Fibroblast Growth Factor-2 Suppression of Tumor Necrosis Factor (cid:97) -Mediated Apoptosis Requires Ras and the Activation of Mitogen-activated Protein Kinase*

Treatment of L929 cells with tumor necrosis factor (cid:97) (TNF (cid:97) ) activates a programmed cell death pathway resulting in apoptosis. We investigated the intracellular signaling pathways activated in L929 cells by TNF (cid:97) . TNF (cid:97) robustly activates Jun kinase (JNK), a member of the mitogen-activated protein kinase (MAPK) family. In addition, p42 MAPK is activated, but a 10-fold greater concentration of TNF (cid:97) was required for substantial MAPK activation than was needed for maximal JNK stimula- tion. Simultaneous treatment of L929 cells with fibroblast growth factor (FGF-2) significantly reduced the apoptotic response to TNF (cid:97) . FGF-2 substantially activated the Raf/MEK/MAPK (where MEK is mitogen-acti- vated protein kinase kinase) pathway but did not affect TNF (cid:97) activation of JNK. These results indicate that although JNK may play an important role in transmitting the TNF (cid:97) signal from the cell surface to the nucleus, activation of the JNK pathway is not sufficient to induce apoptosis. Expression of dominant-negative Asn-17 Ras in L929 cells diminished the FGF-2 stimulation of p42 MAPK and eliminated the protective effect of FGF-2. Asn-17 Ras expression did not affect JNK activity and had no effect on TNF (cid:97) activation of JNK. Pharmacolog-ical inhibition of MEK-1 activity by incubation of cells with the compound that activation of the pathway mediates protective effect against the important role that integration of multiple pathways plays in the regulation of cell milk in Tris-HCl, pH 7.5, buffered saline. Ras was detected with Y-13259 anti-Ras mono- clonal antibody (45) followed by enhanced chemiluminescence (Amer-sham Corp.) using horseradish peroxidase-anti-mouse IgG (Bio-Rad). Quantitation of Data— PhosphorImager analysis of phosphorylated proteins provided a quantitative measure of kinase activation in arbi- trary phosphorimaging units. Statistical analysis was performed using the JMP program (SAS Institute Inc., Cary, NC), and the method of Tukey and Kramer was used to determine statistical differences.

Treatment of L929 cells with tumor necrosis factor ␣ (TNF␣) activates a programmed cell death pathway resulting in apoptosis. We investigated the intracellular signaling pathways activated in L929 cells by TNF␣. TNF␣ robustly activates Jun kinase (JNK), a member of the mitogen-activated protein kinase (MAPK) family. In addition, p42 MAPK is activated, but a 10-fold greater concentration of TNF␣ was required for substantial MAPK activation than was needed for maximal JNK stimulation. Simultaneous treatment of L929 cells with fibroblast growth factor (FGF-2) significantly reduced the apoptotic response to TNF␣. FGF-2 substantially activated the Raf/MEK/MAPK (where MEK is mitogen-activated protein kinase kinase) pathway but did not affect TNF␣ activation of JNK. These results indicate that although JNK may play an important role in transmitting the TNF␣ signal from the cell surface to the nucleus, activation of the JNK pathway is not sufficient to induce apoptosis. Expression of dominant-negative Asn-17 Ras in L929 cells diminished the FGF-2 stimulation of p42 MAPK and eliminated the protective effect of FGF-2. Asn-17 Ras expression did not affect JNK activity and had no effect on TNF␣ activation of JNK. Pharmacological inhibition of MEK-1 activity by incubation of cells with the compound PD 098059 blocked p42 MAPK activation and FGF-2 protection against apoptosis. Interestingly, activated Val-12 Ras expression substantially enhanced TNF␣-mediated apoptosis in L929 cells, but Val-12 Ras did not constitutively activate MAPK in L929 cells and FGF-2 partially protected Val-12 Ras-expressing cells from TNF␣-mediated apoptosis. Our data indicate that activation of the MAPK pathway mediates an FGF-2 protective effect against apoptosis and highlights the important role that integration of multiple intracellular signaling pathways plays in the regulation of cell growth and death.
Tumor necrosis factor ␣ (TNF␣) 1 is a multifunctional cyto-kine secreted primarily by activated monocytes (reviewed in Ref. 1). It has a wide range of biological activities depending upon cell type, stage of differentiation, and transformation state. TNF␣ acts as a growth factor for fibroblasts (2,3), is cytotoxic toward certain cells and tumors (4), induces monocyte differentiation of the human HL-60 myeloid leukemia cell line (5,6), represses adipocyte (7) and myoblast differentiation (8), and mediates endotoxic shock (9). The pleiotropic effects of this cytokine make it an important mediator in processes as diverse as proliferation, differentiation, and cytotoxicity. TNF␣ exerts these responses by binding to two cell surface receptors, the 55-kDa TNFR (p55 TNFR ) and the 75-kDa TNFR (p75 TNFR ) (10 -13). The receptors are single transmembrane spanning glycoproteins present on almost all cells analyzed (10 -15). The extracellular domain of the p55 TNFR is homologous to the extracellular domains of the low affinity nerve growth factor receptor, the Fas/APO1 receptor, CD40, OX40, and CD27 (10,12). The p55 TNFR and Fas share a 65-residue homology region in the cytoplasmic domains (16,17), which deletion studies have implicated in the TNF␣ signaling cascade leading to apoptosis (18,19). Most of the known TNF␣ responses occur by activation of the p55 TNFR . However, thymocyte proliferation is associated with p75 TNFR , and cytotoxicity may be a function of p75 TNFR acting alone or in concert with the p55 TNFR (20).
Apoptosis involves the activation of a specific suicide program within a cell. It occurs when a cell initiates a series of biochemical and morphological events that result in nuclear disintegration and eventual fragmentation of the dying cell into a cluster of membrane-bound apoptotic bodies (21). Apoptosis is responsible for such diverse activities as the elimination of cells during normal embryological development and determination of the immune receptor repertoire (reviewed in Refs. [22][23][24]. Apoptosis can be triggered in multiple ways, but it is not yet known whether different inducers of apoptosis have a common pathway or whether there are multiple pathways with perhaps some common components. In many peptide-hormone receptor systems signal transduction to the nucleus involves the sequential activation of protein kinases. The extracellular response kinase group of mitogenactivated protein kinases (p42 MAPK and p44 MAPK ) are activated by growth factors via a Ras/Raf-dependent signal transduction pathway (reviewed in Refs. 25,26). In contrast, the Jun kinase/ stress-activated protein kinase members of MAPKs are activated by proinflammatory cytokines and environmental * This work was supported by National Institutes of Health Grants DK37871, GM30324, DK48845, and CA58157 (to G. L. J.) and CA60455 (to A. M. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
TNF␣ has been shown to initiate apoptotic cell death and DNA fragmentation in several mammalian cell lines, including the murine fibrosarcoma cell line L929 (32,33). TNF␣ also has been shown to activate p42/p44 MAPK in this cell line (34). Recently JNKs were shown to be activated by TNF␣ (29,30,35), and activation of the JNK pathway correlated with enhanced apoptosis of PC12 cells in response to trophic factor deprivation (36). We have characterized the regulation of MAPKs and JNKs in L929 cells challenged with TNF␣ and fibroblast growth factor (FGF-2). We show that TNF␣ preferentially activates JNK in L929 cells but that JNK activation is not sufficient to induce apoptosis, because FGF-2 mediates a protective effect against TNF␣-mediated apoptosis without affecting JNK activation. Furthermore, our data indicate that p42/p44 MAPK activation is required for FGF-2 suppression of TNF␣-mediated apoptosis.

MATERIALS AND METHODS
Cell Lines and Culture-L929 cells (ATCC CCL1) were maintained in Dulbecco's modified Eagle's medium with 5% newborn calf serum and 5% bovine calf serum supplemented with 100 g/ml streptomycin and 100 units/ml penicillin. The cells were grown in 10-cm dishes at 37°C in 7.5% CO 2 . Cells were made quiescent where indicated by incubation in Dulbecco's modified Eagle's medium and 0.1% bovine serum albumin for 24 h. Recombinant murine TNF␣ and recombinant human FGF-2 (147 amino acids) were from R&D Systems, Minneapolis, MN. Cells were pretreated where indicated with the MEK-1 inhibitor PD 098059 (a generous gift from A. Saltiel, Parke-Davis) for 1 h at 37°C. Cells were stimulated by incubation with the indicated cytokine or growth factor for various times at 37°C. Stimulation was stopped by rinsing the plates twice with ice-cold phosphate-buffered saline (PBS) and lysing the cells in the appropriate lysis buffer. Cells were scraped from the plates, and nuclei were pelleted for 10 min at 14,000 rpm in a microcentrifuge.
JNK Assay-JNK activity was measured using a solid state kinase assay in which glutathione S-transferase-c-Jun (1-79) (GST-Jun) bound to glutathione-Sepharose 4B beads was used to affinity-purify JNK and then JNK activity was measured in an in vitro kinase assay using the Sepharose-bound GST-Jun as a substrate (28). Stimulated or unstimulated cells were lysed in 0.5% Nonidet P-40, 20 mM HEPES, pH 7.2, 100 mM NaCl, 2 mM dithiothreitol, 1 mM EDTA, 1.0 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, and the nuclei were pelleted. Lysates were normalized for protein content. JNK was affinity-purified from 50 to 100 g of cell lysate by the addition of 10 l of GST-Jun-Sepharose slurry (2 g of GST-Jun). Binding to GST-Jun efficiently isolates the two major forms of JNK (p45 and p55), and under the conditions used JNK isolation was linear for 10 -250 g of cell lysate. The lysates were rotated at 4°C for 1-3 h. Beads were washed twice in lysis buffer and then twice in PAN (10 mM PIPES, pH 7.0, 100 mM NaCl, 21 g/ml aprotinin). Kinase reactions were carried out at 30°C for 15 min in 20 mM HEPES, pH 7.2, 20 mM ␤-glycerophosphate, 10 mM p-nitrophenyl phosphate, 10 mM MgCl 2 , 1 mM dithiothreitol, 50 M sodium vanadate, 10 Ci [␥-32 P]ATP 4300 Ci/mmol. The kinase reaction was linear from 0 to 30 min.
MAPK Assay-MAPK activity was measured exactly as described previously (37) with the exception that MonoQ FPLC fractionation was replaced by step elution from a DEAE-Sephacel column using 0.5 M NaCl in loading buffer. The eluate was assayed in triplicate using the epidermal growth factor receptor 662-681 peptide (EGFR 662-681 ) as a selective substrate for MAPK activity (38).
Raf Activation Assay-Cells were serum-starved and challenged in the presence or absence of the appropriate cytokine or growth factor, as described above. Cells were lysed by scraping in ice cold RIPA buffer (50 mM Tris-HCl, pH 7.2, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1.0% Triton X-100, 10 mM sodium pyrophosphate, 25 mM ␤-glycerophosphate, 2 mM sodium vanadate, 2.1 g/ml aprotinin), and the nuclei were pelletted. The supernatants were normalized for protein content and precleared with protein A-Sepharose prior to immunoprecipitation with rabbit antiserum to the C terminus of c-Raf, rabbit antiserum to a-Raf, or rabbit antiserum to b-Raf (Santa Cruz Biotech., Santa Cruz, CA) and protein A-Sepharose for 2-3 h at 4°C. The beads were washed twice with ice-cold RIPA and twice with PAN. A third of the immunoprecipitate was diluted with SDS sample buffer and used for immunoblot analysis. The remainder was resuspended in kinase buffer (20 mM PIPES, pH 7.0, 10 mM MnCl 2 , 150 ng of kinase-inactive MEK-1, 30 Ci of [␥-32 P]ATP, and 20 g/ml aprotinin) in a final volume of 40 l for 30 min at 30°C. Wild-type recombinant MEK-1 was autophosphorylated in parallel as a marker. Reactions were terminated by the addition of 12.5 l of 5 ϫ SDS sample buffer, boiled, and subjected to SDS-PAGE and autoradiography.
Neutral Red Assay-Uptake of the dye neutral red was used as one measure of cell viability following cytokine or growth factor treatment (39). 1.5 ϫ 10 4 -2.5 ϫ 10 4 L929 cells/well were plated in 12-well tissue culture dishes in 1.25 ml of media. Cells were treated for 15-20 h with various concentrations of TNF␣ and/or FGF-2. 2.5 l of 1% neutral red was added to the wells and incubated for 2 h at 37°C. The media were removed and the wells washed three times with 1 ml each of 37°C PBS. The neutral red was extracted with 1.0 ml of 50% ethanol, 50 mM sodium citrate, pH 4.2, and absorbency was measured at 540 nM.
Propidium Iodide Staining-Cells were plated on glass chamber slides (Nunc, Naperville, Il) at a concentration of 0.2-0.6 ϫ 10 5 cells/ml. Ras expression was induced with 5 mM IPTG in Dulbecco's modified Eagle's medium with 0.1% bovine calf serum for 8 -12 h. Cells were exposed to TNF␣ (5 ng/ml) and/or FGF-2 (500 pg/ml) in Dulbecco's modified Eagle's medium with 0.1% bovine calf serum for 16 h. The parental LACI expressing cell line (see below) was used as a control. Cells were washed twice in PBS, fixed in acetone:methanol (1:1) at Ϫ20°C for 5 min, air-dried, washed twice in PBS, stained with 1 g/ml propidium iodide in PBS for 20 min, washed in PBS, washed in H 2 O, and mounted in 25% glycerol/PBS. Propidium iodide fluorescence was observed using a Nikon inverted microscope equipped with epifluorescence and a 580-nm filter. Images were analyzed using IP lab.
Cell Transfections-L929 cells were transfected by CaPO 4 (40) with the vector 3ЈSS (Stratagene, La Jolla, CA) expressing the LACI repressor. Stable clones were selected in 200 g/ml hygromycin (Calbiochem) and screened for LACI expression by indirect immunofluorescence using rabbit antisera to LACI (Stratagene, La Jolla, CA) and FITC-donkey anti-rabbit. One clone expressing a high level of nuclear LACI was then transfected with hemagglutinin-tagged inhibitory Asn-17 Ras (41) or activated Val-12 Ras (42-44) cloned into the LACI repressible pOPRSV1 vector. Stable clones were selected in 500 g/ml G418 and screened for inducible expression of Ha-Ras by immunoblotting. Incubation in 5 mM isopropyl-1-thio-␤-D-galactopyranoside (IPTG) for 8 -24 h was used to induce Ras expression. Several independent, inducible Asn-17 Ras or Val-12 Ras clones were isolated, and two each were chosen for further analysis.
Quantitation of Data-PhosphorImager analysis of phosphorylated proteins provided a quantitative measure of kinase activation in arbitrary phosphorimaging units. Statistical analysis was performed using the JMP program (SAS Institute Inc., Cary, NC), and the method of Tukey and Kramer was used to determine statistical differences.

FGF-2 Protects L929 from TNF␣-mediated Apoptosis-TNF␣
activates a cell death program resulting in the apoptosis of L929 cells (33). Fig. 1A shows that treatment of L929 cells overnight with TNF␣ resulted in substantial cell death using the neutral red assay as a measure of viable cells (see "Materials and Methods"). The time course of cell death was dependent on the concentration of TNF␣. Treatment with 10 ng/ml TNF␣ resulted in greater than 40% of the L929 cells being apoptotic in 15 h; 1 ng/ml TNF␣ required 24 -48 h to induce a similar level of L929 cell death (not shown). Serum and growth factor withdrawal induces apoptosis in several cell systems (46,47), indicating that growth factors have a protective effect against apoptosis. Consistent with this observation was our finding that FGF-2 affected TNF␣-mediated apoptosis (Fig.  1B). Incubation of L929 cells with TNF␣ in the presence of FGF-2 significantly reduced the cell death response. A concentration of 0.5 ng/ml FGF-2 was effective at blocking TNF␣mediated cell death. The protective effect of FGF-2 was not simply due to an increased proliferative response of L929 cells, because FGF-2 in the absence of TNF␣ did not measurably increase cell number (Fig. 1B).
Regulation of JNK and MAPK by TNF␣ and FGF-2-TNF␣ has been previously shown to activate p42/p44 MAPK in L929 cells (34), but recent studies have indicated that TNF␣ is a potent activator of the Jun kinase (JNK) members of the MAPK family (29,30,35). Analysis of the time course and dose response of TNF␣ on L929 cells demonstrated significant differences in the activation of JNK and p42/p44 MAPK activity. Extracts from TNF␣-treated versus control L929 cells were assayed for JNK activity using GST-c-Jun (1-79) as substrate. TNF␣ induced a transient increase in JNK activity that peaked at 10 -15 min and returned to 2-fold above basal JNK activity 1-2 h poststimulation ( Fig. 2A). Maximal JNK activation was achieved at 1 ng/ml TNF␣, and 0.1 ng/ml TNF␣ activated JNK greater than 4-fold (Fig. 2B). TNF␣ stimulation of p42/p44 MAPK activity was slightly more rapid than JNK activation, reaching maximal stimulation in 5-10 min that returned to near basal levels by 30 min (Fig. 2C). The dose-response curve for p42/ p44 MAPK activation is dramatically shifted to higher TNF␣ concentrations than that for JNK (Fig. 2D). Greater than 10 ng/ml TNF␣ was required to stimulate p42/p44 MAPK 2-3-fold; at 1 ng/ml TNF␣ the MAPK activity was stimulated only 20% above basal, a concentration of TNF␣ that gave maximal JNK activation. Thus, TNF␣ preferentially regulates the JNK pathway relative to p42/p44 MAPK in L929 cells. These findings indicate that the localized concentration of cytokines such as TNF␣ will determine the selectivity and magnitude of cellular JNK and p42/p44 MAPK responses.
In contrast to proinflammatory cytokines such as TNF␣, growth factor receptor tyrosine kinases are generally mitogenic in fibroblasts and stimulate the p42/p44 MAPK pathway. The FGF-2 receptors possess intrinsic tyrosine kinase activity and are present on L929 cells. Fig. 3 demonstrates that FGF-2 stimulates a robust activation of MAPK in L929 cells. Concentrations of 0.25-0.5 ng/ml FGF-2 gave maximal stimulation of MAPK activity. Fractionation of stimulated cell lysates by MonoQ fast pressure liquid chromatography indicated that both p42 and p44 MAPK were activated by FGF-2 (not shown). Activation of the MAPK pathway by tyrosine kinases involves Ras and the Raf serine-threonine protein kinases. Immunoblotting demonstrated that b-Raf and c-Raf are expressed in L929 cells (not shown). Treatment of L929 cells with FGF-2 resulted in the activation of both b-Raf and c-Raf (Fig. 4), measured by their ability to phosphorylate a recombinant kinase-inactive MEK-1 protein (37). MEK-1 is the protein kinase phosphorylated and activated by Raf, which in turn phosphorylates MAPK on both a tyrosine and threonine resulting in MAPK activation (48 -51). In contrast, TNF␣ does not significantly activate either isoform of Raf in L929 cells (Fig. 4). Fig. 5 demonstrates that 10 ng/ml TNF␣ has only modest stimulatory effects on MAPK activity (C), and 2.5 ng/ml FGF-2 has little or no effect on JNK activity (B). These concentrations of FGF-2 and TNF␣ give maximal activation of MAPK and JNK, respectively. Co-stimulation of L929 cells with FGF-2, at concentrations that show partial protection against TNF␣-mediated killing, did not alter the magnitude of JNK activation in response to TNF␣. Similarly, co-stimulation of L929 cells with TNF␣, at concentrations capable of causing cell death, had little or no effect on FGF-2 stimulation of MAPK activity (C). Thus, in relation to JNK and MAPK, TNF␣ and FGF-2 receptors independently regulate the activity of these two sequential protein kinase pathways in L929 cells.

FGF-2 and TNF␣ Independently Regulate Cytoplasmic Protein Kinase Cascades-
Inducible Expression of Inhibitory and Activated Ras Influences Apoptosis-Ras activation is required for many of the phenotypic responses resulting from the activation of tyrosine kinases. Signaling by the FGF-2 receptor involves several different effector pathways including Ras activation. To test the involvement of Ras in the FGF-2-protective response, the Lac Switch-inducible expression system (see "Materials and Methods") was used to control the expression of inhibitory Asn-17 Ras and constitutively activated Val-12 Ras in L929 cells. Fig.  6 shows the functional consequence of expressing inhibitory Asn-17 Ras or activated Val-12 Ras on MAPK and JNK activation in response to FGF-2 and TNF␣, respectively. IPTGregulated expression of the hemagglutinin epitope-tagged Ras mutants (Asn-17 and Val-12 Ras) is shown in D. Expression of Asn-17 Ras significantly blunted FGF-2 stimulation of MAPK (A) but had no effect on TNF stimulation of JNK (C). It was surprising that Asn-17 Ras was unable to completely block MAPK activation and implies that the level of Asn-17 Ras expression was too low to completely block endogenous Ras or that FGF-2 activates MAPK via a Ras-independent pathway. We found that the clones with the highest inducible levels of Asn-17 Ras expression were most effective at blocking MAPK (Fig. 6, A and D). Therefore, it is likely that the partial block of MAPK activation is due to insufficient expression of Asn-17 Ras, although we cannot rule out a Ras-independent pathway for activation of MAPK by FGF-2. With two independent clones, expression of Val-12 Ras did not constitutively activate the MAPK pathway but did appear to enhance FGF-2 stimulation of MAPK (B). Val-12 Ras expression also had no effect on TNF␣ stimulation of JNK activity (C). We have previously shown that constitutive activation of MAPK by Val-12 Ras is a cell type-specific phenomenon in that MAPK is constitutively activated in Val-12 Ras expressing NIH 3T3 cells but is not activated by Val-12 Ras in Rat 1a cells (52). Differential regulation of phosphatases that inactivate the MAPK pathway has been proposed to explain this observation (53).
Expression of Asn-17 Ras did not affect TNF␣-induced apoptosis of L929 cells (Figs. 7 and 8); Asn-17 Ras did, however, markedly inhibit the ability of FGF-2 to protect cells against TNF␣-mediated cell death. These findings indicated that functional Ras signaling is not required for the TNF␣-induced apoptotic response but is required for the protective action of FGF-2. Strikingly, constitutively activated Val-12 Ras markedly enhanced TNF␣-stimulated apoptosis but had little or no effect on the apoptotic index of L929 cells in the absence of TNF␣ (Fig. 7). This observation indicates that Val-12 Ras is functional in L929 cells, despite the fact MAPK is not constitutively activated in this cell line and implies that activated Ras likely regulates pathways in addition to MAPK that are involved in apoptosis. Co-stimulation with FGF-2 and TNF␣ resulted in a diminished apoptotic response relative to TNF␣ alone in Val-12 Ras-expressing cells, indicating that FGF-2 pathways required for protection against TNF␣-stimulated cell death were functional in these cells (Fig. 8). Thus, inhibitory Ras expression prevented FGF-2 protective responses and activated Ras-enhanced TNF␣ killing. The results suggest multiple Ras-dependent events are involved in controlling apoptosis and the role of Ras signaling can be either positive or negative in regulating the phenotypic response to cytokines such as TNF␣.
Inhibition of MEK and MAPK Stimulation Prevents FGF-2 Protection from Apoptosis-The Parke-Davis compound, PD 098059, inhibits the dual specificity protein kinase, MEK-1,

FGF-2 Suppression of TNF␣-mediated Apoptosis
which specifically activates p42/p44 MAPK (54). PD 098059 did not inhibit JNK kinase or the activation of JNK (not shown). Pretreatment of L929 cells with PD 098059 inhibited FGF-2 stimulation of MAPK activity (Fig. 9A). The PD 098059 compound had no effect on TNF␣-mediated apoptosis but inhibited the protective action of FGF-2 (Fig. 9B). Thus, MEK activation of MAPK is required for FGF-2 protection against TNF␣-mediated apoptosis. Interestingly, the phosphatidylinositol 3-kinase inhibitor, wortmannin, did not influence the cell death response to TNF␣ nor did it inhibit the protective response to FGF-2 (not shown). Treatment of L929 cells with wortmannin had no effect on the ability of FGF-2 to stimulate MAPK activity. Apparently, phosphatidylinositol 3-kinase activity is not required for the action of either TNF␣ or FGF-2 on the control of the cell death program in L929 cells. DISCUSSION TNF␣ induces apoptosis of L929 cells and FGF-2 is protective against this cell death response. Our results indicate that the activation of JNK in response to TNF␣ stimulation of L929 cells is not sufficient for the induction of cell death. TNF␣ maximally stimulates JNK activity in the presence of FGF-2 concentrations that are capable of protecting against cell death. Signals in addition to JNK activation must be involved in the TNF␣-mediated death response. The FGF-2 protective response was only partial in that not all the cells were prevented from dying in response to TNF␣ treatment. This may, in part, be related to cell cycle-dependent signaling by TNF␣ and FGF-2; the L929 cells used in these studies were asynchronous so we cannot rule out this possibility. Our findings also demonstrate that Ras is involved in integrating responses that control apoptosis. Expression of activated or inhibitory Ras influences TNF␣ killing of L929 cells. The mechanism for enhanced TNF␣ killing of L929 cells resulting from Val-12 Ras expression is unclear, although it has been observed in C3H mouse fibroblasts as well (55). It may involve an alteration in the expression of specific genes such as c-jun, c-fos, and c-myc which appear to be involved in both growth and apoptotic responses (35, 56 -63). In contrast, the effect of inhibitory Asn-17 Ras appears to primarily be the inhibition of MAPK activation in response to FGF-2. This finding is substantiated by the loss of FGF-2 protection against TNF␣-mediated apoptosis by the MEK inhibitor PD 098059. Studies using the fungal metabolite, wortmannin, demonstrated that phosphatidylinositol 3-kinase was not involved in FGF-2 protection against apoptosis in L929 cells.
Recently, it was demonstrated using PC12 cells that the JNK pathway was involved in mediating apoptosis in response to serum deprivation and that activation of the MAPK pathway was protective against serum deprivation (36). Phosphatidylinositol 3-kinase activity has also been reported to be necessary to protect PC12 cells from serum deprivation-induced apoptosis (64). Interestingly, the expression of Asn-17 Ras protected PC12 cells from nerve growth factor withdrawal-induced apoptosis (65). The findings indicated that Asn-17 Ras maintained PC12 cells in a quiescent state that allowed them to survive in the absence of trophic factors. Removal of trophic factors from PC12 cells appeared to induce an aberrant proliferative response that resulted in apoptosis. Our findings using Asn-17 Ras expression in L929 cells contrast with those in PC12 cells. TNF␣ induced apoptosis in growing L929 cells; Asn-17 Ras expression did not affect the apoptotic response, whereas Val-12 Ras expression significantly enhanced apoptosis. Thus, the involvement of Ras-dependent signaling on apoptotic responses of cycling versus quiescent cells may be quite different.
In human B cells, cross-linking of surface IgM stimulated a host of signaling pathways including MAPK but not JNK and resulted in apoptosis (66). CD40, a member of the TNF receptor family, activated JNK while rescuing B cells from anti-IgMmediated apoptosis (66). Thus, in human B cells MAPK activation is insufficient to protect against apoptosis, and signals including the stimulation of JNK are generated during a protective response. Clearly, the integration of multiple signals appears to be required for apoptosis.
The overlap of signals involved in committing cells to growth or apoptosis is also evident in many transformed cell types. Tumors frequently have a high growth rate but also a high apoptotic index (58,67). The growth rate is simply greater than the apoptotic rate so that the net result is tumor expansion. In addition, transformed cells frequently have selected mutations and growth factor autocrine loops to inhibit apoptosis. For example, Ras function has been shown to be involved in both transformation and protection against apoptosis in Bcr-Abltransformed cells (68,69).
Cumulatively, the results in different cell types indicate it is the integration of multiple signals from cytokines and growth factors that determines the commitment to apoptosis. Similarly, integration of multiple signals and not a single dominant signaling pathway is likely involved in the commitment to growth or differentiation. The requirement for signal integration may allow for specific checkpoints so that cells do not die or grow inappropriately. In this regard, cell systems where specific cytokines or growth factors are added or removed are most relevant in defining the integration of signals controlling growth versus death.
The implication of our findings is that it should be possible to define signal pathways and their integration that control apoptosis in specific cell types. As these findings are further defined, it will be possible to develop strategies to selectively induce a cell type-specific apoptotic response. Development of gene therapy, cytokine, and drug treatments may be possible to selectively promote the death of undesirable cell populations in animals.