Histone H2B Phosphorylation in Mammalian Apoptotic Cells

Histone phosphorylation was investigated in several mammalian cells undergoing apoptosis (human HL-60 and HeLa, mouse FM3A and N18 cells, and rat thymocytes). Among the four nucleosomal core histones (H2A, H2B, H3, and H4), H2B, which is not usually phosphorylated in quiescent or growing cells, was found to be phosphorylated after treatment with various apoptotic inducers. The H2B was phosphorylated around the time when nucleosomal DNA fragmentation was initiated and, like this fragmentation, was completely blocked with Z-Asp-CH2-DCB, an inhibitor of ICE or ICE-like caspase. The involved single phosphopeptide of H2B proved to be phosphorylatable in vitro with a protein kinase C, and the site Ser-32 was tentatively identified. Despite typical apoptotic chromatin condensation, the H3 phosphorylation was at a low level, and the sites where phosphorylation did occur did not include any mitosis-specific phosphopeptides. Phosphorylation of H4 was increased, but the other two histone proteins (H1 and H2A) were not appreciably changed. These observations imply that 1) H2B phosphorylation occurs universally in apoptotic cells and is associated with apoptosis-specific nucleosomal DNA fragmentation, 2) chromatin condensation in apoptosis occurs by a different biochemical mechanism from those operating during mitosis or premature chromosome condensation, and 3) this unique phosphorylation of H2B is a useful biochemical hallmark of apoptotic cells.

Apoptotic cells demonstrate dynamic structural changes, such as chromatin condensation and nucleosomal DNA fragmentation (1)(2)(3). The pathway to apoptosis triggered by exogenous agents leads to activation of some major protein kinases (4,5), as reported for cyclin/H1 kinase (6,7), protein kinase A (8 -10), protein kinase C (10,11), and mitogen-activated protein kinase (12). It is thus possible that the level or site of phosphorylation of chromatin-associated proteins changes during apoptosis-specific alteration of chromatin structure.
Chromatin is composed of nucleosomes, which themselves consist of four kinds of core histones (H2A, H2B, H3, and H4), each of which interacts with approximately 200-base pair lengths of DNA. Histone H1, the largest, is located outside the nucleosome (13). It is known that structural changes in chromatin often reflect chemical modification (14, 15), phosphorylation being associated with extensive alteration. The phosphorylation of the five individual histones is known to be specific (13). Histone H1 is phosphorylated cell cycle dependently by cdc2/H1 kinases, and the phosphorylation reaches maximum levels during mitosis (14) in both N-and C-terminal domains (16). H3 is extensively phosphorylated not only during mitosis in the cell cycle (17,18) but also in association with premature chromosome condensation (19 -21) or growth factor stimulation (22,23). The H3 phosphorylation occurs at Ser-10 in the Nterminal domain (24). H2B phosphorylation is usually negligible in both quiescent and growth states in nucleosome structures in vitro (25,26). H2A is constantly phosphorylated throughout the cell cycle, whereas H4 phosphorylation is weak or negligible depending on the cell type. These two histones (H2A and H4) are also phosphorylated at serine residues of the N-terminal end.
In view of the specificity of histone phosphorylation during the cell cycle, it is clearly of interest to investigate in more detail whether apoptotic chromatin has a specific pattern of histone phosphorylation associated with nucleosomal DNA fragmentation or chromatin condensation. I have mainly concentrated attention on the human preleukemic cell line, HL-60, which can be readily induced to undergo apoptosis by various agents including anticancer drugs (27). Here, I report that most mammalian apoptotic cells demonstrate unique phosphorylation of H2B, which does not usually occur in interphase cells.

EXPERIMENTAL PROCEDURES
Induction of Apoptosis in HL-60 Cells by Various Agents-HL-60, human preleukemic cells, were routinely maintained in suspension culture with RPMI 1640 medium supplemented with 10% fetal bovine serum (Life Technologies, Inc.) at 37°C in the presence of 5% CO 2 . For the induction of apoptosis, the HL-60 cells (approximately 4 ϫ 10 5 /ml) were treated with various apoptosis inducers and related chemicals including VP-16 (etoposide), cisplatin, A23187, ethanol, demecolcine, UV, and anti-Fas antibody. In addition, HeLa cells were treated with 0.1 M okadaic acid (28) and mouse FM3A cells with VP-16 (29). Mouse N18TG2 cells were grown (30) and cultured for 24 h in a serum-free medium. Rat thymocytes were incubated with corticosterone or dexamethasone (31), as indicated in Table I. Apoptotic cells were counted as those with apoptophores caused by blebbing (1), assessed under a phase contrast microscope.
Preparation of Phosphorylated Histones-For the preparation of 32 Plabeled histones, cells (1 ϫ 10 7 ) were labeled with [ 32 P]orthophosphate (40 Ci/ml, ICN; [ 32 P]orthophosphate, carrier-free) as described earlier (32) for the final 3 h of each period of treatment with an apoptosisinducing agent. For pulse labeling, HL-60 cells (5 ϫ 10 6 ) were treated with 20 g/ml VP-16 and [ 32 P]orthophosphate for different 2 h periods. 32 P-Labeled histones were prepared as described previously (32), and the proteins (30 g) were analyzed by acid-urea polyacrylamide electrophoresis with Triton X-100-containing gels (33) at 120 V for 48 h. The proteins were stained with 0.2% Amido Black. For examination of the effects of a caspase inhibitor, cells were treated with Z-Asp-CH 2 -DCB (Peptide Inst, Inc) at the concentration of 25 to 100 g/ml together with an apoptosis agent, and then 32 P-labeled histones were extracted. Autoradiography of 32 P-labeled proteins was conducted with an Image Analyzer (Fujix, BAS 2000).
Phosphopeptide Mapping-The 32 P-labeled histones were fractionated with HPLC 1 (35), purified, and digested with 2% trypsin. The products were analyzed by two-dimensional peptide mapping on thin layer chromatograph (TLC) plates (Funacel, FC-2020). Resolution in the first dimension was by electrophoresis with butanol/acetic acid/ water/pyridine (50:25:900:25, pH 4.7) at 550 V for 50 min. After drying, resolution in the second dimension was by ascending chromatography in butanol/acetic acid/water/pyridine (48.8:15.2:60.4:75.6) (36). Phosphopeptides of histones were identified with an image analyzer and eluted from the identified spots. The phosphopeptides were sequenced with an amino acid analyzer (Applied Biosystems 470A).
Extraction of Soluble DNA-Cells (5 ϫ 10 6 ) were lysed with a solution containing 10 mM Tris-HCl, pH 7.4, 10 mM EDTA, and 0.5% Triton X-100. After centrifugation, the supernatant was saved and treated for 1 h with 25 g/ml RNase and then with 25 g/ml proteinase K. The DNA was further treated with 0.5 M NaCl and 50% isopropanol overnight at Ϫ20°C and dissolved in a minimal amount of Tris-EDTA buffer (above solution without Triton X-100). After centrifugation, aliquots (1 g) were separated by 2% agarose gel electrophoresis at 100 V for 1 h and stained with ethidium bromide (1 g/ml).

RESULTS
Phosphorylated histones in various mammalian apoptotic cells were labeled with [ 32 P]orthophosphate and analyzed with acid-urea polyacrylamide gels containing Triton X-100 ( Fig. 1). Examination of total histone patterns in HL-60 cells (Fig. 1, Amido Black-stained protein (Ab)) treated with an apoptosisinducing agent, VP-16, revealed no essential variation (Fig. 1b) between the control (Fig. 1A) and the mitotic cells (Fig. 1c). However, considerable differences were observed regarding histone phosphorylation, as shown by 32 P autoradiography ( Fig.  1, 32 P). In the control cells, among the five histones, H1 and H2A were highly and H4 was weakly phosphorylated, whereas phosphorylation of H2B and H3 was negligible (Fig. 1a), as observed in most mammalian interphase cells. In mitotic arrest cells treated with Colcemid, H3 was highly phosphorylated, in line with chromosome condensation during the cell cycle (Fig.  1c). This H3 phosphorylation is known as a mitosis-specific phosphorylation at Ser-10. However, in the histones of apoptotic cells induced by VP-16, H2B and H4 were found to be highly phosphorylated. An increase of H3 phosphorylation in both H3.1 and H3.2 was also observed to a lesser extent (Fig.  1b). Comparison between Amido Black-stained proteins and autoradiographs demonstrated that the H2B phosphorylation occurred in a partial fraction (Fig. 1b). The amount of phosphorylated protein was approximately 5-10% of the total H2B estimated by the protein shift from the main peak in the elution profile of HPLC (data not shown).
Since separation of core histones is greater in samples without H1, the H1 in some histone samples was removed by extraction with 5% perchloric acid. H2B phosphorylation increased in parallel with prolonged duration of VP-16 treatment for 6 and 16 h ( Fig. 2A, b and c, respectively). Furthermore, H2B phosphorylation was identified with apoptosis induced by the anti-Fas antibody, increasing in parallel with the duration of treatment ( Fig. 2A, d and e). H2B phosphorylation was also consistently observed in apoptotic HL-60 cells after treatment with A23187, ethanol, or UV light (Table I) and in rat thymocyte cells exposed to two kinds of steroid hormones, dexamethasone and corticosterone (Fig. 2B, c and d). Similarly, H2B phosphorylation was found in mouse apoptotic FM3A cells induced by VP-16 and in N18TG2 cells cultured in a serum-free medium. The extent of H2B phosphorylation approximately corresponded to the level of induction of apoptosis in most cases, as shown quantitatively with the ϩ and Ϫ symbols in Table I. The evidence thus indicates that 1) H2B phosphoryl- ation occurs in most cell lines or somatic thymocyte cells derived from three mammalian species (human, mouse, and rat) exposed to various kinds of apoptotic agents, and 2) the H2B phosphorylation increases in parallel with both duration of treatment and the extent of apoptosis.
To examine the specificity of the H2B phosphorylation in apoptotic cells, HL-60 cells were treated with VP-16 in the presence of Z-Asp-CH 2 -DCB, a specific inhibitor for ICE or ICE-like caspases, which is effective at preventing DNA fragmentation (Fig. 3). As indicated in Fig. 3A, DNA degradation started to be inhibited by the peptide at a concentration of 25 g/ml and was completely blocked at 50 and 100 g/ml (c, d, and e in Fig. 3A, respectively). The effect of the peptide on histone phosphorylation in apoptotic cells was also examined (Fig. 3B). The H2B phosphorylation was also completely blocked at 25 and 50 g/ml peptide (Fig. 3B, c and d, respectively). In addition, enhancement of H3.1, H3.2, and H4 phosphorylations by apoptotic agents (Fig. 3B, b) was decreased by the inhibitor. However, effects on H1 and H2A phosphorylation were very weak. At the 25 g/ml concentration of Z-Asp-CH 2 -DCB, the effect on H2B phosphorylation was greater than on DNA fragmentation. In the case of apoptosis induced by ethanol, the inhibitor completely inhibited the H2B and most H4 phosphorylation but had no effect on H2A.1 and H1 phosphorylation (Fig. 3B, f and g). These data indicate that the caspase inhibitor mainly blocks H2B phosphorylation induced by specific apoptotic agents.
To determine the timing of H2B phosphorylation after the treatment with VP-16, HL-60 cells were pulse-labeled with 32 P for 2 h at several time points after the application and analyzed for histone phosphorylation. Fig. 4 illustrates the findings for the four core histones (H2A, H2B, H3, and H4), along with results for DNA fragmentation (Fig. 4, inset). H2B phosphorylation was sharply elevated after 6 h at around the time when DNA fragmentation started to occur, and then it continued to increase up to 18 h later. The rates for H3 and H4 phosphorylations were also slightly increased, whereas no significant change was noted for H2A and H1 (also see Fig. 1b). H3 phosphorylation was lower than expected given the typical apoptotic chromatin condensation. These data indicate that phosphorylation of H2B is most closely associated with apoptotic cells, especially apoptosis-specific nucleosomal DNA fragmentation in VP-16-induced apoptotic HL-60 cells. To access the sites of histone phosphorylation in apoptotic HL-60 cells, 32 P-labeled histones were purified by HPLC and digested with trypsin. The products were analyzed by two-dimensional thin layer chromatography and autoradiographed for 32 P activity (Fig. 5). With regard to the H2B phosphopeptide, a principal single spot migrated into the highly basic, less polar region in the peptide map (Fig. 5a). Elution and sequencing of the H2B phosphopeptide allowed the site to be tentatively assigned to Ser-32.
To explore which protein kinase might be responsible, phosphopeptides were phosphorylated in vitro with available protein kinases including cyclin/cdc2 kinase, mitogen-activated protein kinase, protein kinase A, and protein kinase C. The H2B phosphopeptide was only phosphorylated significantly by protein kinase C and to a limited extent by protein kinase A. Fig. 5b is a tryptic peptide map for H2B phosphorylated by protein kinase C. The H2B phosphopeptides observed here may contain some phosphopeptides that are incompletely cleaved at Lys or Arg residues by trypsin. A mixture of both H2B phosphopeptides (Fig. 5, a and b) showed that the H2B spot in
The major sites of H3 phosphorylation in apoptotic cells with VP-16 (Fig. 5d) or A23187 (Fig. 5e) were clearly different from those associated with mitosis ( Fig. 5f, indicated by the two arrows). Thus, the two spots usually demonstrated very faint or no labeling in normal interphase cells. Phosphopeptide maps of the three other histones (H1, H2A, and H4) did not demonstrate differences in sites from those of the control, whereas phosphorylation of a single H4 phosphopeptide was extensively enhanced (data not shown).

DISCUSSION
The present data for HL-60 cells are consistent with those of other laboratories in indicating that 1) histone H3 phosphorylation at mitosis-associated sites does not occur in apoptosis and 2) H1 and H2A do not show any considerable increase or changes in any spots. In this report, I document that 1) H2B is phosphorylated in most mammalian apoptotic cells, and 2) H3 is weakly phosphorylated at a few sites that are different from Ser-10, 3) H4 was enhanced in the phosphorylation level. It was hard to observe histone H2B modification on SDS gel electrophoresis, since H2B migrates close to H3 and H2A. In this study, H2B phosphorylation was found by using acid-urea Triton-containing gel electrophoresis, which could retard H2A and H3 migration due to the formation of mixed micells between the detergent and the hydrophobic region of protein molecules (33).
The H2B phosphorylation observed here is quite remarkable given the lack in the normal phase of the mammalian cell cycle. It is quite likely that H2B phosphorylation is apoptosis-specific, since 1) it was observed in many mammalian apoptotic cells induced by various apoptotic agents and increased parallel with both treatment duration and dose, 2) H2B started to be phosphorylated around the time when nucleosomal DNA fragmentation was initiated, 3) the phosphorylation was inhibited completely by Z-Asp-CH 2 -DCB, a broad inhibitor for ICE or ICE-associated caspase at concentrations less than that inhibiting apoptosis (37). The evidence thus strongly implies that 1) phosphorylation of H2B is tightly associated with apoptosis in most mammalian cells, and 2) phosphorylation of H2B is related to the mechanism of caspase initiated apoptosis.
At present, the function of H2B phosphorylation in the apoptotic chromatin structure is not known. However, since only a proportion of H2B was phosphorylated, the present data suggest an association with the early phase of DNA fragmentation (Figs. 3 and 4) rather than chromatin condensation. In fact, nucleosomes are released by apoptosis-specific nucleosomal fragmentation (38), and H2B phosphorylation might occur in free nucleosomes or oligonucleosomes. The N-terminal domain of H2B is reported to lie outside the nucleosomes associ- were treated with 20 g/ml VP-16 and divided into two series of cultures. In one of them, cells (5 ϫ 10 6 ) were pulse-labeled for 2 h with 32 P at 0 (untreated), 1.5 (0.5-2.5 h), 3 (2-4 h), 6 (5-7 h), 12 (11-13 h), 18 (17-19 h) h after VP-16 treatment. Histones were prepared and separated by acid-urea polyacrylamide gel electrophoresis as described under "Experimental Procedures" and as indicated in Fig. 1. The proteins were quantitated by scanning at 565 nm, and then they were cut out and assessed for 32 P radioactivity to give values per g of histones protein (cpm/g). Change with time after VP-16 treatment relative to the control value at each time point is indicated in the figure for H2A, H2B, H3, and H4. The data indicate increased rates of phosphorylation obtained by division with the control value. In contrast to the data in Fig. 1b, this resulted in a higher value for H3 than for H4 phosphorylation. Inset, DNA fragmentation during apoptosis. Soluble DNA was extracted at each time point from the remaining set of VP-16 treated cells (5 ϫ 10 6 ) and run on 2% agarose gels (Mupid, Cosmo Bio) at 100 V for 1 h. M, size markers (123-base pair DNA ladder; Life Technologies, Inc.).
FIG . 5. Characterization of apoptotic H2B and H3 phosphopeptides. a-c, 32 P-Labeled H2B histones derived from apoptotic cells or H2B phosphorylated in vitro with protein kinase C were purified with HPLC. They were digested with 2% trypsin and analyzed by twodimensional peptide mapping on TLC plates (Funacel, FC-2020) as described under "Experimental Procedures" and earlier (36). After electrophoresis (from left to right), resolution in the second dimension was conducted by ascending chromatography. a, H2B in apoptotic HL-60 cells with VP-16 (ϳ300 rpm); b, H2B phosphorylated in vitro with protein kinase C (ϳ1, 500 rpm); c, mixture of a and b. The arrow indicates the H2B spot in apoptotic cells comigrating with one of the protein kinase C phosphopeptides. d-e, H3 phosphopeptides in apoptotic cells with the two apoptosis inducers. d, VP-16 (ϳ300 rpm); e, A23187 (ϳ250 rpm); f, H3 phosphopeptides from cells treated with 0.04% Colcemid for 16 h (ϳ350 rpm). Mitosis-specific H3 phosphopeptides are indicated by arrows. The two spots were identified as peptides containing the Ser-10 residue in our earlier study (25). Small crosses indicate the origins of the phosphopeptides. detection of a histone H2B epitope by an immunological method in the early stage of apoptosis in T lymphocytes.
Whether the phosphorylation of H2B in chromatin facilitates its partial disassociation from linker DNA and makes it accessible to endonuclease(s) remains unclear. The H2B phosphorylation site was here tentatively identified as Ser-32. Interestingly, this is located in the inner globular region near the border with the N-terminal H2B tail. The site may be inside the nucleosome structure and therefore not exposed to protein kinases. This would explain why the site is usually not phosphorylated. In earlier studies by Nishizuka and co-workers, it was found that H2B at Ser-32 is phosphorylatable in vitro by a protein kinase C, which was identified earlier as protein kinase M (42) or protein kinase G (43). There is some evidence that H2B phosphorylation is stimulated by phorbol esters in quiescent Reuber H35 hepatoma cells (44). Circumstantial support for a role for protein kinase C␦, an isomer of protein kinase C, has emerged with the observation that it contains a possible cleavage site at Gln-Asp-Asn of the caspase family (ICE) and the finding that the protein kinase C isomer is proteolytically activated in apoptotic U937 cells exposed to ionizing radiation (45). It should now be determined whether H2B is phosphorylated in the nuclei or degraded free nucleosomes exposed to the cytoplasm where activated kinases are abundant.
The available data in the literature indicate that apoptotic cells have no H3 phosphorylation at Ser-10, which is specific for mitosis (24) and premature chromosome condensation (21), but the present findings indicate that apoptotic HL-60 cells have a low level of H3 phosphorylation at different sites. Direct evidence has been reported recently by Hendzel et al. (46) using a Ser-10 antibody in which apoptotic cells do not demonstrate mitotic H3 phosphorylation. I proposed earlier the idea that a high level of H1 phosphorylation was necessary for the onset of mitosis-specific H3 phosphorylation in vivo (25,26). My data and those from other laboratories indicate that apoptotic chromatin condensation does not appear to be linked to H1 phosphorylation (46 -48). This suggests that apoptotic chromatin condensation differs from that occurring during mitosis or premature chromosome condensation, despite the reported similarities (49,50). Apoptotic chromatin condensation appears irreversible with formation of chromatin clumps undergoing degradation, whereas mitotic chromatin condensation is reversed with progression through the cell cycle.
Variation in apoptotic chromatin structure may also reflect other histone modifications dependent on the cell species or agents, such as the absence of ubiquitinated histone H2A in gliomas and neurinomas (47). Using inhibitors or activators of protein kinases, apoptosis can be induced by staurosporine without considerable histone phosphorylation (48) or with H3 phosphorylation by gliotoxin, a protein kinase A activator (51). It was reported that histone H3 and H4 were deacetylated in rat apoptotic thymocytes (52). It would be of interest to ascertain the change in apoptotic histone acetylation under the influence of the caspase inhibitor, Z-Asp-CH 2 -DCB. In the case of HL-60 cells, however, the rate of acetylation during apoptosis and with the caspase inhibitor was not significantly changed. Probably, the H3 and especially H4 in the control cells are acetylated to a major extent in HL-60 cells (Fig. 3). It is now of interest to determine whether the H2B phosphorylation is the cause or result of chromatin condensation or DNA fragmentation through apoptosis.