Distinct Modes of Cell Death Induced by Different Reactive Oxygen Species

Oxidants derived from inflammatory phagocytes compose a key element of the host immune defense system and can kill mammalian cells by one of several different mechanisms. In this report, we compare mechanisms of cell death induced in human B lymphoma cells by the inflammatory oxidants superoxide, H2O2, and HOCl. The results indicate that the mode of cell death induced depends on the nature of the oxidant involved and the medium in which the cells are treated. When human Burkitt's lymphoma cells are exposed to superoxide anion, generated as a flux from xanthine and xanthine oxidase, the cells die by a non-apoptotic mechanism (pyknosis/necrosis) identical to that seen when cells are treated with a bolus of reagent H2O2. Addition of superoxide dismutase has no effect, whereas catalase is completely protective, indicating that exogenously generated superoxide kills cells entirely through its dismutation into H2O2. In contrast, cells treated in culture media with reagent HOCl die largely by apoptosis. HOCl-induced apoptosis is mediated by aminoacyl chloramines generated in the culture media and can be mimicked by treatment of cells with taurine chloramine or with long lived chloramines generated from modified Lys or Arg. The results suggest that in a physiological milieu in which O 2 − and H2O2 are the main oxidants being formed, the principal form of cell death may be necrotic, and under inflammatory conditions in which HOCl is generated, apoptotic cell death may predominate.

numerous different diseases (4). The primary oxidants generated by all normal phagocytes are the superoxide anion (O 2 Ϫ ) and hydrogen peroxide (H 2 O 2 ), which can, in the presence of reducing metals, go on to form hydroxyl and other free radicals. In addition, monocytes and neutrophils contain high levels of the enzyme myeloperoxidase which catalyzes formation of the potent oxidant, hypochlorous acid (HOCl). HOCl can then go on to react with extracellular amino acids to generate chloramines, which maintain some of the oxidizing potential of HOCl but are not as potent or as broadly reactive (5)(6)(7). Each of these different ROS have molecular characteristics that account for their different levels of reactivity with cellular and extracellular macromolecules.
The studies presented here focus on mechanisms of cell death induced by inflammatory oxidants. Mammalian cells can die by one of several different defined pathways, the most common of which are apoptosis and necrosis. Apoptosis is characterized by a discrete set of biochemical steps and morphological changes including activation of caspases, translocation of phosphatidylserine from the inner to the outer layer of the plasma membrane, chromatin condensation, and fragmentation into apoptotic bodies (8,9). In contrast, cells that die by necrosis swell and then lyse, releasing their contents into the extracellular space (10). It is thought that death by apoptosis is physiologically advantageous because early apoptotic cells are cleared by phagocytosis and subsequent intracellular degradation (11). In this manner, apoptotic cells are removed without causing damage to the surrounding tissue. In contrast, necrotic cells are thought to promote an inflammatory response caused by the leakage of intracellular proteins and nucleic acids prior to phagocytosis. In support of this theory, we found that B lymphoma cells treated with H 2 O 2 die by a non-apoptotic mechanism and are not phagocytosed by macrophages until after they begin to lose their plasma membrane integrity (12). In contrast, early apoptotic cells induced by chemotherapy drugs underwent phagocytosis while their membranes were still intact.
Previous research (13)(14)(15)(16)(17)(18)(19)(20)(21) has demonstrated that exogenously added H 2 O 2 can induce either apoptosis or necrosis, depending on the concentration of H 2 O 2 , the cell type being studied, and the level of ATP in the cells. Relatively little has been reported on mechanisms of induction of cell death from exogenously generated O 2 Ϫ or HOCl (22), such as would be produced by activated neutrophils, and we are unaware of any study that does a direct comparison of the modes of cell death induced by the different oxidants in the same cell type. This report compares mechanisms of B lymphoma cell death induced by exogenous O 2 Ϫ , H 2 O 2 , and HOCl. The data reveal that O 2 Ϫ kills cells solely through its dismutation into H 2 O 2 , and both of these ROS kill by a non-apoptotic mechanism referred to as pyknosis/necrosis (20). In stark contrast, HOCl induces either necrosis or apoptosis depending on the cell environment; in buffered saline, cell death is entirely by rapid necrosis, and in growth media, cell death is primarily apoptotic. We demonstrate further that long lived aminoacyl chloramines mediate HOCl-induced apoptosis.

MATERIALS AND METHODS
Cells and Treatments-The Burkitt's lymphoma cell lines JLP-119 and BL-41 were grown in RPMI 1640 medium containing 10% heatinactivated fetal calf serum, 2 mM L-Gln, and 50 M ␤-mercaptoethanol at 37°C in 5% CO 2 in air as described previously (15). Exponentially growing cells were harvested by centrifugation and resuspended in fresh media to achieve a culture density of 5 ϫ 10 5 cells/ml. Reagent H 2 O 2 (50 -200 M) was added to the cell suspensions, and the cells were incubated at 37°C for the times indicated in the figure legends. A flux of O 2 Ϫ and H 2 O 2 was generated by adding xanthine (XA, 50 -400 M) to the cells in culture media 1 h prior to addition of xanthine oxidase (XO, 20 milliunits; Sigma catalog number X-4500), and cells were incubated for the times indicated in the figure legends. Reagent NaOCl (1-500 M; Aldrich) or H 2 O 2 was added to the cell suspensions, and the cells were incubated for the times indicated in the figure legends. Concentrations of stock reagents were based on molar extinction coefficients of 350 M Ϫ1 cm Ϫ1 for NaOCl at 292 nm (23) and of 50 M Ϫ1 cm Ϫ1 for H 2 O 2 at 240 nm (24). Chloramine toxicity studies were carried out by mixing each amino acid (2 mM in PBS on ice) with NaOCl (50 -500 M) and then adding the mixtures (1 ml) immediately (Ͻ30 s) to cell pellets to get a final density of 5 ϫ 10 5 cells/ml. After a 1-h incubation at 37°C, 1 ml of complete media was added to each well, and cells were incubated for an additional 7 h.
Morphological Assessment of Cell Death Using Hoechst/Propidium (PI) Nuclear Staining and Fluorescence Microscopy-Cells (5 ϫ 10 5 cells/ml) were incubated for 15 min at 37°C with Hoechst 33342 dye (5 g/ml in PBS), centrifuged, washed once in PBS, and then resuspended at ϳ2.5 ϫ 10 7 cells/ml. Propidium iodide (PI; 50 g/ml from a 1 mg/ml stock in PBS) was added just before microscopy. Cells were visualized using fluorescence microscopy as described previously (20). A minimum of 200 cells was counted, and cell morphology was classified as follows: (i) live cells (normal nuclei, blue chromatin with organized structure); (ii) membrane-intact apoptotic cells (bright blue chromatin which is highly condensed, marginated, or fragmented); (iii) membrane-permeable apoptotic cells (bright red chromatin, highly condensed or fragmented); (iv) necrotic cells (red, enlarged nuclei with smooth normal structure); and (v) pyknotic/necrotic cells (dense, red, slightly condensed nuclei with no fragmentation).
Assessment of Apoptosis Using FACScan Analysis-To determine the percentage of cells expressing phosphatidylserine (PS) on the exofacial surface of the plasma membrane, cells were centrifuged and resuspended in FACS buffer containing 1.25 g/ml annexin V-FITC (PharMingen) and 0.1 mg/ml PI in 140 mM NaCl, 2.5 mM CaCl 2 , 10 mM HEPES, pH 7.4, and incubated for 15 min at room temperature. Cells (10,000 per sample) were then analyzed on a FACScan (BD Biosciences) using the CELLQUEST flow cytometric analysis software. Cells in the lower right dot plot quadrant (PS-positive and PI-negative) were reported as apoptotic and have intact plasma membranes, whereas cells in the upper right dot plot quadrant (PS-positive and PI-positive) were scored as necrotic and have leaky membranes. In theory, cells in this latter quadrant can also be late apoptotic but were confirmed to be necrotic in our experiments using fluorescence microscopy as described above. In control experiments, we found that the annexin V FACS assay gave comparable results to those obtained by fluorescence microscopy using Hoechst/PI.
Western Blot Analysis for Apoptosis-Cells were harvested by centrifugation, washed twice with PBS, resuspended in lysis buffer containing 62.5 mM Tris, pH 6.8, 2% SDS, 10% glycerol, and 1 mM diethylenetriaminepentaacetic acid, and heated for 10 min at 100°C. The total cell lysates (50 g/lane) were subjected to SDS-PAGE (25), transferred to nitrocellulose membranes, blocked with 5% milk, and incubated with mouse monoclonal anti-human CPP32 (caspase-3) followed by horseradish peroxidase-conjugated secondary antibody as described previously (21). Bands were visualized by chemiluminescence using the ECL kit from PerkinElmer Life Sciences.
ATP Assay-Intracellular ATP levels were determined using luciferin-luciferase (26) as described previously (20,21) with minor modifications. Briefly, cells (5 ϫ 10 5 cells/ml) that had been treated with H 2 O 2 or HOCl were collected by centrifugation, pelleted, lysed in 150 l of 3% HClO 4 , and placed on ice for 15 min. The solutions were then neutralized with 75 l of 1 M KOH and 30 l of potassium phosphate buffer (1 M K 2 HPO 4 /KH 2 PO 4 , pH 7.4) and incubated on ice for 15 min. The suspensions were centrifuged at 13,000 rpm for 1 min to remove the salt precipitate. Supernatants were collected, diluted 1:5 with potassium phosphate buffer (10 mM KH 2 PO 4 , 4 mM MgSO 4 , pH 7.4), and placed on ice. At the time of the assay, 50-l samples were mixed with 50 l of 50 mM NaAsO 2 , 20 mM MgSO 4 , pH 7.4, in a 96-well plate, followed by addition of 80 g of luciferin/luciferase. Chemiluminescence was quantified in a Dynatech ML 3000 microtiter plate luminometer (Chantilly, VA). ATP standard curves were run in all experiments and were linear in the range of 5-2500 nM. Stock ATP concentrations were quantified from the absorbance at 259 nm using a molar extinction coefficient of 15,400. ATP levels are reported as the % of the ATP levels in control (untreated) cells.
GSH Assay-The 5,5Ј-dithiobis(2-nitrobenzoic acid) colorimetric assay (27) was used for the measurement of total cellular non-protein thiols, which are predominantly GSH. Briefly, cells (2-6 ϫ 10 6 cells) treated with H 2 O 2 or HOCl were collected by centrifugation, washed once with PBS, and resuspended in 300 l of a 2% solution of 5-sulfosalicylic acid for cell lysis and deproteinization. The samples were centrifuged at 12,000 rpm for 5 min, and then 250 l of sample was mixed with 250 l of 5,5Ј-dithiobis(2-nitrobenzoic acid) solution (0.3 M sodium phosphate buffer, 10 mM EDTA, 0.2 mM 5,5Ј-dithiobis(2-nitrobenzoic acid), pH 8.0). After incubation at room temperature for 5 min, the absorbance was read at 412 nm. Standard curves were run in all experiments and were linear in the range of 5-100 M glutathione. GSH levels are reported as the percentage of the GSH levels in control (untreated) cells.
Chloramine Production and Decay Assay-Solutions of individual amino acids (2 mM in PBS) were mixed on ice with 50 -500 M HOCl. The absorbance at 252 nm was read immediately, and this value was used as the control (maximum) value (23). The solutions were incubated at 37°C to allow chloramine decay to occur, taking intermittent measurements of the residual absorbance at 252 nm. Results are reported as the percentage of control value. All amino acids were purchased from Sigma except N-␣-acetyl-Arg, which was from ICN.
Aldehyde Production Assay-The 2,4-dinitrophenylhydrazine colorimetric assay was used for measurement of aldehyde production as described by Stadtman and Berlett (28). Stock solutions of amino acids (2 mM in PBS) were mixed on ice with 500 M HOCl. The solutions were incubated at 37°C to allow for decay of the chloramines into aldehydes. At various time points, 100 l of sample was removed and added to 850 l of 0.02% 2,4-dinitrophenylhydrazine in 2 M HCl and allowed to sit at room temperature for 15 min. The solutions were neutralized by addition of 750 l of 2.5 M KOH and then centrifuged at 3,000 rpm for 5 min. The absorbance of the supernatants at 540 nm was measured.

RESULTS
In previous studies, we found that the mode of B lymphoma cell death induced by H 2 O 2 is non-apoptotic (20). We refer to the cell death induced by H 2 O 2 in these cells as pyknotic/ necrotic because the nuclei are slightly condensed, whereas there are no signs of classical apoptotic changes. Fig. 1  Ϫ has the potential to promote hydroxyl radical formation through the Haber-Weiss reaction, in addition to forming H 2 O 2 . The results in Fig. 2 show that cell death induced by exposure to O 2 Ϫ generated from XO and XA was primarily pyknotic/necrotic at all treatment levels, similar to H 2 O 2 . Addition of superoxide dismutase to the incubation medium had no effect on the type or level of cell death induced, whereas catalase inhibited completely cell death induced by XA/XO. Similar results were obtained with JLP-119 cells (data not shown). The results show that cell death from exogenously generated O 2 Ϫ derives mostly from formation of H 2 O 2 .

Quantification of H 2 O 2 -induced cell death (18 h of incubation) in two different Burkitt's lymphoma (BL) cells lines is shown in
One of the main oxidants generated during an inflammatory response involving neutrophils and/or undifferentiated monocytes is HOCl, formed from the action of myeloperoxidase on H 2 O 2 and Cl Ϫ (2). When we exposed BL cells to reagent HOCl, we found that, in stark contrast to H 2 O 2 and O 2 Ϫ , a high level of apoptosis was induced (Fig. 3). The other main form of cell death induced from HOCl was classical necrosis, with only a low level of pyknosis/necrosis seen (determined from the nuclear morphology). Note that BL-41 cells, which are somewhat more resistant to oxidant-induced killing than JLP-119 cells, incurred a higher level of necrosis compared with JLP-119 cells, which were mostly apoptotic. Apoptosis was confirmed by measuring activation of caspase-3, as determined from cleavage of the caspase-3 pro-enzyme by Western blot immunoassay (Fig. 4) and by measuring annexin V binding to externalized PS (see below).
To determine why H 2 O 2 and HOCl induce such different forms of cell death, we compared the effects of these two oxidants on GSH and ATP levels in the cells. As shown in Fig. 5, the differences cannot be attributed to different effects on cellular GSH levels. Concentrations of H 2 O 2 and HOCl that induce roughly 60 -80% cell death in the cells caused mild to moderate reductions in cellular GSH levels (Յ20% decrease in JLP-119 cells and Յ35% decrease in BL-41 cells), and the effects of H 2 O 2 and HOCl were nearly identical. In contrast, the effects of H 2 O 2 and HOCl on cellular ATP levels were markedly different (Fig. 6). As we saw previously (20), H 2 O 2 caused a complete and irreversible loss of intracellular ATP within 15 min after addition to the cells. However, HOCl caused only a mild and transient drop in ATP levels that was restored to control levels within 1 (BL-41 cells) to 2 h (JLP-119 cells). The ability of HOCl to induce apoptosis correlates with its ability to induce significant cell damage without depleting the cells of ATP and is consistent with reports that ATP must be main-  (20 milliunits) in the presence or absence of superoxide dismutase (SOD) or catalase in complete media. Cells were then incubated for 18 h at 37°C, without washing or exchanging the media, harvested, and stained with Hoechst/PI. Cell death was assessed by nuclear morphology using fluorescence microscopy as described under "Materials and Methods." tained at or above 25% of control levels in order for apoptosis to occur (19 -21, 29).
Previous studies into the mechanism of cytotoxicity induced by HOCl found that the main mode of cell death is necrotic (30), with endothelial cells undergoing only low levels of apoptosis in response to HOCl treatment (22). We noted that those earlier studies were carried out by treating the cells in buffer, not growth media, so we set out to determine whether the medium in which the cells are treated influences the mechanism of cell killing. Note that these and all subsequent experiments were performed with BL-41 cells, and apoptosis was quantified primarily by measuring binding of annexin V followed by FACS analysis to detect cells that have externalized plasma mem-brane phosphatidylserine, a hallmark of apoptosis. BL-41 cells were selected because they express significantly higher levels of exofacial PS than JLP-119 cells. Control studies showed that the annexin V assay gave nearly identical results for the quantification of apoptotic and necrotic cells as the assay for nuclear morphology (see "Materials and Methods" for experimental details). The results in Fig. 7 show that when BL-41 cells are treated with HOCl in PBS/glucose, cell death is entirely by necrosis. In contrast, when the cells are treated either in com- plete media (i.e. containing 10% fetal calf serum) or in RPMI alone, cell death is primarily apoptotic. The data in Fig. 7 also demonstrate that higher levels of HOCl are required to kill the cells in the more complex media, i.e. cell death measured at 8 -10 h is only 70% as high when the cells are treated in complete media (RPMI plus 10% serum) than when they are treated in plain RPMI. Significantly higher cell death is achieved at lower HOCl concentrations when the cells are treated in PBS/glucose, and death appears much more rapidly (within 2 h).
It is widely thought that oxidants and other cytotoxic agents induce apoptosis at low concentrations and necrosis at high concentrations (13). To determine whether HOCl might induce apoptosis in PBS if lower concentrations were employed, we treated BL-41 cells with levels of HOCl ranging from 1 to 500 M, and we measured cell death after 8 h. The cells are relatively unstable in PBS/glucose in the absence of protein and die spontaneously when incubated for longer times. Hence, incubations in PBS were limited to 2 h after which complete media were added, and the cells were incubated for an additional 6 h. As shown in Fig. 8, we did not detect any apoptosis when cells were treated in this manner with low concentrations of HOCl in PBS/glucose.
The results in Fig. 7 demonstrate a profound effect of the media on the mode and level of cell killing by HOCl. RPMI contains several types of components including vitamins, salts, glucose, and high levels of free amino acids (6.8 mM total, which is comparable with the level of amino acids found in human plasma) (31). In initial experiments, we found that of these components, only the amino acids influenced HOCl-induced cell killing; treatment of cells in RPMI vitamins or salts gave results similar to treatment in PBS/glucose. It is well known that HOCl can react with amino acids to form chloramines (5)(6)(7) and that these can be cytotoxic (32). The mode of cell killing by aminoacyl chloramines has not been described previously. The following experiments demonstrate that induction of apoptosis by treatment of BL cells with HOCl in RPMI is mediated by formation of long lived aminoacyl chloramines. Unless otherwise noted, the experiments employed 500 M HOCl, which causes 40 -60% apoptotic cell death when cells are treated in RPMI.
First, spectrophotometric studies showed that when different amino acids (2 mM) were incubated with HOCl (500 M) on ice, they were converted immediately (within seconds) to chloramines, which have a characteristic absorbance peak at 252 nm as described by Test et al. (6) (data not shown). At the same time, the absorbance peak for reagent HOCl at 292 nm was lost immediately. At this ratio of amino acids to HOCl (4:1), there was no detectable HOCl remaining after it was added to the amino acid mixtures.
Chloramines have different levels of stability depending on the chemistry of the modified amino group. The data in Fig. 9A show that most of the aminoacyl chloramines tested were short lived and decomposed in the course of a 30-min incubation at room temperature. These short lived chloramines are known to be derived from reaction of the ␣-amino group with HOCl (5,33,34). In contrast, the chloramine formed from the ␤-amino acid taurine was stable, as expected (23), and did not decompose into an aldehyde even after an overnight incubation (data not shown). The results with Lys and Arg were mixed because they have additional amino groups that can react with HOCl to form chloramines of differing stability. As shown in Fig. 9A, when the ␣-amino group of Lys was blocked by N-acetylation, a stable chloramine was formed on the ⑀-amino group. However, when the ⑀-amino group was blocked, an unstable ␣-chloramine was formed that decomposed into an aldehyde (Fig. 9B). The finding that non-acetylated Lys decayed by more than 50% is consistent with the finding of Hazen et al. (33) that the ␣-amino group is more susceptible to chlorination than the ⑀-amino group. A similar result was obtained with Arg; when the ␣-amino group was blocked with an acetyl moiety, a stable chloramine was formed, whereas when the side chain guanidium group was blocked, a relatively labile chloramine was formed. Note also that the rates of decay of the different ␣-chloramines are not uniform; Glu and Gln decayed extremely rapidly, whereas Ile and His decayed more slowly. Formation of aldehydes from each of the amino acids that showed chloramine decay, as shown for N-⑀-acetyl-Lys in Fig. 9B and as characterized previously (33,34), was confirmed as described under "Materials and Methods" (data not shown).
The results in Fig. 10 show the different modes of cell death induced by treating BL-41 cells with the different aminoacyl chloramines. For these experiments, each amino acid (2 mM in PBS) was mixed on ice with HOCl (100 -500 M) and then added to cells in PBS/glucose. PBS/glucose alone was used as the control. The cells were incubated for 1 h at 37°C followed by addition of an equal volume of complete media and incubation for an additional 7 h. The mode of cell death was assessed by measuring PS externalization and permeability to propidium iodide. The control treatment with PBS/glucose for 1 h followed by incubation with media resulted in roughly 18% cell death, mostly by necrosis. This level of cell death was subtracted from the treatment groups in order to depict only the increase in cell death induced by the various HOCl-modified amino acids. The results demonstrate that treatment of the cells with three long lived chloramines (from taurine, N-␣acetyl-Lys and N-␣-acetyl-Arg) caused cell death primarily through apoptosis. HOCl-modified Lys also caused apoptosis but only when tested at lower concentrations (100 M); at higher concentrations (up to 500 M), the aldehydes and subsequent breakdown products from this amino acid caused increasing amounts of necrosis. Ile and His, which formed chloramines with an intermediate half-life, induced cell death that was roughly half-apoptotic and half-necrotic. The products of HOCl plus Gln, Glu, and N-⑀-acetyl-Lys were relatively nontoxic. Control experiments showed no toxicity from free amino acids (2 mM in PBS/glucose) that had not been reacted with HOCl. These included His, Leu, Ile, Lys, Glu, Gln, Arg, taurine, N-⑀-acetyl-Lys, and complete mixtures of RPMI amino acids (Sigma) tested 1 time (total amino acid concentration of 6.8 mM).

DISCUSSION
In this report, we demonstrate two significant new findings as follows: 1) different ROS induce different modes of cell death in human B lymphoma cells, and 2) HOCl-induced apoptosis is mediated by the interaction of HOCl with amino acids in the medium to form aminoacyl chloramines. Exogenously generated O 2 Ϫ kills cells entirely through its dismutation into H 2 O 2 , and both O 2 Ϫ and H 2 O 2 kill lymphoma cells by a non-apoptotic mechanism referred to as pyknosis/necrosis. As described previously for H 2 O 2 , these cells have mildly condensed (pyknotic) nuclei but show none of the classical features of apoptosis such as externalization of PS, caspase activation, formation of apo- ptotic bodies, or DNA ladder formation (15,20). In contrast, HOCl induces either necrosis or apoptosis depending on the medium in which the cells are treated. When cells are treated with HOCl in buffered saline, cell death is entirely by necrosis, and when cells are treated with HOCl in complete media, significant apoptosis is observed. These two findings will be discussed in greater detail below.
Induction of Different Modes of Cell Death by Different ROS-The term "oxidative stress" is a catch-all term intended to describe the condition that exists when the levels of ROS produced exceed the capacity of anti-oxidant systems to remove those ROS such that excess oxidants exist in a cell or tissue. The term is nondescript regarding which ROS are involved.
Our results indicate that the mode of cell death induced by "oxidants" depends on the nature and environment of the ROS that are involved, and this will depend on the source of the oxidative stress. Note that because our research focuses on how oxidants that are generated during inflammation induce cell death in target tumor cells, the experiments described here examine the effects of oxidants generated outside of a cell. The results cannot be employed to deduce how these same oxidants might act if generated inside of a cell, where different intracellular compartments would be affected. Under biological conditions in which O 2 Ϫ and H 2 O 2 are the main oxidants being formed, cell death in human B lymphoma cells is expected to be non-apoptotic, whereas if HOCl is the predominant oxidant, cell death may instead be by apoptosis. It should be pointed out that, experimentally, H 2 O 2 can induce apoptosis, but this only occurs in cells where the ATP levels are maintained above a certain threshold level of roughly 25% of control ATP levels (19 -21, 29). This may naturally be the case for certain cell types such as T cells, which have been shown to undergo apoptosis in response to H 2 O 2 but can also be induced experimentally in B lymphoma cells by inhibiting the activation of poly(ADP-ribose) polymerase that follows H 2 O 2 -induced strand breakage (20,21,35). In fact, even high levels of H 2 O 2 can induce apoptosis as long as cellular ATP levels are maintained. Incidentally, our studies also show that the same mode of cell death is induced by H 2 O 2 regardless of whether the cells are exposed to a bolus (reagent) or flux (from XA/XO) of H 2 O 2 . Physiological conditions in which O 2 Ϫ and H 2 O 2 predominate would include conditions of ischemia-reperfusion such as occur during a heart attack or ischemic stroke, following exposure to ionizing radiation or redox active drugs and chemicals, and during inflammatory conditions involving primarily macrophages. Due to the high levels of myeloperoxidase in human neutrophils, a significant portion of the H 2 O 2 that is generated by these cells is converted to HOCl (1,23,36,37). Hence, physiological conditions that lead to extensive HOCl formation will be acute and chronic inflammatory conditions involving neutrophils and monocytes (which also express myeloperoxidase).
Mechanism of HOCl-induced Cytotoxicity-Most studies of HOCl-induced cytotoxicity have been carried out on cells incubated in buffer solutions instead of cell culture media (30,38,39). In these studies, the mode of cell death was not specifically determined, but cytotoxicity was assayed by methods that are generally reflective of necrosis (e.g. trypan blue exclusion or Cr 51 release). In more recent studies, Vissers et al. (22) treated endothelial cells with HOCl for 15 min in buffered saline and then transferred the cells to complete media. Under these conditions, they found that at most 20% of the cells died by apoptosis and that the predominant form of cell death induced by higher levels of HOCl was by necrosis, consistent with our results. But in vivo, cells will not be exposed to HOCl in the absence of a large number of molecules that will react with the HOCl, primary among which are proteins and free amino acids. Hence, we focused on the mechanisms of cell death induced by HOCl in the presence of biological media.
The novel finding that we present here is that when cells are exposed to HOCl in growth media, death is directed toward apoptosis instead of necrosis, and this induction of apoptosis is due to the formation of aminoacyl chloramines in the medium. HOCl-generated chloramines have long been known to be cytotoxic both to bacteria and mammalian cells (32, 38 -44), but here, too, the mechanism of cell death either was not studied or measured only loss of membrane integrity (release of Cr 51 ). Our results suggest that the mode of mammalian cell death induced may have been apoptotic, but the cells were examined at late time points, by which time the membranes had become leaky. In fact, in many of these studies, taurine was found to protect cells from HOCl toxicity. Consistent with our results, the death from which the cells were protected was HOCl-induced necrosis.
As pointed out by Weiss and colleagues (6,45), long lived chloramines differ significantly from HOCl because they maintain an oxidizing potential yet are stable enough to diffuse some distance before oxidizing susceptible target molecules. In vivo, the most abundant amino acid in cells and tissues is taurine (46,47), which forms a long lived chloramine that we now show causes apoptosis. Taurine is especially abundant in neutrophils, being present at ϳ20 mM (40,48), and our data would predict that the main form of neutrophil cell death should be apoptotic because much of the HOCl is scavenged by taurine to form taurine chloramine (23,45). This prediction has been borne out by studies of Wagner et al. (49) who showed that the cell death induced in myelomonocytic HL-60 cells by H 2 O 2 is mediated by myeloperoxidase, and the mode of cell death is apoptotic. In addition, neutrophils are known to secrete taurine into the extracellular medium (40), thereby increasing the likelihood that neutrophil-induced death in neighboring cells (e.g. target tumor cells) will be apoptotic.
We find that the short lived aminoacyl chloramines, which decay rapidly into aldehydes and tertiary products (33,34,50,51), can have variable effects on cell viability. The chemical structures of 13 of the aldehyde products have been characterized by mass spectrometry by Heinecke and colleagues (33,34,51). Our data show that some aminoacyl chloramines, such as from Glu and Leu, decay into aldehydes (3-carboxy-propanal and 3-methyl-1-butanal, respectively) that are relatively nontoxic to cells and protect them from HOCl toxicity, allowing neither necrosis nor apoptosis. Others decay into highly cytotoxic aldehydes that induce necrosis, as is seen from the degradation of Ala monochloramine into acetaldehyde (data not shown). Amino acids that form chloramines of intermediate half-life kill by both apoptosis and necrosis, with the necrosis probably being induced primarily by the aldehydes and/or tertiary products formed. Because all of these amino acids are present in RPMI at varying concentrations, they compete with each other for the limiting amounts of HOCl used in these studies. The net effect is that treatment of cells in either RPMI or complete media results in at least a substantial proportion of the cells dying by apoptosis instead of necrosis. Serum proteins also react with HOCl and in these studies appear to act as true scavengers, protecting the cells from HOCl toxicity such that higher levels of HOCl are required to induce cell death.
Future studies will examine the molecular targets and pathways that account for chloramine-induced apoptosis in human lymphoma cells. Rapid loss of membrane integrity (lysis) in cells treated with HOCl in PBS (as evidenced by uptake of PI or release of Cr 51 (30)) indicates that HOCl toxicity derives in part from a direct attack of the oxidant on the cell membrane. HOCl is known to modify membrane lipids (52-54) as well as protein and DNA. On the other hand, aminoacyl chloramines are not as reactive as HOCl and hence are expected to have a more selective spectrum of molecular targets. The absence of lysis of cells treated with the long lived aminoacyl chloramines from taurine, N-␣-acetyl-Lys, and N-␣-acetyl-Arg, even at concentrations of 500 M, suggests that the chloramines may not act on membrane lipids. Instead, proteins compose a likely target of chloramine oxidation, with cysteine thiols and the methionine sulfoether being most susceptible (55,56). Taurine chloramine in particular has been shown to modify and inhibit activities of various proteins (38, 56 -58). The critical task for apoptosis research will be to identify the specific molecular targets that initiate caspase activation and the remainder of the apoptotic cascade.