Caspase-dependent cell death-associated release of nucleosome and damage-associated molecular patterns

Apoptosis, which is anti-inflammatory, and necrosis, which is pro-inflammatory, represent the extremes of the cell death spectrum. Cell death is complex and both apoptosis and necrosis can be observed in the same cells or tissues. Here, we introduce a novel combined mode of cellular demise – caspase-dependent regulated necrosis. Most importantly, it is mainly characterized with release of marked amount of oligo- or poly-nucleosomes and their attached damage-associated molecular patterns (DAMPs) and initiated by caspase activation. Caspase-activated DNase has dual roles in nucleosomal release as it can degrade extracellularly released chromatin into poly- or oligo-nucleosomes although it prohibits release of nucleosomes. In addition, osmotically triggered water movement following Cl− influx and subsequent Na+ influx appears to be the major driving force for nucleosomal and DAMPs release. Finally, Ca2+-activated cysteine protease, calpain, is an another essential factor in nucleosomal and DAMPs release because of complete reversion to apoptotic morphology from necrotic one and blockade of nucleosomal and DAMPs release by its inhibition.

oxidoreductase of the endoplasmic reticulum, which are characterized DAMPs (Figures 1D and F). The released DNAs and DAMPs was significantly increased by partial digestion with micrococcal nuclease (MNase) ( Figures 1H and I), indicative of connection of released nucleosomes and DAMPs to cellular bodies. In live imaging, cells loaded with SYTOX Green and membrane-permeable DNA dye, DRAQ5, showed typical apoptotic morphologies, such as vigorous nuclear and cytoplasmic shrinkage, at early times, with abrupt plasma membrane swelling and the release of genomic DNA being evident after 5 h. DNA release continued slowly over the next 5 h together with slow contractions of swelled plasma membrane (Figures 1J and K,Supplementary Movie 1). Transmission electron microscopy (TEM) demonstrated swelled but intact nuclear membrane, partially condensed chromatin in the vicinity of nuclear membrane, sparse chromatin within the nucleoplasm and chromatin-like electron-dense material in the cytosol. Later examination of cells revealed disintegrated plasma membranes and small deflated nuclei with intact nuclear membranes ( Figure 1L). Confocal microscopy revealed nuclei delineated by lamin A/C and nuclear pores appeared to have little DNA. In contrast, almost all DNA was located in the cytoplasm ( Figure 1M), indicating that the cytosolic, chromatin-like, electron-dense material evident in TEM was definitely chromatin. Nucleosomal release is a common phenomenon provoked by various cytotoxic stimuli from various cells. Nucleosomal and DAMPs release is not a phenomenon confined to amino-acid deprivation. As well-known apoptosis inducers including an anticancer drug, VP16, tumor necrosis factor-alpha (TNFα) plus cycloheximide or staurosporine could promote the release of DNA in HeLa cells (Figure 2a). Several human cell lines showed comparable or lesser DNA release with HeLa when staurosporine treated until 450% of cellular viabilities reduced (Figure 2b) indicating that nucleosomal release is a phenomenon occurred by various cytotoxic stimuli formerly known to induce apoptosis, regardless of its cellular specificity. For example, U937 displayed activated caspases, cleavage of poly ADP ribose polymerase-1 (PARP-1) and lamin A/C (Figure 2c), but little or no release of nucleosomes and DAMPs (Figures 2b and d) with definitive DNA laddering and nuclear fragmentations, whereas HeLa did not show any DNA and nuclear fragmentation in both staurosporine treatment (Figures 2e and f) and amino-acid deprivation ( Figure 1I and Supplementary Figure 1a, lower right and 1b). In confocal microscopy, staurosporine treatment showed nucleosomal release in HeLa, in contrast, nuclear fragmentations in U937 ( Figures 1D-F and 2g), suggesting apoptotic cell death without fragmentation of intracellular DNA and nuclei would be a key figure of the cell death releasing nucleosomes and DAMPs.
Release of nucleosomes and DAMPs is not associated with NETosis. Morphological features of dying HeLa appear to be very similar with NETosis, in that cells release DAMPs attached to nucleosomes instead of antimicrobial molecules. Regulatory factors of NETosis include Nox-derived reactive oxygen species (ROS), autophagy and histone citrullination by peptidylarginine deaminase (PAD). 10 Autophagy, ROS and PAD did not confer any effect on nucleosomal and DAMPs release investigated through specific inhibitors and short hairpin RNA (shRNA) transfection (Supplementary Figures 2). Therefore, cell death releasing nucleosomes and NETosis are totally different from one another, although morphologically very similar.
Comparison between cell death involving release of nucleosomes and DAMPs, and primary necrosis. Next, we compared the cell death with primary necrosis. Primary necrosis induced by heating or H 2 O 2 treatment showed earlier peak of DNA release than amino-acid deprivation or staurosporine treatment (Figure 3a). No fragmentation of genomic DNA was evident in all the conditions (Figure 3b, left panel). The released DNA from heated or H 2 O 2 -treated cells were completely degraded, whereas, surprisingly, the DNA released from staurosporine-treated and amino-aciddepleted cells showed classical apoptotic DNA cleavages into inter-nucleosomal fragments of roughly 180-base pairs Figure 1 Release of nucleosomes and DAMPs from amino-acid-depleted HeLa cells. (A) An inverted microscopic image of HeLa cells in the condition of amino-acid depletion. Arrows designate dying HeLa cells. (B) Genomic sequences of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), Fas, cytochrome oxidase subunit 1 (Co1) and ATP synthase subunit 6 (ATP6) were PCR amplified from extracellularly released DNA, genomic or mitochondrial DNAs. (C) Inverted and fluorescent microscopic images were taken from amino-acid-deprived HeLa cells in the presence of SYTOX, a membrane-impermeable DNA dye. HeLa cells deprived of amino acids for 15 h were fluorescence stained with histone H1 or IL6 antibodies (D), histone H2A, H2B, H3 or H4 antibodies (E) or HMGB1, Hsp90 or ERp57 antibodies (F) in combination with histone H1 antibody and 4',6-diamidino-2-phenylindole (DAPI). (G) Amino-acid-deprived HeLa cells were stained with SYTOX to determine viability, fixed and stained with DAPI and histone H1 antibodies. (H) Amino-acid-deprived HeLa cells were untreated or treated with MNase (500 mU/ml) for 10 min. Released DNA was quantitated at the indicated times. Data from triplicate samples are presented as mean ± S.D. (I) Conditioned media from amino-acid-deprived HeLa cells treated or untreated with MNase were western blotted with histone H1, 2B, H3, H4, IL6, ERp57, HMGB1 or Hsp90 antibodies. (J) Images captured every hour from live imaging of amino-acid-deprived HeLa cells with SYTOX (green) and DRAQ5, membrane-permeable DNA dye (red). (K) SYTOX fluorescent intensities were measured from circularized areas of live imaging of amino-acid-deprived cells in 5-min intervals. (L) TEM images of control cells (La) and amino-acid-deprived HeLa cells (Lb-Ld). (M) Amino-acid-deprived HeLa cells were fluorescence stained with lamin and nuclear pore antibodies, or lamin antibody and wheat germ agglutinin (WGE) and their multiples (Figure 3b, right panel). Heated cells released few histones and DAMPs. On the contrary, H 2 O 2treated cells released histones and DAMPs (Figure 3c). Total protein released from staurosporine-treated cells was analogous with that from H 2 O 2 -treated cells, but much more than from heated cells. The amount of released mitochondria, secretory pathway and cytosol proteins of staurosporinetreated cells was markedly higher than those of heated or H 2 O 2 -treated cells, investigated by GFP ELISA with conditioned media from cells expressing mitochondria-targeted GFP or GFP of secretory pathway, or LDH assay for detecting release of cytosolic components (Figures 3d and e). Furthermore, histone H1 and HMGB1 released from staurosporine-treated cells were bound to the released DNA, although those from heated or H 2 O 2 -treated cells were not, according to chromatin immunoprecipitation (ChIP) assay with conditioned medium (Figure 3f), implying that the cell death induced by staurosporine treatment releases proteins containing higher mitochondrial, cytosolic and secretory components, and DNAs as poly-or oligonucleosomes in comparison with primary necrosis in which totally degraded DNA and disjoined histones and DAMPs from DNA are released from the dying cells.
Release of nucleosomes and DAMPs in dying HeLa cells is dependent on caspase. To explore molecular mechanisms of cell death, the effects of cell death inhibitors were examined. There were no effects on release of nucleosome and DAMPs with co-treatment of Necrostatin-1, an inhibitor of necroptosis, or AG14699, an inhibitor of PARP-1-dependent cell death. Co-treatment with 3-methyl adenine (3MA), an inhibitor of autophagy, increased DNA release as well as protein release even further. On the contrary, nucleosomal and DAMPs releases were strikingly reduced by co-treatment with pan-caspase inhibitor zVAD-fmk and caspase 3 inhibitor zDEVD-fmk in staurosporine-treated and amino-aciddeprived cells (Figures 4a and c, and Supplementary  Figures 5a and c). In addition, cellular viability was increased markedly using zVAD-fmk co-treatments with staurosporine, but not in the condition of amino-acid depletion (  Roles of caspase-activated DNase in nucleosomal release in dying HeLa cells. An intriguing finding was that apoptotic DNA laddering was observed only in the released DNA but not in nuclear DNA (Figure 3b). This prompted investigation of the role of caspase-activated DNase (DFF40/CAD) and its mitochondrial equivalent, endonuclease G (EndoG), which both function in DNA fragmentation during apoptosis. Addition of DNase I extracellularly during cell death decreased DNA release ( Figure 5a) and release of histones ( Figure 5b) in dose-dependent manners, implying that DNase activity impaired nucleosomal release. Moreover, knock-down of CAD increased DNA release and overexpression of CAD decreased release of DNA and histones, whereas there was no effect on release of DAMPs except for Hsp90, although knock-down or overexpression of EndoG had no effect on release of both nucleosomes and DAMPs (Figures 5c-e). Overexpression of CAD induced complete degradation of genomic and released DNAs ( Figure 5f). These data are suggesting that CAD inhibits nucleosomal release through degrading nuclear DNA even if implicated in fragmentation of released nucleosomes, possibly in the extracellular space. Supporting this notion, released DNAs were additionally fragmentized by further in vitro incubation (Figure 5g), and CAD as well as inhibitor of CAD (ICAD) were released bound to extracellular released nucleosomes (Figures 5h and i).
Chloride channels are associated with release of nucleosome and DAMPs. Release of genomic DNA coincided with the beginning of cellular swelling (Figure 1j and Supplementary Movie 1), suggesting the possibility that osmotically triggered water movement and succeeding cellular swelling, caused by increased ion influx, may drive the release of nucleosomes and DAMPs. These were tested using blockers of various ion channels in the plasma Requirement of Cl − current in release of DNA was additionally validated by the cells in either Na + -free or Cl − -free medium, and subsequently DNA release was completely inhibited in Cl − -free medium ( Figure 6e and Supplementary Figure 6d). Co-treatment of DIDS with staurosporine produced massive nuclear fragmentation (Figure 6f), increased TUNEL-positive population ( Figure 6g) and inter-nucleosomal DNA fragmentation ( Figure 6h). Therefore, hallmarks of apoptosis became apparent by inhibition of Cl − channel. In fact, intracellular Cl − concentration gradually increased with peak value at 10 h ( Figure 6i) and Cl − current also increased with a peak value 4 h after treatment of staurosporine ( Figure 6j). Furthermore, DIDS partially suppressed Cl − current, although amiloride produced little change ( Figures   6k and l). However, the combined treatment of DIDS and amiloride completely blocked Cl − current in staurosporinetreated cells (Figure 6m), indicating that the voltage gradient induced by Cl − influx may have been neutralized by following Na + influx, resulting in osmotically triggered movement of water and swelling. In addition, co-treatment of zVAD-fmk with staurosporine also inhibited Cl − current (Figure 6n), indicating the probability that caspase activation may be associated with activation of Cl − channels.
Calpain activity is a prerequisite for nucleosome and DAMPs release. To evaluate the roles of intracellular free Ca 2+ , the effects of Ca 2+ chelator, BAPTA-AM and the Ca 2+ ionophore, A23187 were examined. BAPTA-AM reduced DNA release, whereas A23187 considerably increased DNA release in amino-acid-deprived or staurosporinetreated cells (Figure 7a). Furthermore, intracellular free Ca 2+ was increased after staurosporine treatment (Figure 7b), implicating the presence of Ca 2+ -mediated effectors in nucleosomal release. Thus, we estimated the role of Ca 2+ -activated cysteine proteases, calpains because of their association with apoptotic and necrotic pathways. 11,12 Calpain inhibitors ALLN or PD150606 partially blocked the release of nucleosomes and DAMPs except for Hsp90 (Figures 7c and d). Genomic DNA showed inter- In contrast, there was no change in cellular viability ( Figure 7g). Calpain activities were increased at 2 h, when DNA release and activities of caspases 3, 7 and 9 did not reach maximal levels, and decreased to the basal level by 4 h after staurosporine treatment (Figure 7h). Knock-down of calpastatin or overexpression of calpain 2 significantly increased DNA release, but knock-down of calpain S1 or overexpression of calpastatin considerably decreased DNA release (Figure 7i and Supplementary Figures 7a and b). In addition, calpain 2 began to show limited N-terminal degradation at 2 h after staurosporine treatment during the periods of increased calpain activities, whereas calpain 1 did not (Figure 7j), indicative of selective calpain II activation and its association with nucleosomal and DAMPs release. Regardless of previous reports revealing calpastatin degradation by caspases and subsequent calpain activation, [13][14][15] calpastatin cleavage by caspases was not implicated in calpain activation, because DNA release from calpastatin mutant cells of caspases cleavage sites was not considerably different from that of calpastatin-overexpressing cells ( Supplementary Figures 7c and 8a) and cleavage of calpastatin, although blocked by co-treatment of zVAD-fmk, only appeared in the beginning at 4 h at a time of no increased calpain activity (Supplementary Figure 8b). On the other hand, Cl − currents were significantly inhibited by calpain inhibitors, indicating that one of possible roles of calpain could be activation of Cl − channels (Figure 7k).

Discussion
Recent advances in biology have differentiated regulated necrosis from the programmed cell death termed apoptosis. [16][17][18][19] Our data confirm that cell death involving the release of nucleosomes and DAMPs follows morphological necrosis (Figure 1), but is a caspase-dependent process ( Figure 4). Therefore, the process should be classified as a kind of caspase-dependent regulated necrosis most importantly involving the conversion into apoptotic phenotypes by inhibition of the calpain pathway or Cl − channel (Figures 6 and 7). Necroptosis is a best understood caspase-independent regulatory necrosis, initiated by TNFα-TNFR1, FasL-Fas or Toll-like receptor pathway when apoptosis is blocked, requiring involvement of receptor interaction protein kinase 1 and 3 (RIPK1 and RIPK3). [20][21][22] Execution phase of necroptosis has been largely unknown, however, implication of several elements including ROS, reactive nitrogen species, inhibition of mitochondrial adenine nucleotide translocase and phospholipase A2-lipoxigenase have been recently elucidated. 22,23 Necroptosis and caspase-dependent regulated necrosis are presumed to be similar in morphological aspects, although both are different from each other in the requirements of caspases, RIPK1 or ROS (Figure 4 and Supplementary Figure 3). In addition, nucleosomal release in necroptosis is not reported, yet.
Apoptosis does not induce inflammation because its major regulators and processes have focused on eliminating proinflammatory materials. 24,25 Contrastingly, the cell death shown here is completely at odds with this notion because caspases activation was indispensible for releasing the inflammatory materials. Although caspases function in initiating death processes and seem to be required for opening Cl − channels, its precise molecular mechanisms remain unclear (Figures 4 and 6n).
DNA and nuclear DAMPs are reportedly known to be released in different types of cell death: late apoptosis or secondary necrosis, necrosis, or NETosis. 26 Secondary necrosis is a necrotic change of terminal phase apoptotic cells when late apoptotic products are not adequately eliminated by nearby scavengers, 27,28 from which extracellular nucleosome and DAMPs can be generated as membrane-bound vesicles after formation of apoptotic bodies, 29 whereas DAMPs and nucleosomes are directly released in caspasedependent necrosis as DAMPs bound poly-nucleosomes during relatively early phase when effector caspases are fully activated and mitochondrial membrane potential does not completely decline without apoptotic body formation and DNA fragmentation (Figure 1 and Supplementary Figure 1).
Major divergence of primary necrosis produced by H 2 O 2 treatment or heating from caspase-dependent necrosis was a release of completely degraded DNA and detached DAMPs (Figures 3b and f). Moreover, some of DAMPs such as IL6 and ERp57 were not released in the primary necrosis (Figure 3c). Conclusively, both types of cell deaths are ultimately different from each other.
NETosis, neutrophil-specific form of generalized term 'ETosis' is a kind of regulated necrosis, releasing neutrophil extracellular trap (NET) composed of chromatin, and cytoplasmic and granular products, 30 and coordinated by Noxgenerated ROS, PAD4-mediated chromatin decondensation and autophagy. 10,31 In spite of its morphological similarities just as rapid, explosive dislodge of chromatin and attached components, the regulatory factors of NETosis were not connected to release of nucleosome and DAMPs in the caspase-dependent necrosis (Supplementary Figure 2).
Nucleosomal release is associated with various cytotoxic stimuli (Figure 2a (Figure 2b), suggesting the probability that cell death processes accompanying caspase activation can be subdivided into necrotic (caspase-dependent regulated necrosis) or apoptotic (apoptosis) at the execution phase in cell typespecific manners.
One candidate modulator determining the cell fates seems plausible to be calpain systems as its prohibition converted necrotic status to be completely apoptotic (Figures 7e and f) notwithstanding no acquaintance with its exact molecular mechanism. Calpains have been known to have an important role in caspase-dependent demises. [32][33][34][35][36][37][38][39][40][41] According to the data in Figure 7, calpain II was instantaneously activated early and its inhibition strikingly reduced nucleosomal release with no influence on cellular viability, indicating that instant early calpain activation is not crucial for cell death but needed for the imperative process implicated in nucleosomal and DAMPs release, possibly by regulating Cl − channel activity. Consistent with this, it has been previously reported that calpains contribute to necrotic deaths through increase of plasma membrane permeability to ions, progressive disruption of cytoskeleton and plasma membrane proteins, and mitochondrial dysfunction. 42,43 One of hallmarks of apoptosis is oligo-nucleosomal fragmentation orchestrated by DFF40/CAD. Nevertheless, this has been hardly detected in some types of cells despite the presence of effector caspase activation. [44][45][46][47] Moreover, it may not be an essential factor for apoptosis. [48][49][50][51] In this regard, our results reveal the probable role of CAD to execute its  (Figures 5g-i). Apoptotic volume decrease (AVD) concurrent with cell shrinkage is induced by activation of K + and Cl − channels at early apoptosis before caspase activation, 52 whereas necrotic volume increase is caused by activation of Na + channels. 53 Dying HeLa cells initially showed classical AVD but promptly began to release their DNA with cellular swelling (Figure 1j and Supplementary Movie 1), accompanying the increase of intracellular free Cl − concentration and Cl − influx (Figures 6i  and j). Thus, we postulate that osmotically triggered water flow with Cl − influx is a major driving force for the release of nucleosomes and DAMPs, although, hitherto, the specific Cl − channels are not yet certain. Cl − channels have been demonstrated to function mainly in cellular volume regulation and fluid secretion. Recent findings also have shown their close connection with various human diseases. 54,55 In conclusion, we propose a novel entity of cellular demise, caspase-dependent regulated necrosis. It can be defined as follows ( Figure 8): characterized by massive release of polyand oligo-nucleosomes and their attached DAMPs and displaying necrotic morphology with no fragmentation of nuclei and DNA; subdivided from apoptosis at the step of caspases 9 and 3 activation, hence, the cell death can be blocked by zVAD-fmk; instant activation of calpains is needed for release of nucleosomes and DAMPs, subsequently, inhibition of calpain convert the necrotic cellular demise to the apoptotic one; osmotically triggered water movement following Cl − influx and secondary Na + influx is a critical factor, possibly through providing the driving force for nucleosomal and DAMPs release.
Quantification of released DNAs. Cells were cultured in six-well plates for 424 h to 80% confluency and treated with various cytotoxic reagents for the indicated time periods. Released DNA from dying cells were digested with 500 mU/ml MNase (Thermo Scientific) for 5 min. Nuclease activity was stopped with 5 mM EDTA and the culture supernatants were collected and stored at −20°C until quantification. Total genomic DNA from HeLa cells was extracted with 500 μl DNazol (Molecular Research Center, Cincinnati, OH, USA). Total DNA and released DNA were quantified using a PicoGreen dsDNA assay kit (Life Technologies, Seoul, Korea) according to the manufacturer's instructions. Data are presented as % DNA release calculated as (released DNA/total genomic DNA) × 100. 57 Live cell fluorescence imaging. Cells on Lab-Tek two-well glass chamber slide (NUNC, Rockford, IL, USA) were incubated in the presence of membraneimpermeable DNA dye, SYTOX Green (50 nM; Life Technologies), and membranepermeable DNA dye DRAQ5 (5 μM; Cell Signaling Technology). The images were acquired every 5 min for 24 h via a 100 × objective on a Deltavision RT Deconcolution microscope (Applied Precision, Issaquah, WA, USA) using a Photometrics Cool SNAPHQ 2 camera (Roper-Princeton Instruments, Trenton, NJ, USA) controlled by SoftWoRxTM Imaging workstation (Applied Precision).
Confocal microscopy. Cells grown on Lab-Tek four-well glass chamber slides (NUNC) were incubated in HBSS or medium containing appropriate reagents for the indicated times. Cells were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100 for 5 min. They were washed with PBS and incubated with primary antibodies and subsequently with secondary antibody conjugates (Alexa Fluor 594 donkey anti-mouse IgG and/or Alexa Fluor 488 donkey anti-rabbit IgG; Invitrogen). Images were collected using a laser scanning confocal microscope LSM710 (Carl Zeiss, Oberkochen, Germany) equipped with argon (488 nm) and krypton (568 nm) lasers, using an x40 water immersion objective. Images were processed with ZEN 2009 light edition (Carl Zeiss).
Transmission electron microscopy. The cells were pelleted and washed twice with PBS. Fixation was performed with phosphate buffer pH 7.4 containing 2.5% glutaraldehyde for 30 min at 4°C. The pellets were rinsed twice with cold PBS, postfixed in buffered OsO 4 , dehydrated in graded acetone and embedded in Durcupan ACM resin (Fluka, Yongin, Korea). Ultrathin sections were obtained, mounted in copper grids and counterstained with uranyl acetate and lead citrate. The specimens were observed with a Hitachi H-7600 TEM (Schaumburg, IL, USA) at 80 kV.
Assays for cellular viability and cell death, TUNEL staining, Annexin V staining and JC-1 staining. Cellular viability was quantified with Calcein-AM (Invitrogen) as specified by the manufacturer. The cells were aliquoted into 96-black well plates, incubated with Calcein-AM (2.5 μM) for 15 min and fluorescent intensities were measured at excitation and emission wavelength of 490 nm and 520 nm, respectively. Data were the mean ± S.D. of four independent measurements, and are presented as % viability ((fluorescence intensity of treated cells/fluorescent intensity of non-treated cells) × 100). Assays for determining cell death were done using cell-impermeable DNA dyes, PI (Calbiochem) or SYTOX Green (Life Technologies). Cells in triplicate were washed, stained briefly with PI (500 nM) or SYTOX Green (30 nM), and fluorescence intensities were analyzed by flow cytometry, and are presented as % cell death ((dead cellular counts/total cellular counts) × 100). Intracellular DNA fragmentation was detected using APO-BrdU TUNEL assay kit (Invitrogen) as specified by the manufacturer. Suspensions of dying cells were sequentially fixed by adding 1% paraformaldehyde in PBS for 15 min and 70% ethanol for 30 min in ice, washed twice, and DNA-labeled with TdT enzyme and BrdUTP for 60 min at 37°C. The cells were stained with Alexa Fluor 488-conjugated anti-BrdU antibody for 30 min at room temperature and with PI/RNase A for 30 min at room temperature, and finally analyzed by flow cytometry. Mitochondrial membrane potential was measured by staining of a membranepermeable JC-1 dye (MACS Miltenyi Biotech, Bergisch Gladbach, Germany) and analysis by flow cytometry. Annexin V staining was performed for detecting the exposure of phosphatidylserine to outer leaflet of plasma membrane that is a marker of early apoptosis using a Biotin Annexin V staining kit (BD Biosciences, San Jose, CA, USA) according to the manufacturer's protocol. Annexin V-positive population was defined when Annexin V staining was positive and PI staining was negative in the population by flow cytometry.
ChIP assay. The assay was performed using EZ-Magna ChIP A/G assay kit (Millipore) according to the manufacturer's instructions with minor modifications. Briefly, HeLa cells were treated with 1 μg/ml staurosporine for 10 h, with 32 mM H 2 O 2 for 4 h, or heated at 65°C for 1 h and incubated for 3 h. The conditioned media were fixed with 1% formaldehyde for 10 min, quenched with 1X glycine for 5 min and concentrated. The concentrated media were incubated overnight with goat polyclonal anti-histone H1 antibody (Santa Cruz Biotechnology), goat polyclonal anti-HMGB1 antibody (Santa Cruz Biotechnology) or normal goat serum. The antibody-protein-DNA complexes were captured by incubation with EZ-Magna ChIP A/G. Genomic DNA segments from complexes were eluted, digested with proteinase K digestion and purified using spin columns. Aliquots of the DNA were used as a template for PCR amplification using 35 cycles of 55°C annealing temperature. The GAPDH primers for PCR amplification were 5′-CCCCTT CATTGACCTCAACTAC-3′ and 5′-GAGTCCTTCCACGATACCAAAG-3′.
Calpain assay. Calpain activity was measured using Calpain Activity Fluorometric Assay Kit (Biovision, San Francisco, CA, USA) according to the manufacturer's instructions.
DNA laddering analysis. The cells were pelleted, washed twice and DNA extracted with DNazol (Molecular Research Center). DNA was precipitated with two volumes of ethanol and 0.1 volumes of 3 M sodium acetate, pH 5.2. DNA released from dying cells was concentrated, purified by phenol-chloroform extraction and precipitated with two volume of ethanol and 0.1 volumes of 3 M sodium acetate, pH 5.2. The precipitated DNA was resuspended in TE buffer pH 8.0 containing DNasefree RNase, and analyzed using 1.5% agarose gel in 1 × TBE buffer.

Measurements of intracellular chloride ion concentration.
Fluorescence-based microplate assay applying intracellular quenching of N-(ethoxycarbonylmethyl)-6-methoxyquinolinium bromide (MQAE) fluorescence was used for quantifying intracellular Cl − concentrations as described previously. 58 HeLa cells loaded with 5 mM MQAE (Life Technologies) overnight were incubated with or without staurosporine (1 μg/ml) for the indicated times. One set of the cells was washed three times with Cl − containing buffer (HBSS) and MQAE fluorescence was measured at excitation wavelength of 360 nm and an emission wavelength of 460 nm (F t ), and the other set of the cells were washed three times with Cl − -free HBSS, suspended with Cl − -free HBSS containing 10 μM tributyltin chloride (Sigma-Aldrich) and 5 μM nigericin sodium salt (Sigma-Aldrich). MQAE fluorescence was measured (F 0 ). F 0 /F t values were used as representative values for intracellular Cl −