Apoptosis Induced by Gliotoxin Is Preceded by Phosphorylation of Histone H3 and Enhanced Sensitivity of Chromatin to Nuclease Digestion*

The fungal toxin gliotoxin induces apoptotic cell death in a variety of cells. Apoptosis induced in thymocytes by gliotoxin is rapid, and DNA fragmentation is observable within 4 h treatment. Apoptosis induced by gliotoxin is calcium-independent and unaffected by protein synthesis inhibitors. We have previously shown that gliotoxin results in phosphorylation of a 16.3-kDa protein within 10 min treatment of thymocytes. Here we show that this protein is histone H3 and phosphorylation occurs on Ser-10. Cyclic AMP levels and activity of protein kinase A (PKA) are raised in cells treated with gliotoxin. Apoptosis is inhibited by genistein which also inhibits PKA and histone H3 phosphorylation. Apoptosis is also inhibited by a number of specific inhibitors of PKA suggesting apoptosis induced by gliotoxin is modulated by this kinase. The agents forskolin and cholera toxin do not induce rapid phosphorylation of H3 although some increase in phosphorylation of H3 does occur after 8 h with these agents. Forskolin and cholera toxin also induce apoptosis but over a longer time course than gliotoxin. In all cases levels of apoptosis correlate with degree of H3 phosphorylation. Cells treated with gliotoxin show an early sensitivity to micrococcal nuclease and DNase I digestion indicating a functional relationship between DNA fragmentation and H3 phosphorylation.

The fungal toxin gliotoxin induces apoptotic cell death in a variety of cells. Apoptosis induced in thymocytes by gliotoxin is rapid, and DNA fragmentation is observable within 4 h treatment. Apoptosis induced by gliotoxin is calcium-independent and unaffected by protein synthesis inhibitors. We have previously shown that gliotoxin results in phosphorylation of a 16.3-kDa protein within 10 min treatment of thymocytes. Here we show that this protein is histone H3 and phosphorylation occurs on Ser-10. Cyclic AMP levels and activity of protein kinase A (PKA) are raised in cells treated with gliotoxin. Apoptosis is inhibited by genistein which also inhibits PKA and histone H3 phosphorylation. Apoptosis is also inhibited by a number of specific inhibitors of PKA suggesting apoptosis induced by gliotoxin is modulated by this kinase. The agents forskolin and cholera toxin do not induce rapid phosphorylation of H3 although some increase in phosphorylation of H3 does occur after 8 h with these agents. Forskolin and cholera toxin also induce apoptosis but over a longer time course than gliotoxin. In all cases levels of apoptosis correlate with degree of H3 phosphorylation. Cells treated with gliotoxin show an early sensitivity to micrococcal nuclease and DNase I digestion indicating a functional relationship between DNA fragmentation and H3 phosphorylation.
Apoptotic cell death is a widely studied phenomenon with implications extending from toxin-induced cell death through immunological control to developmental biology where it is best referred to as programmed cell death (1)(2)(3). The biochemical and morphological features of final stage apoptosis are well defined and include condensation of chromatin and DNA fragmentation into large (4) and finally nucleosomal size fragments (5). Apoptosis can be induced in cells by a multitude of noxious agents including radiation, hyperthermia, and a plethora of toxins. Different agents result in different initial cellular lesions such as DNA damage or a disturbance in calcium homeostasis. These "private" pathways then presumably converge onto an apoptotic pathway common to all cells. Searches for a common biochemical pathway to apoptosis, while uncovering many possible routes, have yet to yield a signal transduction pathway unique to apoptosis. For example calcium mobiliza-tion occurs in many instances of apoptosis but is not universally required for its progression (6 -8). It is now clear that activation of proteases plays an important role in apoptotic cell death (9,10), and they are proving to be key modulators of apoptosis.
Gliotoxin is a fungal metabolite belonging to the epipolythiodioxopiperazine class of secondary metabolites. It has been shown to possess potent immunomodulating properties allowing selective depletion of mature lymphocytes resulting in successful bone marrow transplantation without graft versus host disease in a murine model (11). Gliotoxin has also been implicated in the etiology of fungal infections caused by the human pathogen Aspergillus fumigatus (12). The selective effects of gliotoxin on immune cells are probably due to the rapid induction of apoptosis in cells treated with gliotoxin together with the relative ease of induction of apoptosis in certain cells of the immune system. Gliotoxin has been shown to induce apoptosis in thymocytes, peripheral lymphocytes, macrophages, and some cell lines such as the mastocytoma cell line P815 (13)(14)(15). The mechanism by which gliotoxin induces apoptosis is currently not known. Gliotoxin has been shown to inhibit NFkB, and this would explain its immunosuppressive effects (16). Gliotoxin has been shown to inhibit protein synthesis, but this cannot be the property responsible for apoptosis in thymocytes since such cells are protected from apoptosis by protein synthesis inhibitors, at least over the time course examined here. In addition gliotoxin induces apoptosis without any accompanying rise in intracellular calcium in thymocytes (8), so perturbation of calcium levels is not involved. We have presented evidence that gliotoxin induces early incorporation of labeled thymidine into treated cells and proposed that these cells die by apoptosis after an aborted entry into the cell cycle (17). Inappropriate entry into the cell cycle has been shown to be a feature of a number of cells undergoing apoptosis (18,19), and it is apparent that there may be common features in processes involved in mitosis and apoptosis. For example premature activation of p34 cdc2 can lead to apoptosis (20). Phosphorylation/ dephosphorylation of proteins involved in proliferation are essential control elements in cell cycle progression, and there is increasing evidence that phosphorylation may be important in control of apoptosis as well as normal cell proliferation (21,22). Histone phosphorylation in particular has been shown to be involved in condensation of chromatin during mitosis (23,24). Here we show that hyperphosphorylation of histone H3 occurs on serine(s) residues within minutes of treatment with gliotoxin and that permeabilized cells at this point are more sensitive to the effect of nucleases. Our results also show that gliotoxin causes a rapid rise in cAMP levels in thymocytes consistent with our results with splenocytes (25) resulting in phosphorylation of H3 at the mitogenic Ser-10 site followed by apoptosis. Using other agents that would be expected to effect cAMP levels, we show that apoptosis correlated with the level of H3 phosphorylation and propose that rapid chromatin phosphorylation may provide a trigger for subsequent DNA dissolution.

EXPERIMENTAL PROCEDURES
Cells and Chemicals-Thymocytes were harvested from 10-day-old BALB/c mice as described previously (17) and cultured in Eagle's minimal essential medium, F15 (Life Technologies, Inc.), with 5% fetal calf serum. P815 cells were cultured in F15 in 10% fetal calf serum. All cells were incubated at 37°C in 5% carbon dioxide. Cells were treated with gliotoxin at 0.5-1 ϫ 10 6 /ml. Gliotoxin, cycloheximide, genistein, DNase I, micrococcal nuclease, and all protease inhibitors were purchased from Sigma. Anti-p34 cdc2 kinase was obtained from Santa Cruz Biotechnology, Inc. Propidium iodide was purchased from Molecular Probes. Radiolabeled inorganic phosphate and ATP were purchased from Bresatec. BMCI 1 was a kind gift from the laboratory of Dr. D. Deady. The PKI-(5-24) peptide was synthesized in the John Curtin School Biomolecular Resources Facility by Kerry McAndrew. 32 P/ 33 P Labeling and Polyacrylamide Gel Electrophoresis-Cells were washed three times in phosphate-free Krebs-Ringer bicarbonate and resuspended at 2 ϫ 10 7 /ml. Carrier-free 32 P-or 33 P-phosphate was added (20 Ci/2 ϫ 10 6 cells), and the cells were incubated for 2 h at 37°C. Cells were then resuspended in complete medium (F15) at 2 ϫ 10 6 /ml and treated with gliotoxin or other agents for varying times. Cells were then pelleted and the supernatant discarded. Cells were lysed in RIPA buffer (PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS) containing PMSF (100 g/ml), aprotinin (60 g/ml), and sodium orthovanadate (1 mM) on ice for 30 min. Samples were then treated with standard electrophoresis sample buffers, and electrophoresis was carried out in 7.5% polyacrylamide gels using Bio-Rad equipment. All experiments were carried out in duplicate and run in duplicate lanes. Only one lane is shown in figures which in each case are results from a single experiment and gel.
Protein Blotting, Tryptic Digestion, and Phosphoamino Acid Analysis-Proteins were electroblotted onto Immobilon-P membranes (Millipore) using Bio-Rad equipment. Transfer buffer consisted of 25 mM Tris, pH 8, 192 mM glycine, and 20% methanol and transfers were carried out at 75 V, 0.5 A for 60 min at 4°C. Following autoradiography to locate the bands, the protein band was excised from the membrane. In each case the same area and location of membrane was excised for comparison purposes. The protein was then either sequenced or removed for phosphoamino acid analysis. Sequencing was carried out using an Applied Biosystems 477 Protein Sequencer using a standard blot cycle. The first 15 residues were detected, and matching was made using a FASTA/GCG software package (Wisconsin University Biotechnology Center).
For amino acid analysis, the membranes were incubated with 300 l of 6 N constant boiling hydrochloric acid in a heating block at 105°C for 60 min. The sample was then dried in a SpeedieVac. The residue was dissolved in 5 l of acetic acid:pyridine:water, 5:0.5:94.5. Electrophoresis was performed in the electrophoresis buffer using Kodak chromatogram cellulose sheets at 1000 V for 60 min. The thin layer chromatography plate was dried and autoradiographed. Cold standards of phosphorylated serine, threonine, and tyrosine for comparison were run at the same time and developed using ninhydrin spray.
For tryptic digests the portion of the dried gel containing the protein located by autoradiography was excised using a scalpel and rehydrated in 2 ml of doubly distilled water. The paper backing was removed and the gel just covered with 0.2 M NH 4 HCO 3 , 50% aqueous acetonitrile and incubated at 30°C for 60 min. The fluid was removed and discarded, and the gel dried under a stream of nitrogen for 40 min. In the gel digestion was accomplished by addition of 10 -25 l of 0.02% Tween 20, 0.2 M NH 4 HCO 3 followed by 25 g of porcine modified trypsin. Further 0.2 M NH 4 HCO 3 may need to be added to fully hydrate the gel. Incubation is continued overnight followed by incubation in the same volume of 0.1% trifluoroacetic acid in 60% acetonitrile for 1-2 h. The supernatant was removed, dried on a SpeedieVac, and rehydrated in a total of 10 l of electrophoresis buffer (formic acid (88%) 50 ml, acetic acid 156 ml, deionized water 1794 ml, pH 1.9). Samples were loaded onto cellulose electrophoresis paper and air-dried, and electrophoresis in one dimen-sion was performed using a Hunter thin layer peptide mapping system (7000). The paper was air-dried and run in the second dimension in n-butyl alcohol:pyridine:acetic acid:water, 1:0.66:0.2:0.8, by volume for 5-6 h. After air drying spots were visualized using autoradiography, scraped, and placed in 50 l of water with polyvinylidene difluoride membrane overnight and sequenced from the membrane.
Apoptosis-Apoptosis was estimated following ethanol fixation and measurement of the subdiploid population using FACS (26). We have previously shown that thymocytes treated with gliotoxin show the classic morphological features of apoptotic cells (8). Briefly, after treatment cells were pelleted and resuspended in 1 ml of PBS followed by 3 ml of 90% ethanol. After fixing for 24 h at 4°C cells were washed three times in PBS and stained with propidium iodide at 40 g/ml and RNase at 10 g/ml and analyzed on a Becton Dickinson FACSCAN. Cells with less than the normal subdiploid DNA content were scored as apoptotic. FACS analysis was always consistent with nucleosomal DNA fragmentation as seen on agarose gels (not shown). Cells were prepared for electron microscopy by standard procedures and apoptosis quantified as described previously (27).
Cyclic AMP Measurements-c-AMP was measured as described previously (25) using an enzyme-liked immunosorbent assay (Cayman Chemical Co., Ann Arbor, MI).
Estimation of PKA Activity-Cyclic AMP-dependent protein kinase activity was assessed by estimating phosphorylation of the peptide Kemptide using the PKA assay kit from Promega. Briefly cells (2 ϫ 10 6 /ml in 2 ml) after treatment were pelleted at 400 ϫ g at 4°C for 5 min. Cells were quickly washed in ice-cold PBS, repelleted, and suspended in 50 l of buffer (25 mM Tris⅐HCl pH 7.4, 0.5 mM EDTA, 0.5 mM EGTA, 10 mM mercaptoethanol, 1 g/ml leupeptin, 1 g/ml aprotinin, 0.5 mM PMSF) and the cell pellet macerated using disposable plastic pestles. The lysate was centrifuged at 10,000 ϫ g for 15 min at 4°C. PKA was estimated by taking 10 l of this lysate and mixing with PKA assay buffer (200 mM Tris⅐HCl, pH 7.4, 100 mM MgCl 2 , 0.5 mg/ml BSA), 10 l of biotinylated Kemptide (stock solution at 0.5 mM), 10 l of PBS or 10 l of 25 M cAMP (for estimation of maximum PKA), and 10 l of [␥-33 P]ATP (100 -200 cpm/pmol ATP) for 5 min at 37°C. The reaction is terminated by adding 25 l of 7.5 M guanidine⅐HCl. A 25-l aliquot was removed and spotted on a streptavidin disc, and the discs were washed immediately according to the manufacturer's protocol and counted in a Beckman beta counter.
Chromatin Sensitivity in Permeabilized Cells-Cells were treated with gliotoxin with or without inhibitors at 37°C in a CO 2 incubator for 30 min. Cells were pelleted at 4°C, and 400 l of permeabilizing buffer per 2 ϫ 10 6 cells was added at 4°C. Permeabilizing buffer consisted of 9 mM HEPES, pH 7.8, 5 mM dithiothreitol. 4.5% Dextran 110,000, 1 mM EGTA, 4.5 mM MgCl 2 , 15 g/ml digitonin. Cells were incubated at 4°C for 20 min before transfer to FACS tubes. The nuclease was then added at the desired concentrations and the solution made 2 mM in calcium chloride. Tubes were then incubated for 20 -30 min at 37°C. The reaction was terminated by addition of 3 ml of ice-cold PBS, and cells were pelleted and fixed in 70% ethanol overnight. Cells were then washed in PBS and stained with propidium iodide at 40 g/ml and RNase at 10 g/ml and analyzed on a Becton Dickinson FACSCAN.
P34 cdc2 Kinase Activity-After treatment cells were pelleted at 4°C and lysed for 10 min in 500 l of lysis buffer consisting of 20 mM Tris, pH 7.4, 10 mM EDTA, 1 mM EGTA, 100 mM NaCl, 1% Triton X-100, 1 mM sodium fluoride, 1 mM ␤-glycerophosphate, 5 mM sodium pyrophosphate, 1 mM sodium vanadate containing 5 g/ml each of L-1-tosylamido-2-phenylethyl chloromethyl ketone, N ␣ -p-tosyl-L-lysine chloromethyl ketone, leupeptin, pepstatin, aprotinin, and 1 mM PMSF. Debris was pelleted at 10,000 ϫ g for 10 min, and the solution was precleared with 30 l of Santa Cruz A/G agarose secondary antibody. After preclearing, 5 l of Santa Cruz anti-p34 cdc2 was added, giving a final dilution of 1:100, and incubated on ice for 1 h followed by 30 l of Santa Cruz A/G agarose and the solution rotated for a further 1 h at 4°C. The agarose was pelleted and washed three times in lysis buffer and twice in HB buffer consisting of 60 mM ␤-glycerophosphate, 25 mM MOPS, pH 7.2. 15 mM MgCl 2 , 15 mM EGTA, 15 mM p-nitrophenyl phosphate, 1 mM dithiothreitol, 0.1 mM sodium vanadate, 1 mM PMSF, 20 g/ml leupeptin, 40 g/ml aprotinin, and 1% Triton X-100. The pellet was suspended in 10 l of KIN buffer for 20 min. KIN buffer consisted of HB buffer containing 1 mg/ml histone H1 and 200 M [␥-32 P]ATP at 10 -40 Ci/ ml. After 20 min 20 l of polyacrylamide gel electrophoresis sample buffer was added, heated at 100°C for 5 min, and run in a 15% polyacrylamide gel electrophoresis gel using Bio-Rad equipment. The gel was dried and H1 phosphorylation detected by autoradiography.

Phosphorylation of a 16.3K Protein in Thymocytes and P815
Cells by Gliotoxin-Treatment of thymocytes with gliotoxin at concentrations causing apoptosis has been shown to result in hyperphosphorylation of a 16.3K protein (17). No other proteins were observed to undergo the same degree of increased phosphorylation. Fig. 1 shows that this phosphorylation occurs rapidly and is detectable within 10 min of gliotoxin treatment. The level of phosphorylation shown in Fig. 1 at 30 min remains unchanged over a period of 2 h. The protein indicated by the arrow had a molecular mass of 16.3 kDa. Similar results were found using 3 M gliotoxin, but at 100 M gliotoxin there was no hyperphosphorylation of this protein. In fact at 100 M gliotoxin phosphorylation generally was greatly reduced (data not shown). Treatment of P815 cells with 1 and 3 M gliotoxin also resulted in enhanced phosphorylation of the same protein after 60 min as shown in Fig. 2.
Sequencing of Phosphorylated 16.3K Protein and Site of Phosphorylation-The labeled proteins were electroblotted onto nylon membranes, and the 16.3K protein was microsequenced from the membrane. The first 15 residues determined were ARTKQTARKXTGGKA (single letter code) matching histone H3 except for residue 10 which could not be determined due to insufficient signal in the sequencing cycle. We also sequenced the band immediately below H3. This protein was an unresolved mixture of histones H2A and H2B. No enhanced phosphorylation of these proteins was observed at any time, and they were used in some experiments as control phosphorylated proteins. Histone H3 was excised, digested, and analyzed for phosphoamino acids. Fig. 3 indicated that enhanced phosphorylation occurs on a serine residue. Increased labeling on threonine or tyrosine was not detected. In gel digestion of the labeled protein after excision followed by two-dimensional peptide mapping showed that most radioactivity was concentrated in a single spot indicated as A in Fig. 4. A second spot marked B with much less radioactivity may correspond to a fragment resulting from a second cleavage site (see below) and gave insufficient material for sequencing. A number of very weak spots (not visible in the figure), corresponding to bands in untreated cells, were also detected. Microsequencing of A indicated that the partial sequence KXTGGK was present in the spot indicating that phosphorylation occurs at Ser-10 exclusively. The detection of lysine (K) as the first residue is consistent with Ser-10 phosphorylation since the phosphorylated serine would tend to inhibit normal cleavage after Lys-9, resulting in cleavage after Arg-8 giving rise to a lysine as first residue detected. Some cleavage may have occurred after Lys-9 giving rise to spot B. The position of B is consistent with lesser positive charge on this residue (1 lysine versus 2 in fragment A). Further evidence for phosphorylation at Ser-10 was the lack of a detectable residue at this position during sequencing of the intact labeled protein after electroblotting. Phosphoamino acid analysis shows exclusive phosphorylation on a serine residue. These data together establish that the primary site of phosphorylation was Ser-10.
Hyperphosphorylation of H3 Is Not Induced by Cyclohexi-mide and Is a Result of Raised Cyclic AMP Levels-Gliotoxin is a known protein synthesis inhibitor although we have shown that this cannot be responsible for apoptosis induced by this toxin in thymocytes or P815 cells (17). In addition, phosphorylation of H3 in fibroblasts has been shown to occur in the presence of some protein synthesis inhibitors (28). We therefore examined if cycloheximide, a potent protein synthesis inhibitor, also resulted in histone phosphorylation. Fig. 5 shows that no phosphorylation of H3 occurs with cycloheximide. Therefore the effect of gliotoxin cannot be due to its ability to inhibit protein synthesis.
We have shown earlier that gliotoxin can modulate cAMP levels in splenocytes (25), and apoptosis has been shown to be induced by agents that raise cAMP levels (29). Fig. 6 confirms that treatment of thymocytes with dibutyryl cyclic AMP, a cell-permeant analogue of cyclic AMP, also results in enhanced phosphorylation of histone H3. The level of phosphorylation caused by dibutyryl cyclic AMP even at 500 M, however, is significantly less than that caused by gliotoxin.
Increased Cyclic AMP Levels Induced by Gliotoxin, Cholera FIG. 4. Two-dimensional tryptic digest map of radiolabeled H3. Spot A was removed and was shown to contain the partial sequence, Lysine-X-threonine consistent with cleavage between Arg-8 -Lys-9 in H3. Serine 10 (indicated as X), being phosphorylated, is not detected in the sequencer. Table I shows that cAMP levels in thymocytes treated with gliotoxin reach levels almost 7 times that found in untreated cells. In some experiments cyclic AMP levels in cells treated with 3 M gliotoxin were up to 20 times untreated cells. Because we examined the effects of forskolin and cholera toxin on thymocytes, we also measured cAMP levels in thymocytes treated with these two agents. In all experiments these latter two agents caused greater increases in cAMP than gliotoxin. Forskolin at 100 M produced levels of cyclic AMP 4 -5 times that seen with cholera toxin.

Toxin, and Forskolin-
Dexamethasone Does Not Cause Phosphorylation of Histone H3 and Phosphorylation Is Inhibited by Genistein-To determine if phosphorylation of H3 was generally associated with induction of apoptosis, we examined the effect of dexamethasone on histone H3. Fig. 7 shows that there is no enhanced phosphorylation in the presence of dexamethasone indicating that this is not a feature common to apoptosis induced by dexamethasone. We also found no increased phosphorylation of H3 with the calcium mobilizing agent thapsigargin (data not shown). In Fig. 7, endogenous phosphorylation of H3 can be detected because of the greater exposure time.
Genistein is nominally a tyrosine kinase inhibitor (30). It has also been shown to inhibit apoptosis induced by gliotoxin (17). The result shown in Fig. 7 also shows that this agent at concentrations which inhibit apoptosis results in abrogation of the hyperphosphorylation of H3. The effect of genistein at 100 M on H3 phosphorylation is also shown in Fig. 9a.
Apoptosis Induced by Gliotoxin, Forskolin, and Cholera Toxin and H3 Phosphorylation-When we examined the effect of forskolin and cholera toxin on apoptosis in thymocytes, we found very different kinetics from that seen with gliotoxin. Fig.  8 shows kinetics observed with gliotoxin, forskolin, cholera toxin, and the cell-permeant analogue of cAMP, dibutyryl cAMP. We consistently observed a far rapid onset of DNA fragmentation with gliotoxin compared with dibutyryl cAMP although both achieved the same level by 12 h as shown in Fig.  8. Forskolin and cholera toxin both showed a slow onset of DNA fragmentation and in addition did not reach the levels seen with gliotoxin even by 16 h. Neither forskolin nor cholera toxin resulted in any rapid phosphorylation of histone H3 at 1 h (Fig.  9a). Although there was no detectable H3 phosphorylation at 1 h, when we measured levels of H3 phosphorylation at 8 h we found a significant but small increase in cells treated with forskolin and cholera toxin correlating with the levels of apoptosis seen with these agents (Fig. 9b). It is notable that detec-tion of H3 phosphorylation at 8 h with gliotoxin was impossible because of extensive apoptosis in the cells at this point. In this experiment we used 33 P rather than 32 P labeling to minimize long exposure of thymocytes to irradiation. We have confirmed by electron microscopy that DNA fragmentation occurred in cells treated with cholera toxin and forskolin with the morphology of apoptosis and parallels the time course as assessed by   DNA fragmentation (data not shown).
PKA Activity and Apoptosis-We used the synthetic peptide Kemptide to assess any increase in PKA following gliotoxin treatment. As expected PKA levels were elevated after 30 min gliotoxin treatment (Table I). Levels of PKA were increased approximately 2.6-fold above untreated cells in both experiments. Genistein, a nominal tyrosine kinase inhibitor, also significantly inhibited the raised PKA levels in the presence of gliotoxin. Maximum cyclic AMP-dependent phosphorylation of Kemptide measured by adding 10 M cyclic AMP to lysates was only 40% higher than that seen with 3 M gliotoxin. To confirm that PKA was part of the signaling pathway of gliotoxin-induced apoptosis, we used the specific peptide inhibitor PKI-(5-24) and 1-n-butylamino-3-methyl-4-cyanoisoquinoline (BMCI) to inhibit apoptosis induced by gliotoxin ( Fig. 10 and Table II). The former peptide is a specific substrate-based PKA inhibitor based on the naturally occurring inhibitor PKI (31). It is a potent inhibitor of PKA with a K i in the micromolar range. With whole cells, however, it must be used at high concentrations to facilitate sufficient transport into the cell (31). BMCI is a recently described specific inhibitor of PKA with little effect on other protein kinases (32). Both these agents inhibit DNA fragmentation induced by gliotoxin. Furthermore PKI-(5-24) at 2 mM inhibits phosphorylation of histone H3 induced by gliotoxin in thymocytes, (Fig. 11).
Gliotoxin Does Not Activate p34 cdc2 Kinase-The cyclin-dependent kinase p34 cdc2 kinase has been shown to be activated by granzyme B, and this has been linked to apoptosis induced in cells by this kinase (20). Because of this and the report that histone H3 has been shown to be a substrate for p34 cdc2 in a cell-free system, we measured the activity of p34 cdc2 kinase in thymocytes after gliotoxin treatment. Fig. 12 shows there is no effect of 3 M gliotoxin on this kinase at 10 or 20 min. Phosphorylation of H3 has already been initiated by this latter time.
Effect of Gliotoxin on Chromatin Sensitivity to Nucleases-Because modification of chromatin has been shown to result in increased sensitivity to nuclease digestion, we initially examined the effect of gliotoxin treatment on chromatin sensitivity to micrococcal nuclease. After 30 min treatment with gliotoxin, permeabilized cells following ethanol fixation and propidium iodide staining showed the same low level of DNA fragmentation as control cells as expected (Fig. 13). Treatment with micrococcal nuclease results in the appearance of an increased subdiploid population of cells typical of cells undergoing apoptotic like DNA fragmentation. Thymocytes pretreated with gliotoxin at a concentration known to increase H3 phosphorylation show a significant increase in DNA fragmentation in the presence of the nuclease (Fig. 12 and Table III). The same effect was seen when DNase I was used as the probe nuclease. Titration of DNase I shows that at 5-25 units/ml there is a significant difference in the subdiploid cell population between untreated cells and those pretreated with 3 M gliotoxin. At 50 units/ml there is no longer any detectable difference due to overwhelming concentration of the nuclease. Simultaneous treatment of cells with genistein and gliotoxin before permeabilization and fixing abrogates the effect of gliotoxin on the DNA at the optimal DNase concentration of 5 units/ml (Table III). DISCUSSION Gliotoxin induces the biochemical and morphological features of apoptosis in thymocytes and P815 cells (8,17), although the mechanism of induction of apoptosis by gliotoxin has been unclear until now. Gliotoxin has no effect on intracellular calcium levels (8), and apoptosis induced by production of reactive oxygen species is also unlikely (17). Protein synthesis inhibition has also been ruled out. Gliotoxin induces rapid hyperphosphorylation of a 16.3-kDa protein in thymocytes (17). Since tyrosine phosphorylation is an early event in receptormediated signal transduction and since we had evidence that gliotoxin was inducing inappropriate entry into the cell cycle (17), we initially examined the effect of the tyrosine kinase inhibitor genistein and found that this agent inhibited apopto- sis induced by gliotoxin. Genistein also inhibited the enhanced thymidine incorporation induced by gliotoxin which precedes apoptosis in these cells (17). In this paper we show that the protein phosphorylated is histone H3 and evidence suggests phosphorylation occurs primarily on serine 10. Phosphorylation is rapid in thymocytes and P815 cells and precedes apoptosis as assessed by DNA fragmentation and morphology. Furthermore, genistein abrogates both apoptosis and H3 phosphorylation induced in thymocytes by gliotoxin. Concentrations of gliotoxin of 100 M which cause necrosis do not result in H3 phosphorylation. These data suggested a possible causal relationship between apoptosis and histone phosphorylation.
H3 phosphorylation also occurs when thymocytes are treated with the cell-permeant cAMP analogue dibutyryl c-AMP, although to a much lesser extent, indicating that phosphorylation is mediated by the raised cyclic AMP levels induced by gliotoxin. Apoptosis caused by dibutyryl cAMP is also not as extensive as that caused by gliotoxin at 4 -6 h again supporting a causal relationship between H3 phosphorylation and apoptosis. This becomes more apparent when the effects of forskolin and cholera toxin are examined. These agents, which are known to raise cAMP by activation of adenylate cyclase (directly and indirectly, respectively), only cause limited but significant H3 phosphorylation at 8 h and likewise induce less apoptosis. PKA clearly modulates apoptosis induced by gliotoxin given the inhibition of apoptosis by the specific inhibitors PKI-(5-24) and BMCI. The former also inhibits H3 phosphorylation establishing the involvement of cyclic AMP-dependent kinase in the pathway of gliotoxin-induced apoptosis. Genistein, although nominally a tyrosine kinase inhibitor, also significantly but not completely inhibits raised PKA levels caused by gliotoxin as well as H3 phosphorylation and apoptosis. Inhibition of apoptosis and H3 phosphorylation, particularly with the specific inhibitor PKI-(5-24), strongly suggests a causal relationship between the two events. We would suggest that PKA itself is the kinase responsible for H3 phosphorylation. Cyclic AMP-dependent protein kinase has been shown to phosphorylate H3 on Ser-10 in a cell-free system when the histone is associated with DNA (33). Histone H3 can also act as substrate for p34 cdc2 kinase but is unaffected by gliotoxin treatment. This is important given that inappropriate activation of p34 cdc2 kinase by granzyme B leading to unscheduled chromatin condensation has been suggested as one route to apoptosis (20). Activation of p34 cdc2 kinase has been shown not to be a general feature of apoptotic cell death induced by dexamethasone in thymocytes (34) nor for apoptosis induced by actinomycin D, hydrogen peroxide, UV radiation, cycloheximide, and ceramide in FT-210 cells (35). Phosphorylation of H3 has been shown to be dependent on stress-induced mitogen-activated protein kinase activity (36), and we are examining the activity of these kinases in gliotoxin-treated cells.
It is significant that H3 phosphorylation is tightly correlated with passage of cells from S phase to M phase in the cell cycle and has been related to chromatin condensation. Phosphoryl-ation of histones, in particular histone H1, has now been shown to be intimately connected with progression of cells through the cell cycle. The activity of the protein kinase p34 cdc2 in conjunction with cyclins A and B plays a pivotal role in controlling passage throughout the cell cycle (37,38). Histone H1 is known to be phosphorylated by p34 cdc2 and appears to be the major substrate in the cell. There are a number of other possible substrates for this kinase (39), and H3 cannot be discounted as a possible substrate although H3 phosphorylation can occur in the absence of p34 cdc2 kinase activity (40). The role of phosphorylation of H3 in cell cycle progression is less well defined. However, it has been shown to occur in late prophase/metaphase in Chinese hamster cells (41), and numerous reports now link chromatin condensation with phosphorylation of H3 during the normal cell cycle (42,43). Phosphorylation in this case also occurs on Ser-10.
H3 phosphorylation on Ser-10 occurs following mitogen stimulation of fibroblasts (28) which may modulate chromatin structure and contribute to subsequent regulation of the expression of early genes since H3 phosphorylation is tightly coupled to c-Fos/Jun expression. It was also shown that so called supra induction of both H3 phosphorylation and c-jun and c-fos expression occurs when cells are pretreated with certain protein synthesis inhibitors before treatment with growth factors. The basis of this synergy is unclear, but it is only seen with some protein synthesis inhibitors such as anisomycin suggesting that other properties of these agents may be important (28). The ability of gliotoxin to inhibit protein synthesis as well as increase cAMP levels may account for the differences seen in the kinetics of apoptosis and H3 phosphorylation with forskolin and cholera toxin. This may be a further example of supra induction of H3 phosphorylation, and we are currently measuring c-Fos and c-Jun levels in cells treated with gliotoxin. The ability of gliotoxin to inhibit NFkB might also be expected to result in more extensive apoptosis given the recognized protective effect of NFkB expression on induction of apoptosis (44) although inhibition of NFkB has also been associated with protection of neuronal cell death by glutamate (45). The differences noted between gliotoxin on the one hand and cholera toxin and forskolin may also be due to activation of different isoforms and/or cellular locations of PKA by the different agents. Forskolin for example has been shown to induce predominantly nuclear localization of cAMP compared with prostaglandin E2 (46). Regardless of these considerations, apoptosis parallels H3 phosphorylation.
Treatment of cells with gliotoxin under the same conditions  as those producing H3 phosphorylation results in increased chromatin fragility toward both micrococcal nuclease and DNase I. The latter has been suggested as a candidate nuclease implicated in apoptotic DNA fragmentation (47). DNase I sensitivity of chromatin is increased in nuclei treated with a phosphatase inhibitor associated with increased phosphorylation of histone H2A (48). Thus inappropriate and/or hyperphosphorylation of H3 may precipitate apoptosis directly by inducing premature chromatin condensation or by allowing access of endogenous nucleases to chromatin. Raised intracellular cAMP levels have been shown to be associated with apoptosis in a number of instances (29, 49 -51), and we have now related this rise to H3 hyperphosphorylation as a possible trigger for DNA fragmentation. During normal cellular function cAMP acts as a signal transducer by regulating gene expression through its action on protein kinase A (PKA). Following cAMP binding, the active catalytic subunit of PKA is translocated to the nucleus where it acts on a number of substrates resulting in expression of cAMP-inducible genes (52). The total number of possible target substrates for PKA, however, has increased dramatically since its discovery, and cAMP-dependent protein kinase has been shown to act on histone H3 in vitro (33). Gliotoxin is a potent immunomodulating agent inducing apoptosis in thymocytes, splenocytes, T blasts, macrophages, and the cell lines P815 and WEHI 7 (8,(13)(14)(15)17). Its effect on protein synthesis cannot explain its ability to induce apoptosis in all cases (17). Gliotoxin has also been shown to result in raised cAMP levels in splenocytes associated with apoptosis in those cells (25). It appears probable therefore that one general mechanism involved in gliotoxin-induced apoptotis is via raised cAMP levels with subsequent hyperphosphorylation of histone H3. Inhibition of both apoptosis and H3 phosphorylation by genistein supports this.
Phosphorylation of H3 does not appear to be generally associated with apoptosis since we did not observe it in dexamethasone-and thapsigargin-induced apoptosis, and it is therefore likely to be a specific trigger for apoptosis caused by cAMP. However, a recent report correlating poly(ADP-ribosylation) of histone H1 with apoptosis (53) suggests that chromatin modification of some sort may be a common feature in apoptosis induced by diverse agents. This could include not only ribosylation or phosphorylation but also acetylation or methylation. Gliotoxin may act at the level of adenylate cyclase, but since forskolin and cholera toxin show very different kinetics of H3 phosphorylation and apoptosis, this may not be the case. The latter differences, however, may be due to secondary specific effects of gliotoxin as suggested above. Gliotoxin has been shown to inhibit a number of thiol requiring enzymes (54), and it may act to inhibit cAMP phosphodiesterases thus indirectly raising cAMP levels. The mechanism of gliotoxin-induced cAMP rises, and subcellular locations of subsequently activated PKA are under investigation.
In conclusion we have correlated PKA-dependent phosphorylation of H3 at the mitogenic Ser-10 site with apoptosis and presented evidence for a causal link between the two. These observations continue to emphasize the close relationship between pathways mediating apoptosis and those mediating cell proliferation.