The role of phosphatidylinositol 3-kinase in neural cell adhesion molecule-mediated neuronal differentiation and survival

Authors


Address correspondence and reprint requests to Elisabeth Bock, The Protein Laboratory, Institute of Molecular Pathology, University of Copenhagen, Panum Institute 6.2., Blegdamsvej 3, DK-2200 Copenhagen N, Denmark. E-mail: bock@plab.ku.dk

Abstract

The neural cell adhesion molecule, NCAM, is known to stimulate neurite outgrowth from primary neurones and PC12 cells presumably through signalling pathways involving the fibroblast growth factor receptor (FGFR), protein kinase A (PKA), protein kinase C (PKC), the Ras-mitogen activated protein kinase (MAPK) pathway and an increase in intracellular Ca2+ levels. Stimulation of neurones with the synthetic NCAM-ligand, C3, induces neurite outgrowth through signalling pathways similar to the pathways activated through physiological, homophilic NCAM-stimulation. We present here data indicating that phosphatidylinositol 3-kinase (PI3K) is required for NCAM-mediated neurite outgrowth from PC12-E2 cells and from cerebellar and dopaminergic neurones in primary culture, and that the thr/ser kinase Akt/protein kinase B (PKB) is phosphorylated downstream of PI3K after stimulation with C3. Moreover, we present data indicating a survival-promoting effect of NCAM-stimulation by C3 on cerebellar and dopaminergic neurones induced to undergo apoptosis. This protective effect of C3 included an inhibition of both DNA-fragmentation and caspase-3 activation. The survival-promoting effect of NCAM-stimulation was also shown to be dependent on PI3K.

Abbreviations used
BDNF

brain-derived neurotrophic factor

BME

basal modified Eagle's

BSA

bovine serum albumin

CGN

cerebellar granule neurones

DMEM

Dulbecco's modified Eagle's medium

EGFP

enhanced green fluorescent protein

FCS

fetal calf serum

FGFR

fibroblast growth factor receptor

GBSS

Gey's balanced salt solution

GDNF

glial cell line-derived neurotrophic factor

HBSS

Hank's balanced salt solution

HS

horse serum

Ig

immunoglobulin

IGF-1

insulin-like growth factor 1

MAPK

mitogen-activated protein kinase

NCAM

neural cell adhesion molecule

NGF

nerve growth factor

6-OHDA

6-hydroxydopamine

PDGFR

platelet-derived growth factor receptor

PI3K

phosphatidylinositol 3-kinase

PKA

protein kinase A

PKB

protein kinase B

PKC

protein kinase C

The neural cell adhesion molecule, NCAM, is a membrane-associated glycoprotein of the immunoglobulin (Ig) superfamily highly expressed by neurones and glial cells throughout the nervous system. NCAM is of importance in numerous processes involved in the formation and maintenance of the nervous system. Thus, NCAM plays a pivotal role in neural development, regeneration and synaptic plasticity associated with learning and memory consolidation (Rønn et al. 1998, 2000a). Mice with a targeted deletion of the gene encoding NCAM exhibit impaired spatial memory (Cremer et al. 1994), and intracranial injections of antibodies interfering with normal NCAM-function results in inhibition of learning in rat and chicken (Doyle et al. 1992; Scholey et al. 1993). NCAM is encoded by a single gene, but a number of different isoforms are generated through alternative splicing, the three major isoforms being NCAM-120, NCAM-140 and NCAM-180. NCAM engages in homophilic (NCAM-NCAM) and heterophilic binding (NCAM binding to, e.g. the neural cell adhesion molecule L1, the fibroblast growth factor receptor (FGFR), or components of the extracellular matrix). Homophilic binding has been shown to lead to activation of a number of intracellular signalling pathways. One well described and measurable result of homophilic NCAM-binding is the induction of neurite outgrowth through pathways involving activation of FGFR, protein kinase C (PKC), the Ras-mitogen activated protein kinase (MAPK) pathway, protein kinase A (PKA) and an increase in intracellular Ca2+ levels (reviewed by Berezin et al. 2000 and by Crossin and Krushel 2000).

NCAM interactions can be stimulated in vitro in different ways. Antibodies against the extracellular part of NCAM capable of clustering NCAM, recombinant proteins comprising parts of the extracellular domain of NCAM, or a coculture system in which neurones are grown on top of fibroblasts expressing NCAM, are often used. Recently, a synthetic peptide ligand, the C3-peptide, binding to the N-terminal IgI module of NCAM has been identified by means of combinatorial chemistry. C3 promotes neurite outgrowth from hippocampal neurones and PC12-E2 cells activating intracellular signalling pathways similar to the pathways activated by homophilic NCAM-binding (Rønn et al. 1999, 2000b).

Cell death plays a key role in normal neuronal development where 50% of the developing neurones are eliminated through programmed cell death, and in neurodegenerative conditions such as Parkinson's and Alzheimer's diseases. Therefore, it is of importance to understand how death and survival of neurones are controlled. Although, the focus has been on soluble neurotrophic factors such as nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF) and insulin-like growth factor 1 (IGF-1), there is increasing evidence indicating that cell-cell or cell-substratum adhesion also is important in neuronal survival. Thus, the neuronal cell adhesion molecule L1, has been reported to be involved in survival of dopaminergic (Hulley et al. 1998), cerebellar and hippocampal (Chen et al. 1999) neurones in culture, and integrin signalling has been reported to increase resistance of hippocampal neurones to apoptosis (Gary and Mattson 2001).

We present here data indicating that phosphatidylinositol 3-kinase (PI3K) plays a role in NCAM-mediated differentiation. Furthermore, we show for the first time that stimulation of NCAM by means of the peptide ligand, C3, prevents neuronal cell death. C3 was able to increase cell survival and to reduce DNA fragmentation and caspase-3 activation in neurones induced to undergo apoptosis. This neuroprotective effect was also dependent on PI3K. Moreover, we show that the serine/threonine kinase Akt/protein kinase B (PKB) was activated (phosphorylated) downstream from PI3K when neurones were stimulated with C3.

Materials and methods

Materials

The PI3K inhibitors LY294002 and wortmannin were purchased from Calbiochem (La Jolla, CA, USA). The C3 peptide (ASKKPKRNIKA) and a peptide comprising the sequence of the C3 peptide with two alanine substitutions C3ala2 (ASKKPAANIKA) were synthesized as dendrimers (C3d and C3ala2d) composed of four monomers coupled to a lysine backbone. The phosphorylated platelet-derived growth factor receptor (PDGFR)-derived peptide [SDGGY(P)MDMS] was synthesized in tandem with the Antennapedia internalization sequence (RQIKIWFQNRRMKWKK). All peptides were from Loke A/S (Aarhus, Denmark). IGF-1 and glial cell line-derived neurotrophic factor (GDNF) were purchased from Gibco BRL (Paisley, UK). 6-hydroxydopamine (OHDA) and sodium metabisulfite were from Sigma (St Louis, MO, USA). Monoclonal mouse antibodies against rat Thy-1 were purchased from Caltag Laboratory (Burlingame, CA, USA), and the polyclonal rabbit antibodies against rat GAP-43 were from Chemicon Int. Inc. (Temecula, CA, USA). Alexa Fluor® goat anti-mouse and goat anti-rabbit antibodies were from Molecular Probes (Leiden, Netherlands). Monoclonal mouse antibodies against tyrosine hydroxylase were from Boehringer Mannheim (Mannheim, Germany), biotinylated sheep anti-mouse Ig antibodies were from Amersham Pharmacia Biotech (Buckinghamshire, UK) and peroxidase conjugated streptavidin was from Dako (Glostrup, Denmark). Polyclonal rabbit antibodies against phospho-Akt (Ser473) and horseradish peroxidase (HRP)-conjugated goat anti-rabbit antibodies were from New England Biolabs (Beverly, MA, USA). The ApoAlert™ DNA fragmentation kit and the ApoAlert® Caspase-3 colorimetric assay kit were from Clontech (Palo Alto, CA, USA). All cell culture plastic was from Nunc (Roskilde, Denmark).

Cell culture

PC12-E2 cells

The PC12-E2 cell line (Wu and Bradshaw 1995) was a gift from Dr Klaus Seedorf, Hagedorn Research Institute, Gentofte, Denmark. PC12-E2 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% (v/v) fetal calf serum (FCS), 10% (v/v) horse serum (HS), 1% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin (Gibco BRL, Paisley, UK). The two subclones of the mouse fibroblastoid cell line L929, LBN110 and LVN212, which are stably transfected with the eukaryotic expression vector pHβ-Apr-1-neo (Gunning et al. 1987) containing full-length cDNA encoding human 140 kDa NCAM (without the exons VASE, a, b, c and AAG) and the vector alone (Kasper et al. 1996), respectively, were maintained in DMEM supplemented with 10% (v/v) FCS, 1% (v/v) glutamax, 100 U/mL penicillin, 100 µg/mL streptomycin, 2.5 µg/mL fungizone, 300 µg/mL neomycin (Gibco BRL, Paisley, UK).

Cerebellar granule neurones

Cerebellar granule neurones (CGN) were prepared from 3- to 4-day-old rats, essentially as previously described by Schousboe et al. (1989). Briefly, the cerebella were cleared of meninges and blood vessels, roughly homogenized by chopping and thereafter trypsinized. The neurones were washed in the presence of DNAse 1 and soybean trypsin inhibitor (Sigma) and then plated on poly-l-lysine coated microtitre plates in basal modified Eagle's (BME) medium supplemented with 10% (v/v) FCS, 1% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin, 3.5 g d-glucose/L, 1% (v/v) sodium pyruvate and 40 mm KCl, or Neurobasal™-A medium supplemented with 2% (v/v) B27, 0.5% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin (Gibco BRL) and 40 mm KCl. Twenty-four hours after plating, cytosine-β-d-arabinofuranoside (Sigma) was added to a concentration of 10 µm to avoid proliferation of glial cells after which the neurones were left to differentiate for another 5 days.

Dopaminergic neurones

Dopaminergic neurones were prepared from embryonic day 14 (E14) rats. A pregnant rat was killed and the uterus was excised and kept on ice in Hank's balanced salt solution (HBSS) (Gibco BRL). The ventral part of the mesencephalon was dissected from the foetuses and kept on ice in Gey's balanced salt solution (GBSS) (Gibco BRL) supplemented with 5 g glucose/L. Dissociated cell cultures were established using the same procedure as described for CGN above. The neurones were plated on poly-d-lysine coated cell culture plates in serum contained medium (25% (v/v) HS, 50% (v/v) Optimem 1, 25% (v/v) HBSS (Gibco BRL), and 5 g glucose/L). After 1–2 h, the medium was changed to Neurobasal™ (Gibco BRL) supplemented with 2% (v/v) B27, 0.5% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin.

All animals were handled in accordance with the national guidelines for animal welfare. All cells were routinely kept at 37°C in a humidified atmosphere containing 5% CO2.

Neurite outgrowth assays

PC12-E2 cells and CGN

For coculture experiments PC12-E2 cells or CGN were seeded at a low density (4000–9000 cells/cm2) on top of confluent monolayers of LVN212 or LBN110 cells and grown for 24 or 17 h (PC12-E2 and CGN, respectively) in DMEM supplemented with 1% (v/v) HS, 1% (v/v) FCS, 1% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin (PC12-E2 cells), or in Neurobasal™ medium supplemented with 2% (v/v) B27, 100 U/mL penicillin and 100 µg/mL streptomycin, 1% (v/v) glutamax, 4 g bovine serum albumin (BSA)/L, 2% HEPES (Gibco BRL; CGN). The cells were fixed with 4% (v/v) formaldehyde and immunostained using primary antibodies against Thy-1 or GAP-43 (PC12-E2 and CGN, respectively), and secondary Alexa Fluor® antibodies, alternatively PC12-E2 cells were transfected with plasmids containing enhanced green fluorescent protein (EGFP) in order to visualize cells. For single cell culture experiments, PC12-E2 cells were grown on fibronectin coated cell culture plates for 48 h in DMEM supplemented with 5% (v/v) FCS, 10% (v/v) HS, 1% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin, cells were fixed with 4% (v/v) formaldehyde and stained with Coomassie [6 g/L Coomassie brilliant blue (Sigma), 45% (v/v) ethanol, 10% (v/v) acetic acid]. Images of 100–300 cells were grabbed for each group in each experiment in a systematic series of fields of view as previously described (Rønn et al. 2000c) by computer-assisted fluorescence microscopy using a Nikon Diaphot inverted microscope with a Nikon Plan 20 × objective (Nikon, Tokyo, Japan), a video camera (Grundig Electronics) and the software package Prima created at the Protein Laboratory (Copenhagen, Denmark). Neurite length was estimated using a stereological procedure (Rønn et al. 2000c).

Dopaminergic neurones  Ventral mesencephalic cells were plated at a density of 100 000 cells/cm2 in poly-d-lysine coated 24-well cell culture plates and left to adhere for 1–2 h in serum containing medium before the medium was changed to Neurobasal™ as described above. After 48 h the cells were fixed and dopaminergic neurones were visualized by immunostaining using primary mouse monoclonal antibodies against tyrosine hydroxylase and secondary biotinylated sheep anti-mouse Ig antibodies followed by peroxidase conjugated streptavidin. Ninety-eight pictures for each group in each experiment were automatically taken employing a motorized, movable microscope stage (LUDL Electronic Production Ltd, LEP) using a 10 × objective and a video 1100 analogue b/w CCD video camera (DFA, Denmark) as previously described (Walmod et al. 2002). Neurite length was estimated as described above.

Survival assays

CGN

CGN were plated at a density of 290 000 neurones/cm2 in BME supplemented with 10% (v/v) FCS, 1% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin, 3.5 g d-glucose/L, 1% (v/v) sodium pyruvate and 40 mm KCl on poly-l-lysine coated 96-well cell culture plates, and 24 h after plating cytosine-β-d-arabinofuranoside was added to a concentration of 10 µm to avoid proliferation of glial cells, after which the neurones were left to differentiate for another 5 days. Apoptosis was induced by changing the medium to BME without serum containing only 5 mm KCl. The number of viable neurones was estimated using the CellTiter 96® assay (Promega, Madison, WI, USA) 48 h after induction of apoptosis.

Dopaminergic neurones

Ventral mesencephalic cells were plated at a density of 150 000 cells/cm2 in poly-d-lysine coated 24-well cell culture plates and left to adhere for 1–2 h in serum containing medium before the medium was changed to Neurobasal™ supplemented with 2% (v/v) B27, 0.5% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin. The neurones were left to differentiate for 6 days after which they were exposed for 2 h to medium containing 6-hydroxydopamine (6-OHDA) at a concentration of 100 µm. A stock solution was prepared by dissolving 6-OHDA to a concentration of 10 mm in 0.1% (w/v) sodium metabisulfite in order to prevent oxidation. Subsequently the medium was changed to fresh Neurobasal™ medium, and after 24 h the cultures were fixed and immunostained for tyrosine hydroxylase. Ninety-eight images of each group in each experiment were grabbed using a computer controlled motorized microscope stage as described for the neurite outgrowth assay.

TUNEL assay

CGN were plated at a density of 60 000 neurones/cm2 in Neurobasal™-A medium supplemented with 2% (v/v) B27, 0.5% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin on poly-l-lysine coated eight-well chamber slides and treated as described for the survival assay. Twenty-four hours after induction of apoptosis, DNA-fragmentation was detected using the ApoAlert™ DNA fragmentation assay kit essentially as described by the manufacturer. Briefly, blunt ends of double-stranded DNA molecules were enzymatically labeled with fluorescein-dUTP, and all cells were stained with propidium iodide (PI) at a concentration of 750 ng/mL. Images of a least 200 neurones were taken for each group in each experiment using a Biorad Radiance laser scanning system 2000 and a Nikon Eclipse TE 200 confocal microscope equipped with an oil immersion 60 × 1.4 NA objective (Nikon, Tokyo, Japan), and the fraction of TUNEL-reactive cells was determined.

Caspase-3 assay

CGN were plated at a density of approximately 290 000 neurones/cm2 in Neurobasal™-A medium supplemented with 2% (v/v) B27, 0.5% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin on poly-l-lysine coated cell culture dishes and treated as described for the survival assay. Eight hours after induction of apoptosis, the activity of caspase-3 was estimated using the ApoAlert® Caspase-3 colorimetric assay kit as described by the manufacturer.

PACE assay

Detection of phosphorylated Akt relative to total number of CGN was carried out largely according to Versteeg et al. (2000). Briefly, CGN were plated at a density of 290 000 neurones/cm2 in 96-well microtitre plates directly in serum-free BME supplemented with 1% (v/v) glutamax, 100 U/mL penicillin and 100 µg/mL streptomycin, 3.5 g d-glucose/L, 1% (v/v) sodium pyruvate and 5 mm KCl. Two to three hours later, PI3K inhibitors were added and at 4 h the neurones were stimulated with 5–10 µm C3 or 50 ng/mL IGF-1 for 30 min The plates were centrifuged at 70 gfor 10 min, fixed in 4% (v/v) formaldehyde and stained using polyclonal antibodies against Akt phosphorylated at Ser473 and peroxidase-conjugated secondary antibodies. The total number of cells was estimated by staining with crystal violet.

Statistics

Statistical evaluation was performed by employing the paired t-test using the commercially available software package fig.-p, version 2.98 (Biosoft, Cambridge, UK).

Results

PI3K is required for NCAM-mediated neurite extension

In order to investigate whether PI3K is required for NCAM-mediated neurite extension, we tested the effect of the pharmacological PI3K inhibitors LY294002 and wortmannin on neurite outgrowth induced by physiological, homophilic NCAM-binding in both PC12-E2 cells and CGN. In dopaminergic neurones, NCAM stimulation was achieved using the synthetic dendrimeric peptide ligand of NCAM, C3d (Rønn et al. 1999). PC12-E2 cells or CGN were cultured for 24 and 17 h, respectively, on top of a confluent monolayer of control, NCAM-negative fibroblasts, or NCAM-positive fibroblasts, and subsequently neurite length was estimated. As shown in Fig. 1, PC12-E2 cells (Figs 1a and b) and CGN (Figs 1c and d) extended longer neurites when cultured on top of NCAM-expressing fibroblasts than when grown on top of control fibroblasts. Rat dopaminergic neurones stimulated with C3d for 48 h likewise extended longer neurites as compared with neurones not exposed to the NCAM-mimetic (Figs 1e and f).

Figure 1.

Effect of NCAM-homophilic binding (a–d) and C3d (e and f) on neurite outgrowth from PC12-E2 cells (a and b), CGN (c and d) and dopaminergic neurones (e and f). PC12-E2 cells were grown for 24 h on confluent monolayers of fibroblasts without (a) or with (b) expression of human NCAM-140 and subsequently immunostained for Thy-1. Likewise, cerebellar neurones were grown for 17 h on confluent monolayers of fibroblasts without (c) or with (d) expression of human NCAM-140 and subsequently immunostained for GAP-43. Mesencephalic neurones were grown on poly-d-lysine-coated cell culture plates for 48 h in the absence (e) or presence (f) of C3d and subsequently immunostained for tyrosine hydroxylase. Scale bars = 20 µm.

The effect of homophilic NCAM-binding on neurite outgrowth from PC12-E2 cells was inhibited in a dose-dependent manner by addition of the PI3K-inhibitor LY294002, whereas this compound did not affect neurite outgrowth under control conditions (Fig. 2a). Treatment with wortmannin, a second inhibitor of PI3K likewise inhibited the NCAM-induced neuritogenic response in PC12-E2 cells, whereas it had no significant effect under control conditions (Fig. 2b). Thus, two different inhibitors of PI3K inhibited NCAM-mediated neurite outgrowth in PC12-E2 cells, indicating a role for PI3K in this process.

Figure 2.

Effect of inhibitors of PI3K on NCAM- or C3d-induced neurite outgrowth in PC12-E2, cerebellar, or dopaminergic neurones. Results from at least four independent experiments are in all cases expressed as a percentage ± SEM, with untreated controls set at 100%, corresponding to an average neurite length of 51, 15, and 170 µm, for PC12-E2, cerebellar, and dopaminergic cells, respectively. +p < 0.05, ++p < 0.01, +++p < 0.005 compared with untreated controls; *p < 0.05, **p < 0.01, ***p < 0.005 compared with NCAM- or C3d-stimulated controls. (a) and (b) PC12-E2 cells were grown for 24 h on confluent monolayers of fibroblasts without (○) or with (●) expression of human NCAM-140 in the presence of 0, 3.5, 7, or 10 µm LY294002 (a) or 0, 100, 300, or 1000 nm wortmannin (b), and subsequently immunostained for Thy-1 (b) or visualized through transfection with EGFP (a). Wortmannin is unstable in cell culture medium, for which reason fresh wortmannin was re-added every 5 h. (c) Cerebellar neurones were grown for 17 h on confluent monolayers of fibroblasts without (○) or with (•) expression of human NCAM-140 in the presence of 0, 7 or 10 µm LY294002 and subsequently immunostained for GAP-43. (d) Ventral mesencephalic neurones were grown on poly-d-lysine-coated cell culture plates for 48 h in the presence of 5 µm of C3d (•), GDNF (□) as a positive control or medium alone (○) and in the presence of 0, 3.5, 7, 10 or 14 µm LY294002. The cells were subsequently immunostained for tyrosine hydroxylase.

The dependence of PI3K activation in NCAM-mediated neurite extension was also evaluated in cultures of primary neurones. In Fig. 2(c), it can be seen that a concentration of 10 µm LY294002 significantly inhibited neurite outgrowth from rat CGN grown on NCAM-expressing fibroblasts, while there was no inhibition of neurite outgrowth from CGN grown on NCAM-negative fibroblasts. We also tested the effect of inhibiting PI3K in rat dopaminergic neurones stimulated with C3d. LY294002 inhibited neurite outgrowth from dopaminergic neurones stimulated with C3d or GDNF in a dose-dependent manner already detectable at the lowest concentration employed (3.5 µm). However, LY294002 also had an inhibitory effect on non-stimulated neurones, although only at a concentration of 7 µm or above (Fig. 2d). Thus, PI3K seems to be necessary for NCAM-mediated neurite extension in three different neuronal cell types.

In order to further evaluate the role of PI3K in neuronal differentiation, we tested the effect of a synthetic peptide derived from the high affinity binding site for the p85-subunit of PI3K on the activated platelet-derived growth factor receptor (PDGFR). The peptide, which is attached to an internalization sequence derived from the third helix of the Antennapedia protein (Derossi et al. 1998), is internalized in living cells, where it interacts with the p85 subunit of PI3K leading to its activation. Addition of the PDGFR-peptide to PC12-E2 cells resulted in a significant increase in neurite outgrowth (Fig. 3). Thus, the direct activation of the endogenous pool of PI3K in PC12-E2 cells was enough to induce a neuritogenic response, supporting the contention that PI3K-activation results in neurite extension in PC12-E2 cells. Taken together, the presented data indicate that PI3K is involved in NCAM-mediated neuronal differentiation.

Figure 3.

Effect of the PDGFR-peptide on neurite outgrowth from PC12-E2 cells in single cell culture. PC12-E2 cells were grown on fibronectin coated cell culture plates for 48 h in the presence of 0, 10, 50 or 100 µg/mL PDGFR-peptide and subsequently stained with Coomassie. Results from eight independent experiments are expressed as a percentage ± SEM, with untreated controls set at 100%, corresponding to an average neurite length of 32 µm. *p < 0.05 as compared with untreated control.

The synthetic peptide ligand of NCAM, C3d, promotes neuronal cell survival

Homophilic NCAM-binding induces a neuritogenic response in a variety of neuronal cell types, and as shown above, PI3K is necessary for this response. As PI3K primarily is known for its role in cell survival, our findings raised the following questions: Does NCAM-stimulation promote cell survival, and, if so, is PI3K involved?

To address the first question, the effect of C3d on survival of primary neurones was tested in two different model systems. In the first model, CGN from 3-day-old rats were induced to differentiate for 6 days in a high potassium-containing medium after which the neurones were grown in a low potassium-containing medium either with or without addition of IGF-1 or C3d. As shown in Fig. 4(a), apoptosis induced by reducing the concentration of potassium in the medium could be prevented by treatment with IGF-1. Likewise, treatment with C3d resulted in a significant increase in the number of surviving neurones after induction of apoptosis (Fig. 4b). The second model employed, consisted of embryonic rat dopaminergic neurones induced to undergo apoptosis by exposure to the neurotoxin 6-OHDA (Lotharius et al. 1999). Ventral mesencephalic neurones from E14 rat embryos were allowed to differentiate for 6 days and then exposed to 6-OHDA for 2 h followed by further incubation for 24 h. As shown in Fig. 4(c), the number of cells was strongly decreased after exposure to 6-OHDA as compared with non-treated cells, and addition of GDNF partially prevented this. When dopaminergic neurones were stimulated with 0.2–5 µm of C3d before and after induction of apoptosis, the number of surviving dopaminergic neurones increased considerably, but at a concentration of 14 µm, C3d decreased survival (Fig. 4d). Taken together, these data suggest that activation of NCAM by C3d induces neuronal survival.

Figure 4.

Effect of C3d on survival of cerebellar and dopaminergic neurones induced to undergo apoptosis. (a) and (b) Cerebellar neurones were allowed to differentiate for 6 days in a high potassium (40 mm KCl) medium before apoptosis was induced by changing the medium to a low potassium (5 mm KCl) medium. Forty-eight hours later, survival was estimated. (a) High KCl: cells not induced to undergo apoptosis. Low KCl: cells induced to undergo apoptosis. Low KCl + IGF: cells induced to undergo apoptosis in the presence of 50 ng/mL IGF-1. (b) Cells induced to undergo apoptosis and exposed to 0, 0.04, 0.2, 1, or 5 µm of C3d. (c) and (d) Ventral mesencephalic cells were allowed to differentiate for 6 days in the presence of 0, 0.2, 1, 5, or 18 µm of C3d, after which they were exposed to 100 µm of 6-OHDA for 2 h. After another 24 h in fresh medium containing C3d, the cells were fixed and immunostained for tyrosine hydroxylase and the number of surviving neurones were counted in 98 automatically taken pictures of each group in each experiment. (c) Control: cells not exposed to 6-OHDA, 6-OHDA: cells exposed to 6-OHDA, 6-OHDA + GDNF: cells exposed to 6-OHDA and 10 ng/mL GDNF. (d) Cells exposed to 6-OHDA and 0, 0.2, 1, 5, or 10 µm of C3d. Results from four to five independent experiments are expressed as a percentage ± SEM, with the cultures induced to undergo apoptosis set at 100%. +p < 0.05, ++++p < 0.001 compared with untreated control; *p < 0.05, ****p < 0.001 compared with cultures induced to undergo apoptosis.

C3d protects neurones against apoptosis

We wanted to test whether the C3d-mediated induction of survival involves interference with the apoptosis associated events, DNA-fragmentation and activation of the caspase cascade. Although the survival promoting effect was largest in dopaminergic neurones, these studies were carried out in CGN, since these cultures contain more than 90% CGN (Schousboe et al. 1989) whereas the ventral mesencephalic cultures contain less than 10% dopaminergic neurones (Hulley et al. 1998).

Death of CGN induced by withdrawal of high KCl concentrations is due to induction of apoptosis (Gerhardt et al. 2001; Li et al. 2001) and, in Fig. 5, it can be seen that after withdrawal of high KCl CGN underwent apoptosis with a large proportion of the cells displaying DNA-fragmentation as reflected by a positive TUNEL staining (Fig. 5a, yellow nuclei). In contrast, CGN maintained at high KCl concentration contained very few cells displaying DNA-fragmentation (Fig. 5b). In CGN cultures treated with 0.3 µm C3d, very little DNA-fragmentation was observed after withdrawal of high KCl (Fig. 5c) whereas treatment with a control peptide, C3ala2d, had no protective effect (Fig. 5d). The effect of C3d and the control peptide on DNA-fragmentation was subsequently quantified. From Fig. 5(e), it appears that withdrawal of high KCl resulted in an increase of 150% in the number of TUNEL-positive neurones. C3ala2d at a concentration of 3 µm had no effect on DNA-fragmentation, whereas C3d markedly decreased the number of TUNEL-positive CGN after withdrawal of high KCl (Fig. 5f). The effect of C3d exhibited a reversed bell-shaped dose-dependent relationship, the concentration of 3 µm being the most efficient, inhibiting DNA-fragmentation by 40%.

Figure 5.

Examples of DNA-fragmentation in cerebellar neurones induced to undergo apoptosis by shifting from high potassium (40 mm KCl) to low potassium medium (5 mm KCl) after 6 days of differentiation in a high potassium medium. (a–d) Confocal micrographs of representative double stained (TUNEL, green; propidium iodide, red) neurones are shown here, scale bar = 20 µm. Cells reacting with both TUNEL and propidium iodide appear yellow. (a) Neurones induced to undergo apoptosis. (b) Neurones maintained at high potassium. (c) Neurones induced to undergo apoptosis and simultaneously exposed to C3d (3 µm). (d) Neurones induced to undergo apoptosis and exposed to the control peptide C3d2ala (3 µm). (e and f) Results from four or more independent experiments are expressed as the percentage of cells showing DNA-fragmentation relative to the total number of cells ± SEM. (e) High KCl: cells not induced to undergo apoptosis. Low KCl: cells induced to undergo apoptosis. Low KCl + C3d2ala: cells induced to undergo apoptosis and exposed to the control peptide, C3d2ala (3 µm). (f) Cells induced to undergo apoptosis and exposed to 0, 0.03, 0.1, 0.3, 3, 5, 10 µm of C3d. *p < 0.05, **p < 0.01, ****p < 0.001 compared with high KCl (a) or low KCl (b).

Subsequently, we measured the activity of the apoptosis indicator, caspase-3. Withdrawal of high KCl resulted in a strong (300%) increase in caspase-3 activity under the chosen conditions (Fig. 6a). C3d inhibited this activation by 30% (Fig. 6b).

Figure 6.

Effect of C3d on caspase-3 activity in cerebellar neurones induced to undergo apoptosis. Results from six independent experiments are expressed as a percentage ± SEM, with the neurones induced to undergo apoptosis set at 100%. ++p < 0.01 compared with non-apoptotic neurones; *p < 0.05 compared with apoptotic neurones.

Thus, our results indicate that the NCAM mimetic C3d protects CGN from apoptosis.

NCAM mediates neuronal cell survival through PI3K

To determine whether NCAM-mediated survival is dependent on PI3K, we evaluated the effect of the PI3K inhibitor LY294002 on C3d-promoted survival of dopaminergic neurones exposed to 6-OHDA. For this experiment the dopaminergic neurones were chosen since the survival effect of C3d was most predominant in this system. From Fig. 7, it appears that LY294002 inhibited the survival-promoting effect of C3d to almost control levels, indicating that PI3K activation is involved in the survival of dopaminergic neurones induced by C3d.

Figure 7.

Effect of LY294002 on C3d-induced survival of dopaminergic neurones induced to undergo apoptosis by treatment with 6-OHDA. Ventral mesencephalic cells were allowed to differentiate for 6 days in the presence of 10 µm C3d after which they were exposed to 100 µm of 6-OHDA for 2 h. After another 24 h in fresh medium containing 10 µm C3d the cells were fixed and immunostained for tyrosine hydroxylase and the number of surviving neurones were counted in 98 automatically taken pictures of each group in each experiment. Results from five independent experiments are expressed as a percentage ± SEM, with the negative control (neurones induced to undergo apoptosis) set at 100%. *p < 0.05 compared with negative controls.

NCAM-induced signal transduction results in PI3K-dependent Akt phosphorylation

A major downstream mediator of PI3K-signalling is the serine/threonine protein kinase Akt. Akt is activated by phosphorylation on Thr308 and Ser473 after binding to the lipid products of PI3K. We therefore tested whether Akt was activated in response to NCAM-stimulation. CGN were stimulated with C3d, and the phosphorylation state of Akt was thereafter analysed using a phosphospecific antibody cell-based ELISA (PACE; Versteeg et al. 2000). As shown in Fig. 8, treatment of CGN with C3d or IGF-1 for 30 min led to a marked increase in phosphorylation of Akt. This phosphorylation was abolished by preincubating the neurones with either of the two PI3K inhibitors, LY294002 or wortmannin (Figs 8a and b, respectively), indicating that Akt is activated downstream of PI3K in NCAM-mediated signalling.

Figure 8.

Effect of LY294002 and wortmannin on IGF-1- or C3d-induced phosphorylation of Akt. Cerebellar neurones were starved for 4 h followed by stimulation with 50 ng/mL IGF-1 (black bars), 5 or 10 µm (b and a, respectively) of C3d (hatched bars), or medium alone (white bars). 0, 10, 14 µm of LY294002 (a) or 0, 100, 300 nm of wortmannin (b) was added after 2–3 h of serum starvation. Results from four (a) or five (b) independent experiments are shown as a percentage ± SEM, with untreated controls set at 100%. +p < 0.05, ++p < 0.01, +++p < 0.005, ++++p < 0.001 compared with unstimulated control; *p < 0.05, **p < 0.01, ***p < 0.005 compared with stimulated control.

Discussion

The role for PI3K in neuronal survival is well known and has been described in many contexts (reviewed by Brunet et al. 2001), whereas data pointing to a requirement for PI3K in neuronal differentiation are limited. Neurite outgrowth stimulated by NGF in PC12 cells has been shown to be inhibited by wortmannin (Kimura et al. 1994) and by expression of a dominant negative p85 subunit of PI3K (Jackson et al. 1996). Moreover, differentiation of dopaminergic neurones induced by GDNF is inhibited by wortmannin (Pong et al. 1998). We show here that PI3K is necessary for NCAM-mediated neurite extension in two types of primary neurones, cerebellar and dopaminergic neurones, and in PC12-E2 cells. Interestingly, PI3K also seems important for neurite extension in non-stimulated dopaminergic neurones. This may be due to a low level of homophilic NCAM interactions between cells in the semiconfluent mesencephalic culture.

It has previously been reported that expression of a constitutively active PI3K (Kobayashi et al. 1997) or microinjection of activated PI3K (Kita et al. 1998) induce neurite outgrowth from PC12 cells. These reports, however, do not address the question whether activation of the endogenous pool of PI3K is sufficient for induction of neurites in PC12 cells or whether the induction merely is a result of a non-physiological strong activation of a number of downstream mediators. This question was addressed in the present report, where we show that stimulation of the endogenous pool of PI3K employing the internalized PDGFR-peptide leads to induction of neurite outgrowth. The PDGFR-peptide has previously been reported to bind to the p85 subunit of PI3K in vivo leading to activation of PI3K (Derossi et al. 1998).

Furthermore, we present here data indicating a survival-promoting effect of the synthetic NCAM-ligand, C3d. Treatment with C3d leads to increased survival in two different systems of apoptosis induction, employing cerebellar and dopaminergic neurones, respectively. In addition, we show that in the former system, the neurones exhibit DNA fragmentation and that the apoptotic process is caspase-dependent, indicating that signalling through NCAM promotes neuronal survival by preventing apoptosis. Other cell adhesion molecules have previously been implicated in neuronal survival, e.g. another member of the immunoglobulin superfamily, the neuronal cell adhesion molecule L1, has been shown to promote survival of dopaminergic neurones. Antibodies against L1 increased the survival of dopaminergic neurones in culture, even though they failed to protect against the neurotoxin MPP+ (Hulley et al. 1998). In another study, the extracellular domain of L1 fused to the Fc part of human Ig (L1-Fc) exerted a positive effect on survival of cerebellar and hippocampal neurones in culture, whereas NCAM-Fc had no effect on neuronal survival (Chen et al. 1999). A possible explanation for the lack of effect of NCAM on survival in this study may be that homophilic binding between the applied fusion proteins is occupying binding sites required for the neuroprotective effect.

We show that the NCAM-ligand C3d has a survival-promoting effect on both cerebellar and dopaminergic neurones induced to undergo apoptosis, and that PI3K is necessary for this effect. As C3d has been shown bind to NCAM and to induce neurite outgrowth (Rønn et al. 2000b), these data indirectly point to a new role of NCAM as a neuronal survival factor. Thus, while previously the focus has been on survival induction through soluble neurotrophic factors such as NGF, IGF-1 and BDNF, our data point to an equally important role for cell–cell adhesion molecules in neuroprotection.

Akt is a common downstream target of PI3K and the majority of Akt substrates are implicated in either survival or in insulin metabolism (reviewed by Coffer et al. 1998; by Datta et al. 1999). We show here that Akt is activated when CGN are stimulated with C3d, and that Akt is located downstream of PI3K in this pathway.

Whether Akt is involved in NCAM-mediated differentiation or survival, or both, cannot be answered at present. There are several possible mechanisms by which activation of NCAM can lead to activation of PI3K and Akt. One possible mechanism involves the FGFR which is activated upon homophilic NCAM-binding (reviewed in Berezin et al. 2000). The FGFR has been shown to induce PI3K and Akt activation through the recruitment of multiple docking proteins. Gab1 is recruited to the activated FGFR via the FRS2-Grb2 complex, leading to tyrosine phosphorylation of Gab1, thereby generating binding sites for the SH2-domain of the p85 subunit of PI3K (Ong et al. 2001). Another possible mechanism involves FAK which is phosphorylated upon binding to the activated NCAM-Fyn complex (Beggs et al. 1997). PI3K has been shown to bind via the SH2-domain of the p85 subunit to phosphorylated FAK, and this binding leads to activation of PI3K (Chen et al. 1996). The collaboration of Ras, which also is activated upon homophilic NCAM-binding, might be required for activation of PI3K through FAK or FGFR, as this has been reported to be the case for activation through the PDGFR (Klinghoffer et al. 1996).

In conclusion, PI3K is necessary for both neurite extension and neuroprotection induced by NCAM-stimulation, and Akt is activated downstream of PI3K following NCAM-activation.

Acknowledgements

The financial support by the EU-biotech programme, the Danish Cancer Society, the Danish Medical Research Council and the Lundbeck Foundation is gratefully acknowledged.

Ancillary