Differential Effects of cAMP in Neurons and Astrocytes

Mitogen-activated protein kinase (MAPK) activation provides cell type-specific signals important for cellular differentiation, proliferation, and survival. Cyclic AMP (cAMP) has divergent effects on MAPK activity depending on whether signaling is through Ras/Raf-1 or Rap1/B-raf. We found that central nervous system-derived neurons, but not astrocytes, express B-raf. In neurons, cAMP activated MAPK in a Rap1/B-raf-dependent manner, while in astrocytes, cAMP decreased MAPK activity. Inhibition of MAPK in neurons decreased neuronal growth factor-mediated survival, and activation of MAPK by cAMP analogues rescued neurons from death. Furthermore, constitutive expression of B-raf in astrocytoma cells increased MAPK activation, as seen in neurons, and enhanced proliferation. These data provide the first experimental evidence that B-raf is the molecular switch which dominantly permits differential cAMP-dependent regulation of MAPK in neuronsversus astrocytes, with important implications for both survival and proliferation.

Recent studies have demonstrated that cyclic AMP (cAMP) provides a powerful survival signal for neurons. The short-term survival of spinal motor neurons in vitro is greatly enhanced by elevated intracellular cAMP (1). In the absence of peptide growth factors, the majority of motor neurons extended processes and survived for 1 week in response to cAMP, while the combination of multiple peptide trophic factors with cAMP elevation extended neuronal survival in serum-free media for as much as 3 weeks. Similarly, survival of superior cervical ganglion neurons (2)(3)(4) and cerebellar granule cells (5,6) can be supported by increasing cAMP levels. However, the inability of phosphatidylinositol 3-kinase (PI3K) 1 inhibitors to block the pro-survival effects of cAMP in both neuronal types suggests that cAMP may promote neuronal survival by additional non-PI3K/protein kinase B (Akt) mechanisms (4,6). cAMP elevation can increase recruitment of the trkB receptor to the plasma membrane in retinal ganglion cells (7), suggesting that cAMP may promote neuronal survival, in part, by increasing neurotrophin receptor availability and signaling.
The traditional receptor tyrosine kinase signaling pathway activates mitogen-activated protein kinase (MAPK) by Ras-dependent recruitment of Raf-1, which subsequently phosphorylates MAPK kinase (MEK) and results in activation of MAPK (8,9). However, an alternative pathway was recently described in PC12 cells, which preferentially utilizes Rap1 and 95-kDa Braf, instead of Ras/Raf-1, to activate MEK (10). These two pathways, Ras/Raf-1 and Rap1/B-raf, differ in their response to cAMP, with cAMP inhibiting MAPK when signaling is through Ras/Raf-1, and activating MAPK when signaling is through Rap1/B-raf (10). This divergent signaling has been reported to reflect cAMP regulation of Rap1 activity and selective interaction of Rap1, instead of Ras, with 95-kDa B-raf (11). Although cAMP effects on Rap1 were attributed to activation of protein kinase A (PKA) in that study, a recent report suggests that cAMP may have non-PKA-dependent actions on Rap1 mediated by cAMP binding and guanine nucleotide exchange factors (cAMP-GEFs) (12).
B-raf mRNA has been detected in brain and spinal cord (13), but the expression of B-raf protein in specific central nervous system cell types has not been determined. We employed cultured primary cortical astrocytes and neurons, astrocytoma cells, and a hypothalamic neuronal cell line to assess expression of B-raf protein in neurons and astrocytes and then compared MAPK activity in these cell types in response to cAMP. Although Rap1 is clearly required for cAMP-dependent inhibition of Ras-mediated MAPK activation (11), the question of whether Rap1 activation is sufficient to inhibit Ras signaling has not been established. In this study, we tested the hypothesis that B-raf is a cell-type-specific molecular switch capable of determining whether cAMP has a stimulatory or inhibitory influence on MAPK activity in central nervous system parenchymal cells, and furthermore, that the Rap1/B-raf response to cAMP is dominant over that of Ras/Raf-1. We generated astrocytoma cell lines expressing various mutant forms of Rap1 and measured MAPK activity to ascertain whether Rap1 directly inhibited MAPK activity. Furthermore, in light of the recently identified cAMP-GEFs, which appear to mediate many cAMP effects formerly attributed to activation of PKA (12), we asked whether cAMP effects on MAPK were reversed by inhibition of PKA. Finally, to determine whether B-raf expression alone was sufficient to mediate divergent MAPK responses to cAMP in neurons and astrocytes, we generated B-raf-expressing astrocytoma cells and evaluated their response to cAMP.
In this report, we characterized the Ras/Raf-1 and Rap1/Braf signaling pathways in neurons and astrocytes, and evaluated the effect of cAMP, acting through these two pathways, on MAPK activity. We then determined the functional implications of cAMP-regulated MAPK signaling for survival and proliferation of neurons and astrocytes.

EXPERIMENTAL PROCEDURES
Materials and Reagents-Unless otherwise specified, drugs and chemicals were obtained from Sigma, and cell culture supplies were purchased from standard suppliers, e.g. Falcon, Life Technologies, Inc., and Hyclone.
Preparation of Primary Cortical Astrocyte and Neuronal Cell Cultures, and GT1-1 trk Cell Cultures-Primary murine neuronal and astrocyte cultures were prepared as described (14). Neuronal cultures were prepared from neocortex of embryonic day 15 Swiss-Webster mice, and plated on Primaria plates previously coated with poly-D-lysine: laminin. Pure neuronal cultures contained Ͻ0.3% glial fibrillary-associated protein positive cells, and were used after 12 days in culture. Astrocytes were derived from neocortex of postnatal day 1-2 mouse pups and were used when confluent. These cultures had rare (Ͻ0.1%) oligodendroglial cells. GT1-1 trk cells expressing trk protein (GT1-1 trk) (15) were maintained in passage in DMEM with 10% horse serum, containing penicillin-streptomycin. C6 glioma-derived cells and U373 astrocytoma cells were originally obtained from the ATCC and were maintained in passage until use.
Generation of a Stable C6 Line Expressing 95-kDa B-Raf-C6 cells were transfected with either pcDNA3 vector alone or B-raf (kindly provided by Deborah Morrison, NCI) cloned into pcDNA3 using Lipo-fectAMINE (Life Technologies, Inc.) according to published protocols. C6 cells were selected in 500 g/ml active geneticin and individual clones selected for B-raf expression by Western blot analysis. Two B-raf expressing clones were identified. Clone 1 expressed higher levels of B-raf and therefore was used for further experiments.
Rap1 Activity Assay-C6 cells were grown to 70% confluence and serum starved for 24 h prior to stimulation with 250 M CPT-cAMP for 15 min prior to direct harvest in MAPK lysis buffer on ice. Lysates were incubated with purified GST-ral-GBD kindly provided by Dr. J. L. Bos and used according to published protocols (16). Bound Rap1 was eluted directly in 1 ϫ Laemmli buffer and separated by 15% SDS-PAGE prior to immunoblotting with monoclonal Rap1 antibodies (Transduction Laboratories).
Cell Proliferation Analysis for Astrocytes and Astrocytoma Cells- Analysis of GT1-1 trk Cell Survival-GT1-1 trk cells were plated at 250,000 cells/well in 12-well plates. One day later, cells (3 wells per condition) were counted to obtain the zero hour time point, and the medium was exchanged with DMEM containing serum. PD 98059 was included in appropriate wells to give a 30-min pretreatment. After 30 min, the medium was changed to serum-free DMEM, and the following drugs were added (in DMEM without neuronal growth factor, NGF): no NGF (DMEM only), NGF (100 ng/ml), NGF ϩ PD 98059 (50 M), dibutyryl-cAMP (250 M), or 8-bromo-cAMP (250 M). At 48 h, cells were stained with trypan blue, treated with trypsin, dissociated in a uniform volume of medium, and placed on a hemocytometer. Trypannegative cells in a standard volume were counted to give the number of surviving cells.

B-raf Is Present in Neuronal but Not Glial Cells-We exam-
ined expression of 95-kDa B-raf protein in cells of neuronal and glial lineage. B-raf exists as a number of isoforms that are reported to be expressed primarily in neural and endocrine tissues. Although the 95-kDa B-raf isoform is found in both brain and spinal cord (13), its location in specific central nervous system cell types has not been systematically explored. We measured B-raf protein expression in primary cultures of cortical astrocytes (99% type I astrocytes, pure glial cultures) and neurons (99.5% neurons, pure neuronal cultures) using Western immunoblotting with an antibody specific for B-raf (Fig.  1A). We also assessed B-raf expression in C6 astrocytoma cells, and in the GT1-1 trk neuronal cell line (Fig. 1B) by immunoprecipitation with anti-B-raf antibody. GT1-1 trk cells are immortalized mouse hypothalamic neurons expressing trk protein which have many characteristics of differentiated neurons, including expression of synaptic vesicle proteins, neuronal ␥-aminobutyric acid receptors, neuronal sodium channels, and neurosecretory proteins (15,17). These cells demonstrate neurite extension in response to NGF (15). Both primary neurons FIG. 1. B-raf protein is present in neurons but not astrocytes. Immunoblots (A) using 80 g of protein extracted from primary cultures of cortical astrocytes (pure glial cultures, PGC) or cortical neurons (pure neuronal cultures, PNC) demonstrate the presence of the 95-kDa B-raf protein in neurons, but not astrocytes. Blots were probed with an anti-B-raf antibody (Santa Cruz Biotechnology) which recognizes both the 62-and 95-kDa B-raf isoforms. A second band at 62 kDa in cortical neurons may represent the second B-raf isoform. Equal protein loading was confirmed by Ponceau S staining, and in some experiments by re-probing membranes with anti-tubulin antibody (not shown). Immunoprecipitation followed by immunoblotting (B) of proteins extracted from C6 glioma cells (C6) or GT1-1 trk cells demonstrate 95-kDa B-raf protein in the neuronal GT1-1 trk cells, but not the astrocytoma C6 cell line. Precipitations were performed with 200 g of protein/sample; bands at the bottom represent IgG. and the neuronal cell line expressed the 95-kDa B-raf isoform. In contrast, no B-raf protein was detected in primary cortical astrocytes (Fig. 1A), C6 astrocytoma cells (Fig. 1B), or U373 astrocytoma cells (data not shown). Cortical neurons appear to have small amounts of the 62-kDa B-raf splice variant (13), as well (Fig. 1A). These results suggest that only neurons, but not astrocyte-derived cells, express the 95-kDa B-raf isoform.
cAMP Stimulates MAPK Phosphorylation in Neuronal Cells-To determine the consequences of elevated intracellular cAMP on neuronal MAPK activity, we measured phospho-MAPK using an antibody specific for the active phosphotyrosine and phosphothreonine MAPK. In both cortical neurons and GT1-1 trk neuronal cells, we found that cAMP increased MAPK phosphorylation (Fig. 2, A and B). This effect was dosedependent in both primary neurons and GT1-1 trk cells. Equal amounts of protein loading were verified using the tubulin monoclonal antibody. As depicted in Scheme 1, one model proposes that in the presence of B-raf and Rap1, cAMP stimulates MAPK activation. In addition, it is possible that Rap1/B-raf may be able to override cAMP/Rap1-dependent inhibition of MAPK through the Ras/Raf-1 system.
Inhibition of MAPK Decreased NGF-mediated Survival of GT1-1 trk Cells, and cAMP-dependent Activation of MAPK Promotes GT1-1 trk Cell Survival in the Absence of NGF-GT1-1 trk cells can be supported by NGF for several days in the absence of serum, and removal of NGF will induce progressive loss of these cells by apoptosis. In many neuronal cell types, PI3K activity appears to mediate neurotrophin effects on cell survival (18). However, we found that MAPK inhibition in GT1-1 trk neurons, even in the presence of NGF, resulted in significant neuronal cell death (Fig. 3). Activation of MAPK by cAMP analogues, conversely, rescued cells from death induced by withdrawal of trophic support (Fig. 3). cAMP analogues, however, did not block GT1-1 trk cell death induced by the MEK inhibitor, PD 98059, and could not further improve neuronal survival produced by treatment with NGF (data not shown). The PKA inhibitor, HA-89 (10 M), also did not alter NGF-mediated cell survival (data not shown). These results suggest that cAMP promotes neuronal survival specifically through activation of MAPK; cAMP cannot rescue cells if MEK is inhibited directly, and cAMP confers no added survival advantage when MEK is maximally stimulated by NGF.
cAMP Blocks MAPK Phosphorylation in C6 Glioma Cells and Primary Astrocytes-Ras-dependent signaling activates MAPK, but can be blocked by cAMP-dependent activation of Rap1 (Scheme 1). In cells lacking B-raf, therefore, cAMP should decrease MAPK activity through a negative Rap1 effect on Ras signaling. Consistent with this idea, we found that all three cAMP analogues tested (CPT-cAMP, Bt 2 cAMP, and 8-bromo-cAMP) caused a dose-dependent reduction in MAPK activity in both C6 glioma cells (Fig. 4A) and cortical astrocytes (Fig. 4B). Inhibition of MAPK phosphorylation by cAMP was less effective in astrocyte than in astrocytoma cells (Fig. 4B), although direct inhibition of MEK by PD 98059 in astrocytes was still able to completely block MAPK phosphorylation (Fig. 4B). Astrocytes demonstrated little MAPK activity in the presence of serum, indicating either minimal Ras activity in the presence of serum, or inhibition of Ras-dependent signaling by factors, such as cAMP, stimulated by serum. The latter appears to be likely because serum-dependent suppression of MAPK phosphorylation was reversed by the PKA inhibitor, HA120 (Fig. 4B).
Rap1 Is Activated by cAMP-Several groups have reported that Rap1 can be activated by cAMP, either through specific cAMP-binding proteins (12) or through presumed activation of PKA (11). Using an activity assay for Rap1, we looked at the effect of the cAMP analogue, CPT-cAMP (250 M) on Rap1 activity in C6 cells (Fig. 5A). This assay was performed using the Ral-GBD glutathione S-transferase fusion protein pulldown technique developed by Bos and colleagues (16). In the absence of serum, C6 cells had a low basal level of Rap1 activity. Rap1 activity increased after treatment with CPT-cAMP for 15 min, confirming that cAMP activates Rap1.
Proliferation of U373 Astrocytoma Cells Is Enhanced by Activation of MAPK: Regulation by Rap1-Since Rap1 is activated by cAMP, we next tested the hypothesis that the cAMP effects on MAPK activity and astrocyte/astrocytoma cell proliferation were the result of Rap1 activation. We established U373 astrocytoma cell lines overexpressing wild-type (WT) and mutant Rap1 proteins (Fig. 5B). Clones overexpressing WT Rap1, dominant-negative N17 Rap1, or the G12V constitutive Rap1 mutant were generated and several independently isolated clones analyzed for Rap1b expression by Western blot (Fig. 5B). All clones demonstrated elevated Rap1 expression compared with vector-transfected control lines. To determine how overexpression of WT or mutant Rap1 would affect astrocytoma MAPK activity and proliferation, these cell lines were analyzed by phospho-MAPK Western blot and direct cell counting (Fig. 5B). Basal levels of MAPK activity were similar in U373 cells containing vector, oncogenic Rap1 (G12V mutant) and WT Rap1. However, increased MAPK activation was observed only in clones containing the N17 dominant-negative Rap1 mutation. These results suggest that Rap1 suppresses and GT1-1 trk cells (B) demonstrate increased MAPK activity in response to cAMP analogues in a dose-dependent manner. After cultures were exposed to cAMP analogues CPT-cAMP, Bt 2 cAMP, or 8-bromo-cAMP for 15 min, proteins were extracted, and lysates containing 80 g of total protein were evaluated by Western blotting using antibody specific for phospho-MAPK (Promega). Epidermal growth factor-stimulated phosphorylation of MAPK served as a positive control. SCHEME 1 MAPK activity (Scheme 1). Previous experiments from our laboratory and others have shown that activated MAPK provides a mitogenic signal for astrocytoma cells (19,20). To determine whether N17 Rap1-induced MAPK activation would result in increased cell proliferation, U373 cell lines expressing Rap1 were counted as a function of time after seeding. In dominant-negative Rap1 clones, increased cell proliferation correlated with increased MAPK activity (Fig. 5C), directly demonstrating that Rap1 functions to inhibit MAPK activation and cell proliferation in astrocytic tumor cells.
Proliferation of Astrocytes and Astrocytoma Cells Is Regu-lated by cAMP Analogue Treatment-To determine the effect of cAMP down-regulation of MAPK activity on astrocyte-like cell proliferation, we measured [ 3 H]thymidine incorporation in astrocyte and astrocytoma cultures exposed to treatments shown to modify MAPK activity. Consistent with our prior observation that Bt 2 cAMP decreased MAPK activity, and HA120 increased MAPK activation, C6 cell proliferation was inhibited by Bt 2 cAMP, and enhanced by the PKA inhibitor (Fig. 6A). PD 98059 only blocked proliferation of astrocytes (Fig. 6B) and astrocytoma cells (Fig. 6A) to about 50% of control levels, suggesting that MAPK is an important, but not the sole proproliferative signal in these cells. Regulation of MAPK activity by cAMP, therefore, has functional consequences for proliferation of astrocyte-derived cells.

MAPK Phosphorylation in C6 Astrocytoma Cells Transfected with B-Raf: Conversion to a Neuron-like Response-As
Rap1 appears to mediate the cAMP growth inhibitory signal for astrocytes, we wanted to determine whether the critical difference in Rap1/cAMP signaling in astrocytes versus neurons was dependent on the presence or absence of B-raf. For these experiments, we generated stable C6 clones expressing the 95-kDa isoform of B-raf. Two stable clones (1 and 4) were generated; clone 1 had higher B-raf expression and was used for further experiments (Fig. 7A). Basal levels of MAPK activity were compared in clone 1 versus cells transfected with pcDNA.3 vector (Fig. 7B). The B-raf clone had high basal levels of phospho-MAPK, while no phospho-MAPK was detectable in the vector control. Treatment with the PKA inhibitor, HA120, increased phospho-MAPK in the vector clone, but decreased phospho-MAPK levels in the B-raf clone. Thus, introduction of B-raf protein alone is sufficient to convert cAMP from a negative to a positive regulator of MAPK activation.
Transfection of C6 Cells with B-raf Enhances Proliferation-Using [ 3 H]thymidine incorporation, we next determined whether the increased levels of MAPK activity resulting from B-raf overexpression were associated with increased C6 cell proliferation (Fig. 7C). The B-raf clone, which we found to have elevated levels of phospho-MAPK compared with the vector clone (Fig. 7B), had increased proliferation in the presence of serum. Removal of serum decreased proliferation to the same degree in both clones. These data indicate that one functional FIG. 4

. Effects of cAMP analogues, PKA inhibition, or MEK inhibition on MAPK activation in astrocytic and neuronal cells.
Immunoblots for phospho-MAPK in C6 cells (A) or cortical astrocytes (B) show a dose-dependent decrease in MAPK activation in response to cAMP analogues. After cultures were exposed to cAMP analogues CPT-cAMP, Bt 2 cAMP, or 8-bromo-cAMP for 15 min, proteins were extracted, and lysates containing 80 g of total protein were separated by SDS-PAGE for Western blotting using antibody specific for phospho-MAPK (Promega). cAMP analogues were less effective at decreasing MAPK phosphorylation in primary astrocytes than in astrocytoma cells. The MEK inhibitor PD 98059 (PD) blocked MAPK phosphorylation, and serum-induced inhibition of MAPK phosphorylation in astrocytes was reversed by the PKA inhibitor, HA120 (30 M) (B).

FIG. 3. Chronic inhibition of MAPK blocks NGF-mediated survival of GT1-1 trk cells, and cAMP activation of MAPK rescues GT1-1 trk cells from NGF deprivation.
Survival of GT1-1 trk cells was assessed 48 h after treatment was initiated by staining cells with trypan blue, followed by blinded cell counts of live cells using a hemocytometer. Cells (A) were exposed to medium without NGF, with NGF alone, or with NGF ϩ PD 980598. PD 98059 (PD) substantially decreased the ability of NGF to support neuronal survival. Cultures deprived of NGF (B) were rescued by cAMP analogues, dibutyryl-cAMP, and 8-bromo-cAMP (250 mM). Values are mean Ϯ S.E., expressed as the percentage of the ϩNGF condition. *, p Ͻ 0.05 versus the NGF condition, and **, p Ͻ 0.05 versus NGF plus PD 98059, by ANOVA and Tukey's test for multiple comparisons. outcome of the molecular switch provided by B-raf is increased astrocytoma cell proliferation.

Divergent Tyrosine Kinase Signaling Pathways in Neurons
and Astrocytes-Our data support the idea that neurons and astrocytes use fundamentally different pathways to mediate tyrosine kinase receptor signaling, as shown in Scheme 1. In this model, in the absence of cAMP, growth factors bind to their cognate tyrosine kinase receptors to recruit Ras and activate MAPK. However, when cAMP is present, Rap1 is recruited and activated by cAMP. In astrocytes and other cells lacking B-raf, Rap1 then directly inhibits Ras-dependent stimulation of MAPK phosphorylation. However, in neurons, which express B-raf, Rap1 activates B-raf, which then leads to MAPK phosphorylation. By activating MEK, B-raf is able to bypass upstream inhibition of Ras signaling by Rap1. Thus cAMP represents one of several potential second messenger systems which can initiate divergent tyrosine kinase signaling in neurons and astrocytes.
FIG. 5. cAMP-activated Rap1 (A) analysis of activated Rap1 in C6 cells was performed using ral-GBD affinity chromatography. Cells were exposed to MEM minus serum with or without CPT-cAMP (250 M) for 15 min followed by cell lysis and incubation of extracts with glutathione S-transferase-Ral-GBD coupled to glutathione-Sepharose beads. Eluted proteins were separated by SDS-PAGE and immunoblotted with Rap1 monoclonal antibodies (Transduction Laboratories). As active GTP-bound Rap1 preferentially binds to the Ral-GBD, the amount of Rap1 detected by Western blot is a direct reflection of the level of activated Rap1. In this representative experiment, low basal Rap1 activity was increased by application of the cAMP analog, CPT-cAMP. Modification of MAPK activity alters proliferation of astrocytic cells. Immunoblot analysis (B) of Rap1 expression in U373 cells transfected with vector (pcDNA3, clones 3 and 5), WT Rap1 (clones 3, 5, and 11), N17 dominant-negative Rap1 (clones 14 and 16), or constitutively active G12V Rap1 (clones 10 and 14) demonstrates increased Rap1 expression in all lines except those containing the pcDNA3 vector. All experiments were done in the presence of serum. Phosphorylation of MAPK activity, which is mitogenic for astrocytoma cells, was increased in the N17 dominant-negative Rap1 clones, detected by immunoblot analysis of protein lysates from the clonal lines (B). Increased MAPK activation correlated with increased cell proliferation (C). Ten thousand cells were seeded in quadruplicate in 24-well plates and harvested for direct cell counts using a hemocytometer after 48, 72, and 96 h in culture.
FIG. 6. Astrocyte and astrocytoma cultures were used at approximately 60% confluence to determine the effects of treatments on cell proliferation. Astrocytoma cell proliferation was attenuated by Bt 2 cAMP, which decreases MAPK activity, but was increased by the PKA inhibitor, HA120 (A). Direct inhibition of MAPK phosphorylation by PD 98059 decreased cell proliferation ([ 3 H]thymidine incorporation) in both astrocytoma (A) and astrocyte cells (B). Values are mean Ϯ S.E., n ϭ 6 for astrocytoma cells, n ϭ 8 for astrocytes. *, p Ͻ 0.05 by ANOVA and Student-Neuman-Keuls.
Although in the scheme, B-raf expression is critical for divergent tyrosine kinase signaling in response to cAMP, it is Rap1 that is activated by cAMP. Members of the highly conserved p21 Ras family of small GTPase proteins, including p21 Ras , Rac, Rho, and Rap1 are ubiquitously expressed signaling molecules (21,22) which contribute to regulation of cell survival, growth, cell fate determination, and differentiation (23). A recent report by Kawasaki et al. (12), suggested that in 293T cells, cAMP activates Rap1 by binding to cAMP-GEFs, not through activation of PKA. However, we demonstrated that CPT-cAMP activated Rap1, that Rap1 could directly regulate MAPK activity, and furthermore, that the PKA inhibitor, HA120, reversed inhibition of MAPK activity in astrocytes. This suggests that in certain cell types, such as astrocytes, PKA is involved in cAMP-induced inhibition of MAPK activity. Since Ras and Rap1 show specific preferences for the adapter proteins which promote their translocation to the plasma membrane, with Ras coordinating with Grb-SOS, and Rap1 binding CrkII-C3G (24), we speculate that interaction of Rap1 with its adapter proteins might influence whether cAMP activates Rap1 via cAMP-GEFs or PKA.
While Rap1 is ubiquitously expressed, B-Raf has a circumscribed pattern of expression. Barnier et al. (13) reported that the 95-kDa isoform of B-raf, derived from B-raf message containing exons 8 and 10, is found predominantly in brain and spinal cord. Our data suggest that B-raf expression may, in fact, be further restricted, because we found the 95-kDa protein only in neurons, but not astrocytes or astrocytoma cells. In Scheme 1, we propose that B-raf is both necessary and sufficient to permit Rap1 to induce MAPK activation. This hypothesis was supported by the observation that only cells expressing B-raf (neurons) had increased phospho-MAPK in response to cAMP. More compelling was the ability of B-raf transfection to convert the inhibitory action of cAMP on MAPK activity to a stimulatory action. Both parental C6 cells and cells transfected with pcDNA vector responded to cAMP with decreased MAPK phosphorylation. However, cells transfected with B-raf had increased MAPK phosphorylation in response to cAMP, suggesting that B-raf alone is sufficient to mediate the switch in signaling. It also indicates that B-raf effects are dominant, since the transfected cells possessed both intact Ras signaling and the newly introduced B-raf, yet the B-raf effect overrode Rap1 inhibition of Ras signaling. Although it is possible to argue that this result reflected overexpression of B-raf compared with Ras, results from neuronal cultures suggest this is not the case, as these cells also show dominance of B-raf effects even in the presence of physiological levels of both pathways. Moreover, in PC12 pheochromocytoma cells, the Rap1/B-raf pathway has been identified as the main signaling pathway from trk receptors to MAPK activation (11).
MAPK Regulation of Neuronal Survival-Although cAMP provides a powerful survival signal for certain neuronal populations (2,3,(25)(26)(27), the mechanism behind this capability has not been fully defined. MAPK has been implicated in the survival of sympathetic neurons exposed to cytosine arabinoside (28), and cerebellar granule cells exposed to low potassium conditions (29). While cAMP can recruit trkB protein to the plasma membrane to enhance neurotrophin signaling (7), recent work on superior cervical ganglion neurons (4) and cerebellar granule cells (6) suggests that cAMP promotes neuronal survival through at least one additional pathway. Survival of cerebellar granule neurons, which undergo programmed cell death when not maintained under depolarizing conditions (25 mM potassium), can be supported by both insulin growth factors and cAMP (5,6). Although insulin growth factor-dependent survival is blocked by the PI3K inhibitor, LY 294002, the pro-survival effects of cAMP were not altered by inhibition of PI3K (6). cAMP also supports survival of superior cervical ganglion neurons (3,4), and the cAMP-dependent survival in these cells is not blocked by inhibition of PI3K (4), suggesting that cAMP engages an alternative survival pathway from the PI3K/Akt pathway. We speculated that this second survival pathway might involve MAPK. We therefore evaluated the effect on MAPK inhibition on GT1-1 trk neuronal survival. Inhibition of MAPK partially reversed the ability of NGF to sustain survival of GT1-1 trk cells, and resulted in death of approximately half of the cells in the presence of NGF. While PD98059 is routinely used as a specific inhibitor of MEK, reported to have no effect on PI 3-kinase or other serine/threonine or tyrosine kinases, the possibility remains that it may act through mechanisms other than MEK inhibition. However, cAMP, which activated MAPK, was able to partially rescue GT1-1 trk cells from death induced by withdrawal of NGF, saving 30 -40% of cells. These data suggest that MAPK may complement PI3K (18,30) in mediating the pro-survival effects of NGF. The ability of cAMP to support a reasonable percentage of neurons in the absence of NGF provides further evidence FIG. 7. MAPK in C6 glioma cells transfected with B-raf shows a neuron-like response. Cell lines were generated with stable expression of 95-kDa B-raf or the pcDNA3 vector. One clone (number 1) showed moderate expression of B-raf protein (A), and was used for further studies. In the vector-containing clone, little basal MAPK activity was detected, but phospho-MAPK levels were increased by PKA inhibition (B). In contrast, the B-raf clone had a higher basal level of phospho-MAPK which was decreased by the PKA inhibitor. C6 cells transfected with B-raf, which have high basal MAPK activity, showed significantly greater proliferation (C) than the pcDNA vector control cells. [ 3 H]Thymidine incorporation was determined in the presence or absence of serum. Values are mean Ϯ S.E., *, p Ͻ 0.05 by ANOVA, Student-Neuman-Keuls, n ϭ 6.
for an alternative survival pathway involving MAPK. Taken together, these data suggest that MAPK, in addition to PI3K/ Akt, may contribute to growth factor-mediated survival in certain neuronal populations.
MAPK Regulation of Astrocyte and Astrocytoma Cell Proliferation-We noted that activation of MAPK, a mitogenic signal in astrocytic cells, was substantially enhanced in C6 cells expressing B-raf compared with vector-transfected controls. We therefore assessed proliferation of C6 cells transfected with B-raf or pcDNA vector, and found that the B-raf clone proliferated more rapidly than the vector clone. Inhibition of MAPK by PD 98059 or cAMP substantially decreased astrocyte and astrocytoma cell proliferation, while activation of MAPK after inhibition of PKA, in contrast, increased astrocytic cell proliferation. Thus, cAMP can regulate astrocyte and astrocytoma cell proliferation, and this effect appears to be mediated through modulation of MAPK activity. For most effects on signaling and cell proliferation, we found similar results between astrocytes and astrocytoma cells, with both astrocytoma and primary astrocyte proliferation blocked by PD 98059.
Finally, studies on Rap1 in the current study also addressed a question raised by previous work from our laboratory (31), which suggested that Rap1 may be regulated by tumor suppressor genes that contribute to the formation of astrocytomas. Individuals affected by the inherited predisposition to cancer syndrome tuberous sclerosis complex develop astrocytic tumors (31). The tuberous sclerosis complex 2 protein product, tuberin, has been shown to function as a negative regulator of Rap1 (32). Previous work from our laboratory (32,33) has demonstrated that loss of tuberin expression is observed in 30% of sporadic astrocytomas. This loss of tuberin expression in astrocytic tumors, could result in increased cell proliferation as a result of elevated Rap1 activation. However, results of the current study argue that increased Rap1 activity does not provide a proliferative signal for astrocytes and suggests that Rap1 activation in astrocytes and astrocytomas leads to decreased cell proliferation and MAPK signaling. It is therefore likely that tuberin functions as a growth regulator for astrocytes through mechanisms unrelated to Rap1 regulation (34).
MAPK as Integration Site for Cell Signaling-Our data support the idea that signal integration at MAPK may have important implications for neuronal survival and function. In addition to Ras-dependent signaling, MAPK may integrate signals coming from cell surface receptors via second messengers including not only cAMP, but protein kinase C (35-37), calcium/calmodulin kinase (38 -40), and arachidonic acid-derived eicosanoids (41,42) and other lipid mediators, such as platelet activating factor (43). These observations suggest the hypothesis that strategies which activate receptors that are negatively coupled to adenylyl cyclase, such as the M 2 and M 4 muscarinic acetylcholine receptors, might be detrimental under circumstances where trophic factor input is impaired.
In summary, our data suggests that: 1) B-raf mediates cAMPdependent MAPK activation in neurons; 2) MAPK activation is an important survival signal in the neurons studied, and may mediate the pro-survival effects previously observed for cAMP in neurons; 3) cAMP suppresses MAPK activation in astrocytes; 4) introduction of B-raf into astrocyte-like cells dominantly enables cAMP to activate MAPK; 5) activation of MAPK by cAMP enhances astrocyte proliferation; and 6) thus B-raf appears to be the molecular switch which causes differential regulation of MAPK in neurons versus astrocytes in response to cAMP.