In order to understand the molecular mechanisms regulating HIV-1 latency, we studied the modifications of cellular transcription using microarrays in the promonocytic U1 and T CD4⁺ lymphocytic ACH-2 chronically HIV-1 infected cell lines after reactivation of latency. The two cell lines were treated with 10 mM of the histone deacetylase inhibitor NaB. Viral reactivation was monitored by coculture with P4 indicating cells (Figure 1A) and measuring gag viral mRNA expression (Figure 1B). Increase in both β-galactosidase activity and gag mRNA expression showed that the viral reactivation after NaB treatment was efficient. Total RNAs were extracted after 24 h and sent to the Affymetrix Microarray Facilities for subsequent hybridization on U-133A microarrays.

The pattern of cellular mRNA from chronically infected cells treated with NaB was compared to that from non-treated cells. We used as specific criteria a log₂ ratio change ≥ 1 with a change p-value ≤ 0.0001 for increased genes and a log₂ ratio change ≤ -1 with a 1-change p-value ≥ 0.9999 for decreased genes. Hybridization experiments were performed once. We identified 740 genes that were upregulated by twofold or higher in NaB treated U1 cells and 896 genes that were downregulated, 482 genes in NaB treated ACH-2 cells that had a level increased greater than twofold and 634 genes that had a level decreased greater than twofold (data not shown). Moreover, 260 genes were commonly increased and 337 genes were decreased in both U1 and ACH-2 NaB-treated cells (data not shown). Pathways involved in regulation of transcription, signal transduction, immune response, protein transport, metabolism, apoptosis and RNAs modifications showed altered expression following treatment with NaB. Some of the genes involved in these pathways are shown in Additional Files 1, 2, 3, 4, 5 and 6. Our analysis identified genes that have previously been associated with HIV-1 replication or latency, such as CDK9 16, Jun 1623, PSMB10 27, MAPK1 26 or OAS1 30. This supported the accuracy of our approach, even though, as the hybridization experiments had been performed once, the statistical relevance of the results could not be estimated.

Among the differentially expressed genes, we chose to focus on two candidate genes encoding transcription factors: NCoA3 and IRF8 (Tables 1 and 2). We selected these two genes based on their biological properties, their described effects on viral replication 3132 and their differential expression observed by microarray experiments. Indeed, NCoA3 and IRF8 gene expression are respectively upregulated and downregulated following treatment with NaB of latently infected cells (Tables 1 and 2). Therefore, NCoA3 and IRF8 could be implicated respectively in the reactivation and maintenance of HIV-1 latency.

NCoA3 gene expression is upregulated following treatment with NaB of both U1 and ACH-2 latently infected cells (Tables 1 and 2). NCoA3 is a nuclear receptor coactivator of the Steroid Receptor Coactivator (SRC) family that interacts with nuclear receptors in a ligand-dependent manner and enhances transcriptional activation via histone acetylation and recruitment of general transcription factors and additional cofactors (reviewed in 28). NCoA3 (Unigene Hs. 382168) gene expression in U1 cells is significantly upregulated by 4.9 to 22.6 fold (U1NaBvsU1 signal log₂ ratio ranging from 2.3 to 4.5 with a change p-value < 0.00015) following treatment with NaB (Table 1). Similarly, NCoA3 gene expression is upregulated in NaB-treated compared to non-treated ACH-2 cells by 2 to 13.9 fold but with a lower significance (ACH2NaBvsACH2 signal log₂ ratio ranging from 1 to 3.8 with a change p-value < 0.0055) (Table 2).

IRF8 gene expression is downregulated following treatment of U1 cells with NaB (Table 1). IRF8 is a transcription factor of the Interferon (IFN) Regulatory Factor (IRF) family that binds to IFN-stimulated response element and regulates expression of genes stimulated by IFNs (reviewed in 29). IRF8 (Unigene Hs. 137427) is expressed in the promonocytic cell line U1 (detection signal of 707.9 with a p-value of 0.000244) (Table 1) but is not expressed in the T CD4⁺ lymphocytic cell line ACH-2 (data not shown). Following NaB treatment, IRF8 gene expression in U1 cells is downregulated by 16 fold (U1NaBvsU1 signal log₂ ratio of -4 with a 1-change p-value of 0.99998) (Table 1).

Real-time RT-PCR quantifications were performed to confirm that NCoA3 and IRF8 genes were differentially expressed in the NaB-treated chronically infected cells compared to the non-treated cells. We performed quantification on RNA samples obtained from five independent NaB treatments of U1 and ACH-2 cells and real-time RT-PCR experiments were run in duplicate. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The results show in Figure 2 represent the NCoA3 expression increase fold (Figure 2A) obtained from U1 and ACH-2 cells and the IRF8 expression decrease fold (Figure 2B) obtained from U1 cells treated with NaB for 24 h and 48 h compared to non-treated cells. Concerning NCoA3, real-time RT-PCR showed an upregulation consistent with microarray data in 24 h NaB-treated U1 cells of 8.34 ± 2.42 fold compared to non-treated cells (Figure 2A). NCoA3 gene expression is also increased with a 48 h NaB treatment (upregulation of 8.40 ± 2.33 fold) (Figure 2A). Similarly, an increase of NCoA3 gene expression can be observed on ACH-2 cells following treatment with NaB (upregulation of 4.56 ± 1.28 fold in 24 h and 6.80 ± 2.34 fold in 48 h NaB-treated ACH-2 cells) (Figure 2A). Concerning IRF8, real-time RT-PCR showed a 14.96 ± 4.85 fold decrease in 24 h NaB-treated U1 cells (Figure 2B) in correlation with the microarray ratio previously obtained. Downregulation of IRF8 gene expression is also observed following 48 h NaB-treatment of U1 cells (22.06 ± 11.29 fold decrease) (Figure 2B). Taken together, results from real-time RT-PCR performed on NCoA3 and IRF8 genes corroborate with those obtained using microarray hybridizations.

We next determined whether NCoA3 and IRF8 gene expression were similarly modified in the uninfected parental cell lines. U937 and CEM cells were subjected to identical treatment and RT-PCR quantifications were performed (Figure 3). NCoA3 is upregulated both in U937 and CEM cells following treatment with NaB (upregulation of 7.32 ± 1.74 fold in 24 h and 11.45 ± 2.95 fold in 48 h NaB-treated U937 cells, upregulation of 1.93 ± 1.04 fold in 24 h and 5.59 ± 0.06 fold in 48 h NaB-treated CEM cells) (Figure 3A). IRF8 is only expressed in the promonocytic cell line U937 and, as in U1 cells, its expression was downregulated after NaB treatment (downregulation of 17.95 ± 4.15 fold in 24 h and 22.32 ± 10.82 fold in 48 h NaB-treated U937 cells) (Figure 3B). Thus, NaB treatment modify NCoA3 and IRF8 gene expression in uninfected parental cell lines U937 and CEM at a similar level than in chronically infected cells.

We then performed additional experiments to determine whether the gene expression variations observed could also be mediated by treatments with the phorbol ester PMA and another HDAC inhibitor, TSA. We thus assessed the differential regulation of NCoA3 and IRF8 gene expression in U1 and ACH-2 cells treated with PMA or TSA (Figure 4). Results indicated that NCoA3 expression is upregulated by 24 h and 48 h PMA treatment of U1 and ACH-2 cells (upregulation of 5.70 ± 1.37 fold in 24 h and 9.85 ± 0.90 fold in 48 h PMA-treated U1 cells, upregulation of 3.12 ± 1.05 fold in 24 h and 7.12 ± 1.20 fold in 48 h PMA-treated ACH-2 cells (Figure 4A). However, TSA treatment had no significant effect on NCoA3 expression in U1 and ACH-2 cells, although TSA increased viral expression (data not shown). Concerning IRF8 expression in U1 cells, PMA and TSA treatments for 24 h induced a decrease of 3.22 ± 0.45 fold and 5.32 ± 1.09 fold, respectively (Figure 4B). These results show that NCoA3 expression is upregulated following phorbol ester but not with other HDAC inhibitor treatments in U1 and ACH-2 cells. Moreover, IRF8 gene expression in U1 cells is downregulated with PMA or TSA treatments but at a lower extent than with NaB.

We also assessed the differential regulation of NCoA3 and IRF8 gene expression in others chronically HIV-1 infected cell lines. The chronically infected promonocytic OM10.1 and T CD4⁺ lymphocytic J1.1 cell lines were treated with NaB for 24 h and 48 h and real-time RT-PCR were performed to measure NCoA3 and IRF8 gene expression. As shown in Figure 5, NCoA3 expression is upregulated by 4.94 ± 0.78 fold in OM10.1 and by 2.56 ± 0.64 fold in J1.1 after 24 h NaB treatment. NCoA3 expression increased with time of NaB treatment in both cell lines (upregulation of 12.89 ± 3.10 fold in OM10.1 and 3.51 ± 0.69 fold in J1.1 cells) (Figure 5). Like ACH-2 and unlike U1 cells, the T CD4⁺ lymphocytic J1.1 and the promonocytic OM10.1 cell lines did not express IRF8 (data not shown). Thus, the differential regulation of NCoA3 but not IRF8 gene expression is similar in two other related latently HIV-1 infected cell line models.

We performed reactivation experiments at different times, sooner than 24 h and until 48 h. Quantitative RT-PCR experiments were carried out on total RNAs. This was done using U1 cells to analyze both NCoA3 mRNA increase (Figure 6A) and IRF8 mRNA decrease (Figure 6B) relative to HIV gag mRNA along with ACH-2 cells (Figure 6C) to analyze NCoA3 mRNA increase relative to HIV gag mRNA.

As observed on Figure 6C, the obtained results, both on ACH-2 and U1 cells, clearly show that gag mRNA activation occurs after NCoA3 mRNA increase and accumulation. Moreover, in U1 cells, gag mRNA activation occurs after IRF8 mRNA decrease. Shorter kinetics (0 to 8 h) correlated with these results (data not shown).

To confirm that the changes seen at the RNA level correlated with protein levels, we performed Western blot experiments on nuclear extract of U1, ACH-2, OM10.1 and J1.1 cells treated or not with NaB for 24 h (Figure 7). Results indicated that NaB increased the expression level of NCoA3 protein in U1, ACH-2, OM10.1 and not in J1.1 cells (Figure 7). Moreover, IRF8 protein expression is strongly downregulated in U1 cells following NaB treatment (Figure 7). These results correlate with the differential expression of NCoA3 and IRF8 genes observed with both microarray and real-time RT-PCR experiments.

We analyzed the functional role of NCoA3 on viral transcription by transfection assays. HEK293 cells were cotransfected with pLTRX-luc reporter plasmid containing the luciferase gene under the control of the HIV-1 U3-R promoter region (nt -640 to +78) with or without Tat and/or NCoA3 expression vectors. As shown in Figure 8, NCoA3 increased Tat-stimulated HIV-1 LTR activity by 2.8 ± 1.4 fold. The presence of NCoA3 had synergistic effect on the HIV-1 LTR activity induced by suboptimal expression of Tat. When HEK293 cells were transfected with pLTRΔTAR-luc reporter plasmid lacking the Tat-transactivation response element TAR, Tat was not able to activate the LTR transcription, as expected, and NCoA3 had no effect on the LTR activity (Figure 8). Thus, functional analyses confirm that NCoA3 synergizes with Tat to enhance HIV-1 promoter transcription, as expected 31, and that this effect is dependent on the presence of the TAR region.

We analyzed the functional role of IRF8 on viral transcription by transfection assays. HEK293 cells were cotransfected with pISRE-TK-luc reporter plasmid corresponding to the HIV-1 IFN-stimulated response element, located downstream transcription start site (nt +194 to +223) 33, with or without IRF1 and/or IRF8 expression vectors. As shown in Figure 9, the basal activity of the ISRE-TK was increased by 7.4 ± 1.0 fold in the presence of IRF1 as expected 32, whereas a decrease was detected in the presence of IRF8 (21.9 ± 10.6 to 41.4 ± 9.5 %). The expression of IRF8 inhibited by 43.5 ± 10.6 to 74.7 ± 2.5 % the IRF1-mediated activation of the ISRE-TK in a dose dependent fashion (Figure 9). The expression of the dominant negative IRF8 DNA-binding domain (IRF8-DBD) inhibited by 76.4 ± 6.5 % the IRF1-mediated activation of the ISRE-TK, as expected 34 (Figure 9). The inhibitory effects of IRF8 and IRF8-DBD expression and activation effect of IRF1 expression was abolished when the ISRE sequence was mutated (pISREmut-TK-luc, Figure 9). These results show that IRF8 represses the ISRE-TK promoter transcription through the ISRE element from the HIV-1 promoter, as expected 32.

The chronically HIV-1 infected T CD4⁺ lymphocytic cell lines ACH-2 47 and J1.1 48 derived from CEM and Jurkat cells respectively, and the chronically HIV-1 infected promonocytic cell lines U1 49 and OM10.1 50 derived from U937 and HL-60 cells respectively, were obtained through the National Institutes of Health (NIH) AIDS Research and Reference Reagent Program. Suspension cell lines were grown in RPMI 1640 (Invitrogen) with 10% fetal bovine serum (Invitrogen), 50 U/mL penicillin, 50 μg/mL streptomycin (Invitrogen) and 2 mM glutamine (Invitrogen). Cells were treated with 10 mM of sodium butyrate (NaB; Sigma), or with 10 ng/mL of PMA (Sigma), or with 300 nM of TSA (Sigma). Cells were harvested generally 24 h and 48 h after treatment and cell viability was estimated before subsequent RNA extraction or nuclear extract preparation. P4 indicator cells are HeLa CD4⁺ cells carrying the lacZ gene under the control of the HIV-1 LTR. P4 and HEK293 cells were grown in DMEM (Invitrogen) containing 5% fetal bovine serum (Invitrogen), 50 U/mL penicillin, 50 μg/mL streptomycin (Invitrogen) and 2 mM glutamine (Invitrogen).

The pLTRX-luc construct contains the luciferase (luc) gene downstream of the HIV-1 BRU U3-R promoter region (nt -640 to +78) 51. The pLTRΔTAR-luc construct corresponds to the pLTRX-luc plasmid in which the TAR region (nt +38 to +78) was deleted 51. The pCMV-Tat expression vector was kindly provided by S. Emiliani (Institut Cochin, Paris, France). The pIRF8 expression vector (pcDNAmycHis-ICSBP) and dominant negative construct pIRF8-DBD, which contains the DNA binding domain of IRF8, were a kind gift of B.Z. Levi (Technion-Israel Institute of Technology, Haifa, Israel). The pNCoA3 expression vector (pcDNA3.1-AIB1) was a kind gift of P.S. Meltzer (NIH, Bethesda, USA) 52. The pIRF1 construct was generated by cloning the fragment excised from pHuIRF-3-1 (a kind gift of T. Taniguchi, University of Tokyo, Tokyo, Japan) by HindIII/NotI digestion in the pcDNA3.1 plasmid (Invitrogen). The pISRE-TK-luc and pISREmut-TK-luc constructs were generated by cloning a wild-type (AGGGACTT

GAAA

GAAA

GCCC

GCCC

Total RNAs were extracted using the RNeasy Mini Kit (Qiagen). The procedure included an "on-column" DNase I digestion step according to the manufacturer's instructions. RNA quality was assessed using the Agilent Bioanalyzer 2100 and spectrophotometric analysis prior to cDNA synthesis.

Microarray experiments were performed using the U133-A microarrays (Affymetrix) containing 22283 oligonucleotides spots. Total RNAs obtained from chronically infected U1 and ACH-2 stimulated or not with NaB for 24 h were sent to Dr. C. Thibault (Affymetrix Microarray Facilities, IGBMC, Strasbourg, France) for amplification, labeling and hybridization. Hybridization experiments were performed once. Results were then analyzed with the Mas5.0 Software (Affymetrix) and interpreted using the Data Mining Tool (Affymetrix) and Microsoft Excel softwares. For individual analyses, the p-value cut off was 0.048 as suggested by Affymetrix. For comparative analyses, a log₂ ratio change ≥ 1 for increased genes and ≤ -1 for decreased genes were defined. Gene expression changes were considered to be significant when the change p-value was ≤ 0.0001 for increased genes and 1-change p-value ≥ 0.9999 for decreased genes.

Quantifications of cellular RNAs were performed using a Light Cycler instrument (Roche Diagnostics). Briefly, cDNAs were synthesized from 1 μg of total RNA with MoMLV reverse transcriptase (Superscript II, Invitrogen) and 1/10^th aliquots of the corresponding samples were used for real-time PCR in a 20 μL reaction mixture containing 1X LightCycler FastStart DNA Master SYBR Green I (Roche Diagnostics), 4 mM MgCl₂, and 500 nM of each primer. The reactions were carried out in duplicate and the results were normalized to the expression of Cyclophilin A. Primers for quantitative PCR were designed using Oligo 6 software. All primer pairs produced single amplification product as determined by melting curve analyses. The sequences of the primers used were (5' to 3'): NCoA3 forward CTTTGGGCATTCCTGAACTTGTC, NCoA3 reverse GCCTCATCACCGCAGCAC, IRF8 forward GGAGTGCGGTCGCTCTGAAA, IRF8 reverse GTCGTAGGTGGTGTACCCCGTCA, Cyclophilin A forward AGTGGTTGGATGGCAAGC, Cyclophilin A reverse GATTCTAGGATACTGCGAGCAAA. PCR reactions were carried out with a denaturation step of 10 min at 95°C followed by forty-five cycles of 10 s at 95°C, 5 s at annealing temperature (55°C for NCoA3 and Cyclophilin A, 59°C for IRF8) and 20 s amplification at 72°C. Quantifications of cDNAs were determined in reference to a standard curve prepared by amplification of serial dilutions of PCR product containing matching sequences. Analyses were performed using the second-derivative-maximum method provided by the Light Cycler quantification software, version 3.5 (Roche Diagnostics).

Quantification of gag viral mRNA was performed by real-time RT-PCR as described in 54.

For nuclear extract preparation, 10.10⁶ cells were harvested, washed and nuclei were isolated by addition of 150 μL of buffer I (50 mM Tris pH 7.9, 10 mM KCl, 10% glycerol, 1 mM EDTA, 0.2% NP40) followed by a centrifugation at 3000 g for 3 min. Nuclear extracts were prepared by addition of 15 μL of buffer II (20 mM Hepes pH 7.9, 400 mM NaCl, 10 mM KCl, 20% glycerol, 1 mM EDTA) for 20 min at 4°C followed by a centrifugation at 15000 g for 10 min. Protein concentrations were determined by the Bio-Rad protein assay.

Nuclear extracts (100 μg) were loaded on 8% SDS-polyacrylamide gel and the proteins were transferred to nitrocellulose membrane (Hybond-C, Amersham) that was subsequently blocked for 1 h with 5% non-fat dry milk in PBS-T (PBS, 0.05% Tween20) and incubated with antibodies directed against NCoA3 (goat polyclonal anti-ACTR C-20, Santa Cruz Biotechnology, Inc.), IRF8 (goat polyclonal anti-ICSBP C-19, Santa Cruz Biotechnology, Inc.) or actin (mouse monoclonal anti-actin, Calbiochem) for 2 h. The membranes were then washed and incubated with secondary antibodies conjugated to horseradish peroxidase (HRP conjugated rabbit anti-goat (DakoCytomation) or goat anti-mouse (Calbiochem) immunoglobulins). Hybridizations were revealed using an ECL enhanced chemiluminescence kit (ECL, Amersham). The quantification was done using the Image Tools (Syngene) software.

HEK293 cells were transfected using calcium phosphate co-precipitation method. Cells were lysed 24 h after transfection with a buffer containing 60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 10 mM MgSO₄, 2.5 mM EDTA, 50 mM β-mercaptoethanol and 0.125% Nonidet P-40. Luciferase activities were measured as previously described 55. Cotransfection with pCMV-LacZ plasmid was performed to normalize transfection efficiency and β-galactosidase activities were determined using Chlorophenol red β-D-galactopyranoside (CPRG, Roche Diagnostics) assay as previously described 55. The normalized luciferase index (NLI) was defined as the ratio of luciferase to β-galactosidase activities.