INTRODUCTION

Blood monocytes infiltrate into the sites of inflammation and play major roles in host defense through their ability to present antigens and to produce various mediators. Although the mechanisms of monocyte infiltration have not been fully understood, locally produced monocyte chemoattractants seem to be responsible for the recruitment of blood monocytes into the sites of inflammatory reactions.

Monocyte chemoattractant protein-1 (MCP-1)¹ is a member of the CC subfamily of the chemokine family and attracts blood monocytes both in vitro and in vivo (1-3). MCP-1 mRNA or protein was detected at high levels in the lesions of several diseases such as atherosclerosis (4, 5), arthritis (6), idiopathic pulmonary fibrosis (7, 8), and various tumors (9-11), strongly suggesting that MCP-1 plays a critical role in the recruitment of monocytes in these diseases. A wide variety of cells, including monocytes, fibroblasts, vascular endothelial cells, and smooth muscle cells, produces MCP-1 in vitro in response to various stimuli such as lipopolysaccharide (LPS), interleukin-1 (IL-1), tumor necrosis factor- (TNF), platelet-derived growth factor (PDGF), IFN-, or 12-o-tetradecanoylphorbol 13-acetate (TPA) (1-3). However, the mechanisms of MCP-1 production remain unknown.

To understand the mechanisms involved in the expression of human MCP-1 mRNA in different types of cells at the molecular level, we previously investigated the transcription of human MCP-1 gene in TPA-, IL-1-, and TNF-stimulated human malignant glioma cell line A172 cells (12). The basal level of human MCP-1 mRNA was detected in A172 cells without any stimulus, and the expression of human MCP-1 mRNA was strongly enhanced after stimulation with these stimuli. The binding of Sp1 to the proximal GC box located between bp 64 and 59 was critical for the maintenance of the basal transcription of this gene. Two NF-B sites (A1 and A2 sites) were located 2.6 kilobases upstream of the transcription initiation site. Mutations in the A2 sequence resulted in a loss of enhancer activity, indicating that the A2 site was critical for the enhancer activity after stimulation. The role of the A1 site was not investigated in the previous study.

The mouse homologue of human MCP-1, termed JE, was cloned from PDGF-stimulated BALB/c 3T3 cells as one of early response genes (13, 14). The transcriptional mechanisms of the mouse MCP-1/JE gene was recently investigated. Ping et al. (15) investigated the regulatory site occupancy during the activation of mouse MCP-1 gene in TNF-stimulated BALB/3T3 cells by using in vivo genomic footprinting. In response to TNF, both distal promoter and proximal regulatory regions became occupied in vivo. The binding of (p65)₂, p50/p65, and (p50)₂ to the distal regulatory region of the mouse MCP-1 gene containing both NF-B sites was detected by electrophoretic mobility shift assay (EMSA), but the trans-activity of each NF-B/rel dimer remains unclear. A remote NF-B site on the mouse MCP-1 gene was also reported to be important for PDGF-induced mouse MCP-1 mRNA expression in BALB/c 3T3 cells (16). These reports indicate the importance of the NF-B sites located in the distal region of both the human and mouse MCP-1 genes.

In our preliminary study, NF-B (p50/p65) was activated in human B-lymphocytic IM9P3 cells and Raji cells without any stimulus. The binding of p50/p65 to the A2 site of the human MCP-1 gene was detected by EMSA. However, the expression of human MCP-1 mRNA was not detected in these cells.² It has recently become evident that certain NF-B/Rel dimers such as (p65)₂ or c-Rel/p65 specifically regulate the transcription of certain genes (17, 18). Thus, more detailed study on the interaction of the NF-B sites and different NF-B/Rel dimers is necessary to understand the mechanisms of the human MCP-1 gene expression. In the present study, we investigated the roles of two NF-B sites, A1 and A2 sites, and different NF-B/Rel dimers in the transcription of the human MCP-1 gene and have found that the binding of (p65)₂ and c-Rel/p65 to both A1 and A2 sites elevates the transcription of this gene.

EXPERIMENTAL PROCEDURES

LPS (Escherichia coli 055:B5) was from Difco Laboratories (Detroit, MI). TPA and cycloheximide were from Sigma. The plasmid pGL3-basic and luciferase assay system were from Promega. The full-length murine p65 expression vector pCMV-LDp65, human p50 expression vector pCMV-399 (this vector coded residues 1-399 of p50), human c-Rel expression vector pCMV-NA, and rabbit polyclonal antibodies against p50, p52, p65, RelB, and c-Rel were kind gifts from Dr. Nancy Rice (NCI-FCRDC, Frederick, MD). These antibodies were raised against the peptide of the N terminus of each corresponding protein to avoid cross-reactivity between members of the Rel family (19, 20). Anti-CEBP/ was from Santa Cruz Biotechnology, Inc. p(Ig)₄LUC was a gift from Dr. Nigel Mackman (The Scripps Research Institute, La Jolla, CA). In the plasmid, four tandem copies of the double-stranded oligonucleotide Ig were cloned into the upstream of minimal simian virus 40 promotor expressing the luciferase (pGL2-promoter vector) (18). Dulbecco's modified Eagle's medium, RPMI 1640, -minimal essential medium, FCS, and TRIzol Reagent were from Life Technologies, Inc. -[³²P]dCTP (3000 Ci/mmol) was from ICN (Costa Mesa, CA). -[³²P]ATP was from DuPont NEN. A random primed DNA labeling kit was from Boehringer Mannheim. Hybond-N⁺ nylon membranes were from Amersham Corp. Immobilon P membranes were from Millipore. -Actin cDNA was from CLONTECH. Oligonucleotides were from Operon (Alameda, CA).

Human glioblastoma A172 cells, osteosarcoma HOS cells, and HeLa cells were cultured in Dulbecco's modified Eagle's medium containing 10% FCS. Human monocytic cell line, THP-1 cells, and lymphoblastic cell line, Raji cells, were cultured in RPMI 1640 containing 10% FCS. P19 mouse embryonic carcinoma cells were cultured in -minimal essential medium containing 10% FCS. All cell lines were maintained at 37 °C in 5% CO₂.

After treatment of the cells with the indicated stimuli, total RNA was extracted with TRIzol Reagent according to the protocol from the manufacturer. Northern blot analysis was performed as described (21) in 1.2% agarose gel containing formaldehyde. Filters were hybridized at 42 °C overnight in 50% formamide, 5 × SSC, 5 × Denhardt's solution, 50 µg/ml sheared denatured salmon sperm DNA, 1% SDS, and 1 × 10⁶ dpm/ml of ³²P-labeled cDNA probe. Filters were washed twice with 2 × SSC, 0.5% SDS at room temperature for 15 min and 0.1 × SSC, 0.5% SDS at 60 °C for 30 min before autoradiographic exposure. The probes used were human MCP-1 cDNA (22), an EcoRI fragment (approximately 300 bp that covered most of the coding sequence) of human IL-8 cDNA (23), and -actin cDNA.

Nuclear extracts were prepared by a method previously reported (12), and aliquots were frozen at 80 °C. EMSA was carried out by using 5% polyacrylamide gels in 0.5 × TBE buffer (90 mM Tris borate, 2 mM EDTA) (24). The A2 probe was prepared by annealing an oligonucleotide 5-AGAGTGGGAATTTCCACTCA-3 with its antisense oligonucleotide 5-TGAGTGGAAATTCCCACTCT-3. The A2 mutant (MA2) was prepared by annealing an oligonucleotide 5-GAGTGGGAATTCGGACTCACTTCTCT-3 with its antisense oligonucleotide 5-AGAGAAGTGAGTCCGAATTCCCACTC-3. The IL-8 NF-B probe (20) was prepared by annealing an oligonucleotide 5-AAATCGTGGAATTTCCTCG-3 with its antisense oligonucleotide 5-CGAGGAAATTCCACGATTT-3. The Ig -chain NF-B probe (25) was prepared by annealing an oligonucleotide 5-AGTTGAGGGGACTTTCCCAGGC-3 with its antisense oligonucleotide 5-GCCTGGGAAAGTCCCCTCAACT-3. Each binding reaction was performed in 10 µl of 1 × binding buffer (12 mM HEPES, pH 8.0, 60 mM KCl, 4 mM MgCl₂, 1 mM EDTA, 12% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) containing a 10-fmol probe end-labeled with -[³²P]ATP and T4 polynucleotide kinase, 1 µg of salmon sperm DNA, and 5 mg of each nuclear extract at 20 °C for 30 min. For competition assay, 50-fold excess amounts of appropriate unlabeled oligonucleotide were added to the binding reaction mixture. For supershift assay, appropriate antibody was incubated with nuclear extract in 1 × binding buffer at 20 °C for 30 min before the addition of ³²P-labeled probe. The reaction was continued for 30 min at 20 °C. The gels were prerun at 200 V for 30 min and then run for another 3 h after samples were loaded.

The pGLM-ENH plasmid was digested with XhoI, and the overhangs were filled in with -[³²P]dCTP, followed by a digestion with NotI. The resulting 400-bp fragment was isolated by gel electrophoresis on an agarose gel and purified. DNase I digestion was performed with 20 µg of nuclear extracts and 1 ng of the 400-bp fragment by the method described by Lefevre et al. (26), and the digests were analyzed on 8% polyacrylamide gels.

To obtain the pGLM-PRM, the 167-bp human MCP-1 promoter region between 107 and +60 was PCR amplified with a sense primer PS 5-GGCTGAGCCCACTTATCACTCATGG-3 and an antisense primer PAS 5-GGAAGCTTGCTGGAGGCGAGAGTGCGAG-3. This PCR product was gel purified, digested with XhoI and HindIII, and then ligated into the XhoI-HindIII site of the pGL3-basic plasmid. To obtain the pGLM-ENH, the 230-bp human MCP-1 enhancer region between 2742 and 2513 was PCR amplified with a sense primer ES 5-GGGGTACCGAGATGTTCCCAGCACAG-3 and an antisense primer EAS 5-GGCTCGAGTGACAGCTGTCTGCCTCCC-3. This PCR product was gel purified, digested with KpnI and XhoI, and then ligated into the KpnI-XhoI site of the pGL3-PRM. To obtain the mutated constructs, pGLM-MA1, pGLM-MA2, and pGLM-MA1A2, mutated antisense probes MA1 5-ACGGGATCTaGaAACTTCCA-3 and MA2 5-GAGTGGGAATTcggACTCACTTCTCT-3 were chemically synthesized, and two-step PCR mutagenesis method was performed. A172 cells and P19 cells were transiently transfected with 10 µg of luciferase plasmid DNA per 100-mm tissue culture plate by a calcium phosphate precipitation method (12). After incubation with the DNA-calcium phosphate precipitate for 24 h, the cells were washed with PBS and incubated in fresh medium for 48 h in the presence or absence of 100 ng/ml TPA. The luciferase assay was performed according to the protocol provided by the manufacturer.

RESULTS

First, we investigated LPS- and TPA-induced MCP-1 mRNA expression in THP-1 cells by Northern analysis (Fig. 1). The expression of MCP-1 mRNA was first detected 2 h after LPS stimulation (lane 3). The expression level increased at 8 h and reached the peak at 16 h (lanes 5 and 6). The same level of MCP-1 mRNA expression was maintained until 24 h (lane 7). In contrast, MCP-1 mRNA was not detected until 8 h after TPA stimulation (lanes 11-14). Addition of cycloheximide did not affect the expression of MCP-1 mRNA by LPS- or TPA-stimulated THP-1 cells (lanes 8, 9, 17, and 18). The expression of IL-8 mRNA was also investigated (Fig. 1). A rapid induction of IL-8 mRNA was detected with both TPA and LPS stimulation (lanes 2 and 11). Addition of cycloheximide superinduced IL-8 mRNA expression in LPS-stimulated THP-1 cells (lanes 8 and 9).

[View Larger Version of this Image (38K GIF file)]

We previously identified two NF-B binding sites, A1 and A2, in the enhancer region of the human MCP-1 gene (12). The enhancer region of the mouse MCP-1 gene was recently characterized by two independent groups (Table I). Freter et al. (16) reported four PDGF-responsive elements, elements I-IV, with PDGF-stimulated NIH 3T3 cells. In contrast, Ping et al. (15) reported that three regions (site A, B-1, and B-2) were occupied by proteins during the activation of the mouse MCP-1 gene in TNF-activated NIH 3T3 cells. As shown in Fig. 2A, element I, element III/site A, B-1, and element IV/B-2 of the mouse MCP-1 gene are highly conserved in the human MCP-1 gene. However, element II of the mouse MCP-1 gene is not conserved in the human MCP-1 gene.

Table I. Reported cis-elements and trans-activators involved in MCP-1 gene expression Factors Species Binding sites Regions References NF-B Human A1 and A2 NF-B sites Distal enhancer region 12 Sp1 Human GC box Proximal promoter region 12 Unknown Human Region 1 Distal enhancer region Present study NF-B Mouse  B-1 and B-2 (element IV) Distal enhancer region 15, 16, 50 Sp1 Mouse GC box Proximal promoter region 15 90-kDa coactivator Mouse Element IV Distal enhancer region 50 AP-1 Mouse Not determined Not determined 52 NF-IL6 Mouse Not determined Not determined 53 Unknown Mouse 7-mer sequence 3-Untranslated region 51 Unknown Mouse Element II Distal enhancer region 50 Unknown Mouse Site A (element III) Distal enhancer region 50

[View Larger Version of this Image (45K GIF file)]

The binding of nuclear proteins to the enhancer region of the human MCP-1 gene was characterized by DNase I footprinting with the nuclear extracts of stimulated or non-stimulated THP-1 cells (Fig. 2B). Three regions were protected from DNase I digestion after LPS or TPA stimulation. Regions 2 and 3 corresponded to the previously reported NF-B binding sites, A1 and A2 (12). Region 1 was occupied before LPS or TPA stimulation (lanes 3 and 4) and seemed to be identical to the element III/site A of the mouse MCP-1 gene (15, 16). Constitutive protein binding to the element III/site A of the mouse MCP-1 gene was previously reported (15). These results suggest that three cis-elements (regions 1, 2, and 3) are likely to be involved in the enhancer activity of the human MCP-1 gene. However, only LPS stimulation induces protein binding to A1 and A2 sites of the human MCP-1 gene.

We performed EMSA to identify which NF-B/Rel dimer bound to the A2 probe using the nuclear extracts of LPS- or TPA-stimulated THP-1 cells. As shown in Fig. 3, four different DNA-nuclear protein complexes (C1, C2, C3, and C4) were found 2 h after LPS stimulation (lanes 1 and 2). The formation of the complexes was inhibited with an excess amount of the unlabeled A2 probe (lane 3) but not with the MA2 probe carrying three nucleotide substitutions (lane 4), indicating that these complexes were specific. We also performed EMSA with the A2 probe and nuclear extracts of TPA-stimulated THP-1 cells. Although a significant amount of MCP-1 mRNA was not detected 2-3 h after TPA stimulation by Northern analysis as shown in Fig. 1 (lane 13), the formation of C3 was detected 2 h after stimulation (lanes 5-7). The formation of C4 was also detected 16 h after stimulation when high level MCP-1 mRNA was detected (lanes 8-10 and Fig. 1, lane 15). Neither C1 nor C2 was detected after TPA stimulation.

[View Larger Version of this Image (48K GIF file)]

To identify the NF-B/Rel subunit in the each complex, supershift assay was performed with a specific antibody against each NF-B/Rel protein (19, 20). As shown in Fig. 4, normal rabbit IgG or antibodies against p52, RelB, or CEBP/ had no effect on the complex formation with the A2 probe and nuclear extracts from LPS-stimulated THP-1 cells (lanes 3, 6-8). However, addition of anti-p50 supershifted both C3 and C4 (lane 2). C1, C2, and C3 were supershifted with anti-p65 (lane 4), and C2 and C4 were supershifted with anti-c-Rel (lane 5). These results indicated that C1, C2, C3, and C4 were (p65)₂, c-Rel/p65, p50/p65, and p50/c-Rel, respectively. The complex formation after TPA stimulation was also characterized. C3, which was detected after a 2-h TPA stimulation (lane 9 and Fig. 3, lane 5), was supershifted by the addition of anti-p50 or anti-p65 (lanes 10 and 12). Normal rabbit IgG and antibodies against p52, c-Rel, RelB, and CEBP/ had no effect on the complex formation (lanes 11, 13-17). These results indicated that a 2-h TPA stimulation could induce the binding of p65/p50 to the A2 site of the human MCP-1 gene. C3 and C4 detected after a 16-h TPA stimulation (Fig. 3, lane 8) contained p65/p50 and p50/c-Rel, respectively (data not shown).

Characterization of NF-B/Rel proteins that bound to the A2 probe.

[View Larger Version of this Image (43K GIF file)]

The Binding of Different NF-B/Rel Proteins to the Human MCP-1 A1 Site, the hIL-8 NF-B Site, and the Ig -Chain NF-B Site

The binding affinity of each NF-B/Rel dimer to a NF-B site depends on the sequence of each NF-B site (25, 27, 28), and NF-B binding sequences of different genes have been classified as the binding site for the heterodimer or homodimer of NF-B/Rel proteins. As shown in Fig. 4, a 2-h LPS-stimulation induced the binding of (p65)₂, c-Rel/p65, p50/p65, and p50/c-Rel to the A2 probe containing 5-GGGAATTTCC-3. Four different NF-B/Rel dimers are capable of binding to the A2 site of the human MCP-1 gene. Since the sequences of the human MCP-1 A1 site (5-GGGAACTTCC-3), hIL-8 NF-B site (5-TGGAATTTCC-3), and Ig -chain NF-B site (5-GGGACTTTCC-3) are different from that of the human MCP-1 A2 site, the binding affinity of NF-B/Rel dimer to each NF-B sequence could be different. As shown in Fig. 5, the A1 probe as well as the A2 probe showed affinity to (p65)₂, c-Rel/p65, p50/p65, and p50/c-Rel (lanes 1-5). In contrast, the IL-8 NF-B probe showed affinity to (p65)₂ and c-Rel/p65 but not to p50/p65 and p50/c-Rel (lanes 6-9). The Ig -chain NF-B probe showed affinity to p50/p65 and p50/c-Rel but not to (p65)₂ or c-Rel/p65 (lanes 10-13). Thus, human MCP-1 NF-B sites bind to a broader range of NF-B/Rel family proteins compared with IL-8 or the Ig -chain NF-B site.

Characterization of NF-B/Rel proteins that bound to the NF-B sites of the MCP-1, IL-8, or Ig -chain genes.

[View Larger Version of this Image (47K GIF file)]

To investigate the trans-activity of each NF-B/Rel dimer for the human MCP-1 gene transcription, P19 mouse embryonic carcinoma cells were cotransfected with pGLM-ENH containing the enhancer region of the human MCP-1 gene and different combinations of pCMV-LD (p65 expression vector), pCMV-339 (p50 expression vector), and pCMV-NA (c-Rel expression vector). In P19 cells, endogenous NF-B/Rel proteins were undetectable (29), but Sp1, essential for the basal transcription of the human MCP-1 gene, was detectable (data not shown).

A basal level of luciferase activity was detected after transfection with 15 µg of pGLM-ENH (Fig. 6, A-C). Cotransfection with p65 expression vector dose-dependently increased the luciferase activity up to 12-fold (Fig. 6A), but cotransfection with p50 or c-Rel expression vector did not increase the basal activity (Fig. 6, B and C). These results indicated that (p65)₂ had a potent trans-activity for the human MCP-1 gene transcription, but (p50)₂ and (c-Rel)₂ did not activate the human MCP-1 gene transcription. To investigate the trans-activity of p50/p65 or c-Rel/p65 heterodimer, p50 or c-Rel expression vector was cotransfected with 600 ng of p65 expression vector into P19 cells along with the pGLM-ENH. As shown in Fig. 6D, the luciferase activity, which was elevated by (p65)₂, was dose-dependently decreased by the cotransfection with the p50 expression vector, suggesting that p50/p65 is not a potent trans-activator for the human MCP-1 gene transcription. In contrast, cotransfection with the c-Rel and p65 expression vectors resulted in high luciferase activities (Fig. 6E), suggesting that c-Rel/p65 also has a potent trans-activity. trans-Activity of the p50/c-Rel heterodimer was also examined in P19 cells by cotransfecting with both p50 and c-Rel expression vector. As shown in Fig. 6F, p50/c-Rel heterodimer did not increase luciferase activity. It is reported that p50/p65 is a potent trans-activator for the transcription of the Ig -chain gene (25). As shown in Fig. 6G, cotransfection of P19 cells with p65 expression vector and p(Ig)₄LUC increased luciferase activity, which was probably due to the binding of (p65)₂ to the Ig -chain (17). Cotransfection of P19 cells with p50 and p65 expression vectors and p(Ig)₄LUC dramatically increased luciferase activity, and the luciferase activity was decreased with high amounts of p50, indicating the formation of the p50/p65 heterodimer and subsequently the formation of (p50)₂ at high concentrations of p50 in P19 cells (Fig. 6H). Recent studies have shown that the p50/p65 heterodimer can be efficiently formed when p50 and p65 expression vectors are cotransfected in different cell lines. This has been shown by EMSA and by increases in the transcriptional activation of p(HIVB)₄CAT reporter genes (30-33). A study in vitro has also suggested that p50/p65 is preferentially formed when purified p50 and p65 are incubated together (25).

Characterization of the trans-activity of each NF-B/Rel subunit.

[View Larger Version of this Image (41K GIF file)]

Correlation between MCP-1 mRNA Expression and the Binding Pattern of Different NF-B Subunits to the A2 Probe in Different Cell Lines

As described above, the binding of (p65)₂ or c-Rel/p65 to the A2 site of the human MCP-1 gene was critical for the activation of pGLM-ENH reporter gene in P19 cells. Different human MCP-1 mRNA expression patterns observed in LPS- or TPA-stimulated THP-1 cells might be due to the availability of (p65)₂ or c-Rel/p65 in these cells. Therefore, we investigated whether there was a correlation between human MCP-1 mRNA expression and the binding of NF-B/Rel dimer to the A2 probe by using Raji, HOS, HeLa, and A172 cells. In Raji cells, human MCP-1 mRNA was not detected with or without TPA stimulation (Fig. 7A, lanes 4 and 5), but the binding of p50/p65 and p50/c-Rel was detected (Fig. 7B). Human MCP-1 mRNA expression was rapidly induced in HOS, HeLa, and A172 cells by TPA stimulation (Fig. 7A, lanes 6-11). The binding of (p65)₂, c-Rel/p65, p50/p65, and p50/c-Rel to the A2 probe was observed with the HeLa cell nuclear extracts (Fig. 7B). With the HOS cell nuclear extracts, the binding of (p65)₂, c-Rel/p65, and p50/p65 was observed (Fig. 7B). The binding of (p65)₂ and p50/p65 was observed with the A172 cell nuclear extracts (Fig. 7B). Thus, there was a strong correlation between human MCP-1 mRNA expression and the binding of (p65)₂ and/or c-Rel/p65 to the A2 probe.

Correlation between human MCP-1 mRNA expression (A) and the binding of NF-B/Rel subunits to the A2 probe (B) in different tumor cell lines.

[View Larger Version of this Image (39K GIF file)]

To investigate the role of the A1 site and a possible cooperation with the A2 site, the sequences of the A1 and the A2 sites were mutated (Fig. 8), and then these constructs were transfected into p65-expressing P19 cells or TPA-stimulated HeLa cells. As shown in Fig. 8, 8-fold higher luciferase activity was detected with pGL-ENH (containing both A1 and A2 sites) comparing with the basal luciferase activity detected with pGL-PRM (containing only the proximal region (GC box)) of the human MCP-1 gene. This enhancer activity was significantly reduced by the mutation in either the A1 or A2 sequence, indicating that the enhancer activity was caused by the binding of (p65)₂ to both A1 and A2 sites. The luciferase activity in TPA-stimulated HeLa cells, in which (p65)₂, c-Rel/p65, p50/p65, and p50/c-Rel were induced (Fig. 8B), was also 12-fold higher with the pGL-ENH comparing with the basal activity obtained with the pGL-PRM. The mutation in either the A1 or A2 sequence resulted in a significant loss of the enhancer activity. These results suggest that the binding of (p65)₂ and/or c-Rel/p65 to both the A1 and the A2 sites is important for the human MCP-1 gene transcription.

[View Larger Version of this Image (20K GIF file)]

DISCUSSION

We previously identified two closely located NF-B binding sites, A1 and A2 sites, in the distal 5-flanking region of the human MCP-1 gene (12). A2 site was found to be important for the transcription of the human MCP-1 gene in TPA-, IL-1-, and TNF-stimulated malignant glioma cell, A172. Although the A1 site was also included in the CAT construct we used for the assay, the role of A1 site was not investigated. Recently, two independent groups investigated the transcriptional regulation of the mouse MCP-1 gene (15, 16). Ping et al. (15) found that two NF-B binding sites in the distal regulatory region of the gene were occupied by proteins upon stimulation of 3T3 fibroblast with TNF. The sequences in the distal regulatory region including two NF-B sites are highly homologous between human and mouse MCP-1 genes, suggesting important roles of the two NF-B sites. In the present study, we investigated the roles of two NF-B sites in the transcription of the human MCP-1 gene by using human monocytic cell line, THP-1 cells. Although MCP-1 is a potent chemoattractant for monocytes, monocytes themselves produce MCP-1 in vitro, and infiltrating monocytes into the lesions of inflammatory diseases are often found positive for MCP-1 in vivo (1). Therefore, investigating the mechanisms of MCP-1 mRNA expression in monocytes is particularly interesting.

Binding of nuclear proteins to both A1 and A2 sites were detected 2 h after LPS or TPA stimulation of THP-1 cells by DNase I footprinting. 2-h LPS stimulation caused accumulation of MCP-1 mRNA and formation of four nuclear protein-A2 probe complexes, whereas a 2-h TPA stimulation did not induce accumulation of MCP-1 mRNA, and only one nuclear protein-A2 probe complex was formed. The formation of the same four complexes was detected with the A1 probe. Kunsch et al. (28) previously investigated the nucleotide sequence that was optimal for the binding of each NF-B/Rel protein. According to the study, the A2 (5-GGGAATTTCC-3) sequence has high affinity to both p65 and c-Rel. Although the A1 (5-GGGAACTTCC-3) sequence does not match any sequence shown by Kunsch et al., this sequence could also be considered to have high affinity to both p65 and c-Rel because of its high similarity. In contrast, the NF-B site of the Ig -chain gene (5-GGGACTTTCC-3) has high affinity to p50, and the hIL-8 NF-B site (5-TGGAATTTCC-3) has high affinity to only p65. In the present study, we showed the binding of (p65)₂, c-Rel/p65, p50/p65, and p50/c-Rel to the A1 and the A2 probes after LPS stimulation of THP-1 cells, thus showing their affinity to both p65 and c-Rel. In contrast, the NF-B site of the human IL-8 gene or the Ig -chain gene showed more selective affinity. These results indicate that the NF-B sites of the human MCP-1 gene have a broader binding affinity to different NF-B/Rel proteins compared with that of the human IL-8 gene or the Ig -chain gene and also suggest the role of different NF-B/Rel dimers in the transcription of each gene.

So-called NF-B, the p65/p50 heterodimer, was originally identified as a nuclear factor that bound to the B enhancer motif of the Ig-chain gene and was reported to be involved in the expression of various genes in different types of cells (34). Since then, different NF-B/Rel dimers have been identified. The binding of (p65)₂ to the NF-B site was first presented by Urban et al. (35) by an in vitro experiment in which purified p65 was used. However, Fujita et al. (25) reported that (p50)₂ could bind to the NF-B sequence with an affinity about 10-fold higher than that of (p65)₂, and an equimolar mixture of p50 and p65 formed almost exclusively p50/p65. The findings by Fujita et al. led to a suggestion that (p65)₂ might not naturally exist in vivo. Presence of (p65)₂ was later shown by detecting the binding of (p65)₂ to the NF-B sites (5-GGGAATTCCC-3) of the MHC class I and IL-2 receptor genes with the cytoplasmic extracts of deoxycholate-treated HeLa cells (36) and the nuclear extracts of TPA-stimulated Jurkat cells (37). The binding of c-Rel/p65 to the NF-B site of the urokinase gene (5-GGGAAAGTAC-3) was reported with the nuclear extracts of TPA-stimulated HeLa and HT1080 cells (38).

Recent studies on the tissue factor (TF) and vascular cell adhesion molecule-1 (VCAM-1) genes indicate the roles of different NF-B/Rel dimer in the transcription of different genes (17, 18). TF is specifically induced in the cells such as monocyte or vascular endothelial cells, whereas VCAM-1 is induced in endothelial cells. Oeth et al. (18) characterized a nuclear protein complex from LPS-stimulated THP-1 cells that bound to the NF-B-like site, 5-CGGAGTTTCC-3, in the 5-flanking region of the human TF gene and found that c-Rel/p65, but not p65/p50, bound to the NF-B-like site and mediated the expression of the TF gene. In the promoter region of the VCAM-1 gene, two NF-B sites are closely located, and both NF-B sites are necessary for the transcription of this gene. Ahmad et al. (17) investigated the role of (p65)₂, p50/p65, and (p50)₂ in the transcription of this gene by using recombinant proteins. Although all dimers were capable of binding to the NF-B sites, (p65)₂, not p65/p50 or (p50)₂, activated the transcription of the VCAM-1 gene. Since both MCP-1 and VCAM-1 are involved in the emigration of blood monocytes, induction of both proteins by (p65)₂ may be an effective way to promote the infiltration of monocytes.

Transcription of the human MCP-1 gene requires synergism between two NF-B/Rel dimers and Sp1 that is essential for the basal transcription of this gene. A similar synergism between NF-B/Rel and Sp1 is important for the activation of the human immunodeficiency virus-long terminal repeat promoter (39-41). In this promoter region, two NF-B sites and three Sp1 sites are closely located. An interaction between p65 and Sp1 that binds to the Sp1 site closest to the NF-B site mediates the inducible HIV-1 gene expression (41).

In the promoter region of the E-selectin gene, two NF-B sites are closely located, and a cooperation of two NF-B sites was necessary for the enhanced transcription of this gene (42, 43). This cooperation was mediated by high mobility group protein I (Y), HMG-I (Y). HMG-I (Y) was originally purified from CV-1 cells as a DNA-binding protein to -satellite DNA of African green monkey (44). HMG-I (Y) specifically binds to AT-rich sequences (45) and changes the chromatin conformation and hence the accessibility to other DNA-binding proteins. In the case of the E-selectin gene, HMG-I (Y) recognizes the central AT-rich region of the NF-B sequence, 5-GGGAATATCC-3, stimulates the binding of NF-B to the NF-B binding site, and enhances the assembly of NF-B and other trans-activators (42, 43). HMG-I (Y) also binds to the IFN- NF-B site, 5-GGGAATTTCC-3 (46), and the melanocyte growth-stimulating activity NF-B site, 5-GGGAATTTCC-3 (47). This sequence is completely identical to the A2 sequence of the human MCP-1 gene. Since two NF-B complexes cooperate for the human MCP-1 gene transcription, it is possible that HMG-I (Y) plays a role in enhancing the assembly of the two NF-B/Rel complexes that bind to the enhancer region of the MCP-1 gene (48). Further investigation is necessary to determine whether HMG-I (Y) is involved in the human MCP-1 gene regulation.

However, there seem to be other mechanisms involved in the human MCP-1 gene expression. An elevated level of human MCP-1 mRNA was detected in TPA-stimulated THP-1 cells 16 h after stimulation, but the binding of neither (p65)₂ nor c-Rel/p65 to the A2 probe was detected by EMSA. Since the stability of MCP-1 mRNA was prolonged in IL-1-stimulated U373 malignant glioma cells (49) and TPA- or LPS-stimulated A172 cells,³ an elevated level of MCP-1 mRNA expression in TPA-stimulated THP-1 cells might be due to the prolonged stability of MCP-1 mRNA. Freter et al. (50) recently reported that the binding of 90-kDa phosphoprotein coactivator to the mouse MCP-1 NF-B site was required for PDGF-induced mouse MCP-1 expression in NIH 3T3 cells. The 90-kDa coactivator was obviously different from HMG-I (Y) because of the much higher molecular mass. Ping et al. (15) also suggested that one or more of the factors that interacted with the mouse MCP-1 regulatory region required modification by protein kinases (15). Thus, transcription of human MCP-1 gene may also require phosphorylation of factors that bind to the regulatory region of the gene. Constitutive binding of unknown protein to the element III (also called siteA by Freter et al.) located near upstream of the mouse MCP-1 NF-B sites was also reported (15, 16). Constitutive binding of a nuclear protein to the same region of the human MCP-1 gene was shown in the present study. The 7-mer sequence 5-TTTTGTA-3 located in the 3-untranslated region of the mouse MCP-1 gene was reported to be involved in the PDGF-induced mouse MCP-1 gene transcription in NIH 3T3 cells (51). AP-1 (52) and CEBP/ (53) were also reported to be involved in the mouse MCP-1 gene regulation. Further investigation is necessary to completely understand the mechanisms of elevated levels of human MCP-1 mRNA expression in different types of cells.