Protein Phosphatase 1A-Mediated Stimulation of Apoptosis Is Associated with Dephosphorylation of the Retinoblastoma Protein

Home , Dephosphorylation, Protein phosphatase 1

Protein phosphatase 1a-mediated stimulation of apoptosis is associated with dephosphorylation of the retinoblastoma protein

Rui-Hong Wang1,2, Cathy WY Liu1, Vassilios I Avramis1 and Norbert Berndt*,1

1Division of Hematology/Oncology, Childrens Hospital Los Angeles, University of Southern California School of Medicine, 4650 Sunset Boulevard, Los Angeles, California, CA 90027, USA

Protein phosphatase 1 (PP1) plays important roles in number. Pathways that control these two processes are many dierent aspects of cellular activities including cell often linked to each other. It is thus not surprising that cycle control. One important function of PP1 is to protein phosphatase 1 (PP1), one of the major activate the retinoblastoma protein pRB. Here we show mammalian serine/threonine-speci®c phosphatases ubi- that pRB is one of PP1's downstream targets during quitously expressed, is involved in dierent aspects of apoptosis. When HL-60 cells synchronized at the G1/S cell growth control (Berndt, 1999) and may be a target boundary were treated with pro-apoptotic cytosine for cancer therapy (Berndt, 2000). arabinoside (araC), PP1a protein increased twofold and The assignment of speci®c roles in various cellular PP1 activity about 30% within 1 h. This was followed by processes to PP1 is complicated by the fact that pRB dephosphorylation, pRB cleavage by caspases, mammalian PP1 comprises four isozymes (1a,1d, DNA fragmentation, the appearance of cells with 52n 1g1, 1g2) which can bind to many non-substrate DNA content and ®nally, dying and dead cells. In vitro, cellular proteins functioning as regulatory subunits pRB was protected from caspase-3 digestion by prior (Damer et al., 1998). These subunits modify PP1's Cdk-mediated phosphorylation, whereas PP1a converted activity or target it to particular locales and substrates. phospho-pRB into an ecient substrate for caspase-3. PP1 appears to play several roles in cell cycle Introduction of active PP1a into HL-60 cells by regulation. In organisms from S. pombe to humans, electroporation was sucient to induce characteristics PP1 activity is required for the exit from mitosis of apoptosis. Similarly, araC-resistant cells, normally (Axton et al., 1990; Booher and Beach, 1989; Cheng et unable to die in response to araC, initiated apoptosis al., 2000; Fernandez et al., 1992; Ohkura et al., 1989), when electroporated with active PP1a. This was also although the essential mitotic substrate(s) for PP1 have accompanied by pRB cleavage. In contrast, introduction not yet been identi®ed. Meanwhile we have found that of inhibitor-2 delayed the onset of araC-induced phosphorylation of a C-terminal threonine (Thr320) in apoptosis, whereas concomitant introduction of PP1a mammalian PP1a by Cdks inhibits its activity in vitro and inhibitor-2 completely prevented PP1a-induced (Dohadwala et al., 1994). This site is phosphorylated in apoptosis. These results suggest that dephosphorylation vivo in S. pombe at the onset of mitosis (Yamano et al., of key proteins by PP1a may be crucial for the initiation 1994) and in human cells in mitosis (Kwon et al., 1997; of apoptosis and further support the concept of PP1 Liu et al., 1999) and in late G1 until early S phase (Liu serving as a potential target for anti-cancer therapy. et al., 1999). This indicates that control of PP1 Oncogene (2001) 20, 6111 ± 6122. function by regulatory subunits may not be sucient, but also requires post-translational modi®cation of the Keywords: protein phosphatase 1; retinoblastoma catalytic subunit itself, at least in the context of the cell protein; caspase; protein phosphorylation; cell cycle; cycle. Several laboratories including ours have identi- apoptosis ®ed PP1 to be responsible for the dephosphorylation of pRB during the M/G1 transition (Alberts et al., 1993; Berndt, 1995; Durfee et al., 1993; Ludlow et al., 1993) Introduction and G0/G1 phase (Berndt, 1995). In fact, the inactivation of PP1a by Cdks during the G1/S Reversible protein phosphorylation plays an important transition appears to be linked to the concomitant role in the regulation of the cell cycle and apoptosis, inhibition of pRB: Thus, in human cells expressing which are the two major mechanisms to control the cell intact pRB, constitutively active PP1a (PP1aT320A) prevents pRB phosphorylation and progression into S phase (Berndt et al., 1997). Consistent with this result, G1/S phosphorylation aects only pRB-asso- *Correspondence: N Berndt; E-mail: [email protected] ciated PP1a (Liu et al., 1999). Furthermore, in cells 2 Current address: NIDDK, National Institutes of Health, Building that cannot phosphorylate pRB or do not express 10, Room 9N104, 9000 Rockville Pike, Bethesda, MD 20892, USA Received 6 April 2001; revised 21 June 2001; accepted 28 June pRB, PP1a causes increased cell death, giving rise to 2001 the speculation that it is the phosphorylated form of Roles of PP1 in apoptosis R-H Wang et al 6112 pRB, rather than RB per se, that protects cells from aphidicolin and exposed to 10 mM araC for 2 h. Then apoptosis (Berndt et al., 1997). the cells were incubated for another 24 h in the These ®ndings appear to be in con¯ict with reports absence of the drug. At several intervals, samples were that suggested an inhibitory role for pRB in apoptosis, collected and subjected to Western blot analysis, PP1 a notion that is dicult to reconcile with its established activity assay, DNA fragmentation assay, trypan blue role as a tumor suppressor. Thus, massive cell death staining and ¯ow cytometry (Figure 1). The Western occurs in mice lacking pRB expression (Clarke et al., blots, PP1 activity and DNA fragmentation assays are 1992; Jacks et al., 1992; Lee et al., 1992). Consistent based on equal cell number or equal protein. The data with these data, overexpression of wild-type pRB in show that following araC treatment, PP1a protein pRB-lacking tumor cells inhibits apoptosis (Berry et more than doubled its amount 3 h after release from al., 1996; Fan et al., 1996; Haas-Kogan et al., 1995). araC, whereas of the other PP1 isozymes, PP1d More recently, pRB was found to be cleaved by ICE- increased approximately twofold at 24 h after release like proteases during apoptosis generating a 42-amino- from araC, and PP1g1 remained unaected (Figure acid or 5-kDa peptide from its C terminus (Chen et al., 1a). We also observed that PP1 activity increased by 1997; JaÈ nicke et al., 1996). These data suggest that about 30% (Figure 1b). Meanwhile, high molecular pRB belongs to the so-called death substrates which weight pRB gradually disappeared in favor of a faster- need to be destroyed in order to launch or execute migrating species (Figure 1c), which is indicative of apoptosis. However, since none of the above studies either dephosphorylation or proteolytic cleavage or addressed the role that the pRB phosphorylation status both (Tan and Wang, 1998). The DNA fragmentation may play in apoptosis, we concluded that the exact assay, trypan blue staining and ¯ow cytometry function of pRB in apoptosis remains ill-de®ned. analysis demonstrated that araC indeed induced Similarly, the function of protein phosphatases in apoptosis in G1/S synchronized HL-60 cells: they apoptosis is controversial. Thus, prolonged exposure of showed the typical features of apoptosis ± the appear- cells to phosphatase inhibitors such as okadaic acid has ance of DNA laddering (Figure 1d), cells with 52n been shown to cause cell cycle arrest (SchoÈ nthal and DNA content (Figure 1e), and dying or dead cells Feramisco, 1993; Taylor et al., 2000; Zheng et al., 1991) (Figure 1e). The degree of synchronization and the or apoptosis (Bùe et al., 1991; Inomata et al., 1995; progression through the cell cycle of treated cells is Morimoto et al., 1997). In contrast, apart from our own summarized in Table 1. Considering the succession of observations described above, a positive role for protein events, we notice that PP1 activity started to increase phosphatases in cell death was suggested by earlier within 1 h of araC treatment and reached its peak at ®ndings that apoptosis in human tumor cells is associated 1 h following the termination of exposure to araC. with the dephosphorylation of several proteins (Baxter DNA laddering and apoptotic peaks in ¯ow cyto- and Lavin, 1992) and that protein phosphatase activity is metric analyses appeared at about 2 h after release necessary during early apoptosis in plants (Dunigan and from araC, and ®nally, cells positive for trypan blue Madlener, 1995). Moreover, induction of apoptosis by showed up at 6 h following araC treatment. The cytosine arabinoside (araC) in HL-60 cells leads to observed order of events suggests that during S phase, dephosphorylation of pRB by an unidenti®ed phospha- PP1a, but not other PP1 isozymes, may critically be tase (Dou et al., 1995). In the present study, we sought to involved in launching apoptosis. characterize the role of PP1 in apoptosis, with an It should be noted that etoposide- and actinomycin emphasis on the mechanism of interaction between D-induced apoptosis was also associated with an PP1, pRB and caspases. Using HL-60 leukemia cells treated with araC, we found that PP1 activity rapidly and transiently increased approximately 30%, which was Table 1 Cell cycle progression of HL-60 cells treated with araC for accompanied by a twofold increase in the amount of 2h PP1a protein. This was followed by pRB dephosphor- % of cells ylation and then the proteolytic release of a short peptide Time (h) 52n DNA G1 S G2/M from pRB's C terminus. We also demonstrate that Async 3 38 49 13 phospho-pRB was a poor substrate for caspase-3 in vitro, 0 2 75 15 10 while prior dephosphorylation by PP1a converts phos- 1 1 76 16 8 pho-pRB into an ecient caspase-3 substrate. We further 2 1 79 9 12 2+1 2 70 15 15 show that PP1a activity induced apoptosis in HL-60 cells 2+2 2 59 16 25 as well as araC-resistant cells, and that inhibitor-2 2+3 2 47 27 26 prevented PP1a-dependent apoptosis. 2+4 5 41 29 30 2+5 27 36 32 32 2+6 31 33 30 37 2+20 51 23 45 32 Results 2+24 58 20 15 65 The data shown correspond to the experiment shown in Figure 1. Induction of PP1a precedes araC-induced apoptosis Note that the cells in the G1, S, and G2/M phases were calculated to yield 100%, although as cells with 52n DNA content accumulate, In order to investigate PP1's role in apoptosis, HL-60 the proportion of cells still proceeding through the cell cycle decrease cells were synchronized at the G1/S boundary by accordingly

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6113 induction of PP1a protein and increase of PP1 activity in HeLa cells (data not shown).

The mobility shift of pRB in response to araC is associated with dephosphorylation and cleavage During the last decade, many laboratories have studied the phosphorylation state of pRB by simple Western blot without prior 32P-labeling. This was justi®able by the early observation that phosphorylation of pRB is associated with an increase of the apparent Mr by about 5 ± 10 kDa on SDS gels. Following the discovery of a caspase-liberated 5 kDa peptide from the C terminus of pRB (Chen et al., 1997), this technique has to be approached with caution, since a Western blot with a generic pRB antibody would not be able to reveal whether the downward mobility shift in pRB is due to dephosphorylation or proteolysis. Therefore, we carried out another experiment in which the cells were labeled with 32P-inorganic phosphate prior to Western blotting with sc-102 (a monoclonal antibody to the A/B pocket region of pRB) and Rb-C (a polyclonal antibody to the C-terminal 15 amino acids), the rationale being that sc-102 would detect both intact and C-terminally truncated RB, and Rb-C would only recognize intact pRB. Following the metabolic labeling and araC treatment (see Materials and methods), pRB was immunoprecipitated with sc-102 and then visualized by autoradiography and immunoblotting with sc- 102 and Rb-C, respectively. Figure 2a shows that, in untreated cells, pRB was being phosphorylated, as was expected at the entry into S phase (upper panel). On the other hand, within 2 h after releasing the treated cells from araC, pRB had lost most of its phosphate. This corresponded to the appearance of fast-migrating pRB, also at 2 h after release from araC (Figure 2b, upper panel). The signal produced by the antibody reacting with the C-terminus of pRB suggested that pRB was being cleaved shortly after its dephosphorylation (Figure 2b, lower panel).

pRB dephosphorylation precedes pRB cleavage To determine whether the cleavage of pRB associated with apoptosis occurs simultaneously with, or after dephosphorylation, we repeated the above experiment, Figure 1 AraC-induced apoptosis is associated with activation using a shorter observation time (2 h after release from of PP1. Human HL-60 cells were synchronized at the G1/S araC) and shorter intervals (15 min) between taking boundary with aphidicolin and exposed to 10 mM araC for 2 h. samples. This revealed that within 15 min of araC Following removal of both drugs by several washes with culture medium, the cells were incubated for another 24 h and subjected removal, pRB was being dephosphorylated, as evi- to dierent analyses. (a) Western blots for the three ubiquitously denced by the loss of phosphate (Figure 3a, lower expressed PP1 isozymes a, g1, and d show that PP1a amounts panel) and the appearance of a low Mr form of pRB at increased about twofold at 2 ± 3 h after release from araC. (b) PP1 30 min after release from araC (Figure 3b, upper 32 activity, measured with P-phosphorylase a as a substrate, panel), whereas the loss of intact pRB occurred much transiently increased by approximately 30%. The graph shows the average of ®ve experiments+standard deviation. (c) Western blots for pRB with sc-102 show that, 1 ± 3 h after release from araC, the slow-migrating form pRB gradually disappeared in favor of fast-migrating pRB, suggesting that it got dephos- after release from araC (grey), whereas dying/dead (trypan-blue phorylated, proteolytically cleaved, or both. (d) Cells degrade positive) cells appeared in two waves: a small increase of dying their DNA, beginning at about 2 h after release from araC. The cells was seen immediately after removal of araC, and a more ®rst lane shows DNA size markers. (e) FACS analysis reveals that signi®cant increase to about 40% was seen between 6 and 20 h cells with a DNA content 52n were beginning to appear at 5 h after release from araC (black)

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6114

Figure 2 Apoptosis is associated with pRB dephosphorylation Figure 3 Phosphorylated pRB is dephosphorylated and then and cleavage. HL-60 cells were prepared as in Figure 1, except cleaved in response to araC. This experiment is identical to the 32 that they were also labeled with P-inorganic phosphate. The one presented in Figure 2, except that the observation period illustrations represent data from one typical experiment. The was shortened to 2 h and the intervals for taking samples were molecular changes in pRB were then analysed by immunopreci- reduced to 15 min, in order to see whether dephosphorylation pitation followed by (a) autoradiography or (b) Western blots and cleavage of pRB are simultaneous or successive events. As with antibodies recognizing both intact and truncated pRB (sc- in Figure 2, all panels were prepared from one experiment. The 102, top panels) or intact pRB only (Rb-C, bottom panels). As vertical arrows indicate noticeable changes in the phosphate expected during this phase of the cell cycle, cells not exposed to content or migration pattern of pRB. (a) Autoradiography of araC were found to contain heavily phosphorylated pRB (a, top immunoprecipitated pRB electrotransferred to Immobilon-P panels). The phosphate content of pRB decreased to about one membranes. Similar to Figure 2, pRB had lost about 80% of ®fth of the maximum within 3 h after release from araC, whereas its phosphate at 75 min after release from araC. (b) Western the ratio slow pRB:fast pRB decreased from *12 to *0.1, and blots for pRB with sc-102 and Rb-C (cf. legend to Figure 2). the amount of pRB protein detected by Rb-C decreased by about At 105 and 120 min after release from araC, the amount of 63%. The vertical arrows indicate a noticeable change in the intact pRB detected by Rb-C was approximately 25% of the migration pattern of pRB. See text for a further discussion maximum

later, namely at 105 min after release from araC that pRB, but not phospho-pRB, is a suitable caspase (Figure 3b, lower panel). The increase in pRB substrate. In other words, does phosphorylation by phosphorylation observed at 105 and 120 min after Cdks protect pRB from proteolytic digestion, and does release from araC was most likely due to imperfect PP1 convert phosphorylated pRB into a substrate for synchronization of the cells in this particular experi- proteolysis? In order to explore this further, we ment. It should be noted that we never saw such an incubated in vitro phosphorylated pRB with caspase-3 increase in phosphate content in four independent with or without prior dephosphorylation by PP1a.In experiments using the longer time intervals (cf. Figure these experiments, we used recombinant pRB, Cdk2/ 2). cyclin A, Cdk2/cyclin E, Cdk4/cyclin D1, and PP1a. Assuming that the ICE-like protease cleavage site in pRB corresponds to the C-terminal amino acids 883 ± Dephosphorylation is required to turn phospho-pRB into 887 (DEADG) (JaÈ nicke et al., 1996), we chose to an efficient caspase-3 substrate in vitro employ caspase-3 whose preferred sequence is a very The above experiments suggest that, when cells similar motif DEVDG (Lazebnik et al., 1994). Equal committed to enter S phase are treated with apoptotic amounts of pRB were incubated with Cdk/cyclin stimuli, pRB is dephosphorylated by increased PP1 complexes, either individually or jointly, in the activity and then cleaved by caspases. This could mean presence of [g-32P]ATP. After phosphorylation, phos-

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6115

Figure 4 pRB needs to be dephosphorylated in order to be a substrate for caspase-3. Human pRB, obtained from overexpression in Sf9 cells, was phosphorylated in vitro with the same three Cdk/cyclin complexes that are believed to phosphorylate pRB in vivo, and then incubated with caspase-3, either before or after dephosphorylation with PP1a. The ®gure shows the autoradiograph of immunoprecipitated pRB electrotransferred to Immobilon-P (top panels), and Western blots for pRB with sc-102 (center panels), and Rb-C (bottom panels) (cf. legend to Figure 2). The various additions to each reaction are indicated at the top of the ®gure

Figure 5 pRB isolated from cells committed to S phase entry is not a substrate for caspase-3. In this experiment we compared the susceptibility of recombinant pRB to caspase-mediated digestion to that of pRB immunoprecipitated from human HL-60 cells that had been synchronized at G1/S. It is believed that at this time most of the pRB sites are already phosphorylated. The left panel is identical to Figure 4, except that unphosphorylated pRB was also included: The results demonstrate that pRB was eectively proteolyzed before phosphorylation (lane 2) or after dephosphorylation (lane 6), but not after phosphorylation (lane 4). `Native' pRB from G1/S synchronized cells showed a proteolytic digestion pattern that was identical to that of recombinant pRB (central panel), while the addition of Cdk/cyclin complexes did not change pRB's reactivity with caspase-3 (right panel). The various additions to each reaction are indicated at the top of the ®gure

pho-pRB was incubated with caspase-3, or with PP1a the in vivo conditions during S phase) was able to ®rst, and then with caspase-3. 32P-incorporation into completely block caspase-3 digestion (Figure 4, bottom pRB shows that the three Cdk/cyclin complexes panels, 4th blot). Remarkably, Cdk2/cyclin E alone phosphorylated pRB either individually or jointly was as ecient in preventing pRB proteolysis (Figure under these reaction conditions (Figure 4, top panels). 4, bottom panels, 2nd blot). The other two kinases, Phosphorylation of pRB by all three Cdk/cyclin Cdk4/cyclin D1 and Cdk2/cyclin A, were able to delay complexes (the situation which most closely resembles caspase-3 action individually. However, prior dephos-

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6116 We reasoned that the observed behavior of PP1 in these cells would enable us con®rm and/or narrow down the role(s) of PP1 in apoptosis. Like HL-60 cells, CEM/0 cells showed a similar, i.e. approximately twofold, increase in PP1a protein (Figure 7a, left panel). This was associated with a small increase in PP1 activity (Figure 7b), an increased mobility of pRB, apparent cleavage of pRB (Figure 7c, left panel) as well as signs of apoptosis in response to araC treatment (Figure 7d, left panel). Whereas a twofold induction of PP1a occurred within 1 h of exposure to araC, we note that PP1d expression also increased about threefold, but not until 5 h after release from araC. Interestingly, Figure 6 Elevated PP1a activity triggers DNA fragmentation. treatment of araC-resistant CEM/I cells with araC did Non-synchronized HL-60 cells were electroporated in the presence not yield any changes in any PP1 isozyme, PP1 of wild-type PP1a, heat-inactivated PP1aT320A (h.i.PP1aCA), or constitutively active PP1aT320A (PP1aCA) as described (Berndt activity, pRB or DNA fragmentation (Figure 7, right et al., 1997). The degree of DNA fragmentation was then assessed panels). On the other hand, electroporation of CEM/I over a period of 3 h as described in Materials and methods. The cells with PP1aT320A resulted in changes identical to ®rst lane shows DNA size markers those produced by araC: The apoptotic index was signi®cantly increased (Figure 8a), pRB moved faster in SDS gels, and full-length pRB disappeared (Figure phorylation of phospho-pRB by PP1a resulted in 8b). The eciency of the electroporation is shown in ecient caspase-3 digestion of pRB (Figure 4, bottom Figure 8c: The amount of PP1aT320A increased panels, last lane of each blot). Thus it appears that approximately threefold. This suggests that CEM/I dephosphorylation of pRB is required to convert cells did not lose their ability to die in general, and that phospho-pRB into an ecient caspase substrate. PP1a may be a target for araC and other apoptotic Next, we compared caspase-3 digestion of in vitro agents. We con®rmed this hypothesis by the observa- phosphorylated pRB with that of pRB immunopreci- tion that both actinomycin D and etoposide resulted in pitated from G1/S arrested HL-60 cells, at which time induction of PP1a and cell death (data not shown). pRB would be almost maximally phosphorylated. The results are shown in Figure 5 and con®rmed our Inhibitor-2 negatively regulates apoptosis conclusion that pRB is only an ecient substrate for caspase-3 before phosphorylation (presumably in G0 Regardless of whether PP1a is necessary or sucient to through mid G1) and after dephosphorylation. pRB initiate or execute apoptosis, one would expect that that was isolated from cells synchronized at G1/S PP1 antagonists play a negative role in apoptosis. To behaved in a fashion identical to the recombinant pRB test this idea, the HL-60 cells shown in Figure 6 were phosphorylated by three Cdk/cyclin complexes (Figure also subjected to electroporation with inhibitor-2, the 5, left and center). question being whether this PP1-speci®c inhibitor counteracts araC- or PP1a-induced apoptosis. As shown in Figure 9a, inhibitor-2 was able to delay the PP1a activity is sufficient to induce signs of apoptosis onset of apoptosis mediated by ara-C. On the other We have shown earlier that introduction of moderate hand, inhibitor-2 eectively blocked apoptosis induced amounts of PP1a can cause both pRB+/+ and pRB7/7 by PP1aT320A (Figure 9c). Western blots for inhibitor- cells to die (Berndt et al., 1997). Since the focus of that 2 or the two PP1a species revealed that the electro- study was to identify roles of PP1 in the cell cycle, this poration resulted in a signi®cant uptake of the foreign ®nding was not stringently evaluated. Therefore, we proteins (Figure 9b,d). now asked the question whether PP1a can induce apoptosis. When HL-60 cells were electroporated in the presence of PP1, wild-type PP1a was unable to induce Discussion DNA fragmentation; likewise, heat-inactivated PP1a T320A had no eect. However, cells treated with the During the last few years, two concepts have emerged constitutively active form, PP1aT320A, showed signs that would suggest a negative role for protein of DNA laddering within 2 h after the electroporation phosphatases in the regulation of the cell number. (Figure 6). The apoptotic index of these cells increased First, protein phosphorylation is a major driving force to almost 20% within 3 h. This suggests that PP1a in cell cycle progression. Second, protein phosphoryla- activity is crucial and perhaps sucient to launch tion appears to support cell survival and to counteract apoptosis. apoptosis, e.g. by phosphorylating pro-apoptotic BAD Next, we examined the status of PP1 and pRB in or directly inhibiting apoptotic proteases of the caspase wild-type CEM/0 and araC-resistant CEM/I cells. The family. Both of these pathways involve protein kinase latter do not undergo apoptosis in response to high B (Cardone et al., 1998; Downward, 1998). Besides, a concentrations of araC (Martin-Aragon et al., 2000). positive role for protein phosphatases in apoptosis is

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6117

Figure 7 AraC-resistant cells fail to die and induce PP1. Human leukemia CEM cells were treated with 10 mM araC as described in Figure 1. Parental wild-type CEM/0 cells are shown on the left, araC-resistant CEM/I cells on the right. Cells were lysed in buer A or EBC buer and subjected to (a) Western blots for PP1, (b) PP1 activity assays, (c) Western blots for pRB, and (d) DNA fragmentation analysis. (a) Like HL-60 cells, parental wild-type CEM/0 cells rapidly induced PP1a in response to araC treatment, whereas the amount of PP1 isozymes did not change in araC-resistant CEM/I cells following exposure to araC. (b) PP1 activity was slightly elevated in CEM/0 cells (white), but not in CEM/I cells (black). The graph shows the average of two independent determinations. (c) In CEM/0 cells, pRB underwent apparent cleavage of the C-terminus, as indicated by increased mobility and loss of the intact molecule, whereas in CEM/I cells, no changes in pRB were detectable. (d) DNA fragmentation was only detectable in CEM/0, but not in CEM/I cells. The ®rst and last (unmarked) lanes of the two panels show DNA size markers, l DNA/HindIII fragments and FX174/HaeIII fragments, respectively supported by the following observations: Thus, PP2B- analysis and an understanding of apoptosis is dicult, mutants insensitive to calcium induce apoptosis in since it is more likely that it requires the co-operation neurons, and this may be due to PP2B-mediated of many proteins and pathways, rather than the dephosphorylation of BAD (Shibasaki and McKeon, activation of a single one (Green, 1998; Martin and 1995; Wang et al., 1999). Furthermore, short-term Green, 1995). A second major challenge is to under- inhibition of phosphatases with okadaic acid, calyculin stand the co-ordinated control of apoptosis with cell A, and cantharidin prevents DNA fragmentation, cycle progression or cellular homeostasis: We have PARP cleavage and pRB mobility shift by several shown previously that one of the major mammalian apoptosis-inducing agents (Morana et al., 1996). The protein phosphatases, PP1a, can induce pRB-depen-

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6118 How can pRB be a tumor suppressor and at the same time inhibit apoptosis? This question was a sucient reason for us to undertake the current study in which we focused on the interaction between pRB and PP1 during experimentally induced apoptosis. That PP1 directly and indirectly controls the cell cycle-dependent phosphorylation state of pRB has been established by several groups (Alberts et al., 1993; Berndt, 1995; Durfee et al., 1993; Ludlow et al., 1993). Using the anticancer drug araC as an apoptotic signal for HL-60 leukemia cells, we show here that PP1 is activated before cells execute cell death. This seemed to be largely due to a rather rapid increase in the expression level of PP1a. The transient activation of PP1a was accompanied by dephosphorylation of pRB, which was followed by (probably caspase-mediated) cleavage of pRB (see Figure 3). All of these events occurred within 2 h after releasing the cells from araC exposure, whereas the appearance of cells with 52n DNA content did not occur until 5 h (cf. Figure 1e). It should also be noted that the activation of PP1 occurred during S phase, when PP1 activity is at its cell cycle-dependent minimum (Dohadwala et al., 1994). Proteolytic digestion of pRB that was phosphorylated by Cdks in vitro, was either delayed or completely inhibited, whereas PP1a-mediated dephosphorylation enabled caspase-3 to cleave pRB (see Figure 4). Likewise, caspase-3 could easily cleave unphosphorylated pRB, the form that prevails during G0 and most of G1 phases (see Figure 5). It is tempting to speculate that araC induces an abrupt halt of DNA replication (S phase), that this alone might be sucient to induce apoptosis, and that the phenomena we observed here are secondary eects. However, such an explanation appears unlikely as PP1a activity was found to induce cell death (Berndt et al., 1997) or apoptosis (this study, see Figure 6). Furthermore, the transient induction and activation of PP1a (see Figure 1) did not depend on expression of Figure 8 PP1a activity induces pRB cleavage and apoptosis in pRB. AraC treatment led to a more marked increase in araC-resistant cells. CEM/I cells were electroporated in the PP1a protein in pRB7/7 Saos-2 cells (data not shown). presence of 10 mg BSA or 10 mg PP1aT320A as stated for Figure The failure of araC-resistant CEM/I cells to induce 6. (a) The apoptotic index (average of two experiments) was PP1a in response to araC treatment would indicate determined over a period of 4 h following electroporation with BSA (green) or PP1aT320A (red). (b) pRB was proteolytically that PP1a activity is necessary for apoptosis (see Figure cleaved in response to PP1aT320A, but not BSA, as judged by the 7a,b). However, another interpretation is more likely: loss in apparent Mr (sc-102), and the loss of the intact molecule The fact that inhibitor-2 completely abolished PP1a- (Rb-C). (c) Western blots of the corresponding samples show that mediated apoptosis, yet only delayed the onset of the levels of PP1aT320A in treated cells were transiently elevated araC-induced apoptosis (see Figure 9), supports the about threefold idea that PP1a may simply be one (of many?) protein(s) that is capable of launching apoptosis. Thus, one or several functions carried out by PP1a may be dent cell cycle arrest (Berndt et al., 1997). The interest sucient to trigger apoptosis. Since inhibition of PP1a, in pRB function, its regulators and downstream targets and presumably other PP1 isoforms, does not prevent is of course justi®ed by its assumed role as a tumor araC-induced apoptosis (see Figure 9a), araC and suppressor. As apoptosis ultimately results in termina- perhaps other pro-apoptotic signals cannot rely on PP1 tion of the cell cycle, it is not surprising that in the last alone, in other words, PP1 is not absolutely required few years, several proteins that control cell cycle for apoptosis. Nonetheless, the induction of apoptosis progression have also been implicated in apoptosis. in CEM/I cells following electroporation with PP1a One of them is pRB. T320A underlines the critical role of PP1 activity. As was alluded to in the Introduction, early studies Together, our ®ndings as well as those of other on the roles of pRB in apoptosis created a paradox: laboratories support the notion that there are several

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6119

Figure 9 Inhibitor-2 functions as an inhibitor of apoptosis. HL-60 cells were induced to undergo apoptosis by (a,b) exposure to araC (as in Figure 1) or (c,d) electroporation with PP1aT320A (as in Figure 6). In both cases the apoptotic index was measured over a time period of 3 h following exposure to araC or electroporation. (a) 1 h before adding araC for 2 h, cells were electroporated in the presence of the inert control protein bovine serum albumin BSA (red) or inhibitor-2 (blue). (b) Western blots show the about twofold increase in the amount of inhibitor-2 following electroporation (left blot vs right blot). (c) The apoptotic index (average of two experiments) was measured for up to 3 h. At time=0 h, the cells were electroporated with BSA (the apoptotic index was zero during the observation period), inhibitor-2 (dark green), PP1a (light green), PP1aT320A, (red), or PP1aT320A+inhibitor-2 (blue). (d) Western blots for PP1a (top) or inhibitor-2 (bottom) show the eciency of the electroporation distinct and independent roads to cell death. Our the case is consistent with a recent study showing that ®ndings appear to be able to resolve the aforemen- PP1a dephosphorylates and activates BAD (AylloÂ n et tioned paradox with regards to pRB function: the anti- al., 2000). We have also observed that in HL-60 and apoptotic function of pRB is most certainly restricted HeLa cells, PP1a was associated with procaspase-9 and to the phosphorylated form of the molecule, which phosphorylated BAD (Rui-Hong Wang and Norbert prevails during S phase, a time when cells are Berndt, unpublished). replicating DNA. This interpretation would also be In summary, our data indicate that hypo- or consistent with a recent study reporting that constitu- unphosphorylated pRB is at least compatible with, if tively active (phosphorylation-resistant) pRB causes not required for, apoptosis. Also, inhibition of both cell cycle arrest and apoptosis (Knudsen et al., apoptosis by phospho-pRB can more easily be 1999). reconciled with its role as a tumor suppressor and The data presented here, together with earlier results with a recent report describing that phospho-pRB from our laboratory, support the hypothesis that the stimulates DNA polymerase (Takemura et al., 1997). PP1-pRB interaction is an important, yet non-essential Most importantly, we have presented evidence that a step for cells launching apoptosis. While it may be protein phosphatase acts to recruit an apparent death necessary to dephosphorylate pRB in cells that express substrate such as pRB to a caspase, thereby it, we cannot rule out that this function may be permitting its proteolytic destruction. Our ®ndings assumed by a phosphatase other than PP1a. Our suggest that PP1a functions to activate apoptosis and ®ndings that PP1d increased in response to araC, albeit may be sucient to trigger apoptosis, and may be much later than PP1a, are in agreement with a report able to overcome drug resistance under speci®c describing a crucial interaction between PP1d and circumstances. Furthermore, since the inability to be pRB, although those experiments involved mitotic phosphorylated (inhibited) by Cdks enabled PP1a to HeLa cells (Puntoni and Villa-Moruzzi, 1999). At any induce apoptosis, these ®ndings suggest yet another rate, we have already reported that Saos-2 cells, which link between regulation of the cell cycle and lack functional pRB, also die following electroporation programmed cell death. A future challenge will be to in the presence of PP1a (Berndt et al., 1997). This understand to what extent PP1's functions in both cell suggests that PP1a is able to utilize more than one cycle control and apoptosis are integrated or regulated pathway to trigger or facilitate apoptosis. That this is dierentially.

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6120 kinases and ATP and incubated with 20 units of PP1a at Materials and methods 308C for 20 min. PP1a-treated pRB (on beads) was washed 6 ± 10 times with EBC buer, resuspended in 150 mlof Cells, vectors and proteins digestion buer (50 mM HEPES, pH 7.4, 100 mM NaCl, Human promyelocytic leukemia HL-60 cells were obtained 0.1% CHAPS, 1 mM EDTA, 10% glycerol, 10 mM DTT) from the ATCC, araC-resistant leukemia CEM/I cells were plus 50 ml of 2 units/ml caspase-3 and incubated at 378C obtained from parental CEM/0 cells as described (Martin- overnight. The samples were concentrated by Microcon YM- Aragon et al., 2000). 10 centrifugal ®lter devices (Millipore), separated by SDS ± Recombinant PP1a was obtained via overexpression in PAGE and analysed by autoradiography and pRB Western E.coli as described (Dohadwala and Berndt, 1998). Phos- blot. phorylase b and phosphorylase kinase, both necessary to prepare PP1 substrate, was a gift from Balwant S Khatra Electrotransfer of purified proteins into HL-60 or CEM/I cells (California State University, Long Beach, CA, USA). Recombinant baculoviruses encoding human pRB was a gift To introduce foreign proteins into cells, we modi®ed the from Barbara Driscoll (Childrens Hospital, Los Angeles, CA, procedure described in detail (Berndt et al., 1997). Approxi- USA), and baculoviruses encoding mammalian Cdk 2, Cdk4, mately 26106 cells in 200 ml were subjected to a pulse of cyclin D1, cyclin E, and cyclin A were kindly provided by 200 V at a capacitance setting of 250 mF in the presence of Charles J Sherr (St. Jude's Childrens Research Hospital, 200 units of PP1a or PP1aT320A, or 10 mg of inhibitor-2. Memphis, TN, USA) and David O Morgan (University of California, San Francisco, CA, USA), respectively. Apoptosis assays Antibodies to the three ubiquitously expressed isozymes of PP1 were obtained from Emma Villa-Moruzzi (University of To determine cell viability, cells were stained with trypan blue Pisa, Italy). Two pRB antibodies were used: sc-102 (speci®c immediately after collection. To determine DNA fragmenta- for the centrally located A/B pocket region, Santa Cruz tion, approximately 106 cells were lysed with a solution Biotechnology) and Rb-C (speci®c for the C-terminus of consisting of 10 mM tris-HCl, pH 8.0, 10 mM EDTA, and pRB, New England Biolabs). Inhibitor-2 was purchased from 1% Triton X-100. The post-nuclear lysates were obtained by Upstate Biotechnology, Inc., the antibody to inhibitor-2 was spinning the cell lysates at 48C at 14 000 r.p.m. for 30 min. from BD Transduction Laboratories, and caspase-3 from Phenol-chloroform extraction was performed and ethanol Bio-Mol. All other materials were from Sigma, unless precipitation was carried out at 7208C overnight. The otherwise stated. precipitated DNA was loaded onto a 2% agarose gel after RNase A treatment (Yi et al., 1997). To determine the apoptotic index, cells were stained with Hoechst 33258 Cell culture and metabolic labeling (Spector et al., 1998); cells showing irregular nuclear staining HL-60 cells were maintained in RPMI-1640 medium (due to chromatin condensation and nuclear fragmentation) supplemented with 10% fetal bovine serum and antibiotics were scored as apoptotic (Spector et al., 1998), and divided (100 units/ml of penicillin G and 100 mg/ml streptomycin). by the number of all Hoechst 33258-positive cells. After synchronizing the cells at the G1/S boundary with 5 mg/ ml aphidicolin for 16 h, the cells were washed twice to Flow cytometry analysis remove the drug, and araC (Bedford Laboratories, dissolved in phosphate-buered saline) was added to reach a ®nal 26106 cells per sample were washed twice with PBS, ®xed in concentration of 10 mM for 2 h. For pRB phosphorylation 70% icecold ethanol and incubated overnight at 7208C. The analyses, 16106 cells were metabolically labeled for 2 h with cells were spun down, washed twice with PBS, and extracted 32 200 mCi P-orthophosphate (ICN) in 1 ml (Berndt et al., with a buer (45 mM Na2HPO4, 2.5 mM citric acid, 0.1% 1997; Liu et al., 1999) in the presence of 10 mM araC. The Triton X-100) at room temperature for 30 min. The cells were washed several times to remove araC (and 32P) and extraction buer was washed away with PBS and the cells incubated further. The cells were collected, washed twice with were digested with 10 mg/ml RNase A (DNase free) at 378C PBS and split into several aliquots for further analyses. for 20 min (Hotz et al., 1994). Propidium iodide was added to each sample at a ®nal concentration of 20 mg/ml. The samples were kept on ice for 1 h before being analysed in a In vitro pRB phosphorylation, dephosphorylation and ¯ow cytometer (FACS 440, Becton-Dickinson). To estimate caspase-3 digestion the portion of cells in G1 and G2, the resulting histograms pRB, Cdk2/cyclin A, Cdk2/cyclin E, Cdk4/cyclin D1 were were analysed with ModFit LT 2.0 software. Cells with a expressed in Sf9 cells by recombinant baculovirus infection. DNA content less than 2n were considered to be undergoing Sf9 cells (107) were co-infected with baculoviruses encoding (or committed to) apoptosis (Spector et al., 1998). cyclin A or cyclin E and Cdk2, cyclin D1 and Cdk4 and lysed in kinase reaction buer (20 m Tris-HCl, pH 7.5, 20 mM M Immunoprecipitation, electrophoresis, and Western blotting MgCl2,10mM b-mercaptoethanol supplemented with 150 mM NaCl and 0.5% NP-40 plus the following protease Washed cells were lysed in EBC buer, and the protein and phosphatase inhibitors: 2 mg/ml aprotonin, 1 mM concentration was determined with the Bio-Rad protein assay benzamidine, 0.5 mg/ml leupeptin, 1 mg/ml pepstatin A, (Bradford, 1976). 1.2 mg of total protein from each sample 0.1 mM PMSF, 25 mM NaF, 1 mM sodium b-glyceropho- was pre-cleared with 1 mg of normal mouse IgG plus 20 mlof sphate, 1 mM Na3 VO4). Clari®ed lysates were used to protein G-sepharose (Santa Cruz Biotechnology) in 1 ml of phosphorylate pRB immunoprecipitated from pRB-over- EBC buer at 48C for 4 h. The suspensions were centrifuged expressing Sf9 cells in the presence of [g-32P]ATP (Mihara and the supernatants were mixed with 2 mg of sc-102 and et al., 1989). After phosphorylation, the protein G beads were rotated overnight at 48C. Twenty ml of protein G-sepharose washed 6 ± 10 times with EBC buer (50 mM tris-HCl, beads were added to the mixture 4 h before the beads were pH 8.0, 150 mM NaCl, 0.5% NP-40) to remove the protein centrifuged, recovered and washed with EBC buer at least

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6121 six times at 48C. Finally, 20 ± 40 mlof26 SDS ± PAGE proteins were visualized with the ECL reagents according loading buer was used to resuspend each precipitate. The to the manufacturer's instructions (Amersham). The slow and samples were boiled for 10 min and centrifuged brie¯y to fast migrating forms of pRB are indicated by arrows on the remove insoluble material. righthand side of the ®gures. Cells were lysed with EBC buer (for pRB Western blot) Autoradiographs and Western blots were scanned with a or lysis buer A (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, GS300 densitometer (Hoefer Scienti®c Instruments) to 1% NP-40, 0.1 mM EDTA, 0.1 mM EGTA, for PP1 Western estimate the relative amount of proteins. The intensity of blot), and the lysates or immunoprecipitated proteins were the bands was determined with GS370 v.3 software. All separated by SDS ± PAGE (Novex, 12% Tris-glycine gel for statements referring to quantitative changes in band PP1, 6% Tris-glycine gel for pRB) (Laemmli, 1970). Care was intensities in the text or ®gure legends are based on these taken to load an equal number of cells or an equal protein analyses. amount per lane. After electrophoresis, the samples were electrotransferred to Immobilon-P (Millipore) at a constant PP1 activity assay voltage of 30 V. Membranes containing 32P-labeled pRB were exposed to Kodak X-Omat AR ®lm. For immunoblotting, PP1 activity was measured by using 32p-phosphorylase a as a the membranes were blocked overnight at room temperature substrate as described previously (Dohadwala et al., 1994). with 5% non-fat milk in TBS (25 mM Tris-HCl, pH 8.0, 137 mM NaCl, 2.7 mM KCl) and incubated with the primary antibodies to PP1 isozymes or pRB. They were diluted with 2% BSA in TBST (TBS containing 0.2% tween-20) and Acknowledgments added to the membranes for 2 h at room temperature. After This work was supported in part by NIH grant CA54167 washing with TBST for 3615 min, secondary antibodies (to N Berndt). We wish to thank Balwant S Khatra, diluted with 2% BSA in TBST were added for 1 h at room Charles J Sherr, David O Morgan, and Emma Villa- temperature. After washing 3615 min with TBST, the Moruzzi for their kind sharing of vectors and antibodies.

References

Alberts AS, Thorburn AM, Shenolikar S, Mumby MC and Dou QP, An B and Will PL. (1995). Proc. Natl. Acad. Sci. Feramisco JR. (1993). Proc. Natl. Acad. Sci. USA, 90, USA, 92, 9019 ± 9023. 388 ± 392. Downward J. (1998). Curr. Opin. Cell Biol., 10, 262 ± 267. Axton JM, DoÂ mbradiV,CohenPTWandGloverDM. Dunigan DD and Madlener JC. (1995). Virology, 207, 460 ± (1990). Cell, 63, 33 ± 46. 466. AylloÂ nV,MartõÂ nez-A C, GarcõÂ a A, Cayla X and Rebollo A. Durfee T, Becherer K, Chen P-L, Yeh S-H, Yang Y, Kilburn (2000). EMBO J., 19, 2237 ± 2246. AE, Lee W-H and Elledge SJ. (1993). Genes Dev., 7, 555 ± Baxter GD and Lavin MF. (1992). J. Immunol., 148, 1949 ± 569. 1954. Fan G, Ma X, Kren BT and Steer CJ. (1996). Oncogene, 12, Berndt N. (1995). Adv. Prot. Phosphatases, 9, 63 ± 86. 1909 ± 1919. Berndt N. (1999). Front. Biosci., 4, d22 ± d42. Fernandez A, Brautigan DL and Lamb NJC. (1992). J. Cell Berndt N. (2000). Emerg. Ther. Targets, 4, 581 ± 608. Biol., 116, 1421 ± 1430. Berndt N, Dohadwala M and Liu CWY. (1997). Curr. Biol., Green DR. (1998). Cell, 94, 695 ± 698. 7, 375 ± 386. Haas-Kogan DA, Kogan SC, Levi D, Dazin P, T'Ang A, BerryDE,LuY,SchmidtB,FallonPG,O'ConnellC,HuS- Fung Y-KT and Israel MA. (1995). EMBO J., 14, 461 ± X, Xu H-J and Blanck G. (1996). Oncogene, 12, 1809 ± 472. 1819. Hotz MA, Gong J, Traganos F and Darzynkiewicz Z. (1994). Bùe R, Gjertsen BT, Vintermyr OK, Houge G, Lanotte M Cytometry, 15, 237 ± 244. and Dùskeland SO. (1991). Exp. Cell Res., 195, 237 ± 246. Inomata M, Saijo N, Kawashima K, Kaneko A, Fujiwara Y, Booher RN and Beach D. (1989). Cell, 57, 1009 ± 1016. Kunikane H and Tanaka Y. (1995). J. Cancer Res. Clin. Bradford MM. (1976). Anal. Biochem., 72, 248 ± 254. Oncol., 121, 729 ± 738. Cardone MH, Roy N, Stennicke HR, Salvesen GS, Franke JacksT,FazeliA,SchmittEM,BronsonRT,GoodellMA TF, Stanbridge EJ, Frisch S and Reed JC. (1998). Science, and Weinberg RA. (1992). Nature, 359, 295 ± 300. 282, 1318 ± 1321. JaÈ nicke RU, Walker PA, Lin XY and Porter AG. (1996). Chen W, Otterson GA, Lipkowitz S, Khleif SN, Coxon AB EMBO J., 15, 6969 ± 6978. and Kaye FJ. (1997). Oncogene, 14, 1243 ± 1248. Knudsen KE, Weber E, Arden KC, Cavenee WK, Feramisco Cheng AY, Dean NM and Honkanen RE. (2000). J. Biol. JR and Knudsen ES. (1999). Oncogene, 18, 5239 ± 5245. Chem., 275, 1846 ± 1854. Kwon Y-G, Lee S-Y, Choi Y, Greengard P and Nairn AC. Clarke AR, Maandag ER, Van Roon M, Van der Lugt (1997). Proc. Natl. Acad. Sci. USA, 94, 2168 ± 2173. NMT, Van der Valk M, Hooper ML, Berns A and Te Riele Laemmli UK. (1970). Nature, 227, 680 ± 685. H. (1992). Nature, 359, 328 ± 330. Lazebnik YA, Kaufmann SH, Desnoyer S, Poirier GG and Damer CK, Partridge J, Pearson WR and Haystead TAJ. Earnshaw WC. (1994). Nature, 371, 346 ± 347. (1998). J. Biol. Chem., 273, 24396 ± 24405. LeeEYHP,ChangC-Y,HuN,WangY-CJ,LaiC-C,Herrup Dohadwala M and Berndt N. (1998). Protein phosphatase K, Lee W-H and Bradley A. (1992). Nature, 359, 288 ± 294. protocols, Vol 93. Ludlow JW (ed). Humana Press: LiuCWY,WangR-H,DohadwalaM,SchoÈ nthal AH, Villa- Totowa, NJ, pp 191 ± 199. Moruzzi E and Berndt N. (1999). J. Biol. Chem., 274, Dohadwala M, Da Cruz e Silva EF, Hall FL, Williams RT, 29470 ± 29475. Carbonaro-Hall DA, Nairn AC, Greengard P and Berndt Ludlow JW, Glendening CL, Livingston DM and DeCaprio N. (1994). Proc. Natl. Acad. Sci. USA, 91, 6408 ± 6412. JA. (1993). Mol. Cell. Biol., 13, 367 ± 372.

Oncogene Roles of PP1 in apoptosis R-H Wang et al 6122 Martin-Aragon S, Mukherjee SK, Taylor BJ, Ivy SP, Fu CH, Spector DL, Goldman RD and Leinwand LA. (1998). Cells ± Ardi VC and Avramis VI. (2000). Anticancer Res., 20, A Laboratory Manual. Cold Spring Harbor Laboratory 139 ± 150. Press: Cold Spring Harbor, NY. Martin SJ and Green DR. (1995). Cell, 82, 349 ± 352. Takemura M, Kitagawa T, Izuta S, Wasa J, Takai A, Mihara K, Cao X-R, Yen A, Chandler S, Driscoll B, Akiyama T and Yoshida S. (1997). Oncogene, 15, 2483 ± Murphree AL, T'Ang A and Fung Y-KT. (1989). Science, 2492. 246, 1300 ± 1303. Tan X and Wang JYJ. (1998). Trends Cell Biol., 8, 116 ± 120. Morana SJ, Wolf CM, Li J, Reynolds JE, Brown MK and Taylor BK, Stoops TD and Everett AD. (2000). Am. J. Eastman A. (1996). J. Biol. Chem., 271, 18263 ± 18271. Physiol., 278, L1062 ± L1070. Morimoto Y, Ohba T, Kobayashi S and Haneji T. (1997). WangHG,PathanN,EthellIM,KrajewskiS,YamaguchiY, Exp. Cell Res., 230, 181 ± 186. Shibasaki F, McKeon F, Bobo T, Franke TF and Reed Ohkura H, Kinoshita N, Miyatani S, Toda T and Yanagida JC. (1999). Science, 284, 339 ± 343. M. (1989). Cell, 57, 997 ± 1007. Yamano H, Ishii K and Yanagida M. (1994). EMBO J., 13, Puntoni F and Villa-Moruzzi E. (1999). Biochem. Biophys. 5310 ± 5318. Res. Commun., 266, 279 ± 283. Yi H-K, Fujimura Y, Ouchida M, Prasad DDK, Rao VN SchoÈ nthal AH and Feramisco JR. (1993). Oncogene, 8, 433 ± and Reddy SP. (1997). Oncogene, 14, 1259 ± 1268. 441. Zheng B, Woo CF and Kuo JF. (1991). J. Biol. Chem., 266, Shibasaki F and McKeon F. (1995). J. Cell Biol., 131, 735 ± 10031 ± 10034. 743.

Oncogene