Introduction

The execution of cell death after initiation by an external stimulus can display a large variety of distinct phenotypes (Leist and Jaattela, 2001; Wyllie and Golstein, 2001; Chipuk and Green, 2005). Two of the best studied and phenotypically clearly distinguishable forms of cell death are apoptosis and necrosis, which are characterized by typical morphological criteria. During apoptosis, cells display membrane blebbing, fragmentation of nuclear DNA, chromatin condensation and exposure of phosphatidylserine, which results in the rapid removal of the dying cell. In contrast, during necrosis, which typically occurs after excessive damage by physical or chemical injury, cells display cytoplasmic vacuolization and swelling of the cytoplasm and organelles concomitant with a loss of membrane integrity. Although apoptosis is considered as the major form of programmed cell death while necrosis is often regarded as a passive and unregulated cell death, it is now clear that a large variety of intermediate forms between these two extreme types of cell death exists (Leist and Jaattela, 2001). Along this line, examples for programmed necrosis have been described (Proskuryakov et al., 2003; Zong et al., 2004). Another well-investigated type of programmed cell death is autophagy, which depicts the cellular self-recycling of cytoplasmic components and disposal of organelles and is characterized by an accumulation of autophagic vesicles (Cuervo, 2004; Shintani and Klionsky, 2004). Additionally, other forms of cell death are known including paraptosis and autoschizis, which are characterized by specific criteria on their own (Sperandio et al., 2000; Jamison et al., 2002). Certainly, the exact phenotype of a dying cell is dependent on many different factors including the cell type, the cellular context, and the specific death stimulus (Leist and Jaattela, 2001; Wyllie and Golstein, 2001). Characteristic changes that differ among the various forms also include modifications of the cell shape and architecture, such as alterations of the cytoskeleton (Bursch et al., 2000; Grzanka et al., 2003).

Mammalian S-adenosylmethionine (SAM)-dependent methyltransferases are involved in the modification of a large number of substrates including proteins, RNA, DNA, and lipids (Clarke and Banfield, 2001; Schubert et al., 2003). The methylation of these substrates can have a massive impact on several cellular processes. For example, methylation of DNA by DNA methyltransferases (Dnmts) regulates DNA replication, gene transcription and DNA repair processes (Chen and Li, 2004). On the other hand, methylation of protein substrates can have important influences on a multitude of cellular events, too, including transcriptional regulation, definition of chromatin domains as in the case of histone methylation (Zhang and Reinberg, 2001; Rice et al., 2003) and apoptosis as exemplified by methylation of the tumour suppressor gene product p53 (Chuikov et al., 2004).

The methylation reaction performed by SAM-dependent methyltransferases leads to the generation of two products, the methylated substrate and the by-product S-adenosylhomocysteine (SAH) (Finkelstein, 1998). It is important to note that SAH by itself functions as a potent inhibitor of SAM-dependent methyltransferases and thus needs to be further broken down into adenosine (Ado) and homocysteine (Hcy) by the enzyme S-adenosylhomocysteine hydrolase (SAHH). Inhibition of the SAHH in turn leads to the inhibition of all SAM-dependent methyltransferase reactions due to an increase in the intracellular concentration of SAH (Clarke and Banfield, 2001, 2002). A potent inhibitor of the SAAH and therefore of SAM-dependent methyltransferases is adenosine dialdehyde (AdOx) (Bartel and Borchardt, 1984). In the present study, we have analysed the influence of AdOx on cell cycle and cell survival. We demonstrate that AdOx causes changes in cell cycle distribution and leads to cell death in different cell types. Interestingly, while lower concentrations of AdOx mainly caused typical apoptosis, higher concentrations led to a novel, caspase-independent form of cell death, which was characterized by incomplete nuclear condensation and actin aggregation. Morphologically, this type of cell death was characterized by a striking protuberation of the nucleus and formation of cytoplasmic extensions.

Results

Inhibition of methyltransferases causes apoptotic cell death and cell cycle arrest

Application of AdOx leads to inactivation of SAHH and therefore functions as a general inhibitor of SAM-dependent methyltransferases. Since distinct methyltransferases are inhibited by different concentrations of AdOx (Clarke and Banfield, 2001), we added increasing amounts of AdOx to HeLa cells and determined effects on cell cycle and survival after treatment over a period of 24 h. Effects on cell cycle and survival were monitored by propidium iodide uptake into nuclei prepared from AdOx-treated HeLa cells and subsequent flow cytometric analysis. The presence of apoptotic cells was determined by appearance of a sub-G1 peak due to nuclear fragmentation and formation of hypodiploid nuclei. As can be seen in Figure 1a, after 24 h of AdOx treatment formation of hypodiploid nuclei occurred dose dependently, starting at 100 M and reaching a maximum at 250–500 M AdOx. The cell death caused by AdOx involved the activation of caspases, since the appearance of the sub-G1 peak was almost completely inhibited by treatment of the cells with the broad-spectrum caspase inhibitor benzoyloxycarbonyl-valyl-alanyl-aspartic acid (O-methyl)-fluoro-methylketone (zVAD-fmk). Concomitant with the formation of hypodiploid DNA the cells detached from the plastic surface and displayed an apoptotic morphology including cellular shrinkage and membrane blebbing (see Figure 3).

Figure 1.

Adenosine dialdehyde (AdOx) causes apoptotic cell death and cell cycle arrest. HeLa cells were either left untreated or incubated with increasing amounts of the methyltransferase inhibitor AdOx in the presence (grey bars) or absence (white bars) of 20 M zVAD-fmk. zVAD-fmk was added 3 h prior to the treatment with AdOx. The proportion of hypodiploid nuclei (a, b) and the content of nuclei in the G1, S, and G2 phases (c) were determined by flow cytometric analysis after 24 h. In Figure 3b whole cell extracts from control cells and cells treated with the indicated concentration of AdOx were separated by SDS–PAGE and analysed by Western blotting with antibodies against the proteins indicated at the right. (a and c) represent meanss.d. from experiments performed in triplicates and (b) shows a representative experiment

Full figure and legend (133K)

Figure 3.

A morphologically distinct cell death phenotype caused by treatment with AdOx. HeLa cells were either left untreated (a, f), or were treated with 0.1 M STS (b) or 0.25 M STS (g), 0.25 mM AdOx (c, h), and 1 mM AdOx (d, e, i, j and k) respectively. To cells shown in e 20 M of the caspase inhibitor zVAD-fmk was added 3 h prior to the treatment with 1 mM AdOx. Samples in a–e were analysed by light microscopy. Samples in f–k were visualized by transmission electron microscopy. Magnification was 7000-fold (f–i) or 20 000-fold (j, k) respectively

Full figure and legend (208K)

Since AdOx-mediated nuclear fragmentation was inhibited by addition of zVAD-fmk, we verified the activation of caspases and cleavage of caspase substrates by Western blot experiments with antibodies against caspase-3 and the nuclear proteins Acinus and PARP, which are known to be cleaved during apoptosis (Figure 1b). Caspase-3 was activated during AdOx treatment as determined by the appearance of its active subunits. Caspase activation was confirmed by detection of the cleavage of Acinus and PARP that paralleled the formation of hypodiploid nuclei. Taken together, our data strongly indicate that AdOx can cause a caspase-dependent apoptotic cell death in HeLa cells.

Although treatment with AdOx concentrations higher than 500 M did not lead to further nuclear fragmentation (Figure 1a), the cells discontinued growing (data not shown). We, therefore, investigated potential effects of AdOx on cell cycle events and determined the proportions of cell cycle stages in the same nuclei, which were used for analysis of hypodiploid nuclei formation in Figure 1a. As can be seen in Figure 1c, effects of AdOx on cell cycle distribution were concentration dependent, with a dominant arrest of the cells in the G2 phase when 100 M AdOx were applied. Higher concentrations of AdOx caused a less significant but detectable increase of nuclei in the S phase, indicating that inhibition of methyltransferases leads to dysregulation of the cell cycle in HeLa cells.

Apoptotic cell death mediated by AdOx is dependent on the presence of the tumour suppressor p53

Inhibition of methyltransferases by AdOx includes blocking of the enzymatic activity of Dnmts. Several reports have demonstrated that loss or inhibition of Dnmts can lead to p53-dependent apoptosis (Jackson-Grusby et al., 2001; Karpf et al., 2001; Stancheva et al., 2001; Schneider-Stock et al., 2004). To analyse whether apoptotic cell death caused by AdOx involves the function of p53, we employed HCT116 colon carcinoma cells that were either wild type (HCT116 wt) or deleted of the p53 tumour suppressor gene (HCT116 p53^-/-) (Bunz et al., 1998). Furthermore, HCT116 cells devoid of the Bcl-2 family member Bax (HCT116 Bax^-/-), which is involved in apoptosis via a mitochondrial pathway, were used (Zhang et al., 2000).

In preliminary experiments, we determined that application of 75 M AdOx caused a significant amount of apoptotic cell death in HCT116 cells (data not shown). Subsequently, HCT116 wt, HCT116 p53^-/-, and HCT Bax^-/- cells were either left untreated or were treated with 75 M AdOx for 24 h (Figure 2a), 48 h (Figure 2b), and 72 h (Figure 2c) and analysed for apoptosis by the flow cytometric detection of hypodiploid nuclei. While after 48 h of treatment significant amounts of cell death occurred in the parental cells, death in HCT116 p53^-/- cells was strongly diminished. In contrast, death of HCT116 Bax^-/- cells was only slightly reduced. Similar to HeLa cells, formation of hypodiploid nuclei was largely inhibited by addition of zVAD-fmk, also indicating that the cell death of HCT116 cells was due to apoptosis. In summary, these results strongly indicate that apoptotic cell death caused by AdOx is largely dependent on the presence of the p53 tumour suppressor gene product.

Figure 2.

Apoptotic cell death caused by AdOx requires the tumour suppressor gene product p53 but not the Bcl-2 family member Bax. HCT116 cells either wild type (wt) or deleted for p53 (p53^-/-) or Bax (Bax^-/-) were employed. The absence of p53 and Bax protein was confirmed by immunoblotting (see inlet in a). Cells were treated with 75 M AdOx, and the formation of hypodiploid nuclei was determined by flow cytometric analysis after 24 h (a), 48 h (b), and 72 h (c). Experiments were performed in the absence (white bars) or presence (grey bars) of 20 M of the caspase inhibitor zVAD-fmk, which was added 3 h prior to the treatment with AdOx. Experiments were performed in triplicates and the meanss.d. are shown

Full figure and legend (108K)

AdOx can cause a novel type of cell death with characteristic morphological alterations

Light microscopical analyses of HeLa cells treated with apoptosis-inducing doses of AdOx (i.e. 250 M) revealed typical morphological alterations of apoptosis including cell shrinkage and membrane blebbing (Figure 3c) when compared to untreated cells (Figure 3a). Changes in cell morphology were similar to those observed after incubation with staurosporine (STS), a classical apoptosis inducer (Figure 3b). Interestingly, treatment with AdOx concentrations above 500 M induced a form of cell death that was markedly different from apoptosis. Instead, at these concentration cell death was characterized by a striking nuclear protuberation and the extensive formation of cytoplasmic extensions, which resulted in an 'octopus'-like cellular morphology (Figure 3d). With ongoing treatment cells increasingly lost their membrane integrity, as determined by the uptake of propidium iodide (data not shown). This type of cell death seemed to be independent of caspase activation, since appearance of the described phenotype could not be prevented by the addition of zVAD-fmk (Figure 3e).

To further investigate this unusual form of cell death, we analysed AdOx- as well as STS-treated cells at the ultrastructural level. As can be seen in Figure 3g and h, apoptotic features including intensive chromatin condensation, disruption of the nuclear envelope, and nuclear fragmentation could be confirmed by electron microscopy when HeLa cells were treated with STS and 250 M AdOx. Interestingly, HeLa cells displaying protuberation of the nucleus could readily be detected after treatment with 1 mM AdOx (Figure 3i and j). The nuclear envelope seemed to remain intact, as highlighted by a larger magnification analysis of the region surrounding the extruded nucleus (Figure 3k). HeLa cells treated solely with the vehicle DMSO displayed a normal and healthy phenotype (Figure 3f).

Since the cell death caused by higher concentrations of AdOx was characterized by major changes in cell shape, we investigated the influence of AdOx treatment on components of the cytoskeleton. For this purpose the localization of actin and tubulin filaments was analysed in HeLa cells by immunofluorescence analyses (Figure 4). While untreated HeLa cells showed the typical filamentous distribution of actin and tubulin (Figure 4a and b), the cytoskeleton underwent major changes after treatment of the cells with 1 mM AdOx. Staining with phalloidin-TRITC revealed a strong and characteristic aggregation of actin filaments (Figure 4m). Changes in the tubulin filaments were less obvious, although after immunostaining tubulin seemed to be located preferentially at the periphery of the cytoplasm (Figure 4n). Noteworthy, the aggregated actin filaments were located at the centre of the cytoplasmic extrusions and at the base of the protuberated nucleus (Figure 4o and p). In agreement with the morphological alterations shown in Figure 3, addition of zVAD-fmk did not prevent changes in the distribution of cytoskeletal components (Figure 4q–t).

Figure 4.

Changes of the actin and tubulin cytoskeleton in cells treated with AdOx and STS. HeLa cells were either left untreated (a–d) or treated with 0.1 M STS (e–h), 0.1 mM AdOx (i–l), and 1 mM AdOx (m–p and q–t) respectively. To cells shown in q–t 20 M of the caspase inhibitor zVAD-fmk was added 3 h prior to the treatment with 1 mM AdOx. The fixed and permeabilized cells were analysed using a phalloidin-TRITC conjugate for actin staining and by immunofluorescence with an antibody against tubulin. While untreated cells presented the typical actin and tubulin network, cells treated with AdOx or STS showed characteristic alterations of the cytoskeleton

Full figure and legend (286K)

In contrast, cells treated with either 100 M AdOx (Figure 4i–l) or STS (Figure 4e–h) showed less dramatic changes in actin and tubulin localization, with a diffuse distribution of actin throughout the cell compared to the intensive actin condensation caused by higher AdOx concentrations (Figure 4e and i). These cells demonstrated a clearly apoptotic phenotype, as can be judged by the typical strong chromatin condensation and nuclear fragmentation. Noteworthy, cells treated with 1 mM AdOx revealed only a partial condensation of DNA at the nuclear periphery and no nuclear fragmentation both in the presence and the absence of zVAD-fmk (Figure 4o and s). These data confirm the results obtained by flow cytometric and electron microscopic analyses (Figures 1, 2 and 3).

AdOx treatment leads to activation of the Bcl-2 family member Bax

Activation of the Bcl-2 family member Bax via an N-terminal conformational change is a characteristic feature of the mitochondrial pathway of apoptosis. To analyse whether Bax activation occurs when cell death is induced by AdOx treatment, we employed an antibody that specifically recognizes the proapoptotic conformation of Bax. As demonstrated by immunofluorescence analysis, no signal could be obtained with this antibody when nonapoptotic untreated HeLa cells were analysed (Figure 5b). In contrast, a significant number of HeLa cells treated with 0.1 M STS or 100 M AdOx displayed activation of Bax, along with typical morphological signs of apoptosis (Figure 5g and l). Interestingly, also HeLa cells treated with 1 mM AdOx, which showed nuclear protuberation with partially condensed chromatin, cytoplasmic extensions, and aggregation of actin filaments, displayed Bax activation (Figure 5q). Taken together, although AdOx can initiate apoptotic pathways, the resulting morphological phenotype of cell death is largely influenced by the AdOx concentration.

Figure 5.

Activation of Bax in AdOx- and STS-treated cells. HeLa cells were either left untreated (a–e) or were treated with 0.1 M STS (f–j), 0.1 mM AdOx (k–o), and 1 mM AdOx (p–t) respectively. After fixation and permeabilization, the actin cytoskeleton and activated Bax were visualized with a phalloidin-TRITC conjugate and by indirect immunofluorescence with a conformation-specific antibody recognizing activated Bax, respectively. Bax activation could be detected in cells treated with STS and AdOx. In a–e, an untreated but spontaneously apoptotic HeLa cell is labelled with an arrow and serves as a positive control for the conformation-specific Bax antibody

Full figure and legend (378K)

Discussion

Cellular methylation reactions using SAM as substrate lead to the addition of methyl groups to a large amount of substrates with a wide range of biological implications. By cleaving the methylation reaction byproduct SAH into Ado and Hcy, the SAHH performs an important function in the cell, since excess SAH would otherwise interfere with SAM-dependent methyltransferase reactions. In this study, we employed the SAHH inhibitor AdOx in order to modulate the activity of methyltransferases and to analyse the consequences of the perturbation of proper substrate methylations. We observed that treatment with AdOx caused cell cycle arrest as well as cell death in different cell types. Dependent on the AdOx concentration applied, cell death was clearly apoptotic, as judged by morphological criteria and inhibition of DNA fragmentation by the caspase inhibitor zVAD-fmk.

Since the equilibrium of the SAHH reaction favours SAH formation, SAH can accumulate in the presence of excess amounts of Hcy and Ado, leading again to an inhibition of SAHH and in turn of SAM-dependent methyltransferases (Clarke and Banfield, 2001, 2002). Owing to this interplay cells can be very sensitive to the balance between SAH and Ado+Hcy. For example, it has been shown that increases in SAH or Hcy in combination with Ado sensitize cells to TNF-induced apoptosis (Bergmann et al., 1994; Ratter et al., 1999; Song et al., 2004). Also, Hcy can cause death in several cell types, including apoptosis of neurons and trophoblast cells (Kruman et al., 2000, 2004; Di Simone et al., 2003). These results are in agreement with our data, showing that interference with SAM-dependent methylation reactions can cause apoptotic cell death. Analyses by Ratter et al. (1999) implicated an involvement of methylation by the isoprenylcysteine carboxyl methyltransferase in the enhancement of TNF-mediated cytotoxicity caused by an excess of Ado and Hcy in tumour cells. Further investigations are required to investigate whether repression of isoprenylcysteine carboxyl methyltransferase reactions plays a role in apoptosis induction in HeLa and HCT116 cells.

SAM-dependent methylation reactions are also involved in the modification of DNA by Dnmts. These enzymatic reactions include maintenance methylation performed by Dnmt1 during replication as well as de novo methylation by Dnmt3a and Dnmt3b. Importantly, data from several groups have shown that loss or inhibition of Dnmts leads to p53-dependent apoptosis. Along this line, Dnmt1-deficient mouse embryonic fibroblasts undergo p53-dependent cell death (Jackson-Grusby et al., 2001), and loss of Dnmt1 causes p53-dependent apoptosis in Xenopus embryos (Stancheva et al., 2001). Furthermore, inhibition of Dnmts with 5-aza-2'-deoxycytidine or of Dnmt3a with 5-aza-cytidine did lead to p53-dependent cell death in HCT116 colon carcinoma cells (Karpf et al., 2001; Schneider-Stock et al., 2004).

Using HCT116 cells, which were either wild type or deleted for the tumour suppressor gene p53 or the Bcl-2 family member Bax, we could show that induction of apoptosis by AdOx was largely dependent on the presence of the p53. Therefore, it is arguable that DNA hypomethylation due to inhibition of Dnmts and subsequent induction of apoptosis via the function of p53 is involved in the action of AdOx. On the other hand, as a recent report showed that the apoptotic function of p53 can be regulated by protein methylation via the methyltransferase Set9 (Chuikov et al., 2004), an involvement of post-transcriptional modification of p53 cannot be excluded.

In addition to apoptosis, inhibition of methyltransferases was also able to induce a novel type of cell death, which to our knowledge has not been described before. This type of cell death is characterized by a distinct cell shape and striking morphological alterations including nuclear protuberation together with incomplete chromatin condensation and aggregation of actin filaments. When higher amounts of AdOx above 500 M were applied the dying cells predominantly displayed this novel cell death phenotype. One of the major characteristics of this kind of cell death is the very pronounced aggregation of actin filaments. Modifications of the actin skeleton have been described during apoptosis, and it has been reported that cells exhibit different fates of the cytoskeleton during autophagic and apoptotic death (Bursch et al., 2000; Suarez-Huerta et al., 2000). In other experimental systems modulation of the actin skeleton was the cause for apoptosis, and alterations of the actin polymerization status had profound effects on apoptotic cell morphology (Rao et al., 1999; Martin and Leder, 2001; White et al., 2001). Nearly all actin species are methylated at histidine 73 (Hennessey et al., 1993). Although the consequences of this post-translational modification are unknown, it can be speculated that a modulation of actin methylation is involved in the phenotypical outcome during AdOx-mediated cell death.

Other described forms of cell death include paraptosis and autoschizis (Sperandio et al., 2000; Jamison et al., 2002). Although these types of cell death might share certain features with the novel type of cell death described here (e.g. lack of response to caspase inhibitors during paraptosis), they are morphologically clearly distinct and the pronounced changes of the actin filaments have not been described.

Interestingly, in a recent report it was demonstrated that an intact actin cytoskeleton but not tubulin network is required for apoptotic nuclear disintegration (Croft et al., 2005). Also, it has been shown that activation of the Rho effector kinase ROCK I by caspase cleavage during apoptosis induces membrane blebbing most likely via ROCK I-promoted generation of actin-myosin force (Coleman et al., 2001; Sebbagh et al., 2001). It is therefore possible that the intensive reorganization of actin filaments observed after AdOx treatment prevents nuclear disruption. Since the actin aggregates are characteristically placed at the base of the protuberated nucleus and at the centre of the cytoplasmic extensions, actin filaments might still be involved in these morphological changes.

Analysis of the cells with an antibody specifically recognizing the conformationally changed and active form of Bax showed that apoptotic pathways are still activated by AdOx. However, as revealed by the only partial chromatin condensation and the absence of nuclear fragmentation, apoptosis is not executed by higher doses of AdOx. Bax was also activated by apoptosis-inducing concentrations of AdOx, but analysis of mutant HCT116 cells showed that elimination of the Bax gene product only slightly reduces cell death. Together, these data indicate that activation of Bax does not play a major role and is not a prerequisite in mediating apoptotic cell death by AdOx.

It has been shown that several chromatin-associated proteins are subjected to methylation. For example, the methylation of histones and its influence on the transcription and definition of chromatin domains has been extensively explored (Zhang and Reinberg, 2001; Rice et al., 2003). Also, methylation of the architectural chromatin-binding protein HMGA1a during apoptosis has been described in leukaemic cells, which was prevented by the application of AdOx (Sgarra et al., 2003). Although the consequences of the latter modification are unknown, it should be considered that inhibition of methylation of chromatin-associated proteins might influence nuclear changes during apoptosis.

In summary, our data show that the methyltransferase inhibitor AdOx can cause apoptosis, which is largely mediated by p53, conceivably via the p53-mediated detection of hypomethylated DNA. In addition to apoptosis, AdOx can cause a formerly undescribed form of cell death with striking morphological features. Although a physiological role of this cell death has yet to be established, the phenotypical attributes point to important roles of methylation reactions in the modulation of actin filaments as well as in nuclear fragmentation and chromatin condensation during programmed cell death.

Materials and methods

Cells, reagents, and antibodies

HeLa cells were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated foetal calf serum, 100 U of penicillin/ml, and 0.1 mg streptomycin/ml (PAA Laboratories GmbH, Cölbe, Germany). HCT116 wt, HCT116 p53^-/-, and HCT166 Bax^-/- cells were cultured in McCoy's 5A medium supplemented as described above. Monoclonal antibodies against PARP-1 and tubulin were from PharMingen Inc. (Hamburg, Germany) and Molecular Probes (Leiden, The Netherlands) respectively. The conformation-specific anti-Bax-NT antibody was obtained from Upstate (Virginia, USA), and goat antibodies recognizing caspase-3 from R&D Systems (Wiesbaden, Germany). A monoclonal antibody against -actin, TRITC-labelled phalloidin, AdOx, staurosporine, and the nuclear dye 4'-6-diamidino-2-phenylindole (DAPI) were purchased from Sigma (Deisenhofen, Germany). The rabbit anti-serum against Acinus has been characterized previously (Schwerk et al., 2003). Secondary antibodies for immunofluorescence and Western blot experiments including chicken anti-mouse and anti-rabbit antibodies coupled to Alexafluor488 as well as horseradish peroxidase-coupled antibodies against mouse or rabbit IgG were purchased from Molecular Probes and BioRad (Munich, Germany) respectively. The broad-spectrum caspase inhibitor zVAD-fmk was obtained from ICN (Eschwege, Germany).

Cell death and cell cycle assays

For analysis of cell cycle and determination of hypodiploid nuclei formation samples were essentially prepared as described (Nicoletti et al., 1991). Briefly, cells were harvested by scraping from the plates, and extracts were prepared by lysis in buffer containing 0.1% sodium citrate, 0.1% Triton X-100, and 50 g/ml propidium iodide. Flow cytometry was performed on a FACScalibur (Becton Dickinson GmbH, Heidelberg, Germany) and analysed using CellQuest software. For each measurement a minimum of 10,000 cells were analysed.

Preparation of cell extracts and Western blot analyses

Total cell extracts were prepared in a high-salt lysis buffer containing 350 mM NaCl, 20 mM HEPES, pH 7.9, 1 mM MgCl₂, 0.5 mM EDTA, 0.1 mM EGTA, 20% glycerol, 1% NP-40, 10 mM -mercaptoethanol, 5 mM NaF, and protease inhibitors. Protein concentrations were determined using the protein assay kit from BioRad (Munich, Germany), and 20 g protein per lane were loaded onto SDS–PAGE 4–16% gradient gels and separated at 200 V. The proteins were transferred onto a 0.2 m pore size PVDF membrane (Amersham Bioscience GmbH, Freiburg, Germany) by electroblotting with a wet blot apparatus from BioRad (Munich, Germany) using standard buffers. The membranes were blocked in TBS containing 3% dry milk and 0.05% Tween-20 for 1 h, followed by an incubation with the primary antibody for 4 h. After washing the membranes extensively, the appropriate secondary antibodies (1 : 5000) were applied for 2 h. Membranes were washed extensively with TBS/0.05% Tween-20, and proteins were visualized using ECL reagents (Amersham Bioscience GmbH).

Light microscopy and immunofluorescence analyses

For microscopical analyses cells were fixed for 20 min with PBS containing 3.7% formaldehyde and permeabilized with 0.1% Triton X-100. Primary antibodies or phalloidin-TRITC conjugate were applied at 4°C overnight at the appropriate concentration. Then, cells were washed three times for 20 min and incubated with the secondary antibody (1 : 500). Finally, samples were washed extensively and mounted in fluorescent mounting medium (Dako Corporation, Carpintera, CA, USA) with 10 ng/ml DAPI. Pictures were taken on an Axiovert135 Microscope (Zeiss, Germany) equipped with OpenLab software (Improvision, Tübingen, Germany) and with confocal laser scanning microscopes (Zeiss, Germany) for immunofluorescence analyses.

Transmission electron microscopy

For transmisission electron microscopy, treated and untreated HeLa cells were harvested by trypsinization, washed with PBS, and fixed in 5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) at 4°C, then further processed, embedded, and prepared using standard methods as described (Mehlhorn et al., 1999). Electron micrographs were taken using a Zeiss 902 electron microscope.

References

1. Bartel RL & Borchardt RT. (1984) Mol. Pharmacol. 25: 418–424. | PubMed | ISI | ChemPort |
2. Bergmann S, Shatrov V, Ratter F, Schiemann S, Schulze-Osthoff K & Lehmann V. (1994) J. Immunol. 153: 1736–1743. | PubMed | ISI | ChemPort |
3. Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown JP, Sedivy JM, Kinzler KW & Vogelstein B. (1998) Science 282: 1497–1501. | Article | PubMed | ISI | ChemPort |
4. Bursch W, Hochegger K, Torok L, Marian B, Ellinger A & Hermann RS. (2000) J. Cell Sci. 113: 1189–1198. | PubMed | ISI | ChemPort |
5. Chen T & Li E. (2004) Curr. Top. Dev. Biol. 60: 55–89. | PubMed | ISI | ChemPort |
6. Chipuk JE & Green DR. (2005) Nat. Rev. Mol. Cell Biol. 6: 268–275. | Article | PubMed | ISI | ChemPort |
7. Chuikov S, Kurash JK, Wilson JR, Xiao B, Justin N, Ivanov GS, McKinney K, Tempst P, Prives C, Gamblin SJ, Barlev NA & Reinberg D. (2004) Nature 432: 353–360. | Article | PubMed | ISI | ChemPort |
8. Clarke S & Banfield K. (2001) Homocysteine in Health and Disease In: Carmel R and Jacobsen D (eds). Cambridge University Press: Cambridge, England pp. 63–78. | ChemPort |
9. Clarke S & Banfield K. (2002) Chemistry and Biology of Pteridines and Folates In: Milstein S, Kapatos G, Levine RA and Shane B (eds). Kluwer Academic Publishers: Boston pp. 557–562. | ChemPort |
10. Coleman ML, Sahai EA, Yeo M, Bosch M, Dewar A & Olson MF. (2001) Nat. Cell Biol. 3: 339–345. | Article | PubMed | ISI | ChemPort |
11. Croft DR, Coleman ML, Li S, Robertson D, Sullivan T, Stewart CL & Olson MF. (2005) J. Cell Biol. 168: 245–255. | Article | PubMed | ISI | ChemPort |
12. Cuervo AM. (2004) Mol. Cell Biochem. 263: 55–72. | Article | PubMed | ISI | ChemPort |
13. Di Simone N, Maggiano N, Caliandro D, Riccardi P, Evangelista A, Carducci B & Caruso A. (2003) Biol. Reprod. 69: 1129–1134. | Article | PubMed | ISI | ChemPort |
14. Finkelstein JD. (1998) Eur. J. Pediatr. 157 (Suppl 2): S40–S44. | PubMed | ISI | ChemPort |
15. Grzanka A, Grzanka D & Orlikowska M. (2003) Biochem. Pharmacol. 66: 1611–1617. | Article | PubMed | ISI | ChemPort |
16. Hennessey ES, Drummond DR & Sparrow JC. (1993) Biochem. J. 291: 657–671. | PubMed | ISI | ChemPort |
17. Jackson-Grusby L, Beard C, Possemato R, Tudor M, Fambrough D, Csankovszki G, Dausman J, Lee P, Wilson C, Lander E & Jaenisch R. (2001) Nat. Genet. 27: 31–39. | Article | PubMed | ISI | ChemPort |
18. Jamison JM, Gilloteaux J, Taper HS, Calderon PB & Summers JL. (2002) Biochem. Pharmacol. 63: 1773–1783. | Article | PubMed | ISI | ChemPort |
19. Karpf AR, Moore BC, Ririe TO & Jones DA. (2001) Mol. Pharmacol. 59: 751–757. | PubMed | ISI | ChemPort |
20. Kruman II, Culmsee C, Chan SL, Kruman Y, Guo Z, Penix L & Mattson MP. (2000) J. Neurosci. 20: 6920–6926. | PubMed | ISI | ChemPort |
21. Kruman II, Wersto RP, Cardozo-Pelaez F, Smilenov L, Chan SL, Chrest FJ, Emokpae R, Jr, Gorospe M & Mattson MP. (2004) Neuron 41: 549–561. | Article | PubMed | ISI | ChemPort |
22. Leist M & Jaattela M. (2001) Nat. Rev. Mol. Cell Biol. 2: 589–598. | Article | PubMed | ISI | ChemPort |
23. Martin SS & Leder P. (2001) Mol. Cell. Biol. 21: 6529–6536. | Article | PubMed | ISI | ChemPort |
24. Mehlhorn H, Mencke N & Hansen O. (1999) Parasitol. Res. 85: 625–637. | Article | PubMed | ISI | ChemPort |
25. Nicoletti I, Migliorati G, Pagliacci MC, Grignani F & Riccardi C. (1991) J. Immunol. Methods 139: 271–279. | Article | PubMed | ISI | ChemPort |
26. Proskuryakov SY, Konoplyannikov AG & Gabai VL. (2003) Exp. Cell Res. 283: 1–16. | Article | PubMed | ISI | ChemPort |
27. Rao JY, Jin YS, Zheng Q, Cheng J, Tai J & Hemstreet GP, III. (1999) J. Cell Biochem. 75: 686–697. | Article | PubMed | ISI | ChemPort |
28. Ratter F, Gassner C, Shatrov V & Lehmann V. (1999) Int. Immunol. 11: 519–527. | Article | PubMed | ISI | ChemPort |
29. Rice JC, Briggs SD, Ueberheide B, Barber CM, Shabanowitz J, Hunt DF, Shinkai Y & Allis CD. (2003) Mol. Cell 12: 1591–1598. | Article | PubMed | ISI | ChemPort |
30. Schneider-Stock R, Diab-Asseff M, Rohrbeck A, Foltzer-Jourdainne C, Boltze C, Hartig R, Schonfeld P, Roessner A & Gali-Muhtasib H. (2004) J. Pharmacol. Exp. Ther. 312: 525–536. | Article | PubMed | ISI | ChemPort |
31. Schubert HL, Blumenthal RM & Cheng X. (2003) Trends Biochem. Sci. 28: 329–335. | Article | PubMed | ISI | ChemPort |
32. Schwerk C, Prasad J, Degenhardt K, Erdjument-Bromage H, White E, Tempst P, Kidd VJ, Manley JL, Lahti JM & Reinberg D. (2003) Mol. Cell. Biol. 23: 2981–2990. | Article | PubMed | ISI | ChemPort |
33. Sebbagh M, Renvoize C, Hamelin J, Riche N, Bertoglio J & Breard J. (2001) Nat. Cell Biol. 3: 346–352. | Article | PubMed | ISI | ChemPort |
34. Sgarra R, Diana F, Rustighi A, Manfioletti G & Giancotti V. (2003) Cell Death Differ. 10: 386–389. | Article | PubMed | ISI | ChemPort |
35. Shintani T & Klionsky DJ. (2004) Science 306: 990–995. | Article | PubMed | ISI | ChemPort |
36. Song Z, Zhou Z, Uriarte S, Wang L, Kang YJ, Chen T, Barve S & McClain CJ. (2004) Hepatology 40: 989–997. | Article | PubMed | ISI | ChemPort |
37. Sperandio S, de Belle I & Bredesen DE. (2000) Proc. Natl. Acad. Sci. USA 97: 14376–14381. | Article | PubMed | ChemPort |
38. Stancheva I, Hensey C & Meehan RR. (2001) EMBO J. 20: 1963–1973. | Article | PubMed | ISI | ChemPort |
39. Suarez-Huerta N, Mosselmans R, Dumont JE & Robaye B. (2000) J. Cell Physiol. 184: 239–245. | Article | PubMed | ISI | ChemPort |
40. White SR, Williams P, Wojcik KR, Sun S, Hiemstra PS, Rabe KF & Dorscheid DR. (2001) Am. J. Respir. Cell Mol. Biol. 24: 282–294. | PubMed | ISI | ChemPort |
41. Wyllie AH & Golstein P. (2001) Proc. Natl. Acad. Sci. USA 98: 11–13. | Article | PubMed | ChemPort |
42. Zhang L, Yu J, Park BH, Kinzler KW & Vogelstein B. (2000) Science 290: 989–992. | Article | PubMed | ISI | ChemPort |
43. Zhang Y & Reinberg D. (2001) Genes Dev. 15: 2343–2360. | Article | PubMed | ISI | ChemPort |
44. Zong WX, Ditsworth D, Bauer DE, Wang ZQ & Thompson CB. (2004) Genes Dev. 18: 1272–1282. | Article | PubMed | ISI | ChemPort |

Acknowledgements

We thank Dr H Mehlhorn and M Nissen for help with electron microscopy, and Drs W Wetzel, F Essmann and members of the Häussinger lab for help with confocal microscopy. This study was supported by grants from the Deutsche Krebshilfe and the Deutsche Forschungsgemeinschaft (SFB 503).