Caspase-dependent cleavage of the retinoblastoma protein is ... - Nature

Oncogene (2000) 19, 2171 ± 2178 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc

Caspase-dependent cleavage of the retinoblastoma protein is an early step in neuronal apoptosis Anne-Laurence Boutillier1, Emmanuelle Trinh1 and Jean-Philippe Loeer*,1 1

UMR 7519 CNRS -21, rue ReneÂ Descartes, 67 084 Strasbourg Cedex, France

Rb-de®cient embryos (Rb7/7) show abnormal degeneration of neurons and die at mid-gestation, suggesting that RB may protect against apoptosis. Having previously shown that cyclin D1 accumulates during K+-induced apoptosis of granule neurons, we chose to investigate the role of RB under these conditions. We show that RB is cleaved in its C-terminus during the onset of neuronal apoptosis. Caspase 3-like activity increases following K+ deprivation and the time course correlates with RB cleavage and apoptosis. Although the use of a speci®c caspase 3-like inhibitor (z-DEBD.fmk) delays RB cleavage and reduces DNA fragmentation, data implicate other caspases in these processes. However, K+ deprivation induces a gradual production of the active p20 subunit of caspase 3 (CPP32) that coincides with RB disappearance at the cellular level. Nuclear detection of a transfected HA-tagged caspase cleavage-resistant RB mutant (DEAG/D to DEAA/D) revealed a signi®cant decrease in apoptosis of neurons expressing the RB mutant (less than 5%) relative to the wild type form of RB (40%) during K+ deprivation. Taken together, these data show that caspase-dependent cleavage of RB is an early permissive step of the apoptosis-inducing signaling pathway in neurons. They indicate a major role of RB in neuronal protection. Oncogene (2000) 19, 2171 ± 2178. Keywords: apoptosis; cerebellar neuron; retinoblastoma protein; caspase; CPP32

Introduction Apoptosis or programmed cell death (PCD) is a major physiological phenomenon that occurs massively during development and participates in the maintenance of homeostasis in the adult. During development of the central nervous system (CNS), PCD along with proliferation and dierentiation, ensures the proper formation of the neuronal network, almost half of the neurons produced during embryogenesis dying before reaching adulthood (Oppenheim, 1991). Apoptosis can be mimicked in primary cerebellar granule neurons (CGN) cultured in vitro, by switching external K+ from high (HK: [KCl]=30 mM) to low (LK: [KCl]=5 mM) concentrations (D'Mello et al., 1993; Kienlen-Campard et al., 1997; Schulz et al., 1996). We have recently shown in this model, that the cell cycle regulator cyclin D1 (CD1) accumulates following K+ deprivation (Boutillier

*Correspondence: J-P Loeer Received 1 November 1999; revised 14 February 2000; accepted 14 February 2000

et al., 1999), in agreement with the hypothesis that neuronal apoptosis could result from an abortive attempt to reinitiate the cell cycle (Heintz, 1993). A direct cellular target of CD1 is the retinoblastoma protein (RB) that governs the restriction point of the G1/S phase of the cell cycle. Interestingly, RB-dependent transcriptional activity is down-regulated by Ca2+ entry and subsequent Ca2+/Calmodulin Kinases activation (Sohm et al., 1999), a pathway that is thought to allow survival of CGN cultures in HK conditions (Boutillier et al., 1999; Gallo et al., 1987). Moreover, RB probably has neuroprotective functions as pRB knockout mice die at mid-gestation, presenting massive cell death in the central and peripheral nervous systems (Clarke et al., 1992; Jacks et al., 1992; Lee et al., 1992) and loss of RB family members is often associated with apoptosis (Haas et al., 1995; Helin et al., 1997; Slack and Miller, 1996). During apoptosis, RB can be cleaved proteolytically by the caspase family of proteases, either in its Cterminal caspase consensus cleavage site (DEADG), generating a 42 aa truncated RB protein (Janicke et al., 1996; Tan et al., 1997) or in an internal cleavage site, producing two fragments (p68 and p48) (An and Dou, 1996). The C-terminus cleavage occurs during TNF or anti-CD95 induced apoptosis of dierent cell lines (Janicke et al., 1996; Tan et al., 1997). However, even though a non-degradable RB mutant could protect from TNF-induced apoptosis, it did not attenuate death induced by anti-CD95. This suggests that RB cleavage by caspases is cell-speci®c, being an important event only in certain types of apoptosis (Dou and An, 1998; Tan et al., 1997; Tan and Wang, 1998). More than 10 interleukin-1b-converting enzyme (ICE)-like proteases or caspases have been identi®ed (Alnemri et al., 1996), which speci®cally cleave after aspartic acid residues. Caspases are clearly involved in the execution of the death program in many cell types (Henkart, 1996; Kidd, 1998; Yuan, 1995; Yuan et al., 1993), and are activated in neurodegenerative diseases (for review, Friedlander and Yuan, 1998). Deletion of the CPP32 (caspase 3) gene in mice results in defects in the brain, primarily aecting neuronal cells which are supernumerary and disorganized, thus leading to premature embryonic lethality (Kuida et al., 1996). We used the model of K+ deprivation of cerebellar neurons to analyse RB function in the apoptosis signaling pathway of primary central neurons. We found that RB is cleaved in its C-terminus during the onset of apoptosis, by a caspase 3-like dependent mechanism. Overexpression of a cleavage-resistant RB mutant prevents nuclei fragmentation, suggesting that cleavage of RB is an obligatory step for neurons to enter apoptosis.

RB-dependent neuronal apoptosis A-L Boutillier et al

2172

Results The RB C-terminus is cleaved during the onset of neuronal apoptosis by a caspase sensitive mechanism We investigated the role of the RB protein, ®rst by carrying out Western blot analysis on 6-day-old CGN cultures, potassium-deprived for dierent periods of time. Using an antibody directed to RB C-terminus (C15; amino acids 914 ± 928), we found that RB rapidly disappears as a function of LK treatment, a signi®cant reduction being observed between 4 and 8 h (Figure 1a). This disappearance of RB correlates with an increase in nuclear condensation/fragmentation in neurons, an event representative of apoptosis in response to LK treatment (Table 1). Degradation of RB is speci®c, since no signi®cant changes in the amount of b-actin were detected after K+ deprivation (Figure 1c). Interestingly, in other models of apoptosis such as TNF-treated HeLa D98 cells or anti-CD95 treated Jurkat/T cells, RB is cleaved by a caspase at its C-terminus 883-DEAD/G-887 site (Janicke et al., 1996; Tan et al., 1997). This cleavage generates a 4.2 KDa fragment and a faster migrating cleaved form of RB (DRB, p100; Tan et al., 1997). In order to visualize whether DRB is produced in our model, we next performed Western blots with the G3-245 antibody that recognizes an internal portion of RB (amino acids 332 ± 344; Ludlow et al., 1989). As seen in Figure 1b, the total amount of RB protein is also reduced following K+ deprivation. A faint band can be seen at 4, 8 and 10 h with the G3-245, that is likely to be DRB, more perceptible on a longer exposure of an 8 h LK treatment (Figure 1b, inset). However, it was never

prominent and could not be formally characterized as being DRB (Figure 1b). An interior cleavage of RB has been described by Fattman et al. (1997), that gives two fragments of 48 (N-terminal) and 68 KDa (C-terminal). It seems unlikely that this occurs in our conditions, as neither the 48 KDa moiety was detected by Western blot with the G3-245 antibody, nor the 68 KDa C-terminus fragment with the C-15 antibody (data not shown). Taken together, these results suggest that RB is cleaved on its C-terminus during CGN apoptosis. To understand whether RB cleavage observed in our model is caspase dependent or not, we used caspase inhibitors. Figure 1d shows that the general caspase inhibitor z-VAD.fmk protects RB C-terminus degradation for up to 15 h. A caspase 3-like inhibitor (zDEVD.fmk) eciently delays RB degradation for up to 9 h. These results also correlate with the protective potency of the two inhibitors against K+-induced apoptosis (Table 1): z-VAD.fmk better protects CGN apoptosis than z-DEVD.fmk, which only reverses the percentage of apoptotic nuclei by less than 50% until 8 h. With a longer period of LK treatment, neuroprotection by z-DEVD.fmk further decreases (Table 1). These data suggest that caspase-dependent degradation of RB on its C-terminus may participate in neuronal apoptosis. Involvement of activated p20 subunit of CPP32 in K+-induced CGN apoptosis Degradation of RB is reversed by inhibitors of the DEVD- and YVAD-types (Dou and An, 1998; Janicke et al., 1996; Tan et al., 1997), suggesting that two

Figure 1 Degradation of RB C-terminus during K+-induced CGN apoptosis is aected by caspases inhibitors z-VAD.fmk and zDEVD.fmk. Lysates from neurons K+-deprived for dierent periods of time were submitted to immunoblot analysis with a polyclonal antibody (C-15), that recognizes the 15 C-terminal amino acids of RB (a) or with a monoclonal antibody (G3-245) directed to a central portion of RB (b). Full-length RB is indicated by arrows and its cleaved product (DRB) is marked with an asterisk. Inset in b: The DRB fragment is seen on a longer exposure of the blot (8 h of LK-treatment. G3-245 antibody). (c) Control immunoblotting with a monoclonal anti b-actin antibody. (d) Immunoblot analysis using C-15 antibody in neurons K+-deprived for 6, 9 and 15 h (upper panel) in the presence of the general caspase inhibitor z-VAD.fmk (50 mM, middle panel) or the caspase 3-like inhibitor z-DEVD.fmk (50 mM, lower panel) Oncogene


Table 1

Eects of the caspases inhibitors: z-VAD.fmk or z-DEVD.fmk on K+ deprivation-induced CGN apoptosis Percentage of apoptotic nuclei+s.e.m. 8h 10 h

4h HK LK LK+z-VAD.fmk LK+z-DEVD.fmk

2.65+0.61 4.01+0.75 2.21+0.25 2.59+0.72

3.95+0.46 19.87+1.43a 6.39+0.91b 10.53+1.35a,b

3.52+0.44 33.48+4.74a 4.92+0.56b 18.75+1.95a,b

2173

16 h 2.71+0.49 40.2+1.38a 5.23+1.16b 32.6+2.6a,c

Neurons were treated with LK medium (K+=5 mM) for 4 ± 16 h with or without z-VAD.fmk (50 mM) or z-DEVD.fmk (50 mM), then ®xed and stained with the Hoechst dye 33342. Apoptotic nuclei were scored on a morphological criteria: condensation of chromatin and nuclear fragmentation (see Figure 4). The results of three dierent experiments are given, expressed as the mean percentage of apoptotic nuclei+s.e.m. ANOVA followed by a Student ± Newman ± Keuls multiple comparison test were performed (Graphpad Instat 2 Software). aStatistically very signi®cant increase (P50.001) relative to HK-treated controls from the same time point. bStatistically very signi®cant decrease (P50.001) relative to LK-treated cells from the same time point. cStatistically signi®cant decrease (P50.05) relative to LK-treated cells from the same time point

distinct caspases are involved in this process. We then tested DEVD- and YVAD-cleaving activities in CGN with the use of ¯uorescent tetrapeptide substrates. The Ac-YVAD-AMC substrate was not cleaved in K+deprived cell extracts (Figure 2a), arguing against the possible intervention of the caspase 1-like family of caspases in RB cleavage during apoptosis of CGN. The Ac-DEVD-AMC substrate was preferentially cleaved in K+-deprived cell extracts (Figure 2b; Armstrong et al., 1997; Lynch et al., 1997; Marks et al., 1998). Cleavage of this substrate started at 3 ± 4 h of K+ deprivation, with a maximum of 10-fold at 10 h. The best characterized DEVD substrate preferring caspase is CPP32, whose proteolytic cleavage by upstream caspases generates an active p20 subunit (Armstrong et al., 1997; Nicholson et al., 1995). Figure 3a shows that p20 is absent in neuroprotective concentrations of K+ (HK), and gradually increases following K+ deprivation. As expected, the induction of p20 by LK treatment (8 h, Figure 3b, lane 2) is not reversed by a z-DEVD.fmk inhibitor (lane 3), while it is by zVAD.fmk (lane 4) which probably blocks upstream caspase (Nicholson et al., 1995). This set of results is consistent with the involvement of CED3 caspases (DEVD-sensitive) rather than ICElike caspases (YVAD-sensitive) in K+-induced CGN apoptosis, and points to the activation of CPP32 at the levels of its processing and activity. Localization of RB and activated CPP32 at the cellular level To examine the possible role of CPP32 in RB cleavage, we looked by immunocytochemistry whether the appearance of active CPP32 p20 fragment coincides with the loss of RB C-terminus. Cultures were either stained with the C-15 antibody recognizing the fulllength RB or with the anti-p20 antibody (Figure 4). The Hoechst labeling allowed us to show three distinct nuclear morphologies: `type 1' nuclei are large (5 mm) and round, and scored as representative of healthy neurons as typically seen in high K+ concentrations (Figure 4a,e); in `type 2' nuclei, chromatin starting to condense is rejected around the periphery of the nucleus; and `type 3' nuclei are very condensed or fragmented. In HK (Figure 4b), the full-length RB protein localizes in nuclei and cytoplasm of all neurons. After 10 h of LK treatment (Figure 4c), localization depends on the nuclear morphology: neurons with `type 1' nuclei display the same RB staining (Figure 4d) as in HK (Figure 4b). In contrast,

Figure 2 Induction of caspase 3- but not caspase 1-like activity after K+ withdrawal. Time course of cleavage of the ¯uorescent tetrapeptide substrates Ac-YVAD-amc (50 mM) (a) or Ac-DEVDamc (50 mM) (b) in extracts of K+-deprived granule neurons. Values represent the mean+s.e.m. of three independent experiments performed in triplicates and are expressed in nmoles of cleaved substrate per mg of protein per minute. Statistical signi®cances compare to HK (0) indicated by **P500.1, (unpaired t-test)

in apoptotic neurons with `type 2' or `type 3' nuclei, the uncleaved RB protein is mainly cytoplasmic and disappears as chromatin condenses (Figure 4d). If we compared these observations to those obtained with the active p20 subunit labeling, we can see that: (i) p20 is absent of healthy neurons harboring `type 1' nuclei (Figure 4f, h); (ii) in neurons with `type 2' nuclei, p20 is nuclear (Figure 4h; and (iii) neurons with `type 3' Oncogene


2174

A

B

Figure 3 K+ deprivation activates CPP32. (a) Time course of p20 active CPP32 appearance after various periods of K+ withdrawal analysed by Western blotting with a rabbit polyclonal antibody to the p20 subunit of CPP32 (Mr20 kDa). (b) Immunoblotting with an anti-p20 antibody on granule neuronal extracts after 8 h of K+ deprivation (lane 2) in the presence of 50 mM z-DEVD.fmk (lane 3) or 50 mM z-VAD.fmk (lane 4). A representative experiment out of four is shown in a and b

nuclei have an intense cytoplasmic p20 staining (Figure 4h). These observations show that CPP32 is active in the nucleus as soon as neurons start to condense, the time when RB C-terminus disappears. At the ®nal condensation step, when only a faint amount of RB remains, CPP32 is still very active. These data strongly suggest that RB cleavage co-localizes with the activation of CPP32 at the cellular level during the dierent steps of nuclei condensation. Overexpression of a caspase-cleavage resistant mutant of RB in neurons prevents DNA fragmentation To ascertain that apoptosis activation depends on RB cleavage at its 883-DEADG-887 site, we overexpressed in neurons hemagglutinin (HA)-tagged vectors encoding a wild type (RBWT) or a non cleavable (RBD/A) form of RB (gifts from Drs R JaÈnicke and A Porter) and compared their neuroprotective eects against K+induced apoptosis. The second aspartic acid residue (asp 886) of the consensus `DEADG' site (RBWT) is substituted to an alanine in RBD/A. This mutant was completely resistant to cleavage in nuclear extracts of apoptotic D98 cells (Janicke et al., 1996), whereas overproduced RBWT could still be cleaved by active caspases (Tan et al., 1997). A typical experiment is shown Figure 5a. Twentyfour hours following transfection, neurons were K+deprived (LK) or not (HK) for 14 h, then ®xed and immunostained for the expression of HA to identify transfected cells by immunocytochemistry. After Hoechst labeling, all HA-positive neurons were blindly counted and scored as apoptotic or not. In LK, 39.55% of HA-positive RBWT transfected cells were apoptotic (Figure 5b), a percentage similar to that of the HA-negative population of neurons counted in the same ®eld (Figure 5c). In contrast, only 4.46% of nuclei from HA-positive RBD/A neurons were found Oncogene

Figure 4 Cellular localization of RB and the active p20 subunit of CPP32 during the dierent phases of apoptosis. The three types of nuclei encountered after Hoechst labeling of HK or LKtreated neurons are summarized on the upper panel. After 10 h in HK (a,b,e,f) or LK (c,d,g,h) medium, granular cells were ®xed and stained with the Hoechst dye 33342. Neurons were then immunolabeled either with the rabbit polyclonal C-15 antibody against full-length RB (b,d) or with the rabbit polyclonal anti-p20 subunit of CPP32 (f,h). Representative ¯uorescence micrographs are shown where Hoechst-labeled nuclei are visualized in blue (a,c,e,g) and C-15 positive (b,d) or p20 positive (f,h) neurons in red (Cy3). `Type 1' nuclei are prominent in HK neuroprotective conditions and correspond to healthy neurons displaying a non condensed chromatin dispersed over the entire nucleus (a,e). `Type 2' and `3' nuclei are scored as apoptotic: in `type 2', chromatin margination at the nuclear periphery corresponds to the early phase of chromatin condensation and `type 3' represents highly refractile and condensed nuclei that can be fragmented into two or more blebs. Compare the pattern of RB and CPP32 cellular localization in LK-treated neurons (d,h) in the dierent types of nuclei (c,g) indicated by numbered arrows (1, 2 and 3, corresponding to the dierent `types' of nuclei). Scale bars: 10 mm

apoptotic. These data show that the caspase cleavage resistant form of RB exerts potent neuroprotective eects and suggest that carboxyterminal cleavage of


2175

Figure 5 Transfected caspase-resistant mutant of RB inhibits K+-induced neuronal apoptosis. Neurons were transfected either with an HA-tagged expression vector encoding RB wild type containing the consensus caspase cleavage site (RBWT: DEAD/G, left panel) or RB mutated on this site (RBD/A: DEAA/G) to give a C-terminus cleavage resistant RB protein (right panel). After 24 h, cells were K+-deprived (LK) or not (HK) for 14 h. On ¯uorescence micrographs (a), transfected cells immunostain red and Hoechst-labeled nuclei are visualized in blue. Arrows indicate nuclei of HA-positive neuron. In HK conditions, all neurons transfected with RBWT or RBD/A were scored as non apoptotic, excluding toxic side eects of the transfection procedure or possible proapoptotic eect generated by the overexpressed proteins. Scale bars: 10 mm. Apoptotic nuclei in LK conditions were scored in the HA-positive (b) or in the HA-negative (c) transfected population of neurons. Data from seven independent experiments performed in triplicate are expressed in percentage of apoptotic nuclei obtained in LK. Unpaired t-test with Welch correction analysis reveals a statistical signi®cant decrease of apoptotic nuclei in HA-positive RBD/A compared to RBWT transfected cells (b, P=0.003). No dierence is seen within the HA-negative population of neurons (c, unpaired t-test, P=0.725)

RB is an early permissive step of the apoptotic process. Discussion Several studies in cancerous cell lines have shown that RB is cleaved by caspases during apoptosis, a cleavage that occurs either at the C-terminus or further within the RB protein (An and Dou, 1996; Fattman et al., 1997; Janicke et al., 1996; Tan et al., 1997). Here we show that in post-mitotic neurons, a speci®c RB Cterminus cleavage is observed and correlates with DNA condensation and fragmentation, suggesting its implication in the apoptotic signaling pathway. Fragments that would result from an internal RB cleavage (p48 and/or p68) are not detected, supporting

the idea that RB cleavage occurs at ®rst on the Cterminus consensus 883-DEADG-887 site. However, the predicted truncated RB form (DRB, 100 KDa) never accumulated, precluding further attempts of characterization. A speci®c decrease in the total amount of RB was also observed with the G3-245 antibody, so it is likely that DRB is generated by a ®rst cleavage on RB C-terminus and that the RB protein is then further degraded. Overexpression of a cleavage resistant mutant on this C-terminus site prevents apoptosis entry, a ®nding that demonstrates the functional importance of the presence of the fulllength RB protein to promote neuroprotection in a model of primary neurons. These ®ndings are in line with those obtained in Rb knock out mice (Rb7/7) which die at mid-gestation probably because of massive apoptotic cell death in the central nervous Oncogene


2176

system (Clarke et al., 1992; Jacks et al., 1992; Lee et al., 1992). Caspase specificity of RB cleavage It is documented that K+-induced apoptosis of CGN is inhibited by broad spectrum caspases inhibitors, such as Asp(D)-coupled derivatives (Dichlorobenzoyl or Fluoromethylketone) (D'Mello et al., 1998; Lynch et al., 1997; Nath et al., 1996; Tanabe et al., 1998). A panel of synthetic tetrapeptide inhibitors is now available and can be used to distinguish between activation of the dierent ICE and CED3 subfamilies of caspases (see classi®cation of caspases in Alnemri et al., 1996). However, the use of these more speci®c inhibitors led to some discrepancies between the dierent subtypes of caspases involved in CGN apoptosis, probably also due to dierent culture conditions. Inhibitors of the caspase-1 (ICE) family (YVAD) are generally found to have no or little neuroprotection potential (Eldadah et al., 1997; Moran et al., 1999), an observation that is in accordance with our result that the ¯uorogenic substrate Ac-YVAD-amc is not cleaved in extracts from LK-treated neurons (Figure 4; Marks et al., 1998). This contrasts with an early study where YVAD.cmk was found to protect CGN death (Schulz et al., 1996). However, the most ecient dose used was very high and it is likely that other caspases were also blocked. Thus, it seems unlikely that the ICE, YVADsensitive caspases family participates in RB cleavage in CGN apoptosis. There is a consensus that z-VAD.fmk inhibitor protects from K+-induced apoptosis, suggesting the involvement of the CED3-subtype of proteases (Table 1; Armstrong et al., 1997). RB cleavage was blocked by the z-VAD.fmk inhibitor, as well as with z-DEVD.fmk (Figure 1d). The ¯uorogenic substrate Ac-DEVD-amc is cleaved with a very high eciency (Figure 2, Armstrong et al., 1997; Lynch et al., 1997; Marks et al., 1998), pointing to CPP32 (caspase 3) as a good candidate for caspases involved in RB cleavage during CGN apoptosis. CPP32 has been shown to be required for developmental apoptosis in neurons, whereas it is not obligatory in some other cell types (Kuida et al., 1996). We show that the active p20 subunit of CPP32 is progressively induced by K+ deprivation of CGN, thus correlating with the decline in RB levels. Immunocytochemistry further demonstrated that CPP32 is activated in the cytoplasm and nucleus of apoptotic neurons. p20 is expressed within nuclei that start to condense and surrounds very condensed or fragmented nuclei (see Figure 4). This ®ts with several reports demonstrating that caspase 3 is required for DNA condensation and fragmentation (Janicke et al., 1998; Woo et al., 1998). One striking observation is that disappearance of RB C-terminus coincides with activation of CPP32 at the cellular level. Taken together, these data show that RB C-terminus cleavage is simultaneous with activated CPP32 and it corroborates the involvement of CPP32 in this RB-dependent apoptotic pathway. However, the demonstration of which speci®c caspase(s) is(are) involved in RB C-terminus cleavage is not complete. The z-DEVD.fmk inhibitor prevents CGN apoptosis by almost 50% at 8 h of treatment and is even

Oncogene

less potent at the longer periods of times tested (see Table 1). The fact that apoptosis is reversed by 90% in neurons overexpressing the non cleavable mutant of RB, demonstrates that this site, if cleaved by CPP32, is likely to also be the substrate of another, as yet unidenti®ed, caspase, probably less sensitive to DEVD inhibition. It has been reported that preponderant active caspases in apoptotic cells were CPP32 and Mch2 (caspase-6) (Faleiro et al., 1997). Interestingly, the authors found that multiple species of these caspases could co-exist, diering by their cleavage site and molecular weight, or their isoelectric point. Thus they could have dierent substrate speci®city and/or localization, e.g. function in the apoptotic process (Faleiro et al., 1997). This supports the idea that dierent variants of caspase 3 are likely to be more or less sensitive to DEVD-type inhibitors. In our Western blot experiments, RB cleavage reversion by zDEVD.fmk was rather delayed than total, relative to that obtained with z-VAD.fmk. Moreover, in our experimental conditions, other caspases are induced, such as ICH-2 (caspase-4), but also nedd 2, ICH-1 (caspase-2) and ICE-LAP6, Mch6 (caspase-9), with dierent kinetics following K+ deprivation (unpublished results). These have all been shown to participate in apoptosis (Bergeron et al., 1998; Faleiro et al., 1997; Kidd, 1998; Li et al., 1997) and this point is currently under investigation. Functional consequences of RB cleavage in the apoptotic process The degradation of the RB protein can either be a side eect of caspases activation during apoptosis or play a major role in the release of this process. The use of a non cleavable mutant allowed us to demonstrate that if RB can not be cleaved on its C-terminus, LK-induced apoptosis of CGN cannot proceed. RB thus functions as a key substrate at the onset of apoptosis: it is rapidly cleaved o by caspases, then further proteolyzed, and RB degradation will probably liberate endogenous factors which will, in this context, favor apoptosis signalings. The alternative hypothesis that the shorter RB fragments generated by caspase cleavage are pro-apoptotic by themselves is unlikely. Indeed, it has been shown that overexpression of DRB or the short C-terminus RB portion do not enhance apoptosis, in cell lines where the DRB fragment accumulates following apoptosis (Janicke et al., 1996). Moreover, in our model, DRB has a short half-life as it is hardly detectable by Western blots, so its proapoptosis inducing potential is not probable. In this respect, one can think about several eector proteins that bind RB and could be involved in CGN apoptotic signalings. One candidate is the p53 regulator mdm2 (see review by Prives, 1998) that directly binds the C-terminus fragment of RB (Xiao et al., 1995). Mdm2 is released from RB following its C-terminus cleavage (Janicke et al., 1996). However, mdm2 seems to have anti-apoptotic functions (Chen et al., 1996; Kondo et al., 1995; Prives, 1998), but several reports show that mdm2 is cleaved during apoptosis by CPP32-like proteases (Chen et al., 1997; Erhardt et al., 1997). This makes it a potential target unveiled by RB degradation. Another candidate is the E2F family of transcription factors. They are critical positive regulators of cell


cycle progression and are regulated through their binding to the RB pocket (Dyson, 1998). When overexpressed (e.g. inappropriately activated), E2F-1 has been shown to induce apoptosis in various models (Field et al., 1996; Phillips et al., 1997). It is further known that binding of E2F to the RB protein protects it from degradation through the ubiquitin/proteasome pathway (Campanero and Flemington, 1997; Hateboer et al., 1996). This proteolytic signaling pathway could come into play since we have previously shown that a direct activator of RB, the cyclin D1, is induced in apoptotic CGN, due to a lack of its degradation by the ubiquitin/proteasome pathway (Boutillier et al., 1999). In the same manner, LK treatment could block E2F1 degradation by the ubiquitin/proteasome pathway, and the loss of the RB protein abrogates its repression. Subsequently, this would result in the activation of the E2F-1 pathway. The DRB can still bind E2F-1 (Janicke et al., 1996), but the rapid degradation of RB in our model is favorable to such a scenario. Finally, it has recently been shown that RB physically interacts with histone deacetylase HDAC1 through its LXCXE motif (Magnaghi et al., 1998). It is then conceivable that RB degradation in neurons will be associated with chromatin alterations that would otherwise not occur in a post-mitotic cell, following the release of HDAC. The involvement of such interactions would be of prime importance, as they will have an impact on chromatin conformation and might direct the accessibility of caspases and nucleases in order to proceed to DNA fragmentation. Materials and methods Neuronal primary cultures Cerebellar granule neurons from 7-day-old mice (fvb strain) were cultured as previously described (Schousboe et al., 1989) with minor modi®cations (Boutillier et al., 1999). Experiments were performed after 6 day in vitro (DIV). HK medium contains 30 mM KCl and LK medium, 5 mM KCl. Peptides and antibodies The various peptidic protease inhibitors z-DEVD.fmk (Benzyloxycarbonyl - Asp- Glu-Val- Asp- ¯uoromethylketone), z-VAD.fmk (Benzyloxycarbonyl-Val-Ala-Asp-¯uoromethylketone), and substrates Ac-YVAD-amc (N-Acetyl-Tyr-ValAla-Asp-7-Amino-4-methylcoumarin), Ac-DEVD-amc (NAcetyl-Asp-Glu-Val-Asp-7-Amino-4-methylcoumarin) were purchased from Calbiochem (La Jolla, CA, USA). Monoclonal antibody anti-Rb (G3-245) was obtained from Pharmingen (San Diego, CA, USA). Rabbit polyclonal antibodies against RB (C-15) or Hemagglutinin (HA) (Y11) were from Santa Cruz Biotechnology (Santa Cruz, CA, USA), and anti-p20 active CPP32 rabbit polyclonal antibody from R&D Systems (Minneapolis, MN, USA). Horseradish peroxidase (HRP) conjugated secondary antibodies, goat anti-rabbit IgG and sheep anti-mouse IgG, were purchased from Pierce (Rockford, IL, USA). Cy32-conjugated goat anti-rabbit or donkey anti-mouse IgGs were from Jackson ImmunoResearch (West Grove, PA, USA) and FITC coupled goat anti-rabbit IgG from Vector (Burlingame, CA, USA). Western blots Western blots were performed as described previously (Boutillier et al., 1999) with typically 50 ± 70 mg of total cell extract ran on 6.5% and 13% SDS-acrylamide gels for RB

and p20 detection, respectively. Speci®c bands were detected by ECL (Amersham).

2177

Morphological evaluations of apoptosis (Hoechst labeling) Cells were ®xed with 4% paraformaldehyde in PBS for 30 min at room temperature, incubated for 30 min with the Hoechst dye 33342 (1 mg/ml). Fixed cells were viewed on a Nikon Diaphot 300 microscope (Nikon, Melville, NY, USA) (430 nm). Stained nuclei were scored by blind analysis and categorized according to the condensation and staining characteristics of chromatin. Nuclei presenting condensed or fragmented chromatin were scored as representative of apoptotic neurons. In each experiment performed in triplicate, eight microscopic ®elds (magni®cation 406) were counted (about 2400 neurons per condition). Immunocytochemistry After ®xation and Hoechst staining as described above, cells were washed and blocked for 10 min in PBS containing Triton 0.1% and sheep serum 5%. Cells were subsequently incubated in primary antibodies diluted in the same buer. After several wash steps, immunostaining was detected with the appropriate ¯uorescent secondary antibodies. Cy3 and FITC were visualized with a Leitz microscope using rhodamine and ¯uoresceine ®lter sets. Photographs were taken with Kodak TMAX 320 ®lms. Transfections Neuronal gene transfer was performed as reported previously, using PEI (25K) as DNA carrier (Boussif et al., 1995). Neurons were grown on glass coverslips and transfected at four DIV for 30 min with 1 mg/ml of expression vector: pClneo-HA-RB-WT or pClneo-HA-RBD/A (Janicke et al., 1996). Plates were spun for 5 min at 1500 r.p.m. Twenty-four hours after transfection in HK, neurons were K+-deprived for 14 h. Apoptotic nuclei were visualized by Hoechst staining and transfected cells were revealed by immunocytochemistry using an anti-HA antibody (Y-11). Caspase activity measurements Caspase activities were determined by measuring the release of ¯uorescent cleavage products from the caspase tetrapeptide substrates. At the appropriate time of treatment, neurons were lysed in cold buer A (lgepal NP-4 0.5%, EDTA 0.5 mM, NaCl 150 mM, Tris pH 7.5 50 mM). One portion of the lysate was used for protein determination (Bradford method, Biorad). Fifty ml of lysate was added to a reaction mixture containing 50 mM of each substrate (Ac-YVAD-amc, Ac-DEVD-amc) in buer B (HEPES pH 7.5 20 mM, NaCl 50 mM, DTT 2.5 mM) and incubated for 1 h at 378C. The reaction was stopped by a sixfold dilution of ice-cold buer A. Fluorescence was monitored at excitation 360 nm, emission 465 nm, using a Perkin-Elmer HTS 7000 Bioassay reader (Foster City, CA, USA). A standard curve of ¯uorescence versus free amc was used to calculate the amount of substrate cleaved from ¯uorescence units.

Acknowledgments Recombinant material (pClneo-HA-RB-WT and pClneoHA-RB-D/A) was a generous gift from Drs AG Porter and RU JaÈnicke (Singapore, Republic of Singapore). This work was supported by ARC (no. 9821), `Ligue reÂgionale (68) pour la lutte contre le cancer' and `ReÂgion Alsace'. Oncogene


2178

References Alnemri ES, Livingston DJ, Nicholson DW, Salvesen G, Thornberry NA, Wong WW and Yuan J. (1996). Cell, 87: 171. An B and Dou QP. (1996). Cancer Res., 56, 438 ± 442. Armstrong RC, Aja TJ, Hoang KD, Gaur S, Bai X, Alnemri ES, Litwack G, Karanewsky DS, Fritz LC and Tomaselli KJ. (1997). J. Neurosci., 17, 553 ± 562. Bergeron L, Perez GI, Macdonald G, Shi L, Sun Y, Juriscova A, Varmuza S, Latham KE, Flaws JA, Salter JC, Hara H, Moskowitz MA, Li E, Greenberg A, Tilly JL and Yuan J. (1998) Genes Dev., 12, 1304 ± 1314. Boussif O, Lezoualc'h F, Zanta MA, Mergny MD, Scherman D, Demeneix B and Behr JP. (1995). Proc. Natl. Acad. Sci. USA, 92, 7297 ± 7301. Boutillier AL, Kienlen-Campart P and Loeer JP. (1999). Eur. J. Neurosci., 11, 441 ± 448. Campanero MR and Flemington EK. (1997). Proc. Natl. Acad. Sci. USA, 94, 2221 ± 2226. Chen J, Wu X, Lin J and Levine AJ. (1996). Mol. Cell. Biol., 16, 2445 ± 2452. Chen L, Marechal V, Moreau J, Levine AJ and Chen J. (1997). J. Biol. Chem., 272, 22966 ± 22973. Clarke AR, Maandag ER, Van Room M, Van der Lugt NM, Van der Valk M, Hooper ML, Berns A and Te Riele H. (1992). Nature, 359 328 ± 330. D'Mello S, Galli C, Ciotti T and Calissano P. (1993). Proc. Natl. Acad. Sci. USA, 90, 10989 ± 10993. D'Mello SR, Aglieco F, Roberts MR, Borodezt K and Haycock JW. (1998). J. Neurochem., 70, 1809 ± 1818. Dou QP and An B. (1998). Front. Biosci., 3, 419 ± 430. Dyson N. (1998). Genes Dev., 12, 2245 ± 2262. Eldadah BA, Yakovlev AG and Faden AI. (1997). J. Neurosci., 17, 6105 ± 6113. Erhardt P, Tomaselli KJ and Cooper GM. (1997). J. Biol. Chem., 272, 15049 ± 15052. Faleiro L, Kobayashi R, Fearnhead H and Lazebnik Y. (1997). EMBO J., 16, 2271 ± 2281. Fattman CL, An B and Dou QP. (1997). J. Cell. Biochem., 67, 399 ± 408. Field SJ, Tsai FY, Kuo F, Zubiaga AM, Kaelin WJ, Livingston DM, Orkin SH and Greenberg ME. (1996). Cell, 85, 549 ± 561. Friedlander RM and Yuan J. (1998). Cell Death Dier., 5, 823 ± 831. Gallo V, Kinsbury A, Balazs R and Jorgensen OS. (1987). J. Neurosci., 7, 2203 ± 2213. Haas KD, Kogan SC, Levi D, Dazin P, T'Ang A, Fung YK and Israel MA. (1995). EMBO J., 14, 461 ± 472. Hateboer G, Kerkhoven RM, Shvarts A, Bernards R and Beijersbergen RL. (1996). Genes Dev., 10, 2960 ± 1970. Heintz N. (1993). Trends Biochem. Sci., 18, 157 ± 159. Helin K, Holm K, Niebuhr A, Eiberg H, Tommerup N, Houggard S, Poulsen HS, Spang TM and Norgaard P. (1997). Proc. Natl. Acad. Sci. USA, 94, 6933 ± 6938. Henkart PA. (1996). Immunity 4, 195 ± 201. Jacks T, Fazeli A, Schmitt EM, Bronson RT, Goodell MA and Weinberg RA. (1992). Nature, 359, 295 ± 300. Janicke RU, Sprengart ML, Wati MR and Porter AG. (1998). J. Biol. Chem., 273, 9357 ± 9360. Janicke RU, Walker PA, Lin XY and Porter AG. (1996). EMBO J., 15, 6969 ± 6978.

Oncogene

Kidd VJ. (1998). Annu. Rev. Physiol., 60, 533 ± 573. Kienlen-Campard P, Crochemore C, ReneÂ F, Monnier D, Koch B and Loeer JP. (1997). DNA & Cell Biology, 16, 323 ± 333. Kondo S, Barnett GH, Hara H, Morimura T and Takeuchi J. (1995). Oncogene, 10, 2001 ± 2006. Kuida K, Zheng TS, Na S, Kuan C, Yang D, Karasuyama H, Rakic P and Flavell RA. (1996). Nature, 384, 368 ± 372. Lee EY, Chang CY, Hu N, Wang YC, Lai CC, Herrup K, Lee WH and Bradley A. (1992). Nature, 359, 288 ± 294. Li H, Bergeron L, Cryns V, Pasternack MS, Zhu H, Shi L, Greenberg A and Yuan J. (1997). J. Biol. Chem., 272, 21010 ± 21017. Ludlow JW, DeCaprio JA, Huang CM, Lee WH, Paucha E and Livingston DM. (1989). Cell, 56, 57 ± 65. Lynch T, Vasilakos JP, Raser K, Keane KM and Shivers BD. (1997). Mol. Psychiatry, 2, 227 ± 238. Magnaghi JL, Groisman R, Naguibneva I, Robin P, Lorain S, Le VJ, Troalen F, Trouche D and Harel BA. (1998). Nature, 391, 601 ± 605. Marks N, Berg MJ, Guidotti A and Saito M. (1998). J. Neurosci. Res., 52, 334 ± 341. Moran J, Itoh T, Reddy UR, Chen M, Alnemri ES and Pleasure D. (1999). Neurochem., 73, 568 ± 577. Nath R, Raser KJ, McGinnis K, Nadimpalli R, Staord D and Wang KK. (1996). Neuroreport 8, 249 ± 255. Nicholson DW, Ali A, Thornberry NA, Vaillancourt JP, Ding CK, Gallant M, Gareau Y, Grin PR, Labelle M, Lazebnik YA, Munday NA, Raju SM, Smulson HE, Yamin TT, Yu VL and Miller DK. (1995). Nature, 376, 37 ± 43. Oppenheim RW. (1991). Annu. Rev. Neurosci., 14, 453 ± 501. Phillips AC, Bates S, Ryan KM, Helin K and Vousden KH. (1997). Genes Dev., 11, 1853 ± 1863. Prives C. (1998). Cell, 95, 5 ± 8. Schousboe A, Meier E, Drejer J and Hertz L. (1989). Preparation of cultures of mouse (rat) granule cells. In: Shahar A, De Vellis J, Vernadahis A and Haber B (eds). A dissection and tissue culture manual for the nervous system. New York: Alan R Liss, 203 ± 206. Schulz JB, Weller M and Klockgether T. (1996). J. Neurosci., 16, 4696 ± 4706. Slack RS and Miller FD. (1996). Dev. Genet., 18, 81 ± 91. Sohm F, Gaiddon C, Antoine M, Boutillier AL and Loeer JP. (1999). Oncogene, 18, 2762 ± 2769. Tan X, Martin SJ, Green DR and Wang J. (1997). J. Biol. Chem., 272, 9613 ± 9616. Tan X and Wang JY. (1998). Trends Cell Biol., 8, 116 ± 120. Tanabe H, Eguchi Y, Shimizu S, Martinou JC and Tsujimoto Y. (1998). Eur. J. Neurosci., 10, 1403 ± 1411. Woo M, Hakem R, Soengas MS, Duncan GS, Shahinian A, Kagi D, Haken A, McCurrach M, Khoo W et al. (1998). Genes Dev., 12, 806 ± 819. Xiao ZX, Chen J, Levine AJ, Modjtahedi N, Xing J, Sellers WR and Livingston DM. (1995). Nature, 375, 694 ± 698. Yuan J. (1995). Curr. Opin. Cell. Biol., 7, 211 ± 214. Yuan J, Shaham S, Ledoux S, Ellis HM and Horvitz HR. (1993). Cell, 75, 641 ± 652.