Cooperation of STAT-1 and IRF-1 in Interferon- (cid:1) -induced Transcription of the gp91 phox Gene*

Interferon (IFN)- (cid:1) induces the expression of the gp91 phox gene both during myeloid differentiation and also in mature phagocytes through several cis -elements and their binding proteins. To find new cis -elements for this induction, transient expression assays were performed using a reporter gene driven by serially truncated gp91 phox promoters in U937 cells. The results suggest that a critical cis -element for induction exists in the region from bp (cid:2) 115 to (cid:2) 96 of the promoter. Site-di-rected mutagenesis showed that a (cid:1) -activated sequence (GAS) element at bp (cid:2) 100 ( (cid:2) 100GAS) of the gp91 phox promoter plays a pivotal role for the IFN- (cid:1) -dependent activity of the bp (cid:2) 115 to (cid:3) 12 region of the gp91 phox promoter. Electrophoretic mobility shift assays using several GAS competitors and specific antibodies indicated that phosphorylated STAT-1 (cid:4) specifically binds to the (cid:2) 100GAS. Site-directed mutagenesis showed that an interferon-stimulated response element (ISRE) at bp (cid:2) 88 ( (cid:2) 88ISRE) mediates the induction of the gene by IFN- (cid:1) in cooperation with (cid:2) 100GAS. Electrophoretic mobility shift assay showed that IRF-1 dominantly binds to (cid:2) 88ISRE in an IFN- (cid:1) -dependent fashion. These results demonstrate a new mechanism for IFN- (cid:1) -induced transcription of the gp91 phox gene by the cooperation of STAT-1 (cid:4) and IRF-1 binding to (cid:2) 100GAS and (cid:2) 88ISRE, respectively. one of the following sequences: W-102A ( lane 2 ), a double-stranded oligonucleotide spanning from bp (cid:1) 102 to (cid:1) 73 of the promoter ( W-102B , lane 3 ), IRFmut ( lane 4 ), and Hetero ( lane 5 ). In the latter experiments ( lanes 6–11 ), 2 (cid:3) g of specific polyclonal IgG and control rabbit IgG ( lane 6 ) were used as follows: anti-IRF-1 ( lane 7 ), anti-IRF-2 ( lane 8 ), anti-p48 ( lane 9 ), anti-ICSAT ( lane 10 ), and anti-ICSBP ( lane 11 ). nuclear extracts prepared from U937 treated with 100 units/ml IFN- (cid:1) for 5 h. The binding was competed with 100-fold molar excess of the following competitors: heterogeneous sequence ( Hetero ), and BID-binding site-mutated or IRF-binding site-mutated oligonucleotide spanning from bp (cid:1) 102 to (cid:1) 73 of the promoter ( BIDmut and IRFmut , respectively). immunoassays, 2 (cid:3) g of control rabbit IgG ( control ) and anti-IRF-1 ( anti-IRF-1 ) were used. The binding of a complex to the probe was completely abolished by BIDmut and anti-IRF-1, but not by Hetero, IRFmut, and control antibody. These results indicate that IRF-1 binds to (cid:1) 88ISRE in the W-95

Interferon (IFN)-␥ induces the expression of the gp91 phox gene both during myeloid differentiation and also in mature phagocytes through several cis-elements and their binding proteins. To find new cis-elements for this induction, transient expression assays were performed using a reporter gene driven by serially truncated gp91 phox promoters in U937 cells. The results suggest that a critical cis-element for induction exists in the region from bp ؊115 to ؊96 of the promoter. Site-directed mutagenesis showed that a ␥-activated sequence (GAS) element at bp ؊100 (؊100GAS) of the gp91 phox promoter plays a pivotal role for the IFN-␥-dependent activity of the bp ؊115 to ؉12 region of the gp91 phox promoter. Electrophoretic mobility shift assays using several GAS competitors and specific antibodies indicated that phosphorylated STAT-1␣ specifically binds to the ؊100GAS. Site-directed mutagenesis showed that an interferon-stimulated response element (ISRE) at bp ؊88 (؊88ISRE) mediates the induction of the gene by IFN-␥ in cooperation with ؊100GAS. Electrophoretic mobility shift assay showed that IRF-1 dominantly binds to ؊88ISRE in an IFN-␥-dependent fashion. These results demonstrate a new mechanism for IFN-␥-induced transcription of the gp91 phox gene by the cooperation of STAT-1␣ and IRF-1 binding to ؊100GAS and ؊88ISRE, respectively.
Phagocytes, such as macrophages and granulocytes, generate superoxide anions by the phagocyte NADPH oxidase to kill ingested microorganisms (1). The gp91 phox gene encodes an essential component of the oxidase, and mutations including a deletion (2,3) of the gene result in X-linked chronic granulomatous disease characterized by severe and recurrent infections due to lack of superoxide generation (4). The gp91 phox gene is expressed almost uniquely in differentiated myeloid cells (5). The human gp91 phox promoter from bp Ϫ450 to ϩ12 directs the transcription of a reporter gene in a subset of mouse monocytes/macrophages in transgenic mice (6). Eosinophils, in particular, express gp91 phox in a patient with chronic granulomatous disease (7). These observations indicate that the expression of the gp91 phox gene is both lineage-and differentiation stage-specific. Transcription of the gp91 phox gene is also modulated by inflammatory mediators such as interferon (IFN)-␥, 1 tumor necrosis factor-␣, and lipopolysaccharide in myeloid cells (8,9). Although enhancer elements for the expression of gp91 phox gene in mature myeloid cells have been suggested to be in a 50-kbp region located upstream of a transcription start site of the gene (10), principal cis-regulatory elements are clustered in a 450-bp proximal promoter region of the gene. Several transcription factors regulate gp91 phox gene expression through these cis-elements. A transcriptional repressor, CCAAT displacement protein (CDP/cut), binds to multiple sites in the proximal promoter in immature myeloid cells (11,12). Coincident with induction of gp91 phox gene expression, the binding activity of CDP to these binding sites decreases during differentiation of the cells (11)(12)(13)(14). Moreover, the induction of the gp91 phox gene is prevented by the constitutive overexpression of CDP in a myeloid cell line (13). Finally, CDP itself disappears in the maturation course to peripheral monocytes and macrophages (14). Although general transcriptional activators such as those with increased binding during differentiation, like (BID)/YY1, CCAAT box-binding factors, CP1, and NF-Y, bind to the promoter, these factors can work only after CDP has been released because their binding sites overlap those for CDP (12, 14 -16). Interferon regulatory factor (IRF)-2 also activates the gp91 phox promoter through two interferonstimulated response elements (ISRE) in the absence of CDP binding (15,17). PU.1 works as a pivotal transcription factor of the gp91 phox gene by binding a PU.1/hematopoietic-associated factor (HAF)-1-binding element centered at bp Ϫ53 in human neutrophils, monocytes, and B-lymphocytes (18). HAF-1 is a multiprotein complex, and its components are still not defined, but PU.1, IRF-1, ICSBP, and Elf-1 are picked out as candidates (19,20). GATA-1, GATA-2, and GATA-3 regulate transcription of the gp91 phox gene in eosinophilic cells (21,22).
IFN-␥ is a cytokine that plays an important role in both innate and adaptive immunities (23). IFN-␥ binds to the specific cell surface receptor and activates the receptor-associated Janus family tyrosine kinase (JAK)1 and JAK2. The kinases phosphorylate and activate a latent cytoplasmic transcription factor, STAT-1␣, which is then translocated to the nucleus as a transcriptionally active homodimer (␥-interferon activation factor) and binds to the ␥-activated sequence (GAS) element of IFN-␥-responsive genes resulting in their activation (24). IFN-␥ also induces gp91 phox gene expression, coinciding with monocytic differentiation of committed progenitors (9) and increases gp91 phox transcription in mature monocytes ex vivo (8). Two positive regulation mechanisms for the IFN-␥-induced transcription of the gp91 phox gene have been proposed. One mechanism is the binding of BID/YY1 to the four binding sites that are scattered in a region from bp Ϫ90 to Ϫ355 of the promoter after dissociation of CDP from the gene (15,16). The other is the binding of an undefined multiprotein complex of HAF-1 to the PU.1/HAF-1-binding element centered at Ϫ53 of the promoter (19,25,26).
In this study, we show that STAT-1␣ bound to the GAS element at bp Ϫ100 of the gp91 phox gene has an unequivocal role in IFN-␥-induced transcription of the gene. We further show that IRF-1 participates in the STAT-1-mediated transcription through binding to the ISRE at bp Ϫ88 of the gene. These results suggest a new mechanism for IFN-␥-induced transcription of the gp91 phox gene in which STAT-1␣ and IRF-1 cooperate with each other.

MATERIALS AND METHODS
Cell Culture-U937 cells (Japanese Collection of Research Bioresources, Tokyo) were maintained in RPMI 1640 medium containing 10% (v/v) fetal calf serum. The cells were either untreated or treated with 100 units/ml human IFN-␥ (kindly provided by Shionogi Corp., Osaka, Japan) for the indicated periods.
Northern Blot Analysis-Total RNA was prepared from cells with Trizol LS reagent (Invitrogen) according to the manufacturer's protocol. The RNA (5 g/lane) was electrophoresed on formaldehyde-containing 0.9% agarose gels, transferred to Hybond TM -Nϩ nylon membranes (Amersham Biosciences Inc.), and fixed by ultraviolet light. A probe of 2.9 kb of human gp91 phox cDNA (2) and an 800-bp cDNA of rat glyceraldehyde-3-phosphate dehydrogenase cDNA were labeled with [ 32 P]dCTP using the Megaprime DNA labeling system (Amersham Biosciences). Messenger RNAs were hybridized with these labeled probes. The hybridization patterns were visualized, and the relative radioactivity of each band was estimated on a Molecular Imager FX (Bio-Rad).
The gp91 phox Promoter Constructs-Constructs pϪ487/Luc and pϪ301/Luc were prepared by the exonuclease III deletion method (27) from pGV-5635/Luc, which had the fragment of the gp91 phox gene extending from its initiation codon to upstream bp Ϫ5635. Constructs pϪ267/Luc, pϪ115/Luc, pϪ95/Luc, and pϪ84/Luc were prepared using the PCR method using pGV-267/Luc which had the fragment of the gene extending from its initiation codon to upstream bp Ϫ267 as the template as described previously in detail (21). A mutant of GAS (pϪ115GASm/Luc) with a two-point mutation (TC to AG) from the bp Ϫ103 to Ϫ102 and a mutant of ISRE (pϪ115IRFm/Luc) with a fivepoint mutation (AGGG to TCCC) from bp Ϫ87 to Ϫ84 were prepared from pϪ115/Luc by PCR using a mutated primer.
Promoter Activity Assays-U937 cells (5 ϫ 10 6 ) were electroporated at 950 microfarads and 280 V in a 4-mm gap cuvette containing 10 g of gp91 phox promoter/firefly luciferase plasmid and 0.5 g of herpes simplex virus thymidine kinase promoter/Renilla luciferase plasmid in 0.25 ml of 20 mM Hepes/RPMI 1640 medium (pH 7.2) using a Bio-Rad Gene Pulser II (Bio-Rad). After a 15-min incubation on ice, the cells were incubated at 37°C under 5% CO 2 , 95% air for 5 h in RPMI 1640 containing 10% fetal calf serum with or without 100 units/ml IFN-␥. Reporter activities were measured as described previously (18).
Preparation of Nuclear Extracts-Nuclear extracts were prepared from U937 cells treated or untreated with IFN-␥ as described previously (18) with some modifications. Approximately 1 ϫ 10 7 cells were swollen in 400 l of ice-cold buffer A (10 mM Hepes (pH 7.9), 10 mM KCl, 0.1 mM EGTA, 0.1 mM EDTA, 1 mM dithiothreitol/Complete TM Protease Inhibitor Mixture (Roche Molecular Biochemicals)/Phosphatase Inhibitor Mixture I and II (Sigma)). After a 15-min incubation at 4°C, the cells were lysed in 0.6% Nonidet P-40 by vigorous vortexing for 10 s. After centrifugation at 5,000 rpm (1,700 ϫ g) for 1 min, supernatants were discarded, and precipitated nuclei were resuspended in 100 l of icecold buffer C (20 mM Hepes (pH 7.9), 0.4 M NaCl, 1 mM EGTA, 1 mM EDTA, 1 mM dithiothreitol, Complete TM Protease Inhibitor Mixture/ Phosphatase Inhibitor Mixture I and II). Nuclear extracts were separated by centrifugation at 15,000 rpm (18,000 ϫ g) for 5 min at 4°C and stored in aliquots at Ϫ80°C. Protein concentration was determined with the Bio-Rad protein assay kit.
Western Blot Analysis-Western blot analysis was carried out using standard methods (30). Nuclear extracts (24 g of protein) were mixed with an equal volume of 250 mM Tris-HCl (pH 6.8) containing 4.6% (w/v) SDS, 20% (w/v) glycerol, 10% (v/v) 2-mercaptoethanol, and 20 g/ml bromphenol blue, and heated at 100°C for 5 min. Samples separated by SDS-PAGE in 10% gel were transferred onto an Immobilon TM polyvinylidene difluoride membrane (Millipore, Bedford, MA) using a semidry type electroblotting apparatus (model BE-300, Bio-Craft, Tokyo, Japan). The protein-blotted polyvinylidene difluoride membrane was stained with Coomassie Brilliant Blue, photographed, and destained with methanol. The membranes were pretreated with Block Ace (Yukijirushi Co., Sapporo, Japan) containing 0.1% Tween 20 for 1 h at room temperature followed by washing with TBS (20 mM Tris (pH 7.4), 150 mM NaCl) containing 0.1% Tween 20 (TBS/Tween) and then incubated with 1 g/ml anti-IRF-1 antibody (C-20, Santa Cruz Biotechnology, Santa Cruz, CA) or nonimmunized rabbit IgG in TBS/Tween containing 20 mg/ml bovine serum albumin for 1 h at room temperature. After incubation, the membrane was washed with TBS/Tween and incubated with 1/2500 diluted anti-rabbit IgG conjugated with horseradish peroxidase (Amersham Biosciences) for 1 h at room temperature. The membrane was then washed extensively with TBS/Tween and then rinsed with TBS. The antigen antibody complexes were detected using an Enhanced Chemiluminescence Kit (Amersham Biosciences Inc.).
Statistical Analyses-Data are expressed as mean Ϯ S.E. Statistical analysis was done on data from at least three independent experiments by Student's t test.

Identification of IFN-␥-responsive Elements in the gp91 phox
Promoter-We analyzed the time course of gp91 phox mRNA levels in human promonocytic U937 cells after addition of 100 units/ml IFN-␥ by Northern blot analysis. As shown in Fig. 1, the amount of gp91 phox mRNA increased ϳ4.5-fold within 5 h and reached its maximal level (ϳ7.5-fold) at 48 h after the addition of IFN-␥ as demonstrated in previous studies (8,9,26) with U937 cells and cultured human monocyte-derived macrophages. Similar studies previously done (15,25,26) for IFN-␥ induction of the gp91 phox gene were focused on the later phase of induction (at 24 -48 h after the addition of IFN-␥). In this study, we focused on the early phase of IFN-␥ induction of the gp91 phox gene because the changes in this phase are the most pronounced. The bp Ϫ450 to ϩ12 fragment of the gp91 phox promoter directs IFN-␥-inducible transcription in a stably transfected PLB-985 myeloid cell line (25). Therefore, we initially prepared a pϪ487/Luc construct that contains the gp91 phox gene fragment from bp Ϫ487 to ϩ12 in front of a luciferase reporter gene and transiently transfected it with the reference plasmid pRL-TK into promonocytic U937 cells. The reporter activity of the cells was assessed following treatment with IFN-␥ (100 units/ml) for 5 h and was compared with that of untreated transfectants (Fig. 2). Reporter activity of the pϪ487/Luc was induced 8.2 (Ϯ0.6)-fold by IFN-␥ treatment and reached levels 12 times higher than that of the control, untreated wild-type pϪ115/Luc, indicating that the bp Ϫ487 to ϩ12 region of the gp91 phox promoter directs transcriptional induction of the gene by IFN-␥ in U937 cells. To determine regions responsible for the IFN-␥ induction in the bp Ϫ487 to ϩ12 fragment of the gp91 phox promoter, five serial deletion constructs were generated from the pϪ487/Luc construct. Progressive 5Ј deletions from bp Ϫ487 to Ϫ116 still retained both a high inducibility and high relative promoter activity in response to IFN-␥ treatment (compare pϪ301/Luc, pϪ267/Luc, and pϪ115/Luc with pϪ487/Luc). The removal of an additional 20-bp fragment from bp Ϫ115 to Ϫ96 remarkably reduced the IFN-␥-dependent inducibility and the activity of the promoter (compare pϪ95/Luc with pϪ115/Luc), suggesting that a pivotal cis-element(s) for IFN-␥ induction lies within the region from bp Ϫ115 to Ϫ96 of the gp91 phox promoter. Although bp Ϫ95/Luc and bp Ϫ84/Luc showed significantly higher activity than that of pXP2N, their inducibilities were not significantly different from that of pXP2N. Therefore, further deleted constructs were not analyzed.
STAT-1␣ Binds to a GAS Element at bp Ϫ100 of the gp91 phox Promoter in IFN-␥-treated U937 Cells-To analyze binding of protein complexes to the bp Ϫ115 to Ϫ96 region, we prepared nuclear extracts from U937 cells at various times after the addition of 100 units/ml IFN-␥, and we performed EMSAs with the fragment from bp Ϫ107 to Ϫ85 (W-107A in Fig. 3) as a probe. As shown in Fig. 4, binding of a protein complex (shown by an arrow) to the probe was induced within 1 h of stimulation with IFN-␥ (Fig. 4A) and could be continuously detected over the entire 48 h after the IFN-␥ addition (Fig. 4B). The binding of this protein complex to the probe was specific because the binding was abolished by an excess of homologous competitor (Ho), with a sequence identical to that of the probe, but not by an excess of heterogeneous competitor (He), with a sequence completely different from that of the probe (Fig. 4B). To identify the DNA-binding protein complex, we searched for an element possibly recruiting a certain transcription factor in the region from bp Ϫ107 to Ϫ85 of the gp91 phox promoter. As shown in Fig.  3, the bp Ϫ104 to Ϫ96 region of the gene (5Ј-TTCTGATAA-3Ј, Ϫ100GAS) perfectly conforms to the GAS element at bp Ϫ137 of the human MHC-II transactivator CIIA (5Ј-TTCTGATAA-3Ј) (31) and was present in the region from bp Ϫ115 to Ϫ96, which was shown to be important for the IFN-␥-induced gp91 phox promoter activity in the functional assay (Fig. 2). Therefore, we focused on GAS-binding proteins and did further experiments. We examined the effect of mutations in the Ϫ100GAS element for the binding of this protein complex to the W-107A probe using a fragment from bp Ϫ115 to Ϫ90 (W-115 in Fig. 3) and a Ϫ100GAS mutant version of W-115 (GASmut in Fig. 3) as competitors. As shown in Fig. 5, the binding of this protein complex (shown by an arrow) was abolished by the W-115 competitor (lane 4) but not by the GASmut competitor (lane 5), indicating that the Ϫ100GAS element is essential for binding of the protein complex to the W-107A probe. Previous studies (32)(33)(34)(35) showed that the STAT-1␣ transcription factor frequently binds to GAS elements in the genes of IFN-␥-induced proteins such as IRF-1, class II transactivator, and CD40. Differentiation-induced factor (DIF) also binds to GAS elements of several IFN-␥-induced genes such as the Fc␥ receptor I (Fc␥R) in U937 cells after the addition of IFN-␥ (28). To examine whether either one or both factors bind to the Ϫ100GAS of the gp91 phox promoter in the IFN-␥-treated U937 cells, we performed EMSAs using IFP GAS and GBP GAS as common competitors to both STAT-1␣ and DIF and Fc␥RI GAS and c-fos SIE GAS as specific competitors to STAT-1␣ (28). As shown in Fig. 5, binding of the protein complex to the W-107A probe was abolished by all four GAS competitors (lanes 6 -9), suggesting that this protein complex includes STAT-1␣ but not DIF. To confirm the binding of STAT-1␣ to the Ϫ100GAS, we performed gel shift immunoassays using three different anti-STAT-1 antibodies (Fig. 5)  ent gp91 phox promoter activity through the bp Ϫ115 to Ϫ96 region, we examined the effect of a Ϫ100GAS mutation (TTCT-GATAA to TGATGATAA), which abolishes the binding of STAT-1␣ to Ϫ100GAS (Fig. 5), on gp91 phox promoter activity of pϪ115/Luc in IFN-␥-treated U937 cells. As shown in Fig. 6, the Ϫ100GAS mutant reduced both the inducibility (from 5.5 Ϯ 0.5-to 1.4 Ϯ 0.1-fold) and IFN-␥-dependent promoter activity (from 5.5 Ϯ 0.5 to 1.7 Ϯ 0.2) to levels comparable with those (1.8 Ϯ 0.2) of the pϪ95/Luc (compare pϪ115/Luc with pϪ115GASm/Luc and pϪ95/Luc). This result indicates that Ϫ100GAS is pivotal for the IFN-␥-induced gp91 phox promoter activity dependent on its bp Ϫ115 to Ϫ96 region in U937 cells. IRFs reportedly mediate IFN-␥-induced expression of genes such as guanylate-binding protein, major histocompatibility complex class I, and inducible nitric-oxide synthase through ISRE (36). A previous study (17) showed that both IRF-1 and IRF-2 bound to the bp Ϫ102 to Ϫ65 fragment of the gp91 phox promoter including an ISRE centered at bp Ϫ88 (Ϫ88ISRE in Fig. 3). These findings suggested a role for the IRF-1 and/or IRF-2 in IFN-␥-induced gp91 phox promoter activity. As shown in Fig. 6, pϪ95/Luc including the Ϫ88ISRE failed to respond to IFN-␥, suggesting that Ϫ88ISRE contributes nothing by itself to the IFN-␥-dependent activation of gp91 phox gene. This result, however, did not eliminate the possibility of a cooperative contribution of the element with other elements lying upstream of bp Ϫ95. To test this possibility, we examined the effect of a FIG. 3. Oligonucleotides used in EMSA. The nucleotide sequence of the bp Ϫ115 to Ϫ65 region of the gp91 phox promoter is illustrated. A GAS and an ISRE centered at bp Ϫ100 and bp Ϫ88, respectively, are indicated above the sequence. Regions of the double-stranded oligonucleotide used in this study as probes and/or competitors are shown below the sequence. Each mutation ablating the binding of STAT (GASmut), IRF (IRFmut), or BID (BIDmut) is indicated.
FIG. 4. Analysis of a protein complex that binds to the fragment from bp ؊107 to ؊96 of the gp91 phox promoter in U937 cells treated with IFN-␥. EMSA was performed as described under "Materials and Methods" using the W-107A probe (Fig. 3) and using nuclear extracts isolated from U937 cells at 0, 1, 2, 3, and 4 h (A) and at 0, 5, 10, 24, and 48 h (B) after the addition of IFN-␥ (100 units/ml). The binding was not competed or was competed with 100-fold molar excess of either unlabeled probe (Ho) or heterogeneous sequence (He). Arrows indicate specific protein complexes bound to the probe.

FIG. 2. Identification of a region for IFN-␥-induced activation of the gp91 phox promoter in U937 cells.
Luciferase reporter constructs containing serially truncated gp91 phox promoter are illustrated on the left. These constructs were transfected into U937 cells incubated with (I) or without (N) IFN-␥ (100 units/ml) for 5 h. Firefly luciferase activities were normalized by accompanied activities of the co-transfected Renilla luciferase reporter gene (see "Materials and Methods"). Each column and bar in the center graph are the relative means of three or four independent data sets and S.E., respectively. The activity of pϪ115/Luc N was set to 1. Fold induction was calculated as the ratio of relative luciferase activity of IFN-␥-treated samples to that of the untreated sample and is shown as the mean Ϯ S.E. on the right. Asterisks indicate a statistically significant difference (p Ͻ 0.001).
Ϫ88ISRE mutation (5Ј-AAGAAAAGGAAACC-3Ј to 5Ј-AA-GAAAAGTCCCCC-3Ј as IRFmut in Fig. 3), which abolishes the binding of IRF-1 and IRF-2 to Ϫ88ISRE, but not that of BID (17), on the IFN-␥-inducible promoter activity of pϪ115/Luc. As shown in Fig. 6, the Ϫ88ISRE mutation abolished the IFN-␥dependent activity of pϪ115/Luc (compare pϪ115IRFm/Luc-I with pϪ115/Luc-I), suggesting that Ϫ88ISRE is also required for the maximal gp91 phox promoter activity induced by IFN-␥. The activity of pϪ115IRFm/Luc after the addition of IFN-␥ was equal to that of pϪ115/Luc (compare pϪ115IRFm/Luc-I with pϪ115/Luc-N). Its fold induction, however, was significantly higher (p Ͻ 0.05) than that of pXP2N (compare pϪ115IRFm/ Luc with pXP2N), because of the low constitutive activity of pϪ115IRFm/Luc. This significant inducibility might show a Ϫ88ISRE-independent mechanism for the activation of the gp91 phox gene through Ϫ100GAS by IFN-␥. It should be noted that the Ϫ88ISRE mutation significantly (p Ͻ 0.001) reduced the constitutive activity of the pϪ115/Luc (compare pϪ115IRFm/Luc-N with pϪ115/Luc-N), suggesting that the element also contributes to the constitutive expression of the gp91 phox gene.
IRF-1 Binds to Ϫ88ISRE of the gp91 phox Promoter-To determine whether a protein complex(es) was binding to Ϫ88ISRE in IFN-␥-treated U937 cells, we performed EMSA using as a probe the fragment from bp Ϫ102 to Ϫ65 (W-102A in Fig. 3) encompassing the Ϫ88ISRE (Fig. 7). The binding of three protein complexes (C1, C2, and C3) in nuclear extracts from IFN-␥-treated U937 cells (each lane 1 in A and B) was abolished by a homologous sequence competitor (each lane 2 in A and B) but not by a heterogeneous one (lanes 6 and 5 in A and B, respectively), indicating that their binding is specific. The fragment from bp Ϫ102 to Ϫ73 (W-102B in Fig. 3) inhibited the binding of all three complexes (Fig. 7, lane 3 in B), indicating that these complexes bind to the region from bp Ϫ102 to Ϫ73. To characterize them, we used the following oligonucleotides as competitors: an IRF consensus element (IRFcons) and two mutated versions of W-102B that have individually been ablated in the IRF-binding site and BID-binding sites (IRFmut and BIDmut, respectively, in Fig. 3). The BIDmut competitor inhibited the binding of all three complexes (Fig. 7, lane 5 in A), indicating that these complexes contained no BID protein. In contrast to IRFcons (Fig. 7, lane 4 in A), IRFmut did not abolish their binding to the probe (Fig. 7, lanes 3 and 4 in A and B,  respectively), suggesting that all the protein complexes contain at least a member of the IRF transcription factor family. The  6. Functional analysis of ؊100GAS and ؊88ISRE cis-regulatory elements for IFN-␥-induced activity of the gp91 phox promoter in U937 cells. Three wild-type constructs (pϪ115/Luc, pϪ95/Luc, and pϪ84/Luc), two mutants of pϪ115/Luc (pϪ115GASm/Luc and pϪ115IRFm/Luc), and a promoter-less construct (pXP2N) were transfected into U937 cells incubated with (I) or without (N) IFN-␥ (100 units/ml) for 5 h. Firefly luciferase activities were normalized using the accompanied activities of the co-transfected Renilla luciferase reporter gene (see "Materials and Methods"). Each column and bar in the graph are the relative means of three independent data sets and the S.E., respectively. The activity of pϪ115/Luc N was set to 1. Fold induction was calculated as the ratio of relative luciferase activity of IFN-␥-treated samples to that of the untreated one and is shown as the mean Ϯ S.E. Double and single asterisks indicate statistically significant differences with p Ͻ 0.001 and p Ͻ 0.05, respectively. bodies (lanes 8 -11). None of antibodies inhibited the binding of C1 and C2 complexes (Fig. 7B, lanes 6 -11). These results indicate that the C3 complex, but not other complexes, contain IRF-1. The C1 and C2 complexes may contain other members of IRF family or other transcription factors whose binding sites overlap with Ϫ88ISRE. We also confirmed binding of IRF-1 to Ϫ88ISRE of pϪ95/Luc by using the fragment from bp Ϫ95 to Ϫ65 (W-95 in Fig. 3) of the gp91 phox promoter as a probe (Fig.  7C). Next we analyzed the expression level of the IRF-1 protein in the nucleus of IFN-␥-treated U937 cells by Western blot analysis (Fig. 8). Consistent with the results of EMSAs, IRF-1 could be detected in the nucleus of U937 cells from 2 h (Experiment 1), and its amount increased until 48 h after addition of IFN-␥ (Experiment 2). Binding of IRF-1 in U937 cells treated with IFN-␥ for 48 h to Ϫ88ISRE was confirmed (data not shown). The expression of IRF-1 appears to be linked kinetically to that of gp91 phox mRNA in IFN-␥-treated U937 cells (Fig. 1), suggesting the cooperation of STAT-1␣ with concomitantly induced IRF-1 efficiently activates the gp91 phox gene.
Binding of STAT-1␣ and IRF-1 to the Element Containing Both Ϫ100GAS and Ϫ88ISRE-Because binding of STAT-1␣ and IRF-1 to their isolated binding elements (Ϫ100GAS in W-107A and Ϫ88ISRE in W-102A) might not reflect binding to the native element containing both Ϫ100GAS and Ϫ88ISRE, we next analyzed whether STAT-1␣ and IRF-1 would also bind to the oligonucleotide containing the entire Ϫ100GAS and Ϫ88ISRE (W-107B in Fig. 3). As shown in Fig. 9, both STAT-1␣ and IRF-1 clearly bound to the combination Ϫ100GAS/ Ϫ88ISRE element of W-107B, and the binding was abolished by a homologous sequence competitor (W-107B) but not by a FIG. 7. Identification of protein complexes that bind to ؊88ISRE in U937 cells treated with IFN-␥. EMSA was performed as described under "Materials and Methods" using the double-stranded oligonucleotide of W-102A and W-95 spanning from bp Ϫ102 to Ϫ65 and from bp Ϫ95 to Ϫ65, respectively, of the gp91 phox promoter (Fig. 3)  . C, nuclear extracts were prepared from U937 cells treated with 100 units/ml IFN-␥ for 5 h. The binding was competed with 100-fold molar excess of the following competitors: heterogeneous sequence (Hetero), and BID-binding site-mutated or IRF-binding site-mutated oligonucleotide spanning from bp Ϫ102 to Ϫ73 of the promoter (BIDmut and IRFmut, respectively). For immunoassays, 2 g of control rabbit IgG (control) and anti-IRF-1 (anti-IRF-1) were used. The binding of a complex to the probe was completely abolished by BIDmut and anti-IRF-1, but not by Hetero, IRFmut, and control antibody. These results indicate that IRF-1 binds to Ϫ88ISRE in the W-95 probe.
heterogeneous one (Hetero). Binding of STAT-1␣ to the combination Ϫ100GAS/Ϫ88ISRE element could only be competed by cold W-107A containing Ϫ100GAS but not by cold W-102A containing Ϫ88ISRE. On the contrary, binding of IRF-1 to the combination element could only be competed by cold W-102A but not by cold W-107A, suggesting that both factors can bind independently to their binding sites in the gp91 phox promoter.
Low molecular mass polypeptide 2 (LMP2) has the overlapped interferon consensus sequence 2/␥-interferon-activated sequence (ICS-2/GAS). A recent study showed that a complex of STAT-1 and IRF-1 binds to the ICS-2/GAS (37), suggesting simultaneous binding of STAT-1␣ and IRF-1 to the gp91 phox promoter. DISCUSSION The gp91 phox gene is a major gene regulated by IFN-␥ (23). In the present study, deletion and mutation analyses show that a GAS element centered at bp Ϫ100 of the gp91 phox promoter (Ϫ100GAS) is critical for the IFN-␥-induced transcriptional activation of the gene. EMSA analysis shows that IFN-␥ stimulation causes the dominant binding of phosphorylated STAT-1␣ to this Ϫ100GAS. These results indicate that STAT-1␣ activated by IFN-␥ directly enhances the gp91 phox promoter activity through binding to Ϫ100GAS. Consistent with our results, the gp91 phox promoter from bp ϩ12 to Ϫ100, which is the center position of the Ϫ100GAS, was not activated by IFN-␥ in stable transfectants of human promyelocytic PLB985 cells carrying a reporter construct with this promoter (15). The authors also showed that at least one of the BIDbinding sites centered at bp Ϫ225 (BID-225) or bp Ϫ145 (BID-145) of the gp91 phox promoter must be intact for IFN-␥ induction of the promoter (15). In contrast, the bp Ϫ115 to ϩ12 fragment of the gp91 phox promoter lacking these two BID sites is strongly activated by IFN-␥ treatment in our transient expression system. This discrepancy may be explained by different localization of the reporter constructs in these cells. In the stably transfected cells, the reporter gene is integrated into the genome where chromatin structure and regulatory sequences surrounding the integration site can affect reporter gene expression. On the other hand, the transiently transfected reporter gene is exempt from these regulations because of its existence in the nucleus as an episome. CDP binds to multiple sites within the bp Ϫ450 to ϩ12 of the gp91 phox promoter (38). It excludes the binding of transcriptional activators by occupying binding sites and represses the expression of the gp91 phox gene (12). A recent study showed that the binding of BID/YY1 to elements BID-225 and BID-145, increased by various differentiation-inducing agents including IFN-␥ (15), is caused by the decreased binding of CDP to overlapping binding sites but not by active binding of BID factors to their sites (16). The binding of BID/YY1 to the gp91 phox promoter did not increase until 12 h after the addition of IFN-␥ in PLB-985 cells (15), suggesting that BID/YY1 is not associated with the early phase of the IFN-␥ induction. A previous study (17) suggested that the ISRE centered at bp Ϫ88 of the gp91 phox promoter (Ϫ88ISRE) is necessary for constitutive activity of the gp91 phox promoter in K562 and HEL erythroleukemia cells expressing no endogenous gp91 phox . In this study, we confirm the necessity of Ϫ88ISRE for the constitutive activity of the gp91 phox promoter. Moreover, we show that Ϫ88ISRE is not sufficient for IFN-␥-dependent gp91 phox promoter activation but is essential for its maximal induction. We also show that IRF-1 dominantly binds to Ϫ88ISRE after IFN-␥ treatment. These results indicate that IRF-1 mediates the IFN-␥-induced maximal activation of the gp91 phox promoter through Ϫ88ISRE in U937 cells. Consistent with our results, IRF-1 binds to Ϫ88ISRE in IFN-␥-treated HeLa cells (17). IRF-2 also binds to Ϫ88ISRE in phorbol 12-myristate 13-acetate-treated PLB-985 cells, HEL cells, and K562 cells (17). However, we could not detect IRF-2 binding to the site in this study. Taken together, our results indicate that the cooperation of STAT-1␣ and IRF-1 through Ϫ100GAS and Ϫ88ISRE, respectively, plays a critical role at least in the early phase of the IFN-␥-induced transcription of the gp91 phox gene (Fig. 10). Continuous binding of STAT-1␣ FIG. 8. Immunoblot analysis of IRF-1 expression in U937 cells after addition of IFN-␥. The results of two independent experiments are shown. Cells were harvested at the indicated times of culture with IFN-␥ (100 units/ml). Proteins of nuclear extracts from the cells were separated by SDS-PAGE and subjected to immunoblotting as described under "Materials and Methods" using anti-IRF-1 polyclonal antibody (Anti IRF-1) and a nonimmunized rabbit IgG (Control). An arrow indicates the position of IRF-1. Other bands are nonspecific because they were also observed using a control antibody.
FIG. 9. Binding of STAT-1␣ and IRF-1 to the element containing both ؊100GAS and ؊88ISRE. EMSA was performed as described under "Materials and Methods" except using 2 g/ml heat-denatured salmon sperm DNA instead of poly(dI-dC)⅐poly(dI-dC). The W-107B probe containing both Ϫ100GAS and Ϫ88ISRE was incubated with nuclear extracts isolated from U937 cells untreated or treated with 100 units/ml IFN-␥ for 5 h. The binding was competed with a 100-fold molar excess of one of the following oligonucleotides: unlabeled probe (W-107B); heterogeneous oligonucleotide (Hetero); oligonucleotide from bp Ϫ107 to Ϫ85 of the promoter (W-107A) containing Ϫ100GAS; oligonucleotide from bp Ϫ102 to Ϫ65 of the promoter (W-102A) containing Ϫ88ISRE. and IRF-1 to the gp91 phox promoter for 48 h after the addition of IFN-␥ in U937 cells suggests that the cooperative mechanism works in a late phase of induction. Phosphorylated STAT-1␣ and IRF-1 are degraded by the ubiquitin-proteasome pathway, decreasing to low levels between 1 and 2 h (39). In contrast, sustained binding of STAT-1␣ and IRF-1 to the class II transactivator promoter was observed over 24 h by continuous treatment of 400 units/ml IFN-␥ but not by 15 min of pulsed treatment (40). Continuous stimulation by IFN-␥ may also be needed for the sustained binding of STAT-1␣ and IRF-1 to the gp91 phox promoter. Previous studies (25) showed that the PU.1/HAF-1 element at bp Ϫ53 of the gp91 phox gene is necessary for constitutive (18) and IFN-␥-induced transcription of the gene. A recent study showed (19) that overexpression of PU.1, IRF-1, and ICSBP significantly activates an artificial minimal promoter linked to multiple copies of the PU.1/HAF-1 element in U937 cells, suggesting it as one mechanism of IFN-␥-induced gp91 phox transcription. Binding of PU.1 alone to the element is accepted in general, but the composition of HAF-1 is still under debate. Consistent with our previous data, one group did not identify PU.1 as a component of HAF-1 (18,20). In contrast, another group (19) identified PU.1 as a component of HAF-1. The latter researchers also clearly identified the IRF-1 as that of HAF-1 by EMSA in untreated U937 cells. We, however, detected neither IRF-1 (Figs. 7A and Fig. 8) nor its mRNA by Northern blot analysis 2 in IFN-␥-untreated U937 cells. The U937 cells used by the latter group might have been autonomously activated as if the cells had been stimulated by IFN-␥. Elf-1 is also a component of HAF-1 (20). 2 A previous study (41) showed that ICSBP was constitutively phosphorylated in U937 cells, and the phosphorylated ICSBP indirectly bound to DNA through interaction with IRF-1 directly bound to DNA in response to IFN-␥ stimulation. ICSBP can interact with PU.1 (19,42). Our results of IRF-1 binding to Ϫ88ISRE may provide efficient formation of a complex of PU.1 and IRF-1 bridged by ICSBP on the gp91 phox promoter after IFN-␥ stimulation (Fig. 10). Ubiquitously expressed CREB-binding protein (CBP) works as a global transcriptional co-activator that has a critical role in a wide variety of cellular events including regulation of IFN-␥-inducible genes and differentiation (43,44). A recent study showed the recruitment of CBP by PU.1, IRF-1, and ICSBP to increase gp91 phox transcription (26). STAT-1␣ has been shown to cooperate with other transcription factors such as upstream stimulatory factor-1, glucocorticoid receptor, and PU.1 to induce transcription of genes (45,46). Cell type-specific induction of gp91 phox transcription by IFN-␥ may be accomplished by cooperation between STAT-1␣ and PU.1. Indirect physical interaction of both transcription factors through a co-activator CBP is possible because two sites of CBP between amino acids 566 and 664 and amino acids 1283 and 1915 independently can interact with the amino-terminal region of active STATϪ1␣ and the transactivation domain of PU.1, respectively (47,48). Based on our results and these previous findings, we propose the formation of a large complex composed of STATϪ1␣, IRF-1, PU.1, ICSBP, and CBP on the proximal promoter of gp91 phox after IFN-␥-stimulation (Fig.  10). Further experiments are required to prove this hypothesis. Binding of IFN-␥ to its receptor induces phosphorylation of STAT-1␣. The phosphorylated STAT-1␣ forms a dimer, moves into the nucleus, and binds to Ϫ100GAS of the gp91 phox promoter. Meanwhile the activated STAT-1␣ also binds to the GAS site of the IRF-1 promoter and induces IRF-1. The IRF-1 binds to Ϫ88ISRE of the gp91 phox promoter and activates it in cooperation with the STAT-1␣ at Ϫ100GAS. IFN-␥-induced gp91 phox transcription may be mediated by the formation of a new large complex composed of STATϪ1␣, IRF-1, PU.1, ICSBP, and CBP in the proximal region of the gp91 phox promoter.