Syncytia arising from the fusion of cells expressing a lymphotropic HIV type 1–encoded envelope glycoprotein complex (Env) with cells expressing the CD4/CXC chemokine receptor 4 complex spontaneously undergo cell death. Here we show that this process is accompanied by caspase activation and signs of mitochondrial membrane permeabilization (MMP), including the release of intermembrane proteins such as cytochrome c (Cyt-c) and apoptosis-inducing factor (AIF) from mitochondria. In Env-induced syncytia, caspase inhibition did not suppress AIF- and Cyt-c translocation, yet it prevented all signs of nuclear apoptosis. Translocation of Bax to mitochondria led to MMP, which was inhibited by microinjected Bcl-2 protein or bcl-2 transfection. Bcl-2 also prevented the subsequent nuclear chromatin condensation and DNA fragmentation. The release of AIF occurred before that of Cyt-c and before caspase activation. Microinjection of AIF into syncytia sufficed to trigger rapid, caspase-independent Cyt-c release. Neutralization of endogenous AIF by injection of an antibody prevented all signs of spontaneous apoptosis occurring in syncytia, including the Cyt-c release and nuclear apoptosis. In contrast, Cyt-c neutralization only prevented nuclear apoptosis, and did not affect AIF release. Our results establish that the following molecular sequence governs apoptosis of Env-induced syncytia: Bax-mediated/Bcl-2–inhibited MMP → AIF release → Cyt-c release → caspase activation → nuclear apoptosis.
Infection by HIV-1 is accompanied by an increased apoptotic turnover of lymphocytes, monocytes, and neurons that can be detected either ex vivo in freshly explanted cells, or in vivo by histocytochemical detection of apoptotic cells. The mechanisms of this increased cellular demise are complex and involve direct effects of the virus and viral products, as well as indirect host-mediated factors 1,2. In vitro, HIV-1 infection can cause apoptosis via a multitude of different mechanisms, including the action of the proapoptotic proteins Tat 3,4 and Vpr 5,6,7 and, perhaps more importantly, via interactions between the glycoprotein (gp)120/gp41 envelope complex (Env) with its receptor (CD4) and a suitable coreceptor (e.g., CXC chemokine receptor [CXCR]4). In cultures of T cells inoculated with lymphotropic HIV-1 strains, Env expressed on the plasma membrane of infected cells interacts with CD4/CXCR4 of uninfected cells, resulting in cell fusion 8,9,10. After several rounds of fusion, syncytia attain volumes equivalent to several dozens or hundreds of individual cells and ultimately lyse, while exhibiting several biochemical characteristics of apoptosis and/or signs of necrosis such as cytoplasmic vacuolization 11,12,13,14. According to several studies, syncytium formation is the principal cause of HIV-1–mediated T cell destruction in vitro 11,15. Env variants interacting with CD4/CXCR4 (rather than those having a preference for CD4/CCR5) are mostly encoded by syncytium-inducing HIV-1 strains, and a strong correlation between CD4+ T cell decline and infection by syncytium-inducing HIV-1 variants has been established by some authors in vitro and in vivo 8,9,15,16.
Apoptosis of single cells is generally associated with the activation of caspases 17, a set of specific proteases which either can serve as molecular switches initiating cell death pathways or, when activated in a massive fashion, can mediate the degradation of essential structural and regulatory proteins, culminating in the activation of the caspase-activated DNase, the enzyme responsible for oligonucleosomal DNA fragmentation 18,19. In addition, apoptosis is accompanied by signs of mitochondrial membrane permeabilization (MMP), including a loss of the inner mitochondrial transmembrane potential (ΔΨm) and the release of soluble intermembrane proteins via the outer mitochondrial membrane 20,21,22,23,24. The functional hierarchy among these events depends on the apoptosis-inducing trigger. In the “extrinsic” pathway, caspase activation is induced as an upstream event, e.g., by recruiting apical caspases to the death-induced signaling complex formed upon the ligation of death receptors expressed on the cell surface (e.g., CD95 and TNFR). Depending on the cell type, receptor-proximal caspase activation then either suffices to set off the caspase activation cascade in an MMP-independent fashion (“type 1 cells”) or must relay to MMP via cleavage of Bid and/or production of ceramide (“type 2 cells” 25,26,27). In contrast, most “intrinsic” apoptosis triggers (e.g., DNA damage, glucocorticoids, and reactive oxygen species) activate caspases in an indirect fashion, namely by first permeabilizing mitochondrial membranes in a Bcl-2–inhibitable fashion 20,21,22,23,28,29. MMP results in the translocation of cytochrome c (Cyt-c) from the mitochondrial intermembrane space to the cytosol, where Cyt-c triggers the assembly of a caspase-9–caspase-3 activation complex, the apoptosome 18,21. In addition to Cyt-c, mitochondria can release a variety of intermembrane proteins 30, including a cell type–specific set of procaspases 31,32,33 and the apoptosis-inducing factor (AIF 34), a flavoprotein oxidoreductase that translocates to the cytosol as well as to the nucleus and stimulates apoptosis via an as yet unknown, caspase-independent mechanism.
As in other models of apoptosis induction, signaling via CD4/CXCR4 induces both caspase activation 35,36,37,38,39 and signs of MMP 14,40. However, the precise mechanisms of Env-mediated apoptosis induction and the molecular order of these events have remained elusive in syncytia. Stimulated by these premises, we studied the role of caspases and MMP in syncytia arising from the fusion between Env- and CD4/CXCR4-expressing cells. Here we report that the death of Env-induced syncytia obeys the rules of the intrinsic pathway of apoptosis induction in the sense that caspase activation occurs downstream of MMP. Our results indicate that MMP is regulated by members of the Bcl-2/Bax family and establish a molecular order of apoptotic signaling in which AIF release occurs upstream of that of Cyt-c, which in turn is required for caspase-dependent nuclear apoptosis.
Materials And Methods
Cells and Culture Conditions.
HeLa cells stably transfected with a vector containing the env gene of HIV-1 LAI (HeLa 243 Env 41) were cultured in complete culture medium (DMEM supplemented with 2 mM glutamine, 10% FCS, 1 mM pyruvate, 10 mM Hepes, and 100 U/ml penicillin/streptomycin) containing 2 μM methotrexate. HeLa cells stably transfected with CD4 (HeLa P4; a gift from P. Charneau, Pasteur Institute, Paris, France 42) were selected in medium containing 500 μg/ml G418. Jurkat cells expressing CD4 and CXCR4 and stably transfected with the human Bcl-2 gene or a neomycin (Neo) resistance vector 43 only were provided by N. Israel (Pasteur Institute, Paris, France). Neo and Bcl-2 U937 cells 44 were a gift from F. Hirsch (Centre National de la Recherche Scientifique, ERS1984, Villejuif, France). Cocultures of different cell types were performed in complete culture medium in the absence of selecting antibiotics by adding trypsinized HeLa cells to adherent HeLa CD4 or HeLa Env cells (density: 1–1.5 × 103 cells/mm2) at an ∼1:1 ratio or by adding Jurkat or U937 cells to adherent HeLa Env cells (National Institutes of Health AIDS Research and Reference Reagent Program, Bethesda, MD). Apoptosis was induced or inhibited by staurosporin (2 μM; Sigma-Aldrich), N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethylketone (Z-VAD. fmk), Boc-Asp-fluoromethylketone (Boc-D.fmk), or N-benzyloxycarbonyl-Phe-Ala-fluoromethylketone (Z-FA.fmk) (all used at 100 μM added every 24 h; Enzyme Systems).
5 × 106 HeLa CD4 cells were cultured in the presence of 2.5 × 106 chronically HIV-1–infected H9/IIIB cells obtained from Dr. R.C. Gallo (National Institutes of Health, Bethesda, MD) in 5 ml of RPMI 1640 supplemented with 10% heat-inactivated FCS, 2 mM glutamine, and 25 IU/ml recombinant human IL-2. Before the addition of chronically infected cells, target cells were preincubated for 30 min at 37°C in the presence or absence of Z-VAD.fmk (50 μM), and Z-VAD.fmk was readded every 12 h throughout the culture period.
Fluorescence Staining of Live Cells and Immunofluorescence.
For the assessment of mitochondrial and nuclear features of apoptosis, cells cultured on a coverslip were stained with 5,5′,6,6′-tetrachloro-1,1′, 3,3′-tetraethylbenzimidazolylcarbocyanine iodide (JC-1, 2 μM; Molecular Probes) and Hoechst 33342 (2 μM; Sigma-Aldrich) for 30 min at 37°C in complete culture medium. A rabbit antiserum generated against a mixture of three peptides derived from the mouse AIF amino acid (aa) sequence (aa 151–170, 166–185, 181–200, coupled to KLH 34) was used (diluted 1:1,000) on paraformaldehyde (4% wt/vol) and picric acid–fixed (0.19% vol/vol) cells and revealed with a goat anti–rabbit IgG conjugated to PE (Southern Biotechnology Associates, Inc.). Cells were also stained for the detection of Cyt-c (mAb 6H2.B4 from BD PharMingen; revealed by a goat anti–mouse IgG1 FITC conjugate from Southern Biotechnology Associates, Inc.), heat shock protein (hsp)60 (mAb H4149 from Sigma-Aldrich; revealed by a goat anti–mouse IgG1 FITC), Cyt-c oxidase (COX subunit IV, mAb 2038C12 from BD PharMingen; revealed by a goat anti–mouse IgG2a FITC conjugate), Bax (66241A; BD PharMingen), and/or chromatin (Hoechst 33342, 2 μM, 15 min of incubation at room temperature). Several stages of nuclear apoptosis were distinguished by staining with Hoechst 33342: stage I with rippled nuclear contours and a rather partial chromatin condensation, stage IIa with marked peripheral chromatin condensation, and stage IIb with formation of nuclear bodies 45. A rabbit polyclonal antiserum, CM1 (detected as for anti-AIF), which recognizes the p18 subunit of cleaved caspase-3 but not the zymogen 46, was employed to detect the proteolytic activation of caspase-3, followed by detection of the fluorescence intensity by confocal microscopy.
DNA Gel Electrophoresis.
For pulse field gel electrophoresis, DNA was prepared from agarose plugs (2 × 106 nuclei 47), followed by electrophoresis in a Bio-Rad Laboratories CHEF-DR II (1% agarose, TBE, 200 V, 24 h, pulse wave 60 s, 120° angle).
Subcellular Fractionation and Immunoblotting.
Cells were washed once in PBS, resuspended in isotonic HIM buffer (200 mM mannitol, 70 mM sucrose, 1 mM EGTA, 10 mM Hepes, pH 7.5) supplemented with a protease inhibitor mixture (added at a 1:100 dilution; Sigma-Aldrich), and homogenized using a Polytron homogenizer (Brinkmann Instruments) at setting 6.5 for 10 s. Nuclei were separated at 600 g for 10 min as the low speed pellet and washed twice (1,200 g, 10 min). The supernatant was centrifuged at 10,000 g for 10 min to collect the heavy membrane pellet enriched in mitochondria. This supernatant was centrifuged at 100,000 g for 30 min to yield the organelle-free cytosols. Samples were stored at −80°C until analysis by SDS-PAGE under reducing conditions (40 μg protein/lane), Western blot, and immunodetection of AIF and Cyt-c as described 34.
All syncytia growing on a premarked V-shaped area of a coverslip (>200 per experiment) were microinjected into the cytoplasm (1–3 injections per syncytium depending on their size, 1 injection per ∼4 nuclei) using a computer-controlled microinjector (pressure 200 hPa, 2 s; Eppendorf) with PBS only (pH 7.2), recombinant AIF protein (500 ng/μl 34), horse Cyt-c (12 μg/μl; Sigma-Aldrich), a neutralizing anti-AIF rabbit Ab (titer ∼105 34; diluted 1:1 with PBS) optionally neutralized by preincubation with 10 μM of AIF immunogenic peptides, a preimmune rabbit antiserum, a neutralizing Cyt-c–specific IgG1 mAb (6H2.B4, 250 ng/μl; BD PharMingen), an irrelevant isotype-matched control mAb (anti-hsp60), Koenig's polyanion (2.5 μM; a gift from Dieter Brdizcka, University of Konstanz, Konstanz, Germany), recombinant human Bcl-2 (aa 1–218, 500 ng/μl), Bcl-2Δα5/6 (Bcl-2 Δ143–184, 500 ng/μl), or murine Bax (aa 1–171, 250 ng/μl), BaxΔα5/6 (Δ106–153, 250 ng/μl 48). After microinjection, cells were cultured for 3–24 h and stained for 30 min with the ΔΨm-sensitive dye JC-1 (2 μM) and the DNA-intercalating dye Hoechst 33342 (2 μM), followed by fixation (which removes the JC-1 staining) and immunostaining for Cyt-c and/or AIF as described above.
The quantitation of different parameters by fluorescence microscopy was performed on at least 200 syncytia for each data point, and was repeated at least 3 times in independent experiments, as stated in the figure legends. In at least one experiment out of each series, quantitations were performed in a blinded fashion, and in an additional experiment quantitations were performed independently by two individuals. Interexperimental variability was generally <15%.
Caspase-dependent Nuclear Apoptosis of Syncytia Induced by the Interaction between Env- and CD4-expressing Cells.
HeLa cells stably transfected with human CD4 (HeLa CD4) formed syncytia when cocultured with HeLa cells expressing a lymphotropic HIV-1 Env gene (HeLa Env 42; Fig. 1). After 24 h of coculture, several morphologically normal nuclei (detected with Hoechst 33342, blue fluorescence) could be clearly distinguished within a common cytoplasm of HeLa CD4/HeLa Env hybrids (Fig. 1 A). However, after prolonged culture (48–72 h) an increasing percentage of nuclei manifested apoptotic chromatin condensation (Fig. 1A and Fig. B). This chromatin condensation was restricted to syncytia, and inhibition of cell fusion by addition of the anti-CD4 mAb Leu3A at the beginning of coculture prevented all signs of apoptosis (Fig. 1 A). As in other models of apoptosis 45, chromatin condensation rapidly evolved to a stage with strong Hoechst 33342–detectable condensation of most of the chromatin (stage II), without (stage IIa, in a minority of syncytia) or with the formation of nuclear apoptotic bodies (stage IIb, in most syncytia) (Fig. 1 B). Nuclear apoptosis attained all nuclei within the same heterokaryon in a coordinated fashion (Fig. 1 A), and was accompanied by oligonucleosomal DNA degradation (not shown) as well as by “large scale” (∼50 kbp) DNA fragmentation (Fig. 1 D). All morphological and biochemical signs of nuclear apoptosis were strongly reduced by the two pancaspase inhibitors Boc-D.fmk and Z-VAD.fmk, but not by the chemically related cathepsin inhibitor Z-FA.fmk (Fig. 1 A). Only a minority of cells exhibited a partial peripheral chromatin condensation (stage I) in the presence of caspase inhibitors (Fig. 1 B). In conclusion, Env-induced syncytia spontaneously undergo caspase-dependent nuclear apoptosis.
Caspase-independent Signs of MMP of Env-induced Syncytia.
Staining of Env-induced syncytia with the ΔΨm-sensitive dye JC-1 revealed a progressive ΔΨm loss. Thus, mitochondria from most newly formed syncytia (24 h) possessed a high ΔΨm (red JC-1 fluorescence; Fig. 1 A), whereas mitochondria from aging syncytia (48–72 h) mostly have a low ΔΨm (green JC-1 fluorescence; Fig. 1A and Fig. C). The ΔΨm loss progressed from the perinuclear area to the periphery (not shown), giving rise to a transient intermediate phenotype (Fig. 1 C), and was accompanied by a moderate perinuclear clustering of mitochondria (stained for the matrix protein hsp60, which is not released during apoptosis 30, or the sessile inner membrane protein COX; green fluorescence in Fig. 2 A and Fig. 3 A, respectively). In addition, syncytia progressively manifested signs of outer MMP, as indicated by the translocation of AIF (red fluorescence in Fig. 2 A) from mitochondria to the cytosol and to the nucleus (blue fluorescence in Fig. 2 A and Fig. 3 A), or that of Cyt-c to the cytosol (red fluorescence in Fig. 3 A). The translocation of AIF and Cyt-c from mitochondria to the extramitochondrial compartment was confirmed by subcellular fractionation followed by immunoblot. In single cells or 24-h-old syncytia, AIF and Cyt-c are confined to the mitochondrial compartment. 48 h after initiation of coculture, ectopic AIF becomes detectable in both cytosols and nuclei (Fig. 2 C), whereas ectopic Cyt-c can be detected only in the cytosol (Fig. 3 C). Neither Boc-D.fmk nor Z-VAD.fmk (added every 24 h at a concentration of 100 μM) prevented the mitochondrial manifestations of apoptosis (ΔΨm collapse, AIF and Cyt-c translocation; Fig. 1,Fig. 2,Fig. 3). Hence, MMP proceeds in a caspase-independent fashion in HIV-1–induced syncytia.
Kinetics of AIF and Cyt-c Translocation.
The percentage of cells manifesting ΔΨm dissipation and AIF translocation was higher than that of cells positive for Cyt-c translocation (compare percentage values in Fig. 1 C, 2 B, and 3 B), and double immunofluorescence staining of cells for AIF and Cyt-c confirmed the existence of cells having translocated AIF to the nucleus and still retaining Cyt-c in mitochondria (but not vice versa; not shown), indicating that ΔΨm loss and AIF release occurred before Cyt-c release. AIF and Cyt-c translocation were also observed in heterokaryons generated by coculturing HeLa CD4 cells with a lymphoid cell line chronically infected with a syncytium-inducing HIV-1 isolate. Caspase inhibition with Z-VAD.fmk failed to prevent signs of MMP, although it did inhibit nuclear apoptosis as an internal control of its efficacy (Fig. 4). Kinetic analyses confirmed that mitochondria from HIV-1–induced syncytia translocate AIF before Cyt-c and before caspase-3 activation or nuclear chromatin condensation could be detected (Fig. 4). Hence, immunodetectable translocation of AIF precedes that of Cyt-c.
Bax and Bcl-2 Regulate MMP in Env-induced Syncytia.
Members of the Bcl-2/Bax family regulate apoptosis via their capacity to modulate MMP 21,22,23,49. To address the mechanisms by which MMP occurs in Env-induced syncytia, cells were stained with Abs directed against Bax and Bcl-2. Whereas no changes in Bcl-2 staining were observed upon prolonged culture of syncytia (not shown), Bax (red fluorescence) was found to translocate from a cytoplasmic, preponderantly nonmitochondrial to a punctate, mitochondrial (counterstained with anti-COX, green fluorescence) localization (Fig. 5 A). This finding is reminiscent of other models of apoptosis in which insertion of Bax into mitochondrial membranes causes MMP 50,51,52,53. Microinjection of recombinant Bax (but not microinjection of the mutant BaxΔα5/6 protein lacking the putative membrane insertion domain) induced a rapid (3 h) ΔΨm dissipation, Cyt-c translocation, and nuclear apoptosis (Fig. 5 B). This effect of Bax was reduced by coinjection of Bcl-2 or Koenig's polyanion (Fig. 5 B), an inhibitor of the mitochondrial voltage–dependent anion channel (VDAC) reported to neutralize the effect of Bax on isolated mitochondria 53. Similarly, injection of recombinant Bcl-2 (but not Bcl-2Δα5/6) into freshly generated syncytia inhibited their spontaneous apoptosis, both at the mitochondrial and the nuclear levels (Fig. 5 B). To confirm the Bcl-2–mediated inhibition of apoptosis in a different experimental system, CD4-expressing Jurkat cells or U937 cells were cocultured with HeLa Env cells, a manipulation that resulted in rapid syncytium formation and apoptosis (in the case of Jurkat Neo cells; Fig. 6A–D) or syncytium-independent apoptosis (in the case of U937 Neo cells 54; Fig. 6 E). This latter type of apoptosis requires CD4- and CXCR4-mediated interactions with HeLa Env cells, as it was blocked by the anti-CD4 Ab Leu3a and the natural CXCR4 ligand SDF-1α (Fig. 6 F). Transfection-enforced overexpression of Bcl-2 suppressed the HeLa Env–induced ΔΨm loss and nuclear condensation in both Jurkat and U937 cell types, compared with controls transfected with the Neo resistance vector only (Fig. 6). In conclusion, Bcl-2–regulated MMP is a critical event of Env-induced apoptosis.
Molecular Ordering of Mitochondrial AIF and Cyt-c Release in Syncytia Undergoing Apoptosis.
To investigate the putative functional relationship between the translocation of AIF and Cyt-c, 24-h-old Env-induced syncytia were microinjected with AIF, Cyt-c, or Abs that neutralize either AIF or Cyt-c, followed by determination of apoptotic parameters after 3 h (Fig. 7 A) or 24 h of culture (Fig. 7 B). Microinjection of both AIF and Cyt-c resulted in the rapid (3 h) induction of nuclear apoptosis, the reduction of ΔΨm, and the translocation of endogenous AIF (induced by Cyt-c) or Cyt-c (induced by AIF; Fig. 7 A). Z-VAD.fmk differentially affected the mitochondrial effects of ectopic (extramitochondrial) AIF and Cyt-c. It suppressed the Cyt-c–induced ΔΨm loss and release of AIF. In contrast, Z-VAD.fmk failed to inhibit the AIF-triggered ΔΨm dissipation and Cyt-c translocation (Fig. 7 A). As an internal control, a neutralizing anti-AIF Ab prevented the acute (3 h) AIF effects when microinjected together with AIF (Fig. 7 A). This Ab also prevents the spontaneous ΔΨm loss, Cyt-c release, and nuclear apoptosis of syncytia (Fig. 7 B). This effect is specific, as it was not observed with a preimmune antiserum nor when the AIF Ab was neutralized by preincubation with an excess of AIF-derived immunogenic peptides (Fig. 7 B). In sharp contrast, neutralization of Cyt-c by microinjection of a specific mAb did not prevent the ΔΨm change nor did it affect the AIF translocation occurring during syncytial aging (Fig. 7 B). However, Cyt-c neutralization did inhibit the Hoechst 33342–detectable chromatin condensation (Fig. 7 B). Altogether, these results suggest that ectopic AIF is sufficient and necessary for caspase-independent mitochondrial Cyt-c release, whereas Cyt-c is critical for caspase-dependent nuclear apoptosis.
The Intrinsic Pathway Governs Apoptosis Triggered by Env.
Here we demonstrate that Env-induced syncytia spontaneously undergo apoptosis and that this apoptotic process obeys the rules of the intrinsic (rather than the extrinsic) cell death pathway. This demonstration is based on several lines of evidence. First, syncytia manifest signs of inner MMP (ΔΨm dissipation; Fig. 1) and outer MMP (release of AIF and Cyt-c; Fig. 2 and Fig. 3), in line with the fact that syncytial mitochondria frequently are dilated 13,14. Swelling of mitochondria is associated with MMP and occurs both in early apoptosis of individual cells (before cell shrinkage 55,56,57) and in necrosis 58. Second, outer and inner MMP occurs well before caspases are activated and before nuclear chromatin is condensed in a caspase-dependent fashion (Fig. 1,Fig. 2,Fig. 3,Fig. 4). Third, inhibition of caspases by oligo- or monopeptidic inhibitors does not prevent MMP, confirming that caspases act downstream of MMP (Fig. 1,Fig. 2,Fig. 3). Fourth, inhibition of MMP by microinjection of recombinant Bcl-2 (Fig. 5 B) or by coculture of Env-positive cells with CD4+ cells expressing a Bcl-2 transgene (Fig. 6) prevents nuclear apoptosis, as this may be expected for the intrinsic (but not the extrinsic) pathway of death induction 25,26,27,36.
Formation of syncytia is a nonphysiological process (with the exception of a few cell types such as syncytiotrophoblasts, spermatogonia, osteoclasts, and myocytes 59), supporting the idea that syncytia could be intrinsically condemned to undergo apoptosis. However, HeLa cells driven to form syncytia by culture with methotrexate or by transfection with fusion-competent proteins from human parainfluenza virus type 4a 60 die more slowly than Env/CD4-induced HeLa syncytia (Ferri, K.F., and G. Kroemer, unpublished observation), suggesting that the receptors involved in the fusion process contribute to the triggering of the intrinsic pathway. Accordingly, engagement of CD4 and CXCR4 can induce lymphocyte death without syncytium formation 38,40,61, and U937 cells die upon contact with Env-transfected cells without prior cell fusion (54,62; Fig. 6 E). Although it remains elusive whether Env-induced syncytium-dependent and syncytium-independent apoptosis are mediated by identical pathways, it appears clear that Bcl-2–regulated, presumably Bax-triggered MMP is a critical event of different types of cell death stimulated via the Env–CD4/CXCR4 interaction (Fig. 5 and Fig. 6).
Cell Type–specific Contribution of AIF and Caspases to Nuclear Apoptosis.
In dying syncytia arising from the fusion of Env- and CD4/CXCR4-expressing cells, mitochondria release both Cyt-c and AIF (Fig. 2 and Fig. 3). Microinjection of Abs neutralizing either AIF or Cyt-c prevents chromatin condensation (Fig. 7 B), suggesting that both factors contribute to nuclear apoptosis: AIF in a caspase-independent fashion, and Cyt-c in a caspase-dependent fashion (Fig. 7 A). In a cell-free system, recombinant AIF causes caspase-independent large scale (∼50 kbp) DNA fragmentation and peripheral chromatin condensation 34 when added to purified nuclei. Paradoxically, however, in Z-VAD.fmk–treated 72-h-old syncytia, AIF is clearly present in the nucleus (Fig. 2), yet no ∼50 kbp DNA fragmentation pattern can be detected (Fig. 1 D) and chromatin condensation is strongly reduced (Fig. 1A and Fig. B). Thus, in contrast to fibroblast cell lines in which large scale DNA fragmentation is caspase independent (and presumably AIF mediated 34,45), HeLa syncytia large scale DNA fragmentation appears to be fully caspase dependent, as this has been reported for Jurkat cells in which the caspase-activated DNase accounts for both oligonucleosomal and large scale DNA fragmentation 63. Of note, microinjection of an excess of exogenous recombinant AIF into freshly formed (24 h) syncytia can trigger acute (3 h) chromatin condensation in a caspase-independent fashion (Fig. 7 A). Thus, in principle, AIF can exert an (presumably direct) effect on nuclei from HeLa syncytia. Nonetheless, in the slow (24–48 h) advancement of chromatin condensation observed in HeLa syncytia spontaneously undergoing apoptosis, AIF clearly acts in a caspase-dependent fashion (Fig. 2). Future work will unravel whether AIF-inhibitory factors and/or the abundance of the nuclear AIF target account for these cell type–specific differences.
Hierarchy of Mitochondrial AIF and Cyt-c Release.
Cyt-c release has been reported to occur in a coordinated, nearly simultaneous fashion in most if not all mitochondria of the same cell 64. Accordingly, syncytia retaining Cyt-c or AIF in a fraction of mitochondria were infrequently observed (<1% of the entire population; Fig. 2 A and 3 A) at any time point, whereas a heterogeneity in the ΔΨm loss was evident in a substantial fraction of cells (Fig. 1 C). Internal feed-forward amplification loops could contribute to the rapid kinetics of Cyt-c release. One such amplification loop has been proposed to be provided by caspases, which, once activated as a consequence of MMP, stimulate MMP 65,66. However, in Env-induced syncytia, caspase inhibition does not affect the kinetics of Cyt-c release (Fig. 3 B). A caspase-independent amplification loop may be provided by AIF based on the observation that the microinjection of an anti-AIF Ab retards all signs of MMP, including ΔΨm collapse and the release of Cyt-c (Fig. 7 B), whereas microinjection of recombinant AIF induces MMP (Fig. 7 A). Intriguingly, the mitochondrial release of AIF precedes that of Cyt-c in Env-induced syncytia by several hours (compare the percentage values in Fig. 2 B and 3 B, and in Fig. 4). This appears counterintuitive because protein-permeant pores formed in the outer mitochondrial membrane should favor the release of small proteins such as Cyt-c (14.5 kD) over that of the much larger AIF (56 kD 53). Cyt-c is known to be associated with inner membrane christae, in electrostatic interaction with cardiolipin 67. It is tempting to speculate that, in addition to outer MMP, (AIF-induced?) changes in inner membrane physicochemistry such as cardiolipin oxidation 44,68 must occur to allow for full Cyt-c release.
Irrespective of the exact mechanism, our results indicate that, at least in the model studied herein, the mitochondrial release of Cyt-c is subordinate to that of AIF. Thus, microinjection of AIF into the cytoplasm of syncytia suffices to cause Cyt-c release in a caspase-independent fashion (Fig. 7 A), mimicking that of the spontaneous (caspase-independent; Fig. 3) Cyt-c release. Moreover, neutralization of AIF by microinjection of a specific Ab prevents the release of Cyt-c as well as nuclear apoptosis, whereas neutralization of Cyt-c has no effect on the AIF translocation and only impedes nuclear condensation (Fig. 7 B). Taken together, these results delineate the following sequence of events: Bax-mediated/Bcl-2–inhibited MMP → AIF release → Cyt-c release → caspase activation → nuclear apoptosis. Future research will unravel whether this hierarchy is only applicable to syncytial apoptosis or whether it can be inscribed into a more general pathway.
This work establishes that MMP is a critical step in Env-induced syncytial apoptosis. Other proapoptotic proteins encoded by HIV-1 (PR, Tat, and Vpr) also favor MMP. Vpr exerts its proapoptotic effect, at least in part, by binding to the mitochondrial adenine nucleotide translocator, thereby directly inducing MMP 7. The HIV-1 protease (PR) can cleave Bcl-2, thereby abolishing its MMP-inhibitory function 69. Tat reduces the expression of the mitochondrial superoxide dismutase 2 isoenzyme 4,70, which is another endogenous MMP inhibitor 71. Tat may also favor apoptosis by physically interacting with mitochondria 72. It thus emerges that HIV-1 employs several independent strategies to induce MMP and apoptosis via the intrinsic pathway. It remains an ongoing conundrum whether these manifold strategies are designed to cooperate among each other in an additive or synergistic fashion, in the same cell, or whether they rather reflect the capacity of HIV-1 to kill a wide array of distinct cell types.
We are indebted to Dr. Dominique Piatier-Tonneau for constant support and to Drs. Marc Alizon, Dieter Brdiczka, Pierre Charneau, Nicole Israel, and Anu Srinivasen, as well as the National Institutes of Health AIDS Research and Reference Reagent Program, for gifts of reagents.
This work was supported by a special grant by the Ligue Nationale contre le Cancer, as well as by grants from Association Nationale pour la Recherche sur le SIDA, Fondation pour la Recherche Medicale, the European Commission, the Picasso Program (to G. Kroemer), Fundació irsiCaixa (to J.A. Este), FIS 00/0893 (to J. Blanco), and grants GM 60554 and National Institutes of Health CA69381 (to J.C. Reed). K.F. Ferri receives a fellowship from the French Ministry of Science, and E. Jacotot receives an Association Nationale pour la Recherche sur le SIDA fellowship. J. Blanco is a researcher from the Fundació per a la Recerca Biomèdica Germans Trias i Pujol.
Abbreviations used in this paper: aa, amino acid(s); AIF, apoptosis-inducing factor; COX, Cyt-c oxidase; CXCR, CXC chemokine receptor; Cyt-c, cytochrome c; ΔΨm, mitochondrial transmembrane potential; Env, envelope glycoprotein complex; gp, glycoprotein; hsp, heat shock protein; MMP, mitochondrial membrane permeabilization; Neo, neomycin.