Cis-acting DNA Elements of Mouse Granulocyte/Macrophage Colony-stimulating Factor Gene Responsive to Oxidized Low Density Lipoprotein*

We previously demonstrated that the induction of granulocyte/macrophage colony-stimulating factor (GM-CSF) played an important role in oxidized low density lipoprotein (Ox-LDL)-induced macrophage growth as a growth priming factor. The present study was undertaken to elucidate the transcriptional regulation of the GM-CSF gene using Raw 264.7 cells, a mouse macrophage cell line. Transient transfection into Raw 264.7 cells of several 5′-flanking regions of GM-CSF gene-luciferase fusion plasmids revealed the presence of two positive regulatory sites in regions spanning from −97 to −59 and from −59 to −37 and one negative regulatory site from −120 to −97 in unstimulated cells. When cells were stimulated by Ox-LDL, there was one positive responsive site from −225 to −120 and one negative responsive site from −97 to −59, which contained the NF-κB binding site. Computer analysis revealed the presence of a putative AP-2 binding site from −169 to −160. Mutagenesis of a putative AP-2 binding site and tandem repeat of this site in plasmid resulted in a complete loss and increased responsiveness to Ox-LDL, respectively. Electrophoretic mobility shift assay showed that Ox-LDL increased the binding of certain nuclear protein(s) to a putative AP-2 binding site but decreased their binding to NF-κB binding site. Supershift assay showed that nuclear proteins bound to NF-κB binding site contained, at least, p50 and p65 but could not demonstrate nuclear protein(s) bound to a putative AP-2 binding site. Our results suggested that a putative AP-2 binding site from −169 to −160 was a positive responsive element to Ox-LDL and that the NF-κB binding site from −91 to −82 was a negative responsive element in Ox-LDL-induced GM-CSF transcription.

Atherosclerosis is an inflammatory fibroproliferative process involving a complex set of interconnected events, including endothelial cell injury; smooth muscle cell migration and phenotypic changes; accumulation of monocytes, macrophages, and T lymphocytes; and formation of lipid-laden foam cells (1). In the early stages of atherosclerotic lesions, monocytes/macrophages are the major source of foam cells, which play an essential role in the development and progression of atherosclerotic lesions via production of various active molecules (1). Thus, to elucidate the pathogenesis of atherosclerosis, it seems reasonable to investigate the mechanisms of macrophage activation, proliferation, and survival processes that are regulated by the colony-stimulating factor (CSF) 1 family, such as macrophage colony-stimulating factor (2, 3), interleukin-3 (4), and granulocyte/macrophage colony-stimulating factor (GM-CSF) (5,6).
We and other groups have demonstrated that oxidized low density lipoprotein (Ox-LDL) exhibits a growth-stimulating capacity for macrophages in vitro (7)(8)(9)(10)(11)(12)(13)(14)(15)(16). This finding strongly suggests that Ox-LDL acts as a growth inducer to macrophages by inducing certain intracellular signaling pathways. Subsequent studies from our laboratory identified the exact intracellular signaling pathways in Ox-LDL-induced macrophage growth. These included a rise in intracellular calcium ion and uptake of lysophosphatidylcholine through the scavenger receptors, which resulted in activation of protein kinase C (PKC) (17). Moreover, expression of GM-CSF at the mRNA level was located downstream the signaling pathway from PKC activation to macrophage growth (12).
GM-CSF is a glycoprotein produced by many cells including lymphocytes (18), fibroblasts (19), vascular endothelial cells (20), eosinophils (21), keratinocytes (22), mast cells (23), and monocytes/macrophages (24). It was first identified as a stimulator of progenitor hemopoietic cells to proliferate or differentiate into mature granulocytes or macrophages (25,26). The presence of a signaling pathway linking PKC activation to GM-CSF induction has been described in T lymphocytes. Sugimoto et al. (27) and Tsuboi et al. (28) demonstrated that two cis-acting DNA elements on the GM-CSF promoter region, CLE2/GC, were required for the induction of GM-CSF by phorbol 12-myristate 13-acetate and calcium ionophore. Moreover, Tsuboi et al. (29) demonstrated that cooperation among AP-1-, NF-B-, and NF-AT-binding sequences was required for the induction of GM-CSF, which was located downstream of PKCand Ca 2ϩ -signaling pathways in T lymphocytes. Furthermore, Ares et al. (30) reported that Ox-LDL could induce the activation of AP-1 in smooth muscle cells. Therefore, it is reasonable to speculate that induction of GM-CSF by Ox-LDL is also regulated by the activation of certain cis-acting DNA element(s) and nuclear transcription factor(s) after PKC activation in macrophages. However, the mechanism of GM-CSF production by Ox-LDL in macrophages remains unknown at present. In this study, we examined the promoter activity of GM-CSF in Ox-LDL-induced GM-CSF production by macrophages.

EXPERIMENTAL PROCEDURES
Materials-Calphostin C was purchased from Sigma and dissolved in Me 2 SO. The final concentrations of Me 2 SO were Ͻ0.1% in the culture medium, which did not affect cell viability and cell growth. [␥-32 P]ATP was from NEN Life Science Products. Antibodies for NF-B p50 and p65, AP-2␣, AP-2␤, and AP-2␥ were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Other chemicals were of the best grade available from commercial sources.
Lipoproteins and Their Modifications-Human LDL (d ϭ 1.019 -1.063 g/ml) was isolated by sequential ultracentrifugation from the plasma of consented normolipidemic subjects obtained after overnight fasting (31). LDL was dialyzed against 0.15 mol/liter NaCl and 1 mmol/ liter EDTA, pH 7.4. Acetyl-LDL was prepared by chemical modification of LDL with acetate anhydride (32). Ox-LDL was prepared by incubation of LDL with 5 mol/liter CuSO 4 for 20 h at 37°C followed by the addition of 1 mmol/liter EDTA and cooling (33,34). The concentration of proteins was determined by BCA protein assay reagent (Pierce) using bovine serum albumin as a standard (35). Endotoxin levels associated with these lipoproteins were Ͻ1 pg/mg protein measured by a commercially available kit (Toxicolor system; Seikagaku Corp., Tokyo, Japan). Moreover, growth and viability of Raw 264.7 cells were not affected by endotoxin at a concentration of Ͻ1 ng/ml in our experimental system.
Oligonucleotides-The oligonucleotides used for electrophoretic mobility shift assay contained the following sequences (only one strand is shown): the element of the region from position Ϫ173 to Ϫ147 of the mouse GM-CSF promoter region 5Ј-AAA CCC CCA AGC CTG ACA ACC TGG GG-3Ј (fragment A) and the region from position Ϫ95 to Ϫ70 of the mouse GM-CSF promoter region 5Ј-CTC AGG TAG TTC CCC CGC CCC CCT GG-3Ј (fragment B). Competitors corresponding to a putative AP-2 and NF-B binding sites were designed from their consensus sequences (competitor A, 5Ј-GAT CGA ACT GAC CGC CCG CGG CCC GT-3Ј; competitor C, 5Ј-AGT TGA GGG GAC TTT CCC AGG C-3Ј). Mutated competitors corresponding to a putative AP-2 and NF-B binding sites were designed from their consensus sequences (competitor B, 5Ј-GAT CGA ACT GAC CGC TTG CGG CCC GT-3Ј; competitor D, 5Ј-AGT TGA GGC GAC TTT CCC AGG C-3Ј).
Cell Culture-Raw 264.7 cells were maintained in suspension at a density of 2 ϫ 10 5 to 1 ϫ 10 6 cells/ml in RPMI 1640 (Life Technologies, Inc.) containing heat-inactivated 10% fetal calf serum (Life Technologies, Inc.), 100 units/ml penicillin and 100 g/ml streptomycin (Life Technologies) (medium A). For experiments, the cells were incubated in 100-mm tissue culture dishes (5 ϫ 10 6 cells/dish) or plated at a density of 1 ϫ 10 6 cells/well in culture dishes (35-mm diameter; Falcon). All cell experiments were performed in a humidified atmosphere under 5% CO 2 in air at 37°C.
Purification of Nuclear Extract-When Raw 264.7 cells reached approximately 80% confluence in 100-mm plates in medium A, cells were washed twice with prewarmed phosphate-buffered saline (pH 7.4, 37°C) and then incubated for 3 or 24 h with 20 g/ml of Ox-LDL. Nuclear extract was purified as described previously by Dignam et al. (36). Protein concentrations were determined by the Micro BCA Protein Assay Reagent (Pierce).
Transient Expression of Luciferase Reporter Plasmids into Raw 264.7 Cells-Five g/l firefly luciferase reporter plasmids (pGL3 Basic; Promega, Madison, WI) containing the 5Ј-upstream regions of the GM-CSF gene was transiently transfected into Raw 264.7 cells by the DEAEdextran method using a commercially available kit (Stratagene, La Jolla, CA), with 5 g/l Renilla luciferase control plasmid (pRL-SV40; Promega). After a 24-h incubation with medium A alone, cells were incubated for 24 h in the presence or absence of 20 g/ml Ox-LDL. A plasmid lacking the 5Ј-upstream region of the GM-CSF gene was used as a negative control (pGL3 Basic, Promega). After incubation, cells were washed twice by phosphate-buffered saline and then lysed with 1ϫ passive lysis buffer (Promega) for 15 min at room temperature. The luciferase activity in the resulting protein lysates was measured using the Dual Luciferase Reporter Assay system (Promega). The results were expressed as normalized firefly luciferase activity divided by Renilla luciferase activity, to adjust any differences in transfection efficiency.
During incubation for 24 h, 300 l of the medium were collected at various time intervals and immediately centrifuged at 10,000 ϫ g for 1 min to remove any particulate material. The supernatant was stored at Ϫ80°C immediately. After completion of all culture experiments, the frozen culture supernatants were quickly thawed to determine GM-CSF levels in the medium. The concentration of GM-CSF protein was determined according to the instructions provided by the manufacturer of the GM-CSF-specific ELISA system (Amersham Pharmacia Biotech) using recombinant murine GM-CSF as a standard (12).
RT-PCR Analysis for GM-CSF mRNA-Standard molecular biology techniques were used (38). After incubation of Raw 264.7 cells (2 ϫ 10 6 cells/well in a six-well plate, 35 mm in diameter; Nunc) with or without Ox-LDL (20 g/ml) for different time intervals (0 -5 h), total RNA was extracted with TRIzol (Life Technologies, Inc.). The first strand cDNA synthesis containing 1 g of total RNA was primed with oligo(dT). Primers used for PCR amplification of GM-CSF and ␤-actin were designed on the basis of murine GM-CSF cDNA (39) and murine ␤-actin cDNA (40) sequences as follows: for GM-CSF, forward primer was TGT GGT CTA CAG CCT CTC AGC AC (nucleotides 64 -86 of murine GM-CSF coding sequence), and reverse primer was CAA AGG GGA TAT CAG TCA GAA AGG T (nucleotides 407-431 of murine GM-CSF coding sequence) (39); for ␤-actin, forward primer was GTG GGC CGC TCT AGG CAC CAA (nucleotides 25-45 of murine ␤-actin coding sequence), and reverse primer was CTC TTT GAT GTC ACG CAC GAT TTC (nucleotides 541-564 of murine ␤-actin coding sequence) (40). The sizes of RT-PCR products of GM-CSF and ␤-actin were expected to be 368 and 540 base pairs, respectively. The cycling conditions in the GeneAmp 9600 System consisted of a first step of 94°C denaturation for 10 min, followed by 35 cycles of annealing at 54°C for 60 s, extension at 75°C for 90 s, and denaturation at 94°C for 30 s, with a final elongation step at 75°C for 10 min. Amplification products were analyzed by 1.5% agarose gel electrophoresis. To verify that the amplification products were consistent with the reported sequences of murine GM-CSF and ␤-actin, they were ligated into pGEM-T (Promega), transfected into Escherichia coli XL1-Blue, and sequenced by using 373A DNA sequencer (Applied Biosystems, Foster City, CA) (12).
Statistical Analysis-All data were expressed as mean Ϯ S.D. Differences between groups were examined for statistical significance using Student's t test. A p value less than 0.05 denoted the presence of a statistically significant difference.

Ox-LDL-induced GM-CSF Production in Raw 264.7
Cells-We previously demonstrated that mouse peritoneal macrophages could produce GM-CSF in response to Ox-LDL (12,16). To confirm whether Raw 264.7 cells also respond to Ox-LDL, we investigated the Ox-LDL-induced GM-CSF production in Raw 264.7 cells at protein and mRNA levels by using ELISA and RT-PCR, respectively. Fig. 1A shows that LDL or acetyl-LDL did not induce GM-CSF release into the medium. However, the addition of 20 g/ml Ox-LDL to Raw 264.7 cells significantly induced GM-CSF release into the medium, with the peak release occurring at 4 h after the addition of Ox-LDL (Fig. 1A). RT-PCR analysis showed that GM-CSF mRNA was increased by Ox-LDL with the peak level occurring at 3 h (Fig.  1B), whereas both LDL and acetyl-LDL had no effect on GM-CSF mRNA expression in these cells (data not shown). These results demonstrated that Ox-LDL also induced GM-CSF expression in Raw 264.7 cells.
Transcriptional Activation of Mouse GM-CSF by Ox-LDL-Previous studies demonstrated that several nuclear factor binding sites existed in GM-CSF gene 5Ј-flanking region from sequence Ϫ133 to Ϫ30, which positively regulated GM-CSF expression in response to PKC activation in T lymphocytes (27)(28)(29). Thus, the GM-CSF gene 5Ј-flanking region from Ϫ225 to ϩ26 was cloned and inserted into the promoterless luciferase reporter plasmid (pGL3-basic) ( Fig. 2A) and transiently transfected into Raw 264.7 cells. Incubation with Ox-LDL significantly increased luciferase activity (Fig. 2B), and luciferase activity reached a plateau level at 24 h (Fig. 2C). In contrast, both LDL and acetyl-LDL had no effect on luciferase activity (Fig. 2B). These results suggested that Ox-LDL-induced GM-CSF mRNA expression might be regulated, at least in part, by the transcriptional activation of GM-CSF promoter from Ϫ225 to ϩ26.
Identification of Cis-acting Elements in Mouse GM-CSF Promoter-To identify the regulatory elements in the mouse GM-CSF promoter, we constructed a series of plasmids containing 5Ј-deletions of GM-CSF promoter fused to the luciferase reporter gene (Fig. 3A). As shown in Fig. 3B, in the unstimulated state, luciferase activity in pGL3GM120-transfected cells was almost equal to that in pGL3GM225-transfected cells. However, deletion extending to position Ϫ97, which removed the CLE1 region, resulted in increased luciferase activity. In contrast, deletion extending to Ϫ59 and to Ϫ37 resulted in a reduction of luciferase activity (Fig. 3B). These results suggested that a negative regulatory site existed in the region extending from Ϫ120 to Ϫ97 and that two positive regulatory sites existed in the regions extending from Ϫ97 to Ϫ59 and from Ϫ59 to Ϫ37 in the unstimulated state. To confirm this notion, we constructed two plasmids containing mutations in a region from Ϫ120 to Ϫ97 and a region from Ϫ97 to Ϫ59 (Fig.  3A). Fig. 3B also showed that luciferase activities in pGL3GM120mt-or pGL3GM97mt-transfected cells were significantly higher or lower than those in wild type plasmidtransfected cells, respectively, demonstrating that a region from Ϫ120 to Ϫ97 was a negative regulatory site and that a region from Ϫ97 to Ϫ59 was a positive regulatory site for GM-CSF expression under unstimulated states. On the other hand, when cells transfected with pGL3GM225 were incubated with Ox-LDL, luciferase activity increased, whereas luciferase activity in pGL3GM120-transfected cells was not changed by Ox-LDL-stimulation (Fig. 3B). These results suggested that the region from Ϫ225 to Ϫ120 contained a positive responsive site for Ox-LDL stimulation. In contrast, luciferase activity was significantly decreased by Ox-LDL in pGL3GM97-transfected cells, which contained a positive regulatory site under unstimulated state, suggesting that a negative responsive site for Ox-LDL existed in the 5Ј-flanking region from Ϫ97 to Ϫ59.
We performed a computer analysis of the region extending from Ϫ225 to Ϫ120, which showed the presence of a putative AP-2 binding site from sequence Ϫ169 to Ϫ160 in the mouse GM-CSF promoter region (Fig. 4). Moreover, the NF-B binding site was reported to exist from Ϫ91 to Ϫ82 (41). We next performed a functional analysis using a luciferase reporter plasmid containing mutations in a putative AP-2 binding site from Ϫ169 to Ϫ160 (Fig. 5A). As shown in Fig. 5B, mutation in the putative AP-2 binding site reduced basal promoter activity by 20% and completely diminished Ox-LDL-induced promoter activity. To confirm that the putative AP-2 binding site is a positive responsive site for Ox-LDL stimulation, we constructed pGL3GM177 plasmid, which contained the sequence spanning position Ϫ177 of GM-CSF promoter region, and pGL3GM177 tandem plasmid, which had two more copies of the putative AP-2 binding site (Fig. 6A). Cells transfected with pGL3GM177 plasmid induced 2.5-fold expression of luciferase in response to Ox-LDL, relative to unstimulated cells (Fig. 6B). Transfection of the pGL3GM177 tandem plasmid increased basal level luciferase activity at 1.5-fold and Ox-LDL-induced luciferase activity at 4.4-fold, relative to unstimulated cells (Fig. 6B). These results demonstrated that Ox-LDL-induced GM-CSF expression was required for a putative AP-2 binding site.
Ox-LDL-stimulated Nuclear Factor(s) Binding to the Mouse GM-CSF Promoter-To elucidate whether nuclear protein(s) would bind to the promoter region of GM-CSF gene, the nuclear protein specific for binding to cis-acting elements from Ϫ173 to Ϫ147, containing a putative AP-2 binding site (Fig. 7A), and from Ϫ95 to Ϫ70, containing an NF-B binding site (Fig. 8B), were analyzed by EMSA. As shown in Fig. 7B using a putative AP-2 binding site as a probe, the nuclear proteins from unstimulated cells produced two faint bands, which became prominent bands by Ox-LDL. These bands were completely diminished by cold excess unlabeled fragment A but not completely competed by cold excess unlabeled AP-2 consensus oligonucleotides. Moreover, supershift analysis using anti-AP-2␣, AP-2␤, and AP-2␥ antibodies (Santa Cruz Biotechnology) did not affect the position and density of the bands (data not shown). These results suggested that the binding of nuclear factor(s) to a putative AP-2 binding site was different from AP-2␣, AP-2␤, or AP-2␥ but might be a nuclear protein highly homologous to the AP-2 family. As shown in Fig. 8B, when the NF-B binding element was used as a probe, a strong band was detected in unstimulated cells. Interestingly, this band was decreased by incubation with Ox-LDL for 24 h (Fig. 8B). Cold excess NF-B consensus oligonucleotides diminished this band. However, cold excess mutated NF-B consensus oligonucleotides completely failed to compete for nuclear protein binding to fragment B. Moreover, the density of this band was reduced by both anti-p50 and anti-p65 antibodies but not by nonimmune IgG (Fig. 8C), suggesting that this band contained, at least, NF-B p50 and p65.
Involvement of PKC in GM-CSF Expression-Our previous study demonstrated that PKC activation by Ox-LDL increased GM-CSF mRNA level in peritoneal macrophages (12). We therefore examined the involvement of PKC activation by Ox-LDL in GM-CSF expression using a PKC inhibitor, calphostin C. Preincubation of cells with 500 nM calphostin C resulted in suppression of Ox-LDL-induced GM-CSF mRNA expression to

FIG. 3. Deletion and mutation analysis of the mouse GM-CSF promoter in Raw 264.7 cells after incubation with Ox-LDL.
A, structure of various GM-CSF promoter-luciferase reporter. The mutated positions are underlined. B, Raw 264.7 cells (5 ϫ 10 6 ) in a 100-mm dish were incubated in 15 ml of medium A, and plasmids (5 g) were transfected into Raw 264.7 cells by the DEAE-dextran method. After a 24-h incubation with medium A alone, cells were treated with 20 g/ml Ox-LDL and then harvested 24 h later. Luciferase activity was determined as described under "Experimental Procedures." Data represent the mean Ϯ SD of four separate experiments. †, p Ͻ 0.001, compared with unstimulated pGL3GM120; † †, p Ͻ 0.001, compared with unstimulated pGL3GM97; † † †, p Ͻ 0.01, compared with unstimulated pGL3GM59; #, p Ͻ 0.005, compared with unstimulated pGL3GM225; ##, p Ͻ 0.01, compared with unstimulated pGL3GM97 (Student's t test).

FIG. 4. Putative Nuclear factorbinding sites in the mouse GM-CSF promoter.
Shown is a schematic representation of a segment of the mouse GM-CSF upstream promoter described by Miyake et al. (39). The sequence motifs CLE1, CLE2, the GC box, CLE0 and a putative AP-2 binding site, which was constructed by computer analysis, are indicated. the level of unstimulated cells (Fig. 9A). Moreover, enhanced activation of luciferase by Ox-LDL using pGL3GM225 plasmid was suppressed by 80% by a 500 nM concentration of calphostin C (Fig. 9B), and Ox-LDL-induced nuclear factor(s) binding to fragment A was significantly suppressed by 500 nM calphostin C (Fig. 9C). These results suggested that Ox-LDL-induced nuclear factor(s) activation might be mediated by PKC activation. DISCUSSION Cytokine induction in macrophages is an important process in the development and progression of the early stages of atherosclerotic lesions (1). Various cytokines and growth factors are produced from macrophages in atherosclerotic lesions (1). Among them, we have recently demonstrated that GM-CSF plays an important role in Ox-LDL-induced macrophage growth (12,16). The mechanism of GM-CSF induction has been extensively studied in T lymphocytes (27)(28)(29), whereas such mechanisms are poorly understood in macrophages. In particular, there are no studies demonstrating cis-acting DNA elements within the 5Ј-flanking region of GM-CSF gene that are required for the response to Ox-LDL in macrophages. We therefore examined in the present study the cis-acting elements in the mouse GM-CSF gene using luciferase plasmids with various modified promoter regions of the mouse GM-CSF gene. Our results demonstrated the presence of two positive cis-acting elements and a negative cis-acting element in the unstimulated state. Furthermore, we also showed a positive responsive cisacting element, a putative AP-2 binding site from Ϫ169 to Ϫ160, and a negative responsive cis-acting element, NF-B binding site from Ϫ91 to Ϫ82, for Ox-LDL stimulation. The results also showed that GM-CSF expression might be regulated by the binding of certain nuclear proteins to these elements; Ox-LDL increased the binding of certain nuclear protein(s) to a putative AP-2 binding site but decreased the binding of NF-B p50 and p65 to NF-B binding site.
Luciferase assays with the deletion plasmids and plasmids containing mutations demonstrated the presence of two positive cis-acting elements in unstimulated cells; between positions Ϫ97 and Ϫ59 the element contained the CLE2 region and GC box, while the element between Ϫ59 and Ϫ37 contained the CLE0 region (Fig. 4). In contrast, there was a negative cisacting element between Ϫ120 and Ϫ97 containing the CLE1 region (Fig. 4). These results suggested that CLE1, CLE2/GC box, and CLE0 region might be important for transcriptional regulation of GM-CSF expression in unstimulated cells. Previous studies demonstrated that the region from Ϫ113 to Ϫ30 could act as a phorbol 12-myristate 13-acetate/ionomycin-re- FIG. 5. Mutation in a putative AP-2 binding site reduces transcriptional activation by Ox-LDL. A, structure of the mutated GM-CSF promoter-luciferase reporter (pGL3GM225mt) that was generated from pGL3GM225 by introducing the mutation using PCR mutagenesis. The position of the mutation in a putative AP-2 binding site is underlined. B, pGL3GM225 and pGL3GM225mt (5 g) were transfected into Raw 264.7 cells (5 ϫ 10 6 ) by DEAE-dextran method. After a 24-h incubation with medium A alone, cells were treated with 20 g/ml Ox-LDL and then harvested 24 h later. Luciferase activity was determined as described under "Experimental Procedures." Data represent the mean Ϯ S.D. of four separate experiments. †, p Ͻ 0.05, compared with unstimulated pGL3GM225; † †, p Ͻ 0.01, compared with unstimulated pGL3GM225 (Student's t test).

FIG. 6. Tandem repeat of AP-2 binding site enhances the transcriptional activation of GM-CSF by Ox-LDL.
A, structure of the GM-CSF promoter-luciferase reporter construct, which has two more copies of putative AP-2 binding site (pGL3GM177tandem). B, pGL3GM225, pGL3GM177, and pGL3GM177 tandem plasmids (5 g) were transfected into Raw 264.7 cells (5 ϫ 10 6 ) by DEAE-dextran method. After a 24-h incubation with medium A alone, cells were treated with 20 g/ml Ox-LDL and then harvested 24 h later. Luciferase activity was determined as described under "Experimental Procedures." Data represent the mean Ϯ S.D. of four separate experiments. †, p Ͻ 0.05, compared with unstimulated pGL3GM225, † †, p Ͻ 0.01, compared with unstimulated pGL3GM225, † † †, p Ͻ 0.001, compared with unstimulated pGL3GM225 (Student's t test). sponsive enhancer on heterologous promoters (42). In particular, NF-B (CLE2/GC box) and AP-1/NFAT (CLE0) binding sites (Ϫ95 to Ϫ73 and Ϫ54 to Ϫ40, respectively) are necessary for induction of GM-CSF, and binding of these nuclear factors to their responsible site was increased upon T cell activation by phorbol 12-myristate 13-acetate/ionomycin (22,43,45). Our results are consistent with those of above studies, under unstimulated state. However, we found that the position from Ϫ225 to Ϫ120 might be important for transcriptional activation of GM-CSF in Ox-LDL-stimulated Raw 264.7 cells (Fig. 3B). Computer analysis using the Transcriptional Factor Database (38,44) showed the presence of a single putative AP-2 binding site (from Ϫ169 to Ϫ160) in the sequences spanning positions Ϫ225 to Ϫ120 of the mouse GM-CSF promoter region. Based on our results of transient transfection of luciferase plasmids with a mutation of a putative AP-2 binding site (Fig. 5) and the two additional copies of a putative AP-2 binding site (Fig. 6), this putative AP-2 binding site probably contributes to Ox-LDLinduced GM-CSF induction in Raw 264.7 cells. Future experiments, such as UV cross-linking, DNase footprinting, methylation interference, and missing contact probe analysis may elucidate the contribution of constituent nucleotides to the binding of certain nuclear proteins.
A number of proteins capable of binding to the GM-CSF promoter have been described, including NF-B (46), NF-GMa and NF-GMb (43), NF-GM2 (47), the Ets family member Elf-1 (48,49), and NF-ATp⅐AP-1 complex (44). However, to our knowledge, there are no studies that have previously reported the presence of trans-acting factors binding to the sequences spanning positions Ϫ169 to Ϫ160 of the GM-CSF gene that are required for the response to Ox-LDL in macrophages. The present study demonstrated that trans-acting factor(s) binding to sequences spanning positions Ϫ169 to Ϫ160 of the GM-CSF promoter was increased by Ox-LDL (Fig. 7). These findings suggested that certain nuclear factor(s) activated by Ox-LDL might be positively involved in Ox-LDL-induced induction of GM-CSF transcription.
The transcription factor AP-2 was originally isolated from HeLa cells as an activating factor that bound to promoter regions of the SV40 and human metallothionein IIa (50,51). AP-2 has subsequently been found to be involved in the transcriptional regulation of many cellular genes and reported to encompass three different isoforms, AP-2␣, AP-2␤, and AP-2␥ (52)(53)(54). In addition, a number of splicing variants generate more diversity in AP-2 isoforms (55,56). Our results demonstrated that the binding of nuclear protein to a putative AP-2 binding site was significantly but partially inhibited by cold excess consensus AP-2 oligonucleotides (Fig. 7). To determine whether the nuclear factor(s) bound to the putative AP-2 binding site was AP-2, we performed the supershift assay using anti-AP-2␣, AP-2␤, and AP-2␥ antibodies. However, nuclear factor(s) binding to the fragment A was not shifted by these antibodies. Moreover, the purified AP-2␣, AP-2␤, and AP-2␥ proteins (Santa Cruz Biotechnology) were weakly bound to fragment A compared with the nuclear proteins from Ox-LDLstimulated cells (data not shown). These results suggested that the nuclear factor(s) bound to fragment A might be other AP-2 isoforms or other protein(s) highly homologous to AP-2 family.  ) is underlined. B, electrophoretic mobility shift assay was performed by using radiolabeled fragment B as a probe. Nuclear extracts prepared from untreated Raw 264.7 cells or cells treated with Ox-LDL (20 g/ml) for indicated time intervals. C, nuclear extracts were prepared from Raw 264.7 cells treated with Ox-LDL (20 g/ml) for 3 h. Nuclear extracts were pretreated with 10 g of antibodies for p50, p65, or nonimmune IgG, and then EMSA was performed as described under "Experimental Procedures." Future experiments are necessary to elucidate these nuclear proteins.
In the present study, we demonstrated that the sequence from position Ϫ97 to Ϫ59 was a positive regulatory site for GM-CSF expression in unstimulated cells but reduced luciferase activity in Ox-LDL-stimulated cells compared with unstimulated cells (Fig. 3), suggesting that Ox-LDL might inhibit the transcriptional activation via the CLE2 region. NF-B is known to bind to CLE2 (46,47). The binding of the nuclear factor to CLE2 was completely inhibited by cold excess consensus NF-B oligonucleotides but not by mutated NF-B consensus oligonucleotides (Fig. 8B). Moreover, density of complex of CLE2 and nuclear proteins was significantly reduced by both anti-p50 and anti-p65 antibodies but not by nonimmune IgG (Fig. 8C). These results demonstrated that the nuclear factor bound to CLE2 contained, at least, NF-B p50 and p65. The nuclear factor(s) binding the CLE2 region was not affected by stimulation of Ox-LDL for 3 h. However, the DNA-nuclear factor complex was significantly reduced by stimulation of Ox-LDL after 24 h (Fig. 8). These results suggested that the reduced transcription of GM-CSF by Ox-LDL via CLE2 might be mediated by a decrease in NF-B activation. This conclusion is supported by the results of Brand et al. (57), who demonstrated that long term exposure to Ox-LDL inhibited LPSinduced NF-B activation and interleukin-8 expression in human THP-1 monocytic cells and that this mechanism was dependent on inhibition of IB-␣ degradation.
Ox-LDL-induced GM-CSF production was transiently increased after 4 -8 h, but diminished to almost basal level after 24 h (Fig. 1). The binding of nuclear factors to a putative AP-2 binding element, positive responsive element for Ox-LDL, was increased by Ox-LDL with peak level at 3 h followed by a gradual decrease (Fig. 7). In addition, the binding of the nuclear protein to the NF-B element did not change at 3 h and then decreased by Ox-LDL (Fig. 8). Thus, it is possible to assume that transient activation of GM-CSF transcription might be mediated by the increased binding of nuclear proteins to a putative AP-2 binding element, and then a decrease in the binding of nuclear proteins both to a putative AP-2 binding element and NF-B element might reduce GM-CSF transcription in long term incubation with Ox-LDL.
Our recent studies demonstrated that Ox-LDL-induced GM-CSF production at mRNA level was mediated by the activation of PKC in mouse peritoneal macrophages (12,17). Therefore, in the present study, we examined whether PKC is also involved in the signaling pathway(s) for GM-CSF production activated by Ox-LDL. Pretreatment of Raw 264.7 cells with calphostin C, a PKC inhibitor, down-regulated Ox-LDL-mediated induction of GM-CSF mRNA (Fig. 9A). Calphostin C also inhibited Ox-LDL-induced increase in luciferase activity (Fig. 9B) and the binding of certain nuclear proteins to a putative AP-2 binding site (Fig. 9C). These results suggested that Ox-LDL-induced GM-CSF production might be controlled by PKC via transcriptional activation with a putative AP-2 binding site. In addition, a recent report from Martens et al. (13) demonstrated that phosphatidylinositol 3-kinase also involved in Ox-LDL-induced macrophage growth. Thus, further studies are needed to elucidate the involvement of phosphatidylinositol 3-kinase in signaling pathway for GM-CSF expression in macrophages.