Nuclear Factor-κB p65 Mediates the Assembly and Activation of the TNF-Responsive Element of the Murine Monocyte Chemoattractant-1 Gene

Cells and cell culture

NF-κB p50^−/−, p65^−/−, p50/p65^−/−, and wild-type embryonic fibroblasts were provided by Drs. A. Hoffmann and D. Baltimore (Massachusetts Institute of Technology, Cambridge, MA). BALB/3T3 clone A31 and NIH-3T3 fibroblasts were purchased from the American Type Culture Collection (Manassas, VA). Cells were grown in DMEM supplemented with 10% calf serum (HyClone, Logan, UT), penicillin (50 U/ml), streptomycin (50 μg/ml), and l-glutamine (1 mM; Life Sciences, St. Petersburg, FL). Human rTNF was purchased from Genzyme (Cambridge, MA) and used at 500 U/ml. The cells were grown to about 90% confluence before TNF treatment. PDTC was purchased from Sigma and was added 90 min before TNF treatment.

Plasmids

NF-κB p50 and p65 mammalian expression vectors were obtained from Dr. T. Collins (Brigham and Women’s Hospital, Boston, MA). The pJECAT2.6 and pJECAT0.3 CAT reporter constructs containing the region from −2642 to +81 and from −329 to +81 of the murine MCP-1 promoter, respectively, were described previously (4).

Oligo-mediated PCR was used to create point mutations, base insertions, and κB site conversion constructs on the wild-type pJECAT2.6 plasmid. Briefly, complementary mutagenizing oligos containing base changes at the target site (Table I⇓) and two oligos corresponding to the ends of the MCP-1 insertion (containing unique restriction sites) were designed. The overlapping PCR products were made using pJECAT2.6 as the template with the appropriate combination of mutant and end primers. The PCR products were purified and annealed together as the template for a second PCR reaction using only the end primers. Both the final PCR product and pJECAT2.6 plasmid were cut with the same restriction enzymes and ligated together to make the desired construct. The changes in the distal regulatory region were made using SalI and SpeI sites. The coding end oligo was PCATM5 (5′-GACCATGATTACGCCAAGCTTGC), and the noncoding end oligo was JEM3 (5′-TGATCTTGCTAGTCACTGTCCTCC). All mutations were verified by DNA sequencing.

Changes in the proximal regulatory region were made by oligo-mediated PCR followed by a three-way ligation. The coding end oligo was JE 5–1 (5′-CCCATCTGACACTGAGTGGGAGGA), and the noncoding end oligo was JE 3–0 (5′-GGTGGTGGAGGAAGAGAGAG). The final PCR product was cut with Bsu36I. The pCAT-basic vector (Promega, Madison, WI) was cut with XbaI, filled in with Klenow fragment, and then cut with SalI. The pJECAT2.6 plasmid was double digested with SalI and Bsu36I, and the 2.3-kb fragment was isolated. The three fragments were ligated together. The mutagenizing oligos for the proximal regulatory region are listed in Table I⇑.

Constructs juxtaposing the distal region 5′ to the proximal regulatory region were created by PCR using pJECAT2.6 as the template and primer pairs of JEκB5–9 (5′-GCTCTTGGGAGGCTGGGGATTGATGT) and JEκB3–2 (5′-AGTTAGCACAGGAGGCAGCGCAA), and JEκB5–4 (5′-CCCGAAGGGTCTGGGAA) and JEκB3–4 (5′-GGATAAGAGCGTGGAAATTCCC). The pJECAT0.3 was cut with SalI and then filled in. Ligation of the linear blunt-end pJECAT0.3 with the PCR products generated both the direct or forward orientation (p0.3/D0.4F and p0.3/κBF) and the reverse orientation (p0.3/D0.4R and p0.3/κBR) constructs.

Cleavage of pJECAT2.3 (4) with HindIII followed by religation produced pJECAT0.5, a CAT construct carrying the region from −526 to +81 of the murine MCP-1 5′-flanking sequence. To create plasmids containing the distal κB region 5′ to the proximal regulatory region in the forward orientation, the κB region was cloned by PCR using the primers κBSTP5 (5′-TATCGGTAACCTCTCCCGAAGGGTCT) and κBSUP3 (5′-CGTGGAATCCTCATCCTATTCTCGCA) in separate reactions with each of the following templates, p2.6 mκB1, p2.6 mκB2, and p2.6 mHS. The PCR products were cut with the restriction enzymes BstEII and Bsu36I. Ligation of the cleaved PCR products with BstEII/Bsu36I double-digested pJECAT2.6, p2.6 mκB1, and p2.6 mκB2 plasmids generated p2.6/κB1, p2.6/κB2, p2.6 mκB1/κB1, p2.6 mκB2/κB1, and p2.6 mκB2/κB2 constructs. Ligation of the cleaved PCR products with BstEII/Bsu36I double-digested pJECAT0.5 produced p0.5/κB1, p0.5/κB2, and p0.5/mHS constructs.

In vivo genomic footprinting

IVGF was conducted as previously described (4) with modifications in the preparation of genomic DNA. A 10-cm tissue culture plate was treated with DMS (1 μl/ml) for 2 min followed by three washes with cold HBSS. Thereafter, 4 ml of DNAzol (Molecular Research Center) was added to the plate. The lysate was collected into a tube, and genomic DNA was precipitated by addition of 2 ml of 100% ethanol. The DNA pellet was washed twice with 1 ml of a solution containing 70% DNAzol and 30% ethanol and twice with 1 ml of 95% ethanol. The DNA was dissolved in a solution of 10 mM Tris (pH 7.5) plus 1 mM EDTA and cleaved by piperidine. The IVGF primers were described previously (4).

Transfection assays

Transient transfections were conducted in NIH-3T3 cells as described previously (4) with addition of 1 μg of an alkaline phosphatase expression vector (pSV2AlkPhos) as an internal control of transfection efficiency (39). TNF (500 U/ml) was added 24 h post-transfection to one plate, and medium alone was added to the second plate. The cells were harvested 36 h post-transfection and assayed by ELISA as described by the manufacturer (Boehringer Mannheim, Indianapolis, IN). Alkaline phosphatase activity was assayed using a kit from Bio-Rad (Richmond, CA).

TNF induction of murine MCP-1 requires multiple cis elements

TNF induces murine MCP-1 expression at the transcriptional level (40). IVGF of resting and TNF-treated BALB/3T3 cells revealed TNF-induced protein occupancy at multiple sites within the MCP-1 5′-flanking DNA spanning 2.6 kb (4). These sites included two NF-κB binding sites (κB-1 and κB-2), an SP-1 binding site, and several novel sites. To more precisely define the roles of these sites in controlling MCP-1 expression, multiple base pair substitutions were introduced into each element such that the site was randomized, and the resulting constructions were analyzed for activity. The base vector for these constructions was the wild-type pJECAT2.6 construct (Fig. 1⇓), which contains the 2.6 kb of DNA 5′ to the start site of transcription and all sites identified by IVGF in the previous study (4). Due to extremely poor transfection efficiency of BALB/3T3 cells, the resulting mutant constructions were transiently transfected into NIH-3T3 cells, a murine fibroblast cell line that is highly transfectable. MCP-1 RNA blot hybridization showed that NIH-3T3 cells responded to TNF with similar kinetics as the BALB/3T3 cell line (data not shown). Transfected cells were either treated with TNF or left untreated, and the level of the CAT reporter protein was measured by ELISA (Fig. 1⇓A). In the distal regulatory region, mutation of either κB-1 or κB-2 ablated the TNF inducibility, suggesting that both sites are required for TNF induction of MCP-1. In the proximal regulatory region, mutation of the SP-1 binding site substantially decreased both basal and TNF-induced expression, indicating that the SP-1 site is critical for MCP-1 expression. Although an AP-1 binding site was not occupied in control or TNF-treated BALB/3T3 cells in IVGF assays (4), mutation of this site also reduced both basal and TNF-induced CAT expression (p2.6 mAP). Thus, it is possible that the expression of the MCP-1 gene is regulated through the AP-1 site. Mutation of the κB-3 site, a site with similarity to a NF-κB binding site that was found to be protected by IVGF (4), did not affect either basal or TNF-induced CAT expression, suggesting that the site is not required for TNF induction. Alternatively, this site may not be occupied in the NIH-3T3 cells. Unfortunately, promoter sequence polymorphisms in an allele of the MCP-1 gene in NIH-3T3 cells prevents IVGF analysis of MCP-1 in this cell line.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

TNF induces murine MCP-1 through multiple cis-acting control elements. Schematic maps of the MCP-1 upstream DNA sequence with the indicated cis elements and the MCP-1 reporter constructions are shown. Transient transfections were conducted as described in Materials and Methods. CAT ELISA data from three separate experiments were normalized by cotransfected alkaline phosphatase activity, averaged, and plotted as optical density (OD) units. A, Point mutations of individual cis elements identified as occupied by protein by IVGF. The mutated sites are marked with an X. B, Partial mutations of the intervening sequence between the κB sites. The constructions were made with the pJECAT2.6-base plasmid. The sequence between the κB-1 and κB-2 sites is shown. TNF-induced protein occupancy and DMS hypersensitivity are indicated by the open and solid circles, respectively. Lowercase letters represent the mutated bases.

Previous IVGF analyses (4) revealed that the sequence between κB-1 and κB-2 sites became highly hypersensitive to DMS treatment after TNF addition (Fig. 1⇑B). Sequence analyses of this region, termed HS, failed to identify a consensus protein binding site, suggesting that the hypersensitivity could be due to either the binding of a novel factor(s) to the HS sequence or to distortion of the DNA caused by interactions between the factors bound to the adjacent κB sites. If the former is the case, sequence substitutions should have a pronounced effect on TNF induction of the MCP-1 gene. Otherwise, the sequence itself should not be critical for the induction. To distinguish between the two possibilities, constructs containing point mutations in the HS region were created, and TNF inducibility of the constructs was assayed by transient transfection (Fig. 1⇑). Random mutagenesis of the entire HS sequence (10 bp) resulted in the loss of TNF-induced expression (Fig. 1⇑A, p2.6 mHS). Scanning mutagenesis of the HS and the immediate flanking DNA in which 3 bp were mutated in each construction revealed a critical sequence located in the center of the HS that was required for TNF-induced expression (Fig. 1⇑B, p2.6 mHS3). Mutations introduced surrounding these central base pairs had little or no effect on induction by TNF (Fig. 1⇑B, p2.6nHS1, p2.6 mHS2, and p2.6 mHS4). These results suggest that the DMS hypersensitivity is due to the binding of a specific protein(s) (the HS factor) to the middle of the sequence.

κB-1, the HS sequence, and κB-2 function together as an enhancer

The above experiments suggest that TNF inducibility of the distal regulatory region involves at least three TNF-responsive elements: κB-1, the HS sequence, and κB-2. To determine the orientation and position dependence of this region, a series of constructs was made and tested in transient transfections (Fig. 2⇓). The pJECAT0.3, a construct carrying only the proximal region of the MCP-1 promoter, was not inducible by TNF (4). When the distal region was juxtaposed 5′ to the proximal region of pJECAT0.3 in either orientation, TNF inducibility was recovered, suggesting that the distal regulatory region could function independent of the 2 kb of DNA separating the distal and the proximal regulatory regions (p0.3/D0.4F and p0.3/D0.4R). TNF inducibility was also restored when only the sequence from the κB-1 to κB-2 sites (κB region) was fused 5′ to the proximal regulatory region, indicating that the κB region can function independent of site A (p0.3/κBF and p0.3/κBR). The inversion of the distal regulatory region resulted in constructs that were expressed at higher levels in both control and TNF-treated cells (p0.3/D0.4F and p0.3/κBR). This result may be due to a more efficient orientation of the activating factors in relation to the downstream, proximal regulatory elements or general transcription machinery. As described above, mutation of κB-1, κB-2, or the HS sequence also ablated the TNF inducibility (p0.5/κB1, p0.5/κB2, and p0.5/mHS). Additional constructions in which the distal κB region was placed 3′ to the CAT gene in either orientation were inducible by TNF (data not shown). These results suggest that the κB region constitutes a single enhancer element mediating TNF induction of the MCP-1 gene.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

The distal κB sites and the HS sequence function as an enhancer regulating MCP-1 induction. Schematic maps of the MCP-1 reporter constructions are shown. Transient transfection assays were performed and plotted as described in Fig. 1⇑.

Altered phasing of the κB sites does not affect induction by TNF

Stereospecific alignment between adjacent cis elements has been observed in several genes, including those containing κB motifs (36, 41). The finding that all three elements, κB-1, the HS sequence, and κB-2, constitute an enhancer suggested that the factors bound to these elements might have special constraints. To determine whether the order of the elements or spacing between the elements was critical for the inducibility, two mutants were made and assayed by transient transfections. Insertion of 5 or 10 bp between κB-1 and κB-2 did not significantly affect the induced CAT expression (Fig. 3⇓, p2.6I5 and p2.6I10), suggesting that the phasing and distance between the two κB sites were not critical for induction. The κB1 and κB2 sites are not identical in sequence, suggesting that they may serve distinct functions. To determine whether this was the case, constructions in which the sites were identical with each other were made and tested. Conversion of the κB-1 site to a κB-2 site or of the κB-2 site to a κB-1 site also had no effect on induction (p2.6κB11 and p2.6κB22), indicating that TNF induction of MCP-1 does not depend on the unique sequences of κB-1 and κB-2 but merely on the presence of two κB sites.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

TNF inducibility of the distal κB region requires close proximity but not stereospecific interactions of the κB sites. Each construction was made as described in Materials and Methods. Transfection assays were performed and plotted as described in Fig. 1⇑.

The above results indicate that while two κB elements are required, it may be possible to separate the sites from each other. To determine whether the κB elements could function together when separated, a series of constructions was made in which the κB region was duplicated between the distal and proximal regions of the wild-type and κB mutant plasmids. In these constructions, one κB site of the inserted κB region was mutated to determine whether two wild-type κB sites were sufficient for TNF induction (Fig. 3⇑). In comparison with pJECAT2.6 control construction in which the distal region is intact, no significant difference in reporter gene expression was observed when an additional κB site was inserted (p2.6/κB1 and p2.6/κB2). Insertion of either a wild-type κB-1 or κB-2 between the distal and proximal regulatory regions also did not restore TNF inducibility of the κB mutants (p2.6 mκB1/κB1, p2.6 mκB2/κB1, and p2.6 mκB2/κB2). These results indicate that the two κB sites must function in close proximity and that there is considerable flexibility in the orientation of the sites with respect to each other.

TNF induces MCP-1 through NF-κB

Activation and translocation of NF-κB from the cytoplasm to the nucleus are involved in TNF induction of many genes (30). To determine whether activation and translocation of NF-κB by TNF are sufficient to induce MCP-1 expression, transfection experiments were conducted in which the pJECAT2.6 plasmid was either transfected into NIH-3T3 cells alone or with NF-κB p50 and/or p65 expression vectors (Fig. 4⇓A). Basal expression of the reporter gene was not affected by p50 cotransfection. However, TNF-induced CAT expression was lower when the p50 expression vector was cotransfected, suggesting that the p50/p50 homodimer not only does not activate MCP-1 gene expression but may compete with active NF-κB isoforms to inhibit MCP-1 induction when overexpressed. The level of reporter gene expression increased dramatically when p65 and p50/p65 expression vectors were cotransfected, indicating that both the p65/p65 homodimer and the p50/p65 heterodimer are capable of inducing MCP-1. Interestingly, TNF was found to further increase expression of the reporter in these trans-activation assays. This result may be due to an increase in the amount of translocated p65/p50 or may result from the activation or induction of another factor important for the regulation of MCP-1.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

TNF induces MCP-1 through NF-κB. A, The NF-κB p50/p65 heterodimer and p65/p65 homodimer induce MCP-1. Twenty micrograms of the pJECAT2.6 plasmid was transfected into NIH-3T3 cells alone or together with 10 μg of p50 and/or 10 μg of p65 mammalian expression vectors. The pCAT-basic vector was not inducible by TNF and was added to ensure that 40 μg of plasmids were transfected for each construction. Addition of TNF and CAT ELISAs were performed as described in Materials and Methods. Assays were conducted as described above. B, NF-κB-mediated MCP-1 expression requires the same cis-acting elements as those required by TNF. Two micrograms of p50 and 2 μg of p65 were cotransfected with 20 μg of the indicated MCP-1 CAT-reporter constructions. Cells were collected 36 h post-transfection, and the level of CAT protein was assayed by ELISA as described above.

If TNF induces MCP-1 through NF-κB, the cis elements that are required for TNF induction should also be important for NF-κB-mediated expression. To test this hypothesis, the MCP-1 mutation constructs and the NF-κB p50 and p65 expression vectors were cotransfected into NIH-3T3 cells, and the level of CAT reporter was assayed (Fig. 4⇑B). Mutation of κB-1, κB-2, AP-1, or SP-1 resulted in a dramatic decrease in p50/p65-induced CAT expression, indicating that these four sites are also required for NF-κB induction of MCP-1 (Fig. 4⇑B). The pJECAT2.3, a construct lacking the distal regulatory region, was not inducible by TNF (4) or by p50/p65, indicating that NF-κB acts through the distal κB region as anticipated. Mutation of the HS site did not display as robust an effect on NF-κB trans-activation (as that observed with TNF treatment), suggesting that the role of the HS site may not be as crucial when NF-κB subunits are overexpressed. Alternately, this site may only be responsive to TNF stimulation. Similar to the TNF-induced expression in Fig. 1⇑, mutations of the κB-3 site did not affect NF-κB induction, implying that this site is not critical for TNF- or NF-κB-mediated expression of MCP-1.

NF-κB p65 is required for TNF induction of MCP-1- and TNF-induced occupancy of the distal κB region

To further characterize the roles of p50 and p65 in TNF induction of MCP-1, a matched series of wild-type, p50 and p65 knockout cell lines was analyzed for TNF induction of MCP-1 and the ability of TNF to induce the distal regulatory region. The level of MCP-1 mRNA in resting and TNF-treated p50^−/−, p65^−/−, p50/p65^−/−, and wild-type murine embryonic fibroblasts was analyzed by RNA hybridization (Fig. 5⇓A). Basal MCP-1 expression was not detected in any of the cell lines. In agreement with the NF-κB trans-activation transfection experiments, addition of TNF rapidly induced high level expression of MCP-1 in the wild-type and p50^−/− cells but not in the p65^−/− and p50/p65^−/− cells, indicating that p65 is required for TNF induction of MCP-1 in vivo. It is important to note that prolonged treatment (>4 h) of the p50⁻/p65⁻ cells resulted in apoptosis, highlighting the importance of NF-κB in protecting cells from TNF-mediated apoptosis.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

NF-κB p65 is required for TNF induction of MCP-1 and TNF-induced binding of transcription factors to the distal κB region. A, Northern blot analyses of RNA prepared from wild-type (wt) and NF-κB knockout embryonic fibroblasts after TNF treatment for the indicated time are shown with two probes, MCP-1 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH; control). IVGF conducted for the MCP-1 gene in cells treated with TNF for the indicated times are shown for the coding (B) and noncoding (C) strands of the κB regions. Lanes marked with V indicate in vitro methylated DNA, and lanes marked with 0 represent the samples from control, untreated cells. The open and closed circles indicate the bases that were either occupied or became DMS hypersensitive after TNF treatment, respectively. Brackets on the left site of the sequence indicate the κB sites. A sequence summary, labeled as described above, is presented.

Previously, we observed that TNF induced both DMS-resistant and DMS-hypersensitive bases in the distal κB region of MCP-1 (4). To investigate whether activation and translocation of p50- and/or p65-containing isoforms to the nucleus by TNF are required for changes in site occupancy, IVGF of the distal regulatory region was conducted in the wild-type and the κB mutant cells (Fig. 5⇑, B and C). As previously observed, no differences between the in vitro and in vivo methylation protection pattern were discerned in the resting cells (lane V and 0 h). After 30 min of TNF treatment, full occupancy of both the κB-1 and κB-2 sites was observed in the wild-type and p50^−/− cells, but not in the p65^−/− or p50/p65^−/− cells, suggesting that TNF-induced occupancy at both κB sites requires p65. The effect is more easily observed in the noncoding strand (Fig. 5⇑C). In BALB/3T3 cells (wild-type), the κB sites remained occupied after 4 h of TNF treatment (4). Interestingly, the sites remained partially occupied in the wild-type embryonic fibroblast cells, but were faintly occupied in the p50^−/− cells after 4 h of TNF treatment, suggesting that the stability of the induced protein-DNA complex may be NF-κB isoform dependent. As observed in BALB/3T3 cells, TNF treatment induced strong DMS-hypersensitive sites in the HS sequence in the wild-type cells and p50^−/− cells (Fig. 5⇑C). However, no DMS-hypersensitive site was detected in the p65^−/− cells and p50/p65^−/− cells. Thus, it appears that the TNF-induced hypersensitivity of the HS sequence is related to the occupancy of the κB sites and directed by p65.

TNF-induced activation of NF-κB p50-containing isoforms is inhibited by PDTC

Antioxidants such as PDTC have been shown to inhibit TNF-induced gene expression (32, 42). The inhibitory mechanism of PDTC is to block TNF-induced phosphorylation and degradation of IκBα, preventing NF-κB translocation (43). However, we previously observed that in the presence of PDTC, TNF could still induce partial occupancy on the coding strand of the distal κB sites and could partially induce MCP-1 expression in BALB/3T3 cells (4). Therefore, it is possible that some NF-κB isoforms are induced by TNF through PDTC-insensitive pathways. Because expression of NF-κB family members is not affected by targeted disruption of either NF-κB p50 or p65 subunits (29, 44), the NF-κB mutant cells provide an opportunity to investigate this issue (Fig. 6⇓). A slight induction by PDTC can be observed in wild-type and p50^−/− cells in this overexposed gel. This is in contrast to the significant levels observed previously in BALB/3T3 cells by PDTC (4). Such differences may reflect subtle cell type differences, such as the differences in κB isoforms. After the cells were treated with PDTC for 90 min, TNF induction of the MCP-1 was blocked in the wild-type cells but not in the p50^−/− cells. Notably, the low level of TNF-induced expression of MCP-1 in the p65^−/− cells was also totally blocked by PDTC. These results suggest that PDTC strongly inhibits TNF-induced activation of NF-κB p50-containing isoforms. Western blot analyses of IκBα in the control and PDTC-treated cells showed that TNF-induced IκBα degradation was blocked by PDTC in the same manner in both the wild-type and p50^−/− cells (data not shown). Although a preference for one NF-κB factor (p50/65) over another may exist in wild-type cells, it appears as though TNF can activate the different NF-κB isoforms through distinct and potentially independent mechanisms.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

PDTC inhibits the activation of NF-κB p50-containing isoforms by TNF. PDTC (0.1 mM) was added 90 min before addition of TNF. TNF treatment was conducted for 2 h. A Northern blot of RNA prepared from control cells and from cells treated with PDTC, TNF, or PDTC and TNF together is shown probed with MCP-1 cDNA or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA. The autoradiograph shown was overexposed to show the slight induction of MCP-1 RNA in the p65^−/− cell lines.