INTRODUCTION

Despite the intensive investigation on the immunopathogenesis of AIDS,¹ many questions concerning the molecular mechanisms of HIV-1 primary infection and progression remain unanswered (1, 2). Recently, the identification of cohorts of HIV-exposed individuals who remain free of infection over a long period of viral exposure (3) as well as the existence of a small subgroup of HIV-1-infected subjects who are long-term non-progressors, were described (4, 5). Together with recent reports on viral life cycle (6, 7), the above evidence argue that HIV infection and disease progression may ultimately result from the levels of viral gene expression. The regulation of HIV-1 gene transcription depends on the recognition of cis regulatory regions in the 5-long terminal repeat (LTR) by a set of transcription factors which interact with the basal transcriptional complex. These include a TATA box, three Sp1 sites, and a strong enhancer composed of two NF-B sites (8, 9). In addition, a number of binding sites for transcriptional regulatory proteins have been identified 5 to the NF-B enhancer, in the so-called negative regulatory element (NRE) of HIV-1 LTR. The NRE includes binding sites for USF, AP1, NF-AT, and ETS transcription complex, whose activity on HIV-1 gene expression is uncertain (10, 11). Inducible activation of the viral LTR appears to depend principally on the generation of functional NF-B complex and requires a trans-activating protein, Tat, that interacts with a trans-activating responsive element (12, 13, 14, 15). NF-B defines a family of transcription factors composed by members of the NF-B/Rel family, namely NF-B1 (p50), NF-B2 (p52), RelA (p65), RelB, v-Rel, and c-Rel, which share a sequence homology over a 300 amino acids ``rel homology domain'' (16). NF-B proteins form homo- or heterodimers that bind with different affinities to the NF-B enhancer of HIV-1 (17). NF-B activation by different stimuli results from proteolytic degradation of IB (18), IB (19), and from processing of p105 and p100 precursors to p50 and p52, respectively (19, 20, 21), followed by nuclear translocation and DNA binding of NF-B complexes (17).

Stimuli, such as LPS, UV light, and the cytokines interleukin (IL)-1 (IL-1), IL-6, and tumor necrosis factor-, that activate NF-B, are also potent inducers of C/EBP proteins (22, 23, 24, 25). The C/EBP family of transcription factors belongs to a class of DNA binding factors named bZIP proteins, which include C/EBP, C/EBP (also termed LAP, NF-IL6, IL6-DBP, AGP/EBP), C/EBP (previously defined Ig/EBP-1), and C/EBP (NF-IL6) (22). C/EBP is mainly involved in the transcription activation of adipose-specific genes of 3T3-L1 preadipocytes (23). C/EBP and C/EBP are induced in response to inflammatory stimuli such as lipopolysaccharide (LPS), IL-1, IL-6, and tumor necrosis factor- (24, 25). These proteins are all characterized by a leucine zipper domain and by a DNA-binding basic region located in the C-terminal half of the proteins. Members of the C/EBP family can all associate through the leucine zipper domain, and in the case of C/EBP and C/EBP are activated by phosphorylation (26, 27). In addition, C/EBP factors can associate with members of the NF-B/Rel family, generating C/EBP·NF-B complexes which efficiently activate transcription of cellular genes (28, 29). Accordingly, both the C/EBP and NF-B cis sequences have been identified in the regulatory regions of many genes involved in inflammation and immune regulation (16). In fact, adjacent or overlapping binding sites for NF-B and C/EBP factors have been identified in the promoter regions of IL-6, IL-8, and angiotensinogen genes (30, 31, 32). In these cases, NF-B and C/EBP factors cooperate in modulating gene expression by binding to the respective cis sequences (33, 34, 35). This suggests that formation of C/EBP·NF-B complexes may represent a common response to selected stimuli, and may also regulate the transcription of viral genes. In support of this possibility, we show in this study that C/EBP· NF-B heterodimers are generated both in vitro and in vivo, and are potent activators of HIV-1 LTR. A C/EBP cis region was identified in the viral LTR 5-upstream to the NF-B enhancer, and functioned as a negative regulatory element, while the NF-B enhancer was bound and activated by C/EBP·NF-B complexes. By using mutant forms of C/EBP and p50, we identified the domains involved in DNA-binding and in transcription activation.

MATERIALS AND METHODS

The plasmid pC15CAT contains the HIV-1 LTR cloned into the HindIII site of plasmid pSVOCAT; this plasmid and derivative mutant plasmids were obtained from ``NIH AIDS Reserch and Reference Reagent Program.'' The mutant plasmids were: pCD23CAT, which contains the HIV-1-LTR sequences from 117 to +80 located in front of the cat gene, and thus lacking the NRE (negative regulatory element) and retaining the NF-B and Sp-1 binding sites and the trans-activating responsive element region; pCD54CAT, which contains the HIV-1 LTR sequences from 48 to +80 located upstream of the cat gene; pCD54E9CAT and pCD54E8CAT harbor the NF-B enhancer cloned at the XbaI site at 48, in forward and reversed orientation, respectively. These plasmids are shown in Fig. 1. The pHD-C/EBP, hereafter referred to as pC/EBP, expressing the C/EBP gene was obtained from G. Ciliberto. The pEF-NFIL6wt, pEF-NFIL6(S288A), and pEF-NFIL6(Spl), hereafter referred to as pC/EBP, were kindly donated by S. Akira (33). These plasmids express the C/EBP wild-type gene, the C/EBP gene lacking the DNA binding domain (S288A), or the transcriptional activation domain (Spl). pMT2Tp65, pMT2Tp50, and pMT2Tp50(59-60), hereafter referred to as p65, p50, and p50(59-60), carrying the wild-type p65 and p50 cDNA or p50 cDNA with single base substitutions of the DNA-binding domain, respectively, were previously described (36). The pCD52(3XC/EBP)CAT plasmids, harboring three copies of the HIV-1 C/EBP binding site identified at 174 to 166 of HIV-1 LTR (5-AGCATTTCGTCACA-3) in a sense and antisense orientation, were generated by cloning a single copy of a 3X HIV-1 C/EBP oligonucleotide at 65 of pCD52CAT, which contains the HIV-1 LTR sequences from 65 to +80 in front of the cat gene (shown in Table II). The correct sequence was analyzed by direct sequencing (37). To generate the pGEX-C/EBP plasmid, the coding region of C/EBP was excised from pC/EBP by EcoRV-BglII digestion and inserted into compatible sites (SmaI-BamHI) of pGEX-4T3 (Pharmacia, Sweden) using standard techniques (37). The pGEX-C/EBP was obtained by inserting the C/EBP cDNA, excised from pBlue610 (obtained from S. Akira) by SalI-EcoRI digestion, in compatible sites of pGEX-4T3.

Table II. Responsiveness of pCD52CAT and pCD52(3XC/EBP)CAT plasmids to C/EBP and to NF-B factors Target plasmidsa Transactivating plasmidsa CAT activityb Inductionc pCD52CAT None 0.2 1.0 pCD52CAT p50 0.2 1.0 pCD52CAT pC/EBP 0.3 1.5 pCD52CAT pC/EBP 0.3 1.5 pCD52CAT pC/EBP + pC/EBP 0.2 1.0 pCD52CAT pC/EBP + p50 0.3 1.5 pCD52CAT pC/EBP + p50 0.2 1.0 pCD52(3xC/EBP)CAT None 0.5 1.0 pCD52(3xC/EBP)CAT p50 0.6 1.2 pCD52(3xC/EBP)CAT pC/EBP 2.5 5.0 pCD52(3xC/EBP)CAT pC/EBP 4.0 8.0 pCD52(3xC/EBP)CAT pC/EBP + pC/EBP 6.7 13.4 pCD52(3xC/EBP)CAT pC/EBP + p50 7.2 14.4 pCD52(3xC/EBP)CAT pC/EBP + p50 10.7 21.4 a  Ntera-2 cells were transfected with 5 µg of the target plasmids alone or together with 10 µg of p50 and 2 µg of C/EBP and C/EBP expressing plasmids. Target plasmids are derivative of the wild type HIV-1 LTR (pC15CAT). PCD52CAT carries the region of 65 to +80 bp of HIV-1 LTR inserted upstream of the cat gene. In the pCD52(3XC/EBP)CAT plasmid a double strand oligonucleotide corresponding to three copies of the HIV-1 C/EBP is positioned upstream of the HIV-1 LTR sequence of pCD52CAT. b  Determined by scintillation counting of unacetylated and acetylated spots. c  Fold induction of chloramphenicol acetylation induced by the transactivating plasmids is expressed as the ratio of percentages acetylated. The data are representative of three independent experiments.

NTera-2 cells, a human teratocarcinoma cell line, were cultured in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) heat-inactived FCS (Flow Laboratories, Milan, Italy), 3 mM glutamine, and 10 mM Hepes buffer, pH 7.2 (Life Technologies, Inc., Milan, Italy). These cells were transfected by electroporation as described previously (38, 39). Briefly, for transient expression assay cells were washed and resuspended in 0.3 ml of cold RPMI 20% FCS at a concentration of 1.4 × 10⁷/ml with 5 µg of reporter plasmids and different amounts of transactivating plasmids as detailed in the legend to the figures. After incubation of 10 min in ice, the cells were subjected to a double electrical pulse (0.2 kV, 960 microfarads) using a Bio-Rad apparatus (Bio-Rad), recovered, and cultured in Dulbecco's modified Eagle's medium supplemented with 10% FCS. 48 h after transfection, cells were harvested, washed once with phosphate-buffered saline and collected for CAT assay. The transient expression experiments were performed at least 5 times with different plasmid preparations. Transfection efficiency was monitored by co-transfecting the cells with 5 µg of pnls-LacZ plasmid. -Galactosidase activity was assayed by using 50 µg of protein extracts as described (38). To obtain nuclear extracts from NTera-2 cells enriched in NF-B or C/EBP proteins, cells were transfected with the expression plasmids coding for NF-B1 (p50) or C/EBP genes. 24 h after transfection the cells were harvested and used for nuclear extracts preparation, as reported (39). Peripheral blood mononuclear cells were isolated by centrifugation over a Ficoll-Hipaque (Sigma, Milan, Italy) density gradient at 400 × g for 30 min, washed twice with phosphate-buffered saline, and resuspended in Dulbecco's modified Eagle's medium supplemented with 10% FCS. Monocytes were isolated from peripheral blood mononuclear cells by centrifugation through a 46% Percoll (Pharmacia, Uppsala, Sweden) density gradient at 400 × g for 30 min. For LPS treatment, monocytes were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FCS containing 1 µg/ml LPS (Sigma) for 24 h.

Cell extracts were prepared by three cycles of freezing-thawing in 0.25 M Tris, pH 7.8, and CAT assays were performed as described previously (39, 41). Proteins were measured in each cell extract with an assay kit (Bio-Rad) and equal amounts of proteins were analyzed for each sample. Each CAT assay contained 20-50 µg of proteins, 20 µl of 4 mM acetyl-coenzyme A (Boehringer Mannheim, Germany), 1 µl (0.5 µCi) of D-threo-[1.2-¹⁴C]chloramphenicol (DuPont NEN) in a final volume of 150 µl of 0.25 M Tris, pH 7.8. Reactions were incubated for 3 h at 37 °C, extracted with ethyl acetate, dried, and spotted on silica gel plates (Polygram Sil G; Macherey-Nagel, Duren, Germany). Plates were run in a thin-layer chromatography (TLC) tank containing chloroform:methanol (95:5). After 20 h of autoradiography, the TLC plates were cut and samples were counted in a scintillation counter (LS5000TD; Beckman Instruments, Inc., Palo Alto, CA).

EMSA was performed as previously reported (42). 24 h after transfection, cells were harvested, washed once in cold phosphate-buffered saline, and resuspended in lysing buffer (10 mM Hepes, pH 7.9, 1 mM EDTA, 60 mM KCl, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 0.2% (v/v) Nonidet P-40) for 5 min. Nuclei were collected by centrifugation (500 × g for 5 min), rinsed with Nonidet P-40-free lysing buffer, and resuspended in 100 µl of buffer containing 250 mM Tris-HCl, pH 7.8, 20% glycerol, 60 mM KCl, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 mg/ml aprotinin, and 0.5 µg/ml leupeptin. Nuclei were then subjected to three cycles of freezing and thawing. The suspension was cleared by centrifugation (7000 × g for 30 min), and aliquots were immediately tested in a gel retardation assay or stored in liquid phase N₂ until use. Oligonucleotide probes used included: B 5-CAAGGGACTTTCCGCTGGGGACTTTCCAG-3; C/EBP from HIV-1 LTR (from 177 to 164) 5-AGCATTTCGTCACA-3 and mutants HIV-1-C/EBP M₁, 5-AGCAGTTCGTCACA-3; HIV-1-C/EBP M₂, 5-AGCATATCGTCACA-3; HIV1-C/EBP M₃, 5-AGCATTTAGTCACA-3; and HIV-1-C/EBP M₄, 5-AGCATTTCGTCCCA-3; C/EBP from IL-6 promoter (hereafter referred to as IL-6 C/EBP) 5-GATCGGACGTCACATTGCACAATCTTAATAAT-3. Each oligonucleotide was annealed to its complementary strand and end-labeled with [-³²P]ATP (Amersham Corp.) using polynucleotide kinase (New England Biolabs, Beverly, MA). Equal amounts (5 µg) of cell extracts were incubated in a reaction mixture consisting of 20 µl of buffer containing 20% glycerol, 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, and 2 µg of poly[d(I-C)] (Boehringer Mannheim) for 10 min in ice. 1 µl of -³²P-labeled double-strand oligonucleotide (0.2 ng, 5-8 × 10⁴ cpm) was then added with or without a 25- and 50-fold molar excess of competitor wild-type or mutant oligonucleotides. The reactions were incubated at room temperature for 30 min and run on a 6% acrylamide/bisacrylamide (30:1) gel in 22.5 mM Tris borate, 0.5 mM EDTA. Gels were dried and autoradiographed. To identify the individual proteins present in the complexes, polyclonal antisera against p50, p65, C/EBP, and C/EBP (Santa Cruz Biotechnology, Santa Cruz, CA) were used in combination with the EMSA. Antisera (5 µg) were incubated with nuclear extracts (5 µg) for 30 min at 4 °C prior to the addition of poly[d(I-C)] and ³²P-labeled probe as described for the EMSA.

To produce GST-C/EBP and GST-C/EBP proteins, the plasmids pGEX-C/EBP and pGEX-C/EBP were introduced in Escherichia coli strain JM101. Bacteria containing the plasmid were grown to 0.4 OD₆₀₀ and induced with 0.5 mM isopropyl-1-thio--D-galactoside (Promega, Madison, WI) for 2 h. Cells were then collected by centrifugation at 4 °C (3,000 × g for 15 min) and resuspended in EBC buffer containing 50 mM Tris-HCl, pH 8.0, 120 mM NaCl, 0.5% Nonidet P-40, 5 mM dithiothreitol, 10 µg/ml aprotinin, 10 µg/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride. Cells were broken with a French press apparatus and the lysates were clarified by centrifugation at 4 °C and 27,000 × g for 30 min. Proteins were recovered and added to glutathione-Sepharose beads (Pharmacia), previously equilibrated in EBC buffer. After an overnight incubation, the beads were extensively washed and GST-C/EBP and GST-C/EBP were eluted in EBC buffer with 10 mM glutathione. The ³⁵S-labeled p50 proteins were in vitro translated by using TNT^TM coupled Reticulocyte Lysate Sistems (Promega) according to the instructions of the manufacturer. For protein interaction studies, 10 µg of GST or GST-C/EBP or GST-C/EBP proteins were incubated with 15 µl of translation mixture in buffer A (20 mM Hepes, pH 7.9, 50 mM NaCl, 0.2 mM EDTA, 4 mM dithiothreitol, and 5% (v/v) glycerol). The samples were incubated for 2 h at room temperature. At the same time the glutathione-Sepharose beads were washed, blocked in buffer A with 1 mg/ml bovine serum albumin for 2 h, and washed again. These beads were added to the samples. After 3 h, the beads were collected by centrifugation (at 2,000 × g for 10 s) and washed 10 times with buffer A. The pellets were then resuspended in SDS-gel sample buffer (1% SDS, 100 mM Tris-HCl, pH 6.8, 1 mM EDTA, 1% bromophenol blue, 10% (v/v) -mercaptoethanol, 7 M urea) and resolved on 10% SDS-polyacrylamide gel. Gels were treated with Entensify Kit (DuPont NEN), dried, and exposed.

25 ng of biotinylated oligonucleotide corresponding to HIV-1 NF-B, previously bound to streptavidin-conjugated Dynabeads (Dynal, Oslo, Norway) were incubated with 200 µg of nuclear extracts from monocytes stimulated with LPS (1 µg/ml), and 20 µg of poly[d(I-C]), in 200 µl of EMSA buffer at room temperature for 90 min with slow agitation. The DNA-protein complexes were washed three times with EMSA buffer plus 0.5% bovine serum albumin and O.1% Nonidet P-40, using a magnetic particle concentrator, and were solubilized in SDS-gel sample buffer. The eluted proteins were analyzed on 10% SDS-polyacrylamide gel followed by electrophoretic transfer of proteins onto nitrocellulose (Schleicher and Schuell, Milan, Italy) at 15 volts for 20 h in 150 mM glycine, 20 mM Tris-HCl buffer. Anti-C/EBP and anti-p50 antiserum (Santa Cruz Biotechnology) were used as the first antibody. The second antibody was horseradish peroxidase-conjugated goat anti-rabbit IgG (Sigma). The chemiluminescent reaction was used for the detection as suggested by the manufacturer (Amersham Corp.).

RESULTS

HIV-1 LTR are activated by a variety of stimuli, including LPS, IL-1, IL-6, tumor necrosis factor-, and UV light (22, 25). These stimuli induce NF-B/Rel as well as C/EBP factors (16, 24, 25, 31), suggesting that C/EBP proteins may modulate the activity of HIV-1 LTR. To test this possibility, we co-transfected NTera-2 cells with C/EBP and C/EBP expression plasmids along with a HIV-1 LTR CAT plasmid (pC15-CAT, shown in Fig. 1). As shown in Table I, both C/EBP and C/EBP activated the LTR-driven expression of CAT and acted cooperatively in activating the HIV-1 LTR.

Table I. Expression of C/EBP and C/EBP results in the activation of pC15CAT Transactivating plasmida CAT activityb % Acetylation % Fold inductionc None 0.6 1.0 pC/EBP 5.6 9.3 pC/EBP 11.4 19.0 pC/EBP + pC/EBP 22.3 37.1 a  NTera-2 cells were transfected with 5 µg of pC15-CAT plasmid, carrying the wild type HIV-1-LTR (shown in Fig. 1), alone, or together with C/EBP and C/EBP expression plasmids (10 µg). b  Determined at 48 h post-transfection by using 50 µg of cell extract. c  Determined as the ratio of percentages acetylated in the presence or absence of C/EBP-expressing plasmids. Transfections, cell extracts, and CAT assay were performed as described (37, 38). The data are representative of four independent experiments.

The above results suggested the existence of a C/EBP-responsive sequence in the HIV-1 LTR. A computer assisted analysis of the U3 and R regions of HIV-1 identified a region that matched the C/EBP(NF-IL6) consensus 5-(A/C)TTNCNN(A/C)A-3. This sequence, ATTTCGTCA, located at 174/166 upstream of the tandem NF-B cis sequence, was defined as HIV-1 C/EBP (shown in Fig. 1). An oligonucleotide representative of this sequence was tested for C/EBP DNA binding activity in EMSA. As shown in Fig. 2A, nuclear extracts from NTera-2 cells transfected with C/EBP expression plasmids strongly bound to HIV-1 C/EBP. The binding was competitively displaced by either cold HIV-1 C/EBP oligonucleotide, or by unlabeled oligonucleotide corresponding to the C/EBP(NF-IL6) cis element of the IL-6 promoter (31). The specificity of the binding was further demonstrated by using mutant oligonucleotides of HIV-1 C/EBP (shown in Fig. 2B). In fact, single base substitutions of the C/EBP oligonucleotides abolished the capability of mutant oligonucleotides to compete with wild-type C/EBP (Fig. 2C). Similar results were seen with extracts obtained from pC/EBP-transfected NTera-2 cells (not shown). In other experiments, ³²P-labeled mutant oligonucleotides were unable to bind to nuclear factors from C/EBP-transfected cells (not shown).

Co-expression of NF-B1(p50) and C/EBP Results in the Increase in HIV-1 C/EBP Binding Activity

C/EBP and NF-B regulatory sequences are often contiguous in certain promoters, such as the IL-8 promoter (33). The HIV-1 C/EBP elements lie about 60 base pairs from the NF-B enhancer, thus raising the question of whether C/EBP·NF-B complexes bind simultaneously to both C/EBP and B DNA sequences, or whether they alternatively bind to a single regulatory element. These possibilities were tested by transfecting NTera-2 cells, which express very low levels of endogeneous NF-B or C/EBP factors, with plasmids expressing NF-B1 (p50) and C/EBP alone or in combination. The results shown in Fig. 3 indicate that the binding to HIV-1 C/EBP was strongly increased by co-expressing C/EBP and p50. Moreover, the complexes were super-shifted by antibodies to C/EBP or to p50, while an antibody to p65 (relA) was ineffective, suggesting that C/EBP·p50 complexes were generated in vivo, and bound strongly to HIV-1 C/EBP.

Co-expression of NF-B(p50) and of C/EBP results in a increase in the binding activity to HIV-1 C/EBP

Next, we tested nuclear extracts of NTera-2 cells transfected with C/EBP and p50 for binding to a HIV-1 NF-B oligonucleotide. These experiments showed that co-expression of p50 together with increasing amounts of C/EBP led to a parallel increase in the binding activity to the HIV-1 B site, suggesting the in vivo generation of increasing amounts of C/EBP·p50 complexes (Fig. 4). Altogether, the above results indicate that either C/EBP homodimers or C/EBP·p50 complexes were able to bind to both HIV-1 C/EBP or NF-B site (Figs. 3 and 4). Similar results were obtained in the case of C/EBP (not shown).

C/EBP enhances the HIV-1 B binding activity of NF-B (p50).

Functional Cooperation of C/EBP and NF-B1 (p50) in Inducing the Transcriptional Activation of HIV-1 LTR

To test whether the C/EBP·p50 complexes were functional on HIV-1 LTR, we co-transfected C/EBP and/or p50 expressing plasmids together with LTR-CAT plasmids carrying the region of 572/+80 (pC15-CAT), or a truncated region (117/+80) 5 to the CAT gene (pCD23, shown in Fig. 1). The plasmid pCD23-CAT lacks the C/EBP cis sequence, while retaining the two B enhancers and the Sp1 sites. We found that C/EBP activated the pC15-CAT plasmid, and significantly cooperated with p50 (Fig. 5). Moreover, a stronger activation was observed when the pCD23-CAT plasmid was used, suggesting that C/EBP homodimers and C/EBP·p50 complexes acted on a region located at 117/+80 of HIV-1 LTR (Fig. 5). The data also indicated that the C/EBP region located upstream to the NF-B enhancer acted as a negative regulator of C/EBP and C/EBP:p50 heterodimers.

Functional co-operation of NF-B(p50) and C/EBP in inducing the transcriptional activation of HIV-1 LTR.

To test whether the negative function of HIV-1 C/EBP was depending on its location upstream of the B enhancer, we generated the pCD52(3XC/EBP)CAT plasmid, where three copies of the HIV-1 C/EBP were positioned at 65 of the viral LTR. As shown in Table II, the pCD52(3XC/EBP)CAT plasmid was responsive to C/EBP factors, indicating that HIV-1 C/EBP is intrinsically functional and behaves as a negative regulatory element only in the context of the HIV-1 LTR.

To test whether C/EBP·p50 complexes activated HIV-1 LTR through the B enhancer, we used the pCD54E9-CAT and pCD54E8-CAT plasmids, in which the two B sites of the viral LTR were placed 5 to the TATA box in forward and backward orientation, respectively (shown in Fig. 1). Co-expression of C/EBP or p50 showed that both C/EBP and C/EBP·p50 activated pCD54E9-CAT (Table III). In these experiments, a strong synergism of C/EBP and p50 was observed. Moreover, we found that the activation induced by C/EBP·p50 was stronger than the one induced by co-transfecting p50- and p65-expression plasmids, indicating that C/EBP·p50 complexes were efficient activators of B sites (Table III). Similar results were obtained by using the pCD54E8-CAT plasmid (not shown).

Table III. Responsiveness of pC15CAT, pCD54CAT, and pCD54E9CAT plasmids to C/EBP and to NF-B factors Target plasmidsa Transactivating plasmidsa CAT activityb Inductionc pC15CAT None 0.4 1.0 pC15CAT p50 0.5 1.25 pC15CAT pC/EBP 4.0 10.0 pC15CAT pC/EBP 9.0 22.5 pC15CAT pC/EBP + p50 10.0 25.0 pC15CAT pC/EBP + p50 16.8 42.0 pC15CAT p50 + p65 5.2 13.0 pCD54E9CAT None 0.4 1.0 pCD54E9CAT p50 0.3 0.7 pCD54E9CAT pC/EBP 1.5 3.7 pCD54E9CAT pC/EBP 1.6 4.0 pCD54E9CAT pC/EBP + p50 3.8 9.5 pCD54E9CAT pC/EBP + p50 4.6 16.5 pCD54E9CAT p50 + p65 2.5 6.2 pCD54CAT None 0.4 1.0 pCD54CAT p50 0.2 0.5 pCD54CAT pC/EBP 0.6 1.5 pCD54CAT pC/EBP 0.6 1.5 pCD54CAT pC/EBP + p50 0.8 2.0 pCD54CAT pC/EBP + p50 0.5 1.2 pCD54CAT p50 + p65 0.5 1.2 a  Ntera-2 cells were transfected with 5 µg of the target plasmids alone or together with 2 µg of p50 and p65 and 10 µg of C/EBP and C/EBP. Target plasmids are derivatives of the wild type HIV 1 LTR (pC15CAT). PCD54CAT carries the TATA box and the trans-activating responsive element region of HIV-1 LTR inserted upstream of the cat gene. In the pCD54E9CAT plasmid the HIV1 NF-B enhancer is positioned upstream of the TATA in place of the Spl sites (shown in Fig. 1). b  Determined by scintillation counting of unacetylated and acetylated spots. c  Fold induction of chloramphenicol acetylation induced by the transactivating plasmids is expressed as the ratio of percentages acetylated. The data are representative of four independent experiments.

C/EBP Factors Associate in Vitro and Bind to HIV-1 NF-B

The above results suggested that C/EBP and p50 proteins could physically associate to generate transcriptional complexes binding to the HIV-1 B element. To verify this possibility, in vitro translated and [³⁵S]Met-labeled p50 proteins were tested for binding to GST, GST-C/EBP, or GST-C/EBP proteins. As shown in Fig. 6A, p50 was selectively eluted from GST-C/EBP or GST-C/EBP fusion proteins, indicating an in vivo physical association of p50 and C/EBP factors.

p50 and C/EBP physically interact

Next, we tested whether C/EBP·p50 complexes could also form in vivo upon stimulation, and could bind to the HIV-1 B element. Human monocytes were stimulated with LPS, a known inducer of C/EBP and NF-B factors (24, 25, 31, 43). After 24 h, nuclear extracts were incubated with biotinylated HIV-1 B oligonucleotides. Proteins binding to the oligonucleotides were eluted and subjected to immunoblotting with antibodies to C/EBP or to p50. As shown in Fig. 6B, both C/EBP and p50 were recovered from the HIV-1 B oligonucleotide, suggesting the in vivo formation of C/EBP·p50 complexes. In this experiment we were unable to detect any base-line or LPS-induced C/EBP proteins (not shown).

To identify the domains of C/EBP and p50 involved in the transcriptional activity of C/EBP:p50 heterodimers, we took advantage of expression plasmids carrying single base pair substitutions or deletions of functional domains in p50 and C/EBP. These plasmids were cotransfected with pC15-CAT or pCD23-CAT. As shown in Fig. 7A, deletion of C/EBP DNA-binding domain (pC/EBP-S288A) did not affect the C/EBP·p50 functional co-operation (lane e). In contrast, deletion of the C/EBP transactivation domain (pC/EBP-Spl) abolished the transcriptional activity (lane f). Thus, the DNA-binding domain of C/EBP was dispensable for the function of C/EBP·p50. Furthermore, the results suggested that C/EBP·p50 could function by utilizing the DNA-binding domain of p50 to bind to the HIV-1 B element, and the transcription activation domain of C/EBP to trigger the LTR-driven transcription of CAT. To verify this possibility, a plasmid carrying mutations at codons 59-60 of p50 (p50(59-60)), and therefore expressing a mutant form of p50 lacking a functional DNA-binding domain (36), was used in combination with pC/EBP plasmid in transient expression experiments. As shown in Fig. 7B, p50(59-60) did not act as an activator of the HIV-1 LTR when co-expressed with pC/EBP (lane e). Moreover, increasing amounts of p50(59-60) down-regulated the transcriptional activity of C/EBP·p50 complexes (lanes f-h), suggesting the in vivo generation of inactive C/EBP·p50(59-60) complexes.

Identification of the p50 and C/EBP domains involved in functional co-operation.

DISCUSSION

HIV-1 is the etiologic agent for AIDS and causes various clinical and immunological abnormalities, including activation of polyclonal B cells that manifests as hypergammaglobulinemia and auto-antibody production, lymphadenopathy, Kaposi's sarcoma, and lymphoma of the B-cell phenotype (46, 47, 48). Studies on small cohorts of subjects exposed to HIV-1 and who do not develop HIV-1 infection, and individuals who harbor HIV-1 but remain disease free for long periods (49, 50), strongly suggest that the development of AIDS may depend on a dynamic interplay between viral and host gene products acting on HIV-1 gene transcription. In fact, many viral and bacterial gene products, such as EBV-EBNA2 and LPS, can activate the HIV-1 LTR (39, 40), while inflammatory stimuli, including the cytokines IL-1, IL-6, and tumor necrosis factor-, enhance the expression of HIV-1 genes by inducing active NF-B complexes (16, 43). These stimuli can also activate C/EBP proteins binding to cognate cis sequences found in the regulatory regions of a variety of cellular and viral genes (34, 35, 36). Recent evidence indicate that members of the NF-B and C/EBP families of transcription factors can physically associate, generating active transcriptional complexes (28, 29). In fact, both B and C/EBP sites are present in the promoters of cellular genes such as IL-6, IL-8, serum amyloid A, and angiotensinogen (31, 32, 33, 35, 44, 45), suggesting that cooperative activation by C/EBP·NF-B complexes could represent a substantial way of achieving high gene expression. In support of this possibility, we have shown here that C/EBP·p50 and C/EBP·p50 complexes are generated in LPS-stimulated monocytes and in p50- and C/EBP-transfected NTera-2 cells, and act as potent activators of HIV-1 LTR driven gene expression. This activity was consistently stronger than the one induced by p50·p65 complexes. We found that C/EBP·p50 complexes acted on the NF-B enhancer by utilizing the DNA-binding domain of p50 and the transcription activation domain of C/EBP. In these experiments, the mutant form of p50 functioned as negative trans-dominant of the activity of C/EBP·p50, suggesting that it could down-regulate the expression of HIV-1 genes. The sequence matching the C/EBP consensus was found at position 174/166, upstream of the B enhancer and immediately downstream of the so-called NRE of HIV-1 LTR. Functional characterization of this HIV-1 C/EBP revealed that, although it efficiently bound C/EBP homodimers and C/EBP·p50 complexes, its deletion led to an up-regulation of the transcription activity exerted by C/EBP and C/EBP·p50 complexes. This indicates that the HIV-1 C/EBP element functions as a negative regulatory region by possibly squelching C/EBP and C/EBP·p50 transcription complexs. This activity may be due to the relative distance of C/EBP site to the TATA box and to the NF-B enhancer, since positioning of the C/EBP site 5 to the HIV-1 LTR basal promoter restored its enhancer function (Table II). It is noteworthy that C/EBP factors distort DNA upon binding, but they do not introduce a large DNA bending (51). This could make the bound complexes incapable of interacting with the transcription machinery.

The HIV-1 C/EBP overlaps with a consensus (E box) which is a binding site for proteins of the B class of basic-helix-loop-helix-leucine zipper (b-HLH-Zip) family (52). Members of this family include c-Myc, Max, Mad, TFE3, TFEB, and Mxi1 (53, 54, 55, 56, 57, 58), which form homo- and hetero-multimers, and are potentially able to associate with C/EBP or NF-B factors. The role of these factors in the regulation of HIV-1 LTR is, however, uncertain. Recently, the b-HLH-Zip USF protein has been shown to function as a positive regulator of LTR-driven transcription (59). The USF consensus is located at 173 to 157, and overlaps with HIV-1 C/EBP site. As USF is an efficient DNA-bending factor, the positive effects of USF on LTR function could be due to a different DNA bending which could bring USF close to the TATA box of HIV-1 LTR. The NRE of LTR also harbors a variety of cis sequence, such as Ets, AP1, NFAT1, LEF/TCF1a, and nuclear receptor responsive elements (10, 11). The role of these factors in HIV-1 gene expression is, however, unclear in cell culture systems, while they may play a role in the transcription of integrated proviruses, since their binding sequences are conserved in HIV-1 isolates of AIDS patients. The NRE also contains at least two other C/EBP sites (491/483 and 300/292) which could positively contribute to the LTR activation induced by Mycobacterium tubercolosis (60), suggesting that C/EBP sites can function as positive or negative regulators of gene expression depending on their location respective to the TATA box of LTR. The biological relevance of the HIV-1 C/EBP is suggested by the fact that the sequence appears to be conserved in primary isolates from AIDS patients over several years (61). Moreover, the region is functional in the production of negative strand RNA transcripts from a novel HIV-1 promoter (62). Furthermore, HIV-1 isolates carrying linker mutations of the HIV-1 C/EBP at 174/166 have an enhanced infectivity (63), suggesting an in vivo negative role of HIV-1 C/EBP sequence, which is in agreement with our results (Fig. 5).

We found that C/EBP homodimers, as well as C/EBP·p50 complexes activate the HIV-1 LTR-driven transcription by utilizing the B enhancer as a promiscuous docking site. As different members of NF-B/Rel family can associate with individual C/EBP factors (28, 29), the number of possible combinations of homo- or hetero-multimers binding to different consensus sequences is extraordinarily high. This redundancy may imply a fine regulation of gene expression mediated by generation of complexes having different DNA-binding affinities and a large range of transcriptional activation potency. Our results indicate an extraordinary complexity of the regulation of HIV-1 gene expression, and point to the first steps of HIV-1 life cycle, where the rate of transcription of the integrated proviral HIV-1 in response to a stimulus inducing cellular transcription factors is the major determinant affecting the viral gene expression and replication.