NF-κB and p53 Are the Dominant Apoptosis-inducing Transcription Factors Elicited by the HIV-1 Envelope

Cells and Culture Conditions for Apoptosis Induction.

HeLa cells stably transfected with the Env gene of HIV-1 LAI (HeLa Env) and HeLa cells transfected with CD4 (HeLa CD4) were cocultured at a 1:1 ratio (6) in DMEM supplemented with 10% FCS, l-glutamine, in the absence or presence of 1 μM roscovitin, 1 μM rapamycin (Sigma-Aldrich), 10 μM cyclic pifithrin-α, 5 μM MG132 (Calbiochem), the NF-κB inhibitory peptide SN50, or its negative control SN50M (20 μM; BIOMOL Research Laboratories, Inc.). HeLa/U937 cell cocultures were performed at different ratios (6). U937 or Jurkat cells were cultured with 500 ng/ml recombinant gp120 protein.

Plasmids, Transfection, and Transcription Factor Profiling.

For transcription factor profiling, HeLa CD4 cells were transfected with different luciferase constructs (Mercury™ pathway profiling system from CLONTECH Laboratories, Inc.; 2 μg DNA) using Lipofectamine 2000™ (2 μl; Invitrogen) 24 h before fusion with HeLa Env cells (25 × 10⁴ in 2 ml), and the luciferase activity was measured 24 h later using the luciferase reporter assay kit from CLONTECH Laboratories, Inc. Transfection with pcDNA3.1 vector only, WT, dominant-negative (DN) Cdk1 mutant (14), the IκB super-repressor (IKSR; a gift from L. Schmitz, University of Bern, Bern, Switzerland), p53 DN plasmids (p53H273; a gift from T. Soussi, Institut Curie, Paris, France), or a p53-responsive enhanced green fluorescent protein (GFP) plasmid (a gift from K. Wiman, Karolinska Hospital, Stockholm, Sweden) was performed 24 h before coculture of HeLa CD4 and HeLa Env cells. The frequency of GFP-expressing cells was assessed by cytofluorometry on a FACSVantage™ (Becton Dickinson) equipped with a 100-μm nozzle, allowing for the analysis of relatively large cells. Afterwards, cells were counterstained with 10 μg/ml Hoechst 33324 (15 min), and the FACS^® gates were set on syncytia (i.e., cells with a DNA content >4 n) (11).

Immunofluorescence and Cytofluorometry.

Rabbit antisera specific for p53S15P, p53S46P (Cell Signaling Technology), or the NH₂ terminus of Bax (N20; Santa Cruz Biotechnology, Inc.) were used on paraformaldehyde (4% wt:vol) and picric acid–fixed (0.19% vol:vol) cells (15) and revealed with a goat anti–rabbit IgG conjugated to Alexa 568 (red) or Alexa 488 (green) fluorochromes obtained from Molecular Probes. Cells were also stained for the detection of the NH₂ terminus of Bak (Ab1; Oncogene Research Products), p53S15P (mAb), cyclin B1 (mAb; BD Transduction Laboratories), and IκBαS32/36P (mAb; Cell Signaling Technology) revealed by anti–mouse IgG Alexa conjugates. Cells were counterstained with Hoechst 33324, which allows discernment of karyogamy and apoptotic chromatin condensation (11).

Immunoblots.

Samples were prepared from HeLa Env and HeLa CD4 single cells mixed at a 1:1 ratio in lysis buffer (single cells control) or from HeLa Env/CD4 syncytia. Protein samples from HeLa Env and U937 mixed at different ratios were also prepared and subjected to immunoblot analysis. Aliquots of protein extracts (40 μg) were subjected to immunoblots using antibodies specific for IκBαS32/36P, p53S15P, p53S46P, p53 (Cell Signaling Technology), Puma (rabbit anti–human Puma antisera; United States Biological or Orbigen; and mouse anti–human Puma antibody; Upstate Biotechnology; all gave similar results), β-tubulin (mAb; Sigma-Aldrich), or glyceraldehyde-3-phosphate dehydrogenase (GAPDH, mAb; Chemicon).

Microarrays, Macroarrays, and Quantitative RT-PCR.

HeLa CD4 and HeLa Env cells were labeled with CellTracker™ red or CellTracker™ green (15 μM, 30 min at 37°C), respectively, and washed extensively before coculture. After 18 or 36 h of coculture, in the presence or absence of 10 μM cyclic pifithrin-α (readded once upon 24 h), the cells were subjected to FACS^® purification of syncytia (which are double positive) or single cells, as described previously (11). mRNA preparations were obtained with the RNeasy Mini kit (QIAGEN), quality-controlled with a bioanalyzer (model 2100; Agilent Technologies), alternatively labeled with Cy3/Cy5 fluorochromes, mixed for pairwise comparisons, and hybridized to 19-k human arrays. Arrays were scanned with a confocal laser scanner (ScanArray model 4000; PerkinElmer), quantified with software (QuantArray; PerkinElmer) at the Ontario Cancer Institute Microarray Center (http://www.microarrays.ca), and analyzed with GeneTraffic software (Iobion Informatics). All experiments were performed in triplicates, while switching the Cy3/Cy5 labeling (six data points for each pair-wise comparison) and up-regulations by 50% or down-regulations by 30% were listed. Data showing the transcriptional modification of genes with unknown function are listed in Table SI (available at http://www.jem.org/cgi/content/full/jem.20031216/DC1). Macroarrays were performed using a TranSignal™ p53 target gene array (Panomics) and quantified using the National Institutes of Health image software. For quantitative RT-PCR, cDNAs were synthesized from 1 μg of total RNA with reverse transcriptase MuLV (Roche). The TaqMan universal PCR was performed on an ABI PRISM^® 7000 Sequence Detection System (Applied Biosystems) following the manufacturer's instructions. The following primer sequences were used: bcl-2, forward, 5′-CATGTGTGTGGAGAGCGTCAA-3′, reverse, 5′-CAGGTGTGCAGGTGCCG-3′; bax α, forward, 5′-CCAAGGTGCCGGAACTGAT-3′, reverse, 5′-AAGTAGGAGAGGAGGCCGTCC-3′; bax β, forward, 5′-CCAAGGTGCCGGAACTGAT-3′, reverse, 5′-GAGGAGGCTTGAGGAGTCTCA-3′; bak, forward, 5′-ACCGACGCKATGACTCAGAGTTC-3′, reverse, 5′-ACACGGCACCAATTGATG-3′; and puma (predeveloped Taqman Assays Reagents; www.appliedbiosystems.com).

Antisense Constructs and RNA Interference.

HPLC-purified antisense phosphorothioate oligonucleotides (antisense Puma, 5′-CATCCTTCTCATACTTTC-3′; and control, CTTTCATACTCTTCCTAC-3′; GeneSet Oligos) (100 nM) were transfected with Oligofectamine (Invitrogen) into HeLa CD4 cells 24 h before coculture with HeLa Env cells. RNA interference of Bax and Bak expression was obtained by means of double-stranded morpholino oligonucleotides (GeneSet Oligos) specific for Bax (sense strand, 5′-rCUrgrgrArCrArgUrArArCrAUrgrgrArgrCTT-3′) or Bak (sense strand, 5′-UrgrgUrCrCrCrAUrCrCUrgrArArCrgUrgTT-3′), transfected into both HeLa CD4 and HeLa Env cells 48 h before coculture.

Patients' Samples and In Vitro Infection with HIV-1.

Axillary lymph node biopsies or peripheral blood samples were obtained from healthy and HIV-1–infected individuals (all males, mean age = 36 yr), who were naive for highly active antiretroviral therapy (HAART) with a plasma viral load >20,000 copies/ml or were receiving HAART with a viral load <5,000 copies/ml. Plasma HIV-1 RNA levels were determined by the bDNA procedure (Versant HIV RNA 3.0) according to the manufacturer's instructions (Bayer). None among the HIV-1⁺ patients received interferons, glucocorticoids, was positive for hepatitis B or C, or exhibited signs of autoimmunity. PBMCs were isolated by Ficoll/Hypaque (Amersham Biosciences) centrifugation of heparinized blood from either healthy donors and HIV-seropositive individuals and fixed with 4% paraformaldehyde in PBS, pH 7.2. Biopsies were fixed with 10% formalin, dehydrated, and paraffin embedded. Deparaffinized tissue sections and PBMCs were subjected to immunocytochemistry with antibodies specific for p53S46P, IκBαS32/36P, or PUMA and counterstained with hematoxylin. PBMCs were treated with TRIzol (Invitrogen) to extract total proteins, followed by immunoblot analysis. Alternatively, freshly purified PBMCs were stained with anti-CD4 antibody (FITC-labeled OKT3) and FACS^® sorted, followed by protein extraction. Primary CD4⁺ lymphoblasts (10 × 10⁶) cells (7) were incubated with the HIV-1^LAI/IIIB strain or different clinical HIV-1 isolates (500 ng of p24; reference 16) for 4 h at 37°C. After washing out unabsorbed virus, cells (10⁶ cells/ml) were cultured in RPMI 1640 medium containing 20% FCS and 10 U/ml IL-2.

Online Supplemental Material.

The transcripts with unknown function that are modulated by Env-dependent syncytium formation, as compared with control HeLa cells, are listed in Table SI available at http://www.jem.org/cgi/content/full/jem.20031216/DC1.

Transcription Factor Profiling of HIV-1 Env-elicited Apoptosis.

In a model culture system of syncytium-dependent cell death, HeLa cells stably transfected with the Env gene from HIV-1_LAI/IIIb (HeLa Env) were fused by coculture with CD4/CXCR4-expressing HeLa cells (HeLa CD4; references 6, 7, 17, 18). Syncytium-induced transcriptional effects were monitored by transfecting each of the two cell lines with a series of luciferase reporter gene constructs containing promoter elements responsive to a series of different transcription factors. This approach revealed that, among a panel of 12 different transcription factors, only NF-κB, p53, and AP1 were activated by the Env-dependent syncytium formation (Fig. 1 A). To determine the hierarchy among these transcription factors, cells were cotransfected with the aforementioned luciferase reporter constructs as well as a nonphosphorylable mutant of IκB (IKSR; reference 19) or with a DN mutant of p53 (p53H273; reference 20). Both IKSR and p53H2731 abrogated the increase of NF-κB, p53, and AP1-dependent transcription found in syncytia (Fig. 1 B). These data point to an obligate cooperation between NF-κB and p53, upstream of AP1, in Env-elicited syncytia.

NF-κB Is Required for Env-elicited Mitosis and p53 Activation.

NF-κB is usually considered as an antiapoptotic transcription factor (21, 22). Intrigued by its putative implication in syncytial apoptosis, we further investigated the possible implication of NF-κB in syncytial apoptosis. Cytoplasmic fusion is followed by nuclear fusion (karyogamy) accompanying an aberrant, cyclin B1-dependent entry into mitosis (11, 18). This allows for the distinction of early (prekaryogamic) and late (karyogamic) syncytia. A significant fraction of prekaryogamic syncytia exhibited the phosphorylation of IκB on serine residues 32 and 36 (IκBS32/36P), as detectable with a phosphoneoepitope-specific antibody. IκBS32/36P was found in the cytoplasm of prekaryogamic syncytia. In karyogamic syncytia, IκBS32/36P was confined to the nucleus (Fig. 2 A). p53 was found to exhibit an activation-associated phosphorylation pattern, both on serine 15 (Fig. 2 B, p53S15P) and serine 46 (Fig. 2 C, p53S46P), only in karyogamic syncytia. Immunoblot analysis confirmed the phosphorylation of IκB and p53 (Fig. 2 D). Kinetic analysis confirmed a precise order of phosphorylation events (IκBS32/36P → p53S15P → p53S46P) (Fig. 2 E).

Transfection of HeLa CD4 and HeLa Env cells with a p53-inducible GFP construct allowed for the quantitative assessment of p53-dependent transcriptional effects within syncytia. p53-dependent transcription was strongly inhibited by transfection with IKSR (Fig. 3 A), as well as by pharmacological inhibition of the NF-κB translocation with SN50 (Fig. 3 B). Simultaneously, both IKSR (Fig. 3 C) and SN50 (Fig. 3 D) inhibited the emergence of karyogamic syncytia, suggesting that NF-κB is required for mitotic progression of syncytia. Indeed, NF-κB inhibition fully prevented the syncytial accumulation of cyclin B1, as determined by quantitative immunofluorescence analysis (Fig. 3 E). Inhibition of the Cdk1 by DN Cdk1 (Fig. 3 A) or roscovitine (Fig. 3 B) inhibited the p53-dependent transcription, while inhibiting karyogamy (Fig. 3, C and D). Altogether, these findings indicate that NF-κB is essential for the aberrant, cyclin B1/Cdk1-mediated entry into karyogamy that occurs upstream of p53 activation. The repression of NF-κB reduced the karyogamy-associated phosphorylation of p53 (both p53S15P and p53S46P). Moreover, transfection with DN p53 partially inhibited the phosphorylation of IκB and totally blocked its translocation into the nucleus (Fig. 3 E). These findings point again to an obligate cross-talk between p53 and NF-κB.

IκB phosphorylation was also detectable in lymph nodes from HIV-1 carriers (but not in those from uninfected controls), mainly among mitotic cells (Fig. 4 B). Moreover, the phosphorylation of p53S46P was induced by HIV-1 infection of primary CD4⁺ lymphoblasts (see below) and was detectable in lymph node biopsies from HIV-1 carriers (Fig. 4 A). Altogether, these data extend the notion of HIV-1–induced NF-κB activation and p53 activation.

HIV-1 Env-elicited Transcriptional Effects Involve p53 and Lead to Puma Overexpression.

To further explore the implication of transcription factors in Env-induced syncytial apoptosis, microarrays were performed on HeLa Env/CD4 cocultures maintained for 18 or 36 h in the presence or absence of cyclic pifithrin-α, a chemical inhibitor of p53. 40 genes of known function (Table I) and 42 genes with unknown function (Table SI) were found to be significantly (by a factor of ≥1.5 or ≤0.67) induced or inhibited in syncytia, as compared with control cells cultured separately. The chemical inhibitor of p53 transactivation, cyclic pifithrin-α, was found to prevent these changes in transcription either at 18 or 36 h for 85% among these genes, indicating that p53 is indeed the principal transcription factor, whose activation directly or indirectly (via effects on other transcription factors) determines the pathophysiology of syncytial apoptosis.

Based on the fact that the microarrays used in this study do not cover the entire transcriptome, we used a p53-specific macroarray for the identification of genes specifically associated with syncytial apoptosis. Using this technique, we found that p53 itself, c-Jun (whose gene product, together with c-Fos, forms AP1), the proapoptotic gene PIG8 (also known as ei24), the proapoptotic Bcl-2 family member Bax, as well as the proapoptotic BH3-only Bcl-2 family member Puma were up-regulated in syncytia by a factor ≥2 (Fig. 5, A and B). Quantitative RT-PCR confirmed the induction of Puma and Bax (and in particular the dominant Bax-α splice variant) transcripts (Fig. 6 A), which is consistent with the fact that syncytia overexpress Bax (7) and Puma (Fig. 6 B) at the protein level.

Up-regulation of Puma Is Required for Syncytial Apoptosis.

Puma has been identified recently as a p53-inducible “BH3-only” protein that can activate the proteins Bax and Bak to adopt a proapoptotic conformation (with exposure of the NH₂ terminus), to insert into the outer mitochondrial membrane and to stimulate the release of cytochrome c, leading to subsequent caspase activation and apoptosis (23–25). Accordingly, transfection with antisense oligonucleotides designed to down-regulate the expression of Puma (Fig. 7 B, inset; reference 24) inhibited the exposure of the NH₂-termini of Bax and Bak (as detected with conformation-specific antibodies), the release of cytochrome c from syncytial mitochondria (as detected by immunofluorescence staining), and the apoptosis-associated chromatin condensation (Fig. 7, A and B). Inhibition of NF-κB with IKSR (Fig. 7 C) or SN50, as well as inhibition of p53 with cyclic pifithrin-α (Fig. 7 D) all inhibited the activation of Bax and Bak. RNA interference of Bax or Bak, the two Puma receptors, significantly reduced syncytial apoptosis, as indicated by suppression of the nuclear chromatin condensation and the mitochondrial cytochrome c release (Fig. 7 E). As a side observation, we found that RNA interference of Bak prevented the exposure of the NH₂ terminus of Bax but not vice versa, suggesting a hitherto unsuspected hierarchy among these proapoptotic proteins. Altogether, the data suggest that Puma acts on Bax and/or Bak to kill syncytia in a p53-dependent fashion.

Induction of Puma in Several Models of HIV-1 and Env-induced Apoptosis In Vitro.

When cocultured with HeLa Env cells, U937 cells (which lack functional p53; reference 26) do not form heterokarya, but rather undergo apoptosis after a transient cell-to-cell interaction (5, 6). As shown in Fig. 8 A, apoptosis correlates with the induction of Puma, and the up-regulation of Puma is not inhibited by cyclic pifithrin-α, underscoring that Puma can be induced in a p53-independent fashion. Similarly, different recombinant gp120 variants (MN and 93TH975 with preference for CXCR4; BaL and SF162 with preference for CCR5) induced cell death and Puma protein expression in U937 cells (which express both CXCR4 and CCR5), correlating with the induction of PUMA (Fig. 8 B). Transfection of U937 cells with p53 accelerated the kinetics of gp120-induced Puma induction at early time points when gp120 alone still had no Puma-inducing effect (Fig. 8 C). Thus, p53-independent and -dependent pathways can cooperate to cause Puma induction. In p53-sufficient Jurkat cells (27), which only express CXCR4, only the R4-tropic gp120 variants induced apoptosis and Puma expression (Fig. 8 D). This effect was strongly inhibited by cyclic pifithrin-α (Fig. 8 E). Primary CD4⁺ lymphoblasts from healthy donors infected with HIV-1_LAI/IIIb also manifested the induction of Puma, at the protein level, well after the phosphorylation of p53 (Fig. 8 F). CD4⁺ lymphoblasts infected with clinical HIV-1 isolates (16) manifested the induction of Puma, which could be detected in yet viable syncytia (Fig. 8 G). Thus, Env and HIV-1 infection induce Puma expression in a variety of experimental systems.

Enhanced Puma Expression in HIV-1–infected Patients.

Lymph nodes from untreated HIV-1 patients with high HIV-1 titers (>20,000 copies/ml) stained more positively for Puma than did biopsies from uninfected controls (Fig. 9 A). Using immunocytochemical methods, Puma⁺ cells were found among all relevant (CD4⁺, CD8⁺, and CD14⁺) cells from HIV-1–infected donors with such high HIV-1 titers, with a preference for CD4⁺ and CD14⁺ cells (Fig. 9 B). Puma expression levels among the PBMCs correlated positively with HIV-1 titers (Fig. 9 C). The enhanced Puma expression could also be detected by immunoblotting among circulating CD4⁺ cells from naive (nontreated) HIV-1 patients as compared with healthy controls or HAART-treated individuals (Fig. 9 D). Moreover, longitudinal studies revealed that Puma levels were reduced to normal levels when patients were undergoing successful antiretroviral therapy (Fig. 9 C), in line with the well-established fact that antiretroviral therapy reduces the increased spontaneous apoptosis of circulating CD4⁺ lymphocytes from HIV-1 carriers (28, 29). These data establish that productive HIV infection is associated with the overexpression of the apoptotic effector Puma.