Sp1 is a component of the cytokine-inducible enhancer in the promoter of vascular cell adhesion molecule-1.

Transcription of the vascular cell adhesion molecule-1 (VCAM-1) gene in endothelial cells is induced by the inflammatory cytokines interleukin-1β, tumor necrosis factor-α, and lipopolysaccharide. Previous studies demonstrated that the cytokine-response region in the VCAM1 promoter contains binding sites for the transcription factors nuclear factor-κB (NF-κB) and interferon regulatory factor-1. Using a saturation mutagenesis approach, we report that the cytokine-inducible enhancer consists of these previously characterized elements and a novel region located 3′ of the NF-κB sites. Electrophoretic mobility shift assays and DNase I footprint studies with endothelial nuclear extracts and recombinant protein revealed that the transcriptional activator Sp1 interacts with this novel element in a specific manner. Transient transfection assays using vascular endothelial cells revealed that site-directed mutations in the Sp1 binding element decreased tumor necrosis factor-α-induced activity of the VCAM1 promoter. The cytokine-induced enhancer of the VCAM1 gene requires constitutively bound Sp1 and induced heterodimeric NF-κB for maximal promoter activity.

Transcription of the vascular cell adhesion molecule-1 (VCAM-1) gene in endothelial cells is induced by the inflammatory cytokines interleukin-1␤, tumor necrosis factor-␣, and lipopolysaccharide. Previous studies demonstrated that the cytokine-response region in the VCAM1 promoter contains binding sites for the transcription factors nuclear factor-B (NF-B) and interferon regulatory factor-1. Using a saturation mutagenesis approach, we report that the cytokine-inducible enhancer consists of these previously characterized elements and a novel region located 3 of the NF-B sites. Electrophoretic mobility shift assays and DNase I footprint studies with endothelial nuclear extracts and recombinant protein revealed that the transcriptional activator Sp1 interacts with this novel element in a specific manner. Transient transfection assays using vascular endothelial cells revealed that site-directed mutations in the Sp1 binding element decreased tumor necrosis factor-␣-induced activity of the VCAM1 promoter. The cytokine-induced enhancer of the VCAM1 gene requires constitutively bound Sp1 and induced heterodimeric NF-B for maximal promoter activity.
Vascular cell adhesion molecule-1 (VCAM-1) 1 is a 110-kDa cell surface glycoprotein (Osborn et al., 1989;Rice and Bevilacqua, 1989), which interacts with cells bearing the integrin counter-receptor ␣ 4 ␤ 1 (very late antigen-4 (VLA-4)) and ␣ 4 ␤ 7 (Carlos et al., 1990;Elices et al., 1990;Chan et al., 1992). VCAM-1 is constitutively expressed in antigen-presenting cells (Freedman et al., 1990;Koopman et al., 1991) and embryonic tissues Gurtner et al., 1995). Vascular endothelium in vivo exhibits little or no VCAM-1 expression under normal conditions but can be rapidly induced by lipopolysaccharide (Fries et al., 1993). The endothelial expression of VCAM-1 has been implicated in the recruitment of circulating monocytes and lymphocytes from the bloodstream (reviewed in Carlos and Harlan, 1994;Springer, 1995) and associated with a variety of chronic inflammatory processes (Rice et al., 1991;. VCAM-1 has a distinct pattern of cytokine induction in cultured endothelial cells. Naive cells do not express VCAM-1 messenger RNA; however, exposure of cells to inflammatory mediators or cytokines (lipopolysaccharide, poly(I:C), interleukin-1␤, and tumor necrosis factor-␣ (TNF-␣)) results in rapid up-regulation of mRNA and surface protein within 3 h (Osborn et al., 1989;Read et al., 1994). To study the molecular mechanisms controlling the VCAM-1 response to cytokine, we have characterized the structure of the human VCAM1 gene  and begun analysis of the 5Ј-regulatory region (Neish et al., 1992;Shu et al., 1993;Neish et al., 1995). In previous work, we and others demonstrated that TNF-␣ induction of VCAM1 expression occurs at the transcriptional level and is dependent on tandem NF-B motifs located at positions Ϫ65 and Ϫ75 relative to the single transcriptional start site (Neish et al., 1992;Iademarco et al., 1992). Heterodimeric p50 and p65 (RelA) associate with these elements in the nuclei of endothelial cells exposed to TNF-␣. Moreover, NF-B physically and functionally cooperates with the transcriptional activator IRF-1 to achieve full induction of VCAM1 transcription .
The regulatory regions conferring cytokine-inducible transcription in the leukocyte adhesion molecule genes are composed of multiple binding elements, which are recognized by a surprisingly finite number of transcription factors (reviewed in Collins et al., 1995). To provide a detailed characterizaton of the cytokine-induced transcriptional enhancer in the VCAM1 promoter, we employed a saturation mutagenesis approach to completely map the boundaries of regulatory elements that occur within the VCAM1 cytokine-response region. Using this approach, we have identified a novel Sp1 binding motif in the proximal promoter and have demonstrated that optimal cytokine activation of the VCAM1 gene requires Sp1.

EXPERIMENTAL PROCEDURES
Cell Culture and Cytokine Treatment-For the in vitro mutagenesis study, human endothelial cells were obtained from Clonetics and cultured in endothelial growth media (Clonetics) on Primaria dishes (Falcon). Cells from passage four were used for linker scan transfection studies. In these experiments, endothelial cells were exposed to recombinant TNF-␣ at a concentration of 20 ng/ml for 6 h before harvesting.
For preparation of cell extracts, human umbilical vein endothelial cells (HUVE) were cultured in M199 with 20% fetal bovine serum, 100 mg/ml porcine intestinal heparin, 50 mg/ml endothelial mitogen, 50 units/ml penicillin, 50 mg/ml streptomycin, and 25 mM HEPES in gelatin-coated plates (Gimbrone, 1976). Bovine aortic endothelial cells (BAEC) were isolated and maintained in culture using previously described procedures (Neish et al., 1992). For experiments involving cytokine activation, confluent monolayers of endothelial cells were exposed to recombinant human TNF-␣ (Genentech, San Francisco, CA) at a final concentration of 200 units/ml in complete media for the time * This work was supported in part by National Institutes of Health Grants HL03011-01 (to A. S. N.), HL 35716, HL 45462, and PO136028 (to T. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Plasmid Constructions-A wild-type fragment (Ϫ2150 to ϩ40) of the VCAM1 promoter was generated by the polymerase chain reaction (PCR) using two gene-specific primers and a genomic clone as a template.
Linker-scanning mutants were generated by PCR as follows. For each construct two complementary primers having individual mutations were designed (see below). The coding primer was used in combination with the 3Ј-BglII primer, whereas the non-coding one was used in combination with the 5Ј-SacI primer. Two subfragments were generated, both carrying the mutation at the 5Ј-and 3Ј-ends, respectively. The fragments were gel purified, aliquots of both were mixed, combined with the two primers 5Ј-SacI and 3Ј-BglII, and subjected to the second round of PCR.
The reactions were carried out as recommended in the manufacturer's specifications (Perkin Elmer Corp.). The resulting subfragments were digested with SacI and BglII and ligated into the SacI and BglII sites of pGL2-Basic (Promega). The integrity of the individual mutants was confirmed by DNA sequence analysis.
A fusion gene carrying the region Ϫ258 to ϩ42 of the VCAM1 promoter coupled to a CAT reporter gene (pF2) was described and characterized earlier (Neish et al., 1992). A reporter bearing internal point mutations in the Sp1 site, pF2-mSp1, was made from two separate PCR amplification products as described (Neish et al., 1992). The amplified products were gel purified and subcloned into the SmaI site of the reporter plasmid pCAT3 (Promega).
Transfections and CAT Assays-HUVE transfections were performed using Lipofectin (Life Technologies, Inc.) as per the manufacturer's recommendations and as described previously (Schindler and Baichwal, 1994). Cells were lysed in 150 l of lysis buffer (Promega), and 10 l of the extract were used for the luciferase assay (Promega). Protein concentrations were measured by the Bradford method (Bio-Rad). Transfections were performed in duplicate at least three times.
BAECs were transfected with a modified calcium phosphate technique (Sambrook et al., 1989). Cells were transfected with 10 g of reporter plasmid. Relative transfection efficiency was determined by cotransfection with pTK-GH (5 g) (Nichols Institute Diagnostics, San Juan Capistrano, CA). After transfection, cultures were washed twice with Hanks' balanced salt solution and refed with complete Dulbecco's modified Eagle's media with or without TNF-␣. After 24 h incubation, samples of the medium were collected and assayed for growth hormone by using a commercially available radioimmunoassay kit (Nichols Institute Diagnostics), and cells were harvested. Confluent monolayers were collected by trypsin dissociation, centrifuged (650 ϫ g for 5 min), washed once in phosphate-buffered saline, and resuspended in 200 l of 0.25 M Tris-HCl (pH 7.8) with 10 g/ml aprotinin (Sigma). Cells were lysed by sonication and one freeze-thaw cycle. Supernatants containing the cell extracts were obtained after centrifugation for 10 min at 14,000 ϫ g and stored at Ϫ80°C. Conversion of radiolabeled acetyl coenzyme A to acetylated chloramphenicol was assayed by the two-phase fluor diffusion technique (Sambrook et al., 1989). For each sample, 30 -50 l of cell extract was incubated at 65°C for 15 min to inactivate endogenous transacetylases. The assay was performed in a reaction mixture of 1.25 mM chloramphenicol (Sigma), 125 mM Tris-HCl (pH 7.8), and 0.1 mCi of 3 H-acetyl coenzyme A (DuPont NEN) in a reaction volume of 200 l. The reaction mixture was overlaid with 5 ml of water immiscible scintillation fluid (Econofluor, Dupont NEN) and incubated at 37°C for 2 h. The activity of the 3 H-acetylated chloramphenicol was measured using an LKB RackBeta scintillation counter.
Transfection experiments were repeated at least three times using at least two independent plasmid preparations. A negative control, the promoterless plasmid pCAT3, showed no inducible activity from any experimental manipulation. Preparation of Nuclear Extracts-Following experimental treatment of human umbilical vein endothelial cells, nuclear extracts were prepared as described previously (Read et al., 1994). Monolayers (3-5 ϫ 10 7 cells) were harvested by scraping, washed in cold phosphate-buffered saline, and incubated in two packed cell volumes of buffer A (10 mM HEPES (pH 8.0), 1.5 mM MgC1 2 , 10 mM KCl, 0.5 mM dithiothreitol (DTT), 200 mM sucrose, 0.5 mM phenylmethylsulfonyl fluoride, 1 mg/ml leupeptin and aprotinin, and 0.5% Nonidet P-40) for 5 min at 4°C. The crude nuclei released by lysis were collected by microcentrifugation (15 s), rinsed once in buffer A, and resuspended in two-thirds packed cell volume of buffer C (20 mM HEPES (pH 7.9), 1.5 mM MgCl 2 , 420 mM NaCl, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 1.0 mM DTT, 1.0 g/ml leupeptin and aprotinin). Nuclei were incubated on a rocking platform at 4°C for 30 min and clarified by microcentrifugation for 5 min. The resulting supernatants were diluted 1:1 with buffer D (20 mM HEPES (pH 7.9), 100 mM KCl, 0.2 mM EDTA, 20% glycerol, 1 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml of both leupeptin and aprotinin). Nuclear extracts were frozen on dry ice and stored at Ϫ80°C.
Electrophoretic Mobility Shift Assay (EMSA)-Double-stranded oligonucleotides were end labeled with [␣-32 P]dCTP (50 Ci at 3000 Ci/ mmol, DuPont NEN) and the Klenow fragment of E. coli DNA polymerase I. Binding reactions in 20 l contained 5 g of nuclear extract protein, binding buffer (10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1 mM DTT, 1 mM EDTA, 5% glycerol), 1 g of poly(dI:dC), 1 g of salmon sperm DNA, and 50,000 cpm labeled DNA). Reactions were incubated at 22°C for 20 min and analyzed by electrophoresis on a 4% nondenaturing polyacrylamide gel at 180 V for 2 h in 45 mM Tris borate-EDTA as described by Ausubel (1989). Following electrophoresis, gels were dried, and DNA-protein complexes were localized by autoradiography for 18 h. Competition studies were performed by adding unlabeled double-stranded oligonucleotides to the binding reaction for 10 min prior to the addition of labeled oligonucleotide. For supershift analysis, nuclear extracts from TNF-␣-treated endothelial cells were incubated with 1 l of antisera or non-immune rabbit serum for 15 min at 22°C prior to addition of binding buffer containing labeled oligonucleotide. Samples were subjected to electrophoresis as described above. Rabbit polyclonal antisera to Sp1, Egr-1, and p50 were obtained from Santa Cruz Biotechnology (San Diego, CA). Oligonucleotides utilized are as follows: VCAM-NFϩSp1, CTGGGTTTCCCCTTGAAGGGATTTCCCTCC-GCCTCTGCAA; VCAM-NFϩmSp1, CTGGGTTTCCCCTTGAAGGG-ATTTCCCTCattgTCTGCAA; VCAM-Sp1, ATTTCCCTCCGCCTCTG-CAA; consensus-Sp1, ATTTCCCCCCGCCCGTGCAA; mutant-Sp1, AT-TTCCCTCattgTCTGCAA; VCAM-IRF, GGAGTGAAATAGAAAGTCT-GTG.
Mutated bases are in lower case. Reverse complement strands were designed to leave G overhangs on one end. All oligonucleotides were polyacrylamide gel purified prior to annealing and labeling.
EMSA Using Recombinant Proteins-Recombinant human p50⅐p65 heterodimers were generously provided by Dr. Dimitris Thanos (Columbia University, NY). Recombinant human p50⅐p50 was prepared using pET expression vectors containing full-length human cDNA in accordance with the manufacturer's instructions (Novagen). Recombinant Sp1, purified from HeLa cells infected with recombinant vaccinia virus containing human full-length Sp1 cDNA, was obtained from Promega. Binding reactions were carried out for 15 min at 22°C in a total volume of 20 l containing 10 mM Tris-HCl, pH 8, 50 mM NaCl, 1 mM EDTA, 5% glycerol, 2 mM DTT, 0.5% Nonidet P-40, 1 mg/ml bovine serum albumin, and 32 P-NF-Sp1 (50,000 cpm). Recombinant Sp1 was prebound 5 min prior to addition of recombinant NF-B. Bound complexes were separated from free probe by electrophoresis using non-denaturing 5% polyacrylamide gels and 1 ϫ TBE running buffer.
In Vitro DNase I Footprinting-A single-labeled DNA fragment spanning the proximal VCAM-1 promoter was prepared by digesting construct pBS-VCAM-F2-R1 with EcoRI and HindIII, dephosphorylating the ends with calf intestinal alkaline phosphatase (New England Biolabs), and isolating the 300-bp fragment by electrophoresis on a 2% agarose gel and Qiagen bead purification. The fragment was ␥-32 P labeled at both ends with T4 polynucleotide kinase and purified using a Chromaspin-10 spin column. Following digestion with PstI, the single end-labeled large fragment was purified using a Chromaspin-100 spin column. Guanine ladders were generated by methylation of the probe with dimethyl sulfate (Aldrich) and subsequent piperidine (Sigma) cleavage.
Footprint binding reactions were performed using the 32 P 5Ј-labeled probe in a total volume of 20 l containing 2 l of 10 ϫ footprint binding buffer (50% glycerol, 50 mM MgCl 2 , 500 mM NaCl, 25 mM HEPES, pH 7.9), 0.5 g poly(dI:dC), 10 g bovine serum albumin, and increasing amounts of the recombinant protein as indicated. The reaction was allowed to proceed for 1 h at 4°C. 20 l of DNase buffer (25 mM NaCl, 10 mM HEPES, pH 7.9, 5 mM MgCl 2 , 1 mM CaCl 2 ) containing 0.02 units DNase I (Promega) was added to each tube and incubated for 5 min at 4°C. The reaction was terminated by the addition of 10 l of stop solution (150 mM EDTA, 5% SDS, 250 g/ml salmon sperm DNA) and immediate phenol/chloroform extraction prior to DNA precipitation with absolute ethanol. The pellets were washed with 70% ethanol, evaporated to dryness, and resuspended in sequencing sample buffer. The samples were boiled for 2 min and applied to a 6% polyacrylamide gel and run at 1500 V. The gels were subsequently dried and autoradiographed overnight at Ϫ80°C.

Identification of a Cytokine-responsive Element within the VCAM1
Promoter-To explore whether the DNA region adjacent to the transcriptional start site of the VCAM-1 promoter is capable of conferring TNF-␣-induced expression to a heterologous gene, a 2190-bp fragment of the human VCAM1 promoter (Ϫ2150 to ϩ 40) was fused to the coding sequences of the firefly luciferase gene. This plasmid was transfected into HUVE cells, and luciferase activity was measured in untreated cells, as well as in cells that had been exposed to TNF-␣ for 6 h. This 2190-bp fragment was sufficient to confer TNF-␣ responsiveness to the luciferase gene (Fig. 1A, full-length). To more precisely delineate the region required for TNF-␣ inducibility, deletion constructs were generated and tested for cytokine inducibility in HUVE cells. These studies revealed that a plasmid containing the proximal 100 bp of the VCAM1 promoter could be induced by TNF-␣ as effectively as the Ϫ2150 construct (Fig. 1A, com-pare full-length to LS-100). In contrast, a DNA fragment retaining only 50 bp of 5Ј-flanking DNA (Ϫ50 to ϩ40) was unresponsive to TNF-␣ (Fig. 1A, deletion). An internal deletion of the promoter, which lacks DNA sequences between Ϫ100 and Ϫ50, was also not induced by TNF-␣ (data not shown). Together, these results indicate that a region essential for TNF-␣ inducibility resides between residues Ϫ100 and Ϫ50. These observations are consistent with previous reports using constructs bearing a different reporter (CAT) in BAEC (Neish et al., 1992 and epithelial cells (Shu et al., 1993). This suggests that the B dependence of the VCAM1 cytokine-induced enhancer is neither species nor cell type specific. However, the specific nature of the Rel family members associating with these elements may vary in different contexts.
To identify cis-regulatory sequences required for cytokineinduced expression, we performed a detailed mutational analysis of the region between Ϫ100 and Ϫ50 (Fig. 1B). In vitro mutagenesis was used to prepare a systematic series of promoter mutants, each differing by only 10 bp from the wild-type construct. The plasmids were transfected into HUVE cells, and luciferase activity was determined in the presence and absence of TNF-␣. As shown in Fig. 1A, most of these promoter mutants responded to TNF-␣ treatment in a manner similar to the wild-type construct. The integrity of one large region of the VCAM1 promoter, Ϫ80 to Ϫ40 (LS-80, LS-70, LS-60, LS-50), was essential for TNF-␣-induced expression in HUVE cells. The Ϫ80 to Ϫ60 region corresponds to two adjacent consensus NF-B elements that have previously been determined to be important for cytokine induction (Neish et al., 1992). Interestingly, the Ϫ50 to Ϫ40 region (LS-50) represents a previously unidentified regulatory element in the VCAM1 promoter. In subsequent studies, we focused on this newly defined regulatory region in the VCAM1 promoter that appeared to be important in TNF-␣-induced expression.
Region Ϫ50 to Ϫ40 in the VCAM1 Promoter Contains an Sp1 Binding Site-To identify nuclear proteins that specifically interact with the novel VCAM1 promoter element, EMSA was performed using extracts prepared from TNF-␣-treated or untreated HUVE cells and a double-stranded end-labeled oligonucleotide probe extending from Ϫ39 to Ϫ59. This probe spans the LS-50 region (Fig. 1B) but does not include the adjacent NF-B elements. A single nucleoprotein complex was formed ( Fig. 2A, lane 2, complex A). The complex was competed with a 100-fold molar excess of unlabeled probe (lane 3), whereas an unrelated oligonucleotide did not compete (lane 6). Therefore, the protein component of complex A corresponds to a constitutively active nuclear factor, which specifically binds to the novel element in the VCAM1 promoter.
The linker scan analysis defined a novel element with a GC-rich core suggestive of an Sp1 site. Competitive EMSA was used to provide suggestive evidence for the involvement of Sp1 in complex A. Complex A was competed by a 100-fold molar excess of an oligonucleotide bearing a consensus Sp1 site ( Fig.  2A, lane 4). When the Sp1-like core sequence in the oligonucleotide was mutated (CCGCCT to CattgT), it could no longer compete for binding ( Fig. 2A, lane 5). To further implicate the involvement of Sp1 in complex A, supershift analysis was used. The probe containing the putative VCAM1 Sp1 site was incubated with HUVE nuclear extract, from control and TNF-␣induced cells, in the presence of polyclonal antisera to DNA binding proteins. Complex A is identical in control and TNF-␣-induced conditions (Fig. 2B, lanes 2 and 6) and largely supershifted by anti-Sp1 antisera (designated "B," lanes 3 and 7). The inability of Sp1 antibody to completely supershift the complex may be due to comigration of a nucleoprotein complex unrelated to Sp1 or incomplete recognition due to the phosphorylation state of Sp1. Partial supershifts involving the commercially available antibody used in this study have been reported elsewhere (Kramer et al., 1994;Minowa et al., 1994;Jensen et al., 1995). Complex A is not affected by the presence of antibodies to Egr-1, a related zinc finger transcription factor that binds GC-rich sequences (Cao et al., 1993) (lanes 4 and 8), or anti-p50, a NF-B subunit abundant in endothelial cells (lanes 5 and 9) (Read et al., 1994).
In vitro DNase I footprint analysis was used to determine whether recombinant Sp1 could interact with the human VCAM1 promoter. Sp1 specifically protected a distinct region in the promoter from partial digestion by DNase I in a dosedependent manner (Fig. 3). Comparison of the electrophoretic mobility of the protected region with the G ladder (Fig. 3) indicates that the site bound by Sp1 corresponds to the GC-rich element. The Sp1 binding site was just 3Ј to the two previously described B-binding elements  protected by recombinant p50 homodimers (Fig. 3). These data indicate that the element bound by Sp1 is in close proximity to the two B sites in the VCAM1 promoter.
FIG. 2. A novel element in the VCAM1 promoter interacts with the transcriptional activator Sp1. EMSA using an oligonucleotide probe spanning the novel element in the VCAM1 promoter (VCAM-Sp1) is shown. A, EMSA with non-induced endothelial extracts. The shifted complex is indicated by A. Unlabeled competitor oligonucleotides with sequences listed under "Experimental Procedures" were used at 100ϫ molar excess. B, supershift analysis with nuclear extracts from noninduced (lanes 2-5) and 1-h TNF-␣-induced endothelial cells (lanes 6 -9). Binding reactions were preincubated for 15 min with the antibodies indicated. The supershifted complex is indicated by B.

FIG. 3. Sp1 binds in close proximity to the 3-NF-B recognition element in the proximal VCAM1 promoter.
In vitro DNase I footprint analysis of the VCAM1 promoter using recombinant proteins is shown. The single end-labeled VCAM1 promoter fragment was incubated with increasing amounts of Sp1 or p50 homodimers for 60 min at 4°C prior to partial digestion with DNase I. Binding and electrophoresis conditions are described under "Experimental Procedures." Endothelial Sp1 Does Not Facilitate NF-B Binding to the VCAM1 Promoter-To investigate binding of Sp1 and NF-B in the context of the entire VCAM1 cytokine response region, EMSA was performed with an oligonucleotide probe spanning this region (VCAM-NFϩSp1), which includes both NF-B sites and the novel Sp1 site. Labeled probe was incubated with nuclear extract derived from unstimulated and TNF-␣ stimulated HUVEC. As shown in Fig. 4A (lanes 2-4), this probe forms a nucleoprotein complex consistent with Sp1 (complex A). This complex was present equally under all conditions of cytokine activation, while, consistent with previous studies , NF-B binds only under induced conditions (designated B) (lanes 3 and 4). The same experiment was performed with an identically sized probe incorporating a mutated Sp1 site (VCAM-NFϩmSp1, CCCGCTC to CCattgC). As expected, this probe was unable to bind Sp1 (lanes 6 -8), while the degree of B binding was unaffected (compare lanes 3 and 4 to  lanes 7 and 8).
Sp1 and NF-B Can Co-occupy the VCAM1 Promoter-EMSA was used to determine whether recombinant Sp1 could facilitate the interaction of recombinant p50⅐p65 heterodimer with its recognition element in the VCAM1 promoter. In the absence of Sp1, p50⅐p65 bound to the probe in a dose-dependent manner (designated B) (Fig. 4B, lanes 2-4). At high concentrations of NF-B (lane 4) a complex of slower mobility was observed representing occupancy of both NF-B sites. Sp1 alone resulted in a complex (designated Sp1, lane 5) with a mobility slower than bound NF-B. When the probe was prebound with Sp1 then challenged with increasing concentrations of p50⅐p65, the apparent affinity of the heterodimer for the probe was unaffected (Fig. 4B, compare B complexes in lanes 2-4 to 6 -8). However, at high concentrations of Sp1 and NF-B an additional band of even slower mobility was detected (designated Sp1ϩB, lane 8), representing occupancy of both NF-B sites and the novel Sp1 site. Thus, while recombinant Sp1 does not affect the ability of recombinant p50⅐p65 to interact with the VCAM1 promoter, both Sp1 and heterodimeric p50⅐p65 are able to co-occupy the VCAM1 cytokine response region. Taken together, these experiments demonstrate that endothelial Sp1 binds to a novel site in the VCAM1 promoter immediately adjacent to the 3Ј-NF-B site. Sp1 binds equally well to its recognition element under quiescent and TNF-␣-stimulated conditions and is able to occupy the promoter simultaneously with NF-B. The findings have not, however, established the functional significance of the Sp1 element in the transcriptional regulation of the VCAM1 gene.
Sp1 Binding Is Necessary for Optimal Cytokine-induced Gene Expression-To implicate a role for Sp1 in the transcriptional activation of the VCAM1 promoter, mutations that abolished the binding of Sp1 (CCCGCTC to CCattgC,demonstrated in Figs. 2A and 4A) were introduced into a VCAM1 promoterreporter construct extending Ϫ258 bp upstream of the transcriptional start site. Wild-type and mutant constructs were transiently transfected into low passage BAEC and incubated with or without TNF-␣. Characteristically, CAT activity driven by the wild-type VCAM1 promoter was induced 5-fold by TNF-␣ (Fig. 5, bar 1 and 2). In contrast, both basal and induced expression driven by a promoter fragment harboring the mutated Sp1 site was attenuated by approximately 3-fold (Fig. 5,  bar 3 and 4). Notably, the mutant construct retains cytokine inducibility. From these data, we conclude that an intact Sp1 binding motif is necessary for optimal levels of cytokine-induced transcription of the VCAM1 gene.

DISCUSSION
Transcriptional enhancer elements are comprised of multiple discrete DNA sequence motifs, each representing a cognate site for a DNA binding protein. The precise number, orientation, and relative positions of these protein binding modules may be critical for proper regulatory function (reviewed by Tjian and Maniatis (1994)). To characterize the cytokine-inducible transcriptional enhancer in the VCAM1 promoter, we have utilized saturation mutagenesis to define the boundaries of relevant positive regulatory regions prior to the characterization of interacting nuclear protein species. Here, we report that both Sp1 and NF-B are required for maximal activity of the cytokine response region of the VCAM1 promoter. Deletion or mutation in the Sp1 binding motif results in significant attenuation in the ability of the promoter to drive induced transcription in endothelial cells.
A key component of cytokine-inducible enhancers is the ubiquitous transcription factor NF-B (reviewed in Collins et al., 1995). The heterodimeric form (p65⅐p50) is critical for the transcriptional activation of the human VCAM1 promoter . It is likely that the specific patterns of transcrip- FIG. 4. EMSA with composite probes spanning the cytokine response region. A, mutation of the Sp1 binding site does not compromise the ability of NF-B to interact with the VCAM1 promoter. EMSA is shown with nuclear extract derived from endothelial cells, non-induced (lanes 2 and 6), 1-h TNF-␣ induction (lanes 3 and 7), and 4-h TNF-␣ induction (lanes 4 and 8). Probes are wild-type (VCAM-NFϩSp1) (lanes 1-4) or mutant Sp1 (VCAM-NFϩmSp1) (lanes 5-8). Shifted complexes are designated with arrows. B, Sp1 and NF-B can co-occupy the VCAM1 promoter. EMSA is shown with recombinant Sp1 and NF-B. Increasing concentrations of full-length p50⅐p65 were incubated alone (lanes 2-4) or combined with a fixed amount of prebound recombinant Sp1 (lanes 5-8). The probe used was wild type (VCAM-NFϩSp1). Shifted complexes are indicated with arrows. tional regulation seen among NF-B-dependent genes is mediated by the combinatorial interactions of NF-B with other transcriptional activators and chromatin components. Physical and functional interactions have been shown between NF-B and ATF-2 (Du et al., 1993), C/EBP␤ (Stein et al., 1993a), c-Jun (Stein et al., 1993b), and HMGI(Y) (Thanos and Maniatis, 1992). Previous work with the VCAM1 promoter in our laboratory demonstrated similar interactions between NF-B and IRF-1 . Interactions between NF-B and Sp1 have also been reported. Cooperative functional and physical interactions between NF-B and Sp1 are required for efficient transcription of the HIV-LTR (Perkins et al., 1993(Perkins et al., , 1994. This effect is mediated by specific physical interactions between the amino terminus of the p65 Rel domain (RelA) and the zinc finger region of Sp1. Interestingly, the HIV-LTR shares structural similarities with the VCAM1 cytokine response region (Fig. 6); both genes have tandem NF-B sites located immediately 5Ј of an Sp1 motif. Our findings are consistent with the proposal that the juxtaposition of Sp1 and NF-B elements promotes an interaction between these transcription factors (Perkins et al., 1994). The exposed surfaces of the DNA binding domains of p65 and Sp1 may contact each other in an interaction facilitated by the close approximation of the B and Sp1 elements. This close physical association in both the HIV-LTR and the VCAM1 enhancer would leave the transactivation domains of both p65 and Sp1 to interact with TATA box binding protein-associated factors, or TAFs. These proteins are coactivators postulated to interact with different types of activation domains and thus integrate signals from distinct bound transcription factors and assemble the basal transcriptional complex. The Sp1 activation domain has been demonstrated to directly and specifically interact with dTAF II 110 (Chen et al., 1994). Similarly, p65 has been reported to interact with the TATA box binding protein and enhance transcription (Kerr et al., 1993). These results add to our understanding of the transcription factors necessary for the cytokine responsiveness of this pathophysiologically important gene and provide insights into a general mechanism for the activation of some B-dependent genes. FIG. 5. Sp1 binding site is necessary for full cytokine-induced activity of the VCAM1 promoter. BAEC were transfected with 10 g of VCAM-CAT reporter plasmids containing either wild-type sequences or bearing a mutant Sp1 site. Accumulated CAT was assayed 24 h post-transfection. Transfected cells were incubated with either media (lanes 1 and 3, open bars)