Activated Phosphatidylinositol 3-Kinase and Akt Kinase Promote Survival of Superior Cervical Neurons

SCG neurons were removed from newborn rat pups and dissociated in 0.025% trypsin for 30 min after which an equal volume of 0.4% collagenase (Worthington, Reading, UK) was added for a further 30 min. The cells were triturated with a 19-gauge needle until a single cell suspension was obtained and preplated for 2 h to allow non-neuronal cells to attach (Deckwerth and Johnson, 1993). The neurons were gently rinsed off, centrifuged, and resuspended in culture medium. Neurons were plated on poly-l-lysine/laminin–coated 13-mm glass coverslips at 8 × 10³ cells/coverslip. The cell culture medium was DME (GIBCO BRL Laboratories, Paisley, UK) containing 10% fetal calf serum (Globepharm Limited, Surrey, UK), 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM glutamine, and 100 ng/ml NGF (Promega, Southampton, UK). To limit the growth of any contaminating non-neuronal cells, 20 μM fluorodeoxyuridine, and 20 μM uridine (Sigma Chemical Co., Dorset, UK) were added. Cells were cultured for 5 to 8 d before use in a humidified 10% CO₂ incubator at 37°C.

Microinjection was carried out using a microscope (Axiovert 135M; Zeiss, Inc., Thornwood, NY) with an Eppendorf transjector (model 517; Campden Instruments, Leicester, UK) and micromanipulator (model 5246; Campden Instruments). Microinjection needles were pulled from glass capillaries using a horizontal electrode puller (model 773; Campden Instruments). DNA was injected into the nucleus in 0.5× PBS with either 5 μg/ml of purified guinea pig IgG (Sigma Chemical Co.) for cell staining experiments or 5 μg/ml of neutral 70 kD Texas red dextran (Molecular Probes, Eugene, OR) for cell survival experiments. We found some protective effect (∼20%) due to injection of plasmid DNA and Texas red dextran and therefore used equivalent DNA concentrations in all microinjection experiments.

After microinjection, cells were left 4–24 h to allow the protein to be expressed, and the number of Texas red dextran-positive cells was then counted (100% value). The cells were washed twice with DME and fresh medium lacking NGF but containing 100 ng/ml of anti-NGF antibody (Boehringer Mannheim, East-Sussex, UK). After 72 h the cells were incubated for 30 min with 1 μM calcein AM, which is converted to a green fluorescent derivative in live cells (Live/Dead kit; Molecular Probes). Cells that were positive for both Texas red dextran and calcein were counted as viable. In experiments that did not involve microinjection, calcein-positive cells in 10 random fields around the perimeter of the coverslip were counted at a magnification of 200. All experiments were done in a blinded manner.

In experiments in which LY294002 (Affiniti Research Products Ltd., Nottingham, UK) was used, 0.6% methyl cellulose was added to the medium to prevent cells from mechanically detaching (Hawrot and Patterson, 1979). Rapamycin was purchased from Calbiochem-Novochem (Nottingham, UK).

When analyzing the protein expression of plasmid-injected neurons, purified guinea pig IgG was co-injected. The cells were injected and stained 24 h later unless stated otherwise. Cells were generally fixed in 3% paraformaldehyde for 20 min, permeabilized for 5 min in 0.5% Triton X-100, and blocked in 50% goat serum/50% antibody diluting buffer (1% BSA in PBS) for 30 min. Cells were incubated with primary antibodies for 1 h at room temperature, rinsed, and incubated in FITC-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) plus rhodamine-conjugated donkey anti–guinea pig IgG for a further 30 min. Cells were costained with 1 μg/ml of Hoechst 33342 for 5 min before mounting in Citifluor (Citifluor Ltd., Canterbury, UK). Primary antibodies included an anti–c-Jun antibody, raised against a gst–c-Jun fusion protein, an anti–phospho-c-Jun monoclonal antibody raised against the phosphopeptide amino acids 57–68 (phospho-serine 63) from the mouse c-Jun sequence, both provided by Dominique Lallemand (Pasteur Institute, Paris, France), an anti-p85 antibody, a gift from Ivor Gout (Ludwig Institute, London, UK), and an anti-myc antibody (9E10; Boehringer Mannheim). In the case of the anti–c-myc antibody, cells were fixed in 50% methanol/ 50% acetone and subsequently processed as above. Photographs of cells were taken on a fluorescent microscope (Microphot-FXA; Nikon) using Kodak 400 ASA film.

The PI 3-kinase plasmids used in this paper were p110* PI 3-K (constitutively active kinase) and p110*Δ PI 3-kinase (kinase dead construct) and have been previously described (Hu et al., 1995). The wild-type, K179M, and viral Akt constructs were supplied by Dr. Philip Tschillis (Fox Chase Cancer Center, Philadelphia, PA); the p85 construct was obtained from Dr. Len Stephens (Babraham Institute, Cambridge, UK) with the permission of Professor Kasuga (Kobe University School of Medicine, Kobe, Japan), and the Δ169 c-Jun and pCDbcl-2 plasmids were obtained from Dr. John Ham (Eisai Laboratories, London, UK).

To determine the effects of inhibiting PI 3-kinase in primary neurons, we incubated cells in 50 μM LY294002 for 3 d in the presence of NGF. We had found that this concentration of LY294002 inhibited neurite outgrowth in newly plated cells (data not shown). However we could detect no decrease in the survival of differentiated cells, using the Live/Dead assay after 72 h incubation in LY294002 (Fig. 1,A). We did not use wortmannin in these experiments since it has a half-life of ∼3 h. The gross morphology of cells grown in LY294002 appeared normal (Fig. 1 B), but they appeared slightly smaller, perhaps due to their decreased substrate attachment, which is consistent with a postulated role of PI 3-kinase in substrate adhesion (Serve et al., 1995).

As an alternative method for inhibiting the action of PI 3-kinase, we microinjected a dominant negative p85 construct. The canonical PI 3-kinase consists of a p110 catalytic subunit and a p85 regulatory unit which is important for receptor binding and interacting with other cellular proteins (Carpenter and Cantley, 1996). The dominant negative p85 lacks a 35-amino acid stretch within the inter-SH2 domain necessary for p110 binding and is able to bind to the TrkA receptor, but not to the PI 3-kinase p110 subunit (Hara et al., 1994). We found this modified p85 to be expressed throughout the cytoplasm (Fig. 2,A), but it did not induce apoptosis in SCG neurons in the presence of NGF (Fig. 2 B), even though it has been reported to block neurite outgrowth in NGF-stimulated PC12 cells (Jackson et al., 1996). These data indicate that the activity of PI 3-kinase is not essential for survival when cells are maintained in NGF.

A constitutively active myc-tagged version of PI 3-kinase (p110*), containing the p110 kinase subunit linked to the inter-SH2 domain of the p85 subunit (Hu et al., 1995), was used in these experiments. Expression of this protein was verified by staining with an anti-myc–tagged antibody (9E10), as shown in Fig. 3,A. For survival experiments, DNA was microinjected into 5- to 8-d-old neurons and the neurons withdrawn from NGF 16–24 h later, having allowed sufficient time for the protein to be expressed. After an additional 72 h, survival was quantified. Three control plasmids were used: pCDbcl-2, known to enhance SCG survival after NGF withdrawal (Garcia et al., 1992; Ham et al., 1995); the backbone vector pCG; and a kinase-dead version of PI 3-kinase, p110*Δ, in which amino acids 917– 950 of wild-type p110 had been removed. PI 3-kinase displayed a striking ability to support survival of NGF-deprived neurons, being approximately as effective as bcl-2. On the other hand, the kinase-dead variant was completely ineffective (Fig. 3,B). If we were indeed promoting survival by introducing activated PI 3-kinase, this effect should be completely reversed by the addition of the PI 3-kinase inhibitor, LY294002. As expected, when added at the time of NGF removal, LY294002 did block the protective effects of PI 3-kinase (Fig. 4). Hence our results indicate that, in the absence of NGF, expression of PI 3-kinase activity is sufficient to prevent apoptosis.

One of the kinases known to lie downstream of PI 3-kinase is Akt, which can be directly activated by products of PI 3-kinase (Franke et al., 1995, 1997). We injected wild-type (wt), viral (v) Akt, or a kinase-dead version, Akt K179M, into 5- to 8-d-old SCG neurons and confirmed that wtAkt and K179M Akt were expressed using an anti-HA antibody to the 3′ tag (data not shown). We found that wtAkt or vAkt maintained survival of SCG neurons withdrawn from NGF but that K179M Akt did not (Fig. 4). To ascertain whether the protection by Akt was specific to the PI 3-kinase pathway, we treated injected cells with LY294002. LY294002 prevented protection by injected PI 3-kinase (Fig. 4) but did not prevent survival promoted by the Akt constructs. p70^S6 kinase is known to lie downstream of PI 3-kinase and can be inhibited by the macrolide antibiotic rapamycin (Chung et al., 1992; Price et al., 1992; Soltoff et al., 1992). Rapamycin did not induce neuronal apoptosis when cells were maintained in NGF (data not shown) and did not inhibit the survival-promoting effect of Akt (Fig. 4).

We have previously shown that after the withdrawal of NGF from neurons there is a marked increase in the expression and phosphorylation of the transcription factor c-Jun (Ham et al., 1995; A. Watson, personal communication). This increase in expression and phosphorylation can be inhibited by a dominant negative form of c-Jun, Δ169 c-Jun, in which the transactivation domain has been deleted. This mutant can also protect the cells from apoptosis. We have therefore postulated that c-Jun plays an important role in the death process in neurons by initiating the transcription of genes essential for their death. To determine, therefore, whether the survival-promoting activity of PI 3-kinase could act by inhibiting the c-Jun pathway, we injected neurons with either activated PI 3-kinase, a vector control, or Δ169 c-Jun. NGF was withdrawn 16–24 h later and the cells stained for either c-Jun or phospho–c-Jun expression after an additional 24 h. The phospho–c-Jun antibody used was raised against a peptide containing phosphorylated serine 63 in the c-Jun sequence, a site known to be phosphorylated in activated c-Jun (Smeal et al., 1991). Activated PI 3-kinase was unable to inhibit either the increase in nuclear staining of c-Jun or of phosphorylated c-Jun that follows NGF withdrawal. However, Δ169 c-Jun prevented both changes as expected (Fig. 5, A–C). This suggests that PI 3-kinase is acting downstream of c-Jun or independently of it, and that apoptosis can be prevented even when c-Jun has been activated.

Recently there have been a number of reports examining the effects of PI 3-kinase inhibition on cell survival. These have included studies on cells of hemopoeitic origin that were maintained with different survival factors. It was noted that PI 3-kinase was important for survival of MC-9 cells when they were grown in IL-3 or -4, but not when they were grown in GMCSF or IL-5 (Scheid et al., 1995). However, other types of hemopoeitic cells were dependent on PI 3-kinase for survival when they were maintained in IGF-1, but not in IL-3 (Minshall et al., 1996). Further, the survival of other types of cells, such as oligodendrocytes and fibroblasts, seems to be mediated by PI 3-kinase under all growth conditions (Vemuri and McMorris, 1996; Yao and Cooper, 1996). PI 3-kinase activity has also been shown to be important in the survival of PC12 cells, a model system for sympathetic neurons (Rukenstein et al., 1991; Yao and Cooper, 1995), and most recently in cerebellar granule neurons grown in IGF-1 (Dudek et al., 1997). Thus it seems that multiple survival pathways accessed via different receptors are often present within cells and that their dependence upon PI 3-kinase cannot necessarily be predicted by the use of a particular survival agent.

To investigate the role of PI 3-kinase in a more physiologically relevant system than PC12 cells, we chose to study NGF-dependent SCG neurons, primary sympathetic neurons. PI 3-kinase activity can be inhibited in a number of ways, and we used two different methods. The first was the PI 3-kinase inhibitor LY294002, a reversible inhibitor that competes with ATP for binding to the enzyme (Vlahos et al., 1994). Addition of LY294002 to neurons did not induce apoptosis (Fig. 1), although this compound effectively inhibited PI 3-kinase, since it blocked the effects of over- expressed enzymatically active kinase (Fig. 4, and below). We also employed an alternative way of inhibiting PI 3-kinase by microinjecting a dominant negative form of p85. This construct lacks the region necessary for binding to the p110 subunit while retaining its ability to interact with the receptor or other docking molecules (Klippel et al., 1993; Hara et al., 1994). It is thus able to block endogenous PI 3-kinase activity. It has been shown to effectively inhibit PI 3-kinase in a number of cell types including CHO cells stimulated with insulin or EGF (Hara et al., 1994; Kido et al., 1995) and it inhibits NGF-induced neurite outgrowth in PC12 cells (Jackson et al., 1996). Dominant negative p85 also did not induce apoptosis in SCG neurons grown in NGF (Fig. 2 B) which, together with the inhibitor data, indicates that in the presence of NGF, PI 3-kinase activity is not necessary for survival of SCG neurons. This is in contrast to the results obtained with PC12 cells (Yao and Cooper, 1995), where PI 3-kinase inhibitors induced death, and highlights the importance of using primary cells.

We also studied the effects of over-expression of activated PI 3-kinase on the survival of NGF-deprived neurons. Surprisingly, this enzyme was effective in enhancing survival (Fig. 3,B). The enzymatic activity of this kinase was necessary, since the activated PI 3-kinase did not promote survival if cells were maintained in the presence of LY294002 and since a kinase-dead version of the construct did not promote survival (Figs. 3 and 4).

More recently, Akt kinase (Bellacosa et al., 1991; Coffer and Woodgett, 1991; Jones et al., 1991) has been shown to be stimulated by the products of PI 3-kinase (Franke et al., 1995, 1997; Klippel et al., 1997) after activation of receptor tyrosine kinases. When expressed in SCG neurons, wtAkt and vAkt were both protective after NGF withdrawal. Franke et al. (1995) found a low level of Akt kinase activity in the absence of PDGF after transfection with wtAkt into NIH 3T3 cells. Perhaps even in the absence of NGF there is sufficient PI 3-kinase activity to stimulate wtAkt kinase when Akt protein levels are high. Alternatively, overexpression of the protein could lead to an increase in membrane-localized Akt, which may lead to enhanced activity (Ahmed et al., 1993; Kulik et al., 1997). Interestingly, we find that K179MAkt, which has been shown to act as a dominant negative protein (Dudek et al., 1997), does not induce apoptosis in the presence of NGF, again suggesting that this pathway is not obligatory in the presence of NGF.

Dudek et al. (1997) have carried out experiments on cerebellar granule neurons (CGN) whose survival in serum-free, low K⁺ medium could be partially supported by insulin-like growth factor 1 (IGF-1). LY294002 treatment of these neurons resulted in their death, suggesting that IGF-1 conferred protection via PI 3-kinase signaling. Transfection of CGN with Akt kinase expression vectors promoted their survival in the absence of IGF-1, suggesting that Akt kinase was important in the survival mediated by this growth factor. However, analogous to the situation we describe with NGF and SCG neurons, PI 3-kinase was not an obligatory survival pathway when cells were grown in serum plus high K⁺. Since the submission of this paper, Miller et al. (1997) have also described survival mediated by PI 3-kinase in CGN, maintained in IGF-1 or by depolarization in low serum (Miller et al., 1997).

We also investigated the effect on survival of p70^S6 kinase thought to be downstream of PI 3-kinase and Akt kinase (Downward, 1994). Rapamycin, a potent inhibitor of p70^S6 kinase activation (Price et al., 1992), did not induce apoptosis itself and did not inhibit the survival-promoting action of Akt kinase. Thus our results suggest that in SCG neurons PI 3-kinase signals downstream to Akt kinase and that Akt kinase is not working via p70^S6 kinase. It is unclear what molecules Akt is signaling to in these neurons. A known substrate for Akt is glycogen synthase kinase 3 (GSK3), which is inhibited upon phosphorylation by Akt. GSK3 was initially described as negatively regulating glycogen synthase, the rate-limiting enzyme in glycogen synthesis, but is now known to phosphorylate many other proteins (Welsh et al., 1996), and it may be the case that molecules downstream of GSK3 play a role in apoptosis.

To investigate further how PI 3-kinase can inhibit death, we also looked at the level and phosphorylation state of the transcription factor c-Jun. We had previously shown that upon the withdrawal of NGF from SCG neurons, c-Jun increases in level and becomes phosphorylated, leading to its activation (Ham et al., 1995; A. Watson, personal communication). In addition, the c-Jun NH₂-terminal protein kinase (JNK) pathway had been implicated in apoptosis of PC12 cells (Xia et al., 1995). We therefore examined by immunofluorescence the level of c-Jun and of phospho–c-Jun in cells microinjected with activated PI 3-kinase. PI 3-kinase did not prevent the accumulation of nuclear c-Jun or phospho–c-Jun in neurons after NGF withdrawal (Fig. 5). This suggests that PI 3-kinase is acting to inhibit the death pathway downstream of c-Jun or independently of it. PI 3-kinase did not induce c-Jun expression itself when injected in the presence of NGF (data not shown). Importantly, these data also indicate that the kinases involved in c-Jun regulation, such as JNK, are not directly affected by the PI 3-kinase pathway, since phosphorylation of c-Jun followed the normal pattern after NGF withdrawal. It is possible therefore that PI 3-kinase is a cytoplasmic survival effector that is able to mediate NGF-induced survival independently of new transcription or protein synthesis.

In summary, we have demonstrated a potential role for PI 3-kinase and Akt kinase in the survival of SCG neurons after NGF withdrawal. Since inhibition of PI 3-kinase does not induce apoptosis in the presence of NGF, it is likely that there are other survival pathways stimulated by NGF. In addition we have shown that the protective effect of PI 3-kinase is not due to prevention of c-Jun activation. Stimulation of the PI 3-kinase pathway, with its survival-promoting effects, could therefore lead to novel therapies for a range of degenerative disorders associated with neuronal apoptosis.

We would like to thank Dr. C. Bazenet for critical reading of this manuscript, Ms. K. Ferguson for editorial assistance, and the laboratories that generously supplied antibodies and plasmids.

CGN

cerebellar granule neurons

GSK

glycogen synthase kinase

IGF

insulin-like growth factor

PI

phosphatidylinositol

SCG

superior cervical ganglion