences at a unique HpaI site found in the sequences common to both clones. For in vitro translation, the TF1^phox sequences were subcloned into the pBluescript plasmid (CLONTECH, Palo Alto, CA).

For expression in Escherichia coli, the 1.5-kb clone sequence was subcloned into the EcoRI site of the plasmid pGEX1 (Pharmacia Biotech Inc.) (26) in the sense and antisense orientations (TF1^phox/pGEX1 and TF1^phox(R)/pGEX1, respectively). For expression in mammalian cells, the 1.5-kb clone sequence was subcloned into the EcoRI site of the plasmid pSR (obtained from T. Gabig, Indiana University, Indianapolis, IN) (27) or pRc(CMV) (Invitrogen, Sorrento, CA). TF1^phox cDNA sequences were epitope-tagged at the C terminus of the coding sequence with the influenza hemagglutinin epitope (HAI; amino acid sequence IDYPYDVPDYAE (28)). Complementary synthetic oligonucleotides encoding the epitope tag were annealed, phosphorylated with polynucleotide kinase (Boehringer Mannheim), and ligated in frame into an EcoRV site found 20 amino acids from the end of the TF1^phox coding sequence. Correct orientation of the cDNA was confirmed for all constructs by dideoxy sequencing (Sequenase system, Amersham Corp.).

Concatenated CYBB promoter sequences were subcloned into the minimal promoter/reporter vector p-TATA/CAT (obtained from Dr. A. Kraft, University of Alabama at Birmingham) (29). Synthetic oligonucleotides with the sense and antisense 194 to 244 bp gp91^phox promoter sequences were annealed, phosphorylated with polynucleotide kinase, and ligated with T4 DNA ligase (Boehringer Mannheim). The concatenated oligonucleotides were subcloned into the p-TATA/CAT vector. Clones that were confirmed by dideoxy sequencing to contain four correct copies of the 194 to 244 bp promoter sequence in the sense or antisense orientation were selected.

Duplex synthetic oligonucleotides with the 194 to 244 bp, 213 to 262 bp, or 76 to 138 bp promoter sequence were subcloned into the plasmid vector pUC19, and correct sequence was confirmed by dideoxy sequencing.

Human gp91^phox cDNA sequence was obtained from Dr. Peter Newburger (University of Massachusetts, Worcester, MA). Human p47^phox (30) and p67^phox (31) cDNA sequences were obtained from Dr. Thomas Leto (National Institutes of Health, Bethesda, MD). -Actin cDNA sequence was obtained from Dr. D. Skalnik (Indiana University, Indianapolis, IN). Human CD18 cDNA (32) was obtained from Dr. Kent Robertson (Indiana University).

Oligonucleotides

Oligonucleotides were synthesized either by the core facility at the Herman B. Wells Center for Pediatric Research or in the laboratory of Dr. Donald Miller at the University of Alabama at Birmingham. CYBB promoter sequence (10) oligonucleotides used were as follows: 194 to 244 bp gp91^phox promoter, 5-agaaattggtttcattttccactatgtttaattgtgactggatcatta-3; 213 to 262 bp gp91^phox promoter, 5-gttatttatctcttagttgtagaaattggtttcattttccactatgttta-3; 215 to 240 bp gp91^phox promoter, 5-gaaattggtttcattttccactatgt-3; gp91^phox BID2-binding site mutant (194 to 244 bp), 5-agaaattggtcctgccttccactatgtttaattgtgactggatcatta-3; gp91^phox BID1-binding site (113 to 182 bp), 5-tttgtagttgttgaggtttaaagatttaagtttgttatggatgcaagcttttcagttgaccaatgattat-3; gp91^phox BID1-binding site mutant (113 to 182 bp), 5-tttgtagttgttgaagctcaaggactcagacctgttatggatacaagcccccaagtcgaccaatgattat-3; 76 to 137 bp gp91^phox CDP-binding site, 5-gcttttcagttgaccaatgattattagccaatttctgataaaagaaaaggaaaccgattgc-3; and also including the 93 to 165 bp gp91^phox CDP-binding site, 5-aaagatttaagtttgttatggatgcaagcttttcagttgaccaatgattattagccaatttctgataaaaga-3. Other oligonucleotide sequences included the following: CCAAT box from the MHC class II gene (15), 5-gtctgaaacatttttctgattggttaaaagttgagtgct-3; GASGBP (-interferon ctivation equence of the uanylate-inding rotein) (33), 5-agtttcatattactctaaatc-3; PU.1-binding site from the MHC class II gene (15), 5-tgaaataacctctgaaagaggaacttggttaggta-3; and binding site for Ets-1 (15), 5-ttccagaggatgtggcttctgcgggagagctt-3.

Cell Culture and Stable Transfections

All cell lines were of human origin. The promyelocytic cell line HL-60 (34), the chronic myelogenous leukemia K562 (35), the epithelial carcinoma line HeLa (36), and the T-cell leukemia line MoLT-4 (37) were obtained from American Type Culture Collection (Rockville, MD). The B-cell line CESS (38) was provided by Yu Chung Yang (Indiana University). The myelomonoblastic line PLB985 (18) was provided by Thomas Rado (University of Alabama at Birmingham). The promyelocytic leukemia cell line U937 (39) was obtained from Dr. A. Kraft (University of Alabama at Birmingham).

Cell lines were maintained and differentiated as described previously (15). HeLa cells were treated for various times with human recombinant interferon- at 50 units/ml. Stably transfected PLB985 cell lines carrying the neomycin resistance gene were supplemented with 1.0 mg of Geneticin/ml (G418, Life Technologies, Inc.).

PLB985, U937, or HeLa cells were transfected by electroporation in phosphate-buffered saline at 960 microfarads and 0.24 V using a Gene Pulser (Bio-Rad). Stable transfection of vector or insert sequences was demonstrated by Southern blotting as described previously (15).

Library Screening and cDNA Sequencing

A human leukocyte gt11 library (CLONTECH) was screened with the 1.5-kb TF1^phox clone sequence by DNA-DNA hybridization as described (40). The TF1^phox sequence was labeled with ³²P by the random primer method (41) to a specific activity of 1 × 10⁸ cpm/µg. Positive plaques were purified through six iterations as described (24).

Expression of TF1^phox as a GST Fusion Protein, Generation of TF1^phox Antiserum, and Western Blotting

E. coli JM109 cells were transformed with pGEX1, TF1^phox/pGEX1, or TF1^phox(R)/pGEX1, and positive clones were selected. Transformed E. coli cells were grown and stimulated with isopropyl-1-thio--D-galactopyranoside at log phase as described (26). Fusion proteins were affinity-purified by binding to glutathione-Sepharose (Pharmacia Biotech Inc.) as described by the manufacturer. Purified fusion protein was used to generate rabbit antiserum by Hazelton Research Labs (Denver, PA). Purified fusion proteins used in EMSA were dialyzed into 20 mM Hepes (pH 7.9), 0.1 M KCl, 0.2 mM EDTA, and 20% glycerol. Western blotting and detection were performed using the ECL system (Amersham Corp.) according to the manufacturer's instructions.

In Vitro Translation of Proteins

In vitro transcription and translation of TF1^phox sequences cloned into pBluescript were performed using the in vitro transcription and rabbit reticulocyte lysate in vitro translation systems (Promega) according to manufacturer's instructions.

Electrophoretic Mobility Shift Assays

Nuclear extract proteins were prepared from tissue culture cells growing at log phase by the method of Dignam et al. (43) with protease inhibitors as described (15). Protein assays were performed by the method of Lowry et al. (44).

Oligonucleotides probes were prepared from plasmid constructs by restriction digestion with EcoRI and XbaI and filled in by Klenow fragment with a nucleotide mixture containing unlabeled dATP, dGTP, and dTTP (Boehringer Mannheim) and [-³²P]dCTP (Amersham Corp.). The specific activity of the probes was ~2 × 10⁶ cpm/pmol DNA.

EMSA and antibody supershift assays were performed as described (15). Anti-CDP serum was obtained from Dr. Elis Neufeld (Boston Children's Hospital, Boston). Anti-IRF-2 and anti-Stat91/84 antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-IRF-1 antiserum was obtained from Dr. Richard Pine (Research Institute, New York) (45). Anti-ICSBP serum was obtained from Dr. Keiko Ozato (National Institutes of Health) (46). Anti-HAI tag antibody was obtained from Boehringer Mannheim.

RNA Preparation and Analysis

Total cellular RNA was isolated from log-phase cells by the single step method as described (47). The polyadenylated RNA fraction was isolated by two rounds of affinity chromatography with oligo(dT)-cellulose (Pharmacia Biotech Inc.) as described (48). RNA was electrophoresed on denaturing formaldehyde gels and transferred to Magnagraph membranes (Micron Separations, Westborough, MA) according to the manufacturer's instructions. Blots were hybridized with random primer-labeled DNA probes and stripped for reprobing according to the manufacturer's instructions.

Polymerase chain reaction-5-rapid amplification of cDNA ends was performed as described (49). Poly(A) RNA isolated from PLB985 cells stimulated for 3 h with PMA (2 µg) was used as a template. Two nested primers complementary to sequences within the first 100 bp of the coding sequence of TF1^phox were used. Products were analyzed by Southern blotting and hybridization to TF1^phox sequences.

Transient Transfection and Reporter Gene Assays

PLB985 cells were transfected by electroporation as described above. PLB985 stable transfectants with either control pSR vector or TF1^phox/pSR were transfected with 100 µg of p-TATA/CAT or p-cybbTATA/CAT and 30 µg of pCMV/-galactosidase (as an internal control for transfection efficiency). Transfectants were incubated for 72 h at 37 °C in 5% CO₂ in 10 ml of RPMI 1640 medium, 10% fetal calf serum, 1% penicillin/streptomycin, and 1 mg/ml Geneticin. Cell extracts were made by hypotonic lysis as described (50) in the presence of 2 mM phenylmethylsulfonyl fluoride. -Galactosidase assays were performed as described (51). Chloramphenicol acetyltransferase assays (CAT) were performed by the extraction method using [³H]chloramphenicol (Amersham Corp.) as described (50).

U937 cells were transfected with 70 µg of p-TATA/CAT or p-cybbTATA/CAT, 30 µg of pRc or TF1^phox/pRc, and 30 µg of pCMV/-galactosidase as described above. Cell extracts were analyzed as described above after 24 h of incubation at 37 °C in 5% CO₂ in 10 ml of Dulbecco's modified Eagle's medium, 10% fetal calf serum, and 1% penicillin/streptomycin.

RESULTS

IFN- Induces Changes in the Protein Complexes That Bind to the CYBB Promoter

To identify DNA-binding proteins that might be involved in gp91^phox transcription, EMSA were performed with a DNA probe representing the 194 to 244 bp sequence of the promoter (Fig. 1). This sequence contained the BID2 complex-binding site (19), a sequence similar to the ISRE sequence (20) and the IRF-1/IRF-2 protein recognition site (224 to 236 bp) (21). This sequence also contained a homeodomain protein recognition core, a potential CDP-binding site (212 to 215 bp) (Table I) (10). To identify DNA-binding proteins that might be myeloid lineage-specific, EMSA with nuclear proteins from the myeloid cell line PLB985 before and after treatment with IFN- were compared with EMSA with nuclear proteins from the IFN--responsive epithelial cell line HeLa.

Sequence of the CYBB promoter (+12 to 267 bp).

Table I. Comparison of gp91phox promoter elements with two known interferon response elements Interferon response element sequences are indicated in boldface; core consensus sequences for homeodomain protein binding are indicated in italics; and CCAAT boxes are underlined. Element Sequence BID1-binding site (94 to 164) aaagatttaagtttgttatggatgcaagcttttcagttgagattattagttctgataaaa BID2-binding site (194 to 262)  gttatttatctcttagttgtagaaattggtttcattttatgtttaattgtgactggatcatt ISRE (consensus)                             agtttcnnttcc                             g IRF-1/IRF-2 (consensus)                             actttcacttt(t)c                             gg gg BID2-binding site mutant (194 to 262)  gttatttatctcttagttgtagaaattggtcctgcctttatgtttaattgtgactggatcatt

EMSA with the 194 to 244 bp probe and nuclear proteins from PLB985 cells or HeLa cells demonstrated binding of a dominant low mobility complex, similar to the mobility previously described for CDP (10, 11). However, the PLB985 complex was a doublet, and the HeLa complex migrated as a single band. These findings were consistently observed with three sets of protein extracts from each of the two cell lines. Homologous double-stranded oligonucleotide efficiently competed for binding of the low mobility complex to the 194 to 244 bp probe, as did competitor oligonucleotides containing the 76 to 137 bp and 93 to 165 bp CYBB promoter sequences (Fig. 1). The latter two sequences included a previously described CDP-binding site (10, 11). There was no competition for the low mobility complex with several unrelated oligonucleotides (Fig. 2A and data not shown).

In EMSA with nuclear proteins from HeLa cells, the low mobility complex was disrupted by incubation with anti-CDP serum. However, incubation of PLB985 nuclear proteins with anti-CDP serum did not result in disruption of the low mobility complex doublet (Fig. 2B). For comparison, control EMSA were performed using the 76 to 137 bp promoter region. As described, in EMSA using nuclear proteins from both HeLa and PLB985 cells, low mobility protein complexes bound to this probe (10, 11). However, the complex generated with PLB985 nuclear extract was slightly faster in mobility than the HeLa complex. These protein complexes were demonstrated to represent specific protein binding (data not shown; consistent with previous studies (10, 11)). Anti-CDP serum disrupted binding of the slow mobility complex to the 76 to 137 bp probe in assays with HeLa nuclear proteins (as described previously (11)); however, in EMSA using PLB985 nuclear extract proteins, the dominant slow mobility complex was not disrupted by anti-CDP serum (Fig. 2C).

Treatment of PLB985 and HeLa cells with IFN- resulted in the disappearance of the dominant low mobility complex from the nuclear proteins of both cell types at 24 h (Fig. 3). In PLB985 cells (Fig. 3A), but not HeLa cells (Fig. 3B), IFN- treatment also resulted in binding of a novel protein complex of higher mobility. This higher mobility complex is the BID2 complex that we have previously characterized (19). We previously reported that the BID2 complex represented sequence-specific binding and that BID2 binding increased upon differentiation of PLB985 cells with IFN-, PMA, or retinoic acid (19). The BID2 complex appeared after 24 h of IFN- treatment (Fig. 3A). These results were consistently observed with at least three nuclear extract preparations for each cell type, under each condition.

IFN- treatment of PLB985 and HeLa cells results in changes of in vitro nuclear protein binding to the promoter of the gene encoding gp91^phox.

As we previously described, EMSA demonstrated binding of two additional protein complexes, of similar mobility to the BID2 complex, to the region of the promoter containing the BID2-binding site (seen in Fig. 3) (19). We previously demonstrated that the complex of slightly slower mobility than BID2 (referred to as 2a) did not change in response to IFN- or PMA; did not interact with antibodies to IRF-1, IRF-2, ICSBP, ISGF3, or Stat91/86; and represented nonsequence-specific protein binding to the probe (19). We also demonstrated that the complex of slightly faster mobility than BID2 had sequence-specific binding (19) and was disrupted by an antibody to IRF-2, but not by antibodies to IRF-1 (45), ICSBP (46), ISGF3, or Stat91/86 (19). IRF2 binding became obscured by the BID2 complex in PLB985 nuclear proteins after treatment with IFN- or PMA (19).

A human leukocyte cDNA expression library was screened for proteins binding to the 214 to 262 bp sequence of the promoter of the gene coding for gp91^phox, and a 1.5-kb clone was identified (24). Since there was an open reading frame to the 5-end of this clone, the library was rescreened under high stringency by DNA-DNA hybridization (40) using the original clone as a random primer-labeled probe. Clones with overlapping sequences were obtained that identified a putative full-length cDNA of 2.69 kb. The 2.69-kb sequence coded for a 1.186-bp open reading frame with a 5-untranslated region of 1.242 kb and a 3-untranslated region of 253 bp. This open reading frame was entirely contained within the original 1.5-kb clone and coded for a protein with a predicted molecular mass of 41.5 kDa. Only one clone obtained by rescreening contained sequence that was 5 to the original 1.5-kb clone, and all of the clones but one had the same 3-end as the original 1.5-kb clone.

The 1.5-kb clone had a Kozak consensus sequence (52) immediately upstream of the first methionine codon, and additional methionine codons farther 5 in the sequence of the 2.69-kb clone all had an in-frame stop codon before reaching the original 1.5-kb clone sequence. As previously reported, the coding sequence of the 1.5-kb clone had a predicted amino acid sequence with sequence similarity to S1-RM (24). Additional analysis of the amino acid sequence determined that there was a potential nuclear localization domain (KRVK) at amino acids 297-300 (53). A GeneBank^TM search revealed that no described ribosomal S1 protein included a nuclear localization domain. The predicted amino acid sequence contained no domains traditionally associated with transcription factors. Additionally, there was no homology to either the Stat or IRF families of IFN response factors, although the predicted amino acid sequence contained an excess of charged residues, similar to IRF-like proteins (54). The protein was referred to as TF1^phox.

The TF1^phox transcript size was determined by Northern blotting and polymerase chain reaction-5-rapid amplification of cDNA ends. The 1.5-kb sequence probe hybridized to an mRNA of 2.7 kb on PLB985 poly(A) Northern blots (Fig. 4A), as did a probe containing only sequence from the 2.69-kb clone not present in the 1.5-kb clone (nonoverlapping sequence). When adjusted for differences in RNA loading (by hybridization to a -actin probe), there was no significant change in TF1^phox mRNA levels in response to IFN- (Fig. 4A) or PMA (data not shown). Polymerase chain reaction-5-rapid amplification of cDNA ends was performed on poly(A) RNA isolated from PLB985 cells. A single sharp band that indicated a 5-untranslated region of 1.2 kb was detected using nested primers in the 5-region of the coding sequence, consistent with the 5-untranslated region of the putative full-length clone (data not shown).

The size of in vitro translated TF1^phox was compared with the calculated molecular mass based upon the predicted amino acid sequence. In vitro translated protein was generated using the 1.5-kb clone as a template in rabbit reticulocyte lysate. SDS-polyacrylamide gel electrophoresis of the in vitro translation product demonstrated production of a protein of ~49 kDa, consistent with the predicted molecular mass. The in vitro translation product using the +1 to +1866 bp cDNA sequence as a template was ~20 kDa, consistent with the predicted molecular mass if the first methionine in the 1.5-kb clone was indeed the site of translation initiation (Fig. 4B). In vitro translation of the full-length 2.69-kb clone also resulted in production of a 49-kDa protein.

DNA-binding site specificity of TF1^phox was investigated with recombinant TF1^phox expressed as a glutathione S-transferase fusion protein and purified by affinity chromatography (TF1^phox/GST). EMSA with TF1^phox/GST and the 194 to 244 bp promoter sequence probe demonstrated binding of a protein complex (Fig. 5). The complex was not demonstrated with purified GST or fusion protein generated by antisense TF1^phox/GST (Fig. 5A). Binding of TF1^phox/GST to the BID2-binding site probe was not competed for by unrelated DNA sequences, including GASGBP (33) (Fig. 5B). A mutant 194 to 244 bp sequence unable to bind the BID2 complex due to disruption of the ISRE-like element (19) also was unable to compete for TF1^phox binding. The overlapping 214 to 262 bp sequence, originally used to identify TF1^phox by expression screening, competed efficiently for TF1^phox binding. A competitor oligonucleotide containing the BID1-binding site (19) also competed efficiently for TF1^phox binding, consistent with the cross-reactive binding specificities of BID1 and BID2 (19). Identical results were found in EMSA using the overlapping 214 to 264 bp promoter sequence as a probe (data not shown). In vitro translated TF1^phox was not used in these experiments because it bound poorly in EMSA using CYBB promoter sequence probes.

To investigate whether TF1^phox was a component of the BID2 protein complex, a polyclonal rabbit antiserum was generated. EMSA were performed with the 194 to 244 bp probe and nuclear proteins from IFN- (200 units/ml for 48 h)-treated PLB985 cells in the presence of anti-TF1^phox serum or preimmune serum. Incubation of either IFN-- or PMA-treated PLB985 nuclear proteins with anti-TF1^phox serum disrupted binding of the BID2 complex to the 194 to 244 bp probe (Fig. 6A). The presence of preimmune serum in binding reactions had no effect upon the BID2 complex under the same incubation conditions. Binding of IRF-2 increased in the presence of TF1^phox antiserum, coincident with disruption of TF1^phox binding (Fig. 6A). This result suggested that disruption of BID2 complex binding by TF1^phox antiserum permitted increased binding of a factor usually excluded from the site by the presence of the TF1^phox protein (as part of the BID2 complex) in differentiated myeloid cells. Identical results were obtained with nuclear proteins treated with PMA (0.1 µM for 48 h) (data not shown). A protein complex immunoreactive with TF1^phox antiserum bound to the probe in EMSA with the nuclear proteins of a variety of cell lines (Fig. 6B). Since the probe used in Fig. 6 (A and B) was generated in the same labeling reaction, panels A and B demonstrate increased intensity of the TF1^phox complex (the BID2 band) in PLB985 cells upon treatment of the cells with IFN- for 48 h (compare lane 1 from A with lane 1 from B).

Both gp91^phox gene expression and binding of the BID2 complex to the promoter region of the gene encoding gp91^phox were increased by differentiation of PLB985 cells with IFN-; however, treatment with IFN- had been demonstrated to decrease gp91^phox expression in myeloid cell lines. Therefore, the effect of these two agents upon TF1^phox protein abundance was investigated by Western blotting. Western blots of total cellular lysate proteins from 1 × 10⁷ PLB985 cells before and after 48 h of treatment with IFN demonstrated an immunoreactive band of ~49 kDa that was not altered in abundance by treatment with IFN- (Fig. 6C). However, a decrease in the 49-kDa TF1^phox immunoreactive protein was found in the total cellular proteins of PLB985 cells treated for 48 h with IFN-.

To investigate the effect of the TF1^phox protein upon the promoter region of the CYBB gene, the TF1^phox coding sequence was subcloned into the mammalian expression vector pSR (27) (TF1^phox/pSR), and stable transfectant pools of PLB985 cells were selected. Stable transfectants of PLB985 cells with empty pSR vector were also selected to serve as a control pool. Expression of TF1^phox mRNA was verified by the presence of a 1.5-kb mRNA that hybridized with the 1.5-kb clone probe in Northern blots of total cellular RNA from TF1^phox/pSR, but not control PLB985 transfectants (data not shown). Western blots of total cell lysates demonstrated increased TF1^phox immunoreactive protein in the PLB985 TF1^phox/pSR transfectants in comparison with the control pSR transfectants (Fig. 7A).

Nuclear proteins were isolated from TF1^phox/pSR and control pSR transfectants for use in EMSA of the 194 to 244 bp promoter sequence (Fig. 7B). Nuclear proteins were extracted from PLB985 transfectants that had not been treated with IFN- or PMA. EMSA with nuclear proteins from PLB985 control pSR transfectants demonstrated a dominant slow mobility complex identical to the complex found with nuclear extract proteins from untreated PLB985 cells. However, EMSA using nuclear extract proteins from PLB985 TF1^phox/pSR transfectants demonstrated no low mobility complex, but showed an increase in the BID2 complex (an apparent decrease in the intensity of the 2a complex in lane 4 relative to lane 3 is due to increased protein loading in lane 4). Therefore, overexpression of TF1^phox in PLB985 cells had the same effect upon in vitro protein binding as treatment of the PLB985 cells with IFN- or PMA: decreased binding of the low mobility complex and increased binding of the BID2 complex. These results were obtained with nuclear extract proteins prepared from at least three independent pools of PLB985 TF1^phox/pSR or pSR stable transfectants.

To determine whether recombinant TF1^phox protein was present in the BID2 complex in EMSA using nuclear proteins from PLB985 TF1^phox/pSR transfectants, HAI-tagged TF1^phox was overexpressed in PLB985 cells in the mammalian expression vector pSR. EMSA of the 194 to 244 bp promoter sequence was performed using nuclear proteins from PLB985 TF1^phox (HAI-tagged)/pSR transfectants. The BID2 complex generated in EMSA using nuclear proteins from epitope-tagged TF1^phox-overexpressing PLB985 cells was supershifted by incubation with anti-HAI antibody, but not by an irrelevant monoclonal antibody (Fig. 7C).

To determine the effect of overexpression of TF1^phox upon endogenous gp91^phox mRNA levels, Northern blotting of RNAs from TF1^phox/pSR and control pSR of PLB985 cells was performed. Each experiment was repeated three times for each of three independent pools of PLB985 cells stably transfected with TF1^phox/pSR or control pSR. Northern blots of total cellular RNA were sequentially probed for hybridization with various DNA probes (Fig. 8). Increased gp91^phox message was consistently detected in the TF1^phox/pSR transfectants in comparison with control pSR transfectants. To investigate whether TF1^phox overexpression was specifically affecting gp91^phox expression or was inducing differentiation and therefore causing increased gp91^phox message indirectly, the blots were probed for p47^phox (30), p67^phox (31), and CD18 (32) messages. All of these messages increase during terminal myeloid differentiation, similar to gp91^phox. The abundance of mRNAs for p67^phox, p47^phox, and CD18 was not increased in PLB985 TF1^phox/pSR transfectants relative to control pSR transfectants. Blots were also probed for -actin to control for RNA loading. Stable transfectant pools of HeLa cells with TF1^phox/pSR or control pSR also were generated. No gp91^phox message was detected in Northern blots of RNA isolated from either of these lines (data not shown).

To investigate whether overexpressed TF1^phox was acting upon the promoter of the gene coding for gp91^phox, the concatenated duplex 194 to 244 bp promoter sequence was subcloned into a vector with a linked minimal promoter and reporter gene (p-TATA/CAT) (29). Plasmids were constructed with four copies of the 194 to 244 bp sequence in the sense orientation. PLB985 stable TF1^phox/pSR and control pSR pools were transiently transfected with either control p-TATA/CAT vector or promoter sequence containing vector (p-cybbTATA/CAT), and cell extracts were analyzed after 72 h for CAT activity. CAT activity was normalized for -galactosidase activity from a cotransfected plasmid (pCMV/-galactosidase) to control for transfection efficiency. Two independent stable transfectant pools of both TF1^phox/pSR and control pSR were used in these experiments.

Cell extracts from PLB985 pSR transfectants with p-TATA/CAT had 58.1 ± 29.5% (n = 7) more CAT activity than transfectants with p-cybbTATA/CAT, consistent with the action of a repressor upon the promoter element of the gene coding for gp91^phox in undifferentiated PLB985 cells (Fig. 9A). In contrast, cell extracts from PLB985 TF1^phox/pSR transfectants with p-cybbTATA/CAT had 120.3 ± 44.0% (n = 7) more CAT activity than p-TATA/CAT transfectants, consistent with the action of a transcriptional activator on the CYBB promoter element in the TF1^phox-overexpressing cells. The absolute values of CAT activity for the p-TATA/CAT construct were not significantly different between the TF1^phox/pSR and control pSR stable transfectant pools. These results were unaffected by the orientation of the promoter sequence relative to the minimal promoter.

These constructs were also analyzed by cotransfection into the human promyelocytic leukemia cell line U937. U937 cells are more committed than PLB985 cells and are able to differentiate to monocyte/macrophages with PMA or IFN- (39). U937 cells were cotransfected with either p-TATA/CAT or p-cybbTATA/CAT and either control pRc mammalian expression vector or TF1^phox/pRc. Cell extracts were analyzed 24 h after transfection, and CAT activity was normalized for -galactosidase activity as described above. Cell extracts from transfectants with p-TATA/CAT had 12.6 ± 7.8% (n = 4) more CAT activity that transfectants with p-cybbTATA/CAT when cotransfected with control pRc, consistent with the action of a repressor upon the promoter element (Fig. 9B). Therefore, less repression was demonstrated in U937 cells than in PLB985 transfectants. However, transfectants with 30 µg of TF1^phox expression construct demonstrated 142.8 ± 23.8% (n = 4) more CAT activity with p-cybbTATA/CAT than with p-TATA/CAT, and transfectants with 15 µg of TF1^phox expression construct (with 15 µg of pRc plasmid, to control for total DNA content) demonstrated 55.6 ± 21.4% more CAT activity with p-cybbTATA/CAT than with p-TATA/CAT, consistent with the dose-dependent action of a transcriptional activator on the CYBB promoter element.

These results were sustained when the data were analyzed in terms of absolute CAT activity. If the absolute CAT activity of the U937 cells transfected with p-TATA/CAT and pRc (30 µg) was normalized to 1.0, the CAT activity of U937 cells transfected with p-TATA/CAT and TF1^phox/pRc (30 µg) was 0.995 ± 0.034. Therefore, no significant effect of TF1^phox was demonstrated on the expression vector. Using the same approach, the CAT activity of U937 transfectants with p-cybbTATA/CAT and pRc (30 µg) was 0.935 ± 0.0.059, that with p-cybbTATA/CAT and TF1^phox/pRc (30 µg) was 2.843 ± 0.183, and that with p-cybbTATA/CAT and TF1^phox/pRc (15 µg) was 1.422 ± 0.115. These results were consistent with those obtained for PLB985 transfectants.

DISCUSSION

During terminal myelopoiesis, a number of phagocyte-specific genes become transcriptionally active. Treatment of leukemia cell lines with IFN- or PMA resulted in monocytoid differentiation over a 48-h time period, accompanied by increased expression of gp91^phox, which was maximal at 48 h (8). BID2 complex binding to the CYBB promoter was temporally associated with transcription of the gene encoding gp91^phox (19), suggesting that BID2 component proteins might be of interest in understanding myeloid gene regulation. We cloned a cDNA by affinity of the expressed protein for the BID2-binding site sequence (24). The predicted amino acid sequence of the protein, referred to as TF1^phox, included an RNA-binding domain and a nuclear localization domain, suggesting that it was a nuclear protein.

The presence of an RNA-binding site in a DNA-binding protein was not without precedent. Y box-binding proteins, a family of highly conserved RNA-binding proteins (55, 56), bind to the MHC class II gene promoter (57) and repress transcription (58). Myef-2, identified by screening a cDNA library for proteins binding the myelin basic protein gene promoter, had homology to the RNA recognition motif of heterogeneous ribonucleoprotein M4 and repressed transcription of the myelin basic protein gene (59). Characterization of TF1^phox was of potential interest in further clarification of the role of dual function DNA/RNA-binding proteins in gene regulation.

Two independent binding assays demonstrated that TF1^phox specifically recognized the BID2-binding site sequence. TF1^phox, expressed in gt11 and immobilized on nitrocellulose filters, bound to the 214 to 262 bp promoter sequence of the gene encoding gp91^phox. TF1^phox binding was competed for by excess homologous (but not heterologous) oligonucleotide or homologous oligonucleotide with a specific mutation in the BID2-binding site.² TF1^phox expressed as a GST fusion protein also had sequence-specific DNA binding in EMSA. Therefore, TF1^phox might be relevant to gp91^phox expression since TF1^phox/CYBB promoter binding was not the result of nonspecific DNA-protein interaction.

Although TF1^phox expressed as a GST fusion protein bound DNA probes, in vitro translated TF1^phox bound the DNA probes poorly. This difference may be due to a characteristic of the reticulocyte lysate, such as the presence of an inhibitor of TF1^phox binding or inappropriate post-translational modification of TF1^phox. Or, the difference may be due to a characteristic of the GST fusion protein, such as improved DNA binding by the TF1^phox/GST protein due to dimerization of the GST regions. The latter explanation suggests a possible partner protein to facilitate TF1^phox DNA binding.

Western blots demonstrated TF1^phox immunoreactive protein in human cells, and anti-TF1^phox serum disrupted BID2 complex binding to the CYBB promoter. However, confirmation of identity of the endogenous BID2 nuclear protein to cloned TF1^phox will require purification of BID2 to homogeneity and peptide sequencing. Although IFN- or PMA treatment of PLB985 cells increased in vitro DNA binding of TF1^phox, neither of these treatments increased TF1^phox mRNA or total cellular TF1^phox immunoreactive protein. Therefore, other mechanisms, such as post-translational modifications or protein-protein interactions, may be involved in regulation of TF1^phox binding.

Overexpression of recombinant TF1^phox in PLB985 cells resulted in increased abundance of gp91^phox mRNA, but did not increase the abundance of several other messages indicative of differentiation in leukemia cells. This result suggested that overexpression of TF1^phox did not stimulate differentiation, but acted more specifically upon gp91^phox mRNA levels.

To demonstrate the effect of overexpression of TF1^phox, artificial promoter constructs were employed. Inclusion of the BID2-binding site sequence in the reporter constructs resulted in repression of minimal promoter function in myeloid cell lines. This suggested that a repressor was acting on this promoter sequence in undifferentiated PLB985 or U937 cells, consistent with the level of expression of gp91^phox. The lesser degree of repression in U937 cells may reflect greater commitment to differentiation in this line. Overexpression of TF1^phox relieved the repression conferred by the CYBB promoter sequence. This suggested that a transcriptional activator was acting on the promoter sequence in TF1^phox-overexpressing myeloid cells. However, it was possible that TF1^phox was unrelated to gp91^phox expression in native myeloid cells, but artificially blocked repressor binding and therefore permitted binding of genuine transcriptional activation factor(s) in the TF1^phox transfectants.

Our investigations suggested the possible action of two repressor proteins on the 194 to 244 bp promoter sequence in undifferentiated myeloid cells. Other investigators had demonstrated that CDP repressed gp91^phox expression (10, 11). Although a slow mobility protein complex that bound the 76 to 137 bp and