Growth differentiation factor-15 prevents low potassium-induced cell death of cerebellar granule neurons by differential regulation of Akt and ERK pathways.

Growth differentiation factor-15 (GDF-15) is a novel member of the transforming growth factor-beta superfamily and has been shown to be induced in neurons subsequent to lesions. We have therefore begun to study its putative role in the regulation of neuron survival and apoptosis. Cultured cerebellar granule neurons (CGN) survive when maintained in high K(+) (25 mm) but undergo apoptosis when switched to low K(+) (5 mm). GDF-15 prevented death of CGN in low K(+). This effect could be blocked by phosphatidylinositol 3-kinase/Akt pathway inhibitors LY294002 or wortmannin. In contrast, mitogen-activated protein kinase (MEK)/extracellular-signal-regulated kinase (ERK) pathway inhibitors U0126 and PD98059 potentiated GDF-15 mediated survival and prevented cell death in low K(+) even without factor treatment. Immunoblots revealed GDF-15-induced phosphorylation of Akt and glycogen synthase kinase-3beta. This activation was suppressed by phosphatidylinositol 3-kinase inhibitors. Low K(+) induced delayed and persistent ERK activation, which was blocked by MEK inhibitors or GDF-15. ERK activation induced c-Jun, a member of the AP-1 transcription factor family. GDF-15 or U0126 prevented c-Jun activation. Furthermore, we show that GDF-15 prevented generation of reactive oxygen species, a known activator of ERK. Together, our data suggest that GDF-15 prevents apoptosis in CGN by activating Akt and inhibiting endogenously active ERK.

All TGF-␤ famliy members, except members of the GDNF subfamily, signal through heteromeric complexes of type I and type II serine/threonine kinases receptors. Within the type I and type II receptor families there are several subgroups, e.g. bone morphogenetic protein receptors, depending on their respective ligands (for reviews, see Refs. 7 and 8). GDNF, neurturin, persephin, and artemin signal through a heteromeric receptor complex consisting of the receptor tyrosine kinase Ret (9,10) and glycosylphosphatidylinositol-linked ␣-receptors (11).
We and others (12,13) have recently cloned a novel member of the TGF-␤ superfamily, GDF-15/macrophage inhibitory cytokine-1 (MIC-1). The protein does not belong to one of the known TGF-␤ subfamilies and represents a divergent member of the TGF-␤ superfamily. GDF-15 is widely synthesized in the central nervous system, most strongly expressed in the choroid plexus, and secreted into the cerebrospinal fluid, from where it may penetrate into the brain parenchyma (14 -16). We have previously shown survival promoting effects of GDF-15 on unlesioned and intoxicated midbrain dopaminergic neurons in vitro and in vivo (15). Serotonergic neurons of the embryonic rat raphe were one additional neuron population that responded to GDF-15 by an increase in transmitter synthesis and uptake. However, motoneurons of the spinal cord and sensory dorsal root ganglionic neurons did not respond at all or only very moderately to GDF-15. At this point, it is not clear whether other types of central nervous system neurons are affected by GDF-15. In a previous study, we had also shown that GDF-15 protein is up-regulated in cortical neurons subsequent to a cold lesion (16). This may indicate functions of the protein in neurons affected by injury and suggest a role of GDF-15 in the execution of either survival or cell death programs.
Cultured cerebellar granule neurons (CGN) from postnatal rat represent a highly homogeneous neuron population that provides an excellent system to study neuronal apoptosis. CGN survive for weeks in vitro and develop characteristics of mature CGN in vivo when maintained in depolarizing concentrations of K ϩ (25 mM), but undergo apoptosis when cultured in physiological low K ϩ (5 mM) conditions (17)(18)(19). Although mechanisms underlying CGN apoptosis are not clear as yet, a requirement of RNA/protein synthesis, generation of reactive oxygen species (ROS), activation of caspases, and phosphorylation of c-Jun have been implicated in this apoptotic model (18,20,21).
Our previous observation that GDF-15 is specifically upregulated in lesioned neurons prompted us to investigate proor anti-apoptotic effects of GDF-15 and underlying mechanisms in CGN. Here we demonstrate that (i) GDF-15 prevents death of CGN in low K ϩ by activating Akt and down-regulating ERK, (ii) low K ϩ -induced ERK activation, in turn, induces c-Jun, which can be inhibited by GDF-15, and (iii) GDF-15 mediated survival is accompanied by prevention of low K ϩ -induced ROS generation in CGN.
Expression of Recombinant Human GDF-15-Full-length GDF-15 cDNA was cloned and sequenced as described previously (13). Recombinant human GDF-15 protein was expressed in baculovirus-infected insect SF9 cells and purified from supernatants as described previously (15). Protein extracts of uninfected cells were treated under the same conditions and used in parallel for controls.
Cell Cultures-Cerebellar granule neurons were isolated and cultured from 8 day old Wistar rats as described previously (22) with slight modifications. Briefly, the freshly dissected cerebellum was trypsinized and triturated in ice cold Ca 2ϩ -and Mg 2ϩ -free Hanks' balanced salt solution. The cells were resuspended in high K ϩ medium (Eagle's basal medium containing 10% fetal calf serum, 25 mM KCl (Sigma), 2 mM glutamine, and 0.5% (v/v) pencillin/streptomycin). Cells were seeded in poly-L-lysine (100 g/ml, Sigma) precoated wells in 12-96-well plates at an average density of 1,500 cells/mm 2 . Cultures were incubated at 37°C with 5% CO 2 in a humidified chamber. Cytosine arabinoside (10 M, Sigma) was added after 24 h in vitro to prevent proliferation of nonneuronal cells. On day 4 in vitro the culture dishes were switched from high K ϩ to low K ϩ medium (Eagle's basal medium, 5 mM KCl, 2 mM glutamine, and 0.5% pencillin/streptomycin; ''5K''). Depending on the experiment at the time of medium change the cells in 5K were treated in the presence or absence of stimulants for indicated time points and were processed for the assays described below. All the cell death analyses were performed after 24 h of treatment.
LDH Measurements-LDH activity was assayed using an LDH assay kit (Promega). Total LDH release (percent) was calculated from maximum release, defined as the amount of LDH obtained after exposure of 25 mM K ϩ cultures to Triton X-100, 0.1% for 10 min at 37°C.
Propidium Iodide Staining-Propidium iodide (PI) was used to determine numbers of dead cells. PI (4.6 g/ml) was added directly to the culture medium for 3 min at room temperature, and the cells were counted after fixation with 4% paraformaldehyde. Typically around 70 -300 PI-stained cells were counted in four randomly selected fields per dish. Numbers of PI-stained cells in low K ϩ medium were normalized to 100%. Cell counts were presented as percentages of low K ϩinduced cell death.
TUNEL (Terminal Deoxyuridynyltransferase Nick End Labeling) Assay-The TUNEL assay was performed according to the manufacturer's manual (Promega). Around 500 cells in four different fields consisting of ϳ20 -180 TUNEL-positive cells were counted per coverslip under phase contrast microscope. Cell numbers were converted into percentage of apoptotic cells calculated from total cell numbers.
Reactive Oxygen Species (ROS) Measurement-ROS measurements were essentially carried out as described in Ref. 20. The culture plates were read on Fluostar OPTIMA plate reader (BMG Labtechnologies, Offenburg, Germany) at 480 nm excitation and 520 nm emmision. Neurons grown in high K ϩ showed negligible fluorescence and were used for background fluorescence.
Western Blot Analysis-Neurons cultured in 12-well plates were harvested in 1ϫ SDS lysis buffer containing 62.5 mM Tris-HCl (pH 6.8), 2% w/v SDS, 10% glycerol, 50 mM dithiothreitol, and 0.01% w/v bromphenol blue. Cell lysates containing equal amounts of protein (25 g/ lane) were loaded on 10 or 12% SDS-polyacrylamide gels. The separated proteins were transfered onto a polyvinylidene difluoride membrane using a wet transfer system (Amersham Biosciences). The membranes were blocked with blocking buffer containing 5% dry milk in Trisbuffered saline with Tween 20 (10 mM Tris-HCl (pH 7.4), 150 mM NaCl, 0.1% Tween 20) and incubated overnight with different primary antibodies. Incubations with secondary antibodies were performed for 1 h, and the bound antibodies were detected with the enhanced chemiluminescence (ECL, Amersham Biosciences) reagents in accordance to the manufacturer's protocol. Densitometric quantification of Western blots was done using NIH image (version 1.61) software.

GDF-15 Prevents Cell Death of CGN Induced by Low
K ϩ -We first assayed the effect of GDF-15 on CGN viability by monitoring the release of LDH into the culture medium. Fig. 1A (left panel) shows that GDF-15 significantly decreased release of LDH from CGN induced by low K ϩ (5K) at an optimal concentration of 10 ng/ml. To further substantiate the protective effect of GDF-15, we quantified neuron death by counting PI-stained (dead) cells. Fig. 1A (right panel) documents that GDF-15 maximally reduced numbers of PI positive neurons at a concentration of 10 ng/ml. TUNEL stainings corroborated these findings by showing (Fig. 1B) an increase in the number of apoptotic cells with DNA fragmentation in low K ϩ , which was prominently decreased by GDF-15. Quantitative evaluations revealed a more than 30% increase in numbers of TUNEL positive cells after 24 h under low K ϩ conditions. However, the cultures maintained in serum-free high potassium showed only ϳ4% TUNEL-positive cells. In the presence of IGF-1, an established anti-apoptotic molecule for CGN following K ϩ withdrawal (18,23), cell death was reduced to 10%. GDF-15 reduced numbers of apoptotic neurons in low K ϩ to 15%. Together, these data indicate that GDF-15 is a potent neuroprotective factor for CGN that matches the effect of IGF-1, the best established protective factor in this culture system.

Effects of PI3K and MEK1/2 Inhibitors on GDF-15-induced
Neuroprotection-To begin to analyze the signaling cascades activated by GDF-15, we chose specific inhibitors that inactivate core units of different signaling pathways. Inhibition of PI3K, which directs activation of Akt, has been shown to abolish protective effects of growth factors in CGN (24,25). We used two selective inhibitors of PI3-kinase, wortmannin and LY294002, to block the activity of the enzyme in CGN. To determine whether GDF-15-induced survival is impaired by PI3K inhibitors, we first performed LDH measurements ( Fig.  2A, left panel) and counted PI-stained cells ( Fig. 2A, right  panel). The protective effect of GDF-15 was significantly reduced in wortmannin (100 nM)or LY294002 (50 M)-pretreated cultures, indicating that PI3K activation is important for GDF-15 signaling. To examine the effect of inhibitors on DNA fragmentation, we performed TUNEL stainings. Fig. 2B shows an increased number of TUNEL-positive neurons in the presence of GDF-15 plus wortmannin compared with GDF-15 alone. However, GDF-15 protection was only partly (about 50%) attenuated by inactivation of the PI3K, suggesting that GDF-15 may employ additional mechanisms to protect CGN.
Activation of extracellular signal-regulated protein kinases (ERKs) has been reported to contribute to neuronal cell survival in certain models of neurotoxicity (26 -28). As our data indicated that GDF-15 is a potent protective factor for CGN, we hypothesized that regulation of ERK phosphorylation might be involved in GDF-15-induced survival. ERK1/2 activity is exclusively regulated through MEK1/2 phosphorylation, an immediate upstream dual-specificity kinase (29). We used U0126 (20 M), which blocks MEK1/2 (25), and PD98059 (60 M), a specific blocker of MEK1 (30), to inhibit ERK phosphorylation. Fig.   2, A and B, show that inactivation of ERKs did not reverse the survival promoting effect of GDF-15. On the contrary, neuronal protection with GDF-15 was even more pronounced in the presence of U0126 (20 M) or PD98059 (60 M) as measured by LDH release and numbers of PI-stained cells. To address the question whether MEK inhibition alone is sufficient to promote CGN survival in low K ϩ , we applied U0126 or PD98059 without GDF-15. Fig. 2A provides evidence that both inhibitors promoted survival, although U0126 showed a somewhat higher protection compared with PD98059. To examine whether the MEK1/2 inhibitor increased cell viability by inhibiting DNA fragmentation, we performed TUNEL stainings and quantified the data. As shown in Fig. 2B, the MEK1/2 inhibitor U0126 significantly decreased numbers of TUNEL positive cells. These results suggest that low K ϩ -induced apoptosis in CGN activates MEK1/2 during neuron death and that inhibition of this pathway clearly attenuates cell death.
GDF-15 Induces Akt and GSK-3 Phosphorylation-The serine/threonine protein kinase Akt is a downstream effector of PI3K and has been reported to play a critical role in promoting the survival of a variety of different cell types (for reviews, see Refs. 31 and 32). As the PI3K inhibitor blocked GDF-15-mediated CGN survival we investigated whether GDF-15 induces Akt phosphorylation in CGN cultured in low K ϩ (5K) conditions. As shown in Fig. 3A, GDF-15 significantly increased levels of phosphorylated Akt within 5 min after factor administration. This effect was still detectable after 24 h and decreased to basal level by 48 h (Fig. 3A). GSK-3 is a critical downstream element of the PI3K/Akt cell survival pathway, and its activity can be inhibited by Akt-mediated phosphorylation at Ser-21 of GSK-3␣ and Ser-9 of GSK-3␤ (33). We used an anti-phospho-GSK-3␤ (Ser-9) antibody to detect phosphorylated GSK-3␤. As shown in Fig. 3A, GDF-15-induced phosphorylation of GSK-3␤ was paralleled by an activation of Akt. Moreover, preincubation of the cultures with the PI3K inhibitors wortmannin (100 nM) or LY294002 (10 M) markedly blocked the GDF-15-induced Akt and GSK3 phosphorylation (Fig. 3B). Total Akt levels were not changed (Fig. 3, A and B). These data corroborate our findings that GDF-15 rescues CGN from cell death via the PI3K/Akt survival pathway. Moreover, the rapid phosphorylation of Akt/GSK within 5 min after factor administration points to a direct effect of GDF-15.
GDF-15 Blocks MEK/ERK Signaling in Low K ϩ CGN Cultures-The neuroprotective effect of the MEK inhibitors U0126 and PD98059 on CGN maintained in low K ϩ alone suggests that ERKs may be activated in this system at a certain time point during apoptosis. To investigate this hypothesis, lysates obtained at various time points from low K ϩ CGN culture were subjected to Western blot analysis to detect phosphorylated (activated) ERK using an anti-phospho-ERK1/2 antibody. The same blot was stripped and reprobed to detect total ERK level. As shown in Fig. 4A, activation of ERK1/2 became apparent at about 6 h of K ϩ deprivation, and the signal was maintained for 24 h, declined at 48 h, and remained elevated compared with 0 or 4 h (not shown). Although timing of robust ERK activation varied slightly between different culture batches, it was consistently observed between 4 and 6 h after K ϩ deprivation. There was no change in the level of total ERK. The MEK1/2, upstream activators of ERK, were also phosphorylated with the same kinetics as seen for ERK (Fig. 4A). ERK phosphorylation was completely blocked when cultures were preincubated with the MEK inhibitor U0126 (20 M) or PD98059 (60 M) without affecting total ERK levels (Fig. 4B). To test whether the cell death-preventing effect of MEK inhibition is dose-dependent, we analyzed cell death and ERK inhibition using various concentrations of the potent MEK1/2 inhibitor U0126. U0126 inhibited cell death as well as ERK phosphorylation in a dosedependent manner. As shown in Fig. 5A, the protective effect was maximal between 5 and 20 M, the concentrations at which ERK phosphorylation was strongly inhibited without affecting total ERK levels (Fig. 5B). Higher concentrations of U0126 were toxic to CGN.
To address the question whether GDF-15 may prevent cell death by down-regulating phosphorylation of ERKs, we performed Western blot analysis using cell lysates prepared after various time points of GDF-15 (20 ng/ml) administration. As shown in Fig. 6A, sustained ERK activation observed in control cultures (5K ϩ BC) was markedly suppressed by GDF-15 with- out affecting total ERK levels. Phosphorylation of MEK1/2 was also affected by GDF-15 (Fig. 6A). However, GDF-15 did not completely inhibit ERK activation, while U0126 did (cf. Fig.  4B). In the presence of both GDF-15 and U0126 suppression of cell death was more pronounced than by individual treatments (Fig. 6B). We hypothesized that this could be due to a complete suppression of the ERK cell death pathway by U0126 and an activation of the Akt survival pathway by GDF-15. To challenge this hypothesis we performed Western blot analysis using cell lysates isolated 24 h after GDF-15 (20 ng/ml) treatment in the presence and absence of U0126 (20 M). Fig. 6B shows a complete suppression of ERK phosphorylation, while Akt remained phosphorylated in the presence of both GDF-15 and U0126.

MEK1/2 Inhibitor and GDF-15 Block Phosphorylation of c-Jun-
The c-Jun protein and its phosphorylation have been shown to be involved in low K ϩ -induced cell death (19,21). Given that ERK can phosphorylate c-Jun in vitro (34), together with our observation that there is a slow and sustained ERK activation in low K ϩ , we hypothesized that ERK could be a potential upstream regulator of c-jun. We performed Western blots using cell lysates obtained from various time points after K ϩ deprivation and probed them with phospho-specific c-Jun and total c-Jun antibodies. As shown in Fig. 7A, c-Jun phosphorylation increased slightly at 3 h after K ϩ deprivation, yet both protein and phosphorylation increased robustly and in parallel to ERK activation at 6 h and remained high for 24 h. Evidence that ERK regulates c-Jun protein and phosphorylation in low K ϩ -induced apoptosis was obtained from ERK inhibition studies. The potent inhibitor of MEK/ERK signaling, U0126, was added to low K ϩ (5K) cultures for 9, 12, and 24 h, and lysates were subjected to Western blot analysis. Fig. 7A shows that in the presence of U0126 (20 M) c-Jun phosphorylation was dramatically reduced, and the protein level was markedly decreased. Interestingly, GDF-15, which attenuated ERK phosphorylation, also markedly reduced c-Jun phosphorylation and protein levels. (Fig. 7A).
The stress-activated kinases, JNK and p38, are known to regulate c-Jun gene expression and phosphorylation (35)(36)(37). In accordance with previous studies (21, 38), we observed a high activation of JNK and a low activation of p38 in low K ϩ that were not affected by U0126 or GDF-15 treatments (Fig.   FIG. 4

. Low K ؉ (5K) induces slow and sustained MEK/ERK activation.
A, after switching the medium to 5K for the indicated time points the cells lysates were prepared and processed to detect phosphorylation of ERK1/2 (p-ERK) by Western blotting. The same blot was stripped for the detection of total ERK (ERK). Similarly, the representative lysates were processed for Western blot analysis to detect phosphorylation of MEK1/2 (p-MEK). B, MEK inhibitors attenuate ERK activation in 5K cultures. Immediately after changing the medium to 5K the cells were treated with 20 M U0126 or 60 M PD98059 for the indicated period of time. The lysates were processed for the detection of phosphorylated ERK, and the same blot was stripped for the detection of ERK. DMSO (dimethyl sulfoxide), vehicle control for inhibitors. The quantification of the intensity was depicted below the respective blots.

FIG. 5. MEK1/2 inhibitor U0126 decreases cell death and ERK activation in a dose-dependant manner.
A, different concentrations of U0126 were added to the culture immediately after switching the medium to low K ϩ (5K). Cell death was measured after 24 h by staining with propidium iodide. Significance was calculated at p Ͻ 0.05 (*) or p Ͻ 0.01 (**) compared with untreated cells. B, U0126 inhibits ERK phosphorylation in a dose-dependent manner. U0126 was added to the 5K cultures immediately after switching to 5K medium. After the above mentioned time points cell lysates were assessed for phosphorylation of ERK (p-ERK), and the same blot was stripped and reprobed for total ERK (ERK). DMSO (dimethyl sulfoxide), vehicle control for inhibitors. The quantification of the intensity was depicted below the respective blots. 7B). Thus, activation of ERK regulates c-Jun protein synthesis and phosphorylation in low K ϩ CGN cultures. Treatment with either U0126 or GDF-15 abolishes this signaling and protects neurons from cell death.
The Anti-apoptotic Effect of GDF-15 Inhibits ROS Formation-ROS generated by oxidative stress are known to be important cell death effector molecules in low K ϩ CGN cultures (20,39). We first studied the time course of ROS generation in low K ϩ cultures using the fluorescent ROS indicator dye DCF-H 2. DCF-H 2 gets oxidized to DCF by ROS, which, upon excitation, emits fluorescence (40,41). Fig. 8A shows two peaks in ROS formation at 6 and 12 h and return to base-line levels at 24 h. We next investigated ROS generation after 6 h in low K ϩ in the presence of GDF-15 (20 ng/ml). As shown in Fig. 8B, GDF-15 effectively prevented ROS generation. The MEK inhibitors (PD 98059 (60 M) and U0126 (20 M)) did not affect DCF fluorescence, corroborating the notion that ROS generation is upstream of ERK activation. DISCUSSION GDF-15 is a novel divergent member of the TGF-␤ superfamily with largely unknown functions. Recent studies demonstrated that the molecule is expressed in the central nervous system, and the first functional characterization of the factor revealed pronounced neurotrophic effects on midbrain dopaminergic and raphe serotonergic neurons in vitro and in vivo (15). However, it is not clear whether other central nervous system neuron populations are also promoted by GDF-15. TGF-␤s can affect neurons in both anti-and pro-apoptotic fashions. Thus, elimination of TGF-␤ in chick embryos has been shown to abolish ontogenetic neuron death (36,42) acting as a master molecule in the regulation of pro-apoptotic signaling cascades. In contrast, TGF-␤ can also synergize with neurotrophic factors, e.g. neurotrophins, fibroblast growth factor-2, and GDNF, in the promotion of neuron survival (43)(44)(45)(46)(47). With regard to GDF-15, a pro-apoptotic role in several cell lines has been suggested (48 -51). Moreover, our observation that GDF-15 is robustly induced in lesioned cortical neurons (16) can be interpreted in terms of both an anti-or pro-apoptotic role. To clarify this, we started to analyze the effect of GDF-15 and underlying mechanisms on neuronal apoptosis using the well established CGN model (18). Reverse transcriptase-PCR analysis showed that cultured granule neurons express GDF-15, but expression is down-regulated upon withdrawal of survival promoting signals (not shown). The present study demonstrates that GDF-15, in contrast to TGF-␤2, which has been shown to accelerate apoptosis in CGN (52), prevents primary CGN from low K ϩ -induced apoptosis. Concerning underlying mechanisms, we first showed that the protective effect of GDF-15 was attenuated by inhibitors of the Akt-activating PI3K. Western blot analyses clearly confirmed the functional results and demonstrated a GDF-15-dependent activation of Akt and its downstream target GSK-3␤. Consistent with our data is the observation that other TGF-␤ members, including GDNF, neurturin, persephin, and TGF-␤1, promote cell survival via the PI3K/Akt pathway in various cell types (53)(54)(55)(56)(57).
We next analyzed whether the MEK/ERK pathway may be involved in GDF-15 dependent neuron rescue. Surprisingly, we found that GDF-15-induced protection in CGN was even more pronounced when ERKs were inactivated via selective inhibition of MEK. Moreover, MEK inactivation alone led to a potent protection against low K ϩ -induced apoptosis of CGN. We observed slow and sustained ERK activation in CGN undergoing low K ϩ -induced cell death. Down-regulation of ERK by MEK inhibitors led to a prevention of low K ϩ -induced apoptosis. These findings corroborate several recent reports showing a pro-apoptotic involvement of ERK in neuronal cell death (58,59). Moreover, recent studies also demonstrate that L-2-chloropropionate-and kainate-mediated cell death in CGN (60,61) is attenuated by MEK inhibition. Most importantly, GDF-15 clearly suppressed the persistent ERK activity in CGN. Prevention of cell death by MEK inhibitors is unexpected, because typically the ERK pathway has been credited with proliferative and cell survival responses (62,63). However, our results together with the above studies expand the functional role of ERKs and suggest that the duration of ERK activation might determine pro-or anti-apoptotic decision of the cell.
The AP-1 transcription factor family members c-fos and c-jun have been implicated with pro-apoptotic functions (see Ref. 64 for a review). The role of c-Fos in low K ϩ -induced CGN cell death is not known. A requirement of c-Jun activation in mediating low K ϩ -induced CGN apoptosis has previously been demonstrated by Watson et al. (21). The present study has FIG. 6. A, GDF-15 blocks low K ϩ (5K)induced sustained ERK activation. GDF-15 was added to 5K cultures for the indicated time points, and the cells were lysed and processed for Western blot analysis to detect phosphorylated ERK (p-ERK). The same blot was stripped to detect total ERK (ERK). Similarly, the representative lysates were processed to detect phosphorylation of MEK1/2 (p-MEK). B, GDF-15-and U0126-treated low K ϩ (5K) cultures show active Akt. The cells in 5K were treated with GDF-15 (20 ng/ml) and/or U0126 (20 M) for 24 h. The cell lysates were processed for Western blotting to detect phosphorylated ERK and ERK. The representative samples were processed to detect phosphorylation of Akt (p-Akt). -, no treatment; BC, control lysate of cultures treated with purified supernatant of uninfected Sf9 cells. The quantification of the intensity was depicted below the respective blots.
provided evidence that c-Jun phosphorylation and protein levels increase in parallel to the increase in ERK activation. Notably, MEK1/2 inhibition or treatment with GDF-15, which both inhibit ERK activation, also inhibited c-Jun protein upregulation and phosphorylation. Although the in vivo role of ERK in regulating c-Jun has yet to be determined (65), evidence that c-Jun can be phosphorylated by ERK in vitro (34) and our finding that MEK1/2 inhibition attenuates the increase in c-Jun protein and phosphorylation argue in favor of the possibility that ERK may be directly involved in regulating c-jun in low K ϩ CGN culture. It is also worthwhile noting that the MEK1/2 inhibitor U0126 was originally characterized based on its ability to inhibit AP-1 transcription elements c-jun and c-fos via the inhibition of MEK/ERK signaling cascade (29). Furthermore, the demonstration that kainate-induced toxicity involves c-Jun activation (66), as well as the ability of MEK inhibitors to prevent kainate toxicity in CGN (61) (67). Several studies have demonstrated a MEK/ERK-mediated regulation of the cyclin-dependent kinases (CDKs) cyclin D1, cyclin E, and cyclin kinase inhibitors p21 and p27. Activation of CDKs and cyclin kinase inhibitors can result in cell cycle arrest or apoptosis (68 -70). In the nervous system CDKs have been shown to be involved in the death of PC12 cells and postmitotic neurons (71)(72)(73)(74). For example, serum-deprived neuronal PC12 cells and sympathetic neurons deprived of trophic support contain inappropriate amounts of cyclin B and cyclin D transcripts prior to death (75,76). Most importantly, Padmanabhan et al. (77) documented the requirement of CDK proteins in low K ϩ -induced CGN cell death. They demonstrated an up-regulation of cyclin D1 and cyclin E after 6 h in low K ϩ . This is consistent with the late and persistent ERK induction observed in our studies. Although the mechanisms underlying CDK-activated neuronal cell death are still unknown, non-neuronal cells have provided model systems for the control of cell death via the MEK/ERK signaling pathway. The ability of MEK inhibitors to regulate CDKs and cyclin kinase inhibitor proteins (70,78,79), the presence of a consensus AP-1 promoter sequence in cyclin D1, and its direct induction by c-Jun (67) suggest that the strong MEK/ERK/c-Jun signals observed in this study may direct the regulation of cell cycle protein kinases for the orchestration of neuronal cell death.
To elucidate the GDF-15-mediated mechanisms that suppress ERK activation in CGN, we continued by documenting that the protective effect of GDF-15 is connected to the regulation of ROS. It has been reported that increased generation of ROS is closely associated with low K ϩ -induced cell death (20,39), and several studies have provided evidence that ROS in-FIG. 7. GDF-15 and U0126 inhibit c-Jun protein expression and phosphorylation. After switching to low K ϩ (5K) medium the cells were treated with U0126 (20 M) or GDF-15 (20 ng/ml) for the above mentioned time points. The cells were lysed and processed for Western blotting. A shows the Western blot analysis using antibodies against phosphorylated c-Jun (p-c-Jun), total Jun (c-Jun), and phosphorylated ERK (p-ERK) and the densitometric quantification of p-c-Jun and c-Jun levels for 9, 12, and 24 h. B shows Western blot analysis using antibodies against phosphorylated JNK (p-JNK) and phosphorylated p38 (p-p38) and the densitometric quantification of pJNK levels for 9, 12, and 24 h. duces ERK activation (80 -82). We now show for the first time that GDF-15 inhibits the formation of ROS. Furthermore, the inability of both MEK inhibitors to prevent ROS generation suggests that ROS generation is upstream of ERK activation in low K ϩ CGN. It should be noted, however, that in other systems, as superior cervical ganglionic neurons, nerve growth factor can prevent ROS generation by activating the ERK pathway (83), suggesting that ROS formation in this system is downstream of ERK.
The present in vitro work has provided clues for the possible signaling pathways that are activated by GDF-15 for executing its functions. Activation of Akt is known to play a critical role in controlling the balance between survival and apoptosis (84,85). The discovery that ERK inhibition promotes survival was surprising but needs to be discussed in the context of the putative in vivo relevance of this finding. Recent in vivo studies have revealed that inhibition of ERK can reduce infarct volume after focal cerebral ischemia, suggesting a deleterious effect of ERK activation (86,87). In addition, an abnormal punctate staining pattern of activated ERK has been described in the brains of patients with Alzheimer's disease (88). While the status of ERK activation in patients with Parkinson's disease is currently unknown, the involvement of ERK in 6-hydroxydopamine toxicity (89), and our previous in vivo studies, in which we demonstrated potent protective effects of GDF-15 in 6-hydroxydopamine-lesioned rat brains (15), suggest that the GDF-15 may exert its protective effect in neurons through an ERK-inactivating mechanism.
In conclusion, we have demonstrated that GDF-15 prevents cell death in low K ϩ CGN cultures by two different modes of action. First, GDF-15 induces survival by activating Akt and GSK via the PI3K. Second, GDF-15 down-regulates the MEK/ ERK/c-Jun cell death signaling pathway. Moreover, GDF-15 also attenuates ROS formation, an inducer of ERK, thereby preventing cell death. Putative cross-talks between these pathways need to be addressed by future experiments. An important issue also to be resolved in the future concerns the identity of GDF-15 receptors. Finally, the question whether GDF-15 may act as a regulator of apoptosis for CGN in vivo remains to be answered. Ongoing studies with a GDF-15-deficient lacZ knock-in transgenic mouse, recently developed in our laboratory, may help to address this question.