The selective and inducible activation of endogenous PI 3 ... - Nature

ã

Oncogene (1999) 18, 4586 ± 4597 1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00 http://www.stockton-press.co.uk/onc

The selective and inducible activation of endogenous PI 3-kinase in PC12 cells results in ecient NGF-mediated survival but defective neurite outgrowth Margaret Ashcroft*,1, Robert M Stephens1, Bengt Hallberg2,3, Julian Downward2 and David R Kaplan1,4 1

ABL-Basic Research Program, NCI-FCRDC, West 7th Street, Frederick, Maryland, MD 21702, USA; 2Imperial Cancer Research Fund, 44 Lincoln's Inn Fields, London WC2A 3PX, UK

The Trk/Nerve Growth Factor receptor mediates the rapid activation of a number of intracellular signaling proteins, including phosphatidylinositol 3-kinase (PI 3kinase). Here, we describe a novel, NGF-inducible system that we used to speci®cally address the signaling potential of endogenous PI 3-kinase in NGF-mediated neuronal survival and di€erentiation processes. This system utilizes a Trk receptor mutant (Trkdef) lacking sequences Y490, Y785 and KFG important for the activation of the major Trk targets; SHC, PLC-gl, Ras, PI 3-kinase and SNT. Trkdef was kinase active but defective for NGF-induced responses when stably expressed in PC12nnr5 cells (which lack detectable levels of TrkA and are non-responsive to NGF). The PI 3-kinase consensus binding site, YxxM (YVPM), was introduced into the insert region within the kinase domain of Trkdef. NGF-stimulated tyrosine phosphorylation of the Trkdef+PI 3-kinase addback receptor, resulted in the direct association and selective activation of PI 3-kinase in vitro and the production of PI(3,4)P2 and PI(3,4,5)P3 in vivo (comparable to wild-type). PC12nnr5 cells stably expressing Trkdef+PI 3-kinase, initiated neurite outgrowth but failed to stably extend and maintain these neurites in response to NGF as compared to PC12 parental cells, or PC12nnr5 cells overexpressing wild-type Trk. However, Trkdef+PI 3kinase was fully competent in mediating NGF-induced survival processes. We propose that while endogenous PI 3-kinase can contribute in part to neurite initiation processes, its selective activation and subsequent signaling to downstream e€ectors such as Akt, functions mainly to promote cell survival in the PC12 system. Keywords: Trk; PC12; NGF; phosphatidylinositol 3kinase; di€erentiation; survival

Introduction In recent years we and others, have studied the signal transduction mechanisms by which nerve growth factor (NGF), mediates neuronal survival and di€erentiation processes through its receptor, Trk (Kaplan and Miller, 1997). Trk rapidly dimerizes upon ligand binding and *Correspondence: M Ashcroft Current addresses: 3 UmeaÊ University, 87 UmeaÊ, Sweden; 4 Montreal Neurological Institute, McGill University, Montreal PQ H3A 2B4, Canada Received 20 November 1998; revised 11 February 1999; accepted 9 March 1999

becomes phosphorylated on several distinct tyrosine residues (Y490, Y670, Y674, Y675, and Y785 in human TrkA). This allows the speci®c recruitment and tyrosine phosphorylation of a number of intracellular Src homology-2 (SH2) domain-containing target proteins including SHC, phospholipase C-gl (PLC-gl), and SHP1 (Kaplan and Miller, 1997). These proteins in turn activate a series of intracellular signaling molecules in distinct pathways, including Gab-1 and PI 3-kinase, PI 3-kinase/Akt/p70S6K, and Grb2/mSOS/Ras/BRaf/ MEK/MAPK/p90rsk (Kaplan and Miller, 1997; Segal and Greenberg, 1996). Other proteins, including sucassociated neurotrophic factor-induced tyrosine phosphorylated target (SNT), and Crk are also rapidly tyrosine phosphorylated in response to NGF binding to Trk (Rabin et al., 1993; Ribon and Saltiel, 1996; Teng et al., 1996). Activation of these signaling molecules culminates in downstream events involved in the promotion and maintenance of neuronal survival and di€erentiation. For example, overexpression, or expression of constitutively active forms of Ras, BRaf, MEK, and MAPK, Gab-1, and PI 3-kinase all result in NGF-independent process formation, while expression of dominant-inhibitory Ras, MEK, Crk, and PI 3kinase inhibit NGF-dependent process formation (BarSagi and Feramisco, 1985; Fukuda et al., 1995; Holgado-Madruga et al., 1997; Jackson et al., 1996; Kobayashi et al., 1997; Noda et al., 1985; Szeberenyi et al., 1990; Teng et al., 1996; Wood et al., 1993; Xia et al., 1995). Additionally, it has recently been shown that overexpression, or expression of constitutively active forms of PI 3-kinase or Akt, contributes to NGFindependent survival functions in sympathetic neurons (Crowder and Freeman, 1998). We and others have previously used the rat pheochromocytoma-derived (PC12) cell line (Greene and Tischler, 1976) and the PC12-derived cell line, PC12nnr5, to assess the signaling potentials of speci®c tyrosines in Trk (Baxter et al., 1995; Cunningham et al., 1997; Inagaki et al., 1995; Kouhara et al., 1997; Loeb et al., 1994; Obermeier et al., 1994, 1993; Peng et al., 1995; Stephens et al., 1994). PC12nnr5 cells lack detectable levels of TrkA, are non-responsive to NGF and have proved useful for assaying Trk mutants (Loeb et al., 1991). These studies have identi®ed a number of key residues in Trk involved in initiating the major NGF signaling pathways. First, residues Y490 and Y785 are responsible for the direct association with, and tyrosine phosphorylation of SHC and PLC-gl respectively. Activation of either SHC or PLC-gl is sucient to mediate cellular survival and maintain neuronal di€erentiation (Kouhara et al., 1997;

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

Obermeier et al., 1994, 1993; Stephens et al., 1994). Second, residues Y670, Y674 and Y675, when mutated either independently or in combination, abolish the activation of PLC-gl and PI 3-kinase, and regulate Trk tyrosine kinase activity (Cunningham et al., 1997). Third, deletion of a three amino acid stretch at residues 441 ± 443 (KFG), within the juxtamembrane region of TrkA abolishes tyrosine phosphorylation of SNT, and prevents cellular hypertrophy, initiation of neurite outgrowth, and cessation of cell proliferation in response to NGF (Peng et al., 1995). Finally, the Y490 site has been shown to be essential for mediating the activation of PI 3-kinase in vitro and for the production of PI-3,4-P2 and PI-3,4,5-P3 in vivo (Baxter et al., 1995; this report). In recent years a number of reports have shown that PI 3-kinase and a major substrate of PI 3-kinase activity, Akt, regulates the survival of several cell types, including primary neurons, ®broblasts, and myeloid progenitor cells (Crowder and Freeman, 1998; Franke et al., 1997a). In PC12 cells, NGF rapidly activates PI 3-kinase leading to the production of the second messengers, PI-3,4-P2 and PI-3,4,5-P3 (Carter and Downes, 1992; Solto€ et al., 1992; this report). PI3,4-P2 has been shown to directly activate Akt in vivo (Franke et al., 1997b). Furthermore, it has recently been shown that Akt is activated in response to NGF treatment of PC12 cells (Andjelkovic et al., 1998; this study). Importantly, however, various studies have reported di€erent conclusions regarding the precise role of the PI 3-kinase pathway in mediating survival and/or neurite outgrowth processes. Yao and Cooper (1995) using the PI 3-kinase inhibitors, wortmannin and LY29002, and a PDGFb receptor mutant that fails to activate PI 3-kinase, showed that PI 3-kinase activation mediates neuronal survival, in a pathway distinct from that which mediates neurite outgrowth. However, Kimura and colleagues (Kimura et al., 1994), showed that wortmannin inhibited NGF-dependent neurite elongation, but not cell survival. While Jackson et al. (1996), also using wortmannin and transient constitutive expression of a dominantinhibitory p85 regulator PI 3-kinase subunit, demonstrated that NGF-induced neurite initiation was dependent upon PI 3-kinase activity. Finally, Kobayashi et al. (1997), used an adenovirus-mediated Cre recombination system to express a constitutively active p110 PI 3-kinase catalytic subunit in PC12 cells, and found that this enzyme induced the initiation of incomplete and unstable processes. The variability in the results and conclusions obtained in these studies is most likely due to the use of several techniques and experimental paradigms, each with their own strengths and weaknesses: (1) Wortmannin, the selective PI 3kinase inhibitor, inhibits all of the PI 3-kinase family members through a common mechanism (Wymann et al., 1996) and might have a variety of e€ects on cell physiology; (2) The dominant inhibitory p85 PI 3kinase subunit may bind to several SH2 domain containing proteins, while activated p110 PI 3-kinase that induces high levels of PI 3-kinase activity in cells may stimulate the formation of PI 3-kinase metabolites that do not quantitatively or qualitatively replicate the levels of phosphoinositides produced with NGF treatment of cells; (3) The PDGFb receptor mutants used to study PI 3-kinase signaling are still competent

in activating other signaling proteins such as c-Src, and this receptor is not normally expressed in PC12 cells; (4) The constitutive expression of PI 3-kinase mutants is toxic to PC12 cells. To clarify the role of PI 3-kinase signaling in neurite outgrowth and survival of PC12 cells, and to avoid some of the experimental paradigms described above, we generated the Trkdef addback receptor (Trkdef+PI 3kinase), a novel mutant Trk receptor that can selectively and inducibly activate PI 3-kinase. Trkdef addback receptor can tolerate the addition of individual SH2 domain binding motifs without compromising intrinsic kinase activity. We show that although this receptor is extremely defective in its ability to activate any of the known NGF-induced signaling proteins, it successfully associates with, and selectively activates PI 3-kinase in vitro and in vivo. Furthermore, we show that while PI 3-kinase activation can contribute in part to neuronal outgrowth processes in response to NGF, it functions primarily to promote survival in the PC12 cell system. Results To assess the role of PI 3-kinase signaling in NGFinduced neuronal survival and di€erentiation of PC12 cells, we initially constructed a triple mutant form of the human Trk receptor containing the KFG deletion at residues 441 ± 443 within the juxtamembrane region, shown previously to be important for NGF-induced SNT phosphorylation (Peng et al., 1995). Tyrosine to phenylalanine point mutations were also made at the Y490 and Y785 sites, important for mediating SHC and PI 3-kinase activation, and PLC-gl activation respectively (Baxter et al., 1995; Stephens et al., 1994). This triple mutant receptor was termed the Trk defective receptor, (Trkdef), shown in Figure 1. Sitedirected mutagenesis was used to insert a PI 3-kinase consensus binding motif (YxxM, Escobedo et al., 1991), YVPM, at residue L605 within the insert region of the kinase domain of Trkdef, as described in Materials and methods. Importantly, this site was introduced within a speci®c sequence context and alignment consistent with the analogous site present in the PDGFb receptor (analogous to Y751 in PDGFb). This mutant receptor was termed the PI 3kinase addback receptor (Trkdef+PI 3-kinase) (Figure 1). The Trkdef and Trkdef+PI 3-kinase mutant cDNAs were introduced into the pLNCX retroviral vector and used to stably transfect PC12nnr5 cells which express no detectable endogenous Trk and are non-responsive to NGF (Loeb et al., 1991). G418-resistant clones were expanded on the basis of their morphological similarity with the parental PC12 and PC12nnr5 lines and assessed by Western blot analysis for Trkdef and Trkdef+PI 3-kinase expression and NGF-induced tyrosine phosphorylation. Four positive independent clonal lines expressing Trkdef and ®ve positive independent clonal lines expressing Trkdef+PI 3-kinase were selected for further analysis. PC12 cells and PC12nnr5 cells stably expressing wild-type (WT) Trk, were used as controls in all experiments. The WT Trk PC12nnr5 line has been described elsewhere (Loeb and Greene, 1993; Peng et al., 1995), and overexpresses Trk at about 3 ± 5-fold higher than PC12 cells. Quantifica-

4587

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

4588

tion by densitometry, showed that the WT line expressed Trk at 4.3+0.15 (n=6) fold higher than for parental PC12 cells. Trkdef and Trkdef+PI 3-kinase are tyrosine phosphorylated in response to NGF To quantitate the expression and to con®rm that the Trkdef and the Trkdef+PI 3-kinase receptors were tyrosine phosphorylated in response to NGF, PC12 cells and PC12nnr5 cells expressing either WT Trk, Trkdef (four independent clonal lines) or Trkdef+PI 3kinase (®ve independent clonal lines) were stimulated with or without NGF (100 ng/ml) for 5 min. Lysates were normalized for protein content and immunoprecipitated with antibodies to Trk. Immunoprecipitates were analysed by Western immunoblotting using antiphosphotyrosine and anti-Trk. Figure 2a shows the NGF-induced tyrosine phosphorylation and expression levels for Trkdef and Trkdef+PI 3-kinase. The Trkdef and Trkdef+PI 3-kinase receptors are kinase active and are tyrosine phosphorylated in response to NGF stimulation (Figure 2a). Importantly, the basal activity of the Trk mutants is almost undetectable compared with the WT line (Figure 2a). Of the four Trkdef clonal lines selected, three expressed Trkdef at levels comparable or slightly higher than the WT line (data not shown), and one expressed Trkdef at levels lower than the WT line (Figure 2a). Similarly, of the ®ve Trkdef+PI 3-kinase clonal lines selected, all expressed at levels several fold lower than the WT line (Figure 2a). Quanti®cation by densitometry, showed that expression levels of mutant Trk in clonal lines 1, 2 and 3, were approximately 2.4+0.05 (n=3), 2.86+0.13 (n=3) and 2.75+0.13 (n=3) fold higher than for parental PC12 cells, respectively. We have recently shown, that there is a biological correlation between the levels of expression of Trk mutants and NGF responsiveness. Indeed, when expression levels exceed that of the WT line by ®vefold or more, neurite outgrowth in response to NGF is enhanced and accelerated (Cunningham et al., 1997). Further biochemical and biological analysis was carried out on one Trkdef and the Trkdef+PI-3 kinase lines 1, 2, 3. Trkdef or Trkdef+PI 3-kinase fail to activate PLC-g1 and SHC upon NGF stimulation Previous studies have shown that upon NGF stimulation, Trk autophosphorylation of Y785 and Y490 is

required for subsequent association and tyrosine phosphorylation of PLC-gl and SHC (Loeb et al., 1994; Obermeier et al., 1994, 1993; Stephens et al., 1994). To con®rm that Trkdef or Trkdef+PI 3-kinase were unable to associate with direct targets of Trk, we assessed the tyrosine phosphorylation and association of PLC-gl and SHC with Trkdef or Trkdef+PI 3-kinase. As shown in Figure 2b (upper panel), PLC-gl associates with NGF-stimulated Trk (WT), and undergoes tyrosine phosphorylation upon activation. However, Trkdef or Trkdef+PI 3-kinase fail to activate or directly associate with PLC-gl (Figure 2b), although both receptors are tyrosine phosphorylated in response to NGF (Figure 2a). As previously shown for mutants of Trk (Stephens et al., 1994), the tyrosine to phenylalanine point mutation at Y785 in Trkdef and Trkdef+PI 3kinase is sucient to prevent NGF-mediated association and tyrosine phosphorylation of PLC-gl. Similarly, SHC has been shown to rapidly associate with NGFstimulated Trk (PC12 cells), where all forms (SHC p46, p52, p66) undergo tyrosine phosphorylation (Stephens et al., 1994). Furthermore, SHC p66 undergoes an electromobility shift upon NGF stimulation and appears as a 68 kDa band (Stephens et al., 1994). The Trkdef or Trkdef+PI 3-kinase receptors fail to associate with, or activate SHC (data not shown), although both receptors are tyrosine phosphorylated in response to NGF (Figure 2a). The tyrosine to phenylalanine point mutation at Y490 in Trkdef and Trkdef+PI 3-kinase is sucient to abolish NGFmediated association and tyrosine phosphorylation of SHC. Analysis of SNT phosphorylation in PC12nnr5 cells expressing Trkdef and Trkdef+PI 3-kinase The SNT protein is tyrosine phosphorylated in response to NGF or FGF treatment of PC12 cells (Rabin et al., 1993). However, the mitogens insulin or EGF are unable to activate SNT. Previous studies have shown that the KFG site (residues 441 ± 443) within the juxtamembrane region of Trk is important for NGFmediated SNT phosphorylation and for mediating somatic hypertrophy, neurite initiation, and cessation of cell proliferation in response to NGF (Peng et al., 1995). We note that FRS-2, which was described as being identical to SNT (Kouhara et al., 1997), has quite di€erent biochemical properties to SNT. Tyrosine phosphorylated SNT is localized to the nucleus and

Figure 1 A schematic representation of the Trkdef and Trkdef+PI 3-kinase receptors. Sites of modi®cation are indicated as follows: Y490F, Y785F and DKFG, with addition of the PI 3-kinase binding, YxxM motif at L605 within the kinase insert domain

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

does not bind Grb-2, while FRS-2 is membrane-bound and binds Grb-2 (Rabin et al., 1993; Kouhara et al., 1997; Michaud, Copeland and Kaplan, unpublished). Furthermore, Meaken and colleagues have recently shown that NGF-mediated tyrosine phosphorylation of FRS-2, unlike SNT, is mediated via the SHC site on Trk (Meaken et al., 1999; personal communication). To assess the ability of Trkdef and Trkdef+PI 3-kinase to induce SNT tyrosine phosphorylation, PC12 cells or PC12nnr5 cells expressing either WT Trk, Trkdef, or Trkdef+PI 3-kinase were stimulated with NGF for 5 min, lysed, precipitated with p13suc-1 agarose beads and analysed by Western blot for tyrosine phosphorylation. As shown previously (Peng et al., 1995), Trk was able to mediate NGF-induced tyrosine phosphorylation of SNT, which has an apparent molecular

weight of approximately 90 kDa (Figure 2c). Both Trkdef and Trkdef+PI 3-kinase had a very limited ability to mediate tyrosine phosphorylation of SNT in response to NGF, and this was greatly impaired compared to the WT control. We cannot rule out the possibility that the weak tyrosine phosphorylation of SNT mediated by Trkdef and Trkdef+PI 3-kinase, may re¯ect receptor expression levels since we have previously observed that Trk mutated at KFG will induce tyrosine phosphorylation of SNT when overexpressed. Alternatively, other novel mechanisms may mediate this phosphorylation. For example, it has recently been shown that phosphotyrosines within the Trk kinase domain interact with two novel Trk substrates, rAPS and SH2-B (Qian et al., 1998). rAPS and SH2-B have been shown to bind Grb2 and mediate

Figure 2 Trkdef and Trkdef+PI 3-kinase are tyrosine phosphorylated in response to NGF, but fail to associate with PLC-g1 or suciently activate SNT. (a) PC12nnr5 cells expressing either; human wild-type Trk (WT), Trkdef, or Trkdef+PI 3-kinase (independent clonal lines; 1, 2 and 3) were treated with or without NGF (100 ng/ml) for 5 min at 378C, lysed in NP40 lysis bu€er, and immunoprecipitated with anti-Trk 203. Western blots were analysed for tyrosine phosphorylation (upper panel) and Trk protein levels (lower panel). (b) The cell lines indicated were treated as described in a, and immunoprecipitated with antibodies to PLC-g1. Western blots were analysed for tyrosine phosphorylation (upper panel) and PLC-gl protein levels (lower panel). (c) The cell lines indicated were treated as described in a, and immunoprecipitated with p13suc-1 agarose beads. Western blots were analysed for tyrosine phosphorylation of SNT

4589

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

4590

morphological di€erentiation of PC12 cells. It remains unclear however, whether rAPS or SH2-B interact with other known Trk targets such as SNT. PI 3-kinase associates with Trkdef+PI 3-kinase and is activated upon NGF stimulation in vitro and in vivo PI 3-kinase is activated in response to a number of growth factor stimuli including NGF (Carter and Downes, 1992; Raoni and Bradshaw, 1992; Solto€ et al., 1992). Trk is believed to mediate its activation via the Y490 SHC site, since mutational analysis at this site completely abolishes NGF-induced PI 3-kinase activity in vitro and in vivo (Baxter et al., 1995; this report). Furthermore, it has been shown that the Grb2associated binder-1 protein (Gab-1) mediates PI 3kinase activation by NGF and has been proposed to be involved in recruiting PI 3-kinase to Trk possibly via Grb2. The PI 3-kinase consensus binding site YxxM, present in the PDGFb receptor at Y751, has been shown to readily associate with the p85 PI 3-kinase subunit and activate PI 3-kinase (Toker and Cantley, 1997). To con®rm that PI 3-kinase associates with Trkdef+PI 3-kinase and is activated in response to NGF stimulation, we ®rst carried out an in vitro association and PI 3-kinase assay using recombinant baculovirus produced Trk and Trkdef+PI 3-kinase proteins. Trkdef+PI 3-kinase, unlike Trk, was able to associate with (Figure 3a) and activate PI 3-kinase in vitro (Figure 3b). Secondly, PC12 cells and PC12nnr5 cells expressing Trkdef or Trkdef+PI 3-kinase were stimulated with NGF for 5 min, lysed, and either immunoprecipitated with anti-Trk and analysed by Western blot for PI 3-kinase association (Figure 4a), or immunoprecipitated with anti-phosphotyrosine and exposed to phosphoinositol in vitro (Figure 4b). Phospholipids were extracted and separated by TLC, as outlined in the Materials and methods. As shown in Figure 4a, PI 3-kinase did not associate directly with wild-type Trk (PC12 control), but was activated in response to NGF treatment of PC 12 cells (Figure 4b). We have shown previously that PI 3-kinase will only associate with Trk when Trk is greatly overexpressed (Solto€ et al., 1992). PI 3-kinase was unable to associate with Trkdef (Figure 4a) and was not activated in response to NGF stimulation (Figure 4b). Importantly, PI 3-kinase (p85) associated with the Trkdef+PI 3-kinase (Figures 3a and 4a) and was activated in response to NGF (Figures 3b and 4b). Furthermore, activation of PI 3-kinase by Trkdef+PI 3kinase, was completely blocked in vitro by the PI 3kinase inhibitor, wortmannin (Figure 4b). Interestingly, PI 3-kinase association and activation occurred in direct parallel to NGF-induced tyrosine phosphorylation of Trkdef+PI 3-kinase which was detectable for up to 24 h (data not shown). Analysis of pholpholipid production in vivo showed that Trkdef+PI 3-kinase was able to mediate production of PI 3-,4-P2 and PI 3-,4,5-P3 upon NGF stimulation to levels comparable to that seen for wild-type Trk (Figure 5), although RasGTP was only slightly increased (Hallberg et al., 1998). Trkdef, Y490FTrk (-SHC), and Y490F/Y785F-Trk (-SHC/-PLCgl), were unable to mediate the production of PI 3-,4-P2 and PI 3-4,5-P3 upon NGF stimulation. Basal, unstimulated lipid levels in the range of 900 ±

1600 c.p.m. were obtained for each cell line and subtracted from each measurement. EGF-stimulated lipid levels were used as a control for each cell line and were in the range 23 000 ± 35 000 c.p.m. for PI-3,4,-P2 and 26 000 ± 37 000 c.p.m. for PI-3,4,5-P3. NGFinduced production of PI-3,4,-P2 and PI-3,4,5-P3 levels were represented as a percentage of control stimulation with EGF (Figure 5). It can be concluded from this data that the PI 3-kinase consensus binding site present in Trkdef+PI 3-kinase is biochemically functional, and not only serves a functional docking site for PI 3kinase, but mediates its activation in an NGF-inducible manner. Analysis of MAP kinase activity in cells expressing Trkdef or Trkdef+PI 3-kinase To determine whether Trkdef+PI 3-kinase could stimulate any of the known downstream targets of Trk, we analysed MAP kinase activity. MAP kinase is rapidly tyrosine phosphorylated in response to NGF treatment of PC12 cells, and this phosphorylation is sustained over time (Qui and Green, 1992; Figure 6a). NGF-induced activation of MAP kinase is mediated by both the SHC and PLC-gl sites on Trk (Stephens et al.,

Figure 3 PI 3-kinase associates with Trkdef+PI 3-kinase in vitro and is activated in response to NGF. Sf9 cells (26106), infected with human wild-type Trk, or Trkdef+PI 3-kinase recombinant baculoviruses were lysed and immunoprecipitated with anti-Trk 203. Immunoprecipitated complexes were phosphorylated in vitro and exposed to an adult mouse brain lysate. Samples were either (a) assessed by Western blot for PI 3-kinase (p85) association (upper panel), and Trk protein (lower panel), or, (b) exposed to phosphoinositol (PI) (0.2 mg/ml) in vitro, and the resulting PI 3-P (PIP) was extracted and separated by TLC

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

4591

Figure 4 PI 3-kinase associates with Trkdef+PI 3-kinase in vitro and is activated in response to NGF. PC12 cells and PC12nnr5 cells expressing either, Trkdef or Trkdef+PI 3-kinase (clone 2) were treated with or without NGF (100 ng/ml) for 5 min at 378C and lysed in NP40 lysis bu€er. Lysates were either, (a) immunoprecipitated with anti-Trk and analysed by Western blot for PI 3-kinase (p85) association or, (b) immunoprecipitated with anti-phosphotyrosine beads and exposed to PI (0.2 mg/ml) in vitro in the absence or presence of wortmannin (WT) at 200 ng/ml. The resulting PI 3-P (PIP) was extracted and separated by TLC

Figure 5 E€ect of NGF stimulation on the formation of PI-3,4-P2 and PI-3,4,5-P3 in PC12nnr5 cells expressing WT and mutant Trks. Production of PI-3,4-P2 (®lled bars) and PI-3,4,5-P3 (open bars) in 32P-orthophosphate-labeled PC12nnr5 cells expressing human wild-type Trk (WT), or the mutant Trks; Trkdef, Trkdef+PI 3-kinase, Y490F (SHC7), Y785F (PLC-g17), and Y490F/ Y785F (SHC7/PLC-gl7). All cell lines were stimulated with NGF and EGF (control). Basal, unstimulated lipid levels obtained for each line were in the range 900 ± 1600 c.p.m. and were subtracted from each measurement. Lipid levels were also expressed as a percentage of the level achieved with EGF treatment (value above each column). For all lines, EGF stimulated lipid levels were in the range 23 000 ± 35000 c.p.m. for PI(3,4)P2 and 26 000 ± 37 000 c.p.m. for PI(3,4,5)P3

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

4592

Figure 6 E€ects of NGF on MAP kinase activity in PC12nnr5 cells expressing mutant Trks. (a) PC12 cells, and (b) PC12 cells and PC12nnr5 cells expressing Trkdef or Trkdef+PI 3-kinase clones 1 and 2, were stimulated with NGF (100 ng/ml) for the time periods indicated. Lysates were immunoprecipitated with anti-Erk (C16), and analysed by Western blot for tyrosine phosphorylation (upper panels) and Erk protein levels (lower panels)

1994). As described above, both Trkdef and Trkdef+PI 3-kinase were unable to associate with or induce the direct tyrosine phosphorylation site of SHC or PLC-gl. However, Trkdef+PI 3-kinase mediated a very weak, but transient tyrosine phosphorylation of MAP kinase which was detectable for at least 1 h (Figure 6b). The Trkdef receptor was unable to mediate MAP kinase tyrosine phosphorylation upon NGF stimulation (Figure 6b) even if prolonged stimulation was carried out (data not shown). NGF usually mediates a sustained activation of MAP kinase which has been described to be important for mediating neuronal outgrowth processes. It is particularly interesting that even without the recruitment of SHC or PLC-gl, the recruitment of PI 3-kinase directly to the Trkdef+PI 3kinase receptor induced this weak and transient activation of MAP kinase in the absence of Ras activation (Hallberg et al., 1998). We have previously shown that MAP kinase can be suciently activated independently of Ras activity through the activation of PLC-gl (Stephens et al., 1994; Hallberg et al., 1998). Although PLC-gl is not directly recruited to Trkdef+PI

3-kinase we cannot rule out the exciting possibility that novel, yet unidenti®ed mediators of this pathway may be responsible for, or potentially contribute to this e€ect. It is unlikely however, that rAPS or SH2-B would be involved here, since these have been shown to function upstream of Ras in activating Erk (Qian et al., 1998). Trkdef+PI 3-kinase mediates Akt activation To assess direct targets of PI 3-kinase activation, we investigated the ability of the serine-threonine kinase, Akt to phosphorylate histone H2B as we have previously described (Franke et al., 1995). NGF has recently been shown to activate Akt (PKB a, b, and g) in PC12 cells (Andjelkovic et al., 1998). PC12 cells and PC12nnr5 cells expressing either Trkdef or Trkdef+PI 3kinase were serum starved overnight and stimulated with NGF as mentioned above. Cell lysates were immunoprecipitated with anti-Akt and an in vitro kinase assay was performed. In agreement with a previous report (Park et al., 1996), Akt was activated

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

in response to NGF stimulation of PC12 cells, shown here to be mediated by the direct activation of Trk (Figure 7). Akt was also activated by NGF stimulation of Trkdef+PI 3-kinase, but not of Trkdef (Figure 7). This data indicates that the NGF-induced PI 3-kinase activity mediated by Trk or Trkdef+PI 3-kinase is sucient to activate Akt, previously shown to be a direct downstream target of PI 3-kinase (Franke et al., 1997a,b, 1995), and con®rms that the PI 3-kinase addback site within the Trkdef receptor context can mediate the activation of a signaling protein that is regulated by PI 3-kinase-generated second messengers. Neurite outgrowth in response to NGF To investigate whether the Trkdef+PI 3-kinase receptor is able to contribute to neuronal di€erentiation processes, we assessed neurite outgrowth in response to NGF. Cells were plated in 1.5% serum supplemented with or without 100 ng/ml NGF and assessed over a period of 4 ± 10 days. The Trkdef+PI 3-kinase receptor mediates very short, unstable processes with visible growth cones in response to NGF (Table 1). In immunohistochemical studies, these processes stained positive for both actin and tubulin, indicating that they were authentic neurites (data not shown). The Trkdef was unable to mediate any signi®cant neurite outgrowth in response to NGF (Table 1). This data suggests that although the NGF-induced PI 3-kinase activity may contribute in part to neuronal initiation processes as previously shown (Kobayashi et al., 1997), other signaling pathways are required for sucient outgrowth and stabilization of neurite formation. Trkdef+PI 3-kinase mediates survival processes To investigate whether Trkdef+PI 3-kinase is able to mediate cell survival, PC12 cells and PC12nnr5 cells expressing either Trk (WT control), Trkdef or Trkdef+PI

3-kinase were washed and plated into serum-free RPMI-1640 medium supplemented with or without NGF for a period of 6 days as described in the Materials and methods section. Cell viability was determined at day 6 using a MTT assay, and was represented as a percentage of the original value (average of triplicate wells) obtained a day zero, for each experiment. A total of three independent experiments were averaged (Figure 8). In agreement with others (Greene, 1978), Trk mediated NGFdependent survival over a 6 day period, where cell numbers were maintained at approximately 85 ± 100% for PC12 cells, and PC12nnr5 cells expressing Trk, respectively (Figure 8). The Trkdef receptor was unable to signi®cantly mediate NGF-dependent survival, although a minimal survival response was observed. The Trkdef+PI 3-kinase receptor however, appeared to maintain cellular viability at levels comparable to that maintained by wild-type Trk in response to NGF (Figure 8) and indicates the importance of the PI 3kinase pathway for NGF-dependent survival. This has been shown in recent studies using over expression of activated forms of PI 3-kinase and Akt in sympathetic neurons (Crowder and Freeman, 1998). Interestingly, we also observed that the selective PI 3-kinase inhibitor, LY294002, in agreement with others (Yao and Cooper, 1995), completely abolished NGFdependent survival of PC12 cells and PC12nnr5 cells expressing Trkdef+PI 3-kinase, and enhanced cell death. The percentage viability of cells maintained in NGF (100 ng/ml) decreased from 97% and 92% to 23% and 28% respectively in the presence of LY294002 (20 mM) at 48 h (data not shown). Finally, we were interested in assessing if the weak, transient activation of MAP kinase mediated by Trkdef+PI 3kinase (Figure 6b), was sucient to mediate the survival and neurite initiation functions observed. In a parallel experiment performed with the PI 3-kinase inhibitor described above, we observed that the

Figure 7 Trkdef+PI 3-kinase activates Akt. PC12 cells and PC12nnr5 cells expressing Trkdef or Trkdef+PI 3-kinase (clones 1 and 2) were stimulated with NGF (100 ng/ml). Lysates were immunoprecipitated with anti-Akt-CT antibodies. Immunoprecipitates were exposed to histone H2B in an in vitro kinase assay, as previously described (Franke et al., 1995)

Table 1 Cell line

(7) NGF

(+) NGF

PC12 WT-Trk (PC12nnr5) Trkdef (PC12nnr5) Trkdef+PI3-K (PC12nnr5) clone 1 Trkdef+PI3-K (PC12nnr5) clone 2 Trkdef+PI3-K (PC12nnr5) clone 5

0.5+0.03 1.4+0.35 1.5+0.14 1.2+0.25 1.5+0.05 1.1+0.35

23.7+1.23 41.3+0.49 2.4+0.27 7.2+1.2 8.3+1.2 6.1+1.2

Total average length of neuritic outgrowth (per cell) for each cell line at day ®ve of NGF treatment (100 ng/ml), where 1 unit=0.16cell diameter+s.e.m

Figure 8 Trkdef+PI 3-kinase mediates survival processes. PC12 cells and PC12nnr5 cells expressing either, human wild-type Trk, Trkdef or Trkdef+PI 3-kinase (clone 5) were washed in serum-free RPMI-1640 medium and plated into 24 well plates (0.56105 cells/ well) containing serum-free medium RPMI-1640 medium supplemented with or without NGF (100 ng/ml). Cell viability (triplicate wells) was assessed by day zero (approximately 4 ± 8 h after plating) and on day six using an MTT assay (Boehringer Mannheim, IN, USA). Average values obtained at day six were presented as a percentage of the original (day zero)

4593

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

4594

selective MEK inhibitor, PD989059 (20 mM), was able to block NGF-mediated survival of PC12 cells from 97% to 53% (unstimulated, 22 to 15% respectively), and impair neurite elongation. However, PD98059 had greatly reduced e€ects on NGF-mediated survival of PC12nnr5 cells expressing Trkdef+PI 3-kinase, where percentage viability decreased from 91% to 74% (unstimulated 33% to 18%), and did not appear to markedly e€ect the initiation of the neuritites observed. Taken together, this data suggests that selective activation of the PI 3-kinase kinase signaling pathway in an NGF-inducible manner is able to primarily mediate survival function and contribute, at least in part, to the initiation of neurite outgrowth. Discussion The aim of this study was to establish a system whereby an endogenous intracellular signaling molecule such as PI 3-kinase could be activated in a selective and inducable manner, in order to examine its contribution to biological and biochemical responses of cells to growth factors. Previous studies on the function of PI 3-kinase in neural cell lines and neurons have relied on the use of non-speci®c pharmacological inhibitors, receptors that were not normally expressed in these cells, and constitutively activated PI 3-kinase mutants (see above). Our ®rst step was to generate the Trkdef receptor, which was defective in activating or inducing the tyrosine phosphorylation of any of the known major Trk targets (SHC, PLC-gl, Erk, PI 3-K, and SNT), and was unable to e€ectively mediate neurite outgrowth or survival processes in response to NGF when stably expressed in PC12nnr5 cells. To the Trkdef we added back a PI 3-kinase consensus binding site, YVPM, and generated the Trkdef+PI 3-kinase addback receptor. Upon NGF stimulation, this receptor was able to rapidly associate with, and activate PI 3-kinase. Although Trkdef+PI 3-kinase was unable to e€ectively mediate neurite outgrowth, it was able to contribute to survival processes. The Trk receptor is particularly suited to selectively analysing signaling protein function. Unlike many other RTK's Trk contains only two major tyrosine phosphorylation sites (Y490 and Y785) outside the kinase domain that are required for the speci®c interaction with signaling molecules. One other site, KFG, mediates SNT tyrosine phosphorylation. Our results show that mutation of Y490, Y785, and deletion of KFG renders Trk extremely defective for biochemical and biological responses to NGF. This is in contrast to the other major receptor addback, generated from the PDGFb receptor (Valius and Kazlauskas, 1993), which still encodes tyrosines capable of interacting with signaling proteins, or that can compensate for the loss of a crucial tyrosine. A second weakness of the PDGFb receptor system is the presence of the Src family association sites, which cannot be eliminated without a loss of receptor kinase activity (Fanger et al., 1997). Trk also has the advantage that it will tolerate the addition of any four amino acid SH2 binding motif within a ten amino acid cassette into the kinase insert region. We have also created Trkdef addback receptors that will selectively bind to Src family members, SHP-2, and Grb-2, and

have generated PC12nnr5 lines stably expressing these molecules (Ashcroft, Stephens and Kaplan, unpublished). In the case of Trkdef+PI 3-kinase, the levels of PI 3-kinase activity generated in vivo and in vitro very closely approximated to the levels generated by wildtype Trk. This is an advantage with regard to constitutively activated PI 3-kinase mutants, which generate at least 50% more PI-3,4,5-trisphosphate than is observed in NGF-treated wild-type PC12 cells (Kobayashi et al., 1997). Biological responses to Trk are very sensitive to receptor expression levels, in that overexpressed receptor mutants in PC12 cells can sometimes mask defects in biological phenotypes (Cunningham et al., 1997). We are currently constructing Tet-regulated Trkdef in adenovirus vectors so that the precise levels of Trkdef expression can be titrated according to endogenous Trk. The role of endogenous PI 3-kinase activity in PC12 cells Our results suggest that the biological potential of NGF-stimulated PI 3-kinase activity in PC12 cells is to mediate cell survival and to elicit the initiation of neurite outgrowth. These results are consistent with those of Yao and Cooper (1995), who demonstrated a survival function for PI 3-kinase. However, their results ruled out a role for PI 3-kinase in neurogenesis. In contrast, three other reports have demonstrated a role for PI 3-kinase in regulating either neurite initiation or elongation (Jackson et al., 1996; Kimura et al., 1994; Kobayashi et al., 1997). However, Kimura et al. (1994) did not observe a survival function for PI 3-kinase. Our data, supports the results of Jackson et al. (1996), obtained with wortmannin, and of Kobayashi et al. (1997), obtained with constitutively activated p110 PI 3-kinase, that PI 3-kinase has a role in neurite initiation. Conclusively, our ®ndings implicate that NGF mediated stimulation of endogenous PI 3-kinase activity, is sucient to mediate two processes, cell survival and incomplete neuritogenesis in PC12 cells. How does PI 3-kinase induce both cell survival and neurite initiation? We have previously shown that Akt is a direct target of PI 3-kinase activity and can be activated by binding to PI 3-,4-P2 (Franke et al., 1997a,b, 1995). We and others have also demonstrated that Akt is necessary or sucient to mediate IGF-1 or IL-3 dependent survival in cerebellar granule cells, myc-immortalized ®broblasts, and myeloid cells (Downward, 1995; Dudek et al., 1997; Kau€mannZeh et al., 1997; Kennedy et al., 1997; Miller et al., 1997; Philpott et al., 1997; Songyang, 1997). Akt is also activated by the Trkdef+PI 3-kinase addback receptor (Figure 7), and expression of activated Akt in sympathetic neurons is sucient to mediate NGFindependent survival (Crowder and Freeman, 1998). It is therefore likely that Akt through the direct activation of PI 3-kinase by Trkdef+PI 3-kinase mediates its survival e€ects. In PC12 cells, Akt and PI 3-kinase are regulated through the Y490 SHC association site on Trk (Hallberg et al., 1998; Figure 5). However, mutation at this site in Trk does not markedly impair NGF-dependent survival (Ashcroft, Stephens and Kaplan, unpublished). Therefore, other signaling proteins, particularly those regulated by PLCgl and subsequently MAP kinase (Stephens et al.,

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

1994), may collaborate with PI 3-kinase and/or Akt to mediate cell survival. The identity of the signaling protein(s) responsible for mediating the neurite initiation responses of Trkdef+PI 3-kinase which are downstream of PI 3kinase is less clear. None of the major targets of cellular PI 3-kinase activity, including Akt, p70S6K, GSK3, PDK1 and 2, adaptin, Grp-1, or the PKC isoforms e, l, Z have been shown to have a role in NGF-induced neuritogenesis in PC12 cells, although PKC-e is involved in neuroblastoma cell neurite outgrowth (Fagerstrom et al., 1996). It is therefore likely, that PI 3-kinase activity together with the limited SNT tyrosine phosphorylation mediated by Trkdef+PI 3-kinase may be sucient to initiate inecient neurogenesis. However, since the Trkdef+PI 3-kinase is unable to provide sustained signaling from other pathways, such as the Ras/MAP kinase pathway, neurite outgrowth and di€erentiation is not maintained. The Trkdef+PI 3-kinase system should prove useful for eventually identifying the components responsible for neurite growth in both PC12 cells and neurons. Mechanisms activating PI 3-kinase In wild-type PC12 cells, PI 3-kinase activity is regulated primarily through the SHC site on Trk (Baxter et al., 1995; Hallberg et al., 1998), and through subsequent binding of PI 3-kinase to Gab-1 (HolgadoMadruga et al., 1997). In Trkdef+PI 3-kinase-expressing PC12nnr5 cells, PI 3-kinase is activated independently of the direct activation of any of the known Trk signaling proteins, including Ras (Hallberg et al., 1998). Here we show that PI 3-kinase can be suciently activated independently of SHC and Ras, if provided with a speci®c docking site on Trk itself. In NGF-treated PC12 cells, Ras activation has been shown to be important for PI 3-kinase activation (Rodriguez-Viciana et al., 1994, 1996). However, our results indicate that the direct activation of PI 3-kinase in this system is not dependent upon the presence of elevated Ras activity (Hallberg et al., 1998), and although appears sucient to mediate competent survival processes, is unable to mediate e€ective neurogenesis. We hypothesize that PI 3-kinase, through the SHC site on Trk and via binding to Ras, may participate in initiating neurite outgrowth and cell survival, which is subsequently completed and maintained by proteins of the Ras/PLC-gl pathways (Stephens et al., 1994).

GAG AGT GTT GAT TAC GTT CCA ATG TTG GAT GCT GGT GGG GAG-3'. Sequencing analysis was performed to isolate clones containing the insert, and a Bpu1102I ± BglII fragment from the resulting construct containing the consensus site was inserted into the trkdef cDNA. The trkdef and trkdef+PI 3-K cDNAs were subcloned into the retroviral vector pLCNX (Miller and Rosman, 1989) and used to transfect PC12nnr5 cells (Loeb et al., 1991) as described elsewhere (Greene et al., 1991). Resistant clones were selected with G418 at 400 mg/ml during the ®rst week and at 200 mg/ml thereafter. Additionally, the trkdef and trkdef+PI 3-K cDNAs were subcloned into the baculovirus transfer vector, pAcC4 for expression in Sf9 cells (Summers and Smith, 1987). Cell culture PC12 cells were maintained in Dulbecco's modi®ed Eagle's medium (DMEM) containing 10% heat-inactivated horse serum and 5% calf serum. PC12nnr5 cells and PC12nnr5 cells expressing Trk mutants were plated onto collagen coated tissue culture dishes and maintained in RPMI-1640 medium containing 10% heat-inactivated horse serum and 5% calf serum (Greene et al., 1991). To assay neurite outgrowth, cells were plated at 0.56105 cells/cm2 and maintained in RPMI1640 containing 1% heat-inactivated horse serum and 0.5% calf serum supplemented with or without 100 ng/ml NGF. To assay cell survival, cells were washed three to four times with serum-free RPMI-1640 medium, detached by repeated titration and washed again in serum free medium. Cells were centrifuged and plated into 24-well (16 mm diameter) collagen coated tissue culture dishes (at approximately 0.56105 cells/well) containing serum-free RPMI-1640 medium supplemented with or without 100 ng/ml NGF at days zero, two and four (Greene et al., 1991). Cell viability was determined at day zero (at least 4 ± 8 h after plating) and at day six (for triplicate wells), using an MTT cell proliferation assay as outlined by the manufacturer (Boehringer Mannheim, Indianapolis, IN, USA). For each experiment, the proportion of viable cells at day six was represented as a percentage of the original number plated. Antibodies, immunoprecipitation and immunoblotting Lysis of cells and immunoprecipitations were carried out as detailed elsewhere (Stephens et al., 1994). All immunoprecipitates were washed using lysis bu€er without protease inhibitors present unless otherwise stated. Anti-Trk 203 was described previously (Hempstead et al., 1992). The anti-Ptyr monoclonal antibody 4G10 was provided by DK Morrison (NCI-FCRDC, MD, USA). Antibodies to SHC, PI 3-kinase and PLC-gl were obtained from UBI, NY, USA. The Erk-1 polyclonal antibody was obtained from Santa Cruz Biotech Inc., CA, USA. The SNT protein was isolated from cell lysates using p13suc-1 agarose beads as described previously (Rabin et al., 1993). Human recombinant NGF (2.5S) was obtained from UBI. NY. The Akt antibody (Akt-CT) was generated and used as described (Datta et al., 1995).

Materials and methods Construction of Trk mutant cDNAs The human trkdef cDNA encoding the triple mutation DKFG/ Y490F/Y785F was generated by replacing the StyI fragment (bases 1 ± 1560) in the trk cDNA encoding the Y490F/Y785F double mutant (Stephens et al., 1994) with the equivalent fragment from the trk cDNA encoding the DKFG (Peng et al., 1995). Sequencing analysis was performed to con®rm that all three mutations were present. The human trkdef+PI 3-K cDNA was generated by site-directed mutagenesis of human trk cDNA (Stephens et al., 1994) at residue L605 using the oligonucleotide primer: 5'-GCC AAG CTG CTG CTG GAT

Baculovirus Trk in vitro association assays To assess the ability of Trk or Trkdef+PI 3-kinase to bind substrates, Sf9 cells (26106) infected with Trk or Trkdef+PI 3-kinase recombinant baculovirus (at an MOI of 10), were harvested and lysed in 700 ml of 1% Nonident P-40 (NP-40) lysis bu€er (20 mM Tris pH 8.0, 137 mM NaCl, 0.5 mM EDTA, 10% glycerol) containing, 1 mM phenylmethylsulphonyl¯uoride (PMSF), 0.15 U/ml aprotinin, 20 mM leupeptin, 1 mM sodium orthovanadate) at 48C for 20 min. Trk proteins were immunoprecipitated with anti-Trk 203 and immunoprecipitates were washed once in lysis bu€er containing 0.1% SDS and 1% deoxycholate (RIPA), twice

4595

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

4596

with 0.5 N LiCl, 50 mM Tris (pH 7.4) containing 0.5% Triton X-100, and once with water. Samples were incubated in 50 ml freshly prepared kinase bu€er (30 mM Tris (pH 7.4), 10 mM MnCl2, 1 mM ATP, 500 mM sodium orthovanadate) for 30 min at RT, to activate intrinsic receptor kinase activity and immunoprecipitates were washed three times with NP-40 lysis bu€er. Activated Trk was incubated with 5 mg of an adult mouse brain solubilized in 1.5 ml NP-40 lysis bu€er for a further 3 h at 48C. Proteins were washed three times in NP40 lysis bu€er, and once with water before subjecting to an in vitro protein kinase assay or Western blot analysis to assess associated proteins. In vitro kinase assays PI 3-kinase activity was determined by minor modi®cation to prior methods (Whitman et al., 1985). For assaying Sf9 cells expressing Trk or Trkdef+PI 3-kinase, cells were lysed and immunoprecipitated with anti-Trk. Intrinsic receptor kinase activity was activated in vitro and immunoprecipitates were exposed to adult mouse brain as described above. Alternatively, PC12 cells and PC12nnr5 cells expressing Trkdef or Trkdef+PI 3-kinase were treated with 100 ng/ml NGF at 378C for 5 min. Cells were lysed in NP40 lysis bu€er and samples were incubated with antiphosphotyrosine-agarose beads for 2 h at 48C (Peng et al., 1995). All immunoprecipitates were washed three times with 30 mM HEPES (pH 7.4) containing 0.5% Triton X-100, then incubated in 50 ml of assay bu€er containing 0.2 mg/ml PI, 30 mM HEPES (pH 7.4), 30 mM MgCl2, 50 mM ATP, 300 mM adenosine, and 10 mCi [g-32P]ATP in the presence or absence of wortmannin (at 200 nM in DMSO) for 20 min at room temperature. The reaction was terminated by addition of 100 ml 1 M HCl. Lipids were extracted with 1 : 1 methanol:chloroform solution, separated by thin layer chromatography in a solvent mixture of chloroform/methanol/4M NH4OH (9 : 7 : 2), and visualized by autoradiography using Kodak AR ®lm. To assay for Akt activation, a histone H2B in vitro kinase assay was performed (Franke et al., 1995). PC12 cells and PC12nnr5 cells expressing either Trk or Trkdef+PI 3-kinase receptors were serum-starved over night, stimulated with NGF and lysed in NP40 lysis bu€er as described above. Cell

lysates were balanced for protein and immunoprecipitated with anti-Akt-CT (1 : 500 dilution). Immunoprecipitates were washed three times with lysis bu€er, once with water and once with the Akt kinase bu€er (20 mM HEPES ± NAOH, 10 mM MgCl2, 10 mM MnCl2, pH 7.4). Kinase assays were carried out in 50 ml Akt kinase bu€er containing 10 mCi [g-32P]ATP and histone H2B at a ®nal concentration of 0.5 ± 1 mg per reaction. Samples were incubated for 30 min at RT prior to addition of loading dye. Samples were analysed by SDS ± PAGE (12.5%), blotted and exposed to Kodak AR ®lm over night. In vivo phospholipid analysis To assess NGF-induced production of PI-3,4,-P2 and PI3,4,5-P3, 80% con¯uent PC12nnr5 cells expressing either; human wild-type Trk (WT), Trkdef, or Trkdef+PI 3-kinase receptors, or the Trk mutants; Y490F (SHC7), Y785F (PLC-gl7), or Y490F/Y785F (SHC7/PLC-gl7) (Stephens et al., 1994), were incubated overnight in 500 mCi 32P-orthophosphate (Amersham) in phosphate-free RPMI-1640 medium. Cells were stimulated with NGF or EGF (control) for 5 min at 378C, and lipids were extracted (from stimulated and unstimulated cells), and analysed by HPLC as described previously (Rodriguez-Viciana et al., 1994).

Acknowledgements We especially thank KH Vousden at the ABL-Basic Research Program for helpful discussion and support. We thank G Vande Woude at NCI for interest in this study. We thank LA Greene for interesting discussion, TF Franke for advice on performing Akt assays, and DK Morrison for 4G10. This research was sponsored in part by the National Cancer Institute, DHHS under contract with ABL (M Ashcroft, RM Stephens, DR Kaplan), and by the National Cancer Institute of Canada (DR Kaplan). DR Kaplan is the recipient of the Harold E Johns Award from the National Cancer Institute of Canada.

References Andjelkovic M, Suidan HS, Meier R, Frech M, Alessi DR and Hemmings BA. (1998). Eur. J. Biochem., 251, 195 ± 200. Bar-Sagi D and Feramisco JR. (1985). Cell, 42, 841 ± 848. Baxter RM, Cohen P, Obermeier A, Ullrich A, Downes CP and Doza YN. (1995). Eur. J. Biochem., 234, 84 ± 91. Carter AN and Downes CP. (1992). J. Biol. Chem., 267, 14563 ± 14567. Crowder RJ and Freeman RS. (1998). J. Neurosci., 18, 2933 ± 2943. Cunningham ME, Stephens RM, Kaplan DR and Greene LA. (1997). J. Biol. Chem., 272, 10957 ± 10967. Datta K, Franke TF, Chan TO, Makris A, Yang SI, Kaplan DR, Morrison DK, Golemis EA and Tsichlis PN. (1995). Mol. Cell. Biol., 15, 2304 ± 2310. Downward J. (1995). Nature, 376, 553 ± 554. Dudek H, Datta SR, Franke TF, Birnbaum MJ, Yao R, Cooper GM, Segal RA, Kaplan DR and Greenberg ME. (1997). Science, 275, 661 ± 665. Escobedo JK, Kaplan DR, Kavanaugh WM, Turck CW and Williams LT. (1991). Mol. Cell. Biol., 11, 1125 ± 1132. Fagerstrom S, Pahlman S, Gestblom C and Nanberg E. (1996). Cell Growth Di€er., 7, 775 ± 785. Fanger GR, Vaillancourt RR, Heasley LE, Montmayeur JP, Johnson GL and Maue RA. (1997). Mol. Cell. Biol., 17, 89 ± 99.

Franke TF, Kaplan DR and Cantley LC. (1997a). Cell, 88, 435 ± 437. Franke TF, Kaplan DR, Cantley LC and Toker A. (1997b). Science, 275, 665 ± 668. Franke TF, Yang SI, Chan TO, Datta K, Kazlauskas A, Morrison DK, Kaplan DR and Tsichlis PN. (1995). Cell, 81, 727 ± 736. Fukuda M, Gotoh Y, Tachibana T, Dell K, Hattori S, Yoneda Y and Nishida E. (1995). Oncogene, 11, 239 ± 244. Greene LA. (1978). J. Cell Biol., 78, 747 ± 755. Greene LA, Sobeith MM and Teng KK. (1991). Methodologies for the culture and experimental use of the PC12 phechromocytoma cell line. MIT Press, Cambridge, Massachusetts. Greene LA and Tischler AS. (1976). Proc. Natl. Acad. Sci. USA, 73, 2424 ± 2428. Hallberg B, Ashcroft M, Greene LA, Kaplan DR and Downward J. (1998). Oncogene, 17, 691 ± 697. Hempstead BL, Rabin SJ, Kaplan L, Reid S, Parada LF and Kaplan DR. (1992). Neuron, 9, 883 ± 896. Holgado-Madruga M, Moscatello DK, Emlet DR, Dieterich R and Wong AJ. (1997). Proc. Natl. Acad. Sci. USA, 94, 12419 ± 12424. Inagaki N, Thoenen H and Lindholm D. (1995). Eur. J. Neurosci., 7, 1125 ± 1133.

Selective and inducible activation of PI 3-kinase in PC12 cells M Ashcroft et al

Jackson TR, Blader IJ, Hammonds-Odie LP, Burga CR, Cooke F, Hawkins PT, Wolf AG, Heldman KA and Theibert AB. (1996). J. Cell Sci., 109, 289 ± 300. Kaplan DR and Miller FD. (1997). Curr. Opin. Cell Biol., 9, 213 ± 221. Kau€man-Zeh A, Rodriguez-Viciana P, Ulrich E, Gilbert C, Co€er P, Downward J and Evan G. (1997). Nature, 385, 544 ± 548. Kennedy SG, Wagner AJ, Conzen SD, Jordan J, Bellacosa A, Tsichlis PN and Hay N. (1997). Genes Dev., 11, 701 ± 713. Kimura K, Hattori S, Kabuyama K, Shizawa Y, Takayanagi J, Nakamura S, Toki S, Matsuda Y, Onodera K and Fukui Y. (1994). J. Biol. Chem., 269, 18961 ± 18967. Kobayashi M, Nagata S, Kita Y, Nakatsu N, Ihara S, Kaibuchi K, Kuroda S, Ui M, Iba H, Konishi H, Kikkawa U, Saitoh I and Fukui Y. (1997). J. Biol. Chem., 272, 16089 ± 16092. Kouhara H, Hadari YR, Spivak-Kroizman T, Schilling J, Bar-Sagi D, Lax I and Schlessinger J. (1997). Cell, 89, 693 ± 702. Loeb DM and Greene LA. (1993). J. Neurosci., 13, 2919 ± 2929. Loeb DM, Maragos J, Martin-Zanca D, Chao MV, Parada LF and Greene LA. (1991). Cell, 66, 961 ± 966. Loeb DM, Stephens RM, Copeland T, Kaplan DR and Greene LA. (1994). J. Biol. Chem., 269, 8901 ± 8910. Meakin SO, MacDonald JIS, Gryz EA, Kubu CJ and Verdi JM. (1999). J. Biol. Chem., in press. Miller AD and Rosman GJ. (1989). Biotechniques, 7, 980 ± 982, 984 ± 986, 989 ± 990. Miller TM, Tansey MG, Johnson Jr EM and Creedon DJ. (1997). J. Biol. Chem., 272, 9847 ± 9853. Noda M, Ko M, Ogura A, Liu DG, Amano T, Takano T and Ikawa Y. (1985). Nature, 318, 73 ± 75. Obermeier A, Bradshaw RA, Seedorf K, Choidas A, Schlessinger J and Ullrich A. (1994). EMBO J., 13, 1585 ± 1590. Obermeier A, Halfter H, Wiesmuller KH, Jung G, Schlessinger J and Ullrich A. (1993). EMBO J., 12, 933 ± 941. Park EK, Yang SI and Kang SS. (1996). Mols Cells, 6, 494 ± 498. Peng X, Greene LA, Kaplan DR and Stephens RM. (1995). Neuron, 15, 395 ± 406. Philpott KL, McCarthy MJ, Klippel A and Rubin LL. (1997). J. Cell Biol., 139, 809 ± 815.

Qian X, Riccio A, Zhang Y and Ginty DD. (1998). Neuron, 21, 1017 ± 1029. Qui MS and Green SH. (1992). Neuron, 9, 705 ± 717. Rabin SJ, Cleghon V and Kaplan DR. (1993). Mol. Cell Biol., 13, 2203 ± 2213. Raoni S and Bradshaw RA. (1992). Proc. Natl. Acad. Sci. USA, 89, 9121 ± 9125. Ribon V and Saltiel AR. (1996). J. Biol. Chem., 271, 7375 ± 7380. Rodriguez-Viciana P, Warne PH, Dhand R, Vanhaesebroeck B, Gout I, Fry MJ, Water®eld MD and Downward J. (1994). Nature, 370, 527 ± 532. Rodriguez-Viciana P, Warne PH, Vanhaesebroeck B, Water®eld MD and Downward J. (1996). EMBO J., 15, 2442 ± 2451. Segal RA and Greenberg ME. (1996). Ann. Rev. Neurosci., 19, 463 ± 489. Solto€ SP, Rabin SL, Cantley LC and Kaplan DR. (1992). J. Biol. Chem., 267, 17472 ± 17477. Songyang Z, Baltimore D, Cantley LC, Kaplan DR and Franke TF. (1997). Proc. Natl. Acad. Sci. USA, 94, 11345 ± 11350. Stephens RM, Loeb DM, Copeland TD, Pawson T, Greene LA and Kaplan DR. (1994). Neuron, 12, 691 ± 705. Summers MD and Smith GE. (1987). A manual of methods for baculovirus vector and insect cell culture procedures, College Station, Texas, Texas Agricultural Center. Szeberenyi J, Cai H and Cooper GM. (1990). Mol. Cell Biol., 10, 5324 ± 5332. Teng KK, Courtney JC, Henegouwen PB, Birge RB and Hempstead BL. (1996). Mol. Cell Neurosci., 8, 157 ± 170. Toker A and Cantley LC. (1997). Nature, 387, 673 ± 676. Valius M and Kazlauskas A. (1993). Cell, 73, 321 ± 334. Whitman M, Kaplan DR, Scha€hausen B, Cantley L and Roberts TM. (1985). Nature, 315, 239 ± 242. Wood KW, Qi H, D'Arcangelo G, Armstrong RC, Roberts TM and Halegoua S. (1993). Proc. Natl. Acad. Sci. USA, 90, 5016 ± 5020. Wymann MP, Bulgarelli-Leva G, Zvelebil MJ, Pirola L, Vanhaesebroeck B, Water®eld MD and Panayotou G. (1996). Mol. Cell Biol., 16, 1722 ± 1733. Xia Z, Dickens M, Raingeaud J, Davis RJ and Greenberg ME. (1995). Science, 270, 1326 ± 1331. Yao R and Cooper GM. (1995). Science, 267, 2003 ± 2006.

4597