CCAAT/Enhancer-binding Proteins (cid:1) and (cid:2) Negatively Influence the Capacity of Tumor Necrosis Factor (cid:1) to Up-regulate the Human Cytomegalovirus IE1/2 Enhancer/Promoter by Nuclear Factor (cid:3) B during Monocyte Differentiation*

Recently we demonstrated that the ability of tumor necrosis factor (cid:1) (TNF (cid:1) ) to stimulate the human cytomegalovirus (HCMV) IE1/2 enhancer/promoter activity in myeloid progenitor-like cells decreases when these cells differentiate into promonocytic cells. In addition, TNF (cid:1) stimulation in the progenitor-like cell line HL-60 was shown to be mediated by nuclear factor (cid:3) B (NF- (cid:3) B) activation and its binding to the 18-base pair sequence motifs of the IE1/2 enhancer. We demonstrate here that the cell differentiation-dependent reduction of TNF (cid:1) stimulation is not due to insufficient NF- (cid:3) B activation but correlates with increased synthesis of the monocyte differentiation-associated factors CCAAT/enhancer-binding protein (C/EBP) (cid:1) and (cid:2) . Overexpression of C/EBP (cid:1) / (cid:2) in HL-60 cells, which normally produce only very small amounts of C/EBP, stimulated the basal activity of the promoter in the absence of NF- (cid:3) B but sup-pressed the stimulatory effect of TNF (cid:1) . A

Several clinical studies have shown that systemic inflammation is a key factor for human cytomegalovirus (HCMV) 1 (re)activation in both immunosuppressed and non-immunosuppressed patients (1)(2)(3)(4)(5)(6). The proinflammatory cytokine tumor necrosis factor ␣ (TNF␣) has been identified as a powerful mediator of HCMV stimulation and reactivation in human and murine monocyte/ macrophage progenitor cells (7,8). These are a main target cell for HCMV latency (7, 9 -11). However, TNF␣ did not directly influence reactivation of the virus in peripheral blood mononuclear leukocytes induced by allogenic stimulation (12) 2 suggesting distinct regulatory events in progenitor and mature peripheral blood mononuclear leukocytes.
The stimulatory effect of TNF␣ on the IE1/2 enhancer/promoter in undifferentiated HL-60 cells is mediated via binding to its p55 type I receptor and subsequent activation of NF-B heterodimer p65/p50 that binds in the nucleus to the 18-bp repetitive sequence motifs in the IE1/2 enhancer (5,15). Interestingly, this TNF␣-induced signal transduction pathway was also identified in U937 and THP-1 cells indicating that the observed differentiation-dependent changes in TNF␣-mediated regulation of the HCMV IE1/2 enhancer/promoter cannot be explained by decreased activation of NF-B p65/p50. The C/EBP family of transcription factors together with AP-1 and CREB/ATF belongs to the bZIP-binding proteins. Its four members C/EBP␣, C/EBP␤, C/EBP␥, C/EBP␦ form homo-or heterodimers in vitro and bind to various C/EBP-binding sites with similar affinities (16 -18). The expression of C/EBP was shown to be differentiation-dependent and is thought to be involved in the regulation of adipocytes and monocyte/macrophages differentiation. While their progenitors express no or only small amounts of C/EBP, mature adipocytes, monocytes/ macrophages, and granulocytes show high levels of this transcription factor (16, 19 -23). C/EBPs are directly involved in the transcriptional regulation of several genes, e.g. adipose-specific genes (24), the serum albumin gene (25,26), the phosphoenolpyruvate-carboxygenase gene (27)(28)(29), the interleukin (IL-) 8 gene (30,31), the IL-6 gene (32), and the human immunodeficiency virus, type I, long terminal repeat (33).
Recent studies have shown that C/EBPs may synergize or antagonize other eukaryotic transcription factors, e.g. NF-B, STAT6, or SP-1 (31,(33)(34)(35)(36). By using IL-8 promoter constructs containing the minimal promoter combined with either NF-Bbinding sites or C/EBP-binding sites or both, synergism and antagonism between these transcription factors have been observed (31,37). The IL-1␣-mediated NF-B-dependent transactivation of the angiotensinogen gene was diminished by C/EBP overexpression (38). On the other hand, STAT6 and C/EBP as well as SP-1 and C/EBP synergistically activated the IL-4 dependent transcription of the germ line ⑀ promoter and the CD11c integrin gene promoter, respectively (35,36).
Here we demonstrate for the first time that C/EBP␣ and -␤ can suppress NF-B-mediated up-regulation of the native HCMV IE1/2 enhancer/promoter activity by interaction with NF-B p65. A novel functionally active C/EBP-binding site directly adjacent to an NF-B-binding site in the IE1/2 enhancer/promoter was identified as the molecular target. In the absence of NF-B, both C/EBP␣ and -␤ stimulated the HCMV IE1/2 enhancer/promoter by binding to this site. In the presence of NF-B the functional interaction between C/EBP and NF-B p65 did not necessarily require binding of C/EBPs to DNA; however, DNA binding enhanced the inhibitory effect of C/EBPs.

EXPERIMENTAL PROCEDURES
Cell Culture, Transfection, and CAT Assays-The cell lines HL-60, U937, and THP-1 (ATCC CCL 240, CRL 1593 and TIB 202) were maintained in RPMI 1640 medium (Biochrom, Berlin, Germany) supplemented with 10% fetal calf serum (both certified for low endotoxin). The cells were shown to be mycoplasma-free by the Mycoplasma Detection Kit (Roche Molecular Biochemicals).
Transfection of the cells by the DEAE-dextran method and the chloramphenicol acetyltransferase (CAT) assay were performed as described previously (14). Aliquots of 5 ϫ 10 6 cells were transfected with 2.5 g of plasmid DNA, and for co-transfection experiments up to 6 g of plasmid DNA per 5 ϫ 10 6 cells were used. After transfection cells were grown in the presence or absence of 5 ng/ml TNF␣ (R & D Systems, Wiesbaden-Nordenstadt, Germany) for 48 h. The protein concentration of cell extracts was determined with the Bradford protein assay kit (Sigma). Transfection efficacy of different plasmids was determined by dot blot hybridization of DNA isolated from comparable cell extracts (identical protein concentration) with a 32 P-labeled BamHI-DNA fragment of plasmid pRR55 containing the CAT gene as described previously (15).
Plasmids, Construction of Enhancer/Promoter Mutants-The CAT reporter gene plasmid pRR55 contains the whole native IE1/2 enhancer/promoter between nucleotide positions Ϫ671 and ϩ52 with respect to the IE1/2 transcription start site upstream of the CAT reporter gene (39). The plasmid p4-18PCAT carries the minimal IE1/2 promoter between positions Ϫ52 and ϩ52 in combination with four in-tandem copies of the 18-bp repetitive sequence motif of the IE1/2 enhancer upstream of the CAT gene (40). Both plasmids were kindly provided by T. Stamminger (Erlangen, Germany). The plasmids p4-18/EBPCAT and p4-18/2EBPCAT were constructed by insertion of one or two copies of the oligonucleotide EBP carrying the HCMV C/EBP-binding motif of the IE1/2 enhancer between nucleotides Ϫ97 and Ϫ80 (Table I) in plasmid p4-18PCAT. Plasmid construction followed the rules described by Stamminger et al. (40). The pMSV/EBP␣ and pMSV/EBP␤ expression plasmids, which contain the cDNA of C/EBP␣ and C/EBP␤ genes under control of the mouse sarcoma virus (MSV) long terminal repeat, were kindly provided by Dr. S. McKnight (Tullarik Inc., South San Francisco, CA). The MSV promoter control plasmid, pMSV, was constructed from the MSV/EBP␣ plasmid. The C/EBP␣-cDNA sequence was eliminated by restriction enzyme digestion with NcoI and religation of the plasmid. The expression plasmids p65 and p50, a kind gift from J. Sinclair, Cambridge, UK, contain the coding sequences for NF-B p65 and p50, respectively, under control of the CMV promoter in the pcDNA3 vector.
Mutant plasmids pEBPmut and pNF-Bmut of the native IE1/2 enhancer/promoter in pRR55 were constructed using the Takara in vitro mutagenesis kit (Boehringer Ingelheim, Germany). For in vitro mutagenesis the HindIII-XbaI fragment of plasmid pRR55 containing the complete IE1/2 enhancer/promoter sequence was recloned in pUC18 HindIII/XbaI. Nucleotide exchanges in the C/EBP-binding site and the adjacent NF-B-binding site were introduced into pRR55 using the EBPmut oligonucleotide (5Ј-TTCCAAAAcGcCGTAACA-3Ј) and the NF-Bmut oligonucleotide (5Ј-ATCAACGcGATTTCCAA-3Ј), respectively. The mutations were confirmed by DNA sequence analysis and re-cloned into pRR55 upstream of the CAT reporter gene (Fig. 1). Electrophoretic Mobility Shift Assay (EMSA)-Preparation of nuclear extracts and EMSA were performed basically as described elsewhere (15). Briefly, 10 g of nuclear extracts from untreated or TNF␣treated (5 ng/ml for 2-5 h) cells were incubated with 0.1-1 ng of radiolabeled oligonucleotides in a 20-l reaction mixture for 20 min at room temperature. The sequences of the oligonucleotides used are listed in Table I. The reaction mixture for C/EBP binding contained 10 mM Hepes, pH 7.9, 50 mM NaCl, 5 mM MgCl 2 , 1 mM dithiothreitol, 1.5 mM EDTA, 1.5 g of poly(dI-dC) and 10% glycerin. Binding of NF-B and CREB-1/ATF-1 was performed as described (15,41). The DNA-protein complexes were separated on a 5% native polyacrylamide gel in Tris glycine buffer. Rabbit antisera against NF-B p65, NF-B p50, ATF-1, C/EBP␣ and C/EBP␤ were obtained from Santa Cruz Biotechnology (Heidelberg, Germany).
Western Blotting-Nuclear proteins (50 g) from untreated or TNF␣treated HL-60, U937, and THP-1 cells were denatured for 3 min at 100°C in Laemmli buffer, separated on a 10% denaturating SDSpolyacrylamide gel, and transferred to nitrocellulose membrane (Schleicher & Schü ll) using a Mini Tank Electroblotter "OWL" (Cambridge, UK). To avoid unspecific binding, membranes were blocked for 24 h in phosphate-buffered saline containing 10% neonatal calf serum, 0.2% Tween, and 0.04% NaN 3 . After washing five times in phosphatebuffered saline, 10% Tween for 10 min at room temperature, the membranes were incubated with the C/EBP␣ antisera (1:2.000; Santa Cruz Biotechnology, Heidelberg, Germany) or, as a control, with an actinspecific antiserum (1:1.000; Santa Cruz Biotechnology, Heidelberg, Germany) for 1 h at room temperature, repeatedly washed, and incubated for 2 h with anti-rabbit antibody Immunopure R Donkey (1:2.000; Pierce), and again washed repeatedly. To visualize the protein-antibody complexes, the membranes were treated with the chemiluminescent substrate "Supersignal R Ultra" (Pierce) and analyzed using a Bioblock scientific (Raytest, Germany). Statistical Analysis-Statistical analysis was performed using the parameter-free Friedman test and Wilcoxson Signed Ranks Test for comparison of paired samples. Differences were considered significant at p Ͻ 0.05.

C/EBP␣ and -␤ Bind to the IE1/2 Enhancer/Promoter
Downstream to NF-B-A novel C/EBP-binding site was identified in the HCMV IE1/2 enhancer between nucleotides Ϫ83 and Ϫ90 with respect to the IE1/2 transcription start site, closely downstream to an NF-B site that is localized between positions Ϫ93 and Ϫ104 (Fig. 1). Specific binding of C/EBP to this sequence was demonstrated in EMSA with nuclear extracts from THP-1 (Fig. 2) and U937 (see below) cells. In the presence of unlabeled oligonucleotides containing either the HCMV-specific or a consensus C/EBP-binding sequence, C/EBP binding could be competed very efficiently (Fig. 2, lanes 2 and 4). In contrast to this, addition of oligonucleotides containing either a mutated C/EBP-binding site ("EBPm") or the HCMV-specific 18-bp NF-B-binding sequence ("18") had no influence on formation of the C/EBP-DNA complex (Fig. 2, lanes 3 and 5). By supershift experiments with antisera specific for C/EBP␣ or -␤, generations of slower migrating protein-DNA complexes were observed indicating that the oligonucleotidebound protein complex consists of both C/EBP␣ and -␤ (Fig. 2, lanes 6, 7, and 9). In control experiments no supershift was observed with antisera specific for ATF-1 (Fig. 2, lane 10), NF-B protein p65 (Fig. 2, lane 11), or NF-B protein p50 (data not shown).
In addition to C/EBP three other proteins of lower molecular weights seem to bind to the EBP oligonucleotide of the HCMV IE1/2 enhancer (marked by asterisks). These proteins did not react with the C/EBP-specific antibodies, and their binding to the DNA was not competed for by the C/EBP consensus oligonucleotide (but by the HCMV EBP oligonucleotide itself, Fig. 2, lanes 2 and 4) indicating that these unknown proteins are different from C/EBP␣/␤. Low amounts of these proteins were also detectable in HL-60 cells; however, the nature of these proteins is still unknown.
Stimulation of the HCMV IE1/2 Enhancer/Promoter by TNF␣ in Different Monocytic Cells Is Inversely Correlated with the Intracellular Level of C/EBP␣/␤-As recently demonstrated, TNF␣ causes a strong stimulation of the HCMV IE1/2 enhancer/promoter activity in HL-60 cells. In the more differentiated premonocytic and monocytic cell lines U937 and THP-1, TNF␣ only moderately stimulates and even reduces the promoter activity, respectively (Fig. 3A, see also Ref. 14). The lack of TNF␣-mediated IE1/2 enhancer/promoter stimulation in U937 and THP-1 is not due to the absence of NF-B activa-tions in these cells. As shown by EMSA NF-B binding activity after TNF␣ treatment was similar in all three cell lines (Fig.  3B, lanes 2, 4, and 6). Furthermore, NF-B activation after TNF␣ treatment followed a comparable time course in all three cell lines (not shown). Interestingly, the expression of the transcription factors C/EBP␣ and ␤ showed a cell-and differentiationdependent pattern. In nuclear extracts from untreated and TNF␣-treated HL-60 cells, no or only very low levels of C/EBP␣ and ␤ were observed using both EMSA (Fig. 3B, lanes 7 and 8) and Western blot analysis (Fig. 3C, I, lanes 1 and 2). Whereas the expression and binding activity of C/EBP␣/␤ in nuclear extracts of premonocytic U937 cells showed intermediate levels, the monocytic THP-1 cells displayed the highest (EMSA: Fig. 3B, lanes 9 and 10 versus 11 and 12; Western blot: Fig. 3C, I, lanes 5 and 6 versus 3 and 4). In comparison, the CREB-1/ ATF-1 binding activities to the 19-bp sequence motif of the IE1/2 enhancer and actin expression were similar in all three cell lines and were not influenced by TNF␣ (Fig. 3B, lanes  13-18; Fig. 3C, II, lanes 1-6), suggesting the specificity of C/EBP expression. The identity of the proteins bound to the 18or 19-bp sequence motifs has been shown earlier (15,41).
C/EBP␣ Exerts Its Inhibitory Effect by Interaction with NF-B p65-In further experiments HL-60, U937, and THP-1 cells were transfected with pRR55 and increasing amounts of the plasmids p65 or p50, encoding the two NF-B proteins found to be involved in TNF␣-dependent regulation of the HCMV IE1/2 enhancer/promoter (15). Whereas p50 alone had no effect on the promoter activity (not shown), NF-B p65 induced a strong dose-dependent stimulation of the IE1/2 enhancer/promoter as observed for TNF␣. As shown for TNF␣ (Fig. 3A), p65-mediated stimulation of the IE1/2 enhancer/ promoter activity was also dependent on differentiation of the target cells (Fig. 6A). In HL-60 cells the effect of p65 was much stronger compared with U937 and THP-1 cells.
To investigate whether C/EBP functionally interacts with NF-B p65, HL-60 cells were co-transfected with pRR55 and increasing amounts of plasmid p65 in the absence or presence of pMSV/EBP␣. As demonstrated in Fig. 6B co-expression of C/EBP␣ resulted in a significant reduction of p65-induced promoter stimulation. Similarly, when HL-60 cells were co-transfected with pRR55, a constant amount of NF-B p65-expressing plasmid, and increasing amounts of pMSV/EBP␣, the stimulatory effect of p65 decreased along increasing expression of C/EBP␣ (not shown).
The Inhibitory Effect of C/EBP␣ and -␤ on NF-B Is Not Dependent on DNA Binding-To analyze further the mechanism by which C/EBP␣/␤ diminishes the NF-B-mediated transcription stimulation, plasmids containing four in-tandem NF-B-binding sites together with one or two C/EBP-binding motifs downstream of the IE1/2 enhancer were constructed (Fig. 7A). HL-60 cells were transfected with these plasmids alone or in combination with the C/EBP␣-expressing plasmid and then incubated in the presence or absence of TNF␣. As HL-60 cells do not express basal levels of C/EBPs, the reporter constructs containing one or two C/EBP-binding sites (p4-18/ EBPCAT and p4-18/2EBPCAT) showed similar basal (not shown) and TNF␣-induced activity as the construct p4-18PCAT containing the NF-B sites only (Fig. 7A, gray bars). There was only a marginal reduction in p4-18/2EBPCAT-compared with p4-18PCAT-and p4-18/EBPCAT-transfected cells which may be due to the increased distance of NF-B sites to the transcription start site. However, if either C/EBP␣ (Fig. 7A) or -␤ (not shown) was overexpressed in the transfected cells as well, the basal activity of promoter constructs increased (not shown), and the stimulatory effects of TNF␣ were inhibited in correlation with the number of C/EBP-binding sites (Fig. 7A, black  bars). While TNF␣ stimulation in p4-18PCAT (without any C/EBP-binding site)-transfected cells reached 7.5 Ϯ 4.9-fold the basal level in the presence of C/EBP␣, in p4-18/EBPCAT-(one C/EBP-binding site) and p4-18/2EBPCAT (two C/EBP-binding sites)-transfected cells CAT expression was stimulated by factor 2.83 Ϯ 0.99 and 1.3 Ϯ 0.22 only (p Ͻ 0.01 by Wilcoxson test for paired samples p4-18CAT versus p4-18/EBPCAT versus p4-18/2EBPCAT and the Friedman test). Differences in transfection efficiency for the different plasmids as a reason for the observed effects could be excluded (Fig. 7B).
Interestingly, however, co-expression of C/EBP␣ strongly reduced the TNF␣-dependent stimulation of CAT expression (from 140-to 8-fold stimulation in the experiment shown in Fig.  7A) in p4-18PCAT-transfected cells (without any C/EBP-binding site), indicating that the inhibitory effect of C/EBP␣ on NF-B is at least in part independent from its binding to DNA. The binding of C/EBPs to DNA (plasmids p4-18/EBPCAT and p4-18/2EBPCAT) further increases its inhibitory activity on NF-B-mediated HCMV IE1/2 enhancer/promoter stimulation. Very similar results were observed in HL-60 cells co-transfected with the C/EBP␤-expressing plasmid pMSV/EBP␤ (data not shown).
These results were confirmed by analyzing a promoter mutant that had lost its ability to bind C/EBP. By in vitro mutagenesis an HCMV IE1/2 enhancer/promoter mutant was constructed in which the C/EBP-binding site was destroyed (pEBPmut) due to the exchange of two thymidine nucleotides by cytidine (Fig. 1). As a control a mutant promoter was constructed in which the NF-B-binding site was disrupted by a G-C transition (pNF-Bmut, Fig. 1). The loss of binding activity for the appropriate transcription factor was confirmed by EMSA (Fig. 8).
By using these mutants and the original plasmid pRR55 containing the wild type (wt) enhancer/promoter, we transfected native HL-60 cells and HL-60 cells already co-transfected with the C/EBP␣-expressing plasmid. Transfection efficiency of the different promoter constructs was comparable (not shown). CAT activity was measured after incubation of the  Fig. 9. TNF␣-dependent stimulation of the wt promoter was reduced in HL-60 cells expressing C/EBP (HL-60/C/EBP␣) when compared with HL-60 cells not expressing significant amounts of C/EBP. As expected, in HL-60 lacking C/EBPs the pEBPmut behaves like the wt enhancer/promoter. In C/EBP␣ (Fig. 9) as well as C/EBP␤ (not shown)-expressing HL-60 cells, pEBPmut (like the wt promoter) showed a reduced response to TNF␣. However, when compared with the wt pro-moter the inhibition was partly but significantly reversed in pEBPmut-transfected HL-60/C/EBP␣. Complete reversion could not be obtained because, as shown above, C/EBPs may act independently from their binding to the DNA. For comparison, the NF-Bmut promoter showed a diminished TNF␣ response in both cell lines independently from C/EBP expression. Basal activities of pRR55 and pNF-Bmut were similar in both native HL-60 and HL-60 cells expressing C/EBPs (HL-60/C/ EBP) but increased in HL-60/C/EBP cells compared with native HL-60 cells as shown for pRR55 (Fig. 4A). However, basal activity of the C/EBP-mutant promoter (pEBPmut) was decreased in HL-60 cells (expressing only very low amounts of C/EBP) by about 50% and strongly decreased (by 90%) in HL-60/C/EBP␣ cells when compared with pRR55 and pNF-Bmut. This indicates that C/EBP realizes its stimulatory effect on the  1 and 2), NFmut-EBP (lanes 3 and 4), NF-EBPmut ( lanes 5 and 6), and NFmut-EBPmut (lanes 7 and 8). The band representing simultaneous binding of C/EBP and NF-B was marked by an arrow. Autoradiograph of one representative experiment is shown.
FIG. 9. Mutations in the C/EBP but not in the NF-B-binding site partly restore the TNF␣ responsiveness of the IE1/2 enhancer/promoter in C/EBP␣-expressing HL-60 cells. HL-60 cells and C/EBP␣-expressing HL-60 cells were each transfected with either plasmid pRR55, pEBPmut, or pNF-Bmut and grown in the absence or presence of 5 ng/ml TNF␣. CAT assay was performed 48 h after transfection. Mean values (Ϯ S.E.) of six independent experiments are shown. pEBPmut versus pRR55 in HL-60/C/EBP␣ was found to be significant (p Ͻ 0.05, Wilcoxon test, symbolized by *). IE1/2 enhancer/promoter predominantly by binding to the DNA.
Cooperative Binding of NF-B and C/EBP to the IE1/2 Enhancer/Promoter-To analyze protein binding to the DNA sequence motif of the IE1/2 enhancer/promoter containing both the NF-B-and the C/EBP-binding site, we performed EMSA with oligonucleotides containing the wt enhancer sequences between nucleotides Ϫ80 and Ϫ109 (compare Fig. 1 and Table  I) and nuclear extracts from unstimulated or TNF␣-treated THP-1 (Fig. 10, A and B) and U937 cells (Fig. 8).
In EMSA with the NF-EBP oligonucleotide and nuclear extracts from TNF␣-treated THP-1 (Fig. 10, A and B) and U937 cells (Fig. 8, lane 2), an additional band with slower migration (Fig. 10A, lane 2, and Fig. 10B, lane 1) was observed that is composed of NF-B and C/EBP proteins. This was indicated by supershift experiments with antisera specific for C/EBP␣, C/EBP␤, and NF-B proteins p65 and p50 (Fig. 10A, lanes  3-12), and by competition experiments using increasing amounts of unlabeled oligonucleotides containing either a NF-B ("18") or a C/EBP-binding site ("EBP") (Fig. 10B, lanes  2-8). In contrast to this, no competition was observed by the EBPmut oligonucleotide or an unrelated control oligonucleotide containing the 17-bp sequence motif (42) of the IE1/2 enhancer/promoter (Fig. 10B, lanes 9 and 10). The C/EBP-and NF-B-containing band did not appear in EMSA with oligonucleotides NFmut-EBP, NF-EBPmut, and NFmut-EBPmut altered either in the NF-B site or the C/EBP site or both (Fig. 8,  lanes 3-8). These results indicate that in the presence of both recognition sites, NF-B and C/EBP can bind the IE1/2 enhancer simultaneously.
To study whether NF-B p65 and C/EBP can form heteromeric complexes that are bound via the NF-B site independent from the presence of a C/EBP site, we performed EMSA using the oligonucleotide NF-EBPmut (in this oligonucleotide the C/EBP-binding site was destroyed by site-directed mutagenesis, see Fig. 8) with nuclear extracts from TNF␣-treated THP-1 cells containing both NF-B and C/EBPs. As shown in Fig. 11 by supershift experiments the protein complex bound by the NF-B-binding site contains NF-B p65 protein (Fig. 11,  lane 3) and at least small amounts of C/EBP␣ (Fig. 11, lane 4), indicating that in the absence of its specific DNA-binding sequence in the DNA molecule C/EBP␣ and NF-B p65 form complexes that bind to the NF-B site. This protein complex could be observed under experimental conditions optimal for NF-B binding but not under conditions optimal for C/EBP binding to DNA. (1-4). TNF␣ seems to be the major player in this process as it is able to activate the HCMV IE1/2 enhancer/promoter in monocytic progenitor cells (2,7,14,15). Interestingly, HCMV-IE antigenpositive monocytes are not detectable in the peripheral blood before day 4 -7 following systemic TNF␣ release, e.g. after OKT3 monoclonal antibody administration (2). The delay between systemic TNF␣ release and the appearance of CMV antigenemia suggests TNF␣ targets latent/persistent HCMV rather in monocytic bone marrow progenitor cells but not in mature monocytes because the latter have a half-life of about 24 h in the blood only. Confirming these in vivo observations, we recently showed that TNF␣ activates the IE1/2 enhancer/ promoter in an NF-B-dependent manner in monocytic progenitor cells but not in the more differentiated monocytic cell lines (14,15). The present study provide a molecular explanation for this interesting phenomenon. TNF␣ activates NF-B in all monocytic cell lines studied, independent from their differenti-ation grade. However, our results suggest that in more mature monocytes the transcriptional activity of NF-B is antagonized by interacting with the transcription factors C/EBP␣ and -␤. This interaction is specific as other pathways triggering the HCMV enhancer/promoter activity (e.g. cAMP/CREB) are not influenced by C/EBPs. 2 By EMSA and Western blot analysis we could demonstrate  6 -8), or NF-B p65 (lanes 9 -11) was added to the reaction mixture. In lane 12 reaction mixture was supplemented with 1.5 l of NF-B p50-specific antiserum. *, supershifted protein-DNA complexes. B, nuclear extract from TNF␣-treated THP-1 cells was incubated with radiolabeled NF-EBP oligonucleotide. For competition nuclear extracts were preincubated with a 100-, 10-, or 5-fold molar excess of unlabeled 18-bp oligonucleotide (18, lanes 2-4), a 100-, 50-, 10-, or 5-fold molar excess of unlabeled EBP oligonucleotide (EBP, lanes  5-8), a 100-fold molar excess of EBPmut oligonucleotide (EBPm, lane 9), or a 100-fold molar excess of an oligonucleotide containing the 17-bp sequence motif of the IE1/2 enhancer/promoter (17, lane 10). that the differentiation-dependent reduction of TNF␣-induced IE1/2 enhancer/promoter stimulation correlates with increased expression of C/EBP␣ and -␤ in higher differentiated monocytic cells, whereas NF-B activation remains unchanged. Undifferentiated HL-60 cells (showing phenotype-like monocyte/granulocyte progenitor cells) express only very low amounts of C/EBPs. Co-transfection of HL-60 cells with pRR55 (IE1/2 enhancer/promoter reporter construct) and C/EBP␣-or ␤-expressing plasmids decreased the stimulatory effect of TNF␣/NF-B on the IE1/2 enhancer/promoter-controlled CAT expression in a concentration-dependent manner. We could also demonstrate that NF-B p65 (but not p50) is responsible and sufficient for stimulation of the HCMV IE1/2 enhancer/promoter. Remarkably, the stimulatory effect of p65 like that of TNF␣ decreases with differentiation of promonocytic cells. Analogous co-expression of C/EBP␣ or -␤ in HL-60 cells that express no or only very low levels of C/EBPs inhibited p65-stimulated transcription. These data strongly indicate that interaction of C/EBPs with NF-B p65 is responsible for the differentiation-dependent changes. In accordance with our findings, others have shown (31,34,37) that C/EBP␣ and -␤ inhibit p65-stimulated transcription of an IL-8 promoter construct containing only NF-B sites or a construct containing an NF-B-and a C/EBP-binding site in F9 cells, although the synergistic or additive effects between the two transcription factors that have also been described were not observed.

In vivo observations have suggested a relationship between systemic inflammation and CMV (re)activation
This complex regulation of the HCMV IE1/2 enhancer/promoter in premonocytic and monocytic cells by TNF␣/NF-B and C/EBP␣/␤ is based on closely spaced adjacent binding sites for NF-B and C/EBP in the IE1/2 enhancer only about 60 bp upstream of the TATA box. Although the nucleotide sequence of the identified C/EBP-binding site does not cognate the C/EBP consensus sequence, binding of C/EBP to this site is highly specific. As shown by EMSA, in TNF␣-treated cells NF-B and C/EBP may bind cooperatively to their adjacent binding sites forming a protein complex composed of both NF-B p50/p65 and C/EBP␣/␤ heterodimers.
However, by a series of different experiments using a promoter mutant lacking the C/EBP-binding site and plasmid constructs containing four NF-B sites and increasing numbers of C/EBP-binding sites, we could clearly demonstrate that although simultaneous binding of C/EBP and NF-B to the DNA is not essential for their antagonistic interaction, it does strengthen the effect, possibly by forming a more stable protein-protein-DNA complex. In the absence of the C/EBP-binding site, C/EBP and NF-B form multimeric complexes bound to the NF-B site. In EMSA with nuclear extracts from TNF␣treated THP-1 cells and the NF-EBPmut oligonucleotide, we found a protein complex bound by the NF-B site which contains both p65 and C/EBP␣. However, at least in vitro the interaction of C/EBP and NF-B in the absence of C/EBP binding to the DNA seems weaker than in the presence of DNA binding as only small amounts of C/EBP were detected in the protein-DNA complex (Fig. 11).
Consistent with our observations C/EBP proteins have been demonstrated to interact physically with NF-B p65 and p50 proteins through their bZIP region that associates with the Rel homology domain of NF-B. This interaction occurs independent from DNA binding and involves the same motifs that play a role in homologous dimer formation of NF-B as well as C/EBP (34,43). Stein et al. (34) also proposed that interaction between NF-B and C/EBP involves heterodimers of NF-B and C/EBP. Consistently, in EMSA we were able to show simultaneous and probably cooperative binding of NF-B p65/ p50 and C/EBP␣/CEBP␤ heterodimers to an oligonucleotide representing the native IE1/2 enhancer region with the adjacent NF-B-and C/EBP-binding sites. From these data it can be speculated that interaction of C/EBP with NF-B p65 interferes with the interplay of p65 with the basal transcription initiation complex. By forming multimeric protein complexes C/EBP may mask the domain of NF-B which interacts with members of the basal transcription complex or an undefined co-factor. Additionally, binding of C/EBP adjacent to NF-B (the C/EBP site is located between NF-B and transcription start site) may cause a sterical hindrance of NF-B to interact with the primary transcription complex. This model is supported by our observation that two (but not only one) C/EBPbinding site(s) between the NF-B and the transcription start sites are necessary for complete inhibition of TNF␣-stimulated transcription.
As demonstrated, in the absence of TNF␣/NF-B, C/EBPs caused a concentration-dependent increase of the HCMV IE1/2 promoter activity. In contrast to the antagonistic inhibition of NF-B, this effect strongly depended on binding to the enhancer. Elimination of the C/EBP-binding site in the promoter mutant pEBPmut abrogated the positive effect of C/EBPs. Moreover, basal activity of the pEBP promoter mutant (pEB-Pmut) but not of the NF-B promoter mutant (pNF-Bmut) was decreased in C/EBP-expressing cells (HL-60/C/EBP␣ and U937 cells) by 50 -90% in comparison with the wt promoter. In HL-60 cells which intrinsically only express very low amounts of C/EBP, the basal activity of the promoter mutant pEBPmut was only slightly reduced. Similarly, using promoter constructs containing only the minimal IE1/2 promoter and one or two C/EBP-binding sites, a strong correlation between the number of C/EBP-binding sites upstream of the minimal IE1/2 promoter and CAT expression in C/EBP-expressing cells was observed. 2 Interestingly, NF-B and C/EBP␣/␤ are both activated by FIG. 11. Binding of an NF-B p65-and C/EBP␣-containing protein complex to the NF-B site. Radiolabeled oligonucleotide NF-EBPmut was incubated with nuclear extract from untreated (co, lane 1) and TNF␣-treated THP-1 cells (T, lanes 2-4). For supershift experiments reaction mixture was supplemented with 1.5 l of NF-B p65specific antiserum (lane 3) or 1.5 l of C/EBP␣-specific antiserum (lane 4). In lanes 3 and 4 supershifted protein-DNA complexes are marked by arrows.
TNF␣ and its binding to the TNF␣ receptor p55 type I (5, 29, 44 -47). NF-B activation involves dissociation of the p50/p65 heterodimer from the cytosolic inhibitor protein IB by phosphorylation and subsequent proteolysis of IB followed by nuclear transport (48,49). In HL-60 cells this process has been shown to be dependent on reactive oxygen radicals and protein kinase C (15). In contrast to this, the mechanism of C/EBP activation by TNF␣ is widely unknown. Yin et al. (45) proposed that the TNF␣-dependent increase of C/EBP in the nuclear fraction involves post-transcriptional mechanisms that promote translocation of the proteins into the nucleus. Similarly, Kowentz-Leutz et al. (50) proposed that C/EBP activation is mediated via de-repression by different kinases, e.g. mitogendependent protein kinases or protein kinase C. Mahoney et al. (51), however, demonstrated that in vitro phosphorylation of C/EBP by protein kinase C resulted in an attenuation of its binding to DNA.
In summary, we were able to demonstrate for the first time that the transcription factors C/EBP␣/␤ expressed in differentiated monocyte/macrophages are involved in regulation of the HCMV IE1/2 enhancer/promoter activity. As shown, C/EBP␣ and -␤ increase the basal activity of the native promoter. But at the same time, C/EBP powerfully suppresses the responsiveness of the promoter to NF-B which is induced, for example, in response to the inflammatory cytokine TNF␣ and which was identified as one key mediator of virus (re)activation in granulocyte/monocyte progenitor cells (1,2,5).
As a result, high TNF␣ serum levels (as observed e.g. in septic patients or OKT3 monoclonal antibody-treated transplant recipients) may trigger the reactivation of the latent HCMV by switching on the IE protein expression in myeloid bone marrow C/EBP-negative progenitor cells via NF-B (1, 2, 5). Up-regulation of C/EBP prevents efficient TNF␣-mediated (re)activation of the virus in differentiated monocyte/macrophages. This explains the delay between systemic TNF␣ release and the appearance of HCMV antigenemia in these patients (2). Moreover, up-regulation of C/EBP in mature monocytes/macrophages by differentiation signals, as well as inflammatory cytokines like TNF␣, may prevent overexpression of HCMV IE1/2 proteins that are involved in positive and negative regulation of virus replication and expression of several cellular genes including interleukin genes, chemokine genes, and genes involved in cell cycle regulation (52)(53)(54).