The Fas-induced Apoptosis Analyzed by High Throughput Proteome Analysis*

The fate of cytosolic proteins was studied during Fas-induced cell death of Jurkat T-lymphocytes by proteome analysis. Among 1000 spots resolved in two-dimensional gels, comparison of control versus apoptotic cells revealed that the signal intensity of 19 spots decreased or even disappeared, whereas 38 novel spots emerged. These proteins were further analyzed with respect tode novo protein synthesis, phosphorylation status, and intracellular localization by metabolic labeling and analysis of subcellular protein fractions in combination with two-dimensional Western blots and mass spectrometry analysis of tryptic digests. We found that e.g. hsp27, hsp70B, calmodulin, and H-ras synthesis was induced upon Fas signaling. 34 proteins were affected by dephosphorylation (e.g. endoplasmin) and phosphorylation (e.g. hsc70, hsp57, and hsp90). Nuclear annexin IV translocated to the cytosol, whereas decreasing cytosolic TCP-1α became detectable in the nucleus. In addition, degradation of 12 proteins was observed; among them myosin heavy chain was identified as a novel caspase target. Fas-induced proteome alterations were compared with those of other cell death inducers, indicating specific physiological characteristics of different cell death mechanisms, consequent to as well as independent of caspase activation. Characteristic proteome alterations of apoptotic cells at early time points were found reminiscent of those of malignant cells in vivo.


From the Institute of Cancer Research, University of Vienna, A-1090 Vienna, Austria
The fate of cytosolic proteins was studied during Fasinduced cell death of Jurkat T-lymphocytes by proteome analysis. Among 1000 spots resolved in two-dimensional gels, comparison of control versus apoptotic cells revealed that the signal intensity of 19 spots decreased or even disappeared, whereas 38 novel spots emerged. These proteins were further analyzed with respect to de novo protein synthesis, phosphorylation status, and intracellular localization by metabolic labeling and analysis of subcellular protein fractions in combination with two-dimensional Western blots and mass spectrometry analysis of tryptic digests. We found that e.g. hsp27, hsp70B, calmodulin, and H-ras synthesis was induced upon Fas signaling. 34 proteins were affected by dephosphorylation (e.g. endoplasmin) and phosphorylation (e.g. hsc70, hsp57, and hsp90). Nuclear annexin IV translocated to the cytosol, whereas decreasing cytosolic TCP-1␣ became detectable in the nucleus. In addition, degradation of 12 proteins was observed; among them myosin heavy chain was identified as a novel caspase target. Fas-induced proteome alterations were compared with those of other cell death inducers, indicating specific physiological characteristics of different cell death mechanisms, consequent to as well as independent of caspase activation. Characteristic proteome alterations of apoptotic cells at early time points were found reminiscent of those of malignant cells in vivo.
Apoptosis represents a genetically programmed suicide process indispensable for the life of higher organisms. Aberrations of apoptotic mechanisms were causally implicated in severe diseases such as cancer (1,2). Because apoptosis was characterized by morphological features such as membrane blebbing, cell shrinkage, chromatin condensation, and formation of apoptotic bodies (3), it became evident that various proteins are affected during this process. The specific proteolytic activities of caspases, cysteinyl-aspartate proteases normally expressed as latent zymogens and activated during apoptosis, were recognized as responsible for many of these morphologic alterations (4,5). Signaling pathways, beginning with the activation of surface receptors ultimately resulting in the activation of caspases and hence the cellular demise, have been well characterized (6). However, several proteins became known to po-tentially inhibit (7) or promote (8,9) the onset of apoptosis by different means. Malignant diseases have been characterized by the imbalance of proliferation and cell death, resulting in a net gain of tumor mass (10,11), with regulatory cross-talk between these processes being evident (12). During malignant transformation of cells, resistance to apoptosis may develop by the expression of antiapoptotic proteins (13), thus giving spreading tumor cells a higher chance to develop metastasis (14).
In this study we addressed the question whether cells induced to undergo apoptosis would solely follow the cell death program or alternatively employ additional mechanisms unrelated with caspase activation. Potential intrinsic antiapoptotic responses could help explain how cells proceed during the establishment of resistance to apoptosis leading to diseases. To gain a broad overview of cell activities upon apoptosis induction, we performed a comprehensive proteome analysis to study Fas-mediated apoptosis in Jurkat cells. This cell system was chosen because much is known about the cell death initiation mechanisms (6), because the onset of apoptosis occurs in a quite synchronous manner, and because virtually all cells are committed to cell death and finally die, as described previously (15). To observe specificities with respect to the cell death induction mechanism, the data were compared with those obtained with staurosporine, camptothecin, and oligomycin.
The totality of proteins expressed from the genome of a cell is referred to as proteome (16). The proteome is highly dynamic and depends on many different parameters affecting cells. Proteome analysis may be performed by high resolution 2D 1 gel electrophoresis, separating proteins according to their molecular weight and electric charge, which yields highly reproducible and characteristic 2D protein patterns (17). Several studies have already focused on apoptosis-related proteins in apoptotic cells (18 -21). In this study we investigated whether proteome alterations during apoptosis originated from new protein synthesis, protein translocation, or the formation of new protein isoforms or degradation products. The focus was on cytosolic proteins; data obtained with regard to other protein fractions will be presented elsewhere. 57 of more than 1000 protein spots resolved were found to be altered during Fas-induced apoptosis. Synthesis of hsp70B, for example, was newly but transiently induced. The translation rate of several signal transduction and cytoskeletal proteins was significantly increased. Phosphorylation of some chaperones and signaling proteins was accompanied by the dephosphorylation of others. Caspase-mediated protein degradation was evidenced by different means for 12 proteins. Investigation of proteome alterations using other apoptosis inducers revealed that the same proteins were cleaved, whereas protein translation and (de)phosphorylation profiles differed somewhat. Necrosis displayed few similarities with apoptosis, such as calpain-mediated protein degradation and a potential stress response. This study demonstrates that many data on a large number of proteins may be obtained by high throughput proteome analysis, providing new insights into the fate of cells during apoptosis.
Cell Culture and Induction of Apoptosis-Jurkat cells were routinely cultivated in RPMI 1640 supplemented with 10% fetal calf serum (FCS) at 37°C in a humidified atmosphere containing 5% CO 2 . For induction of apoptosis, Jurkat cells were washed in serum-free medium and re-seeded at a density of 10 5 cells/ml in medium containing 1% FCS. Apoptosis was induced in Jurkat cells 15 min thereafter by addition of either anti-Fas (CD95) antibody (50 ng/ml) camptothecin (Sigma) to final concentrations of 5 M or Staurosporine (Calbiochem, La Jolla, CA) at 1.25 M. Induction of necrosis with oligomycin was done as described by Leist et al. (22), except that the medium contained 1% FCS and was free of glucose.
Scoring of Apoptosis and Viability-Cells were harvested by combining floating and moderately adhering cells, washed twice in phosphatebuffered saline, fixed for 10 min in 2% formaldehyde/phosphate-buffered saline at room temperature, permeabilized with phosphatebuffered saline, 0.2% Triton X-100 for 1 min, and stained in 1 g/ml Hoechst-33258 (Calbiochem) for 2 min. The percentage of cells displaying typical apoptotic nuclear morphology (crescent-shaped condensed chromatin lining the nuclear periphery; apoptotic bodies), referred to as the apoptotic index, was then determined empirically under the fluorescence microscope (Nikon Eclipse TE300). The trypan blue (Sigma) exclusion assay was used to determine cellular viability.
Subcellular Fractionation-Apoptotic and control Jurkat cells were washed twice with phosphate-buffered saline. Total cellular lysates were obtained by dissolving the cell pellet (170 g, 5 min) in 2DE sample buffer (10 M urea, 4% CHAPS, 0.5% SDS, 100 mM dithiothreitol) with subsequent sonication. For preparation of cytosol, cells were lysed in 0.05% Nonidet P-40 in hypotonic buffer (10 mM Hepes/NaOH, pH 7.4, 10 mM NaCl, 3 mM MgCl 2 ). Nuclei were pelleted at 400 ϫ g for 10 min, and the resulting supernatant was centrifuged at 100,000 ϫ g for 60 min to yield the supernatant cytosol, which was ethanol precipitated, and the cytosol pellet, which was directly dissolved in sample buffer. The nuclear matrix protein fraction was prepared as has been described in detail (23).
Metabolic Labeling with Radionuclides-In the case of the control, Jurkat cells were washed in serum-free medium and re-seeded at a density of 10 5 cells/ml in medium containing 1% FCS. After 3 h, cells were washed and re-seeded in methionine-or phosphate-free medium (Life Technologies, Inc.) and complemented with 1% FCS at a density of 10 6 cells/ml. Labeling occurred by addition of 0.8 mCi/10 7 cells 35 S protein labeling mix (NEG-072, PerkinElmer Life Sciences) or 2.5 mCi/ 10 7 cells 32 P orthophosphoric acid (NEX054, PerkinElmer Life Sciences). In the case of Fas-induction, 15 min after re-seeding Jurkat cells at a density of 10 5 cells/ml in medium containing 1% FCS, apoptosis was induced by addition of 50 ng/ml anti-CD95 antibody. Again, after 3 h, cells were washed and re-seeded in methionine-free medium complemented with 1% FCS and 50 ng/ml anti-CD95 antibody at a density of 10 6 cells/ml. Labeling occurred as in control cells. After a 2-h labeling time, cells were collected by centrifugation and fractionated.
Evaluation of 2D Data-Scanning of gels and films and comparative spot pattern analysis were accomplished with the BioImage Investigator system (BioImage, Ann Arbor, MI), using the 2-D Analyzer™ V 6.1 software package. For semiquantitative estimation of protein translation rates, the integrated optical densities of individual spots of 2D autoradiographs were calculated and expressed as a percentage of the total integrated spot intensities (%IOD). Increased translation was concluded in the case of an at least 1.5-fold increase of the %IOD of a spot.
Identification of Protein Spots-Identification of proteins by 2D Western blotting was performed as described recently (26). For mass spectrometry fingerprinting, Coomassie Blue-stained proteins were directly cut out of preparative gels. Matrix-assisted laser desorption ionization-time-of-flight of tryptic protein hydrolysates was carried out essentially as described by Fountoulakis and Langen (27). Protein identification was accomplished by means of PeptIdent software (28) made accessible by ExPASy (Expert Protein Analysis System). Proteins were considered as identified by means of mass spectrometry fingerprinting when a) at least 8 peptides b) representing at least 70% of the obtained peptide peaks c) covering at least 20% of the whole sequence gave hits with d) compatible molecular mass/pI values calculated or published in 2D data bases and e) exemplifying a significant difference in the number of matched peptides to the next potential hit. Indeed, several mass spectrographs did not yet allow unequivocal identification of the respective proteins (data not shown). Calculation of molecular weight and pI data of proteins and peptides was performed by means of Compute pI/Mw software (29) made accessible by ExPASy.

RESULTS
Jurkat cells were induced to undergo apoptosis by treatment with anti-CD95 antibody. As demonstrated previously (15), virtually all of the cells were committed to die by apoptosis; almost 100% chromatin condensation was observed after 12 h. Indeed, we were unable to recover vital cells when we repeatedly washed the cells with fresh medium 3 h after addition of antibody (data not shown).
Proteome analysis was performed with cells 5 h after antibody treatment. At this time, cells displayed an apoptotic index of about 50%, whereas the overall translation rate was found to be only slightly affected. This time was chosen to allow potential transcriptional regulations to become manifest at the protein level. Indeed, translational up-regulations became evident at that time that were no longer detectable at later time points (see Table II). Cytosol was isolated from control and apoptotic cell populations and analyzed by high resolution two-dimensional gel electrophoresis. Computer-assisted comparative analysis of the respective silver-stained spot patterns revealed 19 spots with decreasing intensity (marked by hexagons in Figs. 1 and 2), 38 new spots or spots with strongly increased intensity (marked by circles in Fig. 2), and more than 1000 spots that apparently remained unaffected (Figs. 1 and 2; see Table II). These data were reproduced in two additional independent experiments. The decrease of protein spot intensities was rather slight at 5 h after anti-Fas treatment but much more pronounced at 8 (data not shown) and 12 h after apoptosis induction (see Table II). The affected proteins were investigated by 2D Western blotting as described previously (26) and mass spectrometry of tryptic digests, resulting in the identification of those listed in Table I. Further experiments were performed to deduce whether the altered spots were due to protein synthesis, modification, translocation, or degradation. In addition, the proteome alteration profile of Fas-induced cells was compared with that of staurosporine-, camptothecin-, and oligomycin-induced cells.
Protein Synthesis-Control and apoptotic Jurkat cells were metabolically labeled for 2 h with a mixture of [ 35 S]Met and [ 35 S]Cys, beginning 3 h after antibody treatment. The resulting, previously silver-stained 2D gels were autoradiographed, which allowed us to record relative 35 S incorporation rates for each protein resolved. Comparison of the spot intensities of control and apoptotic samples, considering the respective silver-stained spot intensities, gave a measure for changes of translation rates during apoptosis. As expected, the 35 S incorporation rates of those proteins degraded during apoptosis were found to be decreased. Calmodulin, gelsolin, kinesin FIG. 1. Silver-stained 2D gel of cytosolic proteins from control Jurkat cells. Jurkat cells were washed in serumfree medium and re-seeded at a density of 10 5 cells/ml in medium containing 1% FCS. After 3 h, cells were washed and re-seeded in methionine-free medium complemented with 1% FCS at a density of 10 6 cells/ml. Labeling occurred for an additional 2 h by addition of 0.8 mCi/10 7 cells 35 S protein labeling mix. This silverstained 2D gel of the cytosol protein fraction was subsequently autoradiographed (see Fig. 3). Cytoskeletal proteins listed in Table III Table III are indicated. Circles, new spots or spots with increased intensity after onset of apoptosis; hexagons, spots displaying decreasing intensities during apoptosis. heavy chain, vimentin, and a few unknown proteins displayed increased 35 S labels, as exemplified for calmodulin in Fig. 3A. Most significantly, some of the newly appearing protein spots displayed high 35 S labels (Table II). H-ras, one of these, was found present as well in the cytosol pellet fraction of control cells. The complete absence of such proteins in other fractions of control cells indicated that de novo synthesis of hsp27 (Fig.  3A), hsp70B, and two further unknown proteins at 16 kDa/pI 6.1 and 18 kDa/pI 4.4 (Fig. 2, Table II) occurred during Fasinduced apoptosis.
Protein Modification-Protein modifications affecting the electric charge of amino acids become evident by changes of the isoelectric point (pI) of the respective protein spot, as observed in the case of ␣-actinin, displaying a more basic isoform in apoptotic cells (Fig. 2). To observe changes in protein phosphorylation, control and apoptotic Jurkat cells were metabolically labeled with [ 32 P]orthophosphate by a proceeding analogous to that described for 35 S labeling; the 32 P incorporation rates for each protein resolved were consequently recorded. In this case, comparison of the autoradiography spot intensities of control and apoptotic samples, in addition to changes of the observed pI, gave a measure for protein phosphorylation and dephosphorylation during apoptosis (Fig. 3B). More than 100 phosphorylated proteins were recorded per gel, some of which were not detectable by silver staining. During apoptosis, 14 proteins were found to be phosphorylated, including elongation factor 1␤, IB-␣, heat shock cognate 70, hsp57, hsp90, myosin light chain, and rho GDI 1, whereas 20 proteins were found to be dephosphorylated, including calreticulin, endoplasmin, kinesin heavy chain, nuclear autoantigen SP100 (lymphoid-restricted isoform), and vimentin (Tables II and III).
Protein Translocation-Protein spots disappearing or appearing as new in the cytosol fraction of apoptotic cells were assessed with respect to a potential change in localization. Their occurrence in cytosol pellet and nuclear protein fractions of control and apoptotic cells was investigated by comparison of the respective 2D spot patterns. Protein spots present in all fractions compared, such as some chaperone proteins (30), served as internal anchors for correct pattern overlay. Proteins newly appearing in the cytosol of apoptotic cells but present, for example, in the nuclear fraction of control cells were regarded as translocating. Translocation of nuclear annexin IV to the cytosol is evidenced in Fig. 3C. The presently observed translocation of yet unknown proteins at 92 kDa/pI 4.7 and 95 kDa/pI 4.6 (Table II) might correspond to data published by Ma et al. (31). Translocation from the cytosol to the nucleus is evidenced in the case of TCP-1␣ in Fig. 3C.
Protein Degradation-Several observations made by proteome analysis were indicative of protein degradation. First of all, several proteins were found to decrease or disappear during apoptosis; among others these included caspases 3 and 8, gelsolin, importin ␤-3, myosin light chain, protein phosphatase 2A, rho GDI 2, and vimentin. However, some of these, including hsp90, hsp110, protein disulfide isomerase, and tropomyosin, were found to be accumulated in the cytosol pellet fraction of apoptotic cells and were therefore not considered as degraded, but rather translocated or aggregated. Secondly, several spots appearing as new during apoptosis were found to represent degradation products. Interestingly, these spots displayed rather few 35 S labels with regard to the respective silver stain intensities (see Fig. 5, D versus B), in contrast to new protein spots emerging because of protein modification or synthesis (e.g. hsp27; Fig. 3A). This may be explained by the fact that only a few percent of the amount of any protein will be labeled within the labeling period of 2 h; consequently the great majority of any degradation product should be expected to be derived from unlabeled protein. In addition, allocation of peptides detected by mass spectrometry analysis of respective spots allowed us to deduce whether the degradation product represented the C-terminal or the N-terminal part of the cleaved protein. Calculation of theoretical molecular weight and pI values of the protein fragments resulted in data close to those obtained experimentally. This corresponding set of data is presented in more detail in the case of spectrin alpha-chain (fodrin) and rho GDI 2, whose degradation during apoptosis has been described (32), demonstrating the aptness of the method. Analogous characterizations of the degradation products of myosin heavy chain and ␣-tubulin are outlined as well. By mass fingerprinting we identified a newly appearing 120-kDa spot as an N-terminal fragment of fodrin (Fig. 2, Tables I and II) made up of the peptides covering amino acids (aa) 8 -1022. The mass fingerprint of another new spot at 150 kDa displayed all the same peptides as the 120-kDa spot in addition to another four peptides located between aa 1093-1209. Calculation of the theoretical values for molecular mass/pI of the amino acid regions covered by mass spectrometry revealed 150 kDa/pI 5.0 for aa 1-1209 and 130 kDa/pI 5.1 for aa 1-1022. These values were close to those experimentally determined, 150 kDa/pI 5.05 and 120 kDa/pI 5.1, respectively.
Whereas rho GDI 2 was found to be decreased significantly during apoptosis (Figs. 1 and 2), we detected two rho GDI 2 protein fragments in silver-stained 2D gels of cytosol from apoptotic cells, with similar pI values but differing in their apparent molecular mass (23 kDa and 21 kDa; Fig. 2, Table  IIB). However, these two protein spots could not be differentiated by mass fingerprinting, i.e. the same peptides were detected in both spots covering aa 21-164 of the intact protein.
Calculation of the theoretical molecular mass/pI of rho GDI-2 (22.9 kDa/pI 5.1) and of the aa 19 -210 fragment (20.7 kDa/pI 6.4) readily explained the observed shift of pI from the intact protein (pI 5.05) to that of the protein fragment (pI 6.7; see Figs. 1 and 2).
Another new 105-kDa spot was identified by mass spectrometry fingerprinting as an N-terminal fragment of myosin heavy chain covering aa 126 -939. Interestingly, the cytosol pellet fraction of apoptotic Jurkat cells displayed another apoptosisinduced protein spot at 95 kDa not present in cytosol fractions of control and apoptotic cells. These protein spots were identified by mass spectrometry fingerprinting as C-terminal fragments of myosin heavy chain covering aa 1220 -1932 of the protein, which consists of 1960 amino acids (Fig. 4). The generation of the cleavage product was inhibited by the treatment of cells with the caspase inhibitor z-VAD-fmk (10 M) prior to the induction of apoptosis (data not shown), indicating that caspases are responsible for the degradation of myosin heavy chain.
The identification of another newly induced protein spot as ␣-tubulin (Fig. 2, Table IIB), displaying only a little difference in molecular mass from that of the intact protein, suggested that apoptosis-induced degradation occurred. Although we cannot exclude post-translational modification of ␣-tubulin, our data suggest removal of 19 C-terminal amino acids by cleavage, because a) aa 428 -431 (LEKD) of ␣-tubulin represent a potential caspase group III (caspases 6, 8, 9, and 10) cleavage consensus site (LEXD) (33), and b) the observed molecular mass/pI shift of ␣-tubulin from 53 kDa/pI 5.1 to that of the fragment (51 kDa/pI 5.35) would be compatible with theoretical values of 50 kDa/pI 5.02 for ␣-tubulin, consisting of 450 amino acids, and If not indicated otherwise, the protein samples referred to as control and Fas correspond to those of Fig. 1 and Fig. 2, respectively. A, induced protein synthesis evidenced by 35 S autoradiography and silver staining. After silver staining, gels were equilibrated with Enlightning, dried, and exposed to x-ray films. The 35 S label corresponds to the translaton rate of the respective protein. Note that the increased translation of calmodulin is not reflected at the silver stain level and that hsp27 is not detectable in control cells. B, (de)phosphorylation evidenced by 32 P autoradiography and silver staining. 32 P labeling was performed analgous to 35 S labeling, utilizing phosphate-free medium and 2.5 mCi/10 7 cells [ 32 P]orthophosphoric acid. After silver staining, gels were dried and exposed to x-ray films. Two isoforms of the indicated proteins are encirled, the more acidic isoform labeled by autoradiography, hence phosphorylated, and the basic isoform unphosphorylated, not labeled, and observed only by silver staining. Note that dephosphorylation of endoplasmin is indicated by a decrease of the 32 P label and a more prominent appearance of the basic, dephosphorylated isoform by silver staining. Phosphorylation of heat shock cognate 70 (hsc70) is indicated by the increased 32 P label and a more prominent appearance of the more acidic, phosphorylated isoform by silver staining. C, translocation evidenced by comparison of subcellular fractions. Cytosol was obtained by the lysis of cells in 0.05% Nonidet P-40 in hypotonic buffer, pelleting of nuclei, and ultracentrifugation of the resulting supernatant at 100,000 ϫ g for 60 min; the pellet corresponds to the cytoplasm pellet. The nuclear matrix protein fraction was prepared as described in detail (23). Whereas annexin IV appeared as new in the cytosol during apoptosis, it disappeared from the nuclear matrix protein fraction. On the other hand, TCP-1␣ decreased in the cytosol during apoptosis, concomitantly becoming more prominent in the nuclear matrix fraction. During chromatin condensation, the nuclear matrix fraction displayed various other spot alterations that will be described in detail elsewhere. 47.9 kDa/pI 5.42 for a fragment spanning aa 1-431.
Protein Alterations Dependent on the Mode of Cell Death Induction-To discriminate proteins altered specifically during induction of the Fas signaling pathway, we compared the re-spective protein alterations to those of staurosporine-and camptothecin-treated apoptotic Jurkat cells (Fig. 5, L and M ;  Table II). No differences were observed with respect to protein degradation products potentially generated by caspase activi-
On the other hand, the proteome alteration profile of oligomycin-treated, thus necrotic, Jurkat cells was analyzed (Fig.  5K, Table II). As expected, no caspase-mediated degradation events were observed; however, the calpain-generated 150-kDa fodrin fragment (34) was readily identified in necrotic Jurkat cells (Fig. 5K), indicating calpain activity during necrosis. The same fragment was the only degradation product observed in Fas-induced Jurkat cells pretreated with caspase inhibitor z-VAD-fmk (data not shown), confirming that no caspase generated this fragment. The most apparent modifications during necrosis, such as phosphorylation of IB-␣, myosin light chain, and elongation factor 1␤ or dephosphorylation of endoplasmin, were observed as during apoptotic cell death. Interestingly, hsp90 and hsp110 were again found to be reduced in the cytosol of necrotic cells.
Establishment of a Data Base by Integration of Proteome Analysis Data-As exemplified above, proteins may be affected during apoptosis by quite different means, with different kinetics, and be dependent on the cell death induction pathway. Some data presented so far were selected for the most significantly altered characteristics observed. However, all variables, such as relative protein amount (determined by silver staining), translation rate ( 35 S label), and phosphorylation state ( 32 P label), were recorded for each detectable protein of the different subcellular fractions analyzed. Considering different time points and different cell death induction agents as well, finally a multidimensional set of data was obtained. Related gel sections, representing a small part of the final generated data base, are illustrated in Fig. 5. Proteins allocated in the 2D system may be searched by this means for apoptosis-dependent alterations. This approach, i.e. searching for alterations of selected proteins of interest, is exemplified in the case of two protein families, cytoskeletal proteins and chaperones, which   Tables I and II. nd, not detectable; AR, autoradiography; W, 2D Western blotting; M, mass spectrometry analysis of tryptic peptides; P, position in 2D gels in relation to published 2D data bases; A, amino acid sequencing; ϭ, intensity not significantly changed; ϩ, spot present; Ϫ, spot absent; Ͼ, increased intensity; Ͻ, decreased intensity; Fas, anti-CD95 antibody-treated; con, control; vs., comparison with. (*) refers to Ref. 28; (**) refers to the respective fragments.

Protein
Identified by 35 S AR (Fas 5 h vs. con) 32 P AR (Fas 5 h vs. con) Degradation evident Cytoskeletal proteins ␤-Actin were found to be predominantly affected. As listed in Table III, six cytoskeletal proteins were found to be cleaved, two were found to be dephosphorylated, and two were found to be translationally up-regulated. On the other hand, no degradation of chaperone proteins became evident in our system; still three members were found to be phosphorylated, two were found to be dephosphorylated, two were found to be translationally upregulated, and two were found to be newly induced (Table III).
To sum up, our proteome analysis data suggest that these two protein families represent key players during active cell death.

DISCUSSION
It was the aim of this study to gain more comprehensive insights into the cellular mechanisms activated during apoptosis by using the proteome analysis. The proteome of a cell is highly dependent on the conditions to which the cell is exposed and may respond in a quite complex manner. Upon Fas induction, 57 protein spots displayed significant and reproducible alterations, whereas more than 1000 cytosolic protein spots were apparently unaffected in silver-stained 2D gels. This gives an estimate of around 5% of the cytosolic proteins affected during apoptosis. We registered new synthesis of, for example, hsp27 and hsp70B; increased translation of, for example, calmodulin, H-ras, and vimentin; phosphorylation of, for example, IB-␣, hsc70, hsp90, and myosin light chain; dephosphorylation of, for example, endoplasmin, kinesin heavy chain, and vimentin; and translocation of TCP-1a and annexin IV. In addition, proteins below silver stain sensitivity displayed altered phosphorylation rates as detected by 32 P autoradiography. Some effects were found to be transient and/or specific with respect to the cell death induction pathways, such as the induction of hsp27 and hsp70B and the up-regulation of H-ras, respectively. Experimental data from spot intensities of selected protein fractions, at different time points after induction of apoptosis and with different apoptosis inducers, have been collected in a data base, which might enable new mechanistic insights.
The best characterized consequence of apoptosis induction so far, i.e. the activation of caspases, became evident by the detection of an active caspase 3 isoform (Fig. 2) and the disappearance of the procaspases 3 and 8 (Fig. 1). Consequent to the caspase activation was the disappearance of eight proteins (besides the procaspases) and the appearance of 17 degradation products, which was not observed when the caspase inhibitor z-VAD-fmk was applied. Most protein degradations, such as those of actin, fodrin, gelsolin, protein phosphatase 2A, and rho GDI 2, have been described recently (35); however, we also observed the degradation of myosin heavy chain and obtained the first evidence for the degradation of tubulin ␣. Degradation of cytoskeletal and cytoskeleton-regulating proteins was found to be responsible for most of the prominent spot alterations detected by silver staining.
A stress response of dying cells is reflected by another set of prominent alterations, such as induced synthesis of calmodulin and hsp27, phosphorylation of IB-␣ and myosin light chain (36), and dephosphorylation of endoplasmin (37). The decrease of cytosolic hsp90 and hsp110 might be due partially to aggregation caused by binding to non-native proteins emerging during cell death, because they became more prominent in other cell fractions during apoptosis (data not shown). Besides induced synthesis, these stress-related alterations were also observed during necrosis. Because no caspases are activated during necrosis, this is considered to be a caspase-independent cellular response upon induction of cell death, which might be a consequence of a rise in intracellular Ca 2ϩ signaling (38) or the generation of non-native proteins.
The broad spectrum of protein alterations observed during apoptosis might justify some speculations upon further biological implications of cell death signaling. Stress-induced decrease of cytosolic hsp90 has been demonstrated to be able to uncover cryptic genetic variations, leading to the expression of new traits in Drosophila (39). hsp90 decreased strikingly during cell death, with the most acid isoform becoming undetectable after 12 h (data not shown). This observation might indicate the onset of microevolution upon induction of cell death. Cell death has been demonstrated to occur in malignant tumors in addition to cell proliferation (11). Considering a great number of affected cells and a prolonged time period, cell death signaling might be considered as a Darwinian driving force to select for cell death incompetence, which is a common feature of many malignant cells (7,40). Several of our data would support this interpretation. hsp27, which was presently found to be induced by anti-Fas treatment, has been described to inhibit the mitochondria-mediated apoptosis induction pathway (41). In addition, some kinds of tumor cells have been demonstrated to exhibit high levels of hsp27 (42). Highly expressed H-ras, one of the most common features of cancer cells, may have positive or negative effects on cell growth, differentiation, and death during multistage carcinogenesis (43). H-ras was found to be induced during Fas-induced apoptosis and might represent another antiapoptotic response mediated by Akt kinase, consequent to the activation of PI3-kinase or NF-B (44). The presently observed phosphorylation of IB-␣ (Table IIB) would be compatible with a transient activation of NF-B. In addition, some antiapoptotic cell responses might also be indicated by the induction of calmodulin, a slight induction of grp78, activation of endoplasmin by dephosphorylation (45), and phosphorylation of hsc70. Calmodulin signaling was reported to activate downstream Akt kinase (46), thereby exerting antiapoptotic activity (44). hsc70 has been demonstrated to represent a target for tyrosine kinases (47) and to cooperate with BAG-1 in mediating antiapoptotic activities (48). These data show that transient events observed in cells on their way to cell death are similar to those established in malignant cells. Thus, our data may indicate that chronic cell death induction, which never results in death in virtually all cells affected in vivo, might be considered as a driving force for the formation of cell death-incompetent cells, paving the way for the formation of malignant tumors.