Chromatin Condensation Is Not Associated with Apoptosis*

Apoptosis plays an important role in the survival of an organism, and substantial work has been done to understand the signaling pathways that regulate this process. Characteristic changes in chromatin organization accompany apoptosis and are routinely used as markers for cell death. We have examined the organization of chromatin in apoptotic PC12 and HeLa cells by indirect immunofluorescence and electron spectroscopic imaging. Our results indicate that de novo chromatin condensation normally seen during mitosis does not occur when cells undergo apoptosis. Instead, the condensed chromatin typically observed results from aggregation of the heterochromatin. We present evidence that, early in apoptosis, there is a rapid degradation of the nuclease-hypersensitive euchromatin that contains hyperacetylated histones. This occurs coincident with the loss of nuclear integrity due to degradation of lamins and reorganization of intranuclear protein matrix. These events lead to collapse of the nucleus and aggregation of heterochromatin to produce the appearance of condensed apoptotic chromatin. This heterochromatin aggregate is then digested by nucleases to produce the oligonucleosomal DNA ladder that is a hallmark of late apoptosis. Unlike mitosis, we have not seen any evidence for the requirement of phosphorylated histones H1 and H3 to maintain the chromatin in the condensed state.

Apoptosis, or programmed cell death, is a tightly regulated mechanism that multicellular organisms employ to replace damaged cells and to remove cells no longer needed during normal growth and development (1)(2)(3)(4). The generation of large condensed chromatin bodies and degradation of DNA into oligonucleosomal fragments are features that are seen in many cells undergoing programmed cell death. Although commonly used as markers of apoptosis, the biochemical mechanism responsible for these features is not known. De novo chromosome condensation occurs during mitosis in cycling cells, and this condensation has been extensively studied. Many of the genes involved in regulating mitosis have been identified (5)(6)(7). The cdc2 gene, which encodes a histone H1 kinase, was identified as the major regulator of mitosis and has also been implicated in apoptosis. This raises the question of whether the same mechanisms regulating the formation of the mitotic chromosome also apply to apoptotic chromatin condensation. The requirement for histone H1 kinase activity in both processes led to the suggestion that cells could be triggering apoptosis by entering into an unscheduled mitosis and that a major player in this process could be the p34 cdc2 kinase (1,4). Activation of the p34 cdc2 kinase was first demonstrated to accompany apoptosis in YAC-1 lymphoma cells (8). Subsequently, it was observed that temperatureinduced inactivation of p34 cdc2 in FT210 cells prevented the induction of apoptosis by treatment with fragmentin and perforin. Since then, the up-regulation of p34 cdc2 during apoptosis has been reported in other cell lines (8 -12), including PC12 (13)(14)(15) and HeLa (16), the cell lines used in the present study. However, there are also reports of apoptosis occurring without the activity of p34 cdc2 protein kinase (17)(18)(19)(20). Thus, although p34 cdc2 is required for apoptosis that is induced by some agents, this requirement is not universal to all apoptotic pathways.
In this study, we examined the organization of the condensed chromatin in early apoptotic cells and during mitosis. Our results suggest that the chromatin condensation associated with apoptosis is not an active process. Instead, our results indicate that apoptotic condensation arises as a consequence of the loss of structural integrity of the euchromatin, nuclear matrix, and nuclear lamina. This allows the heterochromatin to aggregate. In vitro DNase I digestion experiments demonstrate that the integrity of the nuclease-sensitive chromatin is essential to maintain the dispersed state of interphase chromatin. These experiments further demonstrate that chromatin has a propensity to aggregate in the absence of post-translational modifications of histones or other nuclear proteins.

MATERIALS AND METHODS
Cell Cultures and Indirect Immunofluorescence-PC12 cells were maintained in RPMI 1640 medium containing 10% horse serum and 5% fetal bovine serum as described (21,22). HeLa cells were grown in Dulbecco's modified Eagle's medium containing 10% newborn calf serum. For indirect immunofluorescence microscopy, HeLa cells were grown on regular coverslips, whereas PC12 cells were grown on coverslips that were coated with poly-L-lysine. After induction of apoptosis, cells were fixed with 1% paraformaldehyde prepared in PBS, 1 followed by permeabilization with PBS containing 0.1% Triton X-100, with three washes with PBS between treatments. The cells were then incubated with antibodies for 30 min at room temperature. Antibodies to acetylated histones and to histones H1 and H3 were used at 1:200 dilutions. Antibodies to phosphorylated histone H3 as well as lamins A and C were used at 1:1000 dilutions. Secondary antibodies were used at 1:100 to 1:200 dilutions. After three washes with PBS, the cells were incu-bated for 30 min at room temperature with fluorescein isothiocyanateconjugated anti-mouse IgG for anti-lamin antibodies or with Cy3-conjugated anti-rabbit IgG for anti-histone antibodies. Following this incubation, the coverslips were washed three times with PBS and mounted onto microscope slides with 90% glycerol containing 1 mg/ml p-phenylenediamine and DAPI. The cells were then imaged with a Zeiss Universal microscope equipped with epifluorescence using a Princeton Instruments 14-bit cooled charge-coupled device.
Electron Spectroscopic Imaging (ESI)-The procedure for preparation and examination of cells by electron spectroscopic imaging was described by Hendzel and Bazett-Jones (23). Briefly, apoptotic cells were collected by centrifugation and fixed by incubating overnight at 4°C in 2% glutaraldehyde in PBS. The cells were pelleted and dehydrated through a series of increasing concentrations of ethanol, with the final dehydration in 95% ethanol. The final cell pellet was polymerized in Quetol 651, and sectioned into 20 -40-nm thick slices. ESI was then performed as described (23) with a Zeiss EM902 electron microscope equipped with a prism-mirror-prism electron imaging spectrometer.
Histone Analyses-HeLa cells were incubated overnight in regular growth medium containing increasing concentrations of sodium butyrate. The cells were collected by trypsinization and pelleted. Lysis of cells and extraction of histones were as described by Th'ng et al. (24). Extraction buffers contained 2 mM butyrate to suppress the activities of endogenous deacetylases. About 100 g of total histones was loaded into each lane of a 30-cm acid-urea gel as described (24), except that the gel solution contained 6% Triton X-100. Following electrophoresis, the histones were visualized by staining with Coomassie Blue R-250.
DNA Degradation in Apoptotic Cells-Susceptibility of DNA to nuclease digestion in apoptotic cells was measured by release of [ 3 H]thymidine as described by Nishioka and Welsh (26). Briefly, HeLa cells were prelabeled overnight with 2 Ci/ml [ 3 H]thymidine and washed once. They were then dispensed into a 96-well plate at a density of 40,000 cells in 100 l/well in the presence of 50 M C 6 -ceramide and varying concentrations of sodium butyrate, as described above. After incubation for 24 h, 100 l of lysis buffer (20 mM Tris, pH 8.0, 10 mM EDTA, and 0.5% Triton X-100) was added to each well, and the plates were centrifuged to pellet the nuclei. 100 l of the lysate was then removed from each of the wells and transferred to scintillation vials for counting. To determine the total amount of label incorporated into the DNA, 100 l of SDS lysis buffer (0.1 M NaOH and 0.2% SDS) was added to the set of untreated cells, and 100 l was then removed for counting. The amount of labeled DNA released into the cytoplasm after lysis with lysis buffer was then expressed as a percentage of the total amount of labeled DNA in the cells, as determined by lysis with the SDS lysis buffer. The treatments were done in quadruplets and were performed three times for verification. Statistical evaluations and plotting of results were performed with Microsoft Excel.
Cell Viability-The determination of percentage of cell survival after induction of apoptosis was by the MTT assay as described by Carmichael et al. (25). Briefly, HeLa cells were plated onto a 96-well plate, with ϳ40,000 cells/well. The cells were then treated with C 6 -ceramide in the presence of increasing concentrations of sodium butyrate, up to 5 mM. After an overnight incubation, the medium was replaced with phenolphthalein-free Dulbecco's modified Eagle's medium containing 2 mg/ml MTT reagent. Following a 3-h incubation at 37°C, the medium was removed, and 100 l of Me 2 SO was added to the cells. The reading of the MTT reaction was at 460 nm on a microplate reader (Molecular Devices). For each treatment, quadruplet samples were taken, and statistical evaluations were performed with Microsoft Excel.

Immunofluorescence of Acetylated Histones in Apoptotic
Cells-For this study, we compared the nuclear organization of cells at early stages of apoptosis with untreated control cells. PC12 cells were induced to differentiate by incubation with nerve growth factor (NGF), followed by withdrawal of NGF and serum for 24 h to induce apoptosis according to the procedure of Greene (21) and Green et al. (22). Although the PC12 cells were allowed to differentiate for up to 7 days, Ͻ1% of the population continued to cycle and entered mitosis. Induction of apoptosis in HeLa cells was accomplished by incubation with 50 -100 M C 6 -ceramide for 12-24 h (27-29). The apoptotic cells appeared smaller than control cells, and the DAPI-stained nuclei exhibited large domains of condensed chromatin (Fig. 1A). These cells were at the early stages of apoptosis, as determined by the absence of cellular fragmentation and apoptotic bodies and the absence of DNA laddering as determined by agarose gel electrophoresis (data not shown).
We investigated the fate of euchromatin during early apoptosis by indirect immunofluorescence using antibodies to acetylated histones H3 and H4. Transcriptionally competent regions of chromatin contain hyperacetylated histones H3 and H4 (6, 30 -35), and these antibodies have a high degree of specificity toward these regions (36). In particular, the acetylated H3 antibody recognizes a diacetyllysine epitope that is found almost exclusively in the highest acetylated species of histone H3 (37,38). These antibodies have been extensively used for immunofluorescence studies of chromatin organization in mammalian cells (36,37), and Western blot analyses of whole cell extracts have further demonstrated their specificity for acetylated histones. 2 Identical results were obtained with both antibodies, and for simplicity, only the results obtained with the more widely characterized acetylated H4 antibody are shown. Fig. 1 shows a typical staining pattern with antibodies to acetylated histone H4 (A, left panel) in PC12 cells counterstained with DAPI (A, right panel). In control interphase cells, acetylated histone H4 staining was seen throughout the nucleus, except in the nucleoli (Fig. 1A). Highly condensed mitotic chromosomes also stained with this antibody (A, inset). In all cases, the specific nuclear staining seen in the cells further illustrates the absence of cross-reaction with cytoplasmic proteins. However, in cells that contain the highly condensed apoptotic chromatin, as visualized by staining with DAPI (A, right panel, arrowheads), very little or no staining with antibody to acetylated histones was observed. This lack of staining was consistently seen in cells that spontaneously undergo apoptosis ( Fig. 1A) and in cells induced to enter apoptosis by serum withdrawal (Fig. 1B). Interestingly, within this population of cells in early apoptosis, there were cells that stained predominantly in the cytoplasmic regions (Fig. 1B, inset), a staining pattern never seen in normal cells. These results indicate that acetylated chromatin is degraded early during apoptosis and released into the cytoplasm, where the histones continue to be further degraded. Although it is possible that there could be deacetylation in cells undergoing apoptosis, the presence of immunoreactive histone fragments in the cytoplasm indicates that the absence of acetylated histones in the nucleus is most likely due to the nucleolytic release of acetylated chromatin. The percentage of apoptotic cells displaying cytoplasmic staining with antibodies to acetylated histones is usually ϳ20% in PC12 cells, indicating that this leakage is a transient event and that degradation of euchromatin occurs early during apoptosis. It is most likely that this immunoreactivity is against fragments of histones that are degraded and trapped in the dense cytoplasm of the apoptotic cell. The observation that this cytoplasmic staining is transient further indicates that the histones continue to be degraded when released from the nuclei. Indeed, attempts to isolate histones from nucleosomal DNA that leaked out into the cytoplasm during apoptosis have been unsuccessful. 2 This cell phenotype has been generally observed in HeLa, Indian muntjac, SKN, and normal diploid human fibroblast cells entering apoptosis (data not shown). In all these cases, the cells with apoptotic chromatin bodies failed to stain with this antibody. This lack of staining is not likely due to inaccessibility of condensed chromatin to antibodies since condensed metaphase chromosomes stained well with the antibody to acetylated histone H4 (Fig. 1A, inset). High resolution ESI studies showed that metaphase chromosomes and apoptotic chromatin have very similar packing densities (see Fig. 5). Very dense apoptotic bodies only appear after the nucleus has entirely disintegrated (data not shown).
These results indicate that most of the nuclease-sensitive acetylated chromatin is degraded early during apoptosis, and at earlier time points, acetylated chromatin fragments leach out into the cytoplasm. We therefore evaluated the integrity of the nuclear lamina by indirect immunofluorescence of lamins A and C. Fig. 1C shows that apoptotic cells containing condensed nuclear material, as indicated by DAPI staining, failed to stain with antibodies to lamins A and C (left panel, arrows). Identical results were obtained for lamin B1 (data not shown). This is consistent with previous reports showing that lamin degradation plays a major role in the reorganization of the nucleus during apoptosis (39 -41). Loss of lamin staining in early apoptotic cells coincided with loss of acetylated histone staining, and it is likely that degradation of nuclear lamins facilitated leakage of acetylated chromatin into the cytoplasm. Histone Acetylation and Susceptibility to Apoptosis-Immunostaining of apoptotic cells with antibodies suggests that the euchromatin is hypersensitive to nuclease degradation during apoptosis due to the pre-existing levels of acetylated histones. To determine if increasing the overall level of histone acetylation will further enhance accessibility of DNA to nucleases, HeLa cells were induced into apoptosis with 50 M C 6 -ceramide in the presence of varying concentrations of sodium butyrate. HeLa cells were used for this study because they do not aggregate like PC12 cells and can be plated onto 96-well dishes at a consistent density, allowing for measurements to be made quantitatively. Ceramide has been well documented as an inducer of apoptosis through the sphingomyelin-ceramide pathway (27)(28)(29). Sodium butyrate inhibits the deacetylation of histones in mammalian cells, leading to the accumulation of hyperacetylated histones and cell cycle arrest without loss of viability (42). Increased histone acetylation has been demonstrated to open up chromatin (6,43,44), allowing for greater accessibility to transcription-associated complexes as well as to nucleases (30). Migration of histone H4 on a Triton-acid-urea gel was used as an indication of the levels of histone acetylation in the cell ( Fig. 2A). In cycling cells, 70% of histone H4 remained in the unacetylated state, and ϳ20% was monoacetylated (Fig. 2B). With increasing concentrations of butyrate, there was a corresponding increase in levels of histone acetylation, as evident by the appearance of bands above unmodified histone H4 (indicated as A0H4). At 0.2 and 0.5 mM butyrate, the unacetylated form of histone H4 declined to ϳ10%, with a corresponding increase in acetylated forms, up to 90% of total H4 (Fig. 2B). Increasing the concentrations of butyrate to 2 mM led to an overall increase in the levels of acetylation, with the tetraacetylated histone H4 (indicated as A4H4) becoming more prominent. The level of unmodified histone H4 declined to Ͻ5% of the total. A similar increase in histone acetylation was seen with trichostatin A, a specific inhibitor of histone deacetylases. However, this agent was found to be cytotoxic to the cells.
To assess the influence of histone acetylation on the accessibility of DNA to degradation during apoptosis, HeLa cells that were prelabeled with [ 3 H]thymidine were treated with ceramide in the presence of increasing concentrations of butyrate. In the absence of ceramide, treatment with up to 0.5 mM butyrate at these concentrations released Ͻ5% of labeled DNA, and this was increased to only 10% when butyrate was increased to 2 mM. With a 24-h treatment with 50 M ceramide to induce apoptosis, the amount of labeled DNA released was 35%, and raising the ceramide concentration to 100 M did not increase the release of labeled DNA (data not shown). In the presence of up to 0.5 mM butyrate, the release of labeled DNA was increased to 50% (Fig. 2C), and further increases in butyrate concentration did not lead to any increase in DNA released. This plateau in the release of labeled DNA suggests that the threshold of accessibility of the DNA to nucleases was reached. The absence of further release of labeled DNA also indicates that the higher concentrations of butyrate did not induce cells into apoptosis, suggesting that the increased release of labeled DNA is from the increased accessibility.
The concentrations of butyrate employed in the experiment had a minimal effect on the viability of these cells (Fig. 2D), as determined by the MTT method (26). The observed decline of 20% in the presence of butyrate is from the cytostatic effects of butyrate. Butyrate has been demonstrated to arrest the growth of mammalian cells and to induce cellular differentiation in certain cells. However, its effect on HeLa cells that were induced into apoptosis by ceramide was apparent. In the presence of ceramide alone, ϳ65% of the cells remained viable after 24 h. In the presence of increasing concentrations of butyrate, the percentage of viable cells was reduced to between 20 and 30%. These results show that although increased levels of acetylated histones in chromatin do not induce apoptosis by themselves, they do accelerate cell death, most likely by facilitating the cleavage of DNA by nucleases. Cells that were hyperacetylated during the induction of apoptosis often showed some staining with acetylated histone antibodies (data not shown). This was highly variable from cell to cell and was most prominent on the surface and immediately outside of the chromatin bodies stained with DAPI (data not shown).
Absence of Phosphorylated Histones H1 and H3 in Apoptotic Chromatin-The phosphorylation of histones H1 (6) and H3 (45) is correlated with mitotic chromosome condensation. In particular, histone H3 phosphorylation is correlated both spatially and temporally with mitotic chromosome condensation (46). Phosphorylation of histone H3 initiates in mid to late G 2 near centromeric heterochromatin. The bulk of histone H3 is phosphorylated during very late G 2 (46). Thus, if apoptotic chromatin condensation involves mechanisms employed to drive mitotic condensation, hyperphosphorylated histone H1 and phosphorylated histone H3 are expected to be enriched in condensed apoptotic chromatin. This possibility was evaluated using antibodies specific for phosphorylated histone H1 (Fig. 3, left panels, PhosH1) (47), phosphorylated histone H3 (PhosH3) (46), and the globular region of histone H1 (H1) (48). The corresponding DAPI images are also shown in Fig. 3 (right panels). The top panels of Fig. 3 show a field of NGF-treated PC12 cells stained with the antibody recognizing phosphorylated histone H1. As reported previously, G 0 /G 1 cells stained very weakly with the antibody (top left panel, arrows). However, even with prolonged incubation with NGF, a few PC12 cells continued to cycle, giving rise to a few cycling cells that contained elevated levels of phosphorylated histone H1. One such cell, with an S-phase level of phosphorylated histone H1, is seen at the right most part of the top left panel. This population typically constitutes Ͻ1% of the total number of cells. Cells show maximal staining with this antibody during mitosis. 2 Unlike mitotic cells, the apoptotic cells, which are notable by their decreased nuclear diameter and intense staining with DAPI (top right panel), were not stained by the phosphorylated histone H1 antibody (top left panel, arrowheads). Note that the cells with an apparently normal interphase appearance do not show increases in staining expected if phosphorylation was an early event during apoptosis, prior to the adoption of a clear apoptotic phenotype.
Similar results were obtained with an antibody specific for phosphorylated histone H3. Here, the interphase cells stained very weakly with the antibody recognizing phosphorylated histone H3 (Fig. 3, middle left panel, arrows) (46). As with the phosphorylated histone H1 antibody, the apoptotic cells stained very weakly with the antibody to phosphorylated histone H3 (middle left panel, arrowheads). In rare cases, we saw intensely stained apoptotic chromatin that was most likely to be from apoptosis in the subset of cells that were in G 2 , after phosphorylation of histone H3 had initiated (46). Revised estimates on the timing of the initiation of H3 phosphorylation indicate that it occurs very near the S/G 2 transition. Interphase cells containing mitotic phosphorylations normally represent Ͻ10% of an asynchronously growing culture (46). It was shown by Lindenboim et al. (13) that PC12 cells will undergo apoptosis in all phases of the cell cycle upon serum deprivation. Whereas Ͻ1% of the apoptotic cells had increased levels of phosphoryl-ation on histone H3, the bulk of the cells, which were in G 0 , showed no phosphorylation on histone H3 during apoptosis.
The absence of immunoreactivity is not likely due to a general lack of accessibility since apoptotic cells enriched in phosphorylated histone H3 are observed at frequencies expected of a modification present only during a small proportion of interphase. Moreover, mitotic chromosomes stain intensely with this antibody (46). The accessibility was further confirmed by staining cells with an antibody recognizing the globular domain of histone H1 (Fig. 3, bottom panels). Although some apoptotic chromatin bodies fail to stain with this antibody, clearly the result is distinct from those obtained with antibodies to acetylated and phosphorylated histones. The arrowheads indicate the positions of apoptotic cells that are positively stained with this antibody. Thus, the results show that phosphorylations of histones H1 and H3, which occur during mitosis and may be important in reversible mitotic chromosome condensation, do not appear to be required for maintenance of the condensation in apoptotic chromatin.
Electron Spectroscopic Imaging of Apoptotic Nuclei-The above results indicate that euchromatin is degraded prior to terminal condensation and that the structural integrity of euchromatin is essential for the dispersed state of interphase chromatin in normal cells. Therefore, an active mechanism of condensing chromatin during apoptosis may not be necessary. Rather, the apoptotic chromatin bodies arise from the release of nuclear constraints on condensed heterochromatin and the subsequent aggregation of this chromatin. To further evaluate this hypothesis, we examined the structural organization of the intranuclear protein matrix and chromatin during apoptosis by ESI, an analytical transmission electron microscopy method that enables the independent but simultaneous determination of protein and nucleic acid organization in ultrathin sections (23,49,50). This form of transmission electron microscopy images inelastically scattered electrons using an energy filter enabling the collection of both mass-specific and element-specific information on a specimen. These images are of high contrast without requiring the deposition of electron-dense stains. Fig. 4 shows a low magnification image of a PC12 cell undergoing apoptosis (center panels), a normal interphase cell (left panels), and a prophase cell (right panels). The comparison of 120-eV (Fig. 4A) and 155-eV (Fig. 4B) energy loss images enables the discrimination of chromatin and non-chromatin (nuclear matrix) structures within the nucleoplasm. In images collected at 155-eV energy loss, the plastic embedding medium is seen as a white background; the proteinaceous components show up as a intermediate, light gray structure; and the dark material represents the chromatin fibers and ribonucleoprotein particles that are enhanced by the presence of phosphorus (23). In comparing the 120-and 155-eV images, the interchromatin (nuclear matrix) component showed similar contrast relative to the embedding medium, reflecting the low density of phosphorus within this structure. In contrast, regions of chromatin (arrows in 155-eV images) either as isolated 30-nm fibers or, more frequently, in association with larger condensed domains showed significantly improved contrast in the 155-eV images due to the high density of phosphorus within the DNA. One striking feature is that the nuclei in apoptotic cells (center panels) showed a considerably more dense interchromatin protein component compared with control cells, as evident from the higher level of gray regions (examples are indicated by arrows in Fig. 4) and the lesser degree of white background. An enlargement of a section of the nuclei illustrating the increased density of the proteinaceous component (gray areas indicated by the large arrows) is shown in Fig. 4C. For example, in the regions of the interphase nucleus that are chromatin-depleted, the space in which neither protein nor nucleic acid could be detected constituted 21% of the area. In the prophase cell, with partially condensed chromosomes, this constituted 37% of the area. However, in apoptotic cells, this chromatin-depleted region constituted only 11% of the area in the nucleus. It was apparent that nuclear density was increased generally for all structures. Thus, the size and density of the chromatin domains (small arrows) were increased relative to control interphase cells (left panels), indicative of the aggregation of chromatin. The increase in the linear distances transversed by condensed domains (Fig. 4, A and B) in the section of nucleus is further consistent with aggregation occurring between domains. These domains were normally more dispersed in control interphase cell nuclei. Another feature that was routinely seen in apoptotic cells was involution of the nuclei, as indicated by the large arrows in Fig. 4B, indicating a loss of structural integrity. In summary, the nuclei in the cells that are undergoing apoptosis showed an increase in density of proteinaceous and nucleic acid components. Although the morphology may be largely mediated by the decreased volume of the nucleus (see Fig. 1), protease-mediated degradation of the nuclear lamina and internal protein matrix is likely to be an essential prerequisite for this process to occur.
Organization of DNA in Apoptotic Nuclei-Our indirect immunofluorescence results indicate that apoptotic bodies are derived from the aggregation of heterochromatin domains following nucleolytic and proteolytic removal of structural constraints within the nucleus. Thus, apoptotic chromatin "condensation" differs from the biochemically mediated condensation process seen during mitosis. High resolution ESI images of phosphorus using an energy-filtered transmission electron microscope enables the determination of DNA organization within nucleosomes in vitro (50) and chromatin in situ (51). Fig. 5 shows the net phosphorus images of condensed regions of chromatin within mitotic (A and E), apoptotic (C, D, and G), and heterochromatic (B and F) regions. These were obtained by digitizing the images collected at 120 eV, which are mass-dependent, and at 155 eV, which are phosphorus-and mass-dependent (23). The signal resulting from subtraction of the 120-eV image from the 155-eV image repre-sents the distribution of the phosphorus-rich nucleic acid and appears as white fibrils. The large arrows show regions containing "30-nm" chromatin fibers, and the small arrows show ribosomes. In all three types of condensed chromatin, the 30-nm fiber (large arrows) was the predominant structural feature, with no obvious orderly folding of this fiber observable in the ultrathin sections. The results demonstrate that there is no major reorganization at the DNA level when cells undergo apoptosis and that the packing density of the chromatin is very similar in apoptotic chromatin, mitotic chromosomes, and highly condensed interphase heterochromatin. We note, however, that 30-nm chromatin can often no longer be observed in the apoptotic chromatin bodies found at the latest stages of apoptosis, after the cell nucleus is no longer discernible. 2

DISCUSSION
Chromatin condensation and the appearance of oligonucleosomal DNA fragments have been used as hallmarks of apoptosis, although the mechanism driving these processes remains to be elucidated. It has been suggested that condensation of chromatin may be driven by the same histone phosphorylations that are associated with mitotic chromosome condensation and that this premature entry into mitosis would lead to mitotic catastrophe and cell death (1,4). During mitosis, the p34 cdc2 kinase is activated; histone H1 is hyperphosphorylated at up to six sites; and histone H3 is phosphorylated at a single site. The activation of p34 cdc2 kinase reported in several apoptotic cells is consistent with the model of premature entry into mitosis (9 -16). However, this requirement for cdc2 is not universal since there are also cells that enter apoptosis in the absence of activation of the cdc2 gene product (17)(18)(19)(20). In PC12 cells, activation of cdc2 has been reported to precede the onset of apoptosis. In this study, immunostaining of apoptotic PC12 cells showed no increase in phosphorylation of histones H1 and H3 (see Fig. 3), suggesting that the condensation of chromatin during apoptosis may not require mitotic phosphorylations. This absence of histone phosphorylation is consistent with the observations of Marushige and Marushige (52) that these phosphorylations are not detected in apoptotic glioma and neurinoma cells. As pointed out by these authors, it is possible that phosphorylation of histones during apoptosis could be a transient event and could have escaped detection. This would require that these modifications be involved in the initiation, but not the maintenance, of the condensed chromatin state. If this were the case, the preapoptotic cells, not otherwise distinguishable from normal interphase cells, should have shown elevated levels of staining with phosphorylated histone antibodies. No such increase, relative to the uninduced population, was observed. Phosphorylation of histone H3 was recently reported in thymocytes and in P815 cells when induced into apoptosis by gliotoxin (53). However, this modification seems to be specific to gliotoxin-induced apoptosis in these cells. One possibility is that these cells preferentially entered apoptosis in G 2 , resulting in the accumulation of mitotically phosphorylated histone H3. We find that apoptotic chromatin can be generated in the presence of phosphorylated histone H3 (see "Results"), although our results indicate that there is no need for post-translational modifications of histones to initiate chromatin aggregation. To date, the role of the cdc2 gene product in apoptosis remains unclear. The activation of this gene may be coincidental during the activation of apoptosis, although the studies of Shi et al. (8) and Furukawa et al. (10) demonstrated a strict requirement for p34 cdc2 kinase activity during apoptosis. An alternative possibility is that p34 cdc2 is required for the phosphorylation of other proteins, such as lamins, for apoptosis to proceed to completion.
Similarly, a role for histone deacetylation in initiating apoptotic chromatin condensation has recently been proposed for thymocytes (54). Consistent with our immunofluorescence observations, Allera et al. (54) observed, by biochemical analysis, a steady-state increase in the proportion of unacetylated histone H4 during the initial stages of apoptosis. These changes were observed prior to the generation of nucleosomal ladders. Our observations indicate that, rather than bulk deacetylation occurring, acetylated chromatin is degraded prior to the generation of nucleosomal ladders derived from aggregated and principally deacetylated chromatin. Allera et al. (54) also report that ongoing histone acetylation is substantially reduced in populations enriched in apoptotic thymocytes. We argue that this reflects the degradation of the chromatin and associated enzymes responsible for dynamic histone acetylation. Indeed, we have observed that apoptotic cells also fail to stain for CBP, P300, TAFII250, and histone deacetylase 1. 2 Moreover, cells induced into apoptosis in the presence of histone deacetylase inhibitors also preferentially lose highly acetylated epitopes. 2 These data indicate that degradation of highly acetylated chromatin, rather than histone deacetylation, accounts for the loss of the acetylated histone epitopes in apoptotic nuclei.
This study shows that histone acetylation is important in modulating apoptosis. It has been reported in lymphoid and colorectal cancer cells that butyrate synergistically acts with staurosporine to induce apoptosis and that this induction is dependent upon new protein synthesis involved in caspase-3 induction (55). In at least some cases, therefore, it is likely that butyrate acts directly in the induction of transcription at acetylation-sensitive promoters. A second possibility, not necessarily Following induction of apoptosis, nucleases degrade the euchromatin, which may then leak out into the cytoplasm. The degradation of euchromatin relieves structural constraints on the chromatin and results in the collapse of heterochromatin into large aggregates. This is facilitated by the degradation of lamins and the internal nuclear matrix, leading to a loss of structural integrity and collapse of the nucleus, and the formation of the large heterochromatin aggregates observed at the latest stages of apoptosis. exclusive of the first, is that butyrate treatment makes euchromatin more sensitive to nuclease cleavage. This would explain the observation made by Medina et al. (55) that cells are more susceptible to apoptosis when exposed to butyrate, even at concentrations that were too low to induce caspase-3.
We have observed that DNase I digestion of the nucleasesensitive chromatin in vitro can result in apoptosis-like chromatin condensation. 2 Consequently, DNA degradation alone appears sufficient for the generation of "condensed" chromatin in interphase cells. From the results of this study, we suggest that the highly condensed chromatin observed in apoptosis is not the result of an active process, as is the case during mitosis, but arises as a consequence of the release of structural and spatial constraints on preexisting heterochromatin. It is known that upon induction of apoptosis, there is concurrent activation of nucleases and proteases in the nuclei (41,56). Recently, a nuclease complex that requires activation by caspases during apoptosis has been identified (57,58). CAD (caspase-activated DNase) is present in the cytoplasm as an inactive complex with an inhibitor subunit (ICAD). Upon induction of apoptosis, ICAD is cleaved by caspase-3, releasing activated CAD, which then causes degradation of the DNA in the nucleus. When the caspase recognition sequence is altered to prevent degradation of ICAD, the nuclease remains inactive, and DNA degradation is prevented, although other apoptotic processes continue. It is also known that inhibitors of nucleases, such as zinc and aurintricarboxylic acid, delay or prevent the onset of apoptosis (59 -61). In this report, we show that the highly acetylated nuclease-sensitive regions of chromatin are rapidly degraded early during apoptosis, leaving the more condensed heterochromatin. The observed increase in the rate of progression to apoptosis when histone acetylation is elevated indicates that the euchromatic regions of chromatin may have a role in regulating the initiation of apoptosis. Thus, we propose that the degradation of the euchromatic regions of chromatin, which have multiple expression-related dynamic contacts with the nuclear matrix (62), relieves many of the intranuclear structural constraints on the interphase chromosomes, facilitating aggregation of the remaining heterochromatin (see model in Fig. 6). In the absence of protease cleavage, this aggregation occurs at points of major structural association, the nuclear lamina and the nucleolar surface. The activation of proteases and the degradation of both the nuclear lamina and components of the intranuclear protein matrix enable the aggregation of chromatin into more spherical bodies frequently observed at later points in cultured apoptotic cells. Results showing that inhibition of lamin degradation prevents nuclei from acquiring complete apoptotic features (39,40) are consistent with this hypothesis. The heterochromatin aggregates within a nucleus whose integrity has been compromised by degradation of organizational constraints provided by DNase I-sensitive chromatin, the nuclear lamina, and the intranuclear protein matrix. Notably, these compact structures very often adopt organizations that appear to minimize surface area. This would reduce the accessibility of the chromatin to nucleases. Consequently, the slower degradation of DNA then proceeds to produce the oligonucleosomal ladder typically seen when apoptotic chromatin is electrophoretically resolved.