Neuroprotection by Brain-derived Neurotrophic Factor Is Mediated by Extracellular Signal-regulated Kinase and Phosphatidylinositol 3-Kinase*

Apoptosis is a form of programmed cell death that plays a pivotal role during development and in the homeostasis of the adult nervous systems. However, mechanisms that regulate neuronal apoptosis are not well defined. Here, we report that brain-derived neurotrophic factor (BDNF) protects cortical neurons against apoptosis induced by camptothecin or serum deprivation and activates the extracellular-signal-regulated kinase (ERK) and the phosphatidylinositol 3-kinase (PI 3-kinase) pathways. Using pharmacological agents and transient transfection with dominant interfering or constitutive active components of the ERK or the PI 3-kinase pathway, we demonstrate that the ERK pathway plays a major role in BDNF neuroprotection against camptothecin. Furthermore, ERK is activated in cortical neurons during camptothecin-induced apoptosis, and inhibition of ERK increases apoptosis. In contrast, the PI 3-kinase pathway is the dominant survival mechanism for serum-dependent survival under normal culture conditions and for BDNF protection against serum withdrawal. These results suggest that the ERK pathway is one of several neuroprotective mechanisms that are activated by stress to counteract death signals in central nervous system neurons. Furthermore, the relative contribution of the ERK and PI 3-kinase pathways to neuronal survival may depend on the type of cellular injury.

However, most mechanistic studies of apoptosis have been limited to proliferating, non-neuronal cells or neurons derived from the peripheral nervous system, including superior cervical ganglion (SCG) or dorsal root ganglion neurons. Although mature neurons and proliferating non-neuronal cells may share some apoptotic mechanisms, mature CNS neurons do not divide. Consequently, there may be significant differences in the mechanisms for apoptosis in CNS neurons and dividing cells. For example, cytosine arabinoside, a DNA synthesis inhibitor, triggers apoptosis in both dividing cells and post-mitotic neurons (4). However, cycloheximide, a protein synthesis inhibitor, potentiates apoptosis in many dividing cells but inhibits apoptosis in post-mitotic neurons (1,3,(5)(6)(7)(8). Because the biochemical and regulatory properties of post-mitotic CNS neurons are distinct from those of peripheral nervous system neurons and dividing non-neuronal cells, it is crucial to define apoptotic mechanisms specific to CNS neurons; this may ultimately lead to the identification of drug targets for modulation of neuronal apoptosis in the CNS and the development of clinical strategies for treatment of neurodegenerative disorders.
The regulatory mechanisms that control neuronal survival and apoptosis are just beginning to be defined (9,10). Many growth factors and neurotrophins can promote neuronal survival, including insulin, insulin-like growth factor-1, BDNF, nerve-growth factor (NGF), and neurotrophins 3 and 4/5 (11)(12)(13)(14). These factors can also activate several intracellular signaling transduction systems including the ERK and the PI 3-kinase pathways (15,16). Activation of the PI 3-kinase pathway is required for NGF-mediated survival of PC12 cells and SCG neurons (17,18), for insulin-like growth factor-1-mediated survival of cerebellar granule neurons, oligodendrocytes, and PC12 cells (19 -22), and for membrane depolarization-mediated survival of cerebellar granule neurons (20). Furthermore, protein kinase Akt (also known as PKB or RAC) may mediate cellular survival because of activation of PI 3-kinase in cerebellar granule neurons and other non-neuronal cells (23,24). Collectively, these data have established the PI 3-kinase pathway as a major neuroprotective mechanism (14,25).
Although activation of the ERK signaling pathway protects various non-neuronal cell lines against apoptosis (26 -28), its role for promoting the survival of neurons is still controversial. Activation of ERK promotes PC12 cell survival (22,29,30). Furthermore, studies with pharmacological inhibitors suggest that ERK activation may mediate neuroprotection by BDNF in retinal ganglion cells and cerebellar neurons (31,32), by NGF in sympathetic neurons (33), and by the pituitary adenylate cyclase-activating polypeptide  in cerebellar neurons (34). However, other studies using inhibitors have suggested that ERK does not mediate the neuroprotection afforded by neurotrophins (NGF, BDNF, and insulin-like growth factor-1) or membrane depolarization in PC12, SCG, or cerebellar neurons (20,(35)(36)(37).
The objective of this study was to evaluate the roles of the ERK and the PI 3-kinase pathways for BDNF neuroprotection of cortical neurons. Cortical neurons were chosen because the importance of ERK for neuroprotection in cortical neurons had not been examined and neurons in the cortex are frequently damaged during neurodegenerative diseases. Because different neuroprotective mechanisms may be activated to combat distinct types of cellular stress, we used two apoptotic paradigms: apoptosis induced by serum deprivation or by camptothecin treatment. Because the optimal growth and survival of neurons depend on the availability of growth factors and neurotrophic factors, serum withdrawal has been widely used as a model for CNS neuronal apoptosis induced by developmental cues (38,39). Camptothecin is an inhibitor of DNA topoisomerase-1. It induces DNA strand breaks during replication as well as transcription, and it may inhibit transcription (40,41). Camptothecin-induced apoptosis has been used as a model system to study neuronal apoptosis induced by DNA damage, which may contribute to several neurodegenerative diseases and agingrelated neuron loss (40,(42)(43)(44). Our results suggest that the ERK signaling pathway plays a pivotal role in the BDNF protection against camptothecin, whereas the PI 3-kinase pathway is critical for BDNF protection against serum deprivation.

EXPERIMENTAL PROCEDURES
Materials-The plasmid pON260 has been described previously (45). The HA-tagged expression vectors for wild type and constitutive active MKK1 (⌬N3-S218E/S222D) were obtained from N. G. Ahn (46). The anti-ERK antibody used for immunoprecipitation was purchased from Upstate Biotechnology Inc. Anti-phospho-ERK antibody (anti-AC-TIVE TM mitogen-activated protein kinase, polyclonal antibody) and anti-phospho-Ser-473 Akt antibody were purchased from Promega and New England Biolabs, respectively.
Cell Culture-Cortical neurons were prepared from newborn Harlan Sprague-Dawley rats and cultured at a density of 1500 -2000 cells/mm 2 in basal medium Eagle (BME, Sigma) supplemented with 10% heatinactivated bovine calf serum (Hyclone, Logan, UT) (47). Cytosine arabinoside (2.5 M) was added on the second day in vitro after seeding (DIV 2) to inhibit the proliferation of non-neuronal cells.
Transfection-Cortical neurons were transiently transfected at DIV 3 using a calcium phosphate co-precipitation protocol (47) with modifications. The DNA/calcium phosphate precipitate and the transfection medium were prepared as described (47). The transfection medium was then "de-gassed" in a dish at 37°C with no CO 2 for 30 min to raise the pH level before adding it to the cultures. Cells were washed three times with BME, and 1.5 ml of de-gassed transfection media were added to each 35-mm dish, immediately followed by the addition of 60 l of DNA/calcium phosphate precipitates. Plates were incubated at room temperature in ambient air for 5 min and then in a humidified incubator with 5% CO 2 at 37°C for 35-45 min. The incubation was stopped by a 5% glycerol shock. Experimental treatments were initiated 24 -48 h after transfection. This modified protocol improved transfection efficiency while maintaining low transfection toxicity.
Drug Treatment-At DIV 4 -6, cortical neurons were treated with various concentrations of camptothecin, PD98059, or LY294002, all dissolved in dimethyl sulfoxide (Me 2 SO). The final concentration of Me 2 SO was 0.2%. When cultures were co-treated with camptothecin and PD98059 or LY294002, the final concentration of Me 2 SO was below 0.6%. BDNF was diluted in PBS containing 0.1% bovine serum albumin before its addition to the cells. The duration of each drug treatment is described in detail in the figure legends.
Serum Deprivation-Serum deprivation experiments were performed with neurons cultured on glass coverslips at DIV 4 -6. The conditioned medium from the culture was removed and saved ("serumcontaining conditioned medium"). For serum deprivation, cells were washed twice with serum-free BME and then incubated in serum-free BME supplemented with 35 mM glucose, 1 mM L-glutamine, 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 2.5 M cytosine arabinoside. Control cells were washed the same way and then incubated in the serum-containing conditioned medium. This washing process slightly increased the basal cell death, probably because of stress during washing.
Cortical Neuron Survival Assayed by MTT Metabolism-Neuronal survival was assayed by measuring the conversion of the yellow, watersoluble tetrazolium, MTT (3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) to the blue, water-insoluble formazan. This conversion is catalyzed by cellular mitochondrial dehydrogenases. Because the rate of this reaction is proportional to the number of surviving cells, the MTT assay is widely used to quantify viable cells (48). MTT assays were performed in 96-well plates as described (49). Optical blanks, used as controls, were generated by incubating the corresponding drugs and FIG. 1. Camptothecin (CPT) induces cortical neuron death. A, dose response and kinetics for camptothecin-induced cortical neuron death. At DIV 5, cortical neurons were treated with 1, 2, 5, or 10 M camptothecin or with an equal volume of vehicle (0.2% Me 2 SO). Neuronal viability was determined by MTT metabolism at 12, 24, or 48 h after the initial treatment. One hundred percent survival was defined as the level of MTT metabolism in cultures treated with vehicle for only 12 h. B, inhibition of camptothecin-induced cortical neuron death by cycloheximide, a protein synthesis inhibitor. Cortical neurons were pretreated for 30 min with 0.1 g/ml cycloheximide (ϩ cycloheximide) or PBS as a vehicle control (Ϫ cycloheximide) and then were treated with 1, 2, and 5 M camptothecin to induce cell death. Cell viability was measured 48 h later by MTT metabolism. One hundred percent survival was defined as the level of MTT metabolism in cultures treated with vehicle (PBS ϩ 0.2% Me 2 SO) for 12 h. Data are the averages of triplicate determinations. Similar results were obtained in three independent experiments; error bars ϭ S.E..
MTT in the conditioned medium. Data are presented as the percentage of survival relative to vehicle-treated control cultures. All MTT assays were performed in triplicates.
Quantitation of Apoptosis by Nuclear Morphological Changes-To visualize nuclear morphology, cells were fixed in 4% paraformaldehyde and stained with 2.5 g/ml DNA dye Hoechst 33258 (bis-benzimide, Sigma) (29). Apoptosis was quantitated at each time point by scoring the percentage of apoptotic cells in the adherent cell population. Uniformly stained nuclei were scored as healthy, viable neurons. Condensed or fragmented nuclei were scored as apoptotic. To obtain unbiased counting, slides were coded, and cells were scored blind without knowledge of their prior treatment. Statistical analysis of the data was performed using one-or two-way analysis of variance (ANOVA) followed by post hoc tests.
DNA Ladder Assay-To examine DNA cleavage, we plated 4 ϫ 10 6 cortical neurons in each dish and treated these cultures with camptothecin for various times. Soluble cytoplasmic DNA was isolated from each plate and loaded to 1.8% agarose gel for DNA ladder analysis (50).
These cortical neurons were post-mitotic, and apoptotic cells at more advanced stages generally did not adhere to the plates well and were often detached from plates. Consequently, there were less total cells adherent to each plate and less total DNA isolated from each plate at later time points than at earlier time points. This phenomenon could contribute to the decreasing level of DNA laddering at 48 h, and it is also consistent with reports in the literature (17,23).
ERK Kinase Assay-ERK activity was quantitated using an immune complex kinase assay as described previously (29). Quantification of kinase activity was achieved by excising the radiolabeled myelin basic protein from a gel after SDS-polyacrylamide gel electrophoresis and counting radioactivity in the Cerenkov channel of a scintillation counter.
Immunostaining-Transfected cells were detected by immunostaining with a polyclonal antibody against ␤-galactosidase and Texas Red-conjugated goat antibody to rabbit IgG. Cells transfected with ␤-galactosidase stained red. Cells transfected with the hemagglutinin (HA)-epitope-tagged MKK1 were immunostained with monoclonal antibody to HA (0.8 g/ml). These transfected cells were visualized by fluorescein-conjugated goat antibody to mouse IgG and stained green. To visualize the nuclei of transfected cells, we included the Hoechst 33258 dye in the wash after the secondary antibody incubation.

Camptothecin Induces Apoptosis in Primary Cultured
Cortical Neurons-To evaluate the purity of the cortical neuron cultures used in this study, cells were double-immunostained with antibodies against MAP-2 and glial fibrillary acidic protein, which are markers for neurons and astrocytes, respectively. More than 90% of the cultured cells stained positive with anti-MAP-2, and only 8% were positive for glial fibrillary acidic protein. Cortical neurons were treated with varying concentrations of camptothecin (1-10 M) and assayed for cell viability at various times after treatment using the MTT metabolism assay (Fig. 1A). Camptothecin reduced MTT metabolism in a doseand time-dependent manner. For example, only 30% of the neurons survived 48 h after treatment with 10 M camptothecin. Cell death was completely blocked by cycloheximide, a protein synthesis inhibitor, suggesting that camptothecin toxicity may depend upon translation (Fig. 1B).
To determine whether camptothecin induces apoptosis in cortical neurons, neurons were stained with the DNA dye Hoechst 33258 to visualize nuclear morphology. DNA cleavage was assayed by agarose gel electrophoresis. In the absence of camptothecin, the cultured neurons exhibited normal cellular morphology with extending neurites and evenly stained nuclei ( Fig. 2A-C). Camptothecin caused morphological changes characteristic of apoptosis, including degeneration of neurites and shrinkage of cell bodies (Fig. 2D), as well as fragmentation and condensation of nuclei (Fig. 2, E and F). Camptothecin also caused DNA cleavage into oligonucleosome fragments manifested as "DNA laddering," a hallmark of apoptosis (Fig. 2G). DNA cleavage was detectable as early as 18 h after treatment and persisted for a minimum of 48 h. A decrease in the amount of fragmented DNA visible at 48 compared with 18 h (Fig. 2G) is similar to that observed in other reports (17,23).
The percentage of neurons showing an apoptotic phenotype was quantitated as a function of the camptothecin concentration and time after challenge with the drug (Fig. 2H). Evenly stained nuclei were scored as healthy, viable neurons, whereas condensed or fragmented nuclei were scored as apoptotic. Only 4% of the apoptotic cells were glial fibrillary acidic proteinpositive astrocytes (data not shown), indicating that apoptotic signals were due primarily to neurons. The camptothecin concentration dependence and kinetics for apoptosis were comparable with those seen when the cells were assayed for viability using the MTT assay. Nuclear morphological changes were evident 12 h after the addition of camptothecin, and 80% of the nuclei were apoptotic 48 h after treatment with 5-10 M camptothecin. The increase in nuclear fragmentation and condensation generally preceded the reduction in MTT metabolism (compare Figs. 1A and 2H) because early apoptotic cells maintain mitochondria function (51). Furthermore, expression of Bcl-2 or Bcl-xL completely protected cortical neurons from camptothecin-induced apoptosis (data not shown). Collectively, these data indicate that camptothecin evokes a classical apoptotic phenotype in cultured cortical neurons.
BDNF Protects Cortical Neurons from Camptothecin-induced Apoptosis-To define signaling pathways that promote survival of CNS neurons, we tested the effectiveness of several neurotrophins against camptothecin-induced apoptosis. BDNF, a member of the NGF neurotrophic factor family, protects several types of CNS neurons from apoptosis caused by trophic support withdrawal (32, 37, 52, 53). However, it was not known whether BDNF protected CNS neurons from DNA-damaging agents. Therefore, cortical neurons were treated with 10 M camptothecin for 24 h in the presence or absence of varying concentrations of BDNF (Fig. 3). Neuronal apoptosis was measured by nuclear fragmentation and condensation (Fig. 3, A-E), and survival was measured by MTT metabolism (Fig. 3F). BDNF protected camptothecin-induced cortical neuron apopto- sis in a dose-dependent manner. For example, after 24 h of treatment, camptothecin induced apoptosis in 75 or 26% of neurons in the absence or presence of 30 ng/ml BDNF, respectively. Similar levels of protection were also observed when cortical neuron survival was assayed by MTT metabolism (Fig.  3F). In addition, cortical neuron apoptosis induced by camptothecin was also partially blocked by neurotrophic factor-3 (NT-3) but not by NGF (data not shown).
The PI 3-Kinase Signaling Pathway Is Activated by BDNF in Cortical Neurons but Plays a Minor Role in BDNF Neuroprotection against Camptothecin-BDNF can activate several signal transduction systems including the PI 3-kinase and the ERK pathways (16,54). Stimulation of PI 3-kinase can lead to the activation and phosphorylation of protein kinase Akt (25). Therefore, we assayed PI 3-kinase activity by Western analysis using a phospho-Akt antibody that specifically recognizes activated Akt. The same samples were also probed with an anti-Akt antibody to ensure an equal amount of protein loading in each lane (data not shown). Phosphorylation of Akt was detectable in cortical neurons maintained in regular culture conditions (Fig. 4A), probably because of PI 3-kinase activation by growth factors present in serum. Addition of BDNF at concentrations as low as 10 ng/ml caused a large increase in Akt phosphorylation, suggesting that the PI 3-kinase pathway was activated by BDNF. Furthermore, BDNF stimulation of Akt phosphorylation was comparable in the presence or absence of camptothecin (data not shown).
To determine whether activation of the PI 3-kinase pathway contributes to BDNF protection against camptothecin we utilized LY294002, a PI 3-kinase inhibitor (17,55), to block PI 3-kinase activation by BDNF. LY294002 at 30 M completely inhibited Akt phosphorylation under basal culture conditions and after stimulation with BDNF, indicating that LY294002 is an effective inhibitor of the PI 3-kinase pathway under these conditions (Fig. 4A).
We next examined the effect of blocking PI 3-kinase activation on BDNF neuroprotection against camptothecin. Cortical neurons were pretreated with 30 M LY294002, 10 ng/ml BDNF, or both and then challenged in the absence or presence of 10 M camptothecin (Fig. 4B). In the absence of camptothecin, LY294002 alone increased basal apoptosis from 14 to 44%, suggesting that the PI 3-kinase pathway may be required for serum-promoted cortical neuron survival (see Figs. 11-13). Camptothecin induced apoptosis in 70% of the cell population; the addition of LY294002 increased apoptosis to 84%. As shown earlier, BDNF greatly reduced camptothecin-induced apoptosis, in this experiment to 29%. Inhibition of PI 3-kinase by LY294002 partially reversed BDNF neuroprotection against camptothecin (44 versus 29% of apoptosis). This decrease was small but statistically significant (p Ͻ 0.01). However, there was a larger difference in the extent of camptothecin-induced apoptosis when neurons were pre-treated with LY294002 alone or together with BDNF (84 versus 44%, p Ͻ 0.0001). Transient expression of a constitutively active PI 3-kinase provided minimal protection against camptothecin-induced apoptosis (data not shown). These data suggest that although activation of the PI 3-kinase pathway may offer some protection, it is not the major mechanism for BDNF protection against camptothecin. Other principle signaling pathways may play a larger role in BDNF neuroprotection against camptothecin.
The ERK Signaling Pathway Is a Major Mechanism for BDNF Neuroprotection against Camptothecin-Because ERK activation mediates the protective effect of NGF in PC12 cells and of BDNF in retinal ganglion cells (29,31), we tested the hypothesis that activation of the ERK pathway may contribute to BDNF neuroprotection against camptothecin. ERK activation by BDNF was quantitated using an immune complex kinase assay, and BDNF at concentrations as low as 10 ng/ml activated ERK (Fig. 5A). ERK was activated up to 6-fold and remained stimulated for more than 12 h. Furthermore, BDNF activation of ERK was comparable in the presence or absence of camptothecin (data not shown). ERK activation by BDNF was also assayed and confirmed by Western analysis using an antiphospho-ERK antibody that specifically recognizes activated ERK (Fig. 5B).
To determine whether ERK activation is important for BDNF neuroprotection, we utilized PD98059 to block ERK activation by BDNF. PD98059 is a specific inhibitor for MKK1 and -2 (56), upstream kinases that phosphorylate and activate ERK (15,57). Although all drugs have the potential to exhibit nonspecific effects, we used PD98059 at concentrations (20 and 40 M) that are known to have no effect on other related MAP kinases including Jun N-terminal kinase and p38 (56). Cortical neurons were stimulated with various concentrations of BDNF in the presence of 0, 20, or 40 M PD98059. ERK stimulation by BDNF was assayed by ERK phospho-Western analysis using the anti-phospho-ERK-specific antibody. PD98059 at 40 M inhibited BDNF-stimulated ERK activation (Fig. 5B).
To evaluate the influence of ERK inhibition on BDNF neuroprotection, cortical neurons were treated with camptothecin in the presence or absence of 40 M PD98059, 10 ng/ml BDNF, or both, and neurons were analyzed for apoptosis 24 h later (Fig. 6A). PD98059 increased camptothecin-induced apoptosis, whereas BDNF inhibited it. The neuroprotection afforded by BDNF was greatly reduced by PD98059 inhibition of ERK (p Ͻ 0.0001). For example, 68, 81, 30, and 62% of neurons were apoptotic after 24 h of treatment with camptothecin alone, camptothecin plus PD98059, camptothecin plus BDNF, or camptothecin plus BDNF and PD98059, respectively. Similar results were obtained with the MTT assay (data not shown). These data suggest that activation of the ERK signaling pathway plays a significant role in BDNF neuroprotection against camptothecin.
We also treated cortical neurons with combinations of PD98059 and LY294002 to inhibit the activation of both ERK and PI 3-kinase pathways (Fig. 6B). Under these conditions, 86 or 75% of cells underwent apoptosis after treatment with camptothecin alone or camptothecin plus BDNF, respectively. This finding indicates that BDNF is ineffective in protecting cortical neurons against camptothecin when the ERK and the PI 3-kinase pathways are both inhibited.
Is activation of the ERK pathway sufficient to protect cortical neurons from camptothecin toxicity? To constitutively and selectively activate ERK, we transiently transfected cortical neurons with plasmid DNA encoding a constitutively active MKK1 (46), which does not activate Jun N-terminal kinase or p38 in transient transfection experiments (58,59). In control experiments, cortical neurons were transfected with wild type MKK1 DNA or the empty cloning vector. Neurons were also co-transfected with an expression vector encoding ␤-galactosidase (CMV-␤-Gal) as a marker for transfection (Fig. 7, A, B, D, and  E). The expression of wild type or constitutively active MKK1 was also directly confirmed by anti-HA immunostaining, taking advantage of the fact that the MKK1 expression vectors were tagged with an HA epitope (Fig. 7, C and F). In MKK1 transfected plates, transfected cells co-expressed ␤-galactosidase and MKK1 (compare Fig. 7, B and C and E and F). To assess the relative activities of the transfected wild type and constitutively active MKK1 for activation of endogenous ERK, the neurons were immunostained using the anti-phospho-ERK antibody and analyzed by confocal fluorescence microscopy. Phosphorylation of endogenous ERK was slightly increased in cells expressing the wild type MKK1 but greatly increased in cells expressing the constitutively active MKK1 (data not shown). These data indicate that expression of the constitutive active MKK1 activated endogenous ERK.
Cortical neurons were treated with camptothecin 2 days after transfection, and apoptosis in the transfected cell population was scored. Expression of the wild type MKK1 had no significant effect on cortical neuron apoptosis before or after camptothecin treatment (Fig. 8). However, expression of the constitutively active MKK1 markedly reduced apoptosis after camptothecin treatment. For example, after treatment with 10 M camptothecin for 24 h, 62% of the neurons transfected with the empty cloning vector or wild type MKK1 underwent apoptosis (Figs. 7D and 8B). However, only 28% of neurons transfected with 4 g of constitutively active MKK1 DNA were apoptotic (Fig. 8B), although many untransfected cells in the same field were clearly apoptotic (Fig. 7, E and F). The protective effect of constitutively active MKK1 depended on the amount of plasmid DNA used for transfection (Fig. 8B). These results suggest that selective and constitutive activation of ERK is sufficient to protect cortical neurons from camptothecin-induced apoptosis. Collectively, the above data support the hypothesis that activation of both the ERK and the PI 3-kinase pathways is required for full neuroprotection of BDNF against camptothecin toxicity. However, the ERK pathway is more important than the PI 3-kinase pathway for BDNF protection against camptothecin.

The ERK Signaling Pathway Is Activated during Camptothecin-induced Apoptosis and May Be a Cell-intrinsic Survival
Pathway to Counteract Stress-The activities of ERK and PI 3-kinase during camptothecin-induced apoptosis were analyzed by Western analysis using the phospho-ERK and phospho-Akt antibodies, respectively. Akt phosphorylation in cortical neurons was unaffected by camptothecin treatment (data not shown), suggesting that camptothecin does not alter PI 3-kinase activity. Because ERK activity in PC12 cells is inhibited during NGF withdrawal-induced apoptosis (29), one might expect a similar decrease when cortical neurons are treated with camptothecin. Surprisingly, camptothecin treatment of cortical neurons increased ERK phosphorylation, indicative of ERK activation (Fig. 9A). This was confirmed using an immune complex kinase assay for ERK activity (Fig. 9B). Activation of ERK was apparent by 6 h (Fig. 9B) and was dependent on the concentration of camptothecin, reaching maximum stimulation at 5-10 M (Fig. 10A). Although ERK was activated only 2-fold, its activation persisted for at least 20 h. Furthermore, the peak of ERK activation (12 h) preceded the peak of the morphological changes associated with apoptosis (24 -48 h).
Does activation of endogenous ERK by camptothecin contribute to apoptosis or is it a protective mechanism to counteract apoptotic signals? To distinguish between these two possibilities, we measured the effect of ERK inhibition on camptothecin-induced apoptosis using PD98059. If activation of ERK contributes to the induction of apoptosis, pre-treatment with PD98059 should reduce camptothecin-induced apoptosis. Conversely, if ERK is activated as a protective mechanism to antagonize stress, pre-treatment with PD98059 should enhance camptothecin-induced apoptosis. To ensure that PD98059 inhibited camptothecin-induced ERK activation, cortical neurons were treated with various concentrations of camptothecin for 12 h in the absence or presence of 40 M PD98059, and ERK activity was measured by the immune complex kinase assay (Fig. 10A). PD98059 antagonized ERK activation by camptothecin at all concentrations examined. The effect of PD98059 on camptothecin-induced cortical neuron cell death was examined at several camptothecin concentrations (Fig. 10B). The addition of PD98059 reduced cortical neuron survival after camptothecin treatment (p Ͻ 0.0001). For example, although 52% of the cells were still alive after 5 M camptothecin treatment for 48 h, only 30% of the cells were viable when PD98059 was also present. Similar results were obtained when apoptosis was monitored by morphological changes after Hoechst staining (data not shown). These data support the hypothesis that ERK is a cell-intrinsic survival signaling pathway that is activated in response to treatment with camptothecin to counteract the apoptotic signal.
The PI 3-Kinase Pathway Is Necessary and Sufficient for Serum-mediated Cortical Neuron Survival as well as BDNF Neuroprotection against Serum Deprivation-Because inhibition of the PI 3-kinase by LY294002 significantly increased basal apoptosis when neurons were maintained under normal culture conditions in the presence of 10% serum (Fig. 4B), we investigated the contributions of the ERK and the PI 3-kinase pathways to serum-promoted survival of cortical neurons. Cortical neurons were treated with several concentrations of LY294002 for various times and assayed for survival by MTT metabolism. Inhibition of the PI 3-kinase reduced cortical neuron survival in a dose-and time-dependent manner (Fig. 11A). These neurons were also treated with PD98059, LY294002, or FIG. 8. Expression of constitutively active MKK1 protects cortical neurons from camptothecin-induced apoptosis. Cortical neurons (DIV 3) were co-transfected with 0, 2, or 4 g of plasmid DNA encoding either wild type or constitutively active MKK1. All cells were also co-transfected with 1 g of plasmid DNA encoding ␤-galactosidase as a marker for transfection. Empty vector pcDNA3 was used to supplement the total DNA to 5 g in each case. Two days after transfection, one-half of the cortical neurons from one plate were fixed (A), and the other half from the same plate were treated with 10 M camptothecin for 24 h and then fixed (B). Neurons were then immunostained with antibodies to detect expression of ␤-galactosidase or MKK1 as described in Fig. 7. Apoptosis in the transfected cell populations (␤-galactosidasestained cells) were scored. Data are the averages of duplicate determinations. At least 350 transfected cells were scored for each condition. Similar results were obtained in four independent experiments; error bars ϭ S.E.

FIG. 9. Camptothecin activates ERK in cortical neurons. A,
Western analysis using anti-phospho-ERK antibody. Cortical neurons (DIV 5) were treated with 5 M camptothecin for the indicated times. Cell lysates were prepared, and 20 g of total protein were used for Western analysis employing antibodies recognizing either phosphorylated (p-) ERK1/2 (top) or unphosphorylated ERK 2 (bottom). Increased phosphorylation of ERK 1/2 after camptothecin treatment indicates ERK activation. Anti-ERK-2 Western was used to confirm an equal amount of protein loading in each gel lane. Similar results were obtained with three independent experiments. B, kinetics of ERK activation by camptothecin measured by immune complex kinase assay. Cortical neurons (DIV 5) were treated with 5 M camptothecin for the indicated times. Cell lysates were prepared, and 300 g of total protein were used for immune complex kinase assay. Data shown are representatives of three independent experiments. both and were assayed for apoptosis. PD98059 alone did not have much effect on cortical neuron survival nor did it potentiate the toxicity of LY294002 (Fig. 11B). Cortical neurons were also transfected with a dominant interfering form of the p85 subunit to selectively inhibit PI 3-kinase (60). Expression of dominant interfering p85 induced apoptosis, even in the presence of serum (Fig. 11C). This suggests that under normal culture conditions (10% serum), the PI 3-kinase activity, not ERK, is important for serum-promoted cortical neuron survival.
Can BDNF protect cortical neurons from serum withdrawal, and what are the relative contributions of the ERK and the PI 3-kinase pathways for this process? Conditions for induction of apoptosis in cerebellar or embryonic cortical neurons have been published (38,39). We discovered that postnatal cortical neurons cultured on glass coverslips also undergo apoptosis after serum withdrawal; phenotypic changes characteristic of apop-tosis included nuclei fragmentation and condensation (Fig. 12). BDNF completely protected postnatal cortical neurons from serum withdrawal. LY294002, but not PD98059, greatly inhibited the neuroprotection afforded by BDNF. Moreover, LY294002 and PD98059 together did not inhibit BDNF neuroprotection more than LY294002 alone (data not shown).
To test whether selective and direct activation of the PI 3-kinase pathway is sufficient to promote cortical neuron survival after serum deprivation, we transiently transfected cortical neurons with DNA encoding a wild type (p110wt) or constitutively active (p110*) catalytic subunit of PI 3-kinase (61). The cloning vector and a kinase dead-mutant form of p110 were used as negative controls. Expression of these constructs did not affect apoptosis when cortical neurons were maintained under normal culture conditions containing serum (data not shown). However, expression of the wild type or the constitutively active catalytic subunit of the PI 3-kinase protected cortical neurons from serum deprivation (Fig. 13A). In contrast, expression of a wild type or constitutively active MKK1 had no effect (Fig. 13B). This indicates that the PI 3-kinase pathway, but not the ERK, is the primary signal transduction system that mediates the neuroprotective effect of BDNF against serum deprivation.

DISCUSSION
The objective of this study was to define the relative contribution of the ERK and the PI 3-kinase pathways to the protection of CNS neurons from different forms of cellular stress. Because CNS neurons have unique cellular and biochemical features that distinguish them from non-CNS neurons, it is important to identify apoptotic mechanisms for CNS neurons. Furthermore, it is likely that several anti-apoptotic mechanisms are employed for the protection of CNS neurons with specificity for the apoptotic signal and type of neurons. One of the obstacles impeding the study of apoptosis in the CNS has been the difficulty in transfecting post-mitotic CNS neurons with high efficiency and low toxicity. The modified calcium phosphate transfection method used in this study allowed us to transfect neurons in culture with specific genes of interest, e.g. constitutively active MKK1 or PI 3-kinase.
The data presented here indicate that camptothecin treatment or serum deprivation, two distinct forms of stress, induce cell death in cortical neurons with phenotypes characteristic of apoptosis. BDNF protected cortical neurons from both stimuli, albeit by different mechanisms. Both the ERK and the PI 3-kinase signal transduction systems were activated by BDNF in cortical neurons. Blocking the ERK pathway using the pharmacological agent PD98059 inhibited BDNF neuroprotection against camptothecin but not serum deprivation. Furthermore, selective and constitutive activation of ERK by transient expression of a constitutive active MKK1 protected cortical neurons from apoptosis induced by camptothecin but not serum deprivation. Although drugs and transient expression experiments have the potential to exhibit nonspecific effects, both approaches gave similar results in our study, strengthening our conclusion that ERK is important for BDNF protection against camptothecin. In contrast, inhibition of the PI 3-kinase pathway by LY294002 was very effective in reversing the neuroprotection of BDNF against serum deprivation but not camptothecin treatment. Expression of a constitutive active PI 3-kinase protected cortical neurons from apoptosis induced by serum deprivation but not camptothecin treatment. Moreover, inhibition of PI 3-kinase by LY294002 or expression of a dominant interfering form of PI 3-kinase was sufficient to induce apoptosis even when cortical neurons were maintained under normal culture conditions in the presence of serum. These data suggest that the ERK pathway is primarily responsible for BDNF neuroprotection against the DNA-damaging agent camptothecin, whereas activation of the PI 3-kinase pathway contributes to BDNF neuroprotection against serum withdrawal and serum-promoted cortical neuron survival.
These data demonstrate that multiple survival pathways are used in cortical neurons to counteract different forms of apoptotic signals, which probably occurs because different insults activate distinct biochemical pathways. For example, SCG neuron apoptosis induced by NGF withdrawal or DNA-damaging agents requires induction of different caspases as well as cyclindependent kinases (43,44,62,63). Our data also emphasize the importance of defining apoptotic mechanisms for specific types of neuron, because neurons from different regions of the brain have distinct properties. Although activation of the ERK pathway does not contribute to the neuroprotective effect of NGF for SCG or BDNF in cultured cerebellar neurons (20,(35)(36)(37), our results demonstrate that it is neuroprotective for camptoth-Empty cloning vector for ␦-p85 was used to supplement the total DNA to 6 g in each case. Cells were fixed at 24 and 48 h post-transfection and immunostained with anti-␤-galactosidase antibody to identify transfected cells. Apoptosis in transfected cell population was scored. Data shown are the averages of duplicate determinations. Similar results were obtained in three independent experiments; error bars indicate S.E.  ecin-treated cortical neurons. Although activation of the PI 3-kinase/Akt pathway has been suggested to play a major role in neuroprotection by neurotrophic factors in cerebellar and SCG neurons (17)(18)(19)(20)23), it has only a small effect on BDNFmediated neuroprotection against camptothecin in cortical neurons.
We also discovered that ERK was activated, rather than inhibited, when cortical neurons were treated with camptothecin and that inhibition of ERK activation by PD98059 further increased camptothecin-induced apoptosis. These findings suggest that ERK does not actively contribute to apoptosis but may be activated as a neuroprotective mechanism. Although the extent of ERK activation by camptothecin was small (2-fold), ERK activation lasted at least 24 h. This prolonged ERK activation may be critical for neuron survival because it provides a long window of opportunity for the activated ERK to transmit signals to downstream targets. Our data suggest the interesting possibility that ERK may function generally as a cellintrinsic survival pathway in neurons and that this mechanism may apply to other forms of apoptosis in CNS neurons and in non-neuronal cells. For example, ERK is activated by H 2 O 2 and low doses of radiation in several cell lines and may provide protection against these stimuli (26,28). Furthermore, ERK may be only one of several cellular self-defense mechanisms that are mobilized to combat stress. Another example is the activation of the NF-B pathway in response to several forms of stress (64 -66).
The use of ERK as a cell survival mechanism in neurons is also interesting because ERK plays a pivotal role for other neuronal functions. For example, the ERK/MAP kinase signal transduction pathway is activated and required for transcriptionally dependent long-term potentiation as well as Ca 2ϩ stimulation of the cAMP regulatory element transcriptional pathway in CNS neurons (67). The activation of ERK and cAMP regulatory element-dependent transcription have been implicated in growth, differentiation, and neuroplasticity (46,67,68). The dual function of ERK as a survival pathway and to mediate activity-dependent processes ensures the viability of neurons contributing to synaptic plasticity, and it may be an important mechanism for activity-dependent maintenance of neuron populations during development.
In summary, the Erk and PI 3-kinase signaling pathways differentially mediate BDNF neuroprotection against camptothecin and serum deprivation. Our data suggest that signaling pathways that mediate neuroprotection are both stimulus-and cell-type specific. FIG. 13. Direct activation of PI 3-kinase, but not ERK, protects cortical neurons from serum deprivation-induced apoptosis. A, transient expression of a wild type (wt) or constitutively active p110 subunit of PI 3-kinase is sufficient to protect cortical neurons against serum deprivation-induced apoptosis. Cortical neurons (DIV 3) were transfected with 4 g of plasmid DNA encoding vector control, a wild type p110, a constitutively active p110 (p110*), or a kinase-dead mutant (KIN) of p110*. All cells were also co-transfected with 2 g of plasmid DNA encoding ␤-galactosidase as a marker for transfection. Two days after transfection neurons were deprived of serum for 24 h, fixed, and immunostained with anti-␤-galactosidase antibody to identify transfected cells. Apoptosis in the transfected cell populations were scored. B, transient expression of a constitutively active (ca) MKK1, which activates ERK, has no effect on cortical neuron survival after serum deprivation. Cortical neurons (DIV 3) were transfected and treated as in Panel A, except 4 g of plasmid DNA encoding vector control, a wild type, or a constitutively active MKK1 were used instead of p110 plasmid DNA. Data shown are the averages of duplicate determinations. Similar results were obtained in two independent experiments; error bars indicate S.E. values.