The Ras/phosphatidylinositol 3-kinase and Ras/ERK pathways function as independent survival modules each of which inhibits a distinct apoptotic signaling pathway in sympathetic neurons.

Ras promotes robust survival of many cell systems by activating the phosphatidylinositol 3-kinase (PI3-kinase)/Akt pathway, but little is understood about the survival functions of the Ras/ERK pathway. We have used three different effector-loop mutant forms of Ras, each of which activates a single downstream effector pathway, to dissect their individual contributions to survival of nerve growth factor (NGF)-dependent sympathetic neurons. The PI3-kinase pathway-selective protein Ras(Val-12)Y40C was as powerful as oncogenic Ras(Val-12) in preventing apoptosis induced by NGF deprivation but conferred no protection against apoptosis induced by cytosine arabinoside. Identical results were obtained with transfected Akt. In contrast, the ERK pathway-selective protein Ras(Val-12)T35S had no protective effects on NGF-deprived neurons but was almost as strongly protective as Ras(Val-12) against cytosine arabinoside-induced apoptosis. The protective effects of Ras(Val-12)T35S against cytosine arabinoside were completely abolished by the ERK pathway inhibitor PD98059. Ras(Val-12)E37G, an activator of RalGDS, had no survival effect on either death pathway, similar to RasS17N, the full survival antagonist. Thus, Ras provides two independent survival pathways each of which inhibits a distinct apoptotic mechanism. Our study presents one of the few clear-cut cases where only the Ras/ERK, but not the Ras/PI3K/Akt pathway, plays a dominant survival signaling role.

One of the central problems posed by degenerative disorders involving postmitotic cells is how to prevent cell death. We have demonstrated previously that p21 Ras (Ras) protein plays a pivotal role in mediating survival by nerve growth factor (NGF) 1 and other cytokines in rat sympathetic (SCG) neurons (1)(2)(3). However, the mechanisms used by Ras to protect the neurons from apoptosis are not yet fully understood. Active Ras associates with multiple downstream targets to exert its bio-logical effects (4,5) including Raf-1 kinase, the catalytic subunit of a phosphatidylinositol 3-kinase (PI3K), and the Ral guanine nucleotide dissociation stimulator (RalGDS) (6). PI3K, which is directly stimulated by Ras (7) and promotes survival through PKB/Akt in numerous cell systems (8), is persistently activated by NGF in SCG neurons (9) and has been shown to mediate survival in sympathetic (9) and sensory neurons (10) as well as robust neurite outgrowth (9). Moreover, expression of active PI3K (11) or PKB/Akt (9,12) is sufficient to protect SCG neurons from apoptosis induced by NGF withdrawal. In contrast, whereas p42 and p44 mitogen-activated protein kinases (ERKs), which mediate the Ras/Raf-1 pathway, are strongly and persistently activated by NGF (13,14), ERK activity is not required for survival support by NGF or other cytokines (14,15). A third Ras-interacting protein, RalGDS, contributes to cell transformation (4), but its functions in neurons are still unclear. In PC12 cells its overexpression inhibited neurite outgrowth induced by NGF suggesting a dominant-negative effect (16).
Recently, we uncovered a putative role for ERK activity in SCG neuron survival by demonstrating that the MEK inhibitor PD98059, which abolishes ERK activity (14,15), dramatically increased apoptosis induced by araC treatment in the presence of NGF (17). These experiments raised the possibility that besides the Ras/PI3K pathway, the Ras/ERK pathway might also protect against apoptosis of NGF-deprived neurons. There is great interest in understanding the role of the Ras/ERKs pathway in survival since it is becoming increasingly clear that there exist PI3K/Akt-independent survival signaling pathways and that PI3K/Akt activity induced by some cytokines is not being utilized for survival (18 -20), although the identity of the alternative survival signals was not determined. However one study (20) excluded ERK activity from being the alternative pathway mediating Akt-independent survival. In PC12 cells it was suggested that ERKs promoted survival by inhibiting c-Jun N-terminal kinase/p38 stress kinases (21), but in SCG neurons we found no obligate relationship between the two processes (13). In cardiomyocytes ERK activity was induced by oxidative stress and was shown to limit damage by inducing cyclooxygenase-2 expression and production of prostacyclin (22), but the relationship to apoptosis was not clear. Moreover, ERK activity is not induced by oxidative signals in SCG neurons. The role of the Ras/ERK signaling pathway in suppression of apoptosis thus remains largely unresolved.
In this study we investigate the relative importance of each of the signal pathways downstream of p21 Ras to determine whether these pathways are synergistic in protection against apoptosis, whether there are limits to protection by the PI3K pathway, and whether there are circumstances in which the ERK pathway might play a predominant role in survival. To this end, we have studied the effects of three different effectorloop domain point mutants of Ha-Ras(G12V) (Ras Val-12 ) which interact with different single effectors in mammalian cells (4 -7, 23). Ras(G12V,T35S) (Ras Val-12 35S) binds to Raf-1 and activates the mitogen-activated protein kinase pathway with about 20% efficiency of Ras Val-12 (7). It shows no interaction with PI3-kinase p110␣ subunit but interacts with RalGDS very slightly; Ras(G12V,E37G) (Ras Val-12 E37G) binds to RalGDS but not to Raf-1 or p110␣; Ras(G12V,Y40C) (Ras Val-12 40C) binds and activates PI3-kinase p110␣ but not Raf-1 or Ral-GDS. Previously it was suggested that Ras Val-12 40C but not Ras Val-12 35S suppresses c-Myc-induced apoptosis in fibroblasts (23). By using these constructs, we demonstrate that Ras Val-12 -40C is as efficient as Ras Val-12 in protecting against the death of NGF-deprived neurons. In contrast, only Ras Val-12 35S could protect against apoptosis induced by araC, the Ras/PI3K pathway playing no role in this protection. These results were corroborated using active Akt and the ERK pathway inhibitor PD98059. The RalGDS-selective proteins Ras Val-12 E37G had no protective effects. Thus unlike cycling cells where the three Ras pathways are synergistic for transformation (5), in sympathetic neurons the two Ras downstream pathways that are implicated in survival are neither additive nor synergistic. Rather, they function as independent mechanisms to support neuron survival by suppressing two different mechanisms of apoptotic induction.
Cell Culture-Superior cervical ganglia (SCG) were dissected from 1-day-old rat pups, and sympathetic neurons were extracted as described previously (13). Briefly, SCG were digested for 40 min at 37°C in 0.1% trypsin, and a single-cell suspension was obtained by triturating the digested ganglia through a narrow-bore flame-polished Pasteur pipette. To purify the neurons, cells were preplated twice for about 1 h each onto collagen-coated culture dishes in L15-CO 2 medium containing 5% rat serum under 5% CO 2 , 37°C, and nonadhering neurons were collected by centrifugation. HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 1 mM L-glutamine, penicillin (100,100 IU/ml), and streptomycin (100,100 UG/ml) in 5% CO 2 at 37°C.
Transfection and Induction of Apoptosis-Transfection was carried out following the protocol from Life Technologies, Inc., with some modifications. Briefly, newly isolated neurons were plated onto poly-L-lysine/laminin-coated glass coverslips in 24-well plates and cultured for 3 h in L15-CO 2 medium containing 3% rat serum and 20 ng/ml NGF. After one wash with serum-free medium without antibiotics, the cells were incubated for 4 h in the same medium containing 3 l of Lipo-fectAMINE and 0.15 pmol of plasmid DNA per well. Transfections were terminated by replacing the transfection mixture with the culture medium. After 36 h of transfection, cells were washed twice with medium lacking NGF and incubated in the same medium for 20 h in the presence or absence of NGF and/or 1 mM araC. In some cases, 80 M PD98059 was added. HeLa cells were passaged 1 day before transfection and transfected using the same protocol used for neurons. After 1 day of expression, the cells were washed with Dulbecco's modified Eagle's medium, serum-starved for 20 h to remove endogenous signaling through Ras, and then lysed for Western blot analysis.
Immunostaining and Evaluation of Apoptosis-SCG neurons were fixed in 3% paraformaldehyde for 20 min at room temperature. After three washes with PBS, the cells were permeabilized in PBS containing 1% BSA and 0.1% saponin followed by 1 h incubation with primary mouse antibodies to the appropriate tag (anti-c-Myc or -HA). Cells were then stained with Cy3-conjugated anti-mouse IgG for 1 h and washed twice in PBS/BSA/saponin and once in the same buffer containing Hoechst 33342. The coverslips were mounted in Vectashield (Vector). Expression of mutant Ras and apoptosis in the transfected population were determined by direct visual counting under a fluorescence microscope. The scores from the entire cohort of transfected neurons (4 -6 coverslips) were pooled to represent one independent experiment, which was repeated between 2 and 6 times. To determine the amount of apoptosis in the total population of neurons for each treatment (labeled Total in Figs. 4 and 6 -8), random fields of neurons from each well (1000 -1500 cells per well) were scored for apoptosis using Hoechst, and the results were pooled to give a single value per experiment. Figs. 4 and 6 show the total percentage of survival obtained after each treatment and also the difference in percentage survival between the total population of NGF-deprived or araC-treated neurons (set to 0) and each of the other treatments or transfections. Statistical comparisons were performed using two-tailed Student's t test.

Confirmation of Functional Specificity of Ras Constructs-
Newly isolated rat sympathetic neurons were transfected with five different mutant forms of human Ha-Ras as follows: Ras Val-12 , which is constitutively active but nonselective; Ras Val-12 35S, which couples almost exclusively to the Raf/ERK pathway; Ras Val-12 40C, which couples exclusively to the PI3K pathway; Ras Val-12 37G, which couples to RalGDS; and Ras17N, which is dominant-negative. Since the number of transfected sympathetic neurons was not high enough for a biochemical analysis by Western blotting, we first examined the fidelity of gene expression and pathway coupling of our constructs by using transfected HeLa cells (Fig. 1). Antibodies detecting either the c-Myc or HA tag epitopes fused to the N termini of the various Ras mutant proteins recognized two major bands of around 25 kDa in the transfected cells but not in control cells (Fig. 1, A and B). The lower bands represent the mature posttranslationally modified form of the transfected Ras proteins, and the upper bands consist of the unmodified forms of transfected Ras that can accumulate during overexpression. 2 To confirm the specificity of the various mutated forms of Ras for the signaling pathways, we probed the HeLa cell samples with antibodies to active phospho-ERKs, which are downstream of Raf-1, and phospho-Akt, which is downstream of PI3K. Fig. 1C shows that Ras Val-12 strongly enhanced phosphorylation of both Akt and ERKs in the transfected HeLa cells; Ras Val-12 35S only increased ERK phosphorylation whereas Ras Val-12 40C only increased Akt phosphorylation thus confirming the lack of cross-reactivity between the signals induced by the two loop effector mutants. Neither Ras17N nor Ras Val-12 37G had any activatory effect on either kinase, confirming that these mutants do not activate either pathway.
Expression of Mutant Forms of Ras in SCG Neurons-To assess the contribution of the various Ras effector pathways to neuronal survival, we scored the number of living and apoptotic neurons expressing different forms of Ras by fluorescence microscopy after staining with the DNA dye Hoechst 33342. Expression of the transfected Ras proteins was revealed by immunocytochemical staining of the c-Myc or HA tags. The extent of survival of Ras-overexpressing neurons was compared with that of the total population of cells in the same samples. As shown in Fig. 2 (a, c, e, g, and i), all the different forms of tagged Ras proteins were localized primarily to the plasma membrane, as expected of the mature protein, although there was also staining in the Golgi region, where the pre-processed form of Ras is accumulated. Staining was distributed in both neurosoma and neurites. Since these neurons were transfected prior to neurite outgrowth, Ras proteins must have been exported into the growing neurites, presumably by being included within membrane vesicles that provide neurite precursor building blocks. During the initial stages of apoptosis, neurites were collapsed, and Ras staining in the fragmented neurites was usually seen as small dots around the neurosoma (Fig. 2c, lower neuron). Occasionally, overexpression of Ras proteins caused nuclear deformation that showed a crescentic or lobulated pattern of nuclei containing uncondensed DNA masses (Fig. 2d, upper neuron). Deformation seemed to be due to the abundance of protein accumulated in the Golgi prior to export. When these neurons had healthy neurites they were scored as surviving cells.
Ras17N Abolishes NGF-mediated Survival-Expression of Ras Val-12 and Ras17N confirmed our previous results using trituration of proteins that demonstrated the critical function of Ras signaling in SCG neuron survival (1-3). Overexpression of Ras Val-12 had no deleterious effects on neurons maintained in the presence of NGF (Fig. 3, a and b; Fig. 4) (survival of Ras Val-12 -positive cells 91 Ϯ 2.6% compared with 95.4 Ϯ 0.6% in the total population) and completely blocked apoptosis induced by 20 h of NGF deprivation (survival of Ras Val-12 -positive cells 87 Ϯ 4% compared with 69 Ϯ 3.5% in the total population of neurons, p Ͻ 0.001 ϪNGF/total versus ϪNGF/Ras Val-12 ). Most Ras Val-12 -expressing cells showed a healthy profile with robust neurites despite the absence of NGF (compare Figs. 2a and 3a). In contrast, Ras17N caused massive neuronal apoptosis in the presence of NGF (survival being reduced from 96 Ϯ 0.6 to 41 Ϯ 7.6%), there being a significant increase (by 28%; p Ͻ 0.005) in the percentage of cells with apoptotic nuclei compared with the value measured in the total population of NGF-deprived neurons (Fig. 2, c and d; Fig. 4). Little further apoptosis was observed in the Ras17N-expressing neurons that were also NGF-deprived (35 Ϯ 3.3% survival) thus supporting the idea that Ras17N is only dominant-negative in the context of a counter-signal induced by NGF. Thus, most of the apoptosis that occurred in Ras17N-expressing neurons occurred prior to NGF deprivation. Furthermore, no proper neurite outgrowth could be observed in the Ras17N-positive neurons maintained in the presence of NGF (Fig. 2c) Fig. 4; p Ͻ 0.001 compared with ϪNGF/ total) and did not reduce survival in the presence of NGF (93 Ϯ 0.9%). Furthermore, Ras Val-12 40C (like Ras Val-12 ) promoted effusive outgrowth of healthy and long neurites in both presence or absence of NGF (Figs. 2g and 3e). Thus, the activity of the Ras-linked PI3K pathway fully blocks the apoptotic signal and is sufficient for supporting neuronal survival in the absence of a neurotrophic factor. In contrast to Ras Val-12 40C, Ras Val-12 35S expression (which activates the Raf-1/ERK pathway) did not block neuronal death during NGF deprivation, the rate of survival of Ras Val-12 35S-positive cells (60 Ϯ 5%) being similar to that of NGF-deprived cells in the total population (69 Ϯ 3.5%) (Figs. 3, c and d and Fig. 4). In the presence of NGF, Ras Val-12 35S-overexpressing neurons underwent as much apoptosis (69 Ϯ 2.4% survival) as the NGF-deprived neurons, suggesting that it may have some dominant-negative effects when the NGF survival signaling is active. Compared with Ras Val-12 -positive neurons, surviving Ras Val-12 35S-positive cells exhibited a thinner and less dense profile of neurites (Fig.  2, a and e, and Fig. 3, a and c). Thus the Ras/ERK pathway does not appear to deliver a protective signal against NGF withdrawal-induced neuronal death.
To address the function of RalGDS pathway, Ras Val-12 37G was expressed in SCG neurons. As shown in Fig. 2i, Ras Val-12 37G seemed to suppress neurite outgrowth in the presence of NGF, and many expressing cells did not display HA-stained neurites. In addition, a low proportion of survival (64 Ϯ 3%) was scored in Ras Val-12 37G-positive neurons in the presence of NGF (Fig. 4) suggesting a slight dominant-negative effect of Ras Val-12 37G on neuronal survival signals. This interpretation was supported by the lack of further reduction in survival (57 Ϯ 7%) when neurons were deprived of NGF for 20 h.
The Ras/ERK but Not the Ras/PI3K Pathway Protects Neurons from Apoptosis Induced by araC-Treatment with 1 mM araC can induce sympathetic neuron apoptosis in the presence of NGF as efficiently as NGF deprivation (24,25). Our previous data (17) suggested that ERK activity was involved in the protection against araC-induced apoptosis prompting us to explore the contribution of both the Ras/Raf-1 and Ras/PI3K pathways to this protection using mutant Ras constructs. After 36 h of expression of Ras Val-12 , Ras Val-12 40C, or Ras Val-12 35S, the neurons were treated with 1 mM araC in the presence or absence of NGF for 20 h. In the presence of NGF, araC reduced neuronal survival to 70 Ϯ 3% in the total population of neurons (Fig. 6a) and caused similar cell death in neurons expressing all three forms of mutant Ras (70 Ϯ 2, 62 Ϯ 5, and 64 Ϯ 7% survival with Ras Val-12 -, Ras Val-12 40C-, or Ras Val-12 35S-transfected cells) showing that araC induces signals that override the protective effects of these constructs. However, there was no increase in apoptosis in neurons expressing Ras Val-12 35S beyond that induced by araC in the presence of NGF in the total population suggesting that Ras Val-12 35S is no longer conferring a negative signal (unlike the negative effect observed in the presence of NGF alone). Clearly, neither survival signals from NGF nor Ras are sufficient to block completely the neuronal death due to araC treatment.
However, when neurons were treated with araC in the absence of NGF, dramatically higher neuronal survival was observed in the Ras Val-12 -and Ras Val-12 35S-expressing neurons (64 Ϯ 5% for Ras Val-12 and 47 Ϯ 8.3% for Ras Val-12 35S) compared with the total population of neurons (13 Ϯ 0.5% survival; p Ͻ 0.001) demonstrating that the signal from the Ras/ERK pathway reduces araC-induced neuronal death ( Fig. 5 and 6b).
Remarkably, contrary to its protective role during NGF deprivation, Ras Val-12 40C failed to rescue the neurons undergoing araC treatment, and only 18 Ϯ 0.1% survival was observed in the neurons expressing Ras Val-12 40C (Figs. 5 and 6b). Ras Val-12 37G also failed to protect against araC-induced apoptosis (Fig. 6, a and b; 57 Ϯ 7% or 20 Ϯ 5% survival in the presence or absence of NGF).
The Protection of Ras Val-12 35S against araC-induced Apoptosis Is Dependent on ERK Activity-Although we showed above that Ras Val-12 35S promotes ERK (but not Akt) phosphorylation in HeLa cells, these data provide no evidence that Ras Val-12 35S is utilizing the ERK pathway in SCG neurons. Moreover, because Ras Val-12 35S induced some death in the presence of NGF that was not further enhanced when NGF was withdrawn (Fig. 4), it was still technically possible that Ras Val-1235S was delivering a survival signal to the latter neurons. To investigate whether Ras Val-12 35S has survival effects in the absence of NGF, and whether Ras/ERK is the pathway used by Ras Val-12 35S to protect against araC, we blocked the ERK signaling pathway using PD98059 (14,15,17). PD98059 completely inhibited ERK phosphorylation induced by Ras Val-12 35S in HeLa cells (Fig. 7a). However, Fig. 7b shows that PD98059 did not significantly decrease the survival of Ras Val-12 35S-expressing neurons that were NGF-deprived (ϪNGF, 57 Ϯ 3.6%, ϪNGF ϩ PD98059 46 Ϯ 2%; p Ͼ 0.05) although it caused a dramatic reversal in the survival of araC-treated, NGF-deprived neurons (ϪNGF ϩ araC, 41 Ϯ 5%; ϪNGF ϩ araC ϩ PD98059, 12 Ϯ 3%, p Ͻ 0.001), similar to the survival of the total population of neurons under same conditions (ϪNGF ϩ araC, 10 Ϯ 1%; ϪNGF ϩ araC ϩ PD98059, 8 Ϯ 2%). Thus, ERK  activity triggered by Ras Val-12 35S contributes to the protection against araC-induced neuronal apoptosis but not against apoptosis induced by NGF deprivation.
Active Akt Does Not Inhibit Apoptosis Induced by araC-The Ras/PI3K pathway activates Akt in numerous cell systems (8).
To address further the role of the Ras/PI3K pathway signal in araC-induced neuronal apoptosis, we transfected a membranetargeted Akt construct (m/p-Akt) that we showed previously to support robust survival of NGF-deprived neurons and that is activated by NGF in a PI3K-dependent manner (9,17). Expression of m/p-Akt promoted its strong phosphorylation in HeLa cells (Fig. 8a) without increasing ERK activity. Similar to Ras-Val-12 40C, m/p-Akt significantly blocked neuronal death induced by NGF withdrawal (81 Ϯ 4% survival compared with 61 Ϯ 7% survival in total population p Ͻ 0.01) but did not improve neuronal survival after araC treatment (14 Ϯ 0.6% compared with 11 Ϯ 0.4% in the total population) (Fig. 8b). These data suggest that Ras Val-12 40C operates through the Akt pathway, and that none of the components of this pathway can obstruct the death signals triggered by araC.

DISCUSSION
Ras is an important mediator of anti-apoptotic signals induced by survival factors in several cell systems, thus clarification of its roles in survival and apoptosis is of immense interest. By using the technique of "pressure trituration" (26 -28), we (1-3) and others (29) demonstrated that Ras is an essential mediator of survival induced by NGF, cytokines, or cyclic AMP in SCG neurons. Given the pivotal role of Ras in SCG neuron survival, we have analyzed the mechanisms that are used by Ras to prevent neuronal apoptosis to determine whether Ras confers its protection by operating additive, synergistic, or independent pathways. We show that both the Ras/PI3K virgule Akt and the Ras/ERK pathways are crucial mediators of neuronal survival because there is no overlap in the processes that are targeted by each pathway. Thus, only Ras/PI3K/Akt prevents death due to NGF deprivation, but the Ras/ERK pathway is the only pathway that can promote neuroprotection when death is induced by araC. Our study reveals one of the first clear-cut cases where the Ras/ERK but not the Ras/PI3K/Akt pathway plays a dominant survival signaling role.
Ras17N and Ras Val-12 37G Do Not Activate Survival Pathways-To evaluate the technique of transient transfection, we first examined whether the survival of SCG neurons, which had been inhibited previously using anti-Ras blocking antibodies (2), could also be inhibited by the dominant-negative form of Ras, Ras17N. We found that expression of inactive Ras17N promoted a dramatic increase in apoptosis in the presence of NGF (Figs. 2 and 4), more than double the amount induced by NGF deprivation. The most likely reason for the increase in apoptosis in neurons expressing Ras17N is that these neurons underwent a much longer period of Ras signal deprivation since the dominant-negative effect of Ras17N would have begun from the time of its expression, whereas NGF deprivation began only 36 h after transfection. Further evidence that Ras17N functioned by counteracting active Ras rather than by being directly pro-apoptotic is that there was little further increase in the death of Ras17N-expressing neurons after NGF deprivation (when Ras is no longer active).
Our data also indicate that Ras Val-12 37G, which was identified by its ability to activate RalGDS (4), has no protective effects, since it blocked survival in the presence of NGF but failed to either reduce cell death after NGF deprivation or to protect against araC-induced apoptosis. In addition to death induction, expression of Ras Val-12 37G also reduced neurite outgrowth. The roles of RalGDS, which stimulates guanine nucleotide dissociation from Ral, in transmitting signals from Ras The overexpressed m/p-Akt strongly enhanced Akt phosphorylation but had no effect on Erk activity (right panel). B, SCG neurons were transfected with m/p-Akt. After 36 h, the neurons were washed and treated with normal growth medium, or NGF-deprived, or exposed to 1 mM araC in the absence of NGF for 20 h as indicated. 200 -2000 transfected neurons were counted per treatment. m/p-Akt significantly reduced the death of NGF-deprived neurons but showed no protection against araC-induced apoptosis. Data correspond to the mean of three independent experiments; the error bars are standard deviations (p Ͻ 0.01 ϩNGF/total versus ϪNGF/total not significant, ϩNGF/Akt versus ϪNGF/Akt; p Ͼ 0.05 ϪNGFϩaraC/total versus ϪNGFϩaraC/Akt). and its cellular functions are still not clear (30). In PC12 cells, Ral guanine nucleotide exchange factors act opposite to other Ras effectors suppressing cell cycle arrest and neurite outgrowth induced by NGF (16). However, a growth-promoting effect of RalGDS was observed either when it was expressed alone (31,32) or in cooperation with the Ras/Raf pathway (4). Thus, if RalGDS mediates any of the effects of Ras in SCG neurons, it is clear that the Ral pathway is not used by Ras to deliver a survival signal. The negative effect of Ras Val-12 37G on neurite outgrowth may be due to Ral activity but could also be due to inhibition of the Ras/PI3K signal through its considerable dominant-negative effects since PI3K activity is required for NGF-induced neurite outgrowth in SCG neurons (9).
Only the Ras/PI3K Pathway Can Mediate Protection by Ras against NGF Deprivation-The PI3K pathway has been implicated in anti-apoptotic signaling in numerous cell types, including sympathetic neurons (9,11,12), and sensory neurons (10,33). One known effector of PI3K is Akt, a serine-threonine kinase that is activated in response to NGF stimulation (9,17) and that was shown recently to be required for SCG neuron survival (9,12). Although PI3K is activated directly by Ras (7), it has been suggested that a major proportion of PI3K activity induced by NGF through TrkA is mediated upstream of Ras (34). In this report, we show that Ras Val-12 40C reduced neuronal apoptosis due to NGF withdrawal as effectively as Ras  or active Akt. Taken together with similar results obtained with active PI3K (11), these data support the idea that the Ras/PI3K/Akt pathway may mediate the NGF survival signal and exclude a role in survival for the other Ras-linked effector pathways tested here. Overexpression of Ras Val-12 35S not only failed to inhibit neuronal apoptosis caused by NGF withdrawal (Figs. 3 and 4) but in the presence of NGF its overexpression reduced neuronal survival to a similar level as that found in the total NGF-deprived neuron population, suggesting that it has mild dominant-negative effects against NGF-induced active Ras. This idea is supported by the findings that like Ras17N, no further decrease in survival was observed in Ras Val-12 35S-positive cells undergoing NGF deprivation, during which time Ras is not activated. Moreover, the MEK inhibitor PD98059 failed to increase the death of Ras Val-12 35S-expressing neurons deprived of NGF.
Only the Ras/ERK Pathway Can Mediate Protection by Ras against araC-induced Death-Our finding that the Ras/ERK pathway plays no role during NGF-induced survival can be contrasted with our finding a crucial role for Ras/ERK signaling, but not the Ras/PI3K signaling, in counteracting araCinduced apoptosis. AraC causes apoptosis in several types of neurons in vitro by mechanisms that are still unclear (35)(36)(37). Our previous data showed that araC requires the presence of wild type p53 to induce apoptosis in sympathetic neurons and proposed a potential protective role of ERK against araC toxicity mainly through the observation that the MEK inhibitor PD98059 caused a significant increase in araC-induced apoptosis in the presence of NGF and that ciliary neurotrophic factor, which cannot sustain ERK activity (13), also provided no protection against araC (17). When araC was added to sympathetic neurons expressing Ras Val-12 , Ras Val-12 40C, or Ras Val-1235S in the presence of NGF (Fig. 6a), there was no significant difference in the loss of survival between non-transfected or the mutant Ras-expressing neurons, the degree of apoptosis in each case being about 30%. For Ras  or Ras Val-12 40C this is the result predicted because araC does not interfere with NGF signaling (17) so the contribution of the signals from these Ras mutants would be redundant to that induced by NGF. The lack of increased apoptosis in neurons expressing Ras Val-12 35S, which might have been predicted from its dominant-negative actions in the presence of NGF, was most likely because of its ability to counterbalance this negative input by its survivalpromoting effect, revealed most clearly when araC was added in the absence of NGF.
Thus, in the presence of araC but absence of NGF, Ras Val-12 and Ras Val-12 35S reduced neuronal death close to the levels observed in the total population of neurons incubated in presence of NGF and araC (64, 47, and 70%, respectively), against only 13% survival of the total population of neurons. The less substantial survival support by Ras Val-12 35S compared with Ras Val-12 is consistent with its weaker ERK stimulating activity ( Fig. 1b (7)). However, the survival of Ras Val-12 40C-and m/p-Akt-overexpressing cells was negligible, being only 5 and 3%, respectively, higher than that of the non-transfected cells. We noted that even protection by Ras Val-12 did not completely block apoptosis induced by araC, as observed when araC treatment occurs in the presence of NGF. Incomplete protection by NGF or Ras Val-12 may be due to multiple signals induced by araC/p53, only one of which is inhibitable by ERK activity. Therefore, araC always provokes a dominant death phenotype, explaining why Ras Val-12 40C and active Akt fail to exert any protection when neurons are exposed to a combination of NGF deprivation and araC treatment. To demonstrate that Ras Val-12 35S was utilizing the Ras/ERK pathway to inhibit araC-induced apoptosis, we used the MEK inhibitor PD98059. As predicted, PD98059 completely abolished the survivalpromoting effects of Ras Val-12 35S. Therefore, considering that Ras Val-12 and Ras Val-12 35S both target the Raf/ERK pathway but that Ras Val-12 40C only activates the PI3K/Akt pathway, the activation of Raf/ERK but not the PI3K pathway is able to reduce neuronal death induced by araC treatment without synergizing with any other Ras pathway.
The Targets of the Two Survival Pathways Must Be Upstream of Cytochrome c Release-Both NGF deprivation and araC treatment cause very similar apoptotic profiles, cytochrome c translocation from the mitochondria to the cytosol, caspase 3 activity, and poly (ADP-ribose) polymerase cleavage in sympathetic neurons (Ref. 38 and data not shown). Thus, the complete separation of pathways that counteract NGF depri- FIG. 9. Scheme summarizing proposed mechanisms used by Ras to protect SCG neurons from apoptosis. NGF withdrawal and araC induce neuronal death by different mechanisms that converge at a point upstream of cytochrome c release (38). The Ras/PI3K pathway is both necessary and sufficient to deliver the protective signal against apoptosis induced by NGF deprivation, but none of the downstream effects of this pathway (including Akt) can target any of the components mediating araC-induced release of cytochrome c, activation of caspases, or apoptosis execution. Although the Ras/ERK pathway is activated concurrently, it can only protect neurons from araC-induced death. See "Discussion" for further details. vation or araC treatment suggests that the two death stimuli trigger SCG neuronal death using at least two different, independent mechanisms. Since Ras Val-12 35S and Ras Val-12 40C contribute their protective effects to the two types of neuronal apoptosis independently, Ras must exert its protective role before the death mechanisms converge onto the mitochondria. The targets of Ras signals that protect neurons from apoptosis through its different effector pathways are still unclear. It has been suggested that Akt, the downstream target of PI3K, inactivates the proapoptotic function of BAD, a member of Bcl-2 family, by phosphorylation of Ser-136 (39,40). Although evidence for phosphorylation of BAD by Akt in sympathetic neurons is still lacking, this mechanism could potentially explain how Ras rescues the neurons from NGF withdrawal-induced apoptosis. If so, then the pathway that is inhibited by the Ras/ERK pathway must regulate cytochrome c release by another mechanism. It is now of major interest to find out what its targets might be.
The summary scheme in Fig. 9 shows that the power of p21 Ras as a neuroprotective signal generator lies in its ability to promote signaling through the PI3K and ERK pathways simultaneously but non-redundantly to inhibit two independent mechanisms of apoptosis. The results presented here not only provide new insights into the mechanisms of Ras functions but also identify the kind of therapeutic agents that might be effective in preventing neurodegeneration caused by neurotrophin deprivation or external stimulation.