Fibroblast Growth Factor 1 Regulates Signaling via the Glycogen Synthase Kinase-3β Pathway

We hypothesize that in neurodegenerative disorders such as Alzheimer's disease and human immunodeficiency virus encephalitis the neuroprotective activity of fibroblast growth factor 1 (FGF1) against several neurotoxic agents might involve regulation of glycogen synthase kinase-3β (GSK3β), a pathway important in determining cell fate. In primary rat neuronal and HT22 cells, FGF1 promoted a time-dependent inactivation of GSK3β by phosphorylation at serine 9. Blocking FGF1 receptors with heparinase reduced this effect. The effects of FGF1 on GSK3β were dependent on phosphatidylinositol 3-kinase (PI3K)-protein kinase B (Akt) because inhibitors of this pathway or infection with dominant negative Akt adenovirus blocked inactivation. Furthermore, treatment of neuronal cells with FGF1 resulted in ERK-independent Akt phosphorylation and β-catenin translocation into the nucleus. On the other hand, infection with wild-type GSK3β recombinant adenovirus-associated virus increased activity of GSK3β and cell death, both of which were reduced by FGF1 treatment. Moreover, FGF1 protection against glutamate toxicity was dependent on GSK3β inactivation by the PI3K-Akt but was independent of ERK. Taken together these results suggest that neuroprotective effects of FGF1 might involve inactivation of GSK3β by a pathway involving activation of the PI3K-Akt cascades.

family (2,3), FGF1 (or acidic FGF) is abundant in sensory and motor neurons, and FGF2 (or basic FGF) is primarily produced by astrocytes, although it can be taken up by neurons and translocated to the nucleus (4). Of the four FGF receptors (FGFRs), three are found in the brain: FGFR1 is mainly expressed on neurons, while FGFR2 and FGFR3 are found on glial cells (1,2,6,7). Binding of FGF leads to dimerization of FGFR followed by tyrosine kinase activation (2). FGF2 promotes survival of cortical and hippocampal neurons (8,9) and is also capable of rescuing neurons from denervation and injury (1). Similarly, FGF1 protects selective neuronal populations against the neurotoxic effects of molecules involved in the pathogenesis of neurodegenerative disorders such as Alzheimer's disease (10,11) and HIV encephalitis (12).
FGF1 and -2 are potent regulators of central nervous system development (13,14) and maintenance after neuronal injury (1). However, there is no consensus as to the signal transduction pathways initiated by FGF during neuronal differentiation or during neuroprotection. Some studies suggest that during FGF2-induced neuronal differentiation (i) activation of a mitogen-activated protein kinase, such as extracellular signal-regulated kinases (ERK1 and ERK2), is neither necessary nor sufficient, (ii) activation of Src kinases is necessary but not sufficient, and (iii) FGF2 requires at least two signaling pathways activated by Ras and Src (15)(16)(17). These results indicate that the neuroactive effects of FGF depend on signaling pathways other than ERK. Most studies have focused on the neurotrophic effects of FGF2, while fewer studies have investigated the effects of FGF1. We hypothesize that the neurotrophic effects of FGF1 might involve regulation of other signaling cascades such as the glycogen synthase kinase-3␤ (GSK3␤) pathway, which is important in determining cell fate (18,19). Supporting this possibility, a recent study showed that FGF2-mediated tau hyperphosphorylation was inhibited by lithium, an inhibitor of GSK3␤, but not by inhibitors of ERK or the cyclin-dependent kinases (20). For the present study, we investigated the effects of FGF1 on the GSK3␤ pathway in rat primary neuronal and HT22 cells. Our results suggest that the neuroprotective properties of FGF1 might involve phosphorylation-mediated inactivation of GSK3␤ via the phosphatidylinositol 3-kinase (PI3K)-Akt signaling pathway.

EXPERIMENTAL PROCEDURES
Cell Culture and Treatments-All experiments were performed with rat primary cortical neurons prepared from embryonic day 17 Sprague-Dawley rats as described previously (21) and with HT22 cells (22), mouse hippocampal cells derived from the HT4 cell line (23). Briefly, primary cortical neurons were dissociated from the cortex and maintained in tissue culture dishes coated with 100 g/ml poly-D-lysine in minimum Eagle's medium supplemented with 30 mM glucose, 2 mM glutamine, 1 mM pyruvate, and 10% fetal bovine serum. Cultures were used within 1 week after preparation. HT22 cells were maintained at no greater than 70% confluence in 10% fetal bovine serum in Dulbecco's modified Eagle's medium (high glucose, Irvine Scientific, Irvine, CA) with 1 mM L-glutamine, 1 mM sodium pyruvate, and 1% penicillin/ streptomycin (Invitrogen).
To analyze the effects of FGF1 on neurons, cells were placed in N-2-supplemented serum-free medium (Invitrogen) 24 h before treatment. Neurons were then exposed to FGF1 (10 ng/ml, Sigma) for 0, 1, 2, 5, 10, 30, and 60 min and analyzed by immunoblot for levels of GSK3␤, Akt, and ERK1/2. To confirm that FGF1 effects involve signaling through the FGFR pathway, cells were treated with heparinase since FGF signal transduction involves binding both at the high affinity FGF receptor and the low affinity heparin sulfate receptor. Because heparin optimizes the effects of FGF (24,25), additional experiments were performed in which heparin (2.5 units/ml, Sigma) and heparinase (1unit/ml, Sigma) were added 1 h prior to FGF1 treatment (11).
To investigate the intracellular signaling pathways involved in mediating FGF1 effects, cells were pretreated for 30 min with (i) ERK inhibitors U0126 and PD98059 (10 M, Calbiochem) or (ii) PI3K inhibitor LY294002 (10 M, Calbiochem). All experiments were performed at least three times.
Adenoviral and Recombinant Adenovirus-associated Vector Constructs and Transfection-Replication-defective adenovirus vectors expressing mouse Akt protein fused in-frame to the FLAG epitope under the control of the cytomegalovirus (CMV) promoter were constructed as described previously (kind gifts from Dr. Kenneth Walsh, Tufts University, Boston, MA) (26,27). The dominant negative mutant Akt (S473A) protein cannot be activated by phosphorylation (28) and functions in an inactive fashion (29). The constitutively active Akt construct has the c-src myristoylation sequence fused in-frame to the N terminus of the FLAG-Akt (wild-type) coding sequence (27). Adenoviral Akt constructs were amplified in 293A cells and purified by ultracentrifugation through a CsCl gradient (30).
For infections, HT22 neuronal cells were plated at ϳ50% confluence in serum-free medium and cultured for 24 h at 37°C in 5% CO 2 . Half of the conditioned serum-free medium was removed from each sample, pooled, and stored at 37°C in 5% CO 2 . Cells were infected at a multiplicity of infection of 50 in preconditioned serum-free medium for 4 h. Cells were rinsed three times in warm PBS, and medium was replaced with pooled preconditioned serum-free medium followed by a 48-h incubation at 37°C in 5% CO 2 . After 48 h, cells were treated with FGF1 (10 ng/ml) for 10 min, harvested in lysis buffer, stored at Ϫ20°C, and later used for the Akt and GSK3␤ kinase activity assays and Western analyses. For immunocytochemistry, cells were cultured on coverslips and treated as described above, fixed in 4% paraformaldehyde, and blocked overnight at 4°C in 10% horse serum and 5% bovine serum albumin. Cells on coverslips were then labeled overnight at 4°C with primary anti-FLAG (1:50) (Sigma) followed by incubation with secondary biotinylated IgG (Vector Laboratories, Inc., Burlingame, CA) (1: 200) for 1 h at room temperature. FLAG proteins were detected with 3,3Ј-diaminobenzidine (Sigma) and visualized by light microscopy to access FLAG production and transduction efficiency (data not shown). This assay confirmed that transfection efficiency was up to 90%. Experiments were conducted at least three times to ensure reproducibility.
Expression of exogenous GSK3␤ was performed using the recombinant adenovirus-associated virus (rAAV) Helper-Free System (Stratagene, San Diego, CA) according to the manufacturer's instructions with minor modifications. Briefly, pXT7 plasmids bearing wild-type GSK3␤ cDNA (kind gifts from Dr. Zhengui Xia, University of Washington, Seattle, WA and Dr. Isabel Dominguez, Harvard University, Cambridge, MA) were digested with XbaI followed by ligation of the fragments into the XbaI site of the pCMV-MCS expression vector (Stratagene). The resulting pCMV-MCS-GSK3␤ plasmids were then cut with NotI followed by ligation of GSK3␤-containing fragments to the NotI fragments of pAAV-LacZ (Stratagene) to yield the pAAV-GSK3␤.
For generation of the rAAV, subconfluent 293T cells were transfected without or with the pAAV-GSK3␤ together with pAAV-RC and pHelper plasmid (Stratagene) using Superfect (Qiagen, Valencia, CA). After incubation for 72 h, cells were harvested and lysed by four freeze-thaw cycles. The lysates were centrifuged at 10,000 ϫ g for 10 min to remove cell debris and frozen at Ϫ80°C until use. As a positive control, pAAV-GFP (Stratagene) was used to confirm the transfection efficiency.
Subconfluent neuronal cells were preincubated with hydroxyurea (40 mM) and sodium butylate (1 mM) for 6 h. After washing with PBS, cells were incubated with the transfected 293T cell lysates (1/10 dilution) in Dulbecco's modified Eagle's medium with 2% fetal calf serum for 24 h. The cells were then washed with PBS and incubated in Dulbecco's modified Eagle's medium with 2% fetal calf serum for an additional 36 h. Finally, cells treated with or without FGF1 were harvested and further analyzed as described above. As a positive control, HT1080 human fibrosarcoma cells were used to confirm the transfection efficiency. Experiments were conducted at least three times to ensure reproducibility.
Immunocomplex Kinase Assay for GSK3␤ and Akt-Assays were performed essentially as described previously with some modifications (18). Briefly, for the GSK3␤ assay, cells were rinsed twice with cold PBS and incubated for 20 min on ice in lysis buffer (1% Triton X-100, 10% glycerol, 50 mM HEPES, pH 7.4, 140 mM NaCl, 1 mM EDTA, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 5 g/ml aprotinin, and 5 g/ml leupeptin). The cell lysates were then centrifuged for 10 min at 14,000 rpm, and protein concentration was determined using the BCA reagent (Pierce). Two hundred micrograms of the supernatant were preabsorbed with a protein G-Sepharose (Amersham Biosciences) for 1 h, and the precleared lysates were incubated with anti-GSK3␤ monoclonal antibody (1:50, Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C followed by incubation with protein G-Sepharose for 2 h at 4°C. The immune complexes were then washed twice with the lysis buffer and twice with kinase buffer (20 mM HEPES, pH 7.2, 0.1 mM Na 3 VO 4 , 10 mM glycerophosphate, 10 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM EGTA). Finally, the immune complexes were incubated in 30 l of the kinase buffer containing either 2.5 g of phosphoglycogen synthase-2 peptides or the Ala-21 mutant peptides (Upstate Biotechnology, Lake Placid NY) and 10 Ci of [␥-32 P]ATP (6000 Ci/mmol, PerkinElmer Life Sciences) for 20 min at 30°C. Reactions were terminated by the addition of 5 l of 500 mM EDTA and 5 mM ATP. Samples were then spotted onto Whatman P81 phosphocellulose filter paper. The filters were washed with 180 mM phosphoric acid, dried with acetone, and analyzed by scintillation counting.
Immune complex kinase assays for Akt were performed as described for the GSK3␤ assay with minor modifications. Briefly, the precleared lysates were incubated with polyclonal anti-human Akt antibody (antiprotein kinase B-(88 -100)) (1 g/sample) (Calbiochem) followed by incubation in protein G-Sepharose. After washing, immune complex assays were performed in the presence of 1.0 g of GSK3␤ fusion protein (Cell Signaling, Beverly, MA) as substrate. Reactions were terminated by an addition of the SDS sample buffer. The samples were then subjected to SDS-PAGE (15%) analysis followed by autoradiography. Quantification was performed with the PhosphorImager using the Image Quant software (Amersham Biosciences).
Immunocytochemical Analysis of ␤-Catenin by Laser Scanning Confocal Microscopy-Further analysis of catalytically active GSK3␤ was carried out in neuronal cultures by immunocytochemical visualization of the cellular localization of ␤-catenin. Using this method, inactivation of GSK3␤ is associated with nuclear translocation of ␤-catenin (32,33). Briefly, cells were plated on poly-L-lysine-coated coverslips, exposed to FGF1 in the presence or absence of inhibitors as described above, and fixed for 10 min with 4% paraformaldehyde in PBS. Cells were then immunolabeled with the mouse monoclonal antibody against ␤-catenin (1:1000, Transduction Laboratories) followed by incubation with fluorescein isothiocyanate-conjugated horse anti-mouse IgG (1:75) (Vector Laboratories, Inc.). Cells on coverslips were analyzed with a laser scanning confocal microscope (Bio-Rad MRC 1024). For each experimental condition, a total of 20 cells were analyzed in duplicate. For each condition, cellular distribution of ␤-catenin was recorded, and computer-assisted image analysis was performed to estimate the percentage of cells displaying predominantly nuclear versus cytoplasmic distribution of ␤-catenin.
Analysis of Neuronal Cell Viability and Cell Death-To determine whether the effects of FGF1 on the PI3K-Akt, GSK3␤, and ERK pathways influenced cell viability, neuronal cells treated with glutamate (5 mM, Sigma) were analyzed for DNA fragmentation and by the MTT assay and trypan blue exclusion.
Fluorescence-activated cell sorting analysis for the DNA fragmentation was conducted using the flow cytometric method of Nicoletti et al. (34) as previously described. This assay is highly correlated (R 2 ϭ 0.82) with the percentage of cells undergoing apoptosis as determined by annexin-V staining. Briefly, cell nuclei were stained with 50 g/ml propidium iodide in a hypotonic lysis buffer at 4°C for 2 h. Samples were then run on a FACScan (BD PharMingen) to quantify DNA fragmentation on the FL3 channel (propidium iodide). Cell debris were excluded from nuclei by an empirical setting of forward scatter and side scatter threshold levels. Apoptotic nuclear bodies appeared as a broad hypodiploid peak easily discernable from the narrow peak of cells with normal diploid DNA content.
For the MTT assay, 2.5 ϫ 10 3 cells/well were plated in 100 l of Dulbecco's modified Eagle's medium with 10% fetal bovine serum in 96-well microtiter plates. On the following day, after replacing the growth medium with the N-2-supplemented serum-free medium, cells were treated with inhibitors or activators (pharmacological, adenoviral or adenovirus-associated constructs) of the PI3K-Akt or GSK3␤ pathways followed by a 24-h treatment with FGF1 (10 ng/ml). The next day, cells were treated with 5 mM glutamate for an additional 24 h followed by cell survival assays. For the MTT assay (35), 10 l of MTT solution (2.5 mg/ml) was added to each well and incubated at 37°C for 4 -8 h followed by the addition of 100 l of solubilization solution (50% dimethylformamide and 20% SDS, pH 4.8). The next day absorption values were read at 570 nm. Experiments were done in triplicate, and results were averaged.
For trypan blue exclusion, after treating as described, cells were stained with trypan blue (Sigma), and 100 cells/sample were visualized and counted as either dead or alive with an Axiovert inverted microscope. Cells staining blue, indicating cell membrane compromise, were counted as dead.

FGF1 Promotes Akt, GSK3␤, and ERK Phosphorylation in
Neuronal Cells-To investigate the effects of FGF1 on Akt, GSK3␤, and ERK, we first determined the time course for phosphorylation of these protein kinases in rat cortical neurons and HT22 cells. Western blot analysis showed that stimulation with FGF1 resulted in maximum Akt, GSK3␤, and ERK phosphorylation at 10 -15 min followed by a progressive decrease reaching non-detectable levels at 60 min (Fig. 1). Since FGF1 signal transduction involves binding both at the high affinity FGF receptor and the low affinity heparin sulfate receptor, optimizing the effects of FGF (24), experiments were performed with heparin and heparinase. While pretreatment with heparinase blocked primarily Akt and GSK3␤ phosphorylation and to a lesser extent reduced ERK phosphorylation (Fig. 2), heparin enhanced FGF1-mediated phosphorylation of all three kinases (not shown).
FGF1 Regulation of GSK3␤ Activity Is Dependent on the PI3K-Akt Pathway-Since time course experiments (Fig. 1) suggested that the effects of FGF1 on GSK3␤ might be mediated via either the PI3K-Akt or the ERK signaling pathways, further analysis of the molecular events was conducted by treating cells with the PI3K inhibitor LY294002 or the ERK inhibitors U0126 and PD98059 prior to FGF1 stimulation. Inhibition of PI3K, which is upstream of Akt (18), resulted in decreased phosphorylation of Akt and GSK3␤ but had no effect on ERK phosphorylation (Fig. 3). In contrast, inhibitors of the ERK pathway completely blocked FGF1-mediated ERK phosphorylation with only slight effects on Akt and GSK3␤ phosphorylation (Fig. 3). Taken together these results indicate that phosphorylation of GSK3␤ by FGF1 stimulation is mediated via the PI3K-Akt pathway but is independent of ERK.
To further confirm that FGF1-mediated phosphorylation of GSK3␤ via PI3K-Akt resulted in inactivation of GSK3␤, immunocomplex kinase assays were performed. FGF1 treatment decreased GSK3␤ activity by 40%, while inhibition of the PI3K-Akt pathway with LY294002 re-established GSK3␤ activity even in the presence of FGF1 (Fig. 4A). Inhibition of the ERK pathway with U0126 did not interfere with the FGF1-mediated effect on GSK3␤ activity (Fig. 4A). Moreover, while transfection of neuronal cells with dominant negative Akt re-established GSK3␤ activity and blocked FGF1 effects on GSK3␤ (Fig. 4B), constitutively active Akt reduced GSK3␤ activity in a similar fashion to FGF1 (Fig. 4B). Immunocomplex activity assays for Akt confirmed that dominant negative Akt blocked FGF1 effects on Akt (Fig. 4C), and constitutively active Akt increased basal Akt activity levels (Fig. 4C). Consistent with these results, Western blot analysis showed that dominant negative Akt resulted in reduced Akt and GSK3␤ phosphorylation, while constitutively active Akt increased Akt and GSK3␤ phosphorylation (Fig. 4D). Furthermore, these effects were enhanced by FGF1 (Fig. 4D).
Neuronal cells were also infected with a rAAV-expressing wild-type GSK3␤ or with a control vector (AdV-GFP). In cells transfected with rAAV wild-type GSK3␤, the activity of this enzyme (Fig. 5A) as well as cell death (not shown) increased over basal levels compared with vector control. In contrast, pretreatment with FGF1 decreased GSK3␤ activity in neuronal cells infected with rAAV wild-type GSK3␤ (Fig. 5A). Western blot analysis using antibodies against phosphorylated and total GSK3␤ (Fig. 5B) confirmed these findings. Taken together these results support the notion that FGF1 might regulate GSK3␤ activity via PI3K-Akt activation.

Is Associated with ␤-Catenin Translocation to the Nucleus-
Since previous studies show that inactivation of GSK3␤ by phosphorylation at serine 9 results in translocation of ␤-catenin to the nucleus (33), immunocytochemical analyses with ␤-catenin antibodies were performed in neurons treated with FGF1 with or without pharmacological inhibitors. Laser scanning confocal microscopy showed that under basal conditions ␤-catenin immunoreactivity was primarily in the cytoplasm and, to a lesser extent, in the nucleus (Fig. 6, A and B). After stimulation with FGF1, ␤-catenin labeling in the nucleus was significantly increased (Fig. 6, C and D). Consistent with Western blot studies (Fig. 3) and kinase assays (Fig. 4), the effects of FGF1 on ␤-catenin nuclear localization were blocked by LY294002 (PI3K inhibitor) (Fig. 6E) but not by the ERK inhibitors PD98059 (Fig. 6F) and U0126 (Fig. 6G), supporting the idea that FGF1 blocks GSK3␤ activation leading to ␤-catenin degradation.
Neuroprotective Effects of FGF1 against Glutamate Are Mediated by GSK3␤-To investigate the physiological relevance of FGF1 on the PI3K-Akt, GSK3␤, and ERK pathways, cell viability was measured by DNA fragmentation, the MTT assay, and trypan blue exclusion in cells pretreated with FGF1 (10 ng/ml, 24 h) and challenged with glutamate (5 mM, 24 h) in the presence or absence of specific inhibitors. Consistent with results from the MTT assay and trypan blue exclusion (not shown), DNA fragmentation studies showed that FGF1 protected neurons against neurotoxic effects of glutamate and that these effects were blocked by LY294002 but not by U0126 (Fig.  7). Analysis of DNA fragmentation by fluorescence-activated cell sorting showed that treatment of the neuronal cells with glutamate resulted in the formation of apoptotic bodies (Fig. 7), whereas pretreatment with FGF1 reduced the formation of apoptotic bodies in the cells. Treatment with LY294002 (PI3K inhibitor) (Fig. 7A), but not U0126 (Fig. 7B), blocked the neuroprotective effects of FGF1 against glutamate. Similarly, dominant negative Akt blocked the neuroprotective effects of FGF1, while constitutively active Akt was protective against glutamate toxicity in the absence of FGF1 (Fig. 7C).

FIG. 4. Effects of inhibitors on FGF1-mediated activation of Akt and GSK3␤ in HT22 cells.
A, analysis of GSK3␤ activity by immunocomplex assay in the presence of inhibitors of PI3K (LY294002) and ERK (U0126). FGF1 reduced GSK3␤ activity. This effect was produced by LY294002 but not by U0126. B, analysis of GSK3␤ activity by immunocomplex assay in the presence of FGF1 and dominant negative and constitutively active Akt adenoviral infection. FGF1 alone or constitutively active Akt reduced GSK3␤ activity, while dominant negative Akt re-established GSK3␤ activity. C, analysis of Akt activity by immunocomplex assay showed that FGF1 alone or in the presence of constitutively active Akt induced an increase in Akt activity, while dominant negative Akt reduced Akt activity. D, Western blot analysis confirmed that dominant negative Akt reduced Akt and GSK3␤ phosphorylation, while constitutively active Akt increased Akt and GSK3␤ phosphorylation. These effects were enhanced by FGF1. Dom. Neg., dominant negative; Const. Act., constitutively active; GFP, green fluorescent protein.

DISCUSSION
The present study shows that the neurotrophic effects of FGF1 involve signaling via the PI3K-Akt and GSK3␤ pathways. The FGFR tyrosine kinase is linked to the G-protein Ras, which stimulates the ERK signal transduction cascade (15,25) (Fig. 8). In this context, since ERK plays a central role in mediating cellular responses to a variety of signaling molecules (36,37), most studies in both neuronal and non-neuronal cells have concentrated on characterizing the effects of FGF2 (rather than FGF1) on the ERK pathway (16, 17, 25, 36, 38 -42).
Although ERK plays an important role in regulating the trophic effects of FGF, other pathways may also be involved (Fig. 8). For example, the neurotrophic activity of FGF1 is dependent on endogenous FGF1 expression but independent of ERK (43). In agreement with this finding, we showed that stimulation of neuronal cells with FGF1 resulted in GSK3␤ inactivation and ␤-catenin translocation to the nucleus independent of ERK. Trophic factors such as insulin also promote GSK3␤ inactivation by phosphorylation at serine 9 (44), facilitating ␤-catenin nuclear localization (45). While inactivation of GSK3␤ correlates with cell survival, activation of GSK3␤ results in cell death (18). These effects are important for understanding the neurotrophic activity of FGF1 because the GSK3␤ signaling pathway has been shown to play an important role in regulating central nervous system development (32,46) and cell fate (18,19). Phosphorylation of GSK3␤ serine 9 by the dishevelled signal (Wnt) through the frizzled receptor and activated through the Notch receptor results in its inactivation (32) with subsequent ␤-catenin translocation to the nucleus (45). Alterations of this pathway are also currently being recognized as important in the pathogenesis of neurodegenerative disorders such as Alzheimer's disease (47,48) and HIV encephalitis (49). For example, activation of GSK3␤ might facilitate HIV-mediated neurotoxicity since recent studies have shown that the HIV protein Tat may activate GSK3␤ (49). In contrast, the neuroprotective effects of FGF1 against neurotoxins such as HIV-derived proteins and the amyloid ␤ protein of Alzheimer's disease might be associated with its ability to block GSK3␤. In support of this possibility, the present study shows that protection against glutamate toxicity is associated with inactivation of GSK3␤. Furthermore, the neurotoxic effects of gp120 (12) and amyloid ␤ (10) are blocked by inhibitors of GSK3␤ such as LiCl 2 (50 -52). In addition, in individuals with HIV encephalitis high levels of neuronal FGF1 expression correlate with improved cognitive performance and preservation of the dendritic integrity (12). Similarly, neurons that express high levels of FGF1, such as motor neurons, are resistant to amyloid ␤ toxicity (11).
As to the potential mechanisms mediating the effects of FGF1 on GSK3␤, the present study shows that inhibitors of the PI3K-Akt pathway block the effects of FGF1 on GSK3␤, while ERK inhibitors have no effect. Further confirming the involvement of the PI3K pathway, FGF1 treatment of neuronal cells resulted in Akt phosphorylation independent of ERK activation. This is consistent with previous studies showing that the PI3K-Akt signaling inactivates GSK3␤, which is important for cell survival (53,54). Both PI3K and Akt are activated by other growth factors including platelet-derived growth factor (55), insulin (56), and brain-derived neurotrophic factor (57). Growth factor-induced cell survival is dependent on the activation of PI3K and its downstream effector, Akt (18). Akt phosphorylates several intracellular substrates, thus affecting cell survival and programmed cell death (5,18). GSK3␤ has been previously identified as one of the main substrates for the PI3K-Akt pathway (18,54). Overexpression of catalytically active GSK3␤ in neuronal cell lines results in apoptosis, while dominant negative GSK3␤ prevents cell death following the phosphorylation-mediated inhibition by the PI3K-Akt cascade (18). Similarly, recent studies have shown that, in primary neuronal cultures, apoptosis induced by withdrawal of trophic factors or by PI3K inhibition is dependent on GSK3␤ activation (57). These studies also show that both the expression of an inhibitory GSK3␤-binding protein or a dominant-interfering form of GSK3␤ reduced neuronal apoptosis (57). In contrast, expression of a mutant ␤-catenin, not affected by GSK3␤ activity, did not protect against apoptosis. Thus, although stabilization of ␤-catenin as a result of GSK3␤ inactivation is an important physiological effect, several other pathways downstream of GSK3␤ might regulate cell fate (57). Taken together these results suggest that the neuroprotective effects of FGF1 involve a PI3K-Akt-mediated inactivation of GSK3␤.
In conclusion, FGF1 neuroprotective effects might involve inactivation of GSK3␤ by a pathway involving activation of PI3K-Akt cascades (Fig. 8). To the best of our knowledge, this is the first study showing that FGF1 is capable of acting on these intracellular pathways to enhance cell viability. This finding is significant because alterations in FGF1 and GSK3␤ are being recognized as important in the pathogenesis of neurodegenerative disorders such as Alzheimer's disease and HIV encephalitis.