Role of mTOR, Bad, and Survivin in RasGAP Fragment N-Mediated Cell Protection

Partial cleavage of p120 RasGAP by caspase-3 in stressed cells generates an N-terminal fragment, called fragment N, which activates an anti-apoptotic Akt-dependent survival response. Akt regulates several effectors but which of these mediate fragment N-dependent cell protection has not been defined yet. Here we have investigated the role of mTORC1, Bad, and survivin in the capacity of fragment N to protect cells from apoptosis. Neither rapamycin, an inhibitor of mTORC1, nor silencing of raptor, a subunit of the mTORC1 complex, altered the ability of fragment N from inhibiting cisplatin- and Fas ligand-induced death. Cells lacking Bad, despite displaying a stronger resistance to apoptosis, were still protected by fragment N against cisplatin-induced death. Fragment N was also able to protect cells from Fas ligand-induced death in conditions where Bad plays no role in apoptosis regulation. Fragment N expression in cells did neither modulate survivin mRNA nor its protein expression. Moreover, the expression of cytoplasmic survivin, known to exert anti-apoptotic actions in cells, still occurred in UV-B-irradiated epidermis of mouse expressing a caspase-3-resistant RasGAP mutant that cannot produce fragment N. Additionally, survivin function in cell cycle progression was not affected by fragment N. These results indicate that, taken individually, mTOR, Bad, or Survivin are not required for fragment N to protect cells from cell death. We conclude that downstream targets of Akt other than mTORC1, Bad, or survivin mediate fragment N-induced protection or that several Akt effectors can compensate for each other to induce the pro-survival fragment N-dependent response.


Introduction
Activation of executioner caspases was once believed to represent a point of no return in the path to death. However it is now well established that while executioner caspases are indispensable for apoptosis, there are situations when their activation does not lead to death. For example, healthy dividing cells can weakly activate caspase-3 in response to mild stresses [1]. Caspase-3 also participates, in an apoptosis-independent manner, in T and B cell homeostasis [2,3], in microglia activation [4], and in muscle [5], monocyte [6], bone marrow stromal stem cell [7], and erythroid cell differentiation [8].
Low caspase-3 activation in stressed cells induces the partial cleavage of RasGAP into an amino-terminal fragment, called fragment N, that prevents amplification of caspase-3 activation and death in an Akt-dependent manner [9]. Knock-in mice that only express a caspase-3-resistant RasGAP mutant, and hence that cannot generate fragment N in response to stress, are unable to stimulate Akt efficiently and are more sensitive to damage induced by various pathophysiological insults [10]. Fragment N generation can therefore explain why cells having mildly activated caspase-3 do not necessarily die. On the other hand, when caspase-3 activity is strongly stimulated in cells, fragment N is further processed into smaller fragments, called N1 and N2, that no longer have the capacity to activate Akt [11]. The extent of caspase-3 activity in a cell can therefore be sensed by RasGAP to either mount an efficient Akt-dependent protection when the stress is not too strong [1,12] or to abrogate this protective signal in cells faced with strong insults or apoptotic stimuli [11].
Phosphorylation of downstream effectors by Akt leads to diverse cellular responses affecting metabolism, protein synthesis, proliferation, angiogenesis and inhibition of apoptosis [13]. A series of Akt effectors have been shown to favor survival once phosphorylated by Akt [13,14]. These can be either pro-apoptotic proteins that become inactivated once phosphorylated by Akt or antiapoptotic proteins the expression of which is induced by Akt. In the first group lie Bad, Bax, Ask1, and pro-apoptotic transcription factors such as YAP and some Forkhead family members. Mdm2, a p53 inhibitor, members of the inhibitor of apoptosis family (c-IAP1/2, survivin), members of the anti-apoptotic Bcl-2 subfamily (A1, Bcl-X L ) and the NF-kB transcription factor, which is generally inducing survival responses, are found in the second group [15][16][17][18].
Although Akt can lead to activation of the NF-kB transcription factor, fragment N-mediated Akt activation does not result in NF-kB stimulation [19]. In fact, fragment N is able to block NF-kB activation in response to various stimuli including exposure to inflammatory cytokines [20,21]. Hence fragment N does not rely on NF-kB activation to protect cells. Actually, a sustained NF-kB activation could be detrimental at least in certain cell types, such as pancreatic beta cells [22]. In this context, the capacity of fragment N to block NF-kB activation would be beneficial and it has indeed been shown that NF-kB inhibition by fragment N contributes to its anti-apoptotic activity in beta cells [21,23]. These observations rule out activation of NF-kB as an Akt-dependent mechanism used by fragment N to protect cells. Which of the other Akt effectors are required for fragment N to protect cells is not known.
In the present work we investigated whether mTORC1, Bad and survivin play a role in fragment N-mediated apoptosis inhibition. Mammalian TOR (mTOR) is a protein kinase that exists in two different complexes. The mTORC1 complex contains mTOR and Raptor and is inhibited by rapamycin. Akt indirectly activates mTORC1 by preventing the TSC1/TSC2 GTPase-activating proteins from inhibiting Rheb, the small GTPbinding protein that stimulates mTORC1. The mTORC2 complex, which contains mTOR and Rictor, is not sensitive to rapamycin, at least in short-term experiments [24], and positively regulates the activation of Akt by phosphorylating it rather than being an Akt effector itself [25,26]. The involvement of mTORC1 in the control of cell survival responses [27,28] makes it a possible fragment N effector candidate.
Another putative player in fragment N-induced protective pathway is Bad, a pro-apoptotic member of the Bcl-2 family [29,30]. When phosphorylated by Akt, Bad binds to 14-3-3 proteins and this prevents its capacity to interact with antiapoptotic Bcl-2 family members. Consequently, phosphorylated Bad loses its pro-apoptotic activity [31].
Survivin (also called BIRC5), an atypical member of the inhibitor of apoptosis (IAP) family of proteins, is also a component of the chromosomal passenger protein complex that ensures proper segregation of sister chromatids and cytokinesis [32,33], and that mediates many other functions during mitosis [34]. The absence of survivin profoundly alters mitosis and cytokinesis and this eventually leads to cell death [32,35]. The exact contribution of survivin in anti-apoptotic responses is controversial and still not well defined. Survivin may indirectly affects caspases either by binding and stabilizing XIAP [36], blocking Smac-dependent XIAP inhibition [19,24,30,33] or inhibiting caspase-9 by cooperating with HBXIP [23]. Depending on the cell types and the experimental conditions, activation of Akt can lead to increased survivin expression [37,38].

Antibodies
The anti-p70 S6 kinase antibody, the anti-raptor antibody, the anti-phospho Bad antibody sampler kit and the rabbit antisurvivin 71G4 monoclonal antibody were purchased from Cell Signaling Technology (catalogue nu9205, 2280, 9105 and 2808 respectively). The monoclonal anti-a tubulin antibody was purchased from Serotec, Oxford, UK (catalogue nuMCA77G). Secondary antibodies used for Western blotting were Alexa Fluor 680-conjugated anti-rabbit antibody (Molecular Probes, Eugene, Figure 1. mTORC1 is not involved in fragment N-induced cell survival. HeLa cells were transfected with a GFP-expressing plasmid together with empty pcDNA3 or with pcDNA3 encoding the D157A caspase-resistant form of fragment N. Twenty-four hours later, the cells were incubated 90 minutes with 20 ng/ml rapamycin or with vehicle alone (DMSO). A. The cells were lysed after an additional 24-hour period and the extent of S6K phosphorylation was assessed by Western blot. B-C. Following the rapamycin pre-incubation period, the cells were treated or not with 10 mM cisplatin (panel B) or with 5 ng/ml Fas ligand (panel C) for 24 hours and apoptosis was scored (mean 695% CI of 5 and 3 independent experiments, respectively, performed in monoplicate). D. HeLa cells were transfected or not with siRNAs directed at Raptor and one day later transfected with empty pcDNA3 or with pcDNA3 encoding fragment N. After an additional 24-, 48-, or 72-hour period, the cells were lysed to assess the levels of Raptor expression. E. Alternatively, one day following the pcDNA3 plasmid transfection, the cells were treated with cisplatin (30 mM) for 24 more hours and apoptosis was then scored (mean 695% CI of 3 independent experiments performed in monoplicate). Asterisks denote statistically significant differences [one-way ANOVAs followed by pair-wise Dunn (Bonferroni) post hoc t tests]. doi:10.1371/journal.pone.0068123.g001 OR; catalogue nuA21109) or IRDye 800-conjugated anti-mouse antibody (Rockland, Gilbertsville, PA; catalogue nu610-132-121). For immunohistochemistry the secondary antibody was Cy3conjugated AffiniPure Fab fragment goat anti-rabbit IgG (catalogue nu711-165-152) and was purchased from Jackson Immu-noResearch Laboratories.

Lentivirus
Recombinant lentiviruses were produced as described [44]. The minimal amounts of viruses inducing expression of the protein of interest in more than 95% of the cells (as assessed by immunofluorescence) were used.

RNAi
HeLa cells were plated in 6-well plates, with a starting density of 2?10 5 cells per well. After an overnight incubation, the cells were transfected with 200 pmoles of siRNA for 7 hours using the calcium phosphate co-precipitation method [45]. siRNAs were synthesized by Microsynth (Balgach, Switzerland). The Raptordirected siRNA sequence was 59-GGACAACGGCCACAAGUA-39 (nucleotides 2338-2355 found in the coding sequence of human Raptor transcript 1 mRNA; NCBI reference sequence: NM_020761.2). Cells were lysed 24, 48, and 72 hours after transfection for Western blotting analysis.

RNA Extraction and Reverse Transcription
RNA was extracted by lysing cells with 500 ml TRI buffer (1.7 M guanidium thiocyanate, 0.1 M sodium citrate, 0.25% N-laurylsarcosyl sodium, 0.05 M b-mercaptoethanol, 0.1 M sodium acetate), followed by the addition of 200 ml of chloroform. The tubes were then vortexed, kept at room temperature for 5 minutes, and spun at maximal speed in an Eppendorf centrifuge for 15 minutes. After the transfer of the aqueous phase in a new tube, 500 ml of isopropanol was added and the solution was mixed by inversion approximately five times. The samples were then incubated o/n at 220uC. After spinning 20 minutes at maximum speed in an Eppendorf centrifuge, the upper phase (isopropanol) was aspirated; the pellet was washed twice with 800 ml 70% ethanol and dried 5-10 minutes at 50uC. The pellet was finally resuspended in 50 ml of water and RNA was quantitated at 260 nm. Half a microgram of RNA was mixed with 500 ng of random hexamers (Microsynth, Balgach, Switzerland), and water was added to reach a final volume of 11 ml. The samples were incubated 3 minutes at 70uC and then kept on ice. A mix (14 ml) containing 5 ml of 10 nM dinucleotide triphosphates (Promega catalogue nuU120D-123D), 0.5 ml of RNasin (Promega catalogue nuN211A), 5 ml of buffer 56, 2 ml dithiothreitol, 0.5 ml Superscript reverse transcriptase, and 1 ml water (the 56 buffer, the dithiothreitol and the transcriptase are coming from the Superscript II reverse Transcriptase Kit; Invitrogen catalog nu18064-014) was added to the RNA-hexamer mix. After incubation at 39uC for 1 hour followed by 15 minutes at 70uC, the cDNA was diluted 1:3 in water.

Real Time PCR
Quantitative PCR assays were carried out on a real-time PCR detection system (iQ5; Bio-Rad) using iQ SYBR Green Supermix (Bio-Rad catalog nu170-8862) in which 500 nM primers and 1 ml of the cDNA template produced as described above were added (final volume of the reaction: 20 ml). The annealing temperature was 59uC. Melting curve analyses were performed on all PCR to rule out nonspecific amplification. Reactions were carried out in triplicate. The primers used were 59-GCG GAG GCT GGC TTC A-39 (forward primer) and 59-AGA AAA AAC ACT GGG CCA AAT C-39 (reverse primer).   The population of cells to analyze was gated according to their size (FSC forward scatter) and granulosity (SCC side scatter) to exclude dead cells and clumps. Within this cell population cells were gated according to the peak of intensity versus the integrated signals to exclude debris and doublets. Ten thousand cells were analyzed per condition.

Western Blot Analysis
Cells were lysed in monoQ-c buffer [12] and protein quantification was performed by the Bradford technique. Equal amounts of protein were migrated in a polyacrylamide gel and transferred onto a Trans-Blot nitrocellulose membrane (Bio-Rad catalogue nu10484060). Membranes were blocked with 5% nonfat milk and incubated over-night at 4uC with specific primary antibodies. Blots were washed with TBS/0.1% Tween 20, incubated with specific secondary antibodies and visualized with the Odyssey infrared imaging system (LICOR Biosciences, Bad Homburg, Germany). Quantification was performed using the Odyssey infrared imaging software.

Apoptosis Measurements
Apoptosis in Hela cells was determined by scoring the number of transfected cells (i.e. cells expressing GFP) displaying pycnotic or fragmented nucleus [39]. Apoptosis in infected MEFs was similarly assessed but in all cells.

Immunocytochemistry
HeLa cells were seeded in coverslips. Twenty-four hours later, coverslips were transferred to a clean dish and the immunocytochemistry was performed. Cells were fixed with 2% paraformaldehyde diluted in 16 PBS for 15 minutes at room temperature. After washing thrice with PBS, cells were permeabilized with 0.2% Triton X-100 diluted in 16 PBS for 10 minutes at room temperature. After washing with PBS, unspecific binding sites were blocked using DMEM culture medium complemented with 10% newborn calf serum. After 20 minutes at room temperature, the primary anti-HA antibody (1:100 dilution) was added and incubated for 1 hour at room temperature in a dark and humid chamber. The secondary antibody was added after three washes in PBS for 1 hour at room temperature in a dark and humid chamber. After extensive washes in PBS, nuclei were stained with 10 mM Hoechst 33342 (Molecular Probes; catalogue nuH1399) for 10 minutes before mounting the slides in 0.1 g/ml Mowiol, 0.22% (v/v) glycerol, Tris 0.1 M pH 8.5, 0.1% diazobicyclo-octane. Mowiol was from Calbiochem (catalogue nu475904) and diazobicyclo-octane was from Fluka (catalogue nu33480).

Data Presentation and Statistics
Results are expressed as the mean 695% confidence intervals (CI). The statistical tests used were one way ANOVAs unless otherwise stated. Normality of the data was verified with the Shapiro-Wilk test.

Results and Discussion
To prevent cleavage of fragment N into the smaller N1 and N2 fragments, which can potentially generate confounding results, the experiments described below were performed with a form of fragment N that has its caspase-3 cleavage site destroyed [1,12].

mTORC1 and Fragment N-induced Protection
Rapamycin, a macrolide antibiotic, has been widely used as a selective inhibitor of the mTORC1 complex and reported as an inducer of autophagy [47]. Depending on the cellular context and the concentrations of the drug used, rapamycin can either induce or inhibit apoptosis [48]. Rapamycin displays cytotoxic effects when used in micro-molar concentrations [49][50][51]. These studies highlight the impact of rapamycin dosage in cell survival outcome.
In our experiments, we used 20 nM of rapamycin that did not induce, as expected, apoptosis ( Figures 1B, 1C) but that was sufficient to fully block serum-induced phosphorylation of S6 kinase, an mTORC1 substrate ( Figure 1A, first two lanes). This also indicates that mTORC1 is the sole kinase mediating S6K phosphorylation in serum-cultured HeLa cells. In starved conditions ( Figure 1A, third lane), mTORC1 was no longer activated as indicated by the absence of S6K phosphorylation. As expected from its ability to stimulate Akt, fragment N activated mTORC1dependent S6K phosphorylation ( Figure 1A, last two lanes). However, the ability of fragment N to protect cells from cisplatinor Fas ligand-induced apoptosis was unaffected by rapamycin ( Figure 1B), suggesting that mTORC1 activation does not modulate fragment N-mediated protection. To further substantiate this point, we aimed to disrupt mTORC1 signaling by silencing Raptor, a protein of the mTORC1 complex [27]. Figure 1D shows that the strongest reduction in Raptor protein expression levels was achieved 72 hours after the siRNA transfection. We therefore assessed the ability of fragment N to protect cells against Fas ligand in which Raptor was silenced for 3 days. Figure 1E shows that Raptor silencing did not influence the ability of fragment N to inhibit Fas ligand-induced apoptosis. In conclusion, inhibition of the mTORC1 complex by rapamycin or via silencing of one of its components (Raptor) did not compromise the protective capacity of fragment N. Hence, mTORC1 does not appear to be an essential element targeted by fragment N to mediate its anti-apoptotic activity in cells.

Phosphorylation of Bad does not Play a Major Role in Fragment N-mediated Protection
Bad is a pro-apoptotic protein that is blocked from triggering apoptosis by Akt-mediated phosphorylation on serine 136 [31]. Bad can be phosphorylated by other kinases, such as PKA, on serine 112 [52]. As expected from its ability to stimulate Akt, fragment N expressed in HeLa cells led to an increase in Bad phosphorylation on serine 136 but not on serine 112 (Figure 2A). Fragments N1 and N2, the fragment N cleavage products generated by high caspase-3 activity that are unable to stimulate Akt [11], did not induce Bad phosphorylation (Figure 2A).
The potential contribution of Bad in fragment N-mediated cell protection was assessed in MEFs derived from wild-type and Bad knock-out mice. Figure 2B displays the possible outcome of experiments using these cells expressing or not fragment N and stimulated with increasing concentrations of cisplatin. As Bad mediates part of the apoptotic response to cisplatin [53], reduced cisplatin-induced death is expected in Bad knock-out (KO) MEFs (orange lines). If Akt-induced Bad phosphorylation does not play any role in the protection mediated by fragment N, the latter should decrease the extent of apoptosis to similar extent in wildtype and Bad KO MEFs ( Figure 2B, left graph). In other words, the maximal percent decrease in apoptosis induced by fragment N should be similar in both MEF types as shown in the inset above the left graph in Figure 2B. In contrast, if Bad phosphorylation is the sole mechanism by which fragment N protects cells, the sensitivity of Bad KO MEFs to cisplatin should be unaffected by the presence of fragment N ( Figure 2B, right graph). In this case, for Bad KO cells, the difference between the apoptosis curves in the presence or in the absence of fragment N should be minimal (orange lines in the inset above the right graph in Figure 2B). The middle graphs in Figure 2B display a situation where Bad phosphorylation is partially contributing to fragment N-mediated cell protection. The actual experiment ( Figure 2C) generated a pattern that corresponded to the scenario presented on the left of Figure 2B, suggesting that Bad plays no or only minimal role in fragment N-induced cell protection.
We reasoned that if Bad is not important for the capacity of fragment N to inhibit apoptosis, fragment N should be able to block cell death when the extrinsic pathway is activated, after Fas ligand stimulation for example, because this mode of death is not regulated by Bad. A potential caveat with this approach is that there is a possibility of a connection of Fas-induced death signaling to the intrinsic pathway via the cleavage of Bid [54]. This connection with the intrinsic pathway of apoptosis can however be blocked by over-expression of Bcl-2 [55]. Figure 2D shows that the stimulation of the mitochondrial intrinsic pathway activated by staurosporine in HeLa cells was fully prevented by Bcl-2 overexpression (compare the 9 th and 11 th bars). HeLa cells are type 2 cells as inhibition of the intrinsic pathway reduces apoptosis induced by Fas ligand (Figure 2D, compare the 5 th and 7 th bars); see also [56]. Importantly, the remaining apoptotic response can be fully blocked by fragment N (compare the 7 th and 8 th bars). This indicates therefore that fragment N can protect cells against a cell death stimulus in which pro-apoptotic Bcl-2 family members such as Bad play no role. Altogether, the experiments shown in Figure 2 support the notion that Bad is not a critical target of fragment N for its ability to protect cells.

Fragment N does not Regulate Survivin
There is evidence that Akt can induce survivin expression [38,57,58]. As survivin may display anti-apoptotic activity in some conditions [36][37][38]59,60], it could be one of the main targets of fragment N that mediates its survival effects. We therefore assessed whether fragment N can regulate survivin by determining if it modulated its expression in vitro and in vivo and whether it affected survivin capacity to regulate cell division.
The effect of fragment N on survivin transcription was assessed by luciferase assay in which either the minimal or the entire sequence of the mouse survivin promoter was used. Fragment N expression in cells had no significant effect on either promoter activities ( Figure 3A). As a positive control for this experiment, cells were transfected with an E2F1-expressing vector that is known to mediate survivin transcription [61] and this resulted, as expected, in an increase in survivin promoter activity ( Figure 3A). To further confirm these results, real time PCR was performed to measure survivin mRNA in cells expressing or not fragment N. Figure 3B shows that fragment N did not induce an increase in the mRNA coding for survivin. Furthermore, survivin protein levels were not affected by fragment N ( Figure 3C).
We have recently demonstrated that UV-B exposure of the epidermis leads to Akt phosphorylation in a caspase-3 and RasGAP cleavage-dependent manner [10]. Moreover, it has also been shown that survivin expression and relocalization to the cytoplasm, where it is supposed to induce its anti-apoptotic response [62], is induced in mouse skin in response to UV-B light [46]. Given that fragment N induces Akt in the epidermis of UV-B-irradiated mice [10] and that cytoplasmic survivin expression is augmented in this same tissue [46], we tested whether survivin expression in the skin is affected in knock-in (KI) mice that cannot generate fragment N because of a mutation in the first caspase-3 cleavage site of RasGAP. Control and KI mice were exposed to UV-B light 24 hours prior to biopsy and survivin levels were monitored in the skin by immunofluorescence. The percentage of keratinocytes expressing cytoplasmic survivin was increased by UV-B in a dose-dependent manner. This increase was similar in wild-type and KI mice ( Figure 3D). This indicates that even though the epidermis of mice unable to generate fragment N is more sensitive to stress-induced apoptosis [10], it is still able to induce cytoplasmic survivin expression to levels that are observed in wild-type mice ( Figure 3D). These results suggest that cytoplasmic survivin is not involved in fragment N-mediated protection, at least in mouse skin. Of note, the percentage of keratinocytes expressing nuclear survivin was not affected by UV-B light ( Figure 3D).
Survivin is a protein with an important role in mitosis, with a peak of expression at G2/M phase and is therefore highly regulated in a cell cycle dependent manner [61,63]. Even though fragment N does not appear to affect survivin levels, it may regulate its well-described function during mitosis. In order to study whether fragment N affects the cell cycle, cells expressing or not fragment N were synchronized in G1 using a mimosine block and then released from this block to resume cell cycling. Figure 4 shows that ectopic expression of fragment N did not alter the cell cycle of HeLa cells. There is thus no evidence that fragment N regulates survivin in vitro or in vivo.
In conclusion, the results presented in this work did not point to a crucial role of a given Akt effector in fragment N-mediated cell protection. Fragment N-mediated protection was not affected by mTOR inhibition or in BAD KO cells. Furthermore, survivin expression was neither modulated by fragment N nor was its function impaired in cells unable to generate fragment N upon stress. Possibly, fragment N relies on several Akt downstream targets to mount an efficient cell survival response. Alternatively, Akt effectors that we have not tested in the present work may fulfill most of the anti-apoptotic response induced by fragment N. Further studies need to be conducted to determine the exact contribution of Akt targets that mediate the capacity of fragment N to protect cells.