A Novel Role of Sp1 and Sp3 in the Interferon-γ-mediated Suppression of Macrophage Lipoprotein Lipase Gene Transcription*

The regulation of macrophage lipoprotein lipase by cytokines is of potentially crucial importance in the pathogenesis of atherosclerosis. We have shown previously that macrophage lipoprotein lipase expression is suppressed by interferon-γ (IFN-γ) at the transcriptional level. We investigated the regulatory sequence elements and the transcription factors that are involved in this response. We demonstrated that the −31/+187 sequence contains the minimal IFN-γ-responsive elements. Electrophoretic mobility shift assays showed that the binding of proteins to two regions in the −31/+187 sequence was reduced dramatically when the cells were exposed to IFN-γ. Both competition electrophoretic mobility shift assays and antibody supershift assays showed that the interacting proteins were composed of Sp1 and Sp3. Mutations of the Sp1/Sp3-binding sites in the minimal IFN-γ-responsive elements abolished the IFN-γ-mediated suppression of promoter activity, whereas multimers of the sequence were able to impart the response to a heterologous promoter. Western blot analysis showed that IFN-γ reduced the steady state levels of Sp3 protein. In contrast, the cytokine decreased the DNA binding activity of Sp1 without affecting the protein levels. These studies therefore reveal a novel mechanism for IFN-γ-mediated regulation of macrophage gene transcription.

The regulation of macrophage lipoprotein lipase by cytokines is of potentially crucial importance in the pathogenesis of atherosclerosis. We have shown previously that macrophage lipoprotein lipase expression is suppressed by interferon-␥ (IFN-␥) at the transcriptional level. We investigated the regulatory sequence elements and the transcription factors that are involved in this response. We demonstrated that the ؊31/؉187 sequence contains the minimal IFN-␥-responsive elements. Electrophoretic mobility shift assays showed that the binding of proteins to two regions in the ؊31/ ؉187 sequence was reduced dramatically when the cells were exposed to IFN-␥. Both competition electrophoretic mobility shift assays and antibody supershift assays showed that the interacting proteins were composed of Sp1 and Sp3. Mutations of the Sp1/Sp3-binding sites in the minimal IFN-␥-responsive elements abolished the IFN-␥-mediated suppression of promoter activity, whereas multimers of the sequence were able to impart the response to a heterologous promoter. Western blot analysis showed that IFN-␥ reduced the steady state levels of Sp3 protein. In contrast, the cytokine decreased the DNA binding activity of Sp1 without affecting the protein levels. These studies therefore reveal a novel mechanism for IFN-␥-mediated regulation of macrophage gene transcription.
Lipoprotein lipase (LPL 1 ; EC 3.1.1.34) plays a central role in lipid metabolism and transport by catalyzing the hydrolysis of the triacylglycerol component of lipoprotein particles, thereby providing non-esterified fatty acids and 2-monoacylglycerol for tissue utilization (1). LPL is expressed by the parenchymal cells of several extrahepatic tissues and is subject to regulation in response to physiological and pathophysiological changes through the action of hormones, cytokines, and lipid metabolite products (2). The LPL expressed by macrophages is of major importance because of its crucial role in atherogenesis (see Ref. 3 for a recent review). LPL is expressed in the lesion where macrophage-derived foam cells represent the predominant site for the synthesis of the enzyme (4,5). In addition, inbred murine strains with elevated levels of macrophage LPL show an increased susceptibility to atherosclerosis (6). More recently, the importance of LPL in the promotion of foam cell formation and atherosclerosis in vivo has been substantiated by three independent transplantation studies in irradiated mouse model systems using donor macrophages from different backgrounds (7)(8)(9). In those mice receiving macrophages from homozygous and heterozygous LPL-deficient donors, the mean lesion area of diet-induced atherosclerosis was reduced substantially compared with those receiving macrophages that express LPL normally (7)(8)(9). Further support for a pro-atherogenic role of macrophage LPL has been provided by Clee et al. (10) through an alternative approach. Such an atherogenic role of LPL predominantly involves a non-catalytic bridging action in which the enzyme serves as a ligand for mediating the interaction of lipoproteins to cell surface receptors and/or proteoglycans and their subsequent uptake by the cells (3).
The cellular changes in the vascular wall during the initiation and the development of atherosclerosis, including the transformation of macrophages into foam cells, are affected by many factors that are known to be present in the lesion, such as cytokines, growth factors, and modified lipoproteins (11,12). The action of such factors on the expression of macrophage LPL has been implicated in the modulation of the atherosclerotic process (3) and has, therefore, been studied in detail in several laboratories including our own. Macrophage LPL expression is induced by platelet-derived growth factor, macrophage colonystimulating factor, glucose, and activators of peroxisome proliferator-activated receptors and is suppressed by certain cytokines (13)(14)(15)(16)(17). Among these cytokines, interferon-␥ (IFN-␥) possesses a unique ability both to prime macrophages and to synergize with other mediators in the regulation of LPL (18 -19).
IFN-␥ plays an important and complex role in atherogenesis with both pro-and anti-atherogenic actions being reported. IFN-␥-receptor/apoE double knockout mice show reduced lesion formation (20). In addition, IFN-␥ stimulates the expression of vascular cell adhesion molecule on endothelial cells and class II major histocompatibility antigens in macrophages and smooth muscle cells (21,22). On the other hand, IFN-␥ decreases collagen synthesis on smooth muscle cells, blocks smooth muscle cell proliferation, and inhibits macrophage foam cell formation by both preventing the oxidation of low density lipoproteins (LDL) and suppressing the expression of several lipoprotein receptors, including type A scavenger receptor, very low density lipoprotein receptor, LDL receptor-related protein, and scavenger receptor CD36 (23)(24)(25)(26)(27)(28)(29)(30).
In the light of the IFN-␥-mediated inhibition of the expression of several genes in macrophages that are involved in lipoprotein uptake, a detailed understanding is necessary of the mechanisms that are in operation. This will not only better our understanding of the molecular basis of foam cell formation and atherogenesis but, in the longer term, may also lead to the identification of novel targets for therapeutic intervention. With respect to macrophage LPL expression, we have demonstrated previously (31) that the IFN-␥ action is mediated at the transcriptional level. We show here that IFN-␥ decreases macrophage LPL gene transcription through a reduction in the binding of the transcription factor Sp1, and the related member Sp3, to the promoter region. In addition, we identify the potential mechanisms that are responsible for this action of IFN-␥ on Sp1 and Sp3. The studies identify a novel mechanism for IFN-␥-regulated gene transcription.

EXPERIMENTAL PROCEDURES
Reagents-The human myeloid leukemic U937 and the mouse macrophage J774.2 cell lines were from the European Collection of Animal Cell Cultures, and the rat alveolar macrophage NR8383 cell line was obtained from the American Collection of Animal Cell Cultures. Recombinant human and mouse IFN-␥ was from PeproTech. All the cell culture reagents were purchased from Greiner, Helena Biosciences, or Invitrogen. Antisera against Sp1 and Sp3 were from Santa Cruz Biotechnology, and the secondary horseradish peroxidase-conjugated antibodies and the casein kinase 2 (CK2) inhibitors apigenin and emodin were purchased from Sigma.
Cell Culture-The cell lines were maintained in either Dulbecco's modified Eagle's medium (J774.2), RPMI 1640 (U937), or Ham's F-12 medium (NR8383), which was supplemented with 10% (v/v) heat-inactivated fetal calf serum (HI-FCS; 56°C, 30 min), 100 units/ml penicillin, and 100 g/ml streptomycin. The cultures were maintained at 37°C in a humidified atmosphere containing 5% (v/v) CO 2 in air. Alveolar macrophages were isolated from in-house-bred adult male Sprague-Dawley rats by bronchoalveolar lavage (32,33). Rats were sacrificed by a lethal injection of pentobarbital (50 mg/kg) via the intraperitoneal route. The capillary bed of the lungs was perfused with 0.15 M NaCl to remove the blood and the lungs excised intact. Then 10-ml aliquots of 0.15 M NaCl were used to lavage the lungs. This procedure was repeated five times, and the recovered material was pooled. The macrophages were collected by centrifugation of the lavage fluid at 250 ϫ g for 20 min at 10°C. The resultant pellet was resuspended in RPMI 1640 that had been supplemented with HI-FCS, streptomycin, and penicillin, as above, and plated out onto tissue culture flasks. The homogeneity of alveolar macrophages was verified by morphological analysis.
Before stimulation with IFN-␥, the cells were preincubated for 4 h in medium containing reduced (0.5%) HI-FCS (17)(18)(19). For experiments involving the use of apigenin or emodin, the inhibitors were added to the cells 1 h before the addition of IFN-␥ (i.e. pretreatment).
LPL Activity Assay and Northern Blot Analysis-The heparin-releasable LPL activity in conditioned medium was determined as described previously (17). For Northern blots, total RNA was prepared from cells using Tri-Reagent LS (Molecular Research Center) according to the manufacturer's instructions. Samples of RNA (15 g) were size-fractionated by electrophoresis on denaturing agarose gels, transferred onto Hybond Nfp membranes (Amersham Biosciences), and hybridized to radiolabeled LPL or ␤-actin cDNA inserts, as described previously (17)(18)(19).
Preparation of Manipulated LPL Promoter-Reporter DNA Constructs-The deletion series, containing 5Ј-truncations of the LPL promoter that are linked to the luciferase reporter gene in the vector p19, were a generous gift from Dr. J. M. Gimble (34). The other LPL promoter-luciferase DNA constructs were prepared using PCR as described previously (35)(36)(37). The PCRs were carried out using the high fidelity Pwo DNA polymerase (Roche Molecular Biochemicals) to minimize PCR-generated mutations, and this was confirmed by sequencing of all the products. To produce the Ϫ31/ϩ187 promoter-luciferase DNA construct in the pGL2-Basic vector (38), the corresponding LPL promoter construct in p19 (34) was used as a template for PCR using primers F and R that had the XbaI or the XhoI sites included at the 5Ј and 3Ј ends, respectively (5Ј-TCCCACCCGGGGTCACTTAAACAGC-3Ј and 5Ј-CC-CTTCTCGAGCTGCTTTGCTGCT-3Ј, respectively, with the restriction sites shown in bold type). The amplification product was purified, digested with XbaI and XhoI, and subcloned into the pGL2-Basic vector (38). This recombinant plasmid was then used to prepare mutations in the three Sp1 sites using the overlap extension method (36,39). The mutations produced were those of the Sp1 site from ϩ44 to ϩ51 in the antisense strand (GGGCAG; Sp1M44), double mutations of the overla-pping Sp1 sites from ϩ62 to ϩ67 in the antisense strand (GGGCAG), ϩ65 to ϩ71 in the sense strand (CCCTCCC) (Sp1M62/65), and mutations of all three sites (Sp1M44/62/65). The Sp1M44 and Sp1M62/65 were prepared initially, and the latter was then used as a template to produce the triple Sp1 mutant (Sp1M44/62/65). The oligonucleotide primers used to generate these mutations were as follows, in which the mutated bases are shown in bold type: 5Ј-GAGGAATTTTGTTTCCTG-TAA-3Ј (Sp1M44F) and 5Ј-ACAGTTACAGGAAACAAAATTCC-3Ј (Sp1M44R), and 5Ј-GTAACTGTTCTGAAATAAACTTTAAA-3Ј (Sp1M62/65F) and 5Ј-GTCAACCTTTAAAGTTTATTTCAGAACA-3Ј (Sp1M62/65R). These primers were used in conjunction with primers F and R detailed above. The PCR products were subcloned into the pGL2-Basic vector (38) as described above.
Transient Transfection Assays-U937 cells from the third to the seventh passage were transfected with recombinant plasmid DNA using Superfect TM (Qiagen). A day before transfection, the cells were harvested by centrifugation (1000 ϫ g for 5 min), resuspended in fresh medium, and plated out at a density of 3 ϫ 10 5 /ml. On the day of the transfection, these cells were centrifuged as above, resuspended in RPMI containing only 3% (v/v) HI-FCS, subcultured into 6-well plates at a density of 1 ϫ 10 6 cells per well, and incubated for 4 h. Complexes of DNA, containing 2 g of appropriate LPL promoter-luciferase DNA construct and 0.5 g of cytomegalovirus-␤-galactosidase plasmid as an internal control for transfection efficiency (35)(36)(37), and 8 l of Superfect TM were then prepared as described by the manufacturer and added to the cells. These were then differentiated into macrophages for 12 h using phorbol 12-myristate 13-acetate (PMA; 1 M) as the differentiation agent, either in the absence or the presence of IFN-␥ (1,000 units/ ml). The luciferase and the ␤-galactosidase activities in the cell extracts were then determined using commercially available kits (Promega). The luciferase activity was normalized to the ␤-galactosidase values (35)(36)(37), with each transfection being carried out in triplicate and repeated at least three times.
Western Blot Analysis-Nuclear and whole cell extracts were prepared essentially as described previously (36, 40 -42). Protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 10 g/ml aprotinin, 10 g/ml I-S soybean trypsin inhibitor) and dithiothreitol (0.5 mM) were added to all the buffers before use. The concentration of proteins in the extracts was determined using the microBCA protein assay kit as described by the manufacturer (Pierce).
Samples (10 g) were size-fractionated using 7.5% (w/v) polyacrylamide gels, containing SDS under reducing conditions, and transferred by blotting to Immobilon-P polyvinylidene difluoride membranes (Millipore) (17,40). Blotted membranes were first incubated for 1 h at room temperature in 10% (w/v) non-fat milk powder to reduce any nonspecific interaction of the antisera with the membrane. Following washing with TBS-Tween (1% v/v), the membrane was incubated with the primary antibody for 30 min in 0.5% (w/v) non-fat milk powder/TBS-Tween (0.1% v/v). After another wash with TBS-Tween, the membrane was incubated, as above, with secondary horseradish peroxidase-conjugated antibodies (anti-rabbit or goat immunoglobulin) for 30 min at room temperature. After washing once more with TBS-Tween, the membranes were developed using an enhanced chemiluminescence detection kit (Amersham Biosciences) and XAR-sensitive film (Eastman Kodak). The sizes of the proteins were determined by comparison with Rainbow molecular weight markers (Amersham Biosciences) that had been subjected to electrophoresis and blotting on the same gel as the test samples.
The oligonucleotides were radiolabeled by "fill-in" reactions using [␣-32 P]dCTP and Klenow DNA polymerase. Nuclear and whole cell extracts were prepared as described previously (36, 40 -42). Then 5-20 g of whole cell extracts or 4 g of nuclear extracts were incubated in a total reaction volume of 20 l containing 34 mM potassium chloride, 5 mM magnesium chloride, 0.1 mM dithiothreitol, and 3 g of poly(dI-dC). After 10 min on ice, 32 P-labeled probes (50,000 cpm) were added, and the incubation was continued for 30 min at room temperature. Following the addition of 5 l of a 20% (w/v) Ficoll solution to each sample, the free probe and the DNA-protein complexes were resolved on 4% (w/v) polyacrylamide gels in 0.5ϫ TBE (45 mM Tris base, 45 mM boric acid, 1 mM EDTA). The gels were then dried under vacuum and exposed to x-ray film. For antibody supershift assays, samples of nuclear or whole cell extracts were incubated with the appropriate antiserum for 30 min on ice prior to the addition of the radiolabeled probe (36, 40 -42, 45). For competition assays, a 100 -500-fold molar excess of the double-stranded oligonucleotides were added to the samples of nuclear or whole cell extracts prior to the addition of the radiolabeled probe (36, 40 -45).

IFN-␥ Suppresses LPL mRNA Expression and Enzymatic
Activity in a Range of Macrophage Sources-We have shown previously (17) that IFN-␥ produces dose-dependent reductions in the enzymatic activity, mRNA expression, and protein levels of LPL in the murine macrophage J774.2 cell line. In addition, we have shown that IFN-␥ produces a time-dependent reduction in LPL enzymatic activity in these cells (17). To investigate whether there were similar changes at the level of LPL mRNA expression, time course Northern blot analysis was carried out. As shown in Fig. 1A, a time-dependent decrease in steady state LPL mRNA levels were observed with the profile and overall extent of decrease being similar to that seen at the level of enzymatic activity (17).
The IFN-␥-mediated decrease in LPL activity and mRNA expression in J774.2 macrophages was of ϳ60% and seen only after an incubation period of 24 h ( Fig. 1A and Ref. 17). To evaluate whether a similar profile also occurred in other macrophage sources, the dose-response experiments on LPL enzymatic activity were repeated with primary cultures of rat alveolar macrophages using the rat alveolar NR8383 cell line for comparison. As shown in Fig. 1, B and C, IFN-␥ was more potent at decreasing LPL activity in these cells compared with J774.2 macrophages, and additionally, the maximal reduction in activity that was produced was greater than 80%. These results, along with previous studies (46 -47) in other laboratories using human monocyte-derived macrophages, demonstrate that IFN-␥ decreases LPL gene expression in macrophages from a range of species and origin.
Identification of the Minimal Region in the LPL Promoter That Is Required for the IFN-␥ Response-We have shown previously (31) that the IFN-␥-mediated decrease in LPL ex-pression was due to a reduction in gene transcription rather than any changes in mRNA stability. To evaluate whether the LPL promoter contained sufficient information for this response, transient transfection experiments were initiated using a LPL promoter-luciferase DNA construct that contains the Ϫ1824/ϩ187 region in the p19 vector (34). Initial experiments using J774.2 macrophages showed that, similar to the experiences in other laboratory (48), these cells could not be transfected efficiently with exogenous DNA. We therefore tested a range of monocyte/macrophage cell lines and found that the human U937 myeloid leukemic cell line could be transfected most efficiently with DNA. Indeed, these cells have been used widely to investigate the regulation of macrophage gene transcription (49 -51). The LPL promoter was induced dramatically when the transfected monocytes were differentiated into macrophages using PMA (data not shown), which is consistent Northern blot of total RNA was then probed sequentially with radiolabeled LPL or ␤-actin cDNA insert as described under "Experimental Procedures." The LPL:␤-actin ratio in unstimulated cells has been assigned as 100%, with the ratio for the remaining cell samples being represented relative to this control value (designated as Ratio (%) in the figure). The data shown are representative of two independent experimental series. B and C, the rat alveolar NR8383 cell line (B) and primary cultures of rat alveolar macrophages (C) were exposed for 24 h with different concentrations of IFN-␥ as shown. The heparin-releasable LPL activity was then determined as described under "Experimental Procedures." The LPL activity at each time point is represented as a percentage of the activity in the medium from untreated control cells (assigned as 100%). The values represent mean Ϯ S.D. from three independent experiments. with previous reports (52,53) that LPL is expressed at virtually undetectable levels in monocytes and induced transiently at the transcriptional level during their differentiation into macrophages. When the transfected monocytes were differentiated with PMA for 12 h in the presence of IFN-␥, an ϳ65% reduction in LPL promoter activity was observed compared with cells that were only treated with PMA (Fig. 2, construct Ϫ1824/ϩ187). A similar reduction in LPL promoter activity was also seen when the transfected cells were first differentiated for 12 h in the presence of PMA and then exposed to IFN-␥ for 12 h (data not shown). These results therefore show that the Ϫ1824/ϩ187 LPL promoter sequence contains sufficient information for the IFN-␥ response. We decided to investigate the mechanisms that are involved in this action of IFN-␥ in more detail.
To map the IFN-␥-responsive elements (IFN-␥REs) in the LPL promoter, a 5Ј-deletion series, containing a common 3Ј end at position ϩ187, was transfected into U937 cells, and the relative LPL promoter activity was determined in the absence or the presence of IFN-␥. Deletion of the Ϫ1824 to the Ϫ101 region had little effect on the basal promoter activity obtained in the absence of IFN-␥. A further deletion to Ϫ54 produced an approximate halving of this activity, which was not affected further by a truncation to Ϫ31, thereby indicating the existence of important regulatory elements between the Ϫ101 to the Ϫ54 region that are essential for basal promoter activity. However, an IFN-␥-mediated reduction in LPL promoter activity of between 52 and 67%, which was comparable with the decrease in endogenous LPL mRNA levels produced in J774.2 macrophages by this cytokine (Ref. 17 and Fig. 1A), was seen with all the LPL promoter-luciferase DNA constructs used, including the Ϫ31/ϩ187 region (Fig. 2). This suggests that the Ϫ31 to ϩ187 LPL promoter region contains the minimal IFN-␥REs and was therefore investigated in detail. However, instead of mapping further the precise sequences that are involved in the IFN-␥ response by analyzing the effect of finer deletions or specific mutations in the Ϫ31/ϩ187 region, we decided to first investigate the interaction of proteins with this promoter region by EMSA using extracts from cells that were either untreated or exposed to this cytokine. It was hoped that such a strategy would identify sub-regions in the Ϫ31 to ϩ187 sequence to which the interaction of DNA-binding proteins is affected following exposure of the cells to IFN-␥ and therefore form the foundation for further detailed investigation. In addition, EMSA would allow analysis of the action of IFN-␥ in both J774.2 macrophages, where the original detailed studies on the action of IFN-␥ were carried out Ref. 17 and Fig. 1A), and U937 cells that were used for the transfection studies (Fig. 2).
The Binding of Proteins to Two Regions Within the Ϫ31/ ϩ187 Sequence Is Reduced Following Exposure of the Cells to IFN-␥-EMSA were carried out using six double-stranded oligonucleotides that spanned the Ϫ31/ϩ187 region (Ϫ31/ϩ8, ϩ9/ ϩ49, ϩ46/ϩ90, ϩ88/ϩ118, ϩ119/ϩ160, and ϩ159/ϩ195). Initial experiments using both nuclear and whole cell extracts from J774.2 macrophages showed a similar DNA-protein interaction pattern (data not shown), and the latter were therefore used for all subsequent experiments. Using the ϩ88/ϩ118, ϩ119/ϩ160, and ϩ159/ϩ195 oligonucleotides, no DNA-protein complexes could be detected even after long autoradiographic exposures (data not shown). In contrast specific DNA-protein complexes were seen with the other three oligonucleotides with binding to the ϩ9/ϩ49 and ϩ46/ϩ90 region being consistently reduced when extracts were used from cells that were treated with IFN-␥ ( Fig. 3A; other data not shown). Competition EMSA were carried out to evaluate whether the binding of proteins to the ϩ9/ϩ49 and ϩ46/ϩ90 region was specific and whether these two regions interacted with identical or distinct proteins. Fig. 3B shows the results with the ϩ46/ϩ90 oligonucleotide with a similar profile being seen with the ϩ9/ϩ49 region. Thus, the DNA protein complexes could be competed by an excess of oligonucleotide containing the self ϩ46/ϩ90 region but not by the Ϫ31/ϩ8 or ϩ88/ϩ188 sequence. Interestingly, competition was also obtained using the ϩ9/ϩ49 sequence, thereby indicating that the ϩ9/ϩ49 and ϩ46/ϩ90 regions bound identical or related proteins (Fig. 3B).
To determine the time course of the IFN-␥-mediated decrease in binding to the ϩ9/ϩ49 and the ϩ46/ϩ90 region, EMSA were carried out using extracts from J774.2 macrophages that were exposed to IFN-␥ for 15 min and 3, 6, 12, and 24 h with untreated cells at the start and the end of the experiment being used as controls (0 and 24h-). As shown in Fig. 3C, a decrease in binding of proteins to both regions was seen after exposure of the cells for 12 h with maximal reduction being attained at 24 h. Although a slight reduction in DNA binding was seen with the ϩ46/ϩ90 region in this experiment using extracts from untreated cells at 24 h, this was not reproducible, and in any case, the binding was substantially greater than that seen for cells exposed to IFN-␥ for 24 h. The kinetics of the IFN-␥-mediated decreases in DNA binding to the ϩ9/ϩ49 and the ϩ46/ϩ90 regions were also comparable with the decrease in endogenous LPL mRNA expression seen in J774.2 macrophages (Fig. 1A). A similar IFN-␥-mediated reduction in DNA binding was also observed when the EMSA were repeated using extracts from the U937 cell line that were exposed to PMA for 12 and 24 h, either in the absence or the presence of this cytokine (data not shown). Thus, the IFN-␥-mediated reduction in binding of factors to the ϩ9/ϩ49 and ϩ46/ϩ90 sequence was seen with extracts from two distinct macrophage sources that are derived from different species.
Sp1 and Sp3 Interact with the IFN-␥RE in the LPL Promoter-A computer analysis of the ϩ9 to ϩ90 region of the promoter using the GCG and the MatInspector version 2.2 data bases (54) showed the presence of putative consensus sites for four factors that have been shown previously to regulate macrophage gene expression, C/EBP, PU.1, STAT, and Sp1. In order to investigate whether any of these factors interacted with the IFN-␥REs in the LPL promoter, competition EMSA experiments were carried out using an excess of oligonucleo- FIG. 2. Identification of the minimal IFN-␥REs in the LPL promoter. Transient transfection assays with the indicated LPL promoterluciferase DNA construct or the parent p19Luc plasmid was carried out as described under "Experimental Procedures." The transfected U937 monocytes were differentiated into macrophages for 12 h, either in the absence or the presence of IFN-␥. The luciferase activity was normalized to the ␤-galactosidase activity and is represented as relative luciferase activity (RLA). Numbers above the histograms show the IFN-␥mediated percentage reduction in this activity compared with that seen for the construct in untreated cells. Each value is the result of at least four independent experiments. tides containing binding sites for these factors with the AP-1binding site oligonucleotide being employed as a nonspecific competitor. As shown in Fig. 4A, the DNA-protein complexes formed using the ϩ9/ϩ49 and the ϩ46/ϩ90 regions could be competed using an excess of an Sp1-binding site oligonucleotide but not by the other sequences, thereby indicating that a Sp1like factor bound to the IFN-␥REs in the LPL promoter.
Sp1 was the first identified member of what is now a family of transcription factors in which Sp1 and Sp3 represent the major and most extensively characterized members that tend to be co-expressed in a number of tissues/cell types and able to interact with the same recognition sequence in many gene promoters (55). To further investigate whether Sp1 and Sp3 also interacted with the ϩ9/ϩ49 and ϩ46/ϩ90 regions, antibody supershift experiments were carried out using specific antisera against Sp1 and Sp3. Non-immune serum and that for C/EBP␤, which has been used extensively for previous studies in the laboratory (40, 42, 45), were included as controls. As shown in Fig. 4B, the production of a slower migrating, antibody-protein-DNA "super-shift" complex was seen with anti-sera against Sp1 and Sp3 but not C/EBP␤ or non-immune serum, thereby indicating that both members interact with the LPL promoter. A closer examination of the data reveals the existence of three DNA-protein complexes, with complex C1 consisting predominantly of Sp1, whereas complexes C2 and C3 were composed mainly of Sp3.

IFN-␥ Decreases the Binding of Sp1 and Sp3 to a Number of Recognition
Sequences-Although GGGGCGGGG (GC element) and CACCC (or GGGTG) boxes have been proposed as consensus Sp1/Sp3-binding sites, the factors have also been shown to interact with sequences divergent from the consensus (55). For example, there is an additional Sp1/Sp3 site in the LPL promoter (position Ϫ91 to Ϫ83) that contains a CT-rich sequence (5Ј-CCTCCCCCC-3Ј) (56). We wondered whether the IFN-␥-mediated decrease in DNA binding seen in EMSA was specific to the ϩ9/ϩ90 LPL promoter sequence or could also be seen with a number of other Sp1/Sp3 recognition sequences. EMSA were therefore carried out using extracts from untreated or IFN-␥-stimulated J774.2 macrophages and several different Sp1/Sp3-binding sites. These included the upstream CT-rich sequence from the LPL promoter (56), the downstream sites in the ϩ9/ϩ49 and ϩ46/ϩ90 sequences, and the consensus Sp1 site (contains the GC element). In addition, the Sp1 site FIG. 3. Interactions of DNA-binding proteins with sequences in the ؊31/؉187 region. A, EMSA was carried out using radiolabeled oligonucleotides against the ϩ9/ϩ49 and ϩ46/ϩ90 sequence and extracts from J774.2 macrophages that were either left untreated (ϪIFN-␥) or exposed to the cytokine for 24 h (ϩIFN-␥). B, competition experiments were performed using radiolabeled ϩ46/ϩ90 sequence and extracts from untreated cells (ϪIFN-␥) in the presence of a 500-fold molar excess of double-stranded oligonucleotides against the Ϫ31/ϩ8, ϩ9/ϩ49, ϩ46/ϩ90 and ϩ88/ϩ118 sequence, as indicated. ϩIFN-␥ represents the profile obtained with extracts from cells exposed to this cytokine for 24 h. C, EMSA were carried out using the ϩ9/ϩ49 and the ϩ46/ϩ90 oligonucleotides and extracts from J774.2 macrophages that were exposed to IFN-␥ for 15 min and 3, 6, 12, and 24 h. Extracts from untreated cells at the start and the end of the experiment were included for comparison (0 and 24 h, respectively). P represents the profile obtained with the free probe alone, which has migrated off the gel, and the DNA-protein complexes are shown by the vertical line labeled C. The results are representative of three independent experimental series.
FIG. 4. Competition EMSA and antibody supershift experiments. EMSA were carried out using radiolabeled oligonucleotides against the ϩ9/ϩ49 and ϩ46/ϩ90 sequence and extracts from untreated J774.2 macrophages in the absence (Ϫ) or the presence of either a 100-or 500-fold molar excess (ϫ100 and ϫ500, respectively) of oligonucleotides containing binding sites for C/EBP, PU.1, STAT, Sp1, and AP1 (A) or antisera against Sp1, Sp3, or C/EBP, as indicated, or the non-immune serum (NI) (B). In both A and B, lanes labeled P and ϩ show the profile obtained with the free probe, which has migrated off the gel, and extracts from J774.2 macrophages exposed to IFN-␥ for 24 h, respectively. C1-C3 and ss represent the positions of the DNAprotein complexes and antibody-DNA-protein supershift complex, respectively. from the C/EBP␣ gene promoter was included because the expression of this gene in macrophages was also suppressed by IFN-␥ with kinetics that are similar to that for the LPL gene (57), and its promoter region also contains important binding sites for the Sp1 family members (36, 58 -59). Fig. 5 shows that exposure of the cells to IFN-␥ reduced the binding of factors to all the Sp1/Sp3 recognition sequences investigated. However, some subtle differences were identified. For example, a faster migrating complex, whose intensity was increased in extracts from IFN-␥-treated cells, was seen with the consensus Sp1binding site (indicated by an asterisk in Fig. 5). Additionally, the signals obtained from the binding of factors to the upstream CT element in the LPL promoter was less intense compared with the other probes despite an equal amount of input radioactivity.
Mutations of All Three Sp1/Sp3 Sites in the ϩ9 to ϩ90 Region Decrease Basal Promoter Activity and Abolish the IFN-␥ Response-The ϩ9/ϩ90 region contains three Sp1/Sp3 sites centered at positions ϩ44 to ϩ51 (antisense strand), ϩ62 to ϩ67 (antisense strand), and ϩ65 to ϩ71 (sense strand) that are highly conserved between the mouse, human, and rat LPL gene promoters (Fig. 6A). To evaluate the importance of these sites in the IFN-␥ response, three DNA constructs were prepared that contained mutations in these Sp1/Sp3 sites, in the Ϫ31/ ϩ187 context, as follows: (i) mutation in the site at ϩ44 to ϩ51 (Sp1 M44); (ii) mutations in the overlapping sites at ϩ62 to ϩ67 and ϩ65 to ϩ71 (SpM62/65); and (iii) mutations in all three sites (Sp1M44/62/65) (see Fig. 6A). The DNA constructs were then transfected into U937 macrophages, and the relative luciferase activity in the absence or the presence of IFN-␥ was determined. Mutations of these Sp1 sites not only produced a reduction of basal promoter activity but also abolished the IFN-␥ response (Fig. 6B). Interestingly, although mutation of the Sp1 sites at ϩ62 and ϩ65 leads to a reduction of basal LPL activity that was similar to that obtained using the promoterless pGL2-Basic vector, the activity of the DNA construct con-taining mutations of all three Sp1/Sp3 sites (ϩ44, ϩ62 and ϩ65) was greater, with the values obtained in the presence of IFN-␥ also being slightly higher than that seen with the wildtype Ϫ31/ϩ188 construct in the absence of the cytokine. Although the precise reason(s) for such changes are currently unclear, it is possible that some complex interactions may occur between Sp1/Sp3 that is bound to the three sites. The most important conclusion is clearly that mutation of these Sp1 sites leads to an abolition of the IFN-␥ response.
The Sp1/Sp3 Sites in the LPL Promoter Can Impart the IFN-␥ Response to a Heterologous Promoter-To evaluate whether the Sp1/Sp3 sites in the LPL promoter can impart the IFN-␥ response to a heterologous promoter, four copies of the sites centered at positions ϩ44 and ϩ62/65 were linked to the minimal SV40 promoter in the pGL2-promoter vector. A similar construct was also prepared using the ϩ38 to ϩ75 region that harbors all three Sp1/Sp3 sequences. The DNA constructs were then transfected into U937 cells, and the relative luciferase activity was determined in the absence or the presence of FIG. 5. The action of IFN-␥ on the binding of factors to the different Sp1/Sp3 recognition sequences. EMSAs were carried out using extracts from J774.2 macrophages that have been either left untreated (Ϫ) or exposed to IFN-␥ for 24 h (ϩ) and the following radiolabeled oligonucleotides: ϩ9/ϩ49 and ϩ46/ϩ90 region of the LPL gene, consensus Sp1-binding site (Sp1), and Sp1-binding sites from the promoter region of the LPL (CT-rich element) and C/EBP␣ genes (Pro Sp1 and C/EBP Sp1, respectively). P represents the profile obtained with the radiolabeled probe alone, and the DNA-protein complexes are shown by a vertical line labeled C. Asterisk indicates the position of a DNA-protein complex obtained with the consensus Sp1 oligonucleotide whose intensity increases following exposure of the cells to IFN-␥. The results are representative of two independent experiments.
FIG. 6. Analysis of the importance of the Sp1/Sp3 sites in the ؊31/؉187 sequence in the LPL promoter. A, alignment of the ϩ35 to ϩ75 sequence of the mouse LPL promoter with the corresponding sequence of the human and the rat promoter. The bases of the Sp1 sites are indicated in lowercase letters. The sequences of the three Sp1 mutants in the ϩ38 to ϩ75 context are also shown with the mutated base underlined. B and C, transfection experiments using the various manipulated LPL promoter-luciferase DNA constructs (see text for details). These were transfected into U937 cells, and the relative luciferase activity (RLA) in the absence or the presence of IFN-␥ was determined as described under "Experimental Procedures" (each value represents the mean Ϯ S.D. from three independent experiments). C, the activity of each construct in untreated cells has arbitrarily been assigned as 100% with that from IFN-␥ treated cells being represented as a percentage of this value. The IFN-␥ response seen with each of the construct was significant with p Ͻ 0.005. IFN-␥. The values obtained using the pGL2-promoter vector were subtracted from those obtained using the various LPL constructs. As shown in Fig. 6C, an IFN-␥-mediated reduction in reporter gene activity was obtained with all the LPL promoter sequences used, with maximal reduction being seen when all the three Sp1 sites were present in their normal context.
The Steady State Levels of Sp3 but Not Sp1 Are Reduced in Macrophages Following Exposure to IFN-␥-The IFN-␥-mediated reduction in binding of Sp1/Sp3 to the LPL promoter region may either be because of a decrease in the steady state levels of the corresponding proteins and/or a cytokine-mediated suppression of DNA binding activity. To investigate the former possibility, Western blot analysis was carried out using J774.2 macrophages that were either left untreated or exposed to IFN-␥ for various times. Fig. 7 shows a representative result from four independent experiments. The description of the trend in Sp1 and Sp3 expression described here is based on all four experiments. A single immunoreactive complex was seen with antisera against Sp1, the levels of which were not affected by exposure of the cells to IFN-␥. Antisera against Sp3 detected four complexes that formed two doublets with approximate molecular masses of 115 and 70 kDa (Fig. 7). Both alternative use of translation initiation codons and post-translational modifications may account for the four polypeptides. Indeed, Sp3 mRNA has been shown previously to specify for three polypeptides through alternative use of translation initiation codon (55,60). The full-length 115-kDa Sp3 is initiated at a non-AUG codon, whereas two smaller species of ϳ70 kDa arise from internal translational initiation sites (55,60). In addition, the Sp1 family has been shown to undergo phosphorylation and glycosylation (55). In contrast to the steady state levels of Sp1, IFN-␥ produces a time-dependent decrease of all four Sp3 polypeptides (Fig. 7). The decrease started soon after incubation of the cells with the cytokine for 6 h and maximal reduction was seen at 24 h. The kinetics of reduction of Sp3 protein levels are therefore similar to those at the level of DNA binding activity (Fig. 3C) and LPL mRNA expression (Fig. 1A). This suggests that a decrease in Sp3 expression makes a major contribution to the IFN-␥-mediated reduction in LPL mRNA expression.

Casein Kinase 2 (CK2) Is Involved in the IFN-␥-mediated Changes in the Binding of Sp1 and Sp3 to the LPL Promoter-
The outcome from Western blot analysis (Fig. 7) showed that IFN-␥ has no effect on the steady state levels of the Sp1 protein and therefore reduces its DNA binding activity by a posttranslational mechanism. Sp1 is known to be subject to two different forms of post-translational modification, glycosylation and phosphorylation (55). Glycosylation does not appear to affect the ability of the factor to bind DNA but instead influences the transactivation potential and the degradation of the protein (61)(62)(63). In contrast, phosphorylation has a profound effect on the DNA binding activity of Sp1 (55). Interestingly, the Sp1 DNA binding activity has been shown to decrease following CK2-mediated phosphorylation of the protein (64,65). It was therefore possible that the IFN-␥-mediated reduction in the binding of Sp1 to its recognition sequence in the LPL promoter was also mediated through a CK2-mediated phosphorylation. We investigated this possibility using the selective CK2 inhibitors, apigenin and emodin. These inhibitors have been used extensively to investigate the role of the CK2 signal transduction pathway (66 -70). Our hypothesis was that if the IFN-␥-mediated decrease in the binding of Sp1 to the LPL promoter was due to phosphorylation by CK2 then this should be prevented, at least in part, in the presence of apigenin or emodin. This hypothesis was tested by EMSA using extracts from untreated J774.2 macrophages or those incubated with IFN-␥ in the absence or the presence of apigenin or emodin. The IFN-␥-mediated reduction in the binding of Sp1 to the ϩ9/ϩ49 and the ϩ46/ϩ90 regions of the LPL promoter was prevented in a dose-dependent manner by both inhibitors (complex C1 in Fig. 8; other data not shown), thereby providing a substantive link between CK2 and changes in Sp1 DNA binding activity. Furthermore, the results also show that the reduction in Sp3 binding, which was due to decreased levels of steady state polypeptides (see Fig. 7), was also prevented in the presence of the two inhibitors (complex C2 and C3 in Fig. 8; other data not shown). Thus, CK2 was also involved in the IFN-␥mediated reduction in the steady state levels of Sp3 polypeptides, which could be due to either their increased degradation or decreased synthesis. More recently, we have confirmed that IFN-␥ induces CK2 kinase activity in J774.2 macrophages, and this is abolished in the presence of the inhibitors. EMSAs were carried out using radiolabeled oligonucleotides against the ϩ9/ϩ49 and the ϩ46/ϩ90 sequence and extracts from J774.2 macrophages that were either left untreated or exposed to IFN-␥ for 24 h in the absence or the presence of apigenin, as indicated. The DNA-protein complexes are shown by labeled arrows. P represents the profile obtained with the free probe, which has migrated off the gel. The data are representative of three independent experiments.

DISCUSSION
We report in this paper studies on the mechanisms responsible for the IFN-␥-mediated suppression of macrophage LPL gene transcription. Promoter-dissection experiments revealed that the Ϫ31/ϩ187 region contained the IFN-␥REs (Fig. 2). EMSA identified the interaction of proteins to two sub-regions (ϩ9/ϩ49 and ϩ46/ϩ90), the binding of which was reduced dramatically following incubation of the cells with the cytokine (Fig. 3). These two regions contained three evolutionarily conserved Sp1 recognition sequences (Fig. 6A), and both competition EMSA and antibody supershift experiments revealed the interaction of Sp1 and Sp3 with these sites (Fig. 4). The binding of factors to a number of other Sp1/Sp3 recognition sequences was also found to decrease following incubation of the cells with IFN-␥ (Fig. 5). This IFN-␥-mediated suppression was mediated, at least in part, by a reduction of Sp3 polypeptide levels and a decrease in the binding of Sp1 without any changes in the protein levels (Fig. 7). CK2 was found to be involved in the regulation of both Sp1 and Sp3 (Fig. 8).
Sp1 was originally identified as a ubiquitous transcription factor that was implicated in the constitutive expression of several genes (55). However, recent studies (55) have revealed the existence of an Sp1 family, with Sp1 and Sp3 being characterized extensively and known to be co-expressed in several tissues/cell types and to interact with an identical consensus sequence. In addition, the Sp1 family has now been shown to be involved in inducible gene transcription that includes responses to glucose (71,72), serum (73), epidermal growth factor (74), platelet-derived growth factor (75), and transforming growth factor-␤ (76,77). Furthermore, recent work has revealed that Sp1 plays a prominent role in the regulation of many genes in macrophages, including urokinase-type plasminogen activator receptor (51), interleukin-10 (78), hematopoietic cell kinase (79), acid sphingomyelinase (80), lysosomal acid lipase (81), CCAAT-enhancer binding protein (C/EBP)-␤ (50), carboxyesterase (82), proto-oncogene c-fes (83), myeloid integrin CD11b (49), and the membrane glycoprotein CD14 (84). However, the precise role of the Sp1 family in the suppression of gene transcription has hitherto not been investigated in detail. The findings in this paper therefore identify a novel action of the Sp1 family in the IFN-␥-mediated suppression of macrophage LPL gene transcription. Such a mechanism may be widely applicable to other IFN-␥-regulated genes and indicate a need for further detailed studies. Indeed, Fig. 5 and other experiments in our laboratory indicate that such a mechanism may also be involved in the IFN-␥-mediated transcriptional suppression of the C/EBP␣ gene, 3 the promoter region of which contains important binding site(s) for members of the Sp1 family of transcription factors (36,58,59).
The importance of Sp1 and Sp3 in the constitutive expression of LPL has been reported previously (56,85) in relation to a common, naturally occurring Ϫ93T/G transition that is associated with reduced promoter activity. Sp1 and Sp3 were shown to bind to an evolutionarily conserved CT element between positions Ϫ91 and Ϫ83 and activate transcription in THP-1 macrophages (56). In addition, a synergistic action of Sp1/Sp3 and sterol regulatory element-binding protein (SREBP)-1 was observed, which may provide a mechanism for cross-talk between cholesterol and triglyceride metabolic pathways (86). This Sp1/ Sp3 site may have also contributed to the halving of the basal activity with deletion of the Ϫ101 to Ϫ54 LPL promoter region seen in this study (Fig. 2). However, this site was not involved in the IFN-␥ response because its deletion does not abolish the cytokine-mediated suppression of LPL promoter activity (Fig. 2). Although Sp1 acts exclusively as an activator of gene transcription, Sp3 contains a transcriptionally repressive domain and can act as a transcriptional activator or a repressor, depending on the promoter and cell type (55,87). Thus, changes in the Sp1 to Sp3 ratio may represent one mechanism in the regulation of gene transcription. However, such a mechanism is clearly not involved in the IFN-␥-mediated suppression of LPL gene transcription because Western blot analysis (Fig. 7) shows that the cytokine decreases Sp3 levels. In addition, previous studies (56) have shown that both Sp1 and Sp3 are able to activate the LPL promoter in macrophages. The results presented in this paper instead suggest the existence of two distinct mechanisms for IFN-␥ action, both of which require the protein kinase CK2 (Fig. 8). First, the cytokine produces a time-dependent reduction in the abundance of Sp3 polypeptides (Fig. 7). Whether such a reduction is produced via decreased synthesis or increased degradation remains to be determined. It is possible that similar to Sp1 (55, 88), Sp3 also undergoes a proteosome-mediated degradation in response to phosphorylation by CK2. Second, although the abundance of Sp1 does not change following incubation of the cells with IFN-␥, the DNA binding activity of complex C1 in EMSA, which is composed of this factor as judged by antibody supershift assays (see Fig. 4B), decreases. Thus, an IFN-␥-mediated phosphorylation of Sp1 via CK2 is likely to be responsible for its reduced binding.
CK2 is a serine/threonine kinase, which is ubiquitously expressed in both the cytoplasm and the nucleus of eukaryotic cells, and exists as a tetramer composed of two larger catalytic subunits (␣ and/or ␣Ј; 37-44 kDa) and two smaller regulatory ␤ subunits (24 -28 kDa) (89). The enzyme phosphorylates serine or threonine residues in acidic domains, with (S/T)XX(D/E) being the canonical motif (89). The primary sequence of both Sp1 and Sp3 contains a number of consensus sites for CK2 (data not shown). Whether Sp3 is directly phosphorylated by CK2 remains to be determined. For Sp1, however, the CK2 consensus sequence at amino acid 579 in the second zinc finger motif has been shown to be phosphorylated by the enzyme, and this results in a decrease in its DNA binding activity (64). Such CK2-mediated phosphorylation has been implicated in the decrease in Sp1 binding activity during terminal differentiation of the liver and in the transcriptional attenuation of two genes encoding the D-site-binding protein and acetyl-coenzyme A carboxylase (64,65,90).
Binding sites for Sp1 members are present in the promoter regions of a large number of class II genes (55). Despite this, such genes are subject to differential regulation by IFN-␥, with the cytokine either inducing or inhibiting their expression or having no effect (see below). For example, we have shown previously (57,91) that IFN-␥ induces the expression of the C/EBP␤, C/EBP␦, and c-jun genes in J774.2 macrophages, albeit with different kinetics and magnitude of activation. Although the promoter regions of these genes each contain important Sp1 recognition sequence(s) (50,(92)(93)(94), they are up-regulated by IFN-␥ in the same cellular system where this cytokine suppresses the expression of the LPL gene. These findings therefore raise questions on the mechanisms that are responsible for such differential gene regulation by IFN-␥ and the reason(s) why the cytokine does not produce a global reduction in the expression of all class II genes whose promoters bind Sp1 members. The Sp1 family has been shown to interact with sequences that are quite diverged from the proposed consensus (GC element and CACCC boxes) (55). For example, Sp1 interacts with a CT-rich sequence in the promoter regions of the LPL, LDL receptor, and c-myc genes (56,(95)(96). It is therefore possible that variations in the Sp1-binding sites between dif-ferent promoters may be responsible for the differential action of IFN-␥. Thus, the affinity of Sp1 members for the various sites may be different, and this could be regulated further by the cytokine. However, EMSA failed to reveal any gross differences in the binding profiles of several such sites when extracts were used from J774.2 macrophages that are either left untreated or exposed to IFN-␥ for 24 h (Fig. 5).
The most likely explanation for the differential action of IFN-␥ in the regulation of genes whose promoters contain Sp1 recognition sequence is therefore the presence of binding sites for other transcription factors that play a more prominent role in the response. Indeed, both promoter-dissection and DNAprotein interaction studies on the C/EBP␤, C/EBP␦, and c-jun genes (which we have found to be up-regulated in J774.2 macrophages by IFN-␥ (57, 91)) contain regulatory sites for Sp1 that act in concert with other factors: CREB/ATF (C/EBP␤) (50), STAT-1 (C/EBP␦) (92), and CTF and AP1 (c-Jun) (93,94). Similarly, the regulatory sequences of other IFN-␥-activated genes contain binding sites for Sp1 together with either STAT-1 or IRFs, with the latter playing a more prominent role in the cytokine response (97)(98)(99)(100)(101)(102). It should, however, be noted that the kinetics of IFN-␥ action on these genes are different from those for the suppression of LPL, with their transient activation occurring immediately following exposure of the cells to the cytokine, and much earlier than the changes seen in the binding of Sp1/Sp3 to the LPL promoter (97).
In conclusion, we have identified a novel mechanism through which IFN-␥ suppresses macrophage LPL gene transcription. Although IFN-␥ regulation of gene transcription is mainly mediated through Stat1 (97), several recent studies (108,109) have indicated the existence of alternative pathways. The regulation through Sp1 and Sp3, as identified in this study, could represent one such mechanism, at least as far as the suppression of gene transcription is concerned. It is also possible that such a mechanism could extend to the action of other cytokines. For example, tumor necrosis factor-␣ has recently been shown to down-regulate murine hepatic growth hormone receptor expression by inhibiting Sp1 and Sp3 binding (110).