Productive HIV-1 Infection of Primary CD4 (cid:1) T Cells Induces Mitochondrial Membrane Permeabilization Leading to a Caspase-independent Cell Death*

We have explored in vitro the mechanism by which human immunodeficiency virus, type 1 (HIV-1) induces cell death of primary CD4 (cid:1) T cells in conditions of productive infection. Although HIV-1 infection primed phy-tohemagglutinin-activated CD4 (cid:1) T cells for death induced by anti-CD95 antibody, T cell death was not prevented by a CD95-Fc decoy receptor, nor by decoy receptors of other members of the TNFR family (TNFR1/ R2, TRAILR1/R2/OPG, TRAMP) or by various blocking antibodies, suggesting that triggering of death receptors by their cognate ligands is not involved in HIV-induced CD4 T cell death. HIV-1 induced CD4 T cell shrinkage, cell surface exposure of phosphatidylserine, loss of mitochondrial membrane potential ( (cid:2)(cid:3) m), and

The depletion of CD4ϩ T cells is a major determinant of pathogenicity in human immunodeficiency virus type 1 (HIV-1) 1 infection. The finding that the level of the viral load established soon after infection correlates with the rate of CD4 T cell loss and the development of AIDS (1) supports the idea that active HIV-1 replication directly contributes to the depletion of CD4ϩ T cells. Accordingly, in vitro studies have shown that HIV-1 replication induces apoptosis in proliferating primary CD4ϩ T cells stimulated with PHA/IL-2 and in CD4ϩ T cell lines (2)(3)(4)(5)(6). Other findings, however, support the idea that death of uninfected T cells also contributes to AIDS pathogenicity. Both spontaneous and activation-induced apoptosis occur in vitro (7)(8)(9)(10) and in vivo (11)(12)(13)

in both infected and uninfected T cells from HIV-1-infected individuals.
Despite intensive investigations, several important questions remain about the mechanisms through which HIV infection induces CD4 T cell death. The first, as stated above, is whether HIV induces death in infected and/or uninfected CD4ϩ T cells. The second concerns the nature of the signal(s) that initiate cell death. T cells from HIV-1-infected individuals show enhanced cell surface expression of CD95 and exhibit increased susceptibility in vitro to CD95-mediated cell death, induced either by an agonistic anti-CD95 antibody, by soluble CD95 ligand (CD95L) (14 -19), or by the engagement of other members of the tumor necrosis factor receptor (TNFR) family, including TRAILR and TNFR1 (20,21). HIV-mediated death of productively infected CD4ϩ T cells in vitro has, however, been found to be independent of CD95/CD95L interactions (4 -6), and the possible involvement of other members of the TNF receptor family has not been explored. Several findings suggest that viral proteins encoded by HIV-1 (gp120 envelope glycoprotein, Vpr) may induce death of either infected or uninfected CD4ϩ T cells in productively infected CD4 T cell cultures (22)(23)(24)(25)(26)(27)(28).
Finally, another area of uncertainty concerns the identification of the effector pathways that lead to cell death following HIV infection. It has been reported that treatment of HIV-infected cells with caspase inhibitors prevents CD4 T cell death and results in increased viral production (4,29), whereas other studies have found that caspase inhibitors did not prevent the death of infected CD4ϩ T cells (5). Two pathways are known to be important for transducing death signals to the apoptotic machinery. The "extrinsic" pathway involves activation of death receptors and recruit procaspase-8 through FADD (30,31). Downstream of caspase-8, two pathways have been reported; caspase-8 may either directly activate caspase-3 or cleave Bid (a proapoptotic member of the Bcl-2 family), inducing the release of cytochrome c from mitochondria. Cytochrome c, together with Apaf-1, activates caspase-9, leading to the activation of the caspase-3 (32)(33)(34). The "intrinsic" pathway is death receptor-independent; stress signals activate proapoptotic members of the Bcl-2 family (Bax, Bak, etc.) and induce the permeabilization of the mitochondria and the release of apoptogenic factors (35). Although caspases are essential effectors of the nuclear apoptotic phenotype (chromatin condensation and fragmentation), evidence from experimental systems using broad spectrum caspase inhibitors support the notion that programmed cell death can proceed in a caspase-independent manner (36 -39). Moreover, recent reports have suggested that caspase inhibitors, which inhibit apoptosis induced by diverse stimuli, lead to the appearance of dead cells expressing necrotic-like phenotype (40 -42). Cellular mechanisms that account for caspase-independent programmed cell death are still elusive and may involve release by mitochondria of effectors such as apoptosis-inducing factor (AIF) (43).
To further characterize the mechanisms responsible for cell death induced by HIV-1, freshly isolated CD4ϩ T cells were infected with HIV-1 and stimulated with PHA/IL-2. We observed that only around half of the dying CD4ϩ T cells displayed a typical apoptotic phenotype (cell shrinkage, chromatin condensation, and fragmentation), the other half showing a nonapoptotic cell death phenotype (membrane permeabilization and intact nucleus) that only shared one feature with apoptosis (cell shrinkage). Treatment with zVAD-fmk, a broad caspase inhibitor, prevented the induction of an apoptotic phenotype (nuclear chromatin condensation and fragmentation) in the HIV-infected CD4ϩ T cells but did not prevent loss in mitochondrial membrane potential (⌬⌿m) and cell death. Our results support a scenario in which disruption of the mitochondria membrane permeability is a central event in cell death following HIV-1 infection.
Cells and Culture Conditions-Heparinized venous peripheral blood was obtained from HIV-seronegative healthy donors. PBMC were isolated by Ficoll-Hypaque density gradient centrifugation and cultured in RPMI 1640 (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (Summit Biotechnology, Greeley, CO), 2 mM L-glutamine, 1 mM sodium pyruvate (Invitrogen), and penicillin/streptomycin (Invitrogen). When indicated, purified CD4ϩ T cells were obtained by depleting PBMC of B cells, NK cells, and CD8ϩ T cells by negative selection, using CD19, CD56, and CD8 mAbs and magnetic beads coated with anti-mouse IgG (Dynal, Lake Success, NY). PBMC were incubated in the absence or presence of virus at the indicated multiplicity of infection (MOI) for 2 h at 37°C. After two washes, the cells were resuspended in complete medium in the absence or presence of 1 g/ml PHA-P (Sigma) and 100 IU/ml IL-2. When indicated, CD4ϩ T cells were incubated with either agonistic CD95 mAb or Fc decoy receptors. HIV p24 antigen was measured by an enzyme immunoassay as described by the manufacturer (Abbott). Intracellular p24 antigen was also assessed by flow cytometry after fixation and permeabilization of CD4ϩ T cells with Intraprep permeabilization reagent (Coulter Corp.).
Virus Preparation-High titer stocks of the laboratory strain HIV-1 LAI (10 6 TCID 50 /ml) were prepared by inoculating CEM at an MOI of 0.001 and growing the cells for 10 days. 10 ml of this culture were added to 400 ml of uninfected CEM (5 ϫ 10 5 cells/ml) and grown for 5-7 days until abundant syncytia were present. The cells were pelleted (300 ϫ g for 10 min) and resuspended in one-one hundredth of the initial volume for 8 h. The supernatant was clarified by centrifugation (800 ϫ g for 10 min).
Measurement of Cell Death-Nuclear condensation and fragmentation (typical morphological changes of apoptosis) was visualized by UV microscopy using Hoechst 33342 nuclear dye. Cells were also doublestained with propidium iodide and Hoechst 33342 to distinguish apoptotic cells from cells that lost membrane integrity. Intact blue nuclei, condensed fragmented blue-pink nuclei, and intact pink nuclei were considered viable, apoptotic (early and late), and nonapoptotic cells, respectively. Cells were also analyzed by light microscopy using trypan blue dye reagent (Sigma). Live cells showed normal refringent cytoplasm, apoptotic cells displayed typical chromatin condensation and fragmentation and excluded trypan blue, and nonapoptotic cells were trypan blue positive, as previously described (7,15,18,46). By flow cytometry, we determined dying cells using FITC-conjugated annexin-V, and to evaluate ⌬m, cells were stained with DiOC6. Nuclear condensation and fragmentation was also assessed by flow cytometry using propidium iodide.
Western Blotting-For total extracts, cells were incubated in SDS lysis buffer, boiled for 10 min, and centrifuged for 15 min at room temperature. Total protein was measured using the DC protein assay (Bio-Rad). Equal amounts of proteins were boiled for 5 min in 2ϫ Laemmli sample buffer with ␤-mercaptoethanol and run on a 4/20% polyacrylamide gel (Bio-Rad). Proteins were transferred to a polyvinylidene difluoride membrane (Bio-Rad) and immunoblotted with specific antibodies. Western blots were then visualized using horseradish peroxidase-conjugated secondary antibodies (Amersham Biosciences, Inc.), followed by enhanced chemiluminescence (Amersham Biosciences).
For detection of cytochrome c and AIF, cells were incubated in cell extract buffer (50 mM PIPES, pH 7.4, 50 mM KCl, 5 mM EGTA, 2 mM MgCl 2 , 1 mM dithiothreitol, 10 M cytochalasin B, and 1 mM phenylmethylsulfonyl fluoride) for 30 min at 4°C, homogenized with a Dounce homogenizer, and centrifuged at 14,000 ϫ g for 30 min at 4°C. The supernatant was removed and stored as cytosolic fraction.
The mitochondria-enriched heavy membrane (HM) fraction was prepared as follows. Cells were washed in isotonic buffer (10 mM HEPES, pH 7.5, 200 mM mannitol, 70 mM sucrose, 1 mM EGTA) supplemented with a mixture of protease inhibitors (Roche Molecular Biochemicals) and homogenized with a Dounce homogenizer. Nuclei and unbroken cells were separated at 120 ϫ g for 5 min. The supernatant was centrifuged at 10,000 ϫ g for 30 min to collect HM pellet.
Cell-free Extracts-Cytoplasmic extracts were derived from uninfected and HIV-infected primary CD4ϩ T cells stimulated with PHA/ IL-2. Cytoplasmic extracts from Jurkat T cells incubated in the absence or presence of CD95 mAbs (7C11, Coulter Corp.) for 4 h were also used as a control. Cell-free extracts and nuclei from CEM cells were prepared as previously described (47). Briefly, cytoplasmic extracts were prepared as follows. Cells were washed twice in phosphate-buffered saline and incubated on ice for 20 min with cell extract buffer. Cells were lysed with a B-type pestle. Lysis was monitored by phase-contrast microscopy. The cell lysate was centrifuged at 4°C for 15 min at 17,000 ϫ g, and the clear cytosol was carefully removed. CEM nuclei were prepared as follows. CEM cells were washed twice in phosphate-buffered saline and once in nuclei isolation buffer (10 mM PIPES, pH 7.4, 10 mM KCl, 2 mM MgCl 2 , 1 mM dithiothreitol, 10 M cytochalasin B, and 1 mM phenylmethylsulfonyl fluoride), resuspended in nuclei isolation buffer, allowed to swell on ice for 20 min, and gently lysed with a Dounce homogenizer. Liberated nuclei were then layered over 30% sucrose in nuclei isolation buffer and centrifuged at 800 ϫ g for 10 min, followed by washing in nuclei isolation buffer and resuspension in nuclei storage buffer (10 mM PIPES, pH 7.4, 80 mM KCl, 20 mM NaCl, 250 mM sucrose, 5 mM EGTA, 1 mM dithiothreitol, 0.5 mM spermidine, 0.2 mM spermine, 1 mM phenylmethylsulfonyl fluoride, and 50% glycerol) at 2 ϫ 10 8 nuclei/ml. Nuclei were stored at Ϫ80°C until use.
Cell Surface Staining-Two-color flow cytofluorometric analysis (FACScan; Becton Dickinson) was performed by co-staining cells with mAbs (including isotype controls) directly labeled with phosphatidylethanolamine or PercP. Lymphocytes were gated by forward and side scatter parameters.
Statistical Analysis-Statistical significance (p) was assessed using the paired Student's t test as indicated in the figure legends.

Productive HIV Infection of Primary CD4ϩ T Cells Induces both Typical Apoptosis and Nonapoptotic Cell Death-Resting
PBMC from healthy donors were incubated for 2 h with the laboratory strain HIV-1 LAI . After removing unbound residual virus, T cells were stimulated with PHA and IL-2 for 4 days, and cell death and viral production were assessed. Both cell death and viral replication (assessed by p24 expression) increased with the inoculum (Fig. 1A). Absolute numbers of CD4ϩ T cells were decreased in the infected cultures, the extent of depletion being proportional to the viral input (Fig.  1B). Cytofluorimetric analysis of T cell populations indicated that after 4 days of PHA and IL-2 stimulation, ϳ10% of CD4ϩ T cells from six different donors inoculated with HIV-1 LAI remained in the culture, compared with 55% in the uninfected cultures (Fig. 1C). Cell death preceded any significant appearance of syncytia in the culture (data not shown).
In order to exclude the possibility that HIV-induced CD4 T cell death in the cultures resulted (at least in part) from CD8ϩ T cell-mediated killing, we purified CD4ϩ T cells by negative selection prior incubation with HIV-1 LAI and PHA/IL-2 stimulation (Fig. 1D). In these conditions (as in unfractionated PBMC), HIV-1 LAI induced CD4 T cell depletion. This strongly suggested that viral-induced CD4 T cell depletion occurred independently of participation of additional lymphoid cells, such as CD8ϩ T cells, NK cells, or B cells.
Approximately 50% of the HIV-infected cells displayed cell shrinkage (Fig. 2, B and E), phosphatidylserine (PS) exposure ( Fig. 2, C and E) and loss in ⌬m (Fig. 2, D and E). In contrast, nuclear chromatin condensation and fragmentation typical of apoptosis, as visualized by UV microscopy using Hoechst 33342 nuclear dye ( Fig. 2A), was observed in only 20 -30% of CD4ϩ T cells in the HIV-infected cultures (Fig. 2E). Thus, half of the dying CD4ϩ T cells displayed typical apoptosis, and the other half displayed a nonapoptotic cell death phenotype characterized by lack of nuclear chromatin condensation and fragmentation associated with a loss in membrane integrity (Fig. 2E). By double staining using annexin-V, which detects PS exposure, and propidium iodide (PI), which measures membrane integrity, HIV-infected cultures contained both annexin-V-positive/PI-negative CD4ϩ T cells (apoptotic) and annexin-V-positive/PI-positive CD4ϩ T cells (nonapoptotic) (20.5 Ϯ 6 and 32.3 Ϯ 8%, respectively; data not shown). Altogether, these data suggest that HIV induces both typical apoptosis and an additional phenotype of cell death. The latter was distinct from necrosis (membrane permeability and cell swelling), since CD4ϩ T cells displayed a membrane permeability associated with a cell shrinkage.
Antagonists of the TNFR Family Members Do Not Prevent HIV-1 LAI -mediated CD4 T Cell Death-As shown in Fig. 3, A and B, when CD4ϩ T cells were incubated with HIV-1 LAI and stimulated by PHA/IL-2, CD4ϩ T cells were primed for cell death in response to antibody-mediated CD95 ligation at day 4. In contrast, uninfected activated CD4ϩ T cells became sensitive to CD95 ligation-mediated death at day 6, as previously described (48) (data not shown). Although HIV-infected CD4ϩ T cells became sensitive to CD95 ligation, this process appeared not to account for their subsequent death in the absence of any CD95 antibody treatment. Indeed, when cultures were treated 3 days after PHA/IL-2 stimulation with a neutralizing mAb (ZB4) or a CD95-Fc decoy receptor (which prevents interaction between CD95L and CD95 (49)), CD4 T cell death observed 3 days later was not prevented (Fig. 3C). Similarly, no preventive effect was observed when CD95-Fc was added at day 1 following PHA stimulation (data not shown). The cellular localization of CD95L was examined (50) by isolating cytosolic CD95L/CD95 is a death ligand/receptor pair of the TNF/TNF receptor superfamily, which also includes other pairs, such as TRAIL/TRAILR1, TRAIL/TRAILR2, and the orphan death receptor TRAMP(DR3) (51). Using either neutralizing antibodies or decoy receptors that have been previously demonstrated to block the interaction between ligands and receptors (44,45,52) (Fig. 3D), we assessed whether HIV-mediated cell death of CD4ϩ T cells stimulated with PHA/IL-2 involved one or more of these death receptors. In vitro treatment with the decoy receptors or neutralizing antibodies 3 days after PHA/IL-2 stimulation did not prevent CD4 T cell death observed 3 days later (Fig. 3C). These data suggest that HIV-1 triggers a cell death pathway that is independent of the TNF receptor family (CD95, TNFR1/TNFR2, TRAILR1/R2/OPG, TRAMP).
HIV-1 LAI -mediated Cell Death Is Associated with Caspase Activation-Caspases, the main effectors of apoptosis, are synthesized as inactive procaspases that require processing to become active (53). The current model suggests that after apoptotic stimuli, activation of initiator caspases such as caspase-8 and caspase-9 leads to the proteolytic cleavage of effector caspases such as caspase-3. To assess the role of caspases in HIV-mediated cell death, caspases from CD4ϩ T cells were analyzed by Western blotting. We observed that in uninfected and HIV-infected PHA-stimulated CD4ϩ T cells, the amount of proenzymes detected was increased compared with that of unstimulated T cells at both 4 and 6 days. Caspase-3 (32-kDa proenzyme) was reproductively found to be processed into fragments of 20 kDa after PHA/IL-2 stimulation (Fig. 4). None of these fragments (17,19, and 21 kDa) was observed in unstimulated CD4ϩ T cells. The observation of caspase cleavage in PHA-activated CD4ϩ T cells is in agreement with a recent report showing that CD4ϩ T cells, in contrast to transformed T cell lines, rapidly process caspases, including caspase-3, and caspase substrates following T cell activation (54). Although a similar pattern was observed comparing uninfected and HIVinfected CD4ϩ T cells, the amount of caspase-3 proenzyme in HIV-infected CD4ϩ T cells at days 4 and 6 after PHA/IL-2 stimulation was decreased compared with that in uninfected CD4ϩ T cells (Fig. 4), suggesting an increased processing of the proenzyme in the HIV-infected CD4ϩ T cells. Activation of the initiator caspases was also assessed using antibodies against caspase-8 and -9. Western blotting showed that after stimulation with PHA/IL-2 caspase-8 was processed into 43-45-kDa fragments, but the 18-kDa active subunit was detected at day 6 only in HIV-infected CD4ϩ T cells. In addition, the amount of proenzymes in HIV-infected CD4ϩ T cells drastically decreased at day 6 compared with that in uninfected CD4ϩ T cells (Fig. 4). In the same extracts, caspase-9 (46 kDa) was processed into a 37-kDa fragment following PHA/IL-2 stimulation in both uninfected and HIV-infected CD4ϩ T cultures. The amount of proenzyme decreased at days 4 and 6 in HIV-infected CD4ϩ T cells, compared with that of uninfected CD4ϩ T cells (Fig. 4). Altogether, these results suggest that HIV-1 infection induces procaspases processing. Since caspase-3 was activated after PHA/IL-2 stimulation, we determined whether caspase substrates were also processed in stimulated CD4ϩ T cells. PARP (115 kDa) is one of the caspase-3 substrates cleaved during apoptosis (55). Stimulation with PHA/IL-2 increased PARP protein levels, confirming previous reports showing an up-regulation of PARP and its cleavage in stimu-lated primary T cells (54). Concomitant with its induction, 80% of PARP was found processed into an 85-kDa fragment in PHA-stimulated CD4ϩ T cells, whether or not they were infected with HIV-1 (Fig. 4).
Inhibitor of Apoptosis Proteins (IAPs) have been proposed to inhibit caspases in more distal portions of the cell death pathway, downstream of cytochrome c (56,57). Western blotting of c-IAP1 and survivin (Fig. 5) revealed that the amount of c-IAP1 was much lower in the cytosolic fraction of HIV-infected CD4ϩ T cells compared with that in uninfected CD4ϩ T cells (c-IAP1 protein was decreased by 60% in two independent experiments). In contrast, no major difference in the expression of survivin protein was observed when comparing HIV-infected and uninfected CD4ϩ T cells (a difference of less than 10% was observed). These data indicate that cell death of productively HIV-infected CD4ϩ T cells is associated with a decrease in c-IAP1 expression.
We next determined the apoptogenic effect of cytoplasmic extracts on isolated CEM nuclei in a cell-free system. In these studies, cytosolic fractions of PHA-stimulated CD4ϩ T cells from uninfected and HIV-infected cultures were compared. The proportion of dying cells, as assessed by evaluation of PS exposure, was 20.2 and 66.5% in uninfected and HIV-infected cultures, respectively (data not shown). As a control, we used cytoplasmic extracts prepared from Jurkat T cells treated or not with CD95 mAbs. The cytosolic fraction of HIV-infected FIG. 4. Procaspase processing in uninfected and HIV-infected CD4؉ T cells following PHA/IL-2 stimulation. CD4ϩ T cells were purified from PBMC of normal donors using negative selection and then incubated for 2 h with either medium (Ϫ) or HIV-1 LAI (ϩ) at a MOI of 0.01, washed, and cultured for 4 (4d) or 6 days (6d) with PHA/IL-2 (ϩ) or without stimulation (Ϫ). Extracts from uninfected and HIV-infected CD4ϩ T cells were analyzed for caspase-3, caspase-8, caspase-9, and PARP by Western blotting. Extracts from Jurkat cells cultured in the absence (Ϫ) or presence of CD95 mAbs (ϩ) are shown as controls. For caspase-8, the Western blot was exposed for 3 min (proenzyme) and 25 min (cleaved products). As a control of loading, actin was used. Proenzymes and cleaved subunits are indicated on the left by arrows. The percentages of dying cells (% of PS ϩ cells) determined by flow cytometry using annexin-V-FITC are indicated. The experiment is representative of three independent experiments performed.
CD4ϩ T cells mediated nuclear chromatin condensation and fragmentation as visualized by flow cytometric and microscopic analysis (Fig. 6), as was observed in nuclei incubated with the cytosolic fraction of CD95 mAb-treated Jurkat cells. In contrast, the activity of cytosol from uninfected CD4ϩ T cells was quite similar to the activity of cytosol from Jurkat T cells cultured in the absence of CD95 mAbs. Next, using zVAD-fmk, a broad caspase inhibitor, we explored whether the inhibition of caspases prevents nuclear chromatin fragmentation and condensation. When cytoplasmic extracts from Jurkat T cells treated with CD95 mAbs and HIV-infected CD4ϩ T cells were incubated with zVAD-fmk (50 M), the degradation of CEM nuclei was inhibited (Fig. 6). Our data suggest that caspases are the main effectors involved in chromatin condensation and fragmentation during HIV-mediated cell death.
HIV-1 LAI Mediated a Caspase-independent Cell Death-Since caspase inhibitor prevented chromatin condensation and fragmentation in a cell-free system, HIV-infected CD4ϩ T cells were cultured in the presence of zVAD-fmk to assess the role of caspases in HIV-mediated cell death. The apoptotic phenotype (chromatin condensation and fragmentation) was markedly decreased in cells maintained in the presence of zVAD-fmk (Fig.  7A). zVAD-fmk did not prevent ⌬m loss (Fig. 7B), PS exposure (Fig. 7C), nor cell shrinkage (Fig. 7D). In fact, zVAD-fmk treatment turned the apoptotic cell death phenotype into a nonapoptotic cell death phenotype, as visualized by UV microscopic analysis using Hoechst 33342/PI double staining (Fig. 7E) and by light microscopic analysis using trypan blue dye reagent (Fig. 7F). Thus, our data suggest that HIV induces a caspaseindependent cell death pathway that is associated with a disruption of mitochondrial membrane potential.
Involvement of Mitochondria during HIV-1 LAI -mediated Cell Death-A variety of key events during programmed cell death focus on mitochondria, including loss of ⌬m and the release of apoptogenic factors into the cytosol (35). To evaluate the role of apoptogenic factors released by mitochondria in HIV-mediated cell death, cytosolic fractions prepared from uninfected and HIV-infected CD4ϩ T cells were analyzed using specific antibodies for the presence of cytochrome c, a 15-kDa protein that is involved in caspase activation, and AIF, a 57-kDa protein that is involved in caspase-independent cell death (43). Fig. 8 shows that the cytosolic fraction of HIV-infected CD4ϩ T cells at day 6 after PHA/IL-2 stimulation contains more cytochrome c and more AIF than uninfected CD4ϩ T cells, suggesting that mitochondria from HIV-infected CD4ϩ T cells release these two factors.
Members of the Bcl-2 family have been shown to be involved in the regulation of mitochondria permeability during apo-ptosis (35). Bcl-2 expression (an antiapoptotic protein) was assessed by flow cytometry after gating on live and dying cells. The mean intensity of fluorescence (MIF) of Bcl-2 expression in live cells (high forward scatter parameter) from uninfected (MIF ϭ 46.3) and HIV-infected CD4ϩ T cells (MIF ϭ 42.9) was similar (Fig. 9A) and was markedly decreased in dying cells (shrunken cells) of both uninfected and HIV-infected CD4ϩ T cells (MIF ϭ 30.8 and 23.2, respectively), suggesting that loss of Bcl-2 is associated with cell death. By two-color flow cytometric analysis, we assessed Bcl-2 expression in live productively infected CD4ϩ T cells as determined by p24 antigen expression (Fig. 9B). Our data indicate that Bcl-2 expression was equivalent in CD4ϩ T cells in which HIV was or was not actively replicating, suggesting that HIV-mediated cell death is not associated with an early change in Bcl-2 expression. However, after cell fractionation, we observed that more Bax (a proapoptotic protein) was present in the mitochondria enriched HM fraction of infected CD4ϩ T cells (Bax protein was increased by 55% in two independent experiments) than that present in uninfected CD4ϩ T cells (Fig. 9D). In vitro treatment of infected CD4ϩ T cells with the reverse transcriptase inhibitor DDI, which prevented cell death (PS exposure and loss in ⌬⌿m) (Fig. 9C), also prevented targeting of Bax to the mitochondria (Fig. 9D). These data suggest that HIV-1 may induce CD4 T cell death via the targeting of Bax to the mitochondria. DISCUSSION Our results suggest that HIV-1 induces CD4 T cell death through a TNF receptor family-independent pathway. These findings confirm and extend previous findings indicating that productive infection of PHA/IL-2 stimulated primary CD4ϩ T cells, as well as productive infection of transformed CD4ϩ T cell lines, induces a process of CD4 T cell death that does not involve CD95/CD95L interactions (4 -6). Other death pathways, such as TRAIL/DR4 or TNF/TNFR, have been suggested to play a role in apoptosis of T cells from HIV-infected persons (20,21), but decoy receptors corresponding to these and other TNF receptor family members did not prevent HIV-mediated death of primary CD4ϩ T cells stimulated with PHA/IL-2, indicating that death occurred independently of these death receptors (CD95, TNFR1/R2/OPG, and TRAMP). In agreement with an absence of involvement of death receptors in HIVmediated cell death, we did not observe an early full processing of caspase-8. Caspase-8 is generally considered to be an initiator caspase, which associates with the adaptator molecule FADD recruited to the death receptor. Full processing of caspase-8 in the 18-kDa active form was only observed at day 6, suggesting that caspase-8 activation may be a late event in the pathway leading to cell death in response to HIV infection (34).
Although HIV induced the cleavage of caspase-3, -8, and -9 proenzymes and markedly decreased the expression of c-IAP1, an inhibitor of caspase-3, our experiments indicate that caspases are dispensable for HIV-mediated programmed cell death. Nevertheless, these proteases do play a role in the nuclear apoptotic phenotype as evidenced by the effects of zVAD-fmk. Several recent reports have suggested that caspase inhibitors inhibit apoptosis induced by diverse stimuli but do not prevent further cell death (36 -38, 40 -42, 58, 59) and lead to the appearance of cells with a necrotic-like phenotype. Our data show that the nonapoptotic phenotype is characterized by membrane permeabilization and intact nuclei and shares with the apoptotic phenotype the presence of cell shrinkage. This experiments is representative of two independent experiments performed. F, apoptotic and nonapoptotic cells were also determined by microscopic analysis using trypan blue dye exclusion as described in the legend to Fig. 2E. 200 cells were counted in duplicate. The data shown are the mean of three independent experiments performed. suggests that the cell death program induced by HIV replication is distinct from necrosis (cell swelling) (60). Cell shrinkage, mediated by apoptotic inducers, is believed to be tightly linked to enhanced K ϩ efflux (61)(62)(63). Thus, it would be interesting to determine whether cell volume is directly related to an efflux of K ϩ during HIV-mediated cell death and if HIV deregulates the continuous activity of the Na ϩ /K ϩ ATPase pump.
Cell death mediated by HIV is associated with a loss of ⌬m, even when caspase activation is blocked, suggesting a crucial role for mitochondria in the regulation of this cell death. The loss of mitochondrial membrane potential is associated with the release of cytochrome c and AIF from the mitochondria. Among critical effectors, AIF, a flavoprotein, has been reported to directly cause chromatin condensation in isolated nuclei and participates in caspase-independent programmed cell death. Although, AIF protein was detected in the cytosolic fraction of HIV-infected CD4ϩ T cells along with cytochrome c, isolated nuclei treated with these extracts did not display the major features of AIF-induced nuclei morphology (64), even when the nuclei were incubated in the presence of zVAD-fmk. During preparation of this manuscript, Ferri et al. (28) reported that syncitia arising from the fusion of cells expressing gp120 with cells expressing the CD4 and CXCR4 molecules complex spontaneously undergo cell death, displaying evidence of caspase activation, loss of ⌬m, and the release of cytochrome c and AIF from mitochondria. Although AIF has been assumed to be a potent activator of programmed cell death, recent work has suggested that AIF release from the mitochondria and translocation to the nucleus can occur in the absence of chromatin condensation and cell death (38,65). Moreover, the study of AIF knockout mice suggests the existence of a third cell death pathway independent of caspases and AIF (66). Although dispensable for cell death, our data suggest that caspases are the main nuclear effectors involved in HIV-mediated DNA condensation and fragmentation, in which AIF is either not involved or not a major contributory factor. Whether AIF participates in other effector pathways of cell death remains to be investigated.
The mechanism responsible for mitochondrial membrane permeabilization has been reported to involve proapoptotic members of the Bcl-2 family, the permeability transition pore complex, the adenine nucleotide translocator, and/or the voltagedependent anion channel (67). It has been proposed that the HIV-1 protease can cleave Bcl-2, thereby abolishing its mitochondrial membrane potential-inhibitory function (68). However, we observed that Bcl-2 expression was equivalent in CD4ϩ T cells, whether uninfected or HIV-infected, and Western blot analysis did not show the presence of fragmented Bcl-2 (data not shown), suggesting that the loss of Bcl-2 may be a consequence of cell death and not directly involved in the loss of ⌬m. Upon death stimuli, Bax is rapidly translocated to the mitochondria and induces the loss of ⌬m, cytochrome c release, and activation of caspases (69 -72). Here, we showed that the amount of Bax was more important in the mitochondriaenriched heavy membrane fraction of HIV-infected CD4ϩ T cells than in that of uninfected and DDI-treated HIV-infected CD4ϩ T cells. It has been recently reported that HIV-1 induces an up-regulation of Bax via a p53 pathway (73). Thus, the result of Genini et al. (73) and our data suggest that HIV-1 may induce both Bax up-regulation and Bax targeting to the mitochondria. However, the nature of the signals that may lead to Bax translocation during HIV-1 infection remains an open question.
Studies by Xiang et al. (39) have shown that inducible Bax expression triggers rapid death even in the presence of caspase inhibitors, and nonapoptotic death proceeds through the generation of reactive oxygen species (ROS). In our model, HIVinfected CD4ϩ T cells die with a nonapoptotic phenotype when treated with the broad spectrum caspase inhibitor (zVAD-fmk). Preliminary data measuring intracellular oxidant levels using the oxidation-sensitive dye dihydroethidium indicate that the levels of ROS in HIV-infected CD4ϩ T cells were higher than in uninfected CD4ϩ T cells and that ROS levels decreased in the presence of DDI. Thus, independently of caspase activation, the ability of Bax to induce disruption of mitochondrial membrane permeability might be expected to result in the disruption of electron transport following cytochrome c release, loss of ATP and an increase in ROS may ultimately cause a nonapoptotic cell death (74).
At this stage, however, we cannot exclude other death pathways that result in increased mitochondrial membrane permeability. Indeed, it has been recently proposed that Vpr (a protein encoded by HIV) is directly targeted to the mitochondrial permeability transition pore complex and permeabilizes mitochondrial membranes in a cell-free system (27). The Vpr protein has been reported to kill lymphocytes, monocytes, and neurons (26,75,76) and to induce mitochondrial dysfunction in the yeast Saccharomyces cerevisiae (77). However, it may also act as a negative regulator of cell death in human cell lines and thymocytes (78,79), and CEM T cells infected with a triple FIG. 9-continued mutant (nef, vpr, and vpu deleted) derived from HIV-1 also undergo cell death (80), suggesting that Vpr is dispensable for HIV-mediated CD4 T cell depletion. Thus, the role of HIV-1 proteins in inducing cell death through the loss of ⌬m remains to be elucidated, and other viral products may be involved in this process.
In conclusion, our data suggest that HIV-1 infection of cycling primary CD4ϩ T cells results in a cell death process associated with a mitochondrial deregulation (loss in ⌬m). Caspases that mediate chromatin condensation and fragmentation are the main effectors of the apoptotic phenotype but are dispensable for cell death. Our data indicate a crucial role of the mitochondria in the regulation of cell death induced by HIV and suggest that the targeting of Bax to the mitochondria may be a major contributory factor.