Regulation of SIVmac239 Basal Long Terminal Repeat Activity and Viral Replication in Macrophages

CCAAT/enhancer-binding protein (C/EBP) β and C/EBP sites in the HIV-1 long terminal repeat (LTR) are crucial for HIV-1 replication in monocyte/macrophages and for the ability of interferon β (IFNβ) to inhibit ongoing active HIV replication in these cells. This IFNβ-mediated down-regulation involves induction of the truncated, dominant-negative isoform of C/EBPβ referred to as liver-enriched transcriptional inhibitory protein (LIP). Although binding of the C/EBPβ isoform to C/EBP sites in the simian immunodeficiency virus (SIV) LTR has previously been examined, the importance of these sites in core promoter-mediated transcription, virus replication, IFNβ-mediated regulation, and the relative binding of the two isoforms (C/EBPβ and LIP) has not been investigated. Here, we specifically examine two C/EBP sites, JC1 (−100 bp) and DS1 (+134 bp), located within the minimal region of the SIV LTR, required for core promoter-mediated transcription and virus replication in macrophages. Our studies revealed that the JC1 but not DS1 C/EBP site is important for basal level transcription, whereas the DS1 C/EBP site is imperative for productive virus replication in primary macrophages. In contrast, either JC1 or DS1 C/EBP site is sufficient to mediate IFNβ-induced down-regulation of SIV LTR activity and virus replication in these cells. We also characterized the differential binding properties of C/EBPβ and LIP to the JC1 and DS1 sites. In conjunction with previous studies from our laboratory, we demonstrate the importance of these sites in virus gene expression, and we propose a model for their role in establishing latency and persistence in macrophages in the brain.


During acute infection, human immunodeficiency virus-(HIV) 2 and simian immunodeficiency virus (SIV)-infected cells
invade the central nervous system (1,2). However, HIV-associated neurocognitive disorders do not usually develop until late in the disease process. Using an accelerated and consistent SIV/ macaque model for HIV/AIDS and central nervous system disease, we have demonstrated that SIV RNA is detected in the brain as early as 7 days postinoculation (p.i.) and peaks by 10 days p.i. (3). Between 14 and 21 days p.i., SIV RNA expression in the brain is down-regulated, at least in part, at a transcriptional level because SIV DNA levels remain constant during this time (3,4). Previous reports in our model have implicated IFN␤ in the suppression of acute SIV replication in the brain (4,5). IFN␤, the predominant type I IFN induced as part of the innate immune response to viral pathogens in the central nervous system (6), has been shown to inhibit HIV/SIV replication in macrophages (7,8), a major source of productive HIV/SIV replication in the brain (5,9).
Transcriptional suppression of the HIV/SIV long terminal repeat (LTR) by IFN␤ involves the induction of a truncated, dominant-negative isoform of the transcription factor CCAAT/enhancer-binding protein ␤ (C/EBP␤), LIP (7,10). There are three known isoforms of C/EBP␤ also referred to as NF-IL6 (nuclear factor-IL6) and LAP (liver-enriched transcriptional activation protein), which are alternatively translated from the same mRNA (11,12). We use the term C/EBP␤ and LIP in this study to refer to the second and third in-frame AUG start site translated products, respectively. C/EBP␤ and LIP belong to the C/EBP family of transcription factors characterized by their basic leucine zipper and highly conserved DNA binding domain (13). LIP is the truncated isoform that retains the DNA binding domain but lacks the transactivation domain (14). As a result, LIP competes with C/EBP␤ for the same DNAbinding sites, and because of its inability to associate with histone acetyltransferases, it antagonizes C/EBP␤-mediated transcriptional activation (14 -17).
C/EBP␤ has been demonstrated as an important regulator of the HIV LTR in macrophages and promonocytic cell lines in vitro (18). In these cells, but not CD4 ϩ lymphocytes, at least one functional C/EBP␤ site in the HIV LTR is necessary and sufficient for maintaining basal level activity (19,20). Similar to the HIV-1 LTR, the SIV LTR ( Fig. 1) has C/EBP sites that bind C/EBP␤ (21). Nonnemacher et al. (21) demonstrated C/EBP␤ binding to four out of five putative C/EBP sites; however, the functional roles of these sites remain to be elucidated. Reports from our laboratory demonstrated that nucleotide sequences spanning from Ϫ225 to ϩ18 bp of SIV LTR relative to the transcription start are sufficient for basal level activity in U937 promonocytic cells (22,23), and the addition of ϩ19 to ϩ149 bp established the minimal region for Tat-mediated transactivation (22,24). There are two distinct C/EBP sites in the Ϫ225 to ϩ149-bp region. The first site, JC1, is located at Ϫ100 bp from the SIV LTR transcriptional start site. JC1 contains overlapping NF-B and C/EBP-binding sequences based on in silico analysis and is protected in DNase I footprint assays (21,23). The second site, DS1, is located at ϩ134 bp and binds C/EBP␤ (21).
We have previously demonstrated that C/EBP␤ activates transcription of the Ϫ225 to ϩ149-bp region of the SIV LTR, whereas LIP suppresses C/EBP␤-mediated transcriptional activation in a dose-dependent manner (4). Using a chromatin immunoprecipitation assay (ChIP), we have demonstrated that C/EBP␤ binds the Ϫ225 to ϩ149-bp core promoter region of the SIV LTR in U937 monocytic cells as well as the SIV LTR in brains of infected macaques examined at 7, 10, and 21 days p.i. In addition, ChIP results demonstrated decreased histone H4 acetylation at the SIV LTR upon co-transfection with LIP into the U937 cells, consistent with the lack of histone acetyltransferase recruitment by LIP (4). In vivo, ChIP demonstrated a similar decrease in acetylated histone H4 at the SIV LTR in brains of animals during the transition from active SIV replication at 7-10 days p.i. to suppression of SIV RNA expression at 21 days p.i. These observations collectively suggested that LIP also binds to the Ϫ225 to ϩ149-bp region and is crucial in the suppression of SIV transcription and virus replication (4). However, we could not specifically assay binding of LIP to the SIV LTR because there are no antibodies that distinguish C/EBP␤ and truncated (LIP) isoforms. Additionally, the contribution of the JC1 and/or DS1 C/EBP sites to transcriptional regulation remained unclear because the two sites are only 200 bp apart, and binding of C/EBP␤ and LIP to these sites could not be resolved using standard ChIP assays. To address these issues in the current studies, FLAG-tagged LIP and Myc-tagged C/EBP␤ were used to examine binding to JC1 and DS1 C/EBP sites in transient ChIP assays.
In this study, transcriptional activity assays of the SIV LTR in primary macrophages demonstrate that the JC1 but not the DS1 C/EBP site is crucial for basal transcription by the core promoter. In the context of the whole virus, however, virus replication assays demonstrate that the DS1 C/EBP site is necessary for productive virus replication. In contrast, either site can mediate IFN␤-induced suppression of promoter (basal)mediated SIV LTR activity and virus replication, independently. We also demonstrate binding of C/EBP␤ and, for the first time, binding of LIP to the JC1 and DS1 C/EBP sites. Finally, we show that the two isoforms have differential binding affinities for JC1 and DS1, which, together with the functional activities of these sites, enable us to propose a molecular model to describe C/EBP␤/LIP regulation of the SIV LTR and their role in establishing viral latency in macrophages in the brain.

EXPERIMENTAL PROCEDURES
Cell Culture-HEK-293T cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum, 10 mM HEPES (Invitrogen), 2 mM L-glutamine, and 0.5 mg/ml gentamicin (Invitrogen) and cultured as described previously (25). CEMx174 cells were cultured as described previously (24). Blood-derived primary macrophages were prepared from healthy human blood, obtained in compliance to Institutional Review Board protocols, or adult rhesus macaques and cultured as described previously (26). Briefly, peripheral blood mononuclear cells were isolated by the Ficoll-Paque method. Ten million (10 7 ) cells were seeded into each well of a 6-well plate and cultured for 7 days in macrophage differentiation media containing RPMI 1640 medium (Invitrogen) supplemented with human AB serum, 10 mM HEPES, 2 mM L-glutamine, 0.5 mg/ml gentamicin, and 100 units/ml macrophage colony-stimulating factor. Under these culture conditions, lymphocytes do not proliferate, and Ͼ 95% of the cells are macrophages (26).
Plasmids-A 390-bp (Ϫ236 to ϩ154 bp) and a 271-bp fragment (Ϫ117 to ϩ154 bp) of the SIV mac 239 LTR was inserted into the pGL4.11 firefly luciferase vector (Promega Corp.) to generate the mid-LTR (mLTR) and mLTRII construct, respectively. The JC1mC/EBP and DS1mC/EBP constructs were generated from the mLTR and mLTRII constructs by site-directed mutagenesis using the following primers: JC1mC/EBP (forward primer, 5Ј-AACAGCAGGGACTGTCCAACCGGGGAT-3Ј; reverse primer, 5Ј-ATCCCCGGTTGGACAGTCCCTGCT-GTT-3Ј) and DS1mC/EBP (forward primer, 5Ј-GCTTGC-CCGCTTCAAGCCCTCTTCAATAA-3Ј; reverse primer, 5Ј-TTATTGAAGAGGGCTTGAAGCGGGCAAGC-3Ј). The JC1/DS1mC/EBP construct was generated from the JC1mC/ EBP construct using the DS1mC/EBP primer set for site-directed mutagenesis. The mLTR and the derivative constructs were used for luciferase assay, whereas the mLTRII and the derivative constructs were used for chromatin immunoprecipitation assay (ChIP). SIV/17EFr full-length 5Ј-and 3ЈLTR mutant viruses were generated by stepwise cloning. 5ЈLTR mutations were introduced by digesting pEGFP-N1 (Clontech) and the SIV/17EFr-pUC19 plasmid with BsrGI and inserting the 1108 -3291-bp fragment of the 5ЈLTR into pEGFP-N1 plasmid, followed by site-directed mutagenesis using the aforemen- tioned primer pairs. The mutated 5ЈLTRs were cloned back into the SIV/17EFr-pUC19 plasmid after BsrGI digestion. Next, SIV/17EFr-pUC19 and the pEGFP-N1 plasmid were digested with EcoR/NheI, thereby inserting the 10,233-13,940-bp fragment of the 3ЈLTR into the pEGFP-N1 plasmid. This construct was then used as template to introduce mutations in the 3ЈLTR JC1 and/or DS1 C/EBP sites. The mutated 3ЈLTR-pEGFP-N1 constructs were digested with NheI/BlpI, thereby inserting the 10,233-11,771-bp fragment of the 3ЈLTR into the respective SIV/17EFr 5ЈLTR-pUC19 mutant plasmids digested with the same enzymes. This generated JC1mC/EBP, DS1mC/EBP, and JC1/DS1mC/CEBP 5Ј-3ЈLTR mutant virus plasmids. The QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) was used to generate all of the above plasmids. The Renilla luciferase reporter plasmid pGL4.74 (hRluc/TK) used for cotransfection in luciferase assays to control for transfection efficiency was obtained from Promega Corp. Expression vectors for C/EBP, pCMV-C/EBP␤ (LAP), and LIP, pCMV-LIP, have been described elsewhere (4). Two FLAG tags were added to the N terminus of LIP (pCMV-LIP was used as template) by inverse mutagenesis using forward primer 5Ј-GAC-TACAAAGACGATGATGACAAGGCGGCCGGCTTCCC-GTTCGCC-3Ј and reverse primer 5Ј-CTTGTCATCATCGT-CTTTGTAGTCCATGGTGGCGCGAATTCGAAG-3Ј (27). A Myc tag was added to the N terminus of the C/EBP␤(LAP) where C/EBP␤ from the pCMV-C/EBP␤(LAP) construct was inserted into the pCMV-Myc vector (Clontech). All plasmid constructs were sequenced to confirm mutations and the correct insertion of fragments.
Blood-derived primary macrophages were plated in macrophage differentiation media as described above. Cells were cotransfected with 3 g of the indicated luciferase reporter construct pGL4.11 and 1 g of the pGL4.74-Renilla-luciferase vector (Promega Corp.). Cells were treated with or without 100 units/ml recombinant human IFN␤-1a (PBL Biomedical Laboratories) 5 h post-transfection. IFN␤ treatment was carried out for 24 h, after which cells were harvested, and luciferase activity was measured.
Luciferase Assay-Blood-derived primary macrophages cotransfected as described above were used to measure firefly and Renilla activity with the Dual-Luciferase reporter assay kit (Promega Corp.) and Fluoroskan Ascent FL luminometer according to the manufacturer's instructions. Primary human macrophages were used instead of rhesus macrophages because the transfection efficiency in rhesus macrophages was too low to assess the effects of mutations. The relative luciferase activity of the constructs was consistent between human blood donors.
SIV Viral Stock Production, Infection of Rhesus Macaque Macrophages, and p27 ELISA-CEMx174 cells were transfected with 12 g of viral DNA as described previously (28), specifically, SIV/17EFr (a macrophage-tropic recombinant chimeric clone of SIV mac 239 that shares 100% sequence homology in the LTR within the region of interest) and the 5Ј-3ЈLTR mutant virus constructs generated above. The TCID 50 assay was performed as described elsewhere (29). SIV RNA from every virus stock was sequenced to verify the presence of the mutations. Rhesus monocyte-derived macrophages were infected with wild-type and mutant viruses at a multiplicity of infection (m.o.i.) of 0.01. After 6 h of infection, cells were washed extensively and cultured in media (1 ml) with or without 100 units/ml IFN␤. Supernatants were collected (1 ml) daily from 1 to 5 days postinfection and replaced with the appropriate media. The level of SIV capsid protein, p27, in supernatants was analyzed by ELISA (SIV p27 ELISA kit, Zeptomatrix) to assess virus replication.
Chromatin Immunoprecipitation Assay (ChIP)-Transient ChIP assays were performed as published previously (32,33) with modifications using the Magna ChIP kit (Millipore). Briefly, cells were treated with 1% formaldehyde for 10 min at room temperature and neutralized with 1ϫ glycine. Nuclei were isolated and sonicated to obtain 200 -500-bp DNA fragments using the Branson sonicator for 15 12-s pulses at 50% amplitude with 1 min of incubation on ice between pulses. One million (10 6 ) cells were used per immunoprecipitation in a 50-l volume and pre-cleared with protein A/G-PLUS-agarose beads (Santa Cruz Biotechnology) for 3 h followed by a 16-h pre-clear with Dynabeads protein G (Invitrogen) at 4°C. Input was determined from 1% of the cell lysate. For immunoprecipitation, pre-cleared cell lysates were incubated with no antibody, 1-4 g of mouse IgG (12-371, Millipore) as control, anti-Myc (sc-9E10X, Santa Cruz Biotechnology), or FLAG M2 monoclonal antibody (Sigma) antibody at 4°C for 4 h. Reverse crosslinked and eluted samples were then used for real time quantitative PCR by SYBR-Green (Qiagen) and PCR using the following primers: JC1F, 5Ј-AAGCTTCGCTGAAACAGCAG-3Ј, and DS1R, 5Ј-TTTGGCATCTTCCATGGTG-3Ј. To verify that the no antibody and mouse IgG controls were negative, PCR was carried out on the same samples used in the qPCR assay and analyzed by agarose gel (supplemental Fig. S4).
Western Blot-Nuclear extracts from HEK-293T cells were prepared as described above. Protein lysates were resolved on 12.5% gradient Tris-HCl polyacrylamide gel (Bio-Rad) and transferred onto polyvinylidene difluoride membranes (Millipore). Blots were blocked at room temperature with 5% milk diluted in Tris-buffered saline/Tween 20 (TBST) and probed with anti-C/EBP␤ polyclonal antibody (sc-150, Santa Cruz Biotechnology) and anti-lamin A/C antibody (sc-7292; Santa Cruz Biotechnology) as a loading control at appropriate dilutions. Membranes were washed with TBST and incubated with appropriate horseradish peroxidase-conjugated secondary antibodies (Dako) for 1 h at room temperature. Membranes were washed with TBST and analyzed with Super Signal West Dura extended duration substrate (Pierce).
Data and Statistical Analysis-Luciferase and ChIP assay figures depict mean signal from triplicate samples Ϯ S.E., whereas p27 ELISA graphs depict mean signal from duplicate samples Ϯ S.E. from three or more independent experiments as indicated in the figures. ChIP data was analyzed by the ⌬⌬Ct method, where all signals are initially normalized to corresponding input followed by normalization to C/EBP␤ or LIP binding to wild-type, mLTRII, construct as indicated. Statistical significance (p Ͻ 0.05) was evaluated using a Student's two-sided t test.
Densitometric analysis of EMSA complexes was carried out using Kodak MI software. In competition experiments, the intensity of the complexes incubated with labeled oligonucleotide alone was set to 100%, and the intensities of complexes in all other lanes were normalized to this. Standard calculations were performed to determine binding affinities. Unlabeled oligonucleotide concentration (in nanomolar) used in the competition EMSA, expressed as Ϫlog[nM], was plotted against band intensity represented as percentage (%) of protein bound to labeled oligonucleotide. Data were fit using nonlinear regression, sigmoidal dose-response curve described by the following equation: Y ϭ min ϩ (max Ϫ min)/(1 ϩ 10 ∧ ((logEC 50 Ϫ X)⅐n), where max is maximum protein binding at lowest competitor concentration (unlabeled oligonucleotide); min is minimum protein binding at highest competitor concentration (unlabeled oligonucleotide); EC 50 is half-maximal protein binding; X is logarithm of unlabeled oligonucleotide concentration, and n is the Hill coefficient (GraphPad Prism version 4.0a software). The following Reaction 1 was used to estimate the K D value of C/EBP␤ and FLAG-LIP to the canonical, JC1, and DS1 C/EBP sites, where C is protein complex; D H is labeled oligonucleotide, and D C is unlabeled oligonucleotide.
The above is a modification of the reaction derived by Sun and Baltimore (34). Because siRNA could not be used to selectively knock down endogenous C/EBP␤ or LIP, we assumed that ectopically expressed C/EBP␤/C/EBP␤ or FLAG-LIP/ FLAG-LIP homodimers predominate in the complexes bound to the labeled oligonucleotide especially considering that pCMV-C/EBP␤(LAP)-transfected 293T cells express undetectable levels of LIP and that pCMV-FLAG-LIP-transfected cells express undetectable levels of C/EBP␤(LAP) (Fig. 3A). With the addition of excess unlabeled oligonucleotides, complexes bound to labeled oligonucleotide were increasingly competed away, shifting the binding of complexes from labeled to unlabeled oligonucleotides, with unbound labeled oligonucleotides running off the gel. At the unlabeled oligonucleotide concentration D C (1/2) , where 50% of complexes bound to labeled oligonucleotide (D H ) are competed away (i.e. EC 50 from the graph), the concentration of CD H is equal to CD C . Therefore, the K D value is equal to the concentration of unlabeled oligonucleotide at 50% competition, i.e. [D C (1/2) ] to the closest approximation (14,34).

Promoter-mediated Transcriptional Activation of the SIV LTR Requires the JC1 but Not the DS1 C/EBP␤ Site, although the DS1 C/EBP Site Is Crucial for Productive Virus Replication in Primary
Macrophages-ChIP assays previously demonstrated that C/EBP␤ binds to the Ϫ225 to ϩ149-bp region of the SIV-LTR in vitro and in vivo (4). The JC1 C/EBP site resides at Ϫ100 bp within the Ϫ225 to ϩ18-bp region required for basal SIV LTR activity, whereas the DS1 C/EBP site resides at ϩ134 bp within the ϩ19 to ϩ149-bp region required for Tat transactivation and apparently dispensable for basal SIV LTR activity (22)(23)(24). However, no study has examined the specific contribution of either site in regulating SIV LTR transcriptional activity. To this end, we transfected primary human macrophages with firefly luciferase constructs ( Fig. 2A) containing Ϫ236 to ϩ154 bp of the SIV LTR, referred to as mid-sized LTR (mLTR); the mLTR construct with mutations in the JC1 C/EBP site (JC1mC/E), DS1 C/EBP site (DS1mC/E), or both JC1 and DS1 C/EBP sites (JC1/DS1mC/E) along with a Renilla luciferase vector to normalize transfection. Mutation of the JC1 C/EBP site (JC1mC/E) reduced LTR activity by 70.4% (p Ͻ 0.001), whereas mutation of the DS1 C/EBP site (DS1mC/E) did not significantly alter LTR activity, consistent with the lack of requirement for the ϩ19 to ϩ149-bp region for basal activity ( Fig. 2A). Mutation of both the JC1 and DS1 C/EBP sites (JC1/DS1mC/E) decreased LTR activity by 73% (p Ͻ 0.001) further confirming that the JC1 C/EBP site is important for transcriptional activation of the SIV LTR in macrophages, whereas the DS1 C/EBP site is largely dispensable.
To better understand the functions of these sites in the context of the whole virus and productive virus replication in primary rhesus macrophages by endogenous C/EBP␤, the same mutations were introduced into the 5Ј-and 3ЈLTR JC1 and/or DS1 C/EBP sites of the infectious molecular clone of the SIV/ 17EFr virus. SIV/17EFr is a macrophage-tropic, recombinant clone of SIV mac 239 that shares 100% LTR sequence homology (26). Primary rhesus macrophages were infected with SIV/ 17E-Fr (wild type) and individual SIV/17EFr viruses with mutations in the JC1, DS1 and JC1/DS1 C/EBP sites. Virus replication was measured from 1 to 5 days p.i. using an SIV capsid protein, p27 ELISA. SIV/17EFr (wild type) and JC1mC/E virus replication was detected at 24 h p.i., peaked at 48 h p.i., and decreased consistently thereafter ( Fig. 2B; data for 4 and 5 days p.i. not shown). In contrast, replication of DS1mC/E and JC1/ DS1mC/E viruses was not detectable until 48 h p.i. (Fig. 2B). Replication of the mutant viruses was compared with wild type, SIV/17EFr, set to 100% at 24 and 48 h p.i. (Fig. 2, C and D, respectively). Mutation of the JC1 C/EBP site (JC1mC/E) decreased virus replication consistently, particularly at 24 h p.i. when replication was significantly lower than wild-type virus by 56% (p ϭ 0.033; Fig. 2C). Although there was a 31% decrease in virus replication of the JC1mC/E mutant compared with the wild type at 48 h p.i. (Fig. 2D), this decrease was not significant, FIGURE 2. JC1 C/EBP site is crucial for basal transcriptional activation of the SIV LTR, whereas DS1 C/EBP site is critical for virus replication. A, functional analysis of mLTR (wild type), JC1, DS1, and JC1/DS1 mutant C/EBP (mC/E) constructs in primary human macrophages were carried out by luciferase assay. Primary human macrophages were transfected with pGL4-firefly luciferase construct DNA containing mLTR (wild type) insert or mLTR constructs with mutations in the indicated C/EBP sites. Co-transfection with pGL4.74 Renilla luciferase vector was used for normalizing transfection efficiency. The activity of mutant constructs is normalized to wild-type mLTR construct, set to 100%. Results are expressed as the mean Ϯ S.E. (n Ն 4). *, p Ͻ 0.001, Student's two-sided paired t test. B, p27 ELISA was carried out to measure virus replication. Primary rhesus macaque macrophages were infected with SIV/17EFr (wild type; open squares), JC1mC/E (closed triangles), DS1mC/E (gray circles), or JC1/DS1mC/E (open diamonds) 5Ј-3ЈLTR mutant viruses at m.o.i. of 0.01. Supernatants (1 ml) were collected at 24, 48, and 72 h p.i. and were used for p27 analysis. C, JC1mC/E mutant virus replication measured by p27 ELISA normalized to SIV/17EFr (wild type)-infected cells set at 100% after 24 h of infection. DS1mC/E and JC1/DS1mC/E mutant virus replication were below the limit of detection at this time point. D, indicated mutant virus replication measured by p27 ELISA normalized to SIV/17EFr set at 100% after 48 h of infection, when SIV/17EFr peak virus replication was observed. Mock-infected (ϫ) p27 values were below the limit of detection. Results are expressed as the mean Ϯ S.E. from three independent experiments carried out in duplicate; p values calculated by Student's two-sided paired t test.
suggesting that the JC1 C/EBP site may be important for initial cycles of infection within a cell and the initial round of transcription. Of note, Muesing et al. (35) have also demonstrated that upstream LTR regulatory elements are not essential for transactivation in their HIV-1 LTR studies in keeping with our observations here.
The SIV/17E-Fr viruses with mutations in the DS1 (DS1mC/E) and JC1/DS1 C/EBP (JC1/DS1mC/E) sites had dramatically lowered virus replication and were below the limit of detection until 48 h p.i., when replication was 95% (p ϭ 0.013) and 97% (p Ͻ 0.001) lower than wild-type virus, respectively (Fig. 2D). These observations indicate that in the context of the whole virus, the DS1 C/EBP site is crucial for productive virus replication. These functional data again corroborate previous observations that the ϩ19to ϩ149-bp region is necessary for Tat transactivation (22,24). The DS1 C/EBP site is 9 bp downstream of the 3Ј end of the transactivation-response element (TAR), spanning ϩ1 to ϩ126 bp (36,37). This region forms a secondary RNA structure upon transcription (35,38). Tat binds this element in association with transcription elongation factor (p-TEFb), TAR-RNA-binding protein, as well as several other DNA binding transcription factors and functions to enhance RNA polymerase II processivity (39 -41). Using the Mfold software that analyzes secondary RNA structure, we verified that the DS1 C/EBP site mutation did not alter the RNA structure of the TAR element (data not shown).
C/EBP␤ Binds to JC1 and DS1 C/EBP Sites-In the studies above, we have demonstrated that the JC1 C/EBP site is important for basal transcription, and the DS1 C/EBP site is critical for productive virus replication in primary human and rhesus macrophages; therefore, we next characterized binding of C/EBP␤ to these sites individually. To be consistent with previous studies by several groups, including those studies of C/EBP␤-mediated regulation of the HIV LTR (15,20), we used the rat homologs of C/EBP␤ (LAP) and LIP (42)(43)(44)(45)(46). We carried out EMSAs with these proteins after confirming (supplemental Fig. S1) that the binding characteristics of the rat homologs are very similar to the human homologs, hC/EBP␤ (NF-IL6) and hLIP, and recognize the same C/EBP consensus sequence as described by Osada et al. (31).
EMSAs were conducted using nuclear extracts enriched in the respective proteins of interest from HEK-293T cells. The levels of endogenous C/EBP␤ and LIP in HEK-293T cells are undetectable as reported previously (42), and we confirmed this by Western blot (Fig. 3A). We first examined the binding of C/EBP␤ to the JC1 and DS1 C/EBP sites. Complexes formed using the canonical C/EBP, JC1 and DS1 probes (Fig. 3, B and D,  lanes 1 and 3; see arrows), and the complexes were supershifted or abrogated by the addition of anti-C/EBP antibody (Fig. 3, B  and D, lanes 2 and 4), demonstrating the binding of C/EBP␤ to each site. Importantly, no complexes were formed with JC1 and DS1 probes containing point mutations in the core C/EBPbinding site (Fig. 3, B and D, lanes 5 and 6; see arrows). It should be noted as well that the JC1 probe with the mutation in the JC1 site but not the NF-B site retains the ability to bind NF-B (top complex; supplemental Fig. S2), indicating that C/EBP␤ binding to JC1 occurs independently of NF-B binding.
Specificity of C/EBP␤ binding to JC1 and DS1 probes was confirmed by competition EMSAs, where the C/EBP␤-containing complexes were competed with excess unlabeled canonical C/EBP probe (Fig. 3, C and E, lanes 1 and 2; see arrows) but not canonical C/EBP probe containing mutations in the core C/EBP-binding site (Fig. 3, C and E, lane 3). There were no C/EBP␤ complexes formed on mutated JC1 and DS1 probes (Fig. 3, C and E, lane 4), and no competition could be detected with wild-type or mutated canonical C/EBP probes (Fig. 3, C  and E, lanes 5 and 6).
To substantiate the above findings, i.e. binding of C/EBP␤ to these sites, in a more biologically relevant in vivo setting, we performed ChIP assays. With standard ChIP, characterization of binding sites at high resolution is difficult to achieve as shearing DNA yields fragments between 200 and 500 bp, and fragments smaller than this makes the assay inefficient. To resolve C/EBP␤ binding to the JC1 and DS1 C/EBP sites, which are only 200 bp apart, we modified the standard ChIP assay. We initially tried using restriction enzymes to resolve the two sites prior to carrying out immunoprecipitation; however, contamination with incomplete digestion fragments interfered with the assay and could not be overcome (data not shown). Modifying previously published methods (32,33), we transiently transfected HEK-293T cells with a 271-bp fragment of the LTR, mLTRII (wild type), that encompassed just the JC1 and DS1 C/EBP sites with and without the mutations in the JC1, DS1, or both JC1/ DS1 C/EBP sites. In addition, we co-transfected the pCMV-Myc-C/EBP␤ vector, where the Myc tag was used to aid in efficient and clean immunoprecipitation. We carried out standard immunoprecipitation followed by quantitative PCR (qPCR; Fig. 3F). The data are presented as the percentage binding of C/EBP␤ to the indicated constructs normalized to mLTRII binding, set at 100%. Individually mutating the JC1 or DS1 C/EBP sites led to a significant decrease in binding by 63 (p ϭ 0.009) and 66% (p ϭ 0.016), respectively. Furthermore, mutating both sites led to 85% decrease in binding of C/EBP␤ (p ϭ 0.002), confirming that C/EBP␤ binds to both the JC1 and DS1 sites. qPCR signals for no antibody and mouse IgG controls were less than 0.1% relative to input. We further verified that the no antibody and mouse IgG controls were negative by carrying out PCR on the above samples and analyzing them on agarose gel (supplemental Fig. S4A). To ensure there was the same amount of protein in each immunoprecipitation, we measured myc-C/EBP␤ expression by Western blot and observed no significant differences (supplemental Fig. S5). Collectively, the EMSA and ChIP data demonstrate not only that C/EBP␤ binds to both JC1 and DS1 but also that binding requires an intact C/EBP site.

IFN␤-induced Suppression of SIV LTR Transcriptional Activity and Virus Replication in Primary Macrophages
Is Mediated by Either JC1 or DS1 C/EBP Sites-We have previously demonstrated that IFN␤-mediated suppression of SIV LTR activity correlates with increased expression of LIP relative to C/EBP␤ in monocytic cell lines and in the brain of SIV-infected macaques (4,5). Having demonstrated that C/EBP␤ regulates LTR activity through the JC1 and DS1 C/EBP sites, we next examined the functions of these sites in IFN␤-mediated suppression of SIV LTR activity. Primary human macrophages were transfected with mLTR, JC1mC/E, DS1mC/E, and JC1/ DS1mC/E firefly luciferase constructs (Fig. 4A), and each construct was co-transfected with a Renilla luciferase construct, to use as an internal transfection control. Cells were then treated with or without 100 units/ml recombinant human IFN␤ for 24 h post-transfection. Each sample was initially normalized for transfection efficiency (firefly/Renilla), and then to assess the effect of IFN␤, the normalized luciferase activity of the IFN␤-treated cells was compared with the normalized luciferase activity of untreated control cells transfected with the same firefly constructs. Treatment of wild-type mLTRtransfected cells with 100 units/ml IFN␤ for 24 h significantly decreased LTR activity by 59% (p ϭ 0.0035). A similar decrease in LTR activity was observed in cells transfected with the JC1mC/E or DS1mC/E construct treated with 100 units/ml IFN␤, where the decrease in LTR activity was 51 (p ϭ 0.025) and 48% (p ϭ 0.0091), respectively. However, a lack of IFN␤-mediated suppression was observed with mutation of both the JC1 and DS1 C/EBP sites (JC1/DS1mC/E). These data demonstrate that JC1 and DS1 C/EBP sites are individually important and that at least one of these sites is required for the suppressive effect of IFN␤ on SIV LTR transcriptional activity in macrophages.
Initial studies from our laboratory have demonstrated that IFN␤ induces LIP expression at 24 h posttreatment and suppresses SIV replication (4,47). More recent studies from our laboratory have demonstrated that inhibition of IFN␤-induced LIP expression also inhibits IFN␤-mediated suppression of SIV replication (47). Therefore, we investigated the contribution of these C/EBP sites in the IFN␤-induced, LIP-mediated down-regulation of SIV virus replication (Fig. 4, B and C). Primary rhesus macaque macrophages were infected with SIV/17EFr and SIV/17EFr viruses containing mutations in the JC1, DS1, or JC1/DS1 C/EBP sites at an m.o.i. of 0.01 followed by 100 units/ml IFN␤ treatment (or not) 6 h post-infection. The supernatants (1-5 days p.i.) were used to monitor virus replication by quantitating the SIV viral capsid protein, p27. At 24 h p.i. (Fig. 4B), there was a significant decrease in SIV/17EFr (83.5%; p ϭ 0.015) and JC1mC/E (93%; p ϭ 0.0016) virus replication upon treatment with IFN␤, which was maintained at 48 h p.i. (Fig.  4C), when compared with virus replication in the infected untreated controls. In the case of the SIV/17EFr-DS1mC/E and JC1/ DS1mC/E mutated viruses, replication in the infected untreated controls was not detected until 48 h p.i. At this point IFN␤ treatment decreased virus replication significantly, down-regulating it to levels below the limit of detection. We also carried out infections at an m.o.i. of 0.05 followed by 1000 units/ml IFN␤ treatment and observed similar suppressive effects regardless of the mutations (data not shown). These findings were surprising in light of our previous experiments demonstrating that siRNA-mediated knockdown of CUGBP1 (CUG-repeat RNA-binding protein 1), the RNA-binding protein required for IFN␤-induced LIP expression, not only inhibited LIP expression but also the ability of IFN␤ to suppress SIV replication (47). Thus, we suspected that the ability of IFN␤ to suppress replication of SIV/17EFr-JC1mC/E, DS1mC/E, and JC1/DS1mC/E mutant viruses may be due to the  2, 4, and 6). C, nuclear extract incubated with JC1 or JC1mC/EBP-P 32 radiolabeled probes alone (lanes 1 and 4); competed with excess unlabeled canonical C/EBP (lanes 2 and 5) or mutated canonical C/EBP probes (lanes 3 and 6). D, nuclear extract incubated with canonical C/EBP, DS1, or DS1mC/EBP 32 P-radiolabeled probes alone (lanes 1, 3, and 5) or with anti-C/EBP␤ antibody (lanes 2, 4, and 6). E, nuclear extract incubated with DS1 or DS1mC/EBP 32 P-radiolabeled probes alone (lanes 1 and 4), competed with excess unlabeled canonical C/EBP (lanes 2 and 5), or mutated canonical C/EBP probes (lanes 3 and 6). Unlabeled canonical and mutated canonical C/EBP probes were in 1000-or 500-fold molar excess of labeled JC1 and DS1 probes, respectively. These concentrations were chosen as they were in excess of that necessary for complete competition of C/EBP␤-containing complex bound to labeled probes as shown in Fig. 6. Free probe was run off the gel to resolve bands. F, C/EBP␤ binding to the JC1 and DS1 C/EBP sites was assessed by ChIP assays. HEK-293T cells were transfected with mLTRII (Ϫ117 to ϩ154 bp; wild type), JC1mC/E, DS1mC/E, or JC1/Ds1mC/E constructs and co-transfected with myc-C/EBP␤ expression vector. Standard ChIP-qPCR was carried out on these cells as described under "Experimental Procedures." Data are presented as the percentage of C/EBP␤ binding to mutant constructs compared with wild type, mLTRII construct. qPCR signal for no antibody and mouse IgG controls were less than 0.1% of the input. Results are expressed as the mean Ϯ S. E. (n ϭ 3); p values were calculated by Student's two-sided paired t test. ability of LIP to bind JC1 and DS1 C/EBP sites despite the mutations.
Dominant-negative C/EBP␤ Isoform, LIP, Binds to JC1 and DS1 C/EBP Sites-Based on our previous reports described above (47) and the fact that LIP has been shown to bind the albumin D element C/EBP site with higher affinity than C/EBP␤ (14), we investigated whether LIP would bind to the mutated JC1 and DS1 sites, perhaps accounting for the ability of IFN␤ to suppress JC1 and DS1 mutant virus replication. Our previous studies have shown by ChIP that C/EBP␤ binds to Ϫ225 to ϩ149-bp region of the SIV LTR in vitro and in vivo, and at that time, however, we were unable to demonstrate the specific binding of LIP to the LTR because there is no commercially available anti-C/EBP␤ antibody that distinguishes the fulllength C/EBP␤ and the truncated, dominant-negative isoform of C/EBP␤, LIP. Furthermore, we could not use siRNA directed toward C/EBP␤ mRNA because LIP is expressed from the same mRNA as wild-type C/EBP␤ using an alternative translation start site such that an siRNA approach would down-regulate expression of both isoforms; hence, the logic behind our previous strategy to knock down expression of the CUGBP1 protein that is required for the alternative translation of LIP (47). Therefore, in this study we generated a FLAG-tagged LIP expression vector (FLAG-LIP) as a novel tool to examine the direct and specific binding of LIP to JC1 and DS1 C/EBP sites.
EMSAs performed with nuclear extracts prepared from HEK-293T cells that were transfected with the FLAG-LIP expression vector revealed that LIP complexes formed on canonical C/EBP, JC1, and DS1 probes (Fig. 5, A and C, lanes 1  and 3; see arrows) and were supershifted or abrogated by the addition of anti-FLAG antibody (Fig. 5, A and C, lanes 2 and 4), demonstrating the presence of LIP on each probe. (Binding of the FLAG-LIP protein to labeled canonical and JC1 C/EBP probe was also demonstrated with C/EBP␤ antibody (supplemental Fig. S3, A and B)). We also verified that the addition of the FLAG epitope did not interfere with the activity of LIP in the 293T cells (supplemental Fig. S3C). Importantly, no complexes formed with JC1 and DS1 probes containing point mutations in the core C/EBP-binding site (Fig. 5, A and C, lanes 5 and 6; see arrows).
Competition EMSAs were carried out to examine the specificity of LIP binding to JC1 and DS1 probes. LIP-containing complexes on JC1 and DS1 probes (Fig. 5, B and D, lane 1, see  arrows) were competed away with excess unlabeled canonical C/EBP probe (Fig. 5, B and D, lane 2) but not with canonical C/EBP probe containing mutations in the core C/EBP-binding site (Fig. 5, B and D, lane 3). As no LIP complexes formed on mutated JC1 and DS1 probes (Fig. 5, B and D, lane 4), competition was not detected with wild-type or mutated canonical C/EBP probes (Fig. 5 B and D, lanes 5 and 6). Together, these data demonstrate that LIP binds to both the JC1 and DS1 C/EBP sites.
To confirm binding of LIP to the JC1 and DS1 C/EBP sites, we performed ChIP assays (Fig. 5E). HEK-293T cells were transfected with mLTRII (wild type) or the C/EBP site mutant constructs as indicated and co-transfected with the pCMV-FLAG-LIP construct before carrying out qPCR. Mutating either the JC1 or DS1 C/EBP site did not significantly hinder LIP binding to this region, which was surprising considering the EMSA data, where mutating the sites clearly abolished LIP binding thus underscoring the sensitivity of quantitative PCR. However, mutating both the JC1/DS1 C/EBP sites decreased binding by 45.5% (p ϭ 0.009), indicating that at least one intact site is necessary for maximum binding of LIP to this region. These results implicate residual LIP binding even in the presence of mutations when taking the Ϫ117 to ϩ154-bp region into consideration as opposed to just the individual JC1 and DS1 C/EBP sites as in the EMSAs. Importantly, the ability of LIP to associate with these sites, despite mutations, explains in part the IFN␤-mediated suppression of virus replication unaffected by the JC1, DS1, and JC1/DS1mC/EBP mutations. The no antibody and IgG negative controls in the qPCR results confirmed insignificant background, as the qPCR signal for these samples was less than 0.1% of the input. This result was further verified by carrying out PCR on the same samples followed by analysis on agarose gel (supplemental Fig. S4B). We also confirmed that the FLAG-LIP expression levels were consistent in each immunoprecipitation by Western blot and saw no significant differences (supplemental Fig. S5). Collectively, the EMSA and ChIP data presented here demonstrate LIP binding to both JC1 and DS1 C/EBP sites; however, in contrast to C/EBP␤ binding, LIP was observed to have residual binding to the Ϫ117 to ϩ154-bp region despite mutation in both the sites.
C/EBP␤ Has a Higher Affinity for the JC1 C/EBP Site than the DS1 C/EBP Site-The functional analyses, luciferase assay and virus replication assay (p27 ELISA assay), support the conclusions that the JC1 and DS1 C/EBP sites differentially regulate not only basal SIV LTR activity and productive virus replication but also differentially down-regulate SIV LTR activity in response to IFN␤ by endogenous C/EBP␤ proteins in primary macrophages. Therefore, we hypothesized that this may be the result of the two C/EBP␤ isoforms having different affinities for the site, and to this end we determined the binding affinities of C/EBP␤ and the dominant-negative FLAG-LIP isoform for the JC1 and DS1 C/EBP sites using gradient competition EMSA.
To examine this hypothesis and validate the binding approaches used, we first conducted competition EMSA to determine the binding affinity of C/EBP␤ and FLAG-LIP for the canonical C/EBP site, and we compared it with previously published data by Descombes and Schibler (14). To this end, we incubated nuclear extracts containing either C/EBP␤ or FLAG-LIP isoform from HEK-293T cells with labeled JC1 oligonucleotides and increasing concentrations of unlabeled canonical C/EBP oligonucleotides (Fig. 6), the sequence of which was based on PCR-mediated random site selection (31). C/EBP␤containing complexes bound to labeled JC1 oligonucleotides and required ϳ28-fold molar excess unlabeled canonical C/EBP oligonucleotide for complete competition (Fig. 6A, lane  10). FLAG-LIP-containing complexes also bound to labeled JC1 oligonucleotides but required only 10-fold molar excess unlabeled canonical C/EBP oligonucleotide for complete competition (Fig. 6B, lane 11). Quantitation of the band intensities in the EMSAs was carried out and plotted with unlabeled oligonucleotide concentration, expressed as Ϫlog[nM], against percentage of protein complex bound to labeled oligonucleotide (Fig. 6C). Based on the calculations described under  1, 3, and 5) or with anti-FLAG antibody (lanes 2, 4, and 6). B, nuclear extract was incubated with JC1 or JC1mC/EBP 32 P-radiolabeled probes alone (lanes 1 and 4), competed with excess unlabeled canonical (can) C/EBP (lanes 2 and 5), or mutated canonical C/EBP probes (lanes 3 and 6). C, nuclear extract was incubated with canonical C/EBP, DS1, or DS1mC/EBP 32 P-radiolabeled probes alone (lanes 1, 3, and 5) or with anti-FLAG antibody (lanes 2, 4, and 6). D, nuclear extract was incubated with DS1 or DS1mC/EBP 32 P-radiolabeled probes alone (lanes 1 and 4), competed with excess unlabeled canonical C/EBP probes (lanes 2 and 5), or mutated canonical C/EBP probes (lanes 3 and 6). Unlabeled canonical and mutated canonical C/EBP probes were in 1000-or 500-fold molar excess of labeled JC1 and DS1 probes, respectively. These concentrations were chosen, as they were in excess of that necessary for complete competition of FLAG-LIP containing complex bound to labeled probes as shown in Fig. 6. Free probe was run off the gel to resolve bands. E, LIP binding to the JC1 and DS1 C/EBP sites was assessed by ChIP assays. HEK-293T cells were transfected with mLTRII (Ϫ117 to ϩ154 bp; wild type), JC1mC/E, DS1mC/E, or JC1/Ds1mC/E constructs and co-transfected with FLAG-LIP expression vector. Standard ChIP-qPCR was carried out on these cells as described under "Experimental Procedures." Data are presented as percentage of LIP binding to mutant constructs compared with mLTRII (wild type) construct. qPCR signals for no antibody and mouse IgG controls were less than 0.1% of the input. Results are expressed as the mean Ϯ S. E. (n ϭ 3); p values were calculated by Student's two-sided paired t test.
"Experimental Procedures," the canonical C/EBP site has ϳ3.2fold higher binding affinity for FLAG-LIP than C/EBP␤-containing complexes as the K D values of the FLAG-LIP and C/EBP␤ for the canonical C/EBP site were calculated to be 1.30 ϫ 10 Ϫ7 and 4.12 ϫ 10 Ϫ7 M, respectively (Table 1). Although the absolute K D values differ, our results are similar to previous reports by Descombes and Schibler (14) where bacterially expressed recombinant LIP was demonstrated to bind the C/EBP site from the albumin promoter element D and had a 2.5-fold higher affinity than recombinant C/EBP␤ (LAP) for this site. It must be noted that the small difference in our results compared with those of Descombes and Schibler (14) may be due to the fact that we used the consensus C/EBP site identified by Osada et al. (31), which is slightly different in sequence from the C/EBP site in the albumin promoter element D and also because we used nuclear extracts from HEK-293T cells expressing these proteins.
We used this validated technique for competition EMSAs using C/EBP␤ containing nuclear extract from HEK-293T cells incubated with either labeled JC1 or DS1 C/EBP oligonucleotides and increasing concentrations of unlabeled DS1 or JC1 C/EBP oligonucleotides, respectively (Fig. 7, A and  B). The C/EBP␤-containing complexes bound to the labeled JC1 C/EBP oligonucleotide required 1000-fold molar excess of unlabeled DS1 C/EBP oligonucleotide for complete competition (Fig. 7A, lane  11), whereas complete competition of C/EBP␤-containing complexes bound to labeled DS1 C/EBP oligonucleotides required only 400-fold molar excess unlabeled JC1 C/EBP oligonucleotide (Fig. 7B, lane 9). Quantitation of the band intensities in the EMSAs were carried out and plotted with unlabeled oligonucleotide concentration, expressed as Ϫlog[nM], against the percentage of protein complex bound to labeled oligonucleotide (Fig. 7C). Based on the calculations described under "Experimental Procedures," the K D values of C/EBP␤ were determined to be 2.38 ϫ 10 Ϫ6 and 4.97 ϫ 10 Ϫ6 M for JC1 and DS1 C/EBP sites, respectively (Table. 1). Therefore, C/EBP␤ FIGURE 6. Dominant-negative isoform, LIP, binds canonical C/EBP site with 3-fold higher affinity than C/EBP␤. A, competition EMSA was conducted using nuclear extracts from C/EBP␤ expressing HEK-293T cells, incubated with 5 ϫ 10 4 cpm JC1 C/EBP 32 P-radiolabeled oligonucleotide and 0 -35-fold molar excess unlabeled canonical (can) C/EBP oligonucleotide. B, competition EMSA was conducted using nuclear extract from FLAG-LIP expressing HEK-293T and incubated with 5 ϫ 10 4 cpm JC1 C/EBP 32 P-radiolabeled oligonucleotide and 0 -15-fold molar excess unlabeled canonical C/EBP oligonucleotide. Representative EMSAs are shown. Free probe was run off the gel to resolve bands. C, densitometric analysis of the C/EBP␤-containing complexes in the EMSAs was carried out using Kodak MI software. The intensity of the C/EBP␤-containing complex incubated with labeled oligonucleotide alone was set to 100%, and the intensity of complexes in all other lanes was normalized to this. Unlabeled canonical C/EBP oligonucleotide concentration, expressed as Ϫlog[nM], was plotted against percentage (%) of protein bound to labeled oligonucleotide (JC1 C/EBP 32 P). A nonlinear regression, sigmoidal dose-response curve was fit to the data to determine EC 50 . Reaction 1 described under "Experimental Procedures" was used to calculate K D . K D values from the graph were calculated to be 4.12 ϫ 10 Ϫ7 M for C/EBP␤and 1.30 ϫ 10 Ϫ7 M for dominant-negative FLAG-LIP-containing complex for the canonical C/EBP site. has approximately 2-fold higher affinity for the JC1 C/EBP site than the DS1 C/EBP site.
Dominant-negative Isoform, LIP, Binds with Equal Affinity to JC1 and DS1 C/EBP Sites but Has Higher Affinity than C/EBP␤ for These Sites-Virus replication was significantly suppressed by IFN␤ treatment despite mutations in the JC1 and DS1 sites (Fig. 4B). Additionally, the ChIP data for LIP binding suggested that LIP binds these sites (albeit at 45% lower levels) even in the presence of mutations in contrast to the binding of C/EBP␤ isoform being significantly hindered by JC1 and/or DS1 mutations. Therefore, we examined whether the affinity of LIP to the JC1 and DS1 sites varied and how it compared with the affinity of C/EBP␤ to these sites.
Labeled JC1 or DS1 oligonucleotides were incubated with FLAG-LIP-containing nuclear extracts prepared from HEK-293T cells and increasing concentrations of unlabeled DS1 or JC1 oligonucleotides, respectively (Fig. 8, A and B). Complete competition of FLAG-LIP-containing complexes bound either to JC1 or DS1 C/EBP oligonucleotide required 400-fold molar excess unlabeled DS1 or JC1 oligonucleotide, respectively (Fig. 8, A and B, lane 9). Quantitation of the band intensities in the EMSAs were carried out and plotted with unlabeled oligonucleotide concentration, expressed as Ϫlog[nM], against percentage of protein complex bound to labeled oligonucleotide (Fig. 8C). The K D value of FLAG-LIP was 1.49 ϫ 10 Ϫ6 and 1.64 ϫ 10 Ϫ6 M for JC1 and DS1 C/EBP sites, respectively (Table. 1), demonstrating that both C/EBP sites have similar affinities for FLAG-LIP. However, it was observed that LIP has a higher affinity for the JC1 and DS1 C/EBP sites compared with the C/EBP␤ isoform by 1.6-and 3-fold, respectively. These results are consistent with our observations of IFN␤ mediating suppression of virus replication despite the presence of mutations, possibly due to the higher affinity of LIP for these sites. However, further investigation is necessary to confirm the role of the increased affinity of LIP and also whether other multiprotein complexes containing LIP are present at these sites unimpeded by the C/EBP site mutation. Nevertheless, these results are consistent with the previously published observation where ratios of LIP:C/EBP␤ greater than 0.2 are sufficient to suppress C/EBP␤-mediated transactivation when LIP affinities for C/EBP sites exceed that of C/EBP␤ (14) and our previous studies, in vitro and in vivo, where LIP:C/EBP␤ ratios being greater than 0.2 correlate with SIV LTR suppression (4,14).

DISCUSSION
HIV and SIV infection and disease progression are the result of a complex interplay between viral and host cell factors. One such interplay involves specific transcription factor-binding sites found in the LTRs of HIV and SIV, to which cellular transcription factors such as Sp1, NF-B, and C/EBP␤ have been demonstrated to bind and regulate viral gene expression (48 -51). These interactions are cell type-specific. C/EBP sites in the HIV-1 LTR have been demonstrated to be crucial for LTR regulation in monocyte/macrophages but not in lymphocytes (20). In this study we characterized regulatory functions of SIV LTR C/EBP sites in primary macrophages, the major cells associated with productive HIV/SIV replication in brain, lung, and spleen, and the major reservoirs of long term viral persistence in tissues, particularly the brain (52,53). We demonstrate that the two C/EBP sites have different functions in basal transcription and virus replication; however, both sites play a role in the IFN␤-induced down-regulation of basal SIV LTR activity and virus replication in macrophages. Furthermore, we provide specific demonstration of C/EBP␤ and LIP binding to the JC1 and DS1 sites by ChIP and show that the two proteins have different affinities for the two sites consistent with their observed function in the context of both LTR luciferase assays as well as in virus replication.
Previous studies of the HIV-LTR demonstrated that C/EBP sites (C/EBP site I and II) function independently and that only one C/EBP site is necessary for basal LTR activation and virus replication (18,19). In contrast, we found that the SIV JC1 and DS1 C/EBP sites have different regulatory functions in basal LTR activation and productive virus replication; the JC1 site is crucial for basal LTR activation, whereas the DS1 site is dispensable. Additionally, we found that C/EBP␤ binds with higher affinity to JC1 than DS1 site, possibly contributing to the crucial function of the JC1 site in positively regulating basal LTR activity. An intact DS1 site, however, is imperative for productive virus replication, whereas the JC1 site does not seem to play as significant a role after the initial rounds of transcription and infection. The dramatic impairment in productive virus replication upon mutating the DS1 site in the single (DS1mC/E) and double (JC1/DS1mC/E) mutants was an unexpected result. Even though the DS1 site is 9 bp downstream of the TAR element (ϩ1 to ϩ125 bp) and several base pair downstream of the critical 5Ј stem-loop 1 necessary for Tat binding (54), we verified that these mutations did not alter the predicted secondary RNA structure formed by TAR using in silico-assisted RNA secondary structure prediction program Mfold. It is tempting to speculate that the DS1 site may be involved in mediating Tat transactivation and productive virus replication perhaps by tethering and stabilizing Tat binding to the TAR element. Tat has been shown to increase expression of the C/EBP␤ protein and binding of C/EBP␤ to DNA in vitro (55), in addition to physically and functionally interacting (in vitro and in vivo) with C/EBP␤ to induce interleukin 6 and MCP-1 (monocyte chemoattractant protein 1 or CCL2 (55)(56)(57)(58)). Moreover, Mameli et al. (58) demonstrated that the Tat-C/EBP␤ functional and physical association is mediated through cyclin T1/cdk9 in U-87MG astrocytic cells and possibly influences HIV-1 transcription. Thus, it will be interesting to characterize and identify the specific HIV C/EBP sites involved in facilitating the functional outcome of this Tat-C/EBP interaction, especially taking into consideration our findings which indicate that SIV C/EBP sites are not equivalent in function, suggesting the following: 1) they may play differential roles in mediating Tat transactivation, and 2) the proximity of a functional C/EBP-binding site to TAR is important for virus replication.
We have previously demonstrated that acute SIV replication is suppressed in the brain without the decrease in viral DNA levels, supporting the conclusion that transcriptional suppression occurs in the brain (3,4). The innate immune response to virus infection, particularly IFN␤ and the induction of LIP mediated by CUGBP1, is crucial for suppression of SIV transcription (4, 47), thereby providing a mechanism for HIV/SIV latency in macrophages in the central nervous system. Although we have previously demonstrated that knockdown of CUGBP1 leads to inhibition of LIP expression and SIV replication, the specific contributions of the JC1 and DS1 C/EBP sites in LIPmediated suppression of basal transcription and virus replication remained unknown. Thus, in this study we have demonstrated that either the JC1 or the DS1 site can independently function to downregulate basal LTR activity, although at least one intact C/EBP site is necessary. In contrast, we found that in the context of the infectious virus and virus replication, IFN␤ significantly suppressed virus replication despite mutations in the JC1 and/or DS1 sites. These results, combined with studies demonstrating that LIP has a higher binding affinity for C/EBP sites, prompted examination of the ability of LIP to bind the mutated C/EBP sites. Indeed, ChIP assays confirmed that LIP continued to bind the mutated JC1 and DS1 sites in SIV LTR, including the mutant JC1/DS1 C/EBP construct compared with wild type. Additionally, albeit speculative, the binding affinity of LIP to the C/EBP sites suggests that the equal affinity of LIP for the JC1 and DS1 sites allows it to bind to these sites either directly or indirectly through its association with other proteins in a multiprotein complex that might also bind to FIGURE 7. C/EBP␤ has 2-fold higher affinity for JC1 C/EBP site than the DS1 C/EBP site. Competition EMSA was conducted using nuclear extracts from HEK-293T cells transfected with C/EBP␤ (LAP) expression vector in all lanes. A, nuclear extract was incubated with 5 ϫ 10 4 cpm JC1 C/EBP 32 P-radiolabeled oligonucleotide and 0 -1000-fold molar excess unlabeled DS1 C/EBP oligonucleotide (lanes 1-11). Addition of anti-C/EBP␤ antibody led to supershift of C/EBP␤-containing complexes (lane 12). B, nuclear extract was incubated with 5 ϫ 10 4 cpm DS1 C/EBP 32 P-radiolabeled oligonucleotide and 0 -1000-fold molar excess unlabeled JC1 C/EBP oligonucleotide (lanes 1-11). The complex was supershifted with the addition of anti-C/EBP␤ antibody (lane 12). Free probe was run off the gel to resolve bands. C, densitometric measurements and K D calculations of C/EBP␤-containing complexes (indicated by arrows) in the EMSAs was carried out as described under "Experimental Procedures." Unlabeled oligonucleotide concentration, expressed as Ϫlog[nM], was plotted against percentage (%) of protein bound to labeled oligonucleotide (JC1 or DS1 C/EBP). A nonlinear regression, sigmoidal dose-response curve was fit to the data and used to determine EC 50 . C/EBP␤-containing complex has a K D of 4.97 ϫ 10 Ϫ6 M for the DS1 C/EBP site and 2.38 ϫ 10 Ϫ6 M the JC1 C/EBP site. K D calculations are represented as mean Ϯ S.D. in Table  1. Shown are representative EMSAs (n ϭ 2). nearby sites without being significantly impeded by the C/EBP site mutation. Finally, in lieu of the abundant literature exploring other IFN␤-mediated antiviral mechanisms, we cannot ignore the potential contributions of APOBEC3G and the recently identified TRIM22 activities (59 -62), including the expression of myxovirus resistance GTPase, the oligoadenylate synthetase/RNaseLpathway,andtheRNAdependent protein kinase pathway, further underscoring the potency of this antiviral mechanism (63).
The higher binding affinity of LIP for the JC1 and DS1 sites compared with C/EBP␤ (1.6-and 3.0-fold, respectively) provides a mechanistic link between IFN␤-mediated suppression of SIV transcription in macrophages and the establishment of SIV latency in brain (3)(4)(5)47). We postulate that having two functional C/EBP sites that mediate transcriptional suppression (as compared with one for basal transcription) accelerates the transition from active to suppressed transcription. Thus, IFN␤-mediated suppression of SIV LTR activity may require not only substoichiometric protein levels of LIP relative to C/EBP␤ (4, 5) but also higher affinities of LIP relative to C/EBP␤ for JC1 and DS1.
Our studies of the relative affinities of C/EBP␤ and LIP, derived from nuclear extracts, for the C/EBP sites corroborate previous observations that used bacterially expressed recombinant proteins binding to the canonical C/EBP site (14). The higher affinity of LIP relative to C/EBP␤ was found not only for binding to the canonical C/EBP site but also to the JC1 and DS1 sites. Interestingly, the affinities of LIP binding to JC1 and DS1 were equivalent, although the affinity of C/EBP␤ binding differed between the sites. The SIV LTR has multiple C/EBP sites like several other viral (Rous and avian sarcoma virus-LTR) and cellular promoters (albumin gene and interleukin 5 promoter (64 -68)), and the redundancy in C/EBP sites may be important not only for maintaining active virus transcription despite random mutations in HIV/SIV viral genomes but also for efficient regulation of LTR activity in multiple cellular environments.
We propose the following molecular model for the regulation of acute SIV replication in the brain based on this study and our previous reports (3)(4)(5)47). SIV enters the brain during acute infection in infected lymphocytes and monocytes, repli-FIGURE 8. Dominant-negative isoform, LIP, has similar affinity for both the JC1 C/EBP site and DS1 C/EBP sites. Competition EMSA was conducted using nuclear extracts from HEK-293T cells overexpressing FLAG-LIP in all lanes. A, nuclear extract was incubated with 5 ϫ 10 4 cpm JC1 C/EBP 32 P-radiolabeled oligonucleotide and 0 -1000-fold molar excess unlabeled DS1 C/EBP oligonucleotide (lanes 1-11). Addition of anti-FLAG antibody led to supershift of complex (lane 12). B, nuclear extract was incubated with 5 ϫ 10 4 cpm DS1 C/EBP 32 Pradiolabeled oligonucleotide and 0 -1000-fold molar excess unlabeled JC1 C/EBP oligonucleotide (lanes 1-11). The complex was supershifted with the addition of anti-FLAG antibody (lane 12). Free probe was run off the gel to resolve bands. C, densitometric analysis and K D calculations of the dominant-negative C/EBP␤ isoform (FLAG-LIP)-containing complexes in the EMSAs were carried out as described previously. Unlabeled oligonucleotide concentration, expressed as Ϫlog[nM], was plotted against percentage (%) of protein bound to labeled oligonucleotide (JC1 or DS1 C/EBP). A nonlinear regression, sigmoidal dose-response curve was fit to the data and used to determine EC 50 . FLAG-LIP-containing complex has a K D of 1.49 ϫ 10 Ϫ6 M for the DS1 C/EBP site and 1.64 ϫ 10 Ϫ6 M for the JC1 C/EBP site. K D calculations are represented as mean Ϯ S.D. in Table 1. Shown are representative EMSA (n ϭ 2). cates in perivascular macrophages, and spreads to resident macrophages (1,2,8,69). At this early time point there is an abundance of the C/EBP␤ isoform in macrophages (5, 69) that bind the JC1 and DS1 sites of the SIV-LTR. C/EBP␤ recruits histone acetyltransferases such as CBP/p300 and PCAF (15,16,70), leading to chromatin remodeling events. Such events, as demonstrated previously in vitro and in vivo in our SIV/macaque model (4), mediate transcriptional activation of the LTR, most likely through the JC1 site. Once Tat is expressed, it binds to the TAR element, assisted by C/EBP␤ bound to the DS1 site and thus contributes, in part, to the surge of SIV RNA production and virus replication during acute infection (peaking at 10 days p.i.) in the brain (3,69).
Virus replication in macrophages in the brain triggers the innate immune responses and production of IFN␤ (8,69), which activates CUGBP-1 required for IFN␤-induced translation of the dominant-negative LIP isoform (47). Between 10 and 21 days p.i., LIP expression predominates in the brain (5), which effectively competes with the C/EBP␤ for occupancy of JC1 and DS1 sites due to the higher affinity of LIP compared with C/EBP␤ for both sites. LIP lacks the transactivating domain (71) and does not interact with histone acetyltransferases leading to repression of LTR activity (4,15). As a result, chromatin remodeling events subside, evidenced by the decrease in acetylated histone H4 both in vitro and in vivo (4), and expression of SIV RNA decreases (3). This mechanism accounts for the suppression of acute SIV replication observed by 21 days p.i. (3) and the establishment of transcriptional latency in macrophages in brain (4).
It also appears that SIV and likely HIV have evolved to utilize the classical host type I IFN response providing a mechanism for the establishment of transcriptional latency and hence the persistent viral reservoirs in macrophages. Latent SIV/HIV reservoirs in the brain pose a particular challenge to therapy because many antiretroviral agents fail to effectively penetrate the central nervous system (72)(73)(74)(75) and are thus of important consideration with regard to therapeutic strategies to eradicate HIV infection.