Lipopolysaccharide Induction of the Tumor Necrosis Factor-α Promoter in Human Monocytic Cells

Biosynthesis of tumor necrosis factor-α (TNF-α) is predominantly by cells of the monocytic lineage. This study examined the role of various cis-acting regulatory elements in the lipopolysaccharide (LPS) induction of the human TNF-α promoter in cells of monocytic lineage. Functional analysis of monocytic THP-1 cells transfected with plasmids containing various lengths of TNF-α promoter localized enhancer elements in a region (−182 to −37 base pairs (bp)) that were required for optimal transcription of the TNF-α gene in response to LPS. Two regions were identified: region I (−182 to −162 bp) contained an overlapping Sp1/Egr-1 site, and region II (−119 to −88) contained CRE and NF-κB (designated κB3) sites. In unstimulated THP-1, CRE-binding protein and, to a lesser extent, c-Jun complexes were found to bind to the CRE site. LPS stimulation increased the binding of c-Jun-containing complexes. In addition, LPS stimulation induced the binding of cognate nuclear factors to the Egr-1 and κB3 sites, which were identified as Egr-1 and p50/p65, respectively. The CRE and κB3 sites in region II together conferred strong LPS responsiveness to a heterologous promoter, whereas individually they failed to provide transcriptional activation. Furthermore, increasing the spacing between the CRE and the κB3 sites completely abolished LPS induction, suggesting a cooperative interaction between c-Jun complexes and p50/p65. These studies indicate that maximal LPS induction of the TNF-α promoter is mediated by concerted participation of at least two separatecis-acting regulatory elements.

sion in cells of monocytic lineage is quite complex, involving controls at both transcriptional and post-transcriptional levels (2). In addition, both 5Ј and 3Ј nucleotide sequences influence LPS-induced transcription of human TNF-␣ cDNA transfected into the murine macrophage RAW264.7 cell line (3).
The promoter region of the human TNF-␣ gene contains a complex array of potential regulatory elements. In T cells, cooperation between the CRE and the adjacent B3 sites is required for calcium-mediated TNF-␣ promoter activity (4), and cooperation between the CRE and the adjacent Ets sites for PMA-induced activity (5). However, the role or these regulatory elements in transcriptional activation of the TNF-␣ gene in human monocytes remains unclear. To date, published reports indicate that activation of TNF-␣ gene transcription in human monocytes in response to various stimuli is mediated by a region within Ϫ200 bp upstream of the transcriptional start site (6 -12). Using a promonocytic leukemia cell line, U937, several nuclear factor binding elements, including AP-1 (6), Egr-1 (7), CRE (8), C/EBP␤ (9), and AP-2 (10), have been suggested to mediate TNF-␣ transcription in response to PMA or cytokines. These results appear to be incomplete and often conflicting. No consensus has been reached, and no cooperation between these regulatory elements was established. Although the reasons for these discrepancies are unclear, the U937 cell line may be too poorly differentiated to serve as a suitable model of gene expression in cells of monocytic lineage.
Transcriptional activation of the murine TNF-␣ gene in murine macrophages has been demonstrated to be predominantly dependent on a region in the murine TNF-␣ promoter upstream of Ϫ451 bp, which contains NF-B DNA binding motifs (13,14). A major histocompatibility complex class II Y box was also inferred to play a role in LPS inducibility (13,14). To date, the role of NF-B in the transcriptional activation of the human TNF-␣ gene in human monocytes in response to LPS is controversial. An early study using murine monocytic cells failed to identify a role for the B sites in transcriptional activation of human TNF-␣ in response to LPS or virus (15). However, recent reports employing pharmacological agents that block the nuclear translocation of NF-B, such as pyrrolidine dithiocarbamate (16), and sodium salicylate (17), demonstrated a suppression of TNF-␣ gene expression. Moreover, LPS induction of the human TNF-␣ promoter in human monocytic THP-1 leukemia cells was mediated by the B3 site at Ϫ97 bp (11). Taken together, these results suggest a likelihood that murine monocytic cells are not suitable for the analysis of the regulation of the human TNF-␣ promoter.
In this study, we have employed a line of THP-1 cells that exhibits high inducibility for TNF-␣ gene transcription similar to that of freshly isolated human monocytes. We have performed a comprehensive analysis of the role of various cisacting regulatory elements in the transcriptional regulation of the human TNF-␣ gene. Using LPS stimulation as a paradigm, we find that a mechanism involving several transcription factors is required for maximal TNF-␣ promoter activity in human monocytes.
Mutant Series-Plasmids containing site-specific mutations at AP-1, AP-2, or CRE sites (6) were provided by Dr. J. Economou. Additional mutant plasmids in this series were produced according to methods described in the Transformer site-directed mutagenesis kit (CLONTECH, Palo Alto, CA). The sequences of the oligonucleotides with site-specific mutations used for these constructs are listed in Table I.
Heterologous Promoter Series-Multiple copies of oligonucleotides containing sequences from the human TNF-␣ promoter were cloned upstream of the minimal SV40 promoter driving expression of the luciferase reporter gene in pGL2-promoter (Promega Corp.). Sequences of oligonucleotides used for producing these constructs are listed in Table I. All plasmids were verified by DNA sequencing.  Table I. The B3 mutant was generated using the B3 m2 oligonucleotide (Table  I)

DNA Transfection
A DEAE-dextran transfection procedure (18,20) was used. Briefly, 3 ϫ 10 7 THP-1 cells were resuspended in 1 ml of Tris-buffered saline and incubated for 10 to 20 min at 37°C with 5 g of plasmid DNA and 80 g of DEAE-dextran (Pharmacia, Uppsala, Sweden). During incubation, cells were monitored closely for permeability to trypan blue. Transfection was stopped by adding large volumes of Tris-buffered saline, usually after 10 min, when 20 -30% of cells are permeable to trypan blue. After washing with Tris-buffered saline, cells were cultivated in media for 48 h. Cells were stimulated with 5 g/ml LPS (Escherichia coli O111:B4 purchased from Calbiochem, La Jolla, CA) in a 96-well plate at 1 ϫ 10 6 cells/well. After 7 h of incubation at 37°C, cells were harvested and luciferase activity was determined using an assay kit (Promega) and the Monolight 2010 luminometer (Analytical Luminescence Laboratory, San Diego, CA).

Localization of DNA Elements Involved in the Transcriptional Activation of the TNF-␣ Gene in Response to LPS-To
define the 5Ј boundary of LPS responsive elements, THP-1 cells were transiently transfected with a series of plasmids containing progressive truncations of the 5Ј promoter sequence between Ϫ1311 bp and Ϫ36 bp. Deletion of sequences upstream of Ϫ182 had no significant effect on LPS-induction of the TNF-␣ promoter activity (Fig. 1). Removal of a region from Ϫ182 to Ϫ162, which contains an Sp1/Egr-1 overlapping site, reduced LPS inducibility by 50% (from 13.3-fold to 7.8-fold). Removal of a region between Ϫ161 and Ϫ121 had no significant effect on the LPS inducibility. In contrast, deletion to Ϫ95, which removes Ets, CRE, and B (B3) sites, further reduced LPS inducibility (Fig. 1). Finally, deletion to Ϫ36 removed a region containing a Sp1 site and abolished basal promoter activity. These results provide new and substantial evidence that LPS induction of the TNF-␣ gene in monocytes involves at least two regulatory elements; region I (Ϫ182 to Ϫ162) and region II (Ϫ120 to Ϫ96).

Determination of the Roles of Various Binding Sites in the Human TNF-␣ Promoter by Functional Analysis of Mutant
Plasmids-To determine the functional role of nuclear binding motifs in the region identified by 5Ј-truncation analysis, plasmids with specific site-directed mutations were examined. Mutation in the Egr-1 site (Ϫ169 bp), the CRE site (Ϫ106 bp), as well as the B3 site (Ϫ97 bp), markedly reduced LPS induction. In contrast, mutation of the AP-1 (Ϫ57 bp) or AP-2 (Ϫ36 bp) had no effect on LPS induction of the TNF-␣ promoter (Fig. 2). These results are consistent with our findings using the 5Јdeletion series of plasmids shown in Fig. 1.
Functional Analysis of the CRE and B3 Sites in Heterologous Promoter Plasmids-To further explore the potential of these nuclear factor binding motifs to function as enhancer elements, we examined their ability to confer inducibility to heterologous promoter. Neither four tandem copies of the B3 nor two copies of the CRE sites alone conferred LPS inducibility to a SV40 minimal promoter (Fig. 3A). However, when two or three copies of a DNA fragment spanning both the CRE and the B sites were cloned upstream of the SV40 promoter strong LPS inducibility was observed (Fig. 3, A and B). Mutation of either the CRE site (Ϫ110/Ϫ86 m1) or the B3 site (Ϫ110/Ϫ86 m2) completely abolished LPS inducibility (Fig. 3B). These results suggest that the LPS induction of human TNF-␣ transcription requires cooperative interaction between proteins bound to the CRE and B sites. There is only 1 base pair separating the CRE and B3 sites. Whether the close proximity of the CRE site and the B3 site is required for the optimal transactivation of TNF-␣ was investigated. For these experiments, 5, 10, or 15 additional base pairs were added between CRE and B3 sites and oligonucleotides were cloned into pGL2promoter. These insertions created 1/2, 1, or 1 1/2 extra turns of the DNA helix between these two sites. Fig. 3C shows that insertion of DNA between these two sites abolished LPS inducibility.
Identification of Transcription Factors That Bind to the Egr-1, CRE, and B3 Sites-EMSA were performed to determine which transcription factors bind to sites in regions I and II of the human TNF-␣ promoter.
Region I: Egr-1 and Sp1-An oligonucleotide containing only the Egr-1 site bound an LPS and phorbol ester-inducible complex (Fig. 4A). This complex was not observed when an oligonucleotide containing a mutant Egr-1 site was used as the probe (Fig. 4A, lanes 4 -6). In addition, this LPS-inducible complex was supershifted by anti-Egr-1 antibody (Fig. 4A, lane  9), demonstrating that Egr-1 bound to this site.
A prominent complex was observed when nuclear extracts from unstimulated cells were incubated with an oligonucleotide spanning the overlapping Sp1/Egr-1 sites (Ϫ182 to Ϫ157 bp) (Fig. 4B, lane 1). This complex was not induced by LPS (lane 2) but was supershifted using an anti-Sp1 antibody (lane 3). In addition, this complex was not observed using an oligonucleotide containing a mutated Sp1 site (Fig. 4B, lanes 5-8). Taken together, these results demonstrated that Sp1 bound to this site. In addition to the constitutively expressed Sp1 complex, LPS stimulation of cells resulted in the formation of an Egr-1 complex that bound to the oligonucleotide containing overlapping Sp1/Egr-1 sites (Fig. 4B, lane 2). Similarly, this LPS inducible Egr-1 complex was observed using the Sp1 mut /Egr-1 oligonucleotide as a probe (Fig. 4B, lane 6). The Egr-1 complex FIG. 4. LPS induction of Egr-1 nuclear factor. A, nuclear extract from unstimulated THP-1 or THP-1 stimulated with LPS (10 g/ml) or PMA (10 nM) for 1 h were probed with labeled oligonucleotide probes of either wild type sequence or mutant sequence of Egr-1 site (Table I). B, probes used were oligonucleotide spanning overlapping Sp1/Egr-1 site (Ϫ182 to Ϫ157) or oligonucleotide containing base substitutions in Sp1 site, Sp1 mut /Egr-1 (Ϫ182/Ϫ157 m1, Table I). For antibody supershift experiment, nuclear extracts were incubated with antibody (2 g) for 20 min before addition of probes. CAGATGAGCTCATATCGTATCGTATCGTGGGTTTCTCCAC was not affected by the addition of an anti-Sp1 antibody (Fig.  4B, lanes 3 and 7), but was supershifted with an anti-Egr-1 antibody (Fig. 4B, lanes 4 and 8). Region II: B3 and CRE-EMSA were performed with oligonucleotides spanning three putative B sites: B1 (Ϫ594 to Ϫ577), B2 (Ϫ216 to Ϫ199), and B3 (Ϫ104 to Ϫ87) ( Table I). As depicted in Fig. 5A, LPS stimulation of cells resulted in the formation of nuclear protein-DNA complexes with B1 (lane 2) and B3 (lane 6), but not with B2 (lane 4). More protein bound to B1 that B3. Similar results were found using nuclear extracts from human peripheral blood monocytes (Fig. 5B). Monospecific anti-p50 and anti-p65 antibodies were used in supershift experiments to identify the protein composition of the complexes formed with the B1 and B3 sites. As shown in Fig. 5C, the LPS induced complex binding either to the B1 or to the B3 oligonucleotides were supershifted by anti-p50 and also by anti-p65 antibodies (Fig. 5C), indicating that they were composed of p50/p65 heterodimers. Furthermore, mutation of the B3 site (B3 m2, Table I) abolished binding of p50/p65 (data not shown). Importantly, the same mutation significantly reduced LPS inducibility of the TNF-␣ promoter (Fig. 2).
Using a prototypic CRE site as a probe, we demonstrated that CREB was constitutively expressed in unstimulated THP-1 cells and that CREB binding was not induced by LPS (Fig. 6A). In contrast, LPS stimulation increased the amount of protein binding to the non-consensus CRE site from the TNF-␣ promoter (Fig. 6B, compare lanes 1 and 4). Antibody supershift experiments were performed to determine the proteins that bound to the CRE site (Fig. 6B). In unstimulated cells, the majority of the complex was supershifted using an anti-CREB antibody, whereas only a minor supershift was observed using an anti-c-Jun antibody. In LPS-stimulated cells, the anti-c-Jun antibody supershift the majority of the complex, whereas the anti-CREB antibody formed a minor supershift band. These results suggest that LPS does not increase the binding of CREB, consistent with our results using a prototype CRE site (Fig. 6A), and that LPS increases binding of c-Jun-containing complexes to the CRE site from the TNF-␣ promoter.
Using an oligonucleotide containing a mutated CRE site (CREm2 , Table I), we observed a reduction in the total amount of protein binding (Fig. 6C, lane 2). In addition, LPS stimulation did not increase complex formation (lane 3). This complex was supershifted by an anti-CREB antibody (lane 4) but was not recognized by an anti-c-Jun antibody (lane 5), suggesting that small amounts of CREB still bound to the mutated CRE site. Since these same base substitutions completely abolished the functional activity of the CRE site in transfected cells (Fig.  3B), these results provide additional evidence that c-Jun-containing complexes, rather than CREB, play a crucial role in LPS induction of the TNF-␣ promoter.

DISCUSSION
In this report, we have defined two cis-acting regulatory elements in the human TNF-␣ promoter that mediated maximal LPS induction of TNF-␣ gene expression in cells of monocytic lineage. Region I contained an overlapping Sp1/Egr-1 site (Ϫ182 to Ϫ162), whereas region II (Ϫ120 to Ϫ96) contained CRE and B sites.
Functional studies demonstrated that Egr-1 binding to region I was required for LPS induction of the TNF-␣ promoter in monocytes. Egr-1 protein expression was induced by LPS stimulation (21). The Egr-1 site at Ϫ169 bp is part of the Sp1/Egr-1 overlapping sequence motif. In unstimulated monocytes, Sp1  (Table I). C, nuclear extracts were preincubated with antibodies (2 g) for 20 min before adding probes.

FIG. 6. LPS induction of nuclear factors binding to CRE site.
Nuclear extracts from unstimulated or LPS-stimulated THP-1 cells were probed with oligonucleotide probe spanning the prototypic CRE sequence (A), TNF-␣ promoter wild type CRE sequence (B), or mutant CRE sequence (CREm2, C). Sequences of these oligonucleotides are shown in Table I. For antibody supershift experiments, nuclear extracts were incubated with various antibodies (2 g) for 20 min before adding probes.
binds to this site, whereas upon LPS stimulation it is likely that Egr-1 displaces Sp1 to mediate induction of TNF-␣ promoter activity. Previously, we have shown that Egr-1 can displace Sp1 from a similar overlapping Sp1/Egr-1 site (22). The role of Sp1 bound to this upstream Sp1 site at Ϫ172 is unknown, although our results using a plasmid containing a mutation in the Sp1 site suggest that it does not mediate basal expression. In contrast, mutation of the Sp1 site at Ϫ56 bp dramatically reduces basal expression by 65%. 2 Further functional studies showed that the CRE and B3 sites in region II were required for LPS induction of the TNF-␣ promoter. We demonstrated that B3 (Ϫ97) bound p50/p65 heterodimers. In contrast, we found no role for B1 (Ϫ588) despite its ability to bind p50/p65. A recent study by Trede et al. (11) also showed a role for B3 in LPS induction of the TNF-␣ promoter in THP-1 cells. However, these investigators did not identify other regulatory regions, possibly due to the low level of induction (about 4-fold) of the TNF-␣ promoter (11). In addition, our studies are in agreement with an earlier report (16), showing that p50/p65 binds with much less avidity to B3 than B1 (Fig. 5). Therefore, it is possible that protein-protein interaction of p50/p65 with c-Jun proteins bound to the adjacent CRE site is required to stabilize the formation of a transcriptional complex that mediates induction of the TNF-␣ promoter.
In monocytic cells, the CRE site in the TNF-␣ promoter constitutively bound CREB (Fig. 6). Upon LPS stimulation, the amount of CREB binding remains unchanged, but there was a marked increase in c-Jun binding. LPS induces c-jun expression (23), suggesting that the increases in c-Jun-containing complexes observed in this study were due to de novo protein synthesis. Furthermore, the transactivating activity of these c-Jun complexes may be increased due to post-translational phosphorylation (24). As reported recently, LPS stimulation of THP-1 cells resulted in rapid activation of JNK (25). The phosphorylation of c-Jun by JNK significantly enhances the transactivation potential of these factors (26). The non-consensus CRE sequence of TNF-␣ promoter TGAGCTCA was shown to have a lower binding affinity for c-Jun/ATF-2 complex than a consensus CRE sequence (27). These variations of sequences and the resulting variation in binding of these bZIP proteins could play a role in the modulation of TNF-␣ expression in response to various signals. Importantly, base substitution in the CRE site that reduced CREB binding but abolished c-Jun binding abrogated the functional activity of this site, suggesting that c-Jun-containing complexes bound to region II are required for LPS induction of the TNF-␣ promoter. The molecular mechanisms by which binding of the nuclear proteins at this CRE site regulate the activation of TNF-␣ gene expression await further elucidation.
Parallel to our findings for monocytic cells, it was demonstrated recently that both the B3 site and the adjacent upstream CRE site are required for the calcium-stimulated TNF-␣ transcription in human T cells (4). However, in contrast to our data for monocytes, these investigators reported that the B3 site bound NFATp and not p50/p65 (28). Moreover, contrary to our findings in monocytes, no constitutive or induced CREB protein binding to the CRE site was observed in T cells (4). Instead, the CRE-binding complex was shown to consist almost exclusively of c-Jun/ATF-2 heterodimers. Furthermore, both the CRE site and adjacent upstream Ets site were essential for both basal promoter activity in T cells and responsiveness to PMA (5). It appears that the Ets site did not play a significant role in basal TNF-␣ promoter activity in monocytes, since it was not affected by mutation of the Ets site. 2 Together, these results demonstrate the marked difference in the regulation of the TNF-␣ promoter in human T cells and monocytes.
In this study, we showed that LPS induction of a heterologous promoter by region II required both the CRE and B3 sites, suggesting a functional cooperation between the transcription factors bound to these sites. Similar cooperation has also been demonstrated in the induction of another cytokine gene, IFN-␤ (29), as well as in the expression of E-selectin (30). In both cases, ATF-2/Jun proteins were shown to interact with p50/p65 proteins. In addition, p50/p65 has been shown to physically interact with ATF-2 (29), and c-Jun (31). These direct associations are considered an important mechanism by which transcriptional factors cooperate to induce gene expression.
The novel findings presented here support the notion that concerted participation of proteins bound to the Egr-1, CRE, and B3 sites mediates the induction of the TNF-␣ promoter in human monocytic THP-1 cells in response to LPS. Future studies will determine if similar regulatory pathways control TNF-␣ gene expression in monocytes and macrophages. Further elucidation of the cooperative interactions of transcription factors bound to these cis-acting regulatory elements are essential to our understanding of the transcriptional regulation of the TNF-␣ gene in human monocytes. Understanding of the molecular mechanisms by which these nuclear factors regulate TNF-␣ gene expression should lead to design of specific inhibitors that will counteract the pathological effects of TNF-␣ in various diseases.