Oncosis is associated with exposure of phosphatidylserine residues on the outside layer of the plasma membrane: A reconsideration of the specificity of the annexin V/propidium iodide assay

Authors

  • Hervé Lecoeur,

    1. Unité d'Oncologie Virale, Département SIDA et Rétrovirus, Institut Pasteur, Paris, France
    Search for more papers by this author
  • Marie-Christine Prévost,

    1. Unité d'Oncologie Virale, Département SIDA et Rétrovirus, Institut Pasteur, Paris, France
    Search for more papers by this author
  • Marie-Lise Gougeon

    Corresponding author
    1. Unité d'Oncologie Virale, Département SIDA et Rétrovirus, Institut Pasteur, Paris, France
    • Unite d'Oncologie Virale, Département SIDA et Retrovirus, Institut Pasteur, 28 Rue du Dr. Roux, 75724 Paris cedex 15, France
    Search for more papers by this author

Abstract

Background

Following a lethal injury, two modes of cell death can be distinguished, apoptosis and primary necrosis. Cells pass through a prelethal stage characterized by a preservation of membrane integrity, in which they shrink (apoptosis) or swell (oncosis, the early phase of primary necrosis). During apoptosis, a loss of phospholipid asymmetry leads to exposure of phosphatidylserine (PS) residues on the outer leaflet of the plasma membrane. We examined whether the external PS exposure, initially supposed to be specific for apoptosis, was also observed in oncotic cells.

Methods

Human peripheral lymphocytes, Jurkat T cells, U937 cells, or HeLa cells were submitted to either apoptotic or oncotic stimuli. PS external exposure was assessed after binding of FITC-conjugated annexin V as was the loss of membrane integrity after propidium iodide (PI) uptake. Morphological examination was performed by optical or electron microscopy.

Results

Similarly to apoptotic cells, oncotic cells expose external PS residues while preserving membrane integrity. Consequently, oncotic cells exhibit the annexin V+ PI- phenotype, previously considered to be specific for apoptotic cells.

Conclusions

This study concludes that the annexin V/PI assay does not discriminate between apoptosis and oncosis and that it can be a useful tool to study oncosis by flow cytometry. Cytometry 44:65–72, 2001. © 2001 Wiley-Liss, Inc.

Following a lethal injury, two modes of cell death can be distinguished, apoptosis and primary necrosis. It has been generally accepted that apoptosis and primary necrosis are two distinct, mutually exclusive, types of cell death. Apoptosis, frequently referred to as programmed cell death, is an active and physiological process. The cell executes the genetic program of its own demise, which can be initiated by an internal clock or by extracellular agents such as hormones, cytokines, killer cells, chemicals, and viral agents. In contrast, primary necrosis is a passive, catabolic, and degenerative process. It generally represents a response to gross injury and can be induced by an overdose of cytotoxic agents (1–3). Oversimplifications regarding the definition of apoptosis and necrosis have led to confusion. Majno and Joris (3) explained that cells committed to cell death pass through a prelethal process in which they swell or shrink. They proposed the word oncosis to define the early stage of primary necrosis during which cells swell. The term necrosis covers the changes that occur after membrane disruption, which could be applied to cells dying via oncosis or apoptosis (3, 4).

Oncosis occurs in response to nonphysiological severe injuries, such as oxidative stress, inhibitors of ATP synthesis, or heat shock (5–7). During oncosis, cells and mitochondria undergo rapid swelling, with dilution of the cytosol, but with preservation of membrane integrity, formation of cytoplasmic blebs at the plasmalemmal surface, and nuclear chromatin clumping. Rupture of the plasma membrane leads to the necrotic stage (oncotic necrosis), which is associated with the release of proteolytic enzymes responsible for local inflammatory responses and tissue damage (3, 4). Oncosis is also involved in natural killer (NK)-mediated antibody-dependent cell cytotoxicity (8). During cytolysis, oncosis is induced by Entamoeba histolytica (9). In contrast, apoptosis is characterized by cell shrinkage with densification of the cytosol, altered mitochondrial structure, and decreased mitochondrial transmembrane potential (10). It is also associated with activation of tissue transglutaminase, which cross-links cytoplasmic proteins (11), loss of asymmetry of phospholipids on the plasma membrane (12), and condensation of nuclear chromatin (13). During the late stage, apoptotic bodies are emitted and phagocytosed by neighboring cells (12). Apoptosis plays a fundamental role in intrathymic selection of T cells and is involved in tissue and organ development. Apoptosis has also been observed in cells infected with Chlamydia psittaci (14) and it represents a marker of disease progression in blood T lymphocytes from human immunodeficiency virus (HIV)-infected subjects (15).

Flow cytometry has become the method of choice for the analysis of apoptosis in a variety of cell systems (1, 16). Multiparameter analysis combines immunocytochemical detection of individual proteins with apoptotic markers and DNA analysis. It is a powerful approach to characterize cells committed to apoptosis and to define the components of the cell death machinery (1, 10, 16–18). In contrast to apoptosis, which was extensively studied, little is known about the flow cytometric features of oncotic cells. In this study, we examine whether an early event characteristic of the apoptotic process (i.e., the exposure of phosphatidylserine [PS] residues on the outside of the plasma membrane; 12,19) also occurs during oncosis. We used the anticoagulant annexin V, which preferentially binds to PS residues on the surface of apoptotic cells prior to the loss of integrity of the plasma membrane. Combined staining with fluorescein-conjugated annexin V and propidium iodide (PI) distinguishes between early (annexin V+/PI-) and late apoptotic cells (annexin V+/PI+; 19,20). A recent study has shown that annexin V+/PI- murine thymocytes do not always correspond to apoptotic cells (21). In this study, we examine whether this phenotype may correspond to oncotic cells. Primary necrosis or apoptosis was induced in several cell types, including human peripheral T lymphocytes, Jurkat T cells, U937 monocytic cells, or HeLa cells, and PS residue translocation was analyzed with the annexin V/PI assay. This study reveals that PS exposure on the outside layer of the plasma membrane is not specific to apoptotic cells; it also occurs in oncotic cells, thus significantly interfering in the detection of apoptotic cells in the annexin V/PI assay.

MATERIALS AND METHODS

Induction of Apoptosis and Oncosis in Cell Lines and Peripheral Blood Mononuclear Cells (PBMCs)

Jurkat and U937 cells were cultured at 5 × 105 cells/ml in RPMI medium (Gibco, Paisley, UK) supplemented with 10% heat-inactivated fetal calf serum (Institut Jacques Boy, Reims, France), 10 IU/ml penicillin, 10 μg/ml streptomycin, 10 mM HEPES, and 1 mM L-glutamine (Sigma, St. Louis, MO; complete RPMI) at 37°C in a 5% CO2 humidified atmosphere. HeLa cells were cultured in complete Dulbecco's modified Eagle's medium (DMEM; Gibco) in 75-ml flasks and submitted to different stimuli at 75% of confluence. PBMCs from human healthy volunteers were isolated from heparinized blood samples after centrifugation on a Ficoll-Hypaque density gradient (Pharmacia, Uppsala, Sweden). They were washed and cultured overnight in complete RPMI at 1 × 106 cells/ml. Apoptosis was induced in cell lines by cycloheximide (CHX 1.5 or 10 μg/ml; Sigma), actinomycin D (AD 10 μg/ml; Sigma), and coated CD95 antibodies (coating for 1 h at 4°C, 50 μg/ml UB2; Immunotech, Villepinte, France). Apoptosis was also induced in PBMCs after overnight stimulation with a combination of phorbol myristate acetate (PMA 50 ng/ml; Sigma) and ionomycin (Iono 250 ng/ml; Sigma) as reported by Ledru et al. (22). Primary necrosis was induced by hyperthermia at 56°C (7), hypotonic shock in distilled water, or it was induced following treatment with 0.005% Triton X-100 (Sigma), 1% sodium azide (NaN3, Sigma; 6), or various concentrations of hydrogen peroxide (H2O2; Prolabo, Fontenay-sous-Bois, France; 5). Cell viability was analyzed following a 30-s staining with trypan blue (1:2 dilution of the TB stock solution from Sigma), and optical microscopy observation.

Analysis of Cell Morphology by Optical and Transmission Electron Microscopy

Optical microscopy studies were performed following eosin-methylene blue staining; Cells were fixed in 1% formalin (Sigma) for 10 min at room temperature (RT), deposed on slides, and submitted for 5 s to the fixative, 1 min to eosin solution, and 5 s to methylene blue solution, all provided by the RAL 555 Kit (MBL, Rhône-Poulenc, Lyon, France). Slides were examined with a Nikon microphot-FXA optical microscope at ×400 and ×1,000. This procedure was also applied on HeLa cells attached to the bottom of flasks.

Ultrastructural studies were performed by transmission electron microscopy. 1 × 106 apoptotic or oncotic cells were submitted to fixation with 2.5% glutaraldehyde in Soërensen buffer (phosphate 0.1 M, pH = 7.4). The cells were further dehydrated in a series of ethanol solutions (30–100%). Finally, cells were embedded in Epoxy. Sections were performed with a Reichter-Jung ultramicrotome before examination. Cells were examined at ×3,000 and ×5,000 with a Jeol JEM 1200 EX electron microscope. A minimum of 100 dead cells were counted to determine the relative proportions of oncotic and apoptotic cells in these samples.

Flow Cytometric Analysis of External PS Exposure With the Annexin V/PI Assay

5 × 105 cells were double stained with FITC-conjugated annexin V and PI for 15 min at 20°C in a Ca2+-enriched binding buffer (apoptosis detection kit, R&D Systems, Abingdon, UK; 20). They were immediately analyzed on the flow cytometer in their staining solution. Annexin V and PI emissions were detected in the FL-1 (band pass 530 nm, band width 30 nm) and FL-2 (band pass 585 nm, band width 42 nm) channels, respectively. Experiments on adherent HeLa cells required prior cell detachment with 0.5 M EDTA for 2 min at RT. After one washing in phosphate-buffered saline (PBS), HeLa cells were stained as mentioned above. To assess the specificity of annexin V binding to PS residues, inhibition experiments were performed with soluble Lα-PS (Sigma). Cells were incubated for 15 min in the staining solution including 5 μg/ml Lα-PS. Inhibition of annexin V binding was assessed by a decreased mean fluorescence intensity (MFI) of annexin V staining.

Flow cytometric studies were performed with a FACSCalibur cytometer, equipped with a 488-nm argon laser (BDIS, Becton Dickinson, San Jose, CA). For each sample, data from 20,000 cells were recorded in list mode on logarithmic scales. Analysis was performed with the Cell Quest software (BDIS) on cells characterized by their forward/side scatter (FSC/SSC) parameters. Cells analyzed included living cells with normal FSC/SSC parameters and dying cells with altered FSC/SSC. Cell debris characterized by a low FSC/SSC and an annexin V- PI- phenotype were excluded from analysis. Positioning of quadrants on annexin V/PI dot plots was performed as reported (19).

RESULTS

PS Exposure on the Outside of the Plasma Membrane Is Detected During Oncosis of Jurkat Cells

Staining cells with a combination of annexin V and PI allows the distinction among living cells (annexin V-/PI-), apoptotic cells (annexin V+/PI-), and apoptotic necrotic cells (annexin V+/PI+; 20; Fig. 1B). Because plasma membrane integrity is preserved during the initial stages of both apoptosis and oncosis (3), we examined whether PS exposure on the outer layer of the plasma membrane, a feature of apoptosis, was also detected during oncosis. Jurkat T cells were submitted for 16 h to either oncotic (1% NaN3) or apoptotic (1.5 μg/ml CHX) stimuli. NaN3 is a mitochondrial poison that blocks ATP production at the Complex IV of the respiratory chain. CHX is a protein synthesis inhibitor that probably blocks the synthesis of short-lived protective proteins that restrain a pre-existent death machinery (23). In control cultures, cells were incubated in medium. The type of death induced by these stimuli was determined by electron microscopy, in parallel to the annexin V/PI flow cytometric assay (Fig. 1). Control Jurkat cells showed the morphology of living cells, i.e., a round shape, a loose heterochromatin, and a fine cytoplasmic matrix. Annexin V/PI double staining indicated a very low level of cell death because 91.7% of the cells were not stained (subset 1, annexin V-/PI-), 1.2% were apoptotic cells exposing external PS residues (subset 2, annexin V+/PI-), 4.6% were apoptotic necrotic cells with loss of membrane integrity (subset 3, annexin V+/PI+), and 2.5% were late dead cells (annexin V-/PI+; Fig. 1B). Short-term incubation with CHX induced a cell shrinkage characteristic of apoptotic death, which was associated with DNA condensation, an electron-dense nucleus, and a granular cytoplasm. Dual staining with annexin-V/PI revealed an increased proportion of single or double-stained cells compared with control cultures, because 21.8% of the cells were apoptotic (subset 2), 22.5% were apoptotic necrotic (subset 3), and 4.2% were late dead cells (subset 4). Both apoptotic and apoptotic necrotic cells were also observed when Jurkat cells were incubated with CD95 (anti-Fas) monoclonal antibodies (Fig. 1C). On the contrary, 1% NaN3 treatment induced cell swelling, nuclear dissolution (karyolysis), and cytoplasmic breakdown (Fig. 1A). Strikingly, 9.1% of these cells were able to bind annexin V without incorporating PI (Fig. 1B). Because the majority (94%) of the dead cells showed morphological features of primary necrosis, we propose that subset 2 corresponds to oncotic cells described by Majno and Joris (3), a prelethal subset that preserves membrane integrity. Staining of oncotic cells with annexin V indicates that PS exposure also occurs in the early phase of primary necrosis. The annexin V/PI assay also indicated that 36.1% of the cells were oncotic necrotic (subset 3). Similar results were obtained when Jurkat cells were submitted to hyperthermia, hypotonic shock in distilled water, or Triton X-100 treatment (Fig. 1C). Thus, these data indicate that the annexin V/PI assay not only detects apoptotic and apoptotic necrotic cells, but also oncotic and oncotic necrotic cells.

Figure 1.

Multiparametric analysis of Jurkat cells cultured for 12 h in medium, in CHX, or in NaN3. A: Electron microscopy analysis. B: Annexin V/PI double staining of cultured cells. The percentages of the four subsets are indicated. C: Jurkat cells were submitted to the indicated stimuli and stained with annexin V/PI. The annexinV/PI phenotype was determined as in B. The type of death was determined by morphological examination. D: MFI of annexin V staining of Jurkat cells treated with apoptotic stimulus (CHX; white boxes) or oncotic stimulus (NaN3; grey boxes). Boxes and whisker plots represent the median MFI, the range, and the 25th and 75th percentiles obtained for 11 experiments. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

The MFI of annexin V staining was compared between apoptotic and oncotic cells. Eleven experiments were performed with Jurkat cells treated with either CHX or NaN3. Annexin V MFI of oncotic cells was not significantly different from that of apoptotic cells (annexin V+/PI-, Fig. 1D). Similarly, annexin V MFI was comparable in apoptotic necrotic and oncotic necrotic cells (annexin V+/PI+, Fig. 1D). Therefore, this cytometric assay cannot discriminate between apoptosis and primary necrosis.

PS Exposure During Oncosis of Peripheral Lymphocytes Induced by NaN3 or Hyperthermia

Freshly isolated PBMCs from human healthy donors were incubated overnight with either PMA-Iono (22) or 1% NaN3 (6). Cell death was assessed by morphological examination following eosin-methylene blue staining. Figure 2A shows the morphology of living cells following a short-term culture in medium, with restricted cytoplasm and centered nucleus. Treatment of lymphocytes with NaN3 induced a polarized alteration, characterized by a dramatic plasma membrane swelling on one side of the cell, the nucleus being located at the opposite side. Annexin V/PI staining showed that 49.3% of the lymphocytes were dying and, as observed for Jurkat cells, a significant proportion of cells were annexin V+/PI- (oncotic) cells. On the other hand, PMA-Iono stimulation of PBMCs induced apoptotic cell death but very similar dot plots were found following annexin V/PI staining, confirming that this assay does not distinguish apoptotic from oncotic cells (Fig. 2B). The specificity of annexin V binding to PS residues on oncotic cells was assessed by inhibiting annexin V binding on heat-treated lymphocytes with 5 μg/ml LαPS. Figure 2C shows that the addition of soluble Lα-PS completely blocked annexin V binding on lymphocytes induced to oncosis, confirming PS exposure on oncotic lymphocytes.

Figure 2.

Oncosis and apoptosis in human PBMCs. A: PBMCs were incubated in medium, 1% NaN3, or they were stimulated for 16 h with PMA-Iono. Morphological examination (×1,000) was performed after staining with eosin-methylene blue. B: Annexin V/PI costaining was performed on the same samples. C: Hyperthermia-treated lymphocytes were either preincubated with 5 μg/ml of Lα-PS before staining with annexin V (white histogram) or they were directly stained with annexin V (;6>black histogram). Results from one representative experiment are shown. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

To confirm that annexin V+PI- cells correspond to oncotic cells, time-course experiments were performed on peripheral lymphocytes submitted to hyperthermia at 56°C (7). Cell death was analyzed concomitantly by flow cytometry (FSC/SSC criteria and annexin V/PI assay) and optical microscopic examination. Figure 3 shows that following 3 min of heat shock, 45% of the lymphocytes were stained with annexin V but remained PI- (Figs. 3A,B). In this sample, about one half of the cells showed the morphology of oncotic cells (round swollen cells with a preserved membrane integrity) and the other half showed features of living cells. Figure 3C shows a representative oncotic cell. In parallel, the cell swelling was detected by flow cytometry: the mean FSC intensity increased from 248 in living cells to 300 in 3 min in heat-shocked cells and SSC increased from 237 to 313, respectively (Fig. 3D). Following 30 min of heat shock, 81% (Fig. 3A) of the cells were annexin V+/PI+ (Fig. 3B). Light microscopy examination revealed that 94% of these cells had the morphology of oncotic necrotic cells (very important swelling) with membrane disruption and release of cytoplasmic content (Fig. 3C). In addition, these cells showed a larger increase in FSC (363 in oncotic necrotic cells versus 300 in oncotic cells), whereas SSC decreased in these cells (250 in oncotic necrotic cells versus 313 in oncotic cells; Fig. 3D). However, at later time points, FSC rapidly decreased to low values, probably resulting from the drastic decrease of the cellular content and the cellular refractive index. These data confirm that, under these experimental conditions, the annexin V+/PI- cells correspond to oncotic cells, which become rapidly oncotic necrotic cells with a disrupted membrane and the annexin V+/PI+ phenotype.

Figure 3.

Characteristics of the annexin V+/PI- and annexin V+/PI+ subsets in hyperthermia-treated PBMCs. A: Kinetics of appearance of the two indicated subsets among total lymphocytes following 56°C heat shock, defined by the annexinV/PI assay. B: Annexin V/PI assay performed on freshly isolated PBMCs, or following 3 or 30 min of hyperthermia. C: Morphological features of the same samples, determined after eosin-methylene blue staining. D: Histograms of FSC/SSC parameters of PBMCs analyzed ex vivo (1), after 3 min of hyperthermia (2), or after 30 min of hyperthermia (3). Mean FSC and SSC values are indicated. In FSC histograms 2 and 3, the dotted line represents ex vivo lymphocytes. Results from one representative experiment are shown. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

PS Exposure During Oncosis of Adherent HeLa Cells Induced by Hypotonic Shock

HeLa cells were cultured for 24 h in medium, in the presence of 10 μg/ml CHX, or they were submitted to hypotonic shock in distilled water for 10 min. Cell examination was performed following eosin-methylene blue staining (Fig. 4A). Cells treated with CHX showed features of apoptotic cells (cell shrinkage, blebs, cell dehydration detected by an extreme violet color, indicated by grey arrows). Cells submitted to hypotonic shock showed features of oncotic cells (cell swelling and enlarged nucleus). In the early stage of primary necrosis, membrane integrity was preserved and the cytoplasm was still present in a large fraction of cells (white arrow, Fig. 4A). Oncotic necrotic cells were characterized by cytoplasm leakage out of the cell and by nuclear dissolution (black arrow, Fig. 4A). The annexin V/PI assay indicated that following incubation in medium, 25.2% of the cells were apoptotic (annexin V+/PI-), whereas only 3% were apoptotic necrotic (annexin V+/PI+). Following CHX treatment, 55% of HeLa cells were dead, including 16.8% of apoptotic cells (annexin V+/PI-) and 38.3% of apoptotic necrotic cells. Hypotonic shock induced a significant fraction (24.3%) of HeLa cells to oncosis (annexin V+/PI-), the rest of the population having reached the stage of oncotic necrosis (Fig. 4B). Therefore, these data indicate that PS residues are also exposed during primary necrosis of adherent cells and that the annexin V/PI assay does not discriminate between apoptosis and oncosis.

Figure 4.

Oncosis and apoptosis in adherent HeLa cells. A: HeLa cells were cultured in medium, in the presence of 10 μg/ml CHX for 24 h (apoptosis), or in distilled water for 10 min (oncosis). Morphological examination of eosin-methylene blue HeLa-stained cells (×400). B: Annexin V/PI dual staining of the same samples.

PS Exposure During Oncosis of Monocytic U937 Cells Induced by NaN3

Apoptosis was induced in monocytic U937 cells by a 16-h treatment with 10 μg/ml of AD, which inhibits RNA transcription by forming a stable complex with double-stranded DNA. Oncosis was induced by a 16-h incubation in 1.5% NaN3. Cell examination was performed following eosin-methylene blue staining (Fig. 5A, panels 1,3,5) or trypan blue uptake (Fig. 5A, panels 2,4,6). Following incubation in medium, most of the cells had the morphology of living cells and 71.0% were not permeable to trypan blue (Fig. 5A, panel 1). Annexin V/PI dual staining indicated that 13% of the dying cells were apoptotic and 2.8% were apoptotic necrotic (Fig. 5B). AD induced apoptosis in U937 cells, showing the typical morphology of blebbing cells. Permeability to trypan blue was detected in 67% of these cells (Fig. 5A, panels 3 and 4). The annexin V/PI assay revealed that the majority (68%) of U937 cells were apoptotic (annexin V+/PI-) and nearly no cells were apoptotic necrotic (Fig. 5B). NaN3 induced cell death in more than 90% of the cells (evaluated by trypan blue staining), which showed oncotic features including cell swelling, plasma membrane disruption, and cytoplasm release (Fig. 5A, panels 5 and 6). The annexin V-PI assay indicated that 89% of the cells were dying, including 61.8% of annexin V+/PI- cells and 27.7% of annexin V+/PI+ cells (Fig. 5B).

Figure 5.

Oncosis and apoptosis in monocytic U937 cells. A: U937 cells were incubated in medium, in 10 μg/ml AD, or in 1.5% NaN3. Morphological examination following eosin-methylene blue staining (panels 1,3,5) or trypan blue staining (panels 2,4,6). B: Annexin V/PI dual staining of the same samples of U937 cells as in C. C: U937 cells were treated with various concentrations of H2O2 for 24 h. The percentage of oncotic + oncotic necrotic (plain line) and apoptotic + apoptotic necrotic (dotted line) cells was determined after trypan blue staining and morphological examination. Annexin V/PI dual staining was performed (no H2O2, 36 μM, or 60 μM H2O2) and the corresponding dot plots are shown. Results from one representative experiment are shown.;7>

Given the lack of specificity of the annexin V/PI assay for detecting apoptosis, it was important to determine to what extent oncotic cell death could interfere with apoptosis quantification in cell samples including both types of death. U937 cells were submited to a 16-h oxidative stress induced by increasing concentrations of H2O2 (ranging from 0 to 900 μM) and were analyzed both morphologically and following annexin V/PI staining. Incubation in medium alone induced spontaneous death in 21% of the cells. Morphological analysis revealed that the majority of dying cells had the features of apoptotic cells, whereas the rest had the features of oncotic cells (Fig. 5C). Morphological examination also revealed that treatment with 9 μM H2O2 only induced apoptosis and that concentrations from 9 to 36 μM induced both apoptosis (dotted line) and oncosis (bold line; Fig. 5C). With concentrations ranging from 60 to 900 μM H2O2, almost all the cells died and the majority showed the features of oncosis and necrosis (Fig. 5C). Interestingly, this progressive change in the proportions of apoptotic/oncotic cells could not be detected with the annexin V/PI double staining (lower panel, Fig. 5C). When 60 μM H2O2 was added, annexin V-PI staining indicated 68.9% of annexin V+ dead cells, whereas morphological examination revealed that only 14% of them were apoptotic. These data indicate that oncosis can significantly interfere with apoptosis detection when both types of cell death coexist.

DISCUSSION

Cell death is usually classified into two broad categories: apoptosis and primary necrosis. The term necrosis may be misleading because it corresponds to the late stage of death associated with membrane disruption, whether cells are dying of apoptosis or oncosis (3). Apoptotic cells exhibit an early loss of phospholipid asymmetry, leading to the exposure of PS residues on the outer layer of the plasma membrane (12). Because annexin V binds to negatively charged phospholipids such as PS, FITC-conjugated annexin V is classically used to identify apoptotic cells by flow cytometry (20). During apoptosis, the cells become reactive to annexin V prior to the loss of both plasma membrane integrity and the ability to exclude PI. By staining the cells with a combination of annexin V and PI, both apoptotic cells (annexin V+/PI-) and apoptotic necrotic cells (annexin V+/PI+) are detected. The annexin V+/PI- phenotype specifically identified apoptotic cells. In our study, we report that annexin V also binds to oncotic cells, which appear with the annexin V+/PI- phenotype in a variety of cell systems such as human peripheral blood lymphocytes, the Jurkat T-cell line, the U937 monocytic line and the cervical carcinoma HeLa cells. Cells are induced to oncosis by several stimuli such as hyperthermia, NaN3, or hypotonic shock. We have confirmed by inhibition experiments that annexin V binding was due to the specific exposure of PS residues on oncotic cells. Consequently, the criteria of annexin V binding on dying cells can no longer be considered as specific for apoptosis.

Many authors use the annexinV/PI assay for apoptosis detection and quantitation but usually do not confirm the type of cell death by morphological examination. We have shown that the presence of oncotic cells within an apoptotic population could significantly interfere with the quantification of apoptosis by the annexin V/PI assay. Both oncosis and apoptosis can coexist in response to certain stimuli. For example, although apoptosis can be induced by low doses of H202, and oncosis by high doses, both types of cell death are induced by intermediate concentrations of H202 (this report). The oncotic threshold, determined by the cell's inability to sustain ion homeostasis and ATP production in response to the stimulus, varies according to the cell type, stage of cell maturation, and the type of stimulus. A progressive switch from apoptosis to oncosis has been observed in vivo in the brain following hypoxic-ischemic injuries (24, 25) as well as in vitro in leukemia ML-1a cells following treatment with xanthine oxidase (26), in Molt-4 cells following irradiation (27), and in U937 cells submitted to oxidative stress (our study). We highlight the limits of the annexin V/PI apoptosis assay and recommend parallel morphological examination by electron or light microscopy. Morphological examination can be easily performed following eosin-methylene blue coloration, May-Grünwald-Giemsa, or trypan blue stainings. The recent development of laser scanning cytometers (LSC), which combine cytometry and image analysis (28), will be of particular interest for apoptosis studies and will help to discriminate between apoptosis and oncosis (17).

This assay may help in studying oncosis by flow cytometry.We performed preliminary experiments combining annexin V and CMX-Ros to study modifications of mitochondrial transmembrane potential (Δψm) in oncotic lymphocytes. We showed that altered Δψm could be detected in oncotic lymphocytes, supporting the hypothesis that apoptosis and oncosis may represent a continuum (23, 29). Further experiments are being undertaken to determine whether other events involved in apoptosis, such as the activation of caspases, are also involved in oncosis. This will better characterize the two types of cell death and will help to identify the specific flow cytometric markers of oncosis.

Acknowledgements

We are indebted to Marie-Christine Cumont, Christine Schmitt, Denise Guétard, Bruno Hurtrel, and Pr. Luc Montagnier for their support. Supported by grants from the Agence Nationnale de Recherche sur le SIDA (ANRS) and from the Fondation pour la Recherche Médicale (Sidaction) to M.L.G.

Ancillary