Multiple apoptotic death types triggered through activation of separate ...

3363

Journal of Cell Science 107, 3363-3377 (1994) Printed in Great Britain © The Company of Biologists Limited 1994

Multiple apoptotic death types triggered through activation of separate pathways by cAMP and inhibitors of protein phosphatases in one (IPC leukemia) cell line Bjørn Tore Gjertsen1,*, Lill Irene Cressey1, Sandrine Ruchaud2, Gunnar Houge1, Michel Lanotte2 and Stein Ove Døskeland1,* 1University of Bergen, Medical school, Department of Anatomy and Cell Biology, Årstadveien 19, N-5009 Bergen, Norway 2Institut National de la Santé et de la Recherche Médicale U301, Institute of Hematology, Centre Hayem, Hôpital St-Louis,

75010,

Paris, France *Authors for correspondence

SUMMARY The protein phosphatase inhibitors okadaic acid and calyculin A at moderate concentrations induced three types of apoptotic promyelocytic leukemia cell death, distinct with respect to ultrastructure and polynucleotide fragmentation. Calyculin A at higher concentrations (>50 nM) induced a non-apoptotic death type with high ATP and pronounced micromitochondriosis. This suggests that protein phosphorylation pathways are involved in the triggering of several death pathways. Activation of the cAMP kinase induced yet another apoptotic death type, preferentially affecting cells in S-phase. In fact, cAMP acted in two ways to stop IPC promyelocyte proliferation: (1) block in late G1 (preventing new cells from entering DNA replication); and (2) induction of apoptosis in S-phase. cAMP and

INTRODUCTION Cell death has been considered to be either necrotic or apoptotic (Kerr et al., 1972; Wyllie et al., 1980). In necrosis rapid damage to the surface membrane or the ATP production is accompanied by mitochondrial swelling, inability to exclude Trypan Blue, and release of inflammation-promoting factors. Apoptotic cell death inflicts minimal damage on neighboring cells and is described as being morphologically stereotypic with cell volume decrease, characteristic hypercondensation of chromatin, and often nuclear and cell fragmentation. Internucleosomal DNA fragmentation may (Wyllie, 1980) or may not (Bøe et al., 1991; Cohen et al., 1992; Oberhammer et al., 1992) be associated with apoptotic cell death. For further discussion see recent reviews on cell death (Sen, 1992; Schwartz et al., 1993; Schwartzman and Cidlowski, 1993; Williams and Smith, 1993). Recently, specific limited cleavage of 28 S rRNA was reported in cAMP-induced death of a promyelocytic leukemia (IPC-81) cell line (Houge et al., 1993). Little is known about the role of protein phosphorylation in the control of cell survival. The IPC-81 cells were chosen for studies of protein phosphorylation involvement in cell elimi-

phosphatase inhibitors acted via distinct pathways. The inhibitors suppressed cAMP-induced death, but only at concentrations high enough to commit the cells to alternative, less conspicuous death types. The tumor-promoting activity of okadaic acid and calyculin A may therefore not be by protection against apoptosis. DNA fragmentation correlated with the novel feature of limited 28 S rRNA cleavage, suggesting co-ordinated polynucleotide cleavage, possibly directed against illegitimate polynucleotides, in some apoptotic death types. Key words: cell death, cell survival, cell cycle, phosphoprotein phosphatase, adenosine 3′,5′ cyclic monophosphate (cAMP)

nation since they undergo swift death in response to the protein phosphatase inhibitor okadaic acid (Bøe et al., 1991) or activation of the cAMP-dependent protein kinase (Lanotte et al., 1991; Vintermyr et al., 1993). In the present study they were exposed to cAMP or phosphatase inhibitors (calyculin A, okadaic acid), and the ensuing cell death characterized with respect to ultrastructure, biochemically and cytochemically evaluated DNA fragmentation, specific 28 S rRNA fragmentation, ATP level, ability to exclude Trypan Blue, and mitochondrial size. Most effects of cAMP in eukaryotic cells are mediated by the cAMP-dependent protein kinase (for reviews see Taylor et al., 1990; Døskeland et al., 1993; Spaulding, 1993). Evidence will be presented that cAMP induced IPC cell apoptosis, preferentially affecting cells in S-phase. Furthermore, cAMP appeared to block the cells in late G1, revealing a dual and unprecedented system for cAMP-induced cell compartment contraction. Among the mammalian serine/threonine protein phosphatases (Cohen, 1989; Bollen and Stalmans, 1992) at least types 1, 2A and X (Brewis et al., 1993) are sensitive to microbial inhibitors such as okadaic acid, microcystin and

3364 B. T. Gjertsen and others calyculin A. In the present study four death types, biochemically and morphologically distinct from that induced by cAMP, were observed in IPC cells treated with okadaic acid or low, moderate, and high concentrations of calyculin A. To probe for common death-inducing phosphorylation pathways between cAMP and phosphatase inhibitors they were tested for synergism, for similarity in sensitivity towards inhibitors of protein kinases and protein synthesis, and finally phosphatase inhibitor action was compared in IPC cells with normal and subresponsive (Gjertsen et al., 1993) cAMP kinase. It will be concluded that separate protein phosphorylation pathways can induce distinct types of death in one cell line. Both okadaic acid and calyculin A have been reported to protect cells against apoptosis, and their tumor-promoting action has been related to suppression of cell death (Song and Lavin, 1993). It will be shown that the phosphatase inhibitors offered only ‘pseudoprotection’ against cAMP, by inducing IPC cell death types less conspicuous than the one induced by cAMP.

MATERIALS AND METHODS Reagents and media Calyculin A and okadaic acid (sodium salt) were from LC Services (MA). The ionophore A23187, cysteamine-2HCl, L-buthionine[S,R]-sulfoximine, ExtrAvidin-tetramethylrhodamine isothiocyanate (ExtrAvidin-TRITC), 4′,6-diamidino-2-phenylindole (DAPI) and bisbenzimide H 33258 were from Sigma. N-[2-(methylamino)-ethyl]-5isoquinolinesulfonamide dihydrochloride (H-8) was from Gibco. 1(5-isoquinolinesulfonyl)-2-methylpiperazine dihydrochloride (H-7), quin2/AM and bisbenzimide H 33342 were from Calbiochem. KT5720 and KT5823 were from Kamiya Biomedical Co. cAMP analogs were obtained from sources described (Gjertsen et al., 1993). All chemicals (in particular calyculin A, 8-chlorophenylthio-cAMP and bisbenzimide H 33258) were protected from light, also when in solution and after addition to the cell culturing media. Cell culturing and handling The rat promyelocytic leukemia cell lines IPC and its subclone IPCRID336 with mutant cAMP-resistant cAMP kinase I (Gjertsen et al., 1993) were grown in Dulbecco’s Modified Eagle Medium with 10% heat-inactivated horse serum. For some experiments (pulse labeling with [ 35S]methionine) the medium was without methionine and with 0.5% dialyzed serum. For the experiments to be reported the cells were in logarithmic growth at densities from 2×105 to 8×105 per ml, routinely 4×105 cells/ml. For studies of clonogenicity the cells (4×105/ml) were preincubated in the absence or presence of calyculin A, diluted to 800 cells/ml, and plated (Lanotte et al., 1986) in collagen gels (1.5 ml per dish). Colonies formed were counted after 4 days. For studies of reversibility of action of apoptotic agents, cells pretreated with cAMP analog, cycloheximide or phosphatase inhibitors were washed twice (by centrifugation at 800 g, 4.5 minutes) in 20-fold excess of medium. To avoid cell loss during centrifugation some samples (pretreated with calyculin A) were grown within Millicell CM culture plate inserts (0.4 µm pore size). Washing was achieved by moving the insert (containing 0.1 ml medium with calyculin A) first to a vial with 75 ml fresh medium, and thereafter to one with 5 ml fresh medium. The cells were counted and monitored for Trypan Blue (present at 0.2%) staining about 24 hours after the start of washing. In some cases the cells were made to adhere to fibronectin-coated wells and cultured in an incubation chamber integrated in a Zeiss Axiovert microscope equipped for microinjection and video recording (Vintermyr et al., 1993). This

allowed the continuous monitoring of cell outline, duration of mitosis (phase contrast) and of the chromatin distribution (fluorescent DNA dye) in cells exposed to phosphatase inhibitors (microinjected into cells or applied close to cells by a micropipette). Morphological studies The cells were fixed by mixing with 9 volumes of prewarmed (37°C) glutaraldehyde (1.5%, v/v) in 0.1 M Na-cacodylate buffer (pH 7.4). Further processing and electron microscopy were as described by Bøe et al. (1991). Morphometric analysis of cell and organelle volumes was according to the methods described by Gundersen et al. (1988). Each morphometrically calculated volume was based on scrutinization of 300-600 sectioned cells. The average values derived directly from morphometry were, both for IPC cells and rat hepatocytes, 7% lower than the mean volume as determined by Coulter counter analysis assuming spherical cell shape (Bøe et al., 1991). This 7% shrinkage was corrected for before using morphometrically determined volumes in the calculation of e.g. cell ATP concentrations. Relative volumes were determined assuming similar shrinkage of cells regardless of treatment. For scoring of chromatin condensation and distribution, the cells (after fixation in glutaraldehyde and washing in phosphate buffered saline) were incubated with DNAspecific dye and subjected to fluorescence microscopy, examining from 200-300 cells. In situ DNA fragmentation was determined essentially according to Gavrieli et al. (1992), except that avidin was tagged with rhodamine rather than peroxidase. The use of fluorescently labeled avidin avoided difficulties caused by any endogenous peroxidase and allowed the simultaneous visualization of bulk DNA with the DNAspecific dye DAPI. Autoradiographic labeling and determination of cell cycle phase durations Routinely, [methyl-3H]thymidine (0.2 µCi) was presented as a pulse (0.3 hours or most often 0.7 hours). The labeling was arrested by extensive washing (twice) in a 20-fold excess of conditioned medium from parallel cultures made 2 µM in unlabeled thymidine. After washing the incubations continued for various periods of time, to obtain labeled cells in all major phases of the cell cycle. The cells were next treated for 4 hours with 8-chlorophenylthio-cAMP, then cytocentrifuged onto slides, which were fixed, dipped in photographic emulsion (K.5D, Ilford), exposed for 2 weeks, and stained with MayGrünewald-Giemsa solution. The number of labeled and unlabeled cells with apoptotic or normal appearance was evaluated. The IPC cells doubled in 12 hours under standard logarithmic growth conditions. A short (10 minute) pulse of [3H]thymidine resulted in 61% autoradiographically labeled cells. The efficient labeling time was very short (about 5 minutes, since pulses of less than 5 minutes gave few clearly labeled cells) and could be ignored for the purpose of calculation of S-phase duration. Since only cells in S-phase are labeled by [ 3H]thymidine, the percentage of labeled cells should reflect the percentage of the cell doubling time occupied by Sphase cells. The S-phase duration was thus tS = 12 hours × 0.61 = 7.3 hours. The duration of G1+G2+M was 4.7 hours (12-7.3). The time (tM) taken for the M-phase (from early to middle prophase through anaphase) as estimated from video recordings of cells or from the fraction of mitoses labeled by a short pulse of [3H]thymidine [MLI], i.e. tM = 12 hours × MLI/100 ranged between 11 and 15 minutes (mean value 13 minutes, i.e. about 0.2 hours). This means that tG1 + tG2 was 4.5 hours (4.7 hours-0.2 hours). The ratio tG1/tG2 was judged from the relative abundance (2.7:1) of the 2n and 4n peaks upon flow cytofluorimetry. This allowed the calculation of tG2 (1/3.7 × 4.5 hours = 1.2 hours) and of tG1 (3.3 hours). Preparation and analysis of RNA and DNA RNA was prepared, electrophoresed, blotted and hybridized with 5′32P-labeled oligonucleotides complementary to various parts of rat 28

Multiple phosphorylation-induced death types 3365 S rRNA (Houge et al., 1993). For preparation of DNA, the cell pellets were lysed in 10 mM Tris buffer, pH 8.0, containing 0.5% sodium dodecyl sulfate, 100 mM EDTA, and 10 mM EGTA. DNase-free RNase A was added (20 µg/ml of lysate), and incubated at 37°C with gentle shaking, followed by addition of proteinase K (100 µg/ml of lysate) and a further 10 hours incubation at 56°C. The DNA was extracted with water-saturated phenol, precipitated with 2 volumes ethanol/0.2 volume 10 M ammonium acetate, pH 5.6, washed in 2 volumes of 70% ethanol, and dissolved in 10 mM Tris-HCl, pH 8.0, containing 1 mM EDTA. For (semi)quantitative estimation of DNA fragmentation the DNA samples (10 µg) were electrophoresed in 1.5% agarose gels and stained with ethidium bromide (0.5 µg/ml). Each UV-illuminated gel was photographed using 3 different exposure times (10, 20, 40 seconds) on a Polaroid 665 positive/negative film. The negatives were scanned on a LKB Ultroscan XL Laser Densitometer and the density plot for each lane integrated. DNA from nontreated (control) cells migrated 1.2 to 1.8 cm under the conditions used, and was considered nondegraded for operational purposes. DNA migrating 1.8 cm to 15 cm was considered degraded. This distinction was simplified by the fact that only a minor proportion of applied DNA migrated between 1.5 and 3 cm, whether from control cells or cells with degraded DNA. DNA degradation was also estimated as efflux of DNA (i.e. radioactivity in the 5% trichloroacetic acid precipitate of the medium) from cells prelabeled with [3H]thymidine. Determination of cellular ATP and estimation of protein synthetic activity For ATP/ADP determination the IPC cell suspension (1 ml) was quickly mixed with 25 µl of aqueous 70% perchloric acid, and the neutralized high-speed supernatant assayed for ATP and ADP by a luciferin-luciferase method. Protein synthetic activity was assayed by presenting [35S]methionine (200 µCi/ml) to the cells for 0.5 hours before cell harvesting. The amount of labeled protein was either determined by scintillation counting of trichloroacetic acid precipitates of the cell lysates, by autoradiography of one-dimensional SDS-polyacrylamide gel electrophoresis separations, or (occasionally) by autoradiography of two-dimensional gels.

RESULTS Distinct IPC cell death types induced by calyculin A, okadaic acid and cAMP As shown in Figs 1-8 and summarized in Table 1, agents acting via protein phosphorylation pathways, i.e. cAMP analog and the protein phosphatase inhibitors okadaic acid and calyculin A, induced IPC cell death with different biochemical and morphological features. The diagnosis of apoptotic cell death [ACD] and its differentiation from necrotic cell death [NCD] is based mainly on morphological criteria. Therefore, representative ultrastructural features of the IPC cell death types are presented first (Figs 1, 2). Although most micrographs shown represent cells treated for 6 hours, the summary of morphological features (Table 1) is based on treatment periods from 10 minutes up to 9 hours. The death types induced by cAMP, okadaic acid and 50 nM calyculin A was not apoptotic since the cell was swollen and the chromatin was uniformly and only moderately condensed. On the other hand, the death was not

Fig. 1. Ultrastructure of IPC cells exposed to cAMP or okadaic acid. Representative transmission electron micrographs of IPC cells grown in standard medium (a), for 6 hours with added 1 µM okadaic acid (b), or 200 µM 8-chlorophenylthio-cAMP (c). Nuclear fragmentation, invaginations of the surface membrane and central segregation of mitochondria were particularly pronounced in cells treated with cAMP analog (c). In cells exposed to okadaic acid, the mitochondria and vesicles typically dislocated towards one cell pole and the chromatin was hypercondensed also in the absence of nuclear fragmentation (b). Bars, 1 µm.

3366 B. T. Gjertsen and others Table 1. Classification of death types in IPC cells Calyculin A Inducing agent

cAMP

Okadaic acid

50 nM calyculin A as pseudonecrosis [pNCD]. The ATP level typically drops in necrosis (Wyllie et al., 1980). In cells treated with calyculin A (100 nM) the ATP level (on a per cell basis) actually increased between 1 and 3 hours of treatment (Fig. 3a), indicating functioning ATP synthesis and providing yet another distinction between NCD and pNCD. The ATP concentration dropped slowly in response to cAMP challenge (Fig. 3b). The ADP level was about 10% of the ATP level, except in cells treated for 4 hours or longer with cAMP analog, in which ADP amounted to 15-35% of the ATP level (data not shown). This rise in ADP was insufficient to account for the drop in ATP, implying a late loss of adenine nucleotides in cAMP-induced death. The death classes (NCD, pNCD, ACD) were subclassified (Table 1) according to the presence and degree of fragmentation of DNA (d,d+,d++), nucleus (n,n +,n++), and cell (c,c+,c++). The cell fragmentation was judged by transmission electron microscopy, and was always preceded in time by evaginations containing organelles like mitochondria. It is noteworthy that cells treated with >50 nM calyculin A showed cell fragmentation, budding (Fig. 2d,e) and loss of microvilli, features generally associated with apoptotic cell death (Table 1). Nuclear fragmentation was revealed by electron microscopy (Figs 1, 2) and by fluorescence microscopy of cells stained with DNA-specific dye (Fig. 4). It was most pronounced in apoptotic death induced by cAMP (Figs 1, 2; Table 1). DNA fragmentation was judged by cytological in situ labeling of free DNA ends (Fig. 4) and by electrophoresis of

isolated DNA. In response to cAMP, okadaic acid or moderate concentrations of calyculin A the degradation was considerable and the fragment sizes corresponded to multimers of the internucleosomal length of about 185 base pairs (Figs 5, 6). However, pseudonecrotic cells treated with >50 nM calyculin A showed no degradation of DNA by either in situ labeling (not shown) or electrophoresis (Fig. 5). IPC cells, human NB4 leukemia cells, thymocytes and mammary MCF-7 carcinoma cells were exposed to cytotoxic agents like tubercidin, 2-Cl-adenosine, and 8-Cl-adenosine and their ultrastructure examined. All death types observed in response to these agents could be classified morphologically within one of the four phosphatase-inhibitor induced classes in Table 1, suggesting at least some generality of the death types proposed. Either of the death types induced by phosphatase inhibitors appeared ‘spontaneously’ (in from 0.05-0.5% of the control cells), suggesting that phosphatase inhibitors could accelerate or mimic spontaneous death processes. The cAMPinduced death type was not observed in control cells. The latency between the application of phosphatase inhibitor and its effect was studied by applying calyculin A solution (1 mM) by a micropipette 20-30 µm from IPC cells. The latter typically reacted with violent blebbing within 15 seconds (data not shown). Similar changes were induced in less than one minute after intracellular injection of the impermeable phosphatase inhibitor microcystin-LR (0.2 mM in the injected solution). It is concluded that cell death features can be induced nearly immediately by phosphatase inhibitors, and that altered protein phosphorylation is closely upstream of the putative final trigger(s) of cell death. Internucleosomal DNA cleavage correlated with limited 28 S rRNA cleavage, but could be dissociated from chromatin condensation and nuclear envelope disintegration The different extent of DNA fragmentation between death types occurring in one cell line (Fig. 5; Table 1) provided an experimental opportunity to correlate DNA fragmentation with chromatin condensation (Arends et al., 1990; Sun et al., 1994), nuclear envelope intactness (Ucker et al., 1992), and with the recently described apoptosis-associated parameter of limited 28 S rRNA fragmentation (Houge et al., 1993). Chromatin condensation preceded DNA fragmentation in cells treated with okadaic acid (Fig. 7). Within single cells treated with 50 nM calyculin A the nuclear envelope disappeared, but the DNA did not undergo internucleosomal fragmentation. In contrast, the DNA of cAMP-treated cells was degraded even if the nuclear envelope was present. This means that DNA fragmentation can occur in nuclei with preserved envelope, and that even complete absence of the latter does not necessarily lead to DNA fragmentation. In order to know if specific, limited cleavage of 28 S rRNA

3368 B. T. Gjertsen and others

Fig. 4. Comparison of subnuclear distribution of condensed chromatin and fragmented DNA in cells treated with cAMP analog or calyculin A. IPC cells were treated for 12 hours with 50 µM 8-chlorophenylthio-cAMP (a-c) or 1 nM calyculin A (d-f), fixed to glass slides, and free DNA ends nick end-labeled with biotinylated poly dUTP, which was detected with rhodamine-labeled avidin (ExtrAvidin-TRITC). Bulk DNA was stained with DAPI. (a,d) Staining of bulk DNA, (b,e) localization of fragmented (end-labeled) DNA, (c,f) double exposures of a/b and d/e, respectively. Note hypercondensed DNA without accompanying fragmentation (d-f), and DNA fragmentation in non-condensed chromatin. Note also the variable chromatin configurations in cells treated with 1 nM calyculin A.

was restricted to cAMP-induced IPC cell death, RNA was isolated from cells treated with other death-inducing agents,

electrophoresed and either stained with ethidium bromide or transferred to northern blots for hybridization against probes

Multiple phosphorylation-induced death types 3369

Fig. 5. Semiquantitative evaluation of internucleosomal DNA fragmentation in response to cAMP analog and phosphatase inhibitors. DNA was isolated from cells treated for various periods of time with 0.1 nM, 0.3 nM or 100 nM calyculin A (upper panel), 1 µM okadaic acid (middle panel), or 200 µM 8-chlorophenylthio-cAMP (lower panel). The DNA (routinely about 10 µg) was loaded into shallow (1×10×5 mm) wells in 1.5% agarose gels, stained with ethidium bromide after electrophoresis, and photographed with positive/negative film. Ordinary photos of the gels are shown on the left. The uppermost lane shows the staining pattern of BstEII-digested λ DNA used as fragment length standard. The right hand side (main panels) show laser densitometric scans of the part of the negative corresponding to one lane of each gel. Note the absence of DNA migrating further than about 1.5 cm into the gel in samples from cells treated with either very high (100 nM) or very low (0.3 nM or lower) calyculin A (upper panel), the marked accumulation of nucleosomal-sized DNA in okadaic acid-treated cells (middle panel), and the very marked accumulation of such fragments in cells treated with cAMP analog (lower panel).

corresponding to selected regions of the 28 S rRNA molecule. Preferential depletion of 28 S rRNA relative to 18 S rRNA was observed in cells treated with okadaic acid as well as with cAMP analog (Fig. 8). In all cases the 28 S rRNA degradation correlated closely with the degree of DNA fragmentation, as determined in parallel samples (not shown). This serves to link cytoplasmic rRNA cleavage to nuclear internucleosomal DNA cleavage, and suggests that limited polynucleotide cleavage of

both RNA and DNA occurs in certain cell death types. Our previous observation (Houge et al., 1993) that the cAMPinduced cleavage of 28 S rRNA is limited to the variable regions D2 (V3) and D8 (V13) of the molecule was confirmed and extended using several probes against subregions of D2 and D8. Only hybridization with probes against the extreme 5′ and 3′ ends of 28 S rRNA is shown. Cleavage at two sites in D2 (one near nucleotide 890 and another near nucleotide 1190) led


% nonfragmented DNA

100

a 50

% nuclei with hypercondensed chromatin

0 100

b 50

0 0

3

6

9

12

Time of incubation (hours)

Fig. 6. Internucleosomal DNA fragmentation in IPC cells treated with low nM concentrations of calyculin A. The figure shows ethidium bromide staining patterns of electrophoresed DNA from cells treated for 6, 12 or 24 hours with 1 nM calyculin A (three right hand lanes), or for 3, 6 or 12 hours with 3 nM calyculin A (three left lanes). The sizes of the DNA standards (middle lane) are shown on the right. A DNA ladder of fragments, multiples of the nucleosomal 0.185 kb size, became evident after 6 hours with 3 nM or 24 hours with 1 nM calyculin A. Further details are given in the legend to Fig. 5.

to the excision of a 0.3 kb fragment. Closely spaced cleavages (between bases 3200 and 3300) occurred in D8. The probe against the 3′ region of 28 S rRNA revealed the 3.6/3.9 kb doublet resulting from cleavage within D2 and ~1.5 kb fragments (not resolved) from cleavage in D8. Such fragments were found whether the cells had been treated with cAMP or okadaic acid. A probe complementary to the 5′ region of 28 S rRNA revealed ~3.3 kb fragments (not resolved) expected from cleavage in D8, in cells treated with either okadaic acid or cycloheximide. These fragments disappeared in advanced cAMP-induced apoptosis, but were clearly present in cells treated for 3-5 hours with cAMP analog (data not shown). It is concluded that the same sites in 28 S rRNA are cleaved in response to cAMP, cycloheximide and okadaic acid. The effect of cycloheximide shows that 28 S rRNA degradation could occur in the absence of active protein synthesis, i.e. affect inactive ribosomes. Treatment with 5 nM calyculin A gave a 28 S rRNA fragmentation pattern similar to that observed in response to cAMP or okadaic acid (not shown). Cleavages associated with normal turnover of 28 S rRNA, with aging of the cell extracts or with calyculin A treatment of the cells, did not occur in D2, but was seen in the D8 region. Treatment with 50 nM calyculin A gave rise to a distinct 28 S rRNA degradation pattern (Fig. 8). Obviously, this 28 S rRNA degradation did not correlate with DNA fragmentation, which was absent in such cells (Figs 5, 7). It is concluded that pNCD differed from the other death types also with respect to rRNA fragmentation.

Fig. 7. The time course of chromatin condensation and of DNA fragmentation in response to cAMP analog and phosphatase inhibitors. The figure shows the time course of internucleosomal DNA fragmentation (a) and of chromatin condensation (b) in IPC cells exposed to 100 nM calyculin A (s), 1 µM okadaic acid (n), or 200 µM 8-chlorophenylthio-cAMP (h). The data are based on four separate experiments in which parallel samples were taken for determination of DNA fragmentation (see the legend to Fig. 5 for details) and chromatin condensation (see the legend to Fig. 4 for details). The s.e.m. is represented by error bars.

cAMP contracts the IPC cell compartment by a dual mechanism: induction of death preferentially in the S-phase and blocking of cell cycle traversal in late G1 IPC cells exposed to brief (20 minute) pulses of [3H]thymidine at half-hourly intervals after cAMP challenge showed a decreased percentage of autoradiographically labeled nuclei, becoming significant already after 2.5 hours. By flow cytometric analysis the proportion of cells in S-phase (between 2n and 4n DNA content) decreased and cells in G1 (2n) increased. It is concluded that cAMP blocked the IPC cells at a (restriction) point closely before the onset of DNA replication. Many obviously apoptotic nuclei were labeled in cAMPtreated cells that had been exposed to a [3H]thymidine pulse just before fixation (Fig. 9). This indicated that S-phase cells could undergo apoptosis and led us to examine closer the cell cycle dependence of cAMP susceptibility (Fig. 10). It was reasoned that the first cells to undergo apoptosis (e.g. those apoptotic already after 4 hours) had been in the most vulnerable part of their cycle during the first 2 hours of cAMP exposure, i.e. the period required to irreversibly commit about half the IPC cells to death (Fig. 11). To identify the cycle position of the cells during this critical period, they were prelabeled by a pulse (0.7 hours) of [3H]thymidine, starting either 12.3, 9.3, 6.3, 3.3 or 0.3 hours before the cAMP challenge. Note that a narrower ‘window’ of the cell cycle was defined

Multiple phosphorylation-induced death types 3371

Fig. 8. Similar cleavages of 28 S rRNA variable regions D2 (V3) and D2 (V13) in apoptotic IPC cells treated with okadaic acid, cAMP analog or cycloheximide, but not in cells treated with 50 nM calyculin A. IPC cells were treated for 6 hours with either 1 µM okadaic acid (OA), 200 µM 8-chlorophenylthio-cAMP (8-CPT-cA), or 30 µg/ml cycloheximide (CHex.), or for 7.5 hours with 50 nM calyculin A (Cal.A). Total cellular RNA was isolated, agaroseelectrophoresed (25 µg/lane) and stained with ethidiumbromide. Northern blots of total RNA were hybridized with oligonucleotide complementary to the 5 prime (5′) or the 3 prime (3′) region of rat 28 S rRNA. The figure shows most extensive cleavage of 28 S rRNA in cells treated with cAMP analog. Based on the sizes of the fragments and additional analysis with probes from a number of regions of the 28 S rRNA molecule, similar cleavage patterns were distinguished in cells treated with either cAMP analog, okadaic acid, or cycloheximide: the probe against the 3′ region of 28 S rRNA revealed a 3.6/3.9 kb doublet resulting from cleavage within D2 (3′V3) and 1.5 kb fragments (not resolved) from cleavage in D8 (3′V13). A probe complementary to the 5′ region revealed unresolved 3.3 kb fragments (5′V13) expected from cleavage in D8. These fragments disappeared in advanced cAMP-induced apoptosis, but were clearly present in cells treated for 3-5 hours with cAMP analog (data not shown). The RNA from cells treated with 50 nM calyculin A showed uncharacteristic fragments that differed in electrophoretic mobility from those observed for RNA from cells treated with cAMP, okadaic acid, or cycloheximide. Cleavage of 18 S rRNA was not observed.

by unlabeled (U.L.) than by labeled cells (Fig. 10, upper inset). After incubation with cAMP for 4 hours the cells were evaluated with respect to apoptotic morphology and [3H]thymidine labeling of their nuclei. An increased proportion of unlabeled apoptotic nuclei was observed for cells pulse labeled about 6 hours before cAMP (Fig. 10). As shown in the upper inset of Fig. 10 the cells that escaped labeling by this pulse (U.L.) should have reached about the mid-S-phase at the start of cAMP challenge. It is concluded that IPC cells were most sensitive to cAMP during their DNA replication, although no major part of the cell cycle seemed to be spared. Leakage of degraded DNA into the medium was maximal when cells with labeled DNA were in S-phase at the time of cAMP challenge (Fig. 10; note the inverse relationship between unlabeled apoptotic nuclei and leakage of labeled DNA), supporting the conclusion that IPC cells were most vulnerable in S-phase. This conclusion was further corroborated

Fig. 9. Cyclic AMP-induced apoptosis of IPC cells in S-phase. Autoradiographic appearance of IPC cells treated with 200 µM 8chlorophenyl-cAMP for 4 hours, whereof the last 0.3 hours with [3H]thymidine. Details of the cytocentrifugation, fixation, staining with May-Grünewald-Giemsa solution, and processing for autoradiography are given in Materials and Methods. Note the heavy labeling of an obviously apoptotic nucleus. Bar, 5 µm.

by experiments using aphidicolin to synchronize the cells, and flow cytometry for cell cycle position. The short pulses of [3H]thymidine given in the above experiments did not alter the sensitivity to cAMP-induced apoptosis. However, continuous prelabeling for 7 hours or more with thymidine (0.7 µCi/ml) enhanced the sensitivity to cAMP challenge as evidenced by an about 0.7 hour decreased lag-time for 50% of the cells to become morphologically apoptotic after addition of cAMP analog. There was no increased ‘spontaneous’ apoptosis in cells preincubated for up to 18 hours with labeled thymidine. These observations demonstrate synergism between weak radiation and cAMP in inducing apoptosis, raising the possibility that moderately elevated cAMP may preferentially eliminate cells with marginally damaged DNA. Interactions between cell death inducing agents: mechanistic aspects To know when the IPC cells became committed to death they were exposed to death-inducing agents for various periods of time, washed and monitored for intactness 24 hours later. It was found that 50% of the cells were irreversibly committed to disrupt after less than 2 hours exposure to cAMP analog and after slightly more than 2 hours exposure to 1 µM okadaic acid. After 3 hours nearly all the cells were committed to disintegrate (Fig. 11). Cells exposed continuously for 24 hours to okadaic acid or to maximal cAMP challenge disintegrated. Cells treated with >50 nM calyculin A displayed a puzzling behavior. When exposed to a short (50 nM calyculin A (pNCD) was distinct from NCD. The protein synthetic activity in cells incubated with calyculin A dropped simultaneously with their commitment to


Fig. 10. IPC cells are preferentially vulnerable to cAMP-induced apoptosis in the S-phase of the cell cycle. IPC cells were pulse labeled with [3H]thymidine for 0.7 hours, washed to remove [3H]thymidine, and challenged with 200 µM 8-chlorophenylthio-cAMP for 4 hours, starting at different time points (lower abscissa) after the pulse of [3H]thymidine. Samples of such treated cells were taken for determination of cell and medium [3H]DNA by scintillation counting, and for autoradiographic determination of labeled apoptotic and non-apoptotic nuclei. The main panel shows a peak in the proportion of unlabelled apoptotic cells (s) and a corresponding nadir in leakage of labeled DNA into the medium (n) for cells that had been exposed to [3H]thymidine about 6 hours before the cAMP challenge. Reference to the upper part of the figure shows that the cells that escaped labeling of their DNA (marked U.L.) during a pulse 6 hours before start of cAMP challenge were in the middle part of their S-phase when the cAMP challenge started and in late S-phase after 2 hours of cAMP-challenge, which is the time when most cells have become committed to death by cAMP (Fig. 12). This means that cells were most vulnerable to cAMP-induced death while in S-phase. As explained in Materials and Methods the duration of S-phase was 7.3 hours and the total cell cycle lasted 12 hours. Therefore, the 0.7 hour pulse of [3H]thymidine would label cells in 8 out of the 12 hours of their cycle. The cycle position of such labeled cells at the commencement of the cAMP challenge is shown by vertical stripes. After 2 hours of cAMP challenge the labeled cells have moved 2 hours further along the cell cycle, their new position being shown by horizontal lines. This means that only a narrow 2 hour segment of the cell cycle contained unlabeled cells during the whole 2 hour commitment period.

death, whereas drop in protein synthesis lagged 1-2 hours behind commitment in cells treated with cAMP analog or okadaic acid (data not shown). The protein synthesis inhibitor cycloheximide was used to probe further the role of protein synthesis. This drug committed 50% of the cells to death in 3 hours, but 15% were still not disrupted after 24 hours of incubation (Fig. 11). A drop in protein synthesis cannot be primarily responsible for apoptosis in response to cAMP and phosphatase inhibitors since they acted faster than cycloheximide in inducing apoptosis. Cycloheximide protected against cAMP-induced cell death, but not against phosphatase

inhibitor-induced death (Table 2) providing another distinction between cAMP- and phosphatase inhibitor-induced death. Since phosphatase inhibitors can induce hyperphosphorylation only if a kinase is active, we investigated whether okadaic acid action required cAMP kinase activity. Several approaches were used to probe this possibility. Firstly, a mutant IPC subclone with a dominantly negative point mutation in the regulatory subunit of cAMP kinase I (Gjertsen et al., 1993) was as sensitive to okadaic acid as the wild-type IPC cells (Fig. 12). Secondly, the cAMP kinase inhibitors H-8 and KT-5720 (Kase et al., 1987; Hidaka et al., 1991) preferentially prevented

Multiple phosphorylation-induced death types 3373 Table 2. Effects of phosphatase inhibitors on cAMP-induced apoptosis % Nuclei with hypercondensed chromatin Mean ± s.e.m. (n)

Agent(s) tested

Type of death

% Degraded DNA Mean ± s.e.m.(n)

Control Cycloheximide (30 µg/ml) Okadaic acid (1 µM) Calyculin A (10 nM) Calyculin A (100 nM) 8-CPT-cAMP (200 µM) Prostaglandin E1 (1 µM) Prostaglandin E2 (10 µM)

2±1 (8) 32±4 (6) 94±3 (9) 100±1 (2) 100±0 (5) 99±1 (5) 98±2 (2) 96±3 (2)

ACD ACD ACD pNCD ACD ACD ACD

− d+n++c++ d+nc d+n++c++ c d++n++c++ d++n++c++ d++n++c++

0±0 (8) 14±2 (3) 35±3 (5) 13±2 (2) 0±0 (5) 86±10 (5) 79±6 (2) 83±7 (2)

8-CPT-cAMP + Cycloheximide (30 µg/ml) + Calyculin A (10 nM) + Calyculin A (100 nM) + Okadaic acid (0.5 µM) + Okadaic acid (1.0 µM) Okadaic acid + Cycloheximide (30 µg/ml)

18±6 (5) 100±0 (3) 100±0 (4) 100±0 (3) 100±0 (3) 97±1 (4)

ACD ACD pNCD ACD ACD ACD

d+n++c++ d+n++c++ c d+nc; d++n++c++ d+nc d+nc

7±3 (5) 10±5 (3) 0±0 (4) 70±5 (3) 40±5 (3) 21±3 (4)

Cell density (x105 per ml) after 24h incubation

The upper part of the table shows that cAMP-elevating agents and the cAMP analog 8-chlorophenylthio-cAMP (8-CPT-cAMP) induced a similar death type, characterized by pronounced internucleosomal DNA fragmentation, in IPC cells. When combined with either cycloheximide (inhibitor of protein synthesis), 1 µM okadaic acid or calyculin A (lower part of Table) this death type was suppressed and replaced by the death type seen when the latter agents acted alone. Note that cAMP-analog and cycloheximide acted antagonistically, and that a mixture of the cAMP-specific and the okadaic acid-specific death patterns (without overlapping types) were observed when cells were exposed to a combination of 0.5 µM okadaic acid and cAMP analog. The time of incubation was 6 hours. Scoring of hypercondensed chromatin and DNA fragmentation was as described in the legends to Figs 4 and 5. The death types (Table 1) were based on DNA fragmentation and ultrastructural features.

Table 3. Test of selected agents for ability to inhibit cAMP- and okadaic acid-induced apoptosis

3

Agent(s) tested 2

1

0 0

120 240 360 Time of preincubation (min) with apoptotic agents before washing

Fig. 11. Time-dependent disintegration of IPC cells after preincubation with agents modulating protein phosphorylation or protein synthesis. The figure shows the decrease of cell density (ordinate) as a function of preincubation time (abscissa) with 200 µM 8-chlorophenylthio-cAMP (h), cycloheximide at 30 µg/ml (.), or the phosphatase inhibitors okadaic acid at 1 µM (∆), or calyculin A at 100 nM (s; g). Cells without any of the above agents (d) were controls. After extensive washing involving centrifugation (s) or involving transfer of culture inserts (g) the cells were resuspended in ordinary medium, and counted 24 hours later. Results shown are the mean of eight independent experiments. The time required for 50% death commitment was S hour for 100 nM calyculin A, 2 hours for cAMP, 2S hours for okadaic acid, and 3S hours for cycloheximide. The error bars represent s.e.m.

cAMP-induced, but not okadaic acid-induced apoptosis. In fact, H-8 enhanced the effect of okadaic acid (Table 3), particularly during the first 4-5 hours of okadaic acid action (not

% Nuclei with hypercondensed chromatin; Mean ± s.e.m. (n) 8-CPT-cAMP Okadaic acid — (200 µM) (0.8 µM)

Control

3±1 (14)

90±1 (11)

77±3 (9)

EGTA (3 mM) EGTA (3 mM) + Quin2/ AM (70 +13 µM)

6±2 (3) 3±2 (2)

94±6 (2) 85±7 (2)

67 49±9 (2)

A23187 (3 µM)

2±1 (4)

51±6 (5)

not determined

KT5823 (2 µM) KT5823 (10 µM)

3±2 (4) 5±3 (3)

100±1 (4) 100±1 (3)

73±9 (3) 72±12 (4)

KT5720 (3 µM) KT5720 (14 µM)

2±1 (5) 7±7 (3)

99±1 (5) 84±9 (3)

67±5 (5) 93±1 (3)

H-8 (19 µM) H-8 (47 µM) H-8 (140 µM)

1±1 (3) 7±2 (4) 8±2 (6)

80±1 (3) 15±4 (3) 3±1 (5)

not determined 87±5 (4) 79±2 (4)

1 2

98 98

not determined not determined

Cysteamine* (200 µM) BSO* (150 µM)

IPC cells were incubated for 6 hours with cAMP analog or okadaic acid after preincubation for 15 minutes with ion-chelators (Quin2/AM was added to a concentration of 70 µM at preincubation and a further 13 µM added after 1 hour of incubation), calcium-ionophore or protein kinase inhibitors. The ability of the kinase inhibitor H-8 to inhibit cAMP-induced and enhance okadaic acid-induced apoptosis was even more pronounced after 4 and 5 hours of incubation. *The data in the table are mean values of two experiments when cells were preincubated for 20 hours. Similar results were obtained after 18 and 24 hours of preincubation with concentrations of cysteamine or L-buthionine-(S,R)sulfoximine (BSO) ranging from 30-250 µM.

shown). Finally, cells exposed to a combination of cAMP analog and weakly apoptotic concentrations (0.8 µM) suppressed the cAMP-specific death type, apparently because the okadaic acid-specific death type (d+,n,c) was dominant (Table 2; data not shown). It is concluded that okadaic acid and calyculin A could prevent the cAMP-induced features of death, but could not save the cells from dying. In view of studies implicating increased Ca2+ (McConkey et al., 1989b; Jiang et al., 1994) or oxidation (see Buttke and Sandstrom, 1994 for review) as mediators of apoptosis, agents perturbing the intracellular level of Ca2+ or the red/ox status were tested for ability to modulate IPC cell death. Depletion of extracellular Ca2+ (using a Ca2+-free medium with EGTA) and intracellular free Ca2+ (using quin2/AM) only marginally decreased the apoptotic response to either okadaic acid or cAMP (Table 3), indicating that Ca2+ influx or elevated Ca2+ was not required for apoptosis. In fact, elevation of intracellular Ca2+ by the ionophore A23187 partially protected against cAMP-induced cell death (Table 3). IPC cell apoptosis was

Fig. 13. Calyculin A is a potent inhibitor of IPC cell clonogenicity. IPC cells (4×105/ml) were preincubated for 1 hour (,), 6 hours (u), or 12 hours (s) with various concentrations (abscissa) of the phosphoprotein phosphatase inhibitor calyculin A. The cells were next diluted to 800/ml in collagen gels with fresh medium (1.5 ml), and the number of cultures counted 4 days later. Some cells were not pretreated, but grown in the continuous presence of various concentrations of calyculin A during the 96 hours of cloning (n). The clonogenicity (left ordinate) is given as the percentage of seeded cells giving rise to clones. The IC50 for calyculin A was 0.1 nM for continuous presence, 0.5 nM for 12 hour preincubation, 1.5 nM for 6 hour preincubation, and 4 nM for 3 hour preincubation. In one experiment cells were coincubated for 6 hours with various concentrations of calyculin A and 200 µM 8-chlorophenylthio-cAMP (d), followed by scoring of the percentage of cells (right ordinate) with the characteristic cAMP-induced pattern of cell fragmentation (see e.g. Figs 1,4). Note the similar potency of calyculin A as inhibitor of clonogenicity and of cAMP-induced apoptosis. The size of each colony was similar for all treatments except continuous presence of 0.1 nM or higher of calyculin A, where colony size was decreased.

unaffected by manipulation of the level of glutathione by cysteamine or L-buthionine-[S,R]-sulfoximine (Table 3), which respectively enhance and inhibit the synthesis of glutathione (Griffith and Meister, 1979; Djurhuus et al., 1990). It is concluded that IPC cell apoptosis can occur without increase of Ca2+ or decrease of glutathione. DISCUSSION Agents acting on enzymes controlling protein phosphorylation induced several, surprisingly different, death patterns in IPC promyelocytic leukemia cells. The death types were distinct with regard to ultrastructure and often with regard to fragmentation of the whole cell, the nucleus, DNA and 28 S rRNA as well (Table 1). The variety of death types observed in one cell line raises the possibility that some cells may have available a repertoire of death type options. Most death types had features of apoptosis (Kerr et al., 1972; Wyllie et al., 1980), but one of the types induced by >50 nM of the protein phosphatase inhibitor calyculin A (Figs 2c, 3; Table 1) was

Multiple phosphorylation-induced death types 3375 neither apoptotic (margination of hypercondensed chromatin was absent, and the cell volume increased) nor necrotic (the mitochondria were not swollen, the plasma membrane appeared intact, the ATP level was high). This clear example of non-necrotic, non-apoptotic death (Table 1) is tentatively classified as ‘pseudonecrosis’ (pNCD). It differs from ‘atypical’ death types described in other systems, like the developing nervous system (Clarke, 1990). cAMP had a dual action on IPC cells, blocking proliferation in G1/S and inducing death preferentially in S-phase (Fig. 10). This allows efficient cell compartment contraction, since cells escaping the G1 block have a high risk of death in the ensuing S-phase. That cAMP acts in S-phase is unprecedented. In S49 lymphoma cells cAMP slowly induces death secondary to growth arrest in G1 (Lemaire and Coffino, 1977). The cAMP signaling pathway was triggered by β-agonist and prostaglandins (Table 2), indicating that it could operate in vivo. The cAMP-induced death was characterized by particularly pronounced fragmentation of both DNA (Figs 4, 5, 7; Table 2) and 28 S rRNA (Fig. 8). It was ultrastructurally distinct from the death induced by other agents and all types of IPC cell death occurring spontaneously. This indicates that cAMP actively signals cell compartment contraction and not merely enhances a ‘default’ pathway of cell death, as presumably is the case when cells are deprived of survival-promoting agents (Raff, 1992), like interleukin-3 (Metcalf, 1989). Withdrawal of this hematopoetic growth factor leads to a death induction in granulocyte progenitor cells that is about 5 times slower than that induced by cAMP in IPC cells (Williams et al., 1990; Collins et al., 1992). The interleukin-3 withdrawal effect is counteracted by overexpression of the bcl-2 gene product (Fairbairn et al., 1993), which presumably protects against necrosis and apoptosis in cells exposed to oxidative stress with decreased level of reduced glutathione (Hockenbery et al., 1993; Kane et al., 1993). The cAMP-induced death was not modulated by agents decreasing or increasing the level of glutathione (Table 3), and was not accompanied by oxidation of glutathione (B. T. Gjertsen, R. Svardal, P. M. Ueland and S. O. Døskeland, unpublished data). The various death types appeared to result from triggering of intrinsically different pathways. IPC cells incubated with 50 nM calyculin A lacked nuclear envelope (Fig. 2), but had apparently intact DNA (Figs 5, 7) suggesting that influx of cytoplasmic factors is insufficient to induce nuclear DNA fragmentation. Calyculin A (at >50 nM) appeared to induce disaggregation of the nuclear envelope in any major phase of the IPC cell cycle, since the envelope was completely absent in cells treated with the phosphatase inhibitor for 1 hour (i.e. less than 10% of the cell cycle time). Phosphatase inhibitor could induce nuclear envelope disintegration and premature chromatin condensation in S-phase, but not in G1 phase of fibroblasts (Yamashita et al., 1990). Mitochondria have generally been considered to be spared in apoptotic cell death (Wyllie et al., 1980; Tepper and Studzinski, 1993), but closer examination revealed a decreased mean mitochondrial volume in most instances of induced IPC cell death (Table 1). Particularly cells treated with >50 nM calyculin A had pronounced micromitochondriosis, indicating that also mitochondria can become fragmented during cell death. The micromitochondria were presumably intact with respect to ATP production capability (Fig. 3). In conclusion, the present study has pointed to altered protein phosphorylation as triggering distinct pathways leading to different types of death in one homogeneous cell line. Cell death, being irreparable, should be under strict regulatory controls and countercontrols, and a pivotal role of protein phosphorylation is as plausible for the regulation of cell elimination as for the regulation of other complex functions, like proliferation and cell cycle progression. Expert technical assistance was provided by Erna Finsås, Nina Lied Larsen and Edith Fick. The work was supported by grants from The Norwegian Cancer Society, The Novo Nordic Foundation (S.D.), Ligue Nationale Contre le Cancer, CNRS (M.L.), and the EC Biomedical and HCM Programmes.

REFERENCES Araki, S., Simada, Y., Kaji, K. and Hayashi, H. (1990). Role of protein kinase C in the inhibition by fibroblast growth factor of apoptosis in serumdepleted endothelial cells. Biochem. Biophys. Res. Commun. 172, 10811085. Arends, M. J., Morris, R. G. and Wyllie, A. H. (1990). Apoptosis. The role of the endonuclease. Am. J. Pathol. 136, 593-608. Baxter, G. D. and Lavin, M. F. (1992). Specific protein dephosphorylation in apoptosis induced by ionizing radiation and heat shock in human lymphoid tumor lines. J. Immunol. 148, 1949-54. Bollen, M. and Stalmans, W. (1992). The structure, role, and regulation of type 1 protein phosphatases. Crit. Rev. Biochem. Mol. Biol. 27, 227-281. Brewis, N. D., Street, A. J., Prescott, A. R. and Cohen, P. T. W. (1993). PPX, a novel protein serine threonine phosphatase localized to centrosomes. EMBO J. 12, 987-996. Buttke, T. M. and Sandstrom, P. A. (1994). Oxidative stress as a mediator of apoptosis. Immunol. Today 15, 7-10. Bøe, R., Gjertsen, B. T., Vintermyr, O. K., Houge, G., Lanotte, M. and Døskeland, S. O. (1991). The protein phosphatase inhibitor okadaic acid induces morphological changes typical of apoptosis in mammalian cells. Exp. Cell Res. 195, 237-246. Clarke, P. G. H. (1990). Developmental cell death: morphological diversity and multiple mechanisms. Anat. Embryol. 181, 195-213. Clem, R. J. and Miller, L. K. (1993). Apoptosis reduces both the in vitro replication and the in vivo infectivity of a baculovirus. J. Virol. 67, 37303738.

Cohen, G. M., Sun, X. M., Snowden, R. T., Dinsdale, D. and Skilleter, D. N. (1992). Key morphological features of apoptosis may occur in the abscence of internucleosomal DNA fragmentation. Biochem. J. 286, 331-334. Cohen, P. (1989). The structure and regulation of protein phosphatases. Annu. Rev. Biochem. 58, 453-508. Collins, M. K. L., Marvel, J., Malde, P. and Lopezrivas, A. (1992). Interleukin-3 protects murine bone marrow cells from apoptosis induced by DNA damaging agents. J. Exp. Med. 176, 1043-1051. Djurhuus, R., Svardal, A. M. and Ueland, P. M. (1990). Cysteamine increases homocysteine export and glutathione content by independent mechanisms in C3H/10T1/2 cells. Biochem. Pharmacol. 38, 327-332. Døskeland, S. O., Maronde, E. and Gjertsen, B. T. (1993). The genetic subtypes of cAMP-dependent protein kinase - functionally different or redundant. Biochim. Biophys. Acta. 1178, 249-258. Eriksson, J. E., Brautigan, D. L., Vallee, R., Olmsted, J., Fujiki, H. and Goldman, R. D. (1992). Cytoskeletal integrity in interphase cells requires protein phosphatase activity. Proc. Nat. Acad. Sci. USA 89, 11093-11097. Fairbairn, L. J., Cowling, G. J., Reipert, B. M. and Dexter, T. M. (1993). Suppression of apoptosis allows differentiation and development of a multipotent hemopoietic cell line in the absence of added growth factors. Cell 74, 823-832. Gavrieli, Y., Sherman, Y. and Bensasson, S. A. (1992). Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation. J. Cell Biol. 119, 493-501. Gjertsen, B. T., Duprez, E., Bernard, O., Lanotte, M. and Døskeland, S. O. (1993). Antiapoptotic effect of heterozygously expressed mutant RI (Ala336→Asp) subunit of cAMP kinase-I in a rat leukemia cell line. J. Biol. Chem. 268, 8332-8340. Griffith, O. W. and Meister, A. (1979). Potent and specific inhibition of glutathione synthesis by buthionine sulfoximine (S-n-butyl-homocysteine sulfoximine). J. Biol. Chem. 257, 7558-7560. Gundersen, H. J. G., Bagger, P., Bentsen, T. F., Evans, S. M., Korbo, L., Marcussen, N., Møller, A., Nielsen, K., Nyengaard, J. R., Pakkenberg, B., Sørensen, F. B., Vesterby, A. and West, M. J. (1988). The new stereological tools: dissector, fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis. Acta Path. Microbiol. Scand. 96, 857-881. Guy, G. R., Cao, X., Chua, S. P. and Tan, Y. H. (1992). Okadaic acid mimics multiple changes in early protein phosphorylation and gene expression induced by tumor necrosis factor or interleukin-1. J. Biol. Chem. 267, 18461852. Hidaka, H., Watanabe, M. and Kobayashi, R. (1991). Properties and use of H-series compounds as protein kinase inhibitors. In Protein Phosphorylation (ed. T. Hunter and B. M. Sefton), pp. 328-339. Academic Press, Inc., San Diego. Hockenbery, D. M., Oltvai, Z. N., Yin, X. M., Milliman, C. L. and Korsmeyer, S. J. (1993). Bcl-2 functions in an antioxidant pathway to prevent apoptosis. Cell 75, 241-251. Houge, G., Døskeland, S. O., Bøe, R. and Lanotte, M. (1993). Selective cleavage of 28 S rRNA variable regions V3 and V13 in myeloid leukemia cell apoptosis. FEBS Lett. 315, 16-20. Ishida, Y., Furukawa, Y., Decaprio, J. A., Saito, M. and Griffin, J. D. (1992). Treatment of myeloid leukemic cells with the phosphatases inhibitor okadaic acid induces cell cycle arrest at either G1/S or G2/M depending on dose. J. Cell. Physiol. 150, 484-492. Jiang, S., Chow, S. C., Nicotera, P. and Orrenius, S. (1994). Intracellular Ca2+ signals activate apoptosis in thymocytes: studies using the Ca2+ATPase inhibitor Thapsigargin. Exp. Cell Res. 212, 84-92. Kane, D. J., Sarafian, T. A., Anton, R., Hahn, H., Gralla, E. B., Valentine, J. S., Örd, T. and Bredesen, D. E. (1993). Bcl-2 inhibition of neural death: decreases generation of reactive oxygen species. Science 262, 1274-1277. Kase, H., Iwahashi, K., Nakanishi, S., Matsuda, Y., Yamada, K., Takahashi, M., Murakata, C., Sato, A. and Kaneko, M. (1987). K-252 compounds, novel and potent inhibitors of protein kinase C and cyclic nucleotide-dependent protein kinases. Biochem. Biophys. Res. Commun. 142, 436-440. Kerr, J. F., Wyllie, A. H. and Currie, A. R. (1972). Apoptosis: a basic biological phenomenon with wide-ranging implications in tissue kinetics. Br. J. Cancer 26, 239-57. Lanotte, M., Gombaud-Saintonge, G., Tertian, G. (1986). Selective inhibition of the proliferation of various murine hemopoietic progenitor cells by cholera toxin. Exp. Hematol. 14, 724-731. Lanotte, M., Riviere, J. B., Hermouet, S., Houge, G., Vintermyr, O. K., Gjertsen, B. T. and Døskeland, S. O. (1991). Programmed cell death

Multiple phosphorylation-induced death types 3377 (apoptosis) is induced rapidly and with positive cooperativity by activation of cyclic adenosine monophosphate-kinase I in a myeloid leukemia cell line. J. Cell. Physiol. 146, 73-80. Lemaire, I. and Coffino, P. (1977). Cyclic AMP-induced cytolysis in SV9 cells: selection of an unresponsive “Deathless” mutant. Cell 11, 149-155. Lucocq, J. (1992). Mimicking mitotic Golgi disassembly using okadaic acid. J. Cell Sci. 103, 875-880. McConkey, D., Hartzell, P., Jondal, M. and Orrenius, S. (1989a). Inhibition of DNA fragmentation in thymocytes and isolated thymocyte nuclei by agents that stimulate protein kinase C. J. Biol. Chem. 264, 13399-13402. McConkey, D. J., Hartzell, P., Nicotera, P. and Orrenius, S. (1989b). Calcium-activated DNA fragmentation kills immature thymocytes. FASEB J. 3, 1843-1849. Metcalf, D. (1989). The molecular control of cell division, differentiation commitment and maturation in haemopoietic cells. Nature 339, 27-30. Oberhammer, F. A., Pavelka, M., Sharma, S., Tiefenbacher, A. F., Purchio, A. F., Bursh, W. and Schulte-Hermann, R. (1992). Induction of apoptosis in cultured hepatocytes and regressing liver by transforming growth factor β1. Proc. Nat. Acad. Sci. USA 89, 5408-5412. Ohoka, Y., Nakai, Y., Mukai, M. and Iwata, M. (1993). Okadaic acid inhibits glucocorticoid-induced apoptosis in T cell hybridomas at its late stage. Biochem. Biophys. Res. Commun. 197, 916-921. Peitsch, M. C., Polzar, B., Stephan, H., Crompton, T., MacDonald, H. R., Mannherz, H. G. and Tschopp, J. (1993). Characterization of the endogenous deoxyribonuclease involved in nuclear DNA degradation during apoptosis (programmed cell death). EMBO J. 12, 371-377. Raff, M. C. (1992). Social controls on cell survival and cell death. Nature 356, 397-399. Robaye, B., Døskeland, A. P., Suarez-Huerta, N., Døskeland, S. O. and Dumont, J. E. (1994). Apoptotic cell death analyzed at the molecular level by two-dimensional gel electrophoresis. Electrophoresis 15, 503-510. Schwartz, L. M. and Osborne, B. A. (1993). Programmed cell death, apoptosis and killer genes. Immunol. Today 14, 582-590. Schwartzman, R. A. and Cidlowski, J. A. (1993). Apoptosis - the biochemistry and molecular biology of programmed cell death. Endocrine Rev. 14, 133-151. Sen, S. (1992). Programmed cell death: concept, mechanism and control. Biol. Rev. Cambridge. Phil. Soc. 67, 287-319. Song, Q. Z., Baxter, G. D., Kovacs, E. M., Findik, D. and Lavin, M. F. (1992). Inhibition of apoptosis in human tumour cells by okadaic acid. J. Cell. Physiol. 153, 550-556.

Song, Q. Z. and Lavin, M. F. (1993). Calyculin A, a potent inhibitor of phosphatases-1 and 2a, prevents apoptosis. Biochem. Biophys. Res. Commun. 190, 47-55. Spaulding, S. W. (1993). The ways in which hormones change cyclic adenosine 3′,5′-monophosphate-dependent protein kinase subunits, and how such changes affect cell behavior. Endocr. Rev. 14, 632-650. Sun, D. Y., Jiang, S., Zheng, L.-M., Ojcius, D. M. and Young, J. D.-E. (1994). Separate metabolic pathways leading to DNA fragmentation and apoptotic chromatin condensation. J. Exp. Med. 179, 559-568. Takizawa, T., Matsukawa, S., Higuchi, Y., Nakamura, S., Nakanishi, Y. and Fukuda, R. (1993). Induction of programmed cell death (apoptosis) by influenza virus infection in tissue culture cells. J. Gen. Virol. 74, 2347-2355. Taylor, S. S., Buechler, J. A. and Yonemoto, W. (1990). cAMP-dependent protein kinase: framework for a diverse family of regulatory enzymes. Annu. Rev. Biochem. 59, 971-1005. Tepper, C. G. and Studzinski, G. P. (1993). Resistance of mitochondrial DNA to degradation characterizes the apoptotic but not the necrotic mode of human leukemia cell death. J. Cell. Biochem. 52, 352-361. Ucker, D. S., Obermiller, P. S., Eckhart, W., Apgar, J. R., Berger, N. A. and Meyers, J. (1992). Genome digestion is a dispensable consequence of physiological cell death mediated by cytotoxic T lymphocytes. Mol. Cell. Biol. 12, 3060-3069. Vintermyr, O. K., Gjertsen, B. T., Lanotte, M. and Døskeland, S. O. (1993). Microinjected catalytic subunit of cAMP-dependent protein kinase induces apoptosis in myeloid leukemia (IPC-81) cells. Exp. Cell Res. 206, 157-161. Williams, G. T., Smith, C. A., Spoocer, E., Dexter, T. M. and Taylor, D. R. (1990). Haemopoietic colony stimulating factors promote cell survival by suppressing apoptosis. Nature 343, 76-79. Williams, G. T. and Smith, C. A. (1993). Molecular regulation of apoptosis genetic controls on cell death. Cell 74, 777-779. Wyllie, A. H. (1980). Glucocorticoid induced thymocytes apoptosis is associated with endogenous endonuclease activation. Nature 284, 555-556. Wyllie, A. H., Kerr, J. F. R. and Currie, A. R. (1980). Cell death: The significance of apoptosis. Int. Rev. Cytol. 68, 251-306. Yamashita, K., Yasuda, H., Pines, J., Yasumoto, K., Nishitani, H., Ohtsubo, M., Hunter, T., Sugimura, T. and Nishimoto, T. (1990). Okadaic acid, a potent inhibitor of type 1 and type 2A protein phosphatases, activated cdc2H1 kinase and transiently induces a premature mitosis-like state in BHK21 cells. EMBO J. 9, 4331-4338. (Received 25 April 1994 - Accepted 24 August 1994)