The Matrix Attachment Region-binding Protein SATB1 Interacts with Multiple Elements within the gp91 phox Promoter and Is Down-regulated during Myeloid Differentiation*

The gp91 phox gene encodes a component of the respiratory burst NADPH oxidase complex and is highly expressed in mature myeloid cells. The transcriptional repressor CCAAT displacement protein binds to at least five sites within the proximal gp91 phox promoter and represses expression prior to terminal phagocyte differentiation. The DNA binding activity of CCAAT displacement protein decreases during terminal phagocyte differentiation, thus permitting the binding of transcriptional activators and induction of gp91 phox expression. We report here that the matrix attachment region-binding protein SATB1 interacts with at least seven sites within the (cid:1) 1542 to (cid:2) 12-base pair gp91 phox promoter. Four additional binding sites for CCAAT displacement protein were also identified. Further-more, the most proximal SATB1-binding site within the gp91 phox promoter binds specifically to the nuclear matrix fraction in vitro . SATB1 expression is down-regulated during terminal myeloid cell differentiation, coincident with induction of gp91 phox expression. Transient transfection assays demonstrate that a SATB1-binding site derived from the gp91 phox promoter represses promoter activity in cells expressing SATB1. These findings underscore the importance of transcriptional repression in the regulation of gp91 phox expression and reveal a candidate myeloid cloned into a growth hormone reporter gene vector (5, 47). DNA probes were isolated following two sets of successive restriction enzyme diges-tions. For the first set, the promoter/growth hormone plasmid was digested with Bam HI and Sal I, and the (cid:1) 1542 to (cid:2) 12-bp promoter fragment was recovered. Half of this DNA was digested with Ase I, and the (cid:1) 1542 to (cid:1) 1421-bp fragment was recovered. The other half of the DNA was digested with Hin dIII, and the (cid:1) 137 to (cid:2) 12-bp fragment was isolated. For the second set of probes, the promoter/growth hormone plasmid was digested with Hin dIII, and the (cid:1) 1535 to (cid:1) 137-bp frag- ment was recovered. This fragment was digested with Mae III, and the (cid:1) 207 to (cid:1) 137-, (cid:1) 439 to (cid:1) 207-, and (cid:1) 1206 to (cid:1) 1065-bp fragments were recovered. The (cid:1) 1065 to (cid:1) 439-bp fragment was further digested with Bsp LUII1, and the (cid:1) 624 to (cid:1) 439-bp fragment was purified. The (cid:1) 1065 to (cid:1) 624-bp fragment was digested with Ban I and (cid:1) 1065 to (cid:1) 815- and (cid:1) 815 to (cid:1) 624-bp fragments were isolated. The (cid:1) 1536 to (cid:1) 1206-bp fragment was digested with Ase I, and the (cid:1) 1421 to (cid:1) 1206-bp fragment was isolated. Western Blot Analysis— Whole cell extract was quantitated via the Bradford method (48), using the Bio-Rad reagent (Bio-Rad). SDS load- ing dye was added to 100 (cid:6) g of whole cell extract and boiled for 10 min.

The gp91 phox gene encodes a component of the NADPH oxidase complex that is required for the production of a respiratory burst and microbicidal activity by phagocytes (1). The catalytic unit of the oxidase is a membrane-associated heterodimer cytochrome composed of gp91 phox and p22 phox . Other oxidase components, such as p47 phox and p67 phox , are cytosolic but migrate to the membrane upon phagocyte stimulation and associate with the cytochrome to form a functional oxidase. The gp91 phox , p47 phox , and p67 phox genes are transcriptionally silent until terminal myeloid differentiation and are then expressed until cell death. Absence of any of these proteins leads to chronic granulomatous disease (CGD), 1 an immunodeficiency syndrome caused by the absence of a phagocyte respiratory burst (1).
Transcriptional regulation of gp91 phox expression is complex. The Ϫ450 to ϩ12-bp fragment of the gp91 phox promoter is sufficient to direct tissue-specific but variegated transgene expression in a subset of mature phagocytes in transgenic mice (2). However, elements 10 -60 kb upstream of the transcription start site are additionally required to direct appropriate tissuespecific gene expression in the full spectrum of mature phagocytes (3). In addition, the Ϫ450 to ϩ12-bp region of the gp91 phox promoter directs interferon (IFN)-␥-induced expression of a linked reporter gene when stably introduced into myeloid cells (4).
Lineage-restricted regulation of gp91 phox expression requires the combinatorial interaction of multiple repressors and activators with the proximal promoter (Fig. 1). The transcriptional repressor CCAAT displacement protein (CDP) binds to at least five sites within the Ϫ450 to ϩ12-bp promoter region (5)(6)(7). The DNA binding activity of CDP is down-regulated during myeloid cell differentiation, coincident with the induction of gp91 phox expression (5,7). Post-translational regulation of the DNA binding activity of CDP occurs during myeloid cell differentiation, as CDP protein 2 and mRNA (8) persist in mature phagocytes that lack CDP DNA binding activity. Although the mechanism for post-translational regulation of CDP during myeloid cell development has not been established, others (9 -11) have demonstrated that in vitro phosphorylation or acetylation of the DNA-binding domains of CDP inhibits DNA binding activity. Down-regulation of CDP DNA binding activity is necessary for induction of gp91 phox expression, as constitutive overexpression of CDP prevents induction of gp91 phox expression upon terminal differentiation of a myeloid cell line (8). Furthermore, other lineage-restricted genes that encode the secondary granule proteins neutrophil collagenase, neutrophil gelatinase, and lactoferrin are also repressed under these conditions, suggesting that they are additional targets of CDP-mediated repression (12,13). Importantly, down-regulation of CDP has also been observed in differentiated tissues that do not express gp91 phox , such as kidney cells and myotubes (14,15). Thus, modulation of CDP DNA binding activity is necessary but not sufficient to direct myeloid cell-specific expression of the gp91 phox gene.
The mechanism of CDP-mediated transcriptional repression includes blocking interaction between widely expressed transcriptional activators and the gp91 phox promoter (Fig. 1). These activators include IFN regulatory factors (IRF)-1 and -2, the CCAAT box-binding factor CP1, and YY1 (5, 14 -17). In addition, gp91 phox promoter mutations have been identified in several CGD patients. These mutations are clustered in the Ϫ55-bp region, and each ablates a binding site recognized by the Ets family members PU.1 and Elf-1 (18 -20). GATA-1 and GATA-2 also bind and transactivate the gp91 phox promoter specifically in eosinophils (21).
CDP-binding sites often co-localize with binding sites of the matrix attachment region (MAR)-binding protein SATB1. Examples include the T cell receptor ␤ gene enhancer (22), the mouse CD8a gene upstream regulatory region (23), and the mouse mammary tumor virus long terminal repeat (24). Unlike many other MAR-binding proteins, SATB1 is a cell type-specific protein and is predominantly expressed in T cells and thymocytes (24 -27). A gene targeting study has recently shown that SATB1 coordinates temporal and spatial expression of a large number of genes during T cell differentiation (28). Both CDP and SATB1 contain cut repeat and homeodomain DNAbinding domains (26,29) and exhibit activity as repressors of gene expression (5-7, 24, 29, 30). SATB1 recognizes DNA sequences in which one strand consists of A, T, and C but not G nucleotides (ATC sequences) (25). Clustered ATC sequences exhibit a high propensity to unwind or base-unpair when placed under negative superhelical strain, and such regions are called base-unpairing regions (31,32). Base-unpairing regions are typically found within MARs, and SATB1 binds in vivo to ATC sequence stretches at the bases of chromatin loop domains (33). CDP, on the other hand, does not exhibit a DNA binding specificity to ATC sequences, per se. However, in vitro studies show that CDP preferentially binds to sequences containing the homeodomain-binding motif (ATTA) and CCAAT boxes (34) that are often found within AT-rich regions of DNA such as MARs.
Given the multiple CDP-binding sites within the gp91 phox promoter, we examined whether SATB1 also binds to this promoter. SATB1 was found to interact with numerous elements within the Ϫ1542 to ϩ12-bp gp91 phox promoter. Similar to CDP, SATB1 DNA binding activity is abundant in immature myeloid cells but is down-regulated upon terminal phagocyte differentiation. We further demonstrate that the gp91 phox promoter associates with the nuclear matrix fraction and that SATB1 represses gp91 phox promoter activity.

EXPERIMENTAL PROCEDURES
Construction of Plasmids and Oligonucleotides-Plasmids containing the proximal human gp91 phox promoter linked to a luciferase reporter gene (35) and a cytomegalovirus promoter-␤-galactosidase plasmid have been described previously (7). Various oligonucleotides were inserted upstream of the Ϫ102 to ϩ12-bp gp91 phox promoter fragment following digestion with BamHI and SalI (see below for nucleotide sequences). These include a SATB1-binding site derived from the gp91 phox promoter (gp91phoxSATB1), a mutated version of the gp91 phox promoter element that no longer binds SATB1 (gp91SATB1-mut), and an unrelated fragment of the gp91 phox promoter (Ϫ624 to Ϫ579 bp) that fails to bind SATB1. The nucleotide sequence of each luciferase con- FIG. 2. CDP and SATB1 interact with multiple elements within the gp91 phox promoter. EMSA was performed as described under "Experimental Procedures" using gp91 phox promoter element probes and nuclear extract isolated from undifferentiated PLB-985 cells. 1st to 3rd lanes, Ϫ1542 to Ϫ1421-bp probe; 4th to 6th lanes, Ϫ1421 to Ϫ1206-bp probe; 7th to 9th lanes, Ϫ1206 to Ϫ1065-bp probe; 10th to 12th lanes, Ϫ1065 to Ϫ815 probe; 13th to 15th lanes, Ϫ815 to Ϫ624-bp probe; 16th to 18th lanes, Ϫ624 to Ϫ439-bp probe; 19th to 21st lanes, Ϫ439 to Ϫ207-bp probe; 22nd to 24th lanes, Ϫ207 to Ϫ137-bp probe; 25th to 27th lanes, Ϫ137 to ϩ12-bp probe. Antisera directed against CDP or SATB1 were added to the indicated samples. Arrows indicate positions of CDP and SATB1 complexes. struct was determined. Plasmid DNA was isolated using Qiagen Maxiprep kits (Qiagen, Inc., Valencia, CA).
Cell Culture and Transfection Analysis-The human myelomonocytic cell line PLB-985 (36) was a kind gift of Thomas Rado (Birmingham, AL). The human T cell line Jurkat (37), the human erythroleukemia cell line HEL (38), the human chronic myelogenous leukemia cell line K562 (39), the human cervical choriocarcinoma cell line HeLa (40), and the monkey kidney cell line COS-7 (41) were obtained from the American Type Culture Collection (Manassas, VA). PLB-985, HEL, Jurkat, and K562 cells were grown in RPMI supplemented with 10% Fetal Clone III (bovine serum product, HyClone, Logan, UT), 0.2 mM glutamine, 50 units/ml penicillin, and 50 g/ml streptomycin (Life Technologies, Inc.). HeLa and COS-7 cells were grown using similarly supplemented Dulbecco's modified Eagle's media.
PLB-985 cells were differentiated into monocyte/macrophages by treatment with 0.1 M phorbol 12-myristate 13-acetate (PMA) for 48 h (5) or differentiated into neutrophils by treatment with 0.5% dimethylformamide (Sigma) for 6 days. The degree of differentiation of PLB-985 cells was assessed by measuring respiratory burst activity using a nitro blue tetrazolium (NBT) test. Briefly, 20 l of cells in 1ϫ PBS were incubated with 200 l of a supersaturated solution of NBT with 0.1 nM PMA for 30 min at 37°C. Cell cultures were considered NBT-positive when 80% of the cells were purple. COS-7 cells were transiently transfected with the SATB1 overexpression plasmid pECH-SATB1 (25) or empty plasmid using the Lipo-fectAMINE Plus reagent (Life Technologies, Inc.). One day after transfection, cells were harvested for isolation of whole cell extract as described (42). HEL cells were transiently transfected as described (16) with 5 g of luciferase test plasmid and 1 g of cytomegalovirus-␤galactosidase vector to serve as an internal control for transfection efficiency. Sixteen hours after plating, cells were lysed using 1ϫ lysis buffer (Promega, Inc., Madison, WI), and 20 l of cell extract was assayed for luciferase activity using a luciferase assay kit (Promega, Inc., Madison, WI) per the manufacturer's instructions. Measurement of ␤-galactosidase activity was performed as described (43). Each construct was transfected at least four times in duplicate using at least two independent plasmid preparations.
In Vitro DNA Binding Assays-Nuclear extracts were prepared as described by Dignam et al. (44) with slight modifications to inhibit protease activity as described (45). Briefly, cells were pelleted and washed with PBS. Cells were then treated with 2.7 mM diisopropylflu-FIG. 3. Elements of the gp91 phox promoter exhibit cooperative binding to a GST-SATB1 fusion protein. EMSA was performed as described under "Experimental Procedures" using the indicated regions of the gp91 phox promoter as probes and purified GST-SATB1 fusion protein (26). The concentration of GST-SATB1 fusion protein in each reaction is indicated above each lane. The positive control probe is an oligonucleotide, denoted (25) 7 , for which SATB1 exhibits a dissociation constant of ϳ0.2 nM (25). Mobility of complexes cannot be directly compared between panels.
orophosphate. After vortexing and incubating on ice for 10 min, cells were washed three times with PBS. Soluble cell extracts and insoluble nuclear matrices were prepared as described (42,46). Whole cell extracts were prepared by boiling pelleted cells in 100 mM Tris, pH 6.8, 2% SDS, 20% glycerol, 10% ␤-mercaptoethanol, and 0.02% bromphenol blue. A glutathione S-transferase (GST) fusion protein that contains the DNA-binding domain of SATB1 (amino acid residues 346 -763) was purified from Escherichia coli as described previously (26).
Complementary oligonucleotide probes for use in electrophoretic mobility shift assays (EMSA) were annealed and radiolabeled using T4 polynucleotide kinase (New England Biolabs, Beverly, MA) and [␥-32 P]ATP. Restriction fragment EMSA probes were fill-in labeled using Klenow enzyme (Roche Molecular Biochemicals) and [␣-32 P]dCTP. All EMSA probes were purified as described (17). EMSA with nuclear extracts was performed as described previously using 5 g of nuclear extract, 2.3 g of poly(dI-dC), and 15,000 cpm of probe in a 20-l reaction volume (5). Antisera directed against SATB1 (25) or CDP (29) were diluted 1:10, and 1 l was added to the appropriate reaction. The binding reactions were incubated on ice for 20 -30 min and then loaded onto a 0.5ϫ TBE, 3.5% polyacrylamide gel, and electrophoresis was performed at 25 mA. EMSA using recombinant SATB1 was performed as described previously (25), using a GST fusion protein consisting of the 346 -763-amino acid region of human SATB1 that contains both the MAR-binding domain and homeodomain (26).
In vitro nuclear matrix binding assays were performed as described (46). The 0.2-kb EcoRI-PstI fragment containing a portion of the CD8a upstream regulatory MAR was used as a positive control (23). An oligonucleotide corresponding to the CDP-␣-binding element of the gp91 phox promoter (Ϫ137 to Ϫ76 bp) was released from pUC19 by digestion with BamHI and used as the experimental probe. pUC19 linearized by digesting with BamHI was used as a negative control. All three DNA fragment probes, CD8a, CDP-␣, and pUC19, were fill-in labeled with Klenow enzyme and [␣-32 P]dCTP. The CDP-␣ and CD8a probes were gel-isolated, whereas the pUC19 probe was purified using an S-200 spin column per the manufacturer's instructions (Amersham Pharmacia Biotech). Briefly, insoluble nuclear matrices were mixed with 30,000 cpm of each probe and the indicated mass of E. coli DNA in a total volume of 100 l of binding buffer (10 mM Tris, pH 7.4, 50 mM NaCl, 2 mM EDTA, and 0.25 mg/ml bovine serum albumin). Binding reactions were rocked at room temperature for 1 h and then nuclear matrices were pelleted and washed three times with binding buffer. The final pellet was resuspended in 0.2 ml of proteinase K solution (10 mM Tris, 2 mM EDTA, 5 g/ml salmon sperm DNA, and 0.4 mg/ml proteinase K) and incubated at 37°C for 10 -15 h. Reactions were extracted with phenol and chloroform and were ethanol-precipitated. The binding assay samples were then resuspended and subjected to electrophoresis on a 4% polyacrylamide gel.
Restriction Enzyme Fragments Used in EMSA-The Ϫ1542 to ϩ12-bp fragment of the proximal gp91 phox promoter was previously cloned into a growth hormone reporter gene vector (5,47). DNA probes were isolated following two sets of successive restriction enzyme digestions. For the first set, the promoter/growth hormone plasmid was digested with BamHI and SalI, and the Ϫ1542 to ϩ12-bp promoter fragment was recovered. Half of this DNA was digested with AseI, and the Ϫ1542 to Ϫ1421-bp fragment was recovered. The other half of the DNA was digested with HindIII, and the Ϫ137 to ϩ12-bp fragment was isolated. For the second set of probes, the promoter/growth hormone plasmid was digested with HindIII, and the Ϫ1535 to Ϫ137-bp fragment was recovered. This fragment was digested with MaeIII, and the Ϫ207 to Ϫ137-, Ϫ439 to Ϫ207-, and Ϫ1206 to Ϫ1065-bp fragments were recovered. The Ϫ1065 to Ϫ439-bp fragment was further digested with BspLUII1, and the Ϫ624 to Ϫ439-bp fragment was purified. The Ϫ1065 to Ϫ624-bp fragment was digested with BanI and Ϫ1065 to Ϫ815and Ϫ815 to Ϫ624-bp fragments were isolated. The Ϫ1536 to Ϫ1206-bp fragment was digested with AseI, and the Ϫ1421 to Ϫ1206-bp fragment was isolated.
Western Blot Analysis-Whole cell extract was quantitated via the Bradford method (48), using the Bio-Rad reagent (Bio-Rad). SDS loading dye was added to 100 g of whole cell extract and boiled for 10 min. Electrophoresis was performed on a 7% SDS-polyacrylamide gel for 6 -8 h at 25 mA at room temperature, and proteins were transferred to polyvinylidene difluoride membrane (Bio-Rad) overnight at 4°C. The membrane was incubated in blocking buffer (1ϫ PBS, 0.1% Tween 20, 5% dry milk) for 10 -18 h at room temperature. Polyclonal rabbit anti-SATB1 serum and either antibody 1583 (25) or a generous gift of Paul Gottlieb (University of Texas) were diluted 1/2000 in blocking buffer and incubated 10 -18 h at room temperature with orbital shaking. The membrane was washed three times in wash buffer (1ϫ PBS, 0.1% Tween 20). Secondary ␣-rabbit antibody conjugated to horseradish peroxidase was diluted 1/20,000 in blocking buffer and incubated at room temperature for 8 h. Membranes were washed three times in wash buffer and subjected to chemiluminescent detection via the manufacturer's instructions (Amersham Pharmacia Biotech).
Northern Analysis-Total RNA was isolated from cell lines using TriReagent (Molecular Research Center, Cincinnati, OH). Forty micrograms of total RNA was loaded onto a 1% formaldehyde gel and transferred to MagnaGraph nylon membrane (Micron Separations, Inc., Westborough, MA). The 0.9-kb AvaI-BglII SATB1 cDNA fragment from pECH-SATB1 was used as a probe for SATB1 expression (49). Probe was random-primed using the NEBlot kit (New England Biolabs, Beverly, MA) and purified on an S-200 spin column per the manufacturer's instructions (Amersham Pharmacia Biotech). Blots were hybridized using 2 ϫ 10 6 cpm/ml of probe and Sigma PerfectHyb Plus (Sigma) at 68°C for 18 h.

SATB1 Binds to Multiple
Sites within the gp91 phox Promoter-The transcriptional repressor CDP binds to multiple sites within the proximal gp91 phox promoter and excludes the binding of activator proteins (5, 7) (Fig. 1). Previous studies dem-  5. SATB1 fails to interact with previously described CDPbinding sites within the proximal gp91 phox promoter. EMSA was performed as described under "Experimental Procedures" using the indicated oligonucleotide probes and nuclear extract isolated from undifferentiated PLB-985 cells. The location of each CDP-binding site probe within the gp91 phox promoter is illustrated in Fig. 1. Lanes 1-3, CDP-␦ probe (Ϫ382 to Ϫ331 bp); lanes 4 -6, CDP-␥ probe (Ϫ241 to Ϫ192 bp); lanes 7-9, CDP-␤ probe (Ϫ182 to Ϫ112 bp); and lanes 10 -12, CDP-⑀ probe (Ϫ117 to Ϫ90 bp). Antisera directed against CDP or SATB1 were added to the indicated samples. The arrow indicates the CDP complex. onstrated that the CDP-␣ element probe derived from the gp91 phox promoter (Ϫ137 to Ϫ76 bp) produces a doublet of low mobility DNA-protein complexes in EMSA (5,7). However, although both low-mobility complexes exhibit a similar DNAbinding specificity, only the upper complex is disrupted by antiserum directed against CDP. Several regulatory elements in other promoters exhibit a similar doublet of low mobility EMSA complexes in which the upper complex contains CDP and the lower complex contains the MAR-binding protein, SATB1 (22)(23)(24). In addition, the proximal gp91 phox promoter contains several regions of ATC sequence (5), the preferred binding site for SATB1. Hence, studies were performed to determine whether SATB1 interacts with the gp91 phox promoter.
EMSA experiments using restriction fragment probes that cover the entire Ϫ1542 to ϩ12-bp gp91 phox promoter resulted in the detection of multiple SATB1 and CDP-binding sites (Fig. 2). CDP and SATB1 both bind to six regions of the gp91 phox promoter as follows: Ϫ1421 to Ϫ1206 bp, Ϫ1206 to Ϫ1065 bp, Ϫ1065 to Ϫ815 bp, Ϫ624 to Ϫ439 bp, Ϫ439 to Ϫ207 bp, and Ϫ137 to ϩ12 bp. For each of these probes, addition of antiserum directed against SATB1 specifically disrupts the lower member of the doublet, whereas addition of antiserum directed against CDP disrupts the upper complex. Addition of normal serum did not affect either complex (data not shown). In addition, the Ϫ1542 to Ϫ1421-bp probe binds SATB1 but not CDP (Fig. 2, 1st to 3rd lanes). Although a doublet pattern was found using this probe, the upper member of the doublet complex was not disrupted by addition of antibody directed against CDP. Neither CDP nor SATB1 bind the Ϫ815 to Ϫ624-bp probe (13-15th lanes), even in the presence of low concentrations of poly(dI-dC) (data not shown), which sometimes permits visualization of low affinity CDP interactions (6). Together with the five previously described CDP-binding sites in the proximal gp91 phox promoter, these data reveal at least nine binding sites for CDP and seven binding sites for SATB1 within the Ϫ1542 to ϩ12-bp gp91 phox promoter.
EMSA was also performed with a purified GST fusion protein that contains the DNA-binding domain of SATB1 (26) and probes corresponding to restriction enzyme fragments derived from the gp91 phox promoter that contain ATC sequences (5) (Fig. 3). The affinity of GST-SATB1 for gp91 phox promoter fragments was compared with that of a probe with high base unpairing potential. A 25-bp sequence derived from the core unwinding element of the MAR 3Ј of the immunoglobulin heavy chain enhancer exhibits high affinity to SATB1 (25,32). Upon multimerization, a pentamer (wild type (25) 5 ) has a K d of ϳ1 nM (33) and a heptamer (wild type (25) 7 ) has a K d of ϳ0.2 nM) (Fig. 3). Consistent with the finding that native SATB1 binds to the gp91 phox promoter (Fig. 2), purified GST-SATB1 fusion protein binds to restriction fragment probes containing the Ϫ1535 to Ϫ1178-, Ϫ607 to Ϫ138-, and Ϫ137 to ϩ12-bp elements of the gp91 phox promoter. These regions contain at least one and typically multiple stretches of ATC sequence. SATB1 binds to the Ϫ1535 to Ϫ752-bp probe with a K d of ϳ0.015 to 0.02 nM and to the Ϫ607 to Ϫ138-bp probe with a K d of ϳ0.04 nM. Binding of SATB1 to the Ϫ137 to ϩ12-bp probe is weaker, with a K d of ϳ0.8 nM. Importantly, the affinity for the Ϫ1535 to Ϫ752-bp probe is 10 -20-fold greater than that exhibited for each half of this promoter region (Fig. 3). The Ϫ1535 to Ϫ1167and Ϫ1178 to Ϫ752-bp probes exhibit dissociation constants of ϳ0.1 and 0.2 nM, respectively. These data suggest that ATC sequences found at multiple sites within the gp91 phox promoter exhibit synergistic binding to SATB1. SATB1 is predominantly expressed in T cells and thymocytes (25). However, SATB1 expression has also been detected in a number of myeloid cell lines such as U937, HL60, HEL, and K562 cells but not in the non-hematopoietic HeLa cell line (49,50). EMSA was performed to assess the cell lineage distribution of the putative SATB1 complex that forms with the CDP-␣ element probe (Fig. 4). This element is contained within the Ϫ137 to ϩ12-bp probe depicted in Fig. 2. Consistent with the distribution of SATB1, the CDP/SATB1 EMSA doublet was dramatically abundant in Jurkat cells (lanes 1-3) and was also produced by nuclear extract derived from HEL cells (Fig. 4,  lanes 4 -6). However, the SATB1 complex was not apparent in nuclear extract isolated from HeLa cells (lanes 7-9). Interestingly, a faint SATB1 complex is apparent in nuclear extract isolated from non-hematopoietic (monkey kidney) COS-7 cells (lanes 10 -12). These results also reveal that the CDP-␣ element derived from the gp91 phox promoter contains binding sites for both CDP and SATB1.
Because SATB1 and CDP share similar DNA-binding domains and exhibit overlapping DNA-binding specificities (22)(23)(24), we examined whether SATB1 also binds additional CDPbinding sites previously described within the proximal gp91 phox promoter (6, 7) (Fig. 1). EMSA was performed using oligonucleotide probes that correspond to four distinct CDP-binding sites within the human gp91 phox promoter (5, 7) (Fig. 5). Each probe produced a low mobility CDP complex that was disrupted by the addition of antiserum directed against CDP. However, no complexes were disrupted with addition of antibody directed against SATB1, indicating that SATB1 fails to bind to these CDP-binding site elements. Thus, although overlapping, the DNA-binding specificities of CDP and SATB1 are distinct.
SATB1 Is Down-regulated during Myeloid Cell Differentiation-Previous reports (5, 7) demonstrated that CDP DNA binding activity is down-regulated during terminal myeloid differentiation. To examine whether SATB1 DNA binding activity is similarly modulated during myeloid cell development, EMSA was performed using nuclear extract isolated from PLB-985 cells terminally differentiated into monocyte/macrophages or neutrophils. These studies reveal that both SATB1 and CDP DNA binding activity are abolished during terminal phagocyte differentiation (Fig. 6A, lanes 5-8). Nuclear extract integrity is demonstrated by the presence of faster migrating complexes in all lanes, including a complex previously shown to contain the CCAAT box-binding protein CP1 (5,7).
Additional studies were performed to characterize the mechanism of down-regulation of SATB1 DNA binding activity during myeloid differentiation. Northern blot analysis reveals that the SATB1 mRNA is dramatically down-regulated during myeloid differentiation (Fig. 6B, lanes 3 and 4). RNA from HeLa and Jurkat cells were analyzed as negative and positive controls, respectively (lanes 1 and 2). Western blot analysis of soluble cell extract demonstrates a similar decrease in SATB1 protein levels upon differentiation of PLB-985 cells to monocyte/macrophages (Fig. 6C, lanes 3 and 4). Similar results were obtained for soluble nuclear extracts (data not shown). SATB1 was overexpressed in COS-7 cells as a control for the mobility of SATB1 protein (lane 2). Although SATB1 is detected in the soluble extracts of undifferentiated PLB-985 cells, it has also been found associated with the nuclear matrix (51). Hence, Western blot analysis was also performed with whole cell extracts to assess whether the loss of SATB1 DNA binding activity in soluble extracts is due to translocation to the insoluble nuclear matrix. These results demonstrate that the level of SATB1 protein in whole cell lysates is also dramatically downregulated upon terminal myeloid differentiation (Fig. 6D, lanes  3-5). We conclude that SATB1 expression is down-regulated at the transcriptional level during myeloid cell differentiation.
The Proximal gp91 phox Promoter Associates with the Nuclear Matrix Fraction-Because the MAR-binding proteins CDP and SATB1 interact with the CDP-␣ (Ϫ137 to Ϫ76 bp) element, we examined whether this fragment of the gp91 phox promoter associates with the nuclear matrix. An in vitro nuclear matrix binding assay was performed using nuclear matrices isolated from PLB-985 cells. An MAR derived from the CD8a upstream regulatory element (23) was used as a positive control probe, and pUC19 DNA was used as a negative control probe. Both the CD8a and gp91 phox elements associate tightly with the nuclear matrix fraction in the presence of E. coli competitor DNA, whereas the pUC19 DNA fails to bind the nuclear matrix under these conditions (Fig. 7). We conclude that the CDP-␣ element of the gp91 phox promoter interacts with the nuclear matrix fraction.
SATB1 Represses the gp91 phox Promoter-The CDP-␣ element (Ϫ137 to Ϫ76 bp) of the gp91 phox promoter represses promoter activity in non-phagocytic cells and contains binding sites for several transcription factors, including CP1, CDP, SATB1, YY1, GATA-1, GATA-2, and HoxA10 (5,7,16,21,52). Because of the complexity of binding sites within this element, it has proven difficult to specifically ablate the SATB1-binding site to assess functional significance. As an alternative approach, a variety of oligonucleotides containing defined binding sites were inserted upstream of the Ϫ102 to ϩ12-bp gp91 phox promoter. This promoter fragment lacks numerous upstream repressor binding sites ( Fig. 1 and 2) and exhibits strong promiscuous promoter activity following introduction into the non-phagocytic HEL cell line (6,7). We demonstrated previously that deletion of the Ϫ137 to Ϫ103-bp element removes a CDPbinding site (7). Conversely, the Ϫ137 to ϩ12-bp gp91 phox promoter, which retains the CDP-␣ element, is transcriptionally repressed in this assay.
The ability of SATB1 to bind to each experimental oligonucleotide was examined in EMSA (Fig. 8A). As illustrated in Fig.  2, the Ϫ1542 to Ϫ1421-bp region of the gp91 phox promoter contains a strong SATB1-binding site. However, a CDP complex is not evident, even under conditions of low poly(dI-dC) concentration (data not shown). Analysis of oligonucleotides that span this promoter region revealed that the SATB1-binding site lies within the Ϫ1542 to Ϫ1483-bp promoter region (Fig. 8A, lanes 1-4, and data not shown), a region that contains a 58-bp ATC sequence. This SATB1-specific binding site is hereafter denoted gp91phoxSATB1. The slower mobility complex that appears upon addition of antiserum directed against SATB1 (lane 4) appears to be a supershifted SATB1 complex, as it is not disrupted by the addition of antiserum directed against CDP (data not shown). A mutated version of this binding site was created by changing 11 T nucleotides to G or C. This oligonucleotide, gp91SATB1-mut, no longer contains a binding site for SATB1 (Fig. 8A, lanes 5-8). A region of the gp91 phox promoter (Ϫ624 to Ϫ579 bp) that fails to bind to either SATB1 or CDP (Fig. 8A, lanes 9 -11) was used as a negative control. Each promoter/reporter gene plasmid construct was transiently transfected into HEL cells that express SATB1, and luciferase expression was determined. Insertion of a SATB1binding site derived from the upstream gp91 phox promoter (gp91phoxSATB1) resulted in a 96% decrease in reporter gene expression compared with the Ϫ102 to ϩ12-bp gp91 phox promoter construct (Fig. 8B). Mutation of this SATB1-binding site (gp91SATB1-mut) resulted in approximately a 6-fold increase in reporter expression compared with the wild type gp91phoxSATB1 element (p Ͻ 0.05). Similarly, replacement of the gp91phoxSATB1 oligonucleotide with another region of the gp91 phox promoter that lacks a SATB1-binding site (Ϫ624 to Ϫ579 bp) results in approximately a 10-fold increase in reporter gene expression compared with the wild type gp91phoxSATB1 element (p Ͻ 0.05). For reasons not understood, the Ϫ102 to ϩ12-bp construct exhibits exceptionally high promoter activity in HEL cells (7). Insertion of any sequence upstream of this promoter fragment results in a partial dimunition of promoter strength (data not shown). Hence, it is not surprising that the gp91SATB1-mut and Ϫ624 to Ϫ579-bp constructs do not exhibit a level of reporter gene expression comparable with that observed with the Ϫ102 to ϩ12-bp construct. We conclude that the Ϫ1542 to Ϫ1421-bp element that specifically binds SATB1 acts as a negative regulator of the proximal gp91 phox promoter. DISCUSSION Restriction of gp91 phox promoter activity to mature myeloid cells requires a regulated balance between transcriptional repression and transcriptional activators. The repressor CDP interacts with multiple gp91 phox promoter elements and excludes the binding of transcriptional activators (5-7). We re-port here that another repressor, the MAR-binding protein SATB1, interacts with multiple elements within the gp91 phox promoter. In total, nine CDP-binding sites and seven SATB1binding sites have been identified within the Ϫ1542 to ϩ12-bp gp91 phox promoter. Furthermore, adjacent independent SATB1binding elements exhibit synergistic binding to SATB1. This remarkable aggregation of repressor binding sites underscores the importance of regulated repression in the control of myeloid-restricted gene expression.
SATB1 is abundantly expressed in T cells and is required for normal T cell development (25,28). However, lower levels of SATB1 expression have been detected in other hematopoietic lineages (50,(53)(54)(55), and SATB1-binding sites flank the ␥-globin promoter (49,53). We find abundant SATB1 DNA binding activity in nuclear extracts isolated from immature PLB-985 myeloid cells. Importantly, SATB1 is transcriptionally downregulated during terminal myeloid cell differentiation, coincident with the induction of gp91 phox transcription (56, 57). Sim- FIG. 7. The proximal gp91 phox promoter associates with the nuclear matrix fraction. The in vitro nuclear matrix binding assay was performed as described under "Experimental Procedures," using nuclear matrices isolated from undifferentiated PLB-985 cells and either CD8a (lanes 1 and 2) or CDP-␣ (lanes 3 and 4) probes. E. coli DNA (40 ng) was added as a nonspecific competitor in lanes 2 and 4. The pUC19 probe is a control for specificity of binding. Arrows indicate positions of bound probes.
FIG. 8. Functional analysis of SATB1 binding to the gp91 phox promoter. A, EMSA was performed as described under "Experimental Procedures," using nuclear extract isolated from undifferentiated PLB-985 cells and the following probes: lanes 1-4, gp91phoxSATB1 (Ϫ1542 to Ϫ1483 bp of the gp91 phox promoter); lanes 5-8, gp91SATB1-mut (mutated version of the Ϫ1542 to Ϫ1483-bp element that no longer binds SATB1); lanes 9 -12, Ϫ624 to Ϫ579-bp element of the gp91 phox promoter that lacks SATB1 and CDP-binding sites. Antisera directed against CDP or SATB1 were added to the indicated samples. Arrow indicates the SATB1 complex. B, transient transfection assays were performed with HEL cells as described under "Experimental Procedures." Each indicated oligonucleotide was cloned upstream of the Ϫ102 to ϩ12-bp gp91 phox promoter linked to a luciferase reporter gene. Data are represented as the percentage of the Ϫ102 to ϩ12-bp-luc construct. Solid circles represent SATB1. ϫ indicates the absence of a SATB1binding site. pXP2 is the parental luciferase reporter vector. Each construct was analyzed in at least four independent experiments, using at least two independent preparations of each plasmid. Error bars represent S.E. ilar down-regulation of SATB1 is observed upon dimethyl sulfoxide-induced neutrophil differentiation of the myelomonocytic cell line HL60 and hemin-induced erythroid differentiation of the chronic myelogenous leukemia cell line K562. 3 Interestingly, down-regulation of SATB1 appears specific for terminal differentiation of myeloid cells, as no change in SATB1 protein was detected in HL60 myeloid cells upon induction of apoptosis (50,55). Consistent with this distribution, the addition of SATB1-binding sites to the promiscuously active Ϫ102 to ϩ12-bp gp91 phox promoter results in dramatic repression of promoter activity in HEL cells. To our knowledge, the gp91 phox promoter represents the first candidate target gene for SATB1 action in myeloid cells, suggesting that SATB functions in myeloid cell development as well as in T cells. Further studies are therefore warranted to assess myeloid cell development and function in SATB1 knock-out mice (28).
The Ϫ137 to Ϫ67-bp region of the gp91 phox promoter binds a large number of functionally important transcription factors. These include the repressors SATB1, CDP, and HoxA10 (5,7,52) and the activators YY1, CP1, GATA-1, GATA-2, IRF-1, and IRF-2 (5,7,16,17,21). Interaction and/or competition between these factors presumably plays a role in appropriate gp91 phox regulation. The regulation of gp91 phox expression is further complicated by the finding of distant regulatory elements that are required for appropriate expression in the full complement of mature phagocytes (3). Inclusion of 60 kb upstream of the gp91 phox gene in a transgene construct results in appropriate expression in all mature phagocytes following hematopoietic differentiation of murine embryonic stem cells, whereas successive truncations result in a variegated expression pattern in which a progressively smaller percentage of cells is capable of transcribing the transgene (3). However, cells expressing the transgene do so at apparently full strength. A similar phenomenon was observed in transgenic mice carrying up to 2.6 kb of the upstream gp91 phox flanking region, which exhibited strong transgene expression in only a few percent of phagocytic cells (2). Such variegated expression patterns are often seen for genes controlled by active repression and have been interpreted as evidence that gene regulation is fundamentally a binary (on versus off) phenomenon (58,59). Interestingly, a similar variegated expression pattern is observed in a subset of CGD patients who carry single base pair mutations in the proximal gp91 phox promoter that ablate an Ets factor-binding site (18 -20, 60). In these patients, ϳ95% of phagocytes fail to express the gp91 phox gene and are unable to generate a respiratory burst, whereas 5% of the lineage appears normal. Hence, the stochastic balance for generation of a transcription preinitiation complex can be affected by alterations within either the proximal promoter or within distant elements. In this context, it is of interest that MAR-binding proteins (e.g. ϪCDP and SATB1) interact with the proximal gp91 phox promoter, and a fragment of this promoter interacts with the nuclear matrix fraction in vitro. Changes in nuclear matrix composition have been reported in response to differentiation (61,62), cell proliferation (63), and transformation (64 -67). Unfortunately, the in vitro nuclear matrix assay used for our studies do not permit conclusions regarding the lineage specificity of the detected interaction (46,68,69). Additional studies are required to assess the relationship between regulated chromatin structure of the gp91 phox gene and interactions with the nuclear matrix, interactions between distant locus regulators and the proximal promoter, and the role of SATB1 in regulating these processes.