¹Développement, vieillissement et pathologie de la rétine, INSERM U. 450, affiliée CNRS, Association Claude Bernard - 29, rue Wilhem, 75016 Paris, France

²INSERM U. 348, IFR Circulation, 75010 Paris, France

Correspondence to: Frédéric Mascarelli, Développement, vieillissement et pathologie de la rétine, INSERM U. 450, affiliée CNRS, Association Claude Bernard - 29, rue Wilhem, 75016 Paris, France

Retinal pigmented epithelial (RPE) cells are of central importance in the maintenance of neural retinal function. Changes in the RPE cells associated with repair activities have been described as metaplasia, while RPE cell apoptosis is responsible for the development of a variety of retinal degenerations. We investigated the regulation of the anti-apoptotic properties of the fibroblast growth factors (FGF) 2 in serum-free cultures of RPE cells. In the absence of serum, confluent stationary RPE cells died by apoptosis via a caspase 3-dependent pathway. The addition of FGF2 greatly reduced apoptosis over a 7-day culture period. We demonstrated the involvement of an autocrine loop involving endogenous FGF1 in the mechanisms that govern FGF2-induced resistance to apoptosis by showing: (1) higher levels of apoptosis in cells treated with antisense FGF1 oligonucleotide or after neutralization of excreted FGF1; (2) the long-term activation of FGFR1 and of ERK2, (3) the inhibition of FGFR1 and ERK2 activation and an increase in apoptosis if excreted FGF1 was neutralized. FGF2 also increased the de novo synthesis and the production of Bcl-xl before the onset of apoptosis. Both inhibition of ERK2 activation, which decreased Bcl-xl synthesis, and downregulation of Bcl-x by antisense oligonucleotide treatment inhibited the survival-promoting activity of FGF2. Thus, FGF2-induced cell survival is a progressive adaptive phenomenon involving ERK2 activation by excreted FGF1 and ERK2-dependent Bcl-x production.

FGF; Bcl-x; ERK2; apoptosis; retinal pigmented epithelial cells

Introduction

The prototype members of the fibroblast growth factors (FGFs) family, FGF1 and FGF2 were originally isolated as mitogens with multiple biological activities including angiogenesis, mitogenesis, cellular differentiation and repair (for reviews see Burgess and Maciag, 1989; Baird and Klagsbrun, 1991). FGF1 and 2 have been implicated in tumor development and malignant progression (for review see Rifkin and Moscatelli, 1989; Courlier et al., 1997). FGF1 and 2 lack a classic signal peptide (Abraham et al., 1986; Jaye et al., 1986), implying that they are not secreted by the classical secretion pathway. There is evidence that FGF1 and 2 are exported from the cell and subsequently act as autocrine or paracrine factors (Vlodavski et al., 1982; Sato and Rifkin, 1988). FGFs act through tyrosine-kinase receptors (FGFR1-R4) (for reviews see Jaye et al., 1992; Partanen et al., 1992; Givol and Yayon, 1990). FGFRs activation causes tyrosine phosphorylation of the receptor itself and of intracellular proteins, including phospholipase C (PLC) (Burgess et al., 1990), and extracellular signal-regulated kinases (ERKs) (Cobb et al., 1991). Overexpression of FGFs and FGFRs in cancer cells suggest that paracrine stimulation of cells may play an important role in cancer progression. Furthermore, the opposite effects of FGF2 on cell growth in human cancer cells, depending on the activation level of the ERK pathway, also indicate that modulation of the level of FGF intracellular signaling controls cellular activity in tumors (Liu et al., 1998). In addition, the widespread presence of FGFs and FGFRs in non-dividing cells suggests that FGF signaling plays important roles in cell survival. Various studies have shown that FGF1 and 2 increase survival by preventing apoptosis. However, little is known about the mechanisms underlying exogenous FGF-induced cell survival. In addition, it is now a matter of debate whether FGF1 and 2 actually do promote cell survival. Recently, it has been shown that FGF2 induces apoptosis in rat myofibroblasts (Funato et al., 1997), increases apoptosis of neural retina cells in developing chicks (Yokoiyama et al., 1997) and promotes apoptosis in human breast cancer cells (Wang et al., 1998), whereas FGF1 increases peroxynitrite-induced apoptosis in murine fibroblasts (Shin et al., 1996) and PC12 cells (Spear et al., 1998).

In normal adult retina, all cell layers express FGF2 as cells no longer differentiate or proliferate. In vivo, despite the presence of FGF2 in the retinal interphotoreceptor matrix, retinal pigmented epithelial (RPE) cells have a limited proliferation capacity, corresponding to the normal increase in retinal space associated with growth and age. Increased levels of FGF2 with elevated levels of FGF2 excretion have been detected in cells derived from pigment cell tumors and RPE-associated choroidal melanomas in humans (Enzmann et al., 1998). Although RPE cells are unable to divide in vivo, they may still require survival factors to inhibit their apoptosis. The survival or RPE cells is critical to the maintenance of retina function. In retinal dystrophy in rat, cell apoptosis is prevented by a single subretinal injection of exogenous FGF2, demonstrating that FGF2 can act as a survival-promoting trophic factor in the retina in vivo (Faktorovich et al., 1990). In culture, RPE cells produce FGF1 and FGF2 (Chen et al., 1996), FGF high affinity receptors (FGFR) and in particular FGFR1 (Guillonneau et al., 1997). In growing RPE cells, ERKs are rapidly and transiently activated in response to exogenous FGFs (Malecaze et al., 1993). We have shown that FGF2 stimulated the production of endogenous FGF1 (Guillonneau et al., 1997) and that high-level expression of endogenous FGF1 is correlated with a reduction in apoptosis during RPE cell aging (Guillonneau et al., 1998b). This suggests that there is an FGF paracrine pathway supporting RPE cell survival in vitro, though no loop of FGF1 activation involving endogenous FGF2 was demonstrated in vivo. We investigated several aspects of FGF2 signaling, including the activation and synthesis of FGFR1 and ERK2 and the production of anti-apoptotic members of the Bcl-2 protein family in quiescent, confluent primary RPE cell cultures as a function of cell survival. We used antisense oligonucleotides (ODNs) and pharmacological strategies. We found that the sustained upregulation of Bcl-x production mediated by long-term activation of the ERK2 pathway and the FGF1 paracrine pathway plays a key role in integrating and transmitting exogenous FGF2 signals for cell survival.

Results

After serum depletion, FGF2-induced RPE cell rescue from caspase 3-dependent apoptosis requires FGF1 synthesis

In the presence of serum, there was no significant cell death of confluent quiescent RPE cells over a 7-day culture period. In the absence of serum, we found that during the first 3 days of culture, the number of TUNEL-labeled nuclei of RPE cells, undergoing programmed cell death (PCD), was similar to that for cells cultured in the presence of serum (Figure 1a). Thereafter, the number of cells undergoing PCD increased dramatically and was 13 and 21 times higher on days 5 and 7 respectively, than the basal level of PCD in RPE cell cultures observed on day 1. Apoptosis involves a tightly regulated death pathway including the activation of cysteine proteases of the caspase family, so we investigated apoptosis signaling with peptide inhibitors designed to mimic known sequences of caspase substrates. Treatment of serum-deprived RPE cells with YVAD, which is a direct substrate for caspase1-like activity, did not inhibit cell apoptosis, suggesting that this event is not caspase1-dependent (Figure 1a). In contrast, treatment with DEVD, a direct substrate for caspase 3-like activity, reduced apoptosis levels by factors of 2.7 on day 5 and 7 respectively, indicating that RPE cell apoptosis induced by serum withdrawal is dependent on caspase 3-like activity. The addition of a single dose of FGF2 (20 ng/ml) on day 0 reduced by factors of 4 and 3.2 the rate of apoptosis on days 5 and 7 respectively (Figure 1b).

In RPE cells, addition of exogenous FGF2 caused a large increase in FGF1 production and excretion after 3 days of culture (Guillonneau et al., 1997). Thus, we tested whether FGF1 was involved in apoptosis protection using an antisense strategy to deplete FGF1. Treatment with FGF1 antisense ODNs specifically inhibited FGF1 synthesis by 90% over the 7-day period of culture (Figure 2a). FGF1 sense ODNs had no significant effect on FGF1 synthesis. The addition to FGF1-depleted cells of FGF2 had no effect on cell apoptosis during the first 3 days, then PCD increased and was 2.8 times higher on day 5 and 2.3 times higher on day 7 than the basal level of PCD on day 1 of culture (Figure 2b). FGF1 sense ODNs had no effect on FGF2 survival activity. Thus, these data strongly suggests that FGF2 survival activity may be mediated by the synthesis of endogenous FGF1 by RPE cells.

Sustained FGF2-stimulated FGFR1 and ERK2 activation is dependent on FGF1 excretion and is necessary for lower levels of apoptosis

In PC12 cells, trophic activity of endogenous FGF1 is FGFR and ERK activation-dependent (Renaud et al., 1996). Thus, we assessed autophosphorylation of FGFR1 (the only FGF tyrosine-kinase receptor at the bovine RPE cell surface) after serum withdrawal in the absence and in the presence of FGF2. FGFR1 underwent weak but detectable tyrosine phosphorylation, whereas a single addition of FGF2 induced strong and surprisingly sustained phosphorylation over the 5-day period of culture, with nine times higher phosphorylation levels than were detected in the absence of the growth factor (Figure 3). The addition of a neutralizing anti-FGF1 antibody to FGF2-stimulated cultures reduced FGFR1 autophosphorylation to undetectable level (Figure 3b), indicating that the sustained FGFR1 activation detected after FGF2 stimulation was due to excreted FGF1. Stimulation of RPE cells by FGF2 had no significant effect on the production of FGFR1 during the 5-day culture period showing that the effects of excreted FGF1 on activation of FGFR1 were not due to the stimulation of FGFR1 production (Figure 3c).

The relationship between the FGF2-stimulated synthesis and excretion of FGF1, FGFR1 activation and cell survival, led us to investigate the activation of ERK2 as a potential downstream pathway. The state of ERK2 phosphorylation was investigated by Western blotting, using an antibody that specifically recognized active ERK1 and 2 (ERKs-P). Addition of a single dose of FGF2 rapidly induced ERK1 and 2 phosphorylation within 5 min (Figure 4a). Surprisingly, high levels of ERK1 and 2 activation were observed over the next 12 h. After serum withdrawal, only weak basal levels of phosphorylation were detected by reduced electrophoretic mobility and ERK2 activity over the 7-day culture period. In contrast, after FGF2 stimulation, there was sustained ERK2 phosphorylation and activity at levels five to 11 times higher than that in unstimulated cells (Figure 4b,c).

Then, we investigated the role of ERK2 on FGF1-mediated cell survival. Sustained inhibition of FGF2-induced ERK2 phosphorylation was obtained in the presence of a neutralizing anti-FGF1 antibody over the 7-day culture period (Figure 5a). Similarly, the addition of a blocking anti-FGFR1 antibody to FGF2-stimulated cells led to the complete inhibition of ERK2 activation, demonstrating that the sustained effects of excreted FGF1 on ERK2 were mediated by FGF1 binding to FGFR1 (Figure 5b). Neutralization of excreted FGF1 and inhibition of the binding of FGF1 to FGFR1 had no significant effect on the apoptosis of FGF2-stimulated cells during the first 3 days of culture (Figure 5e). In contrast, on days 5 and 7, RPE cell death increased by factors of 7.1 and 4.1 respectively. Finally, MEK1/2 inhibition by PD098059 (Figure 5c) and UO126 (Figure 5d) also abolished the sustained activation of ERK2 over the 7-day period of culture and resulted in a 6.2 to 7.5 and 4.1 to five times increase in cell death on days 5 and 7 respectively (Figure 5e), indicating that ERK2 activation was required for the survival of FGF2-stimulated RPE cells. Taken together, these data strongly suggested that the survival-promoting activity of FGF1 excreted by FGF2-stimulated RPE cells was mediated by FGFR1 binding and may involve ERK2 long term activation.

FGF2 inhibits RPE cell death through the stimulation of Bcl-x synthesis via an ERK2-dependent mechanism

Little is known about the mechanism and proteins involved in the survival activity mediated by FGF2. Many members of the Bcl-2 family, including Bcl-x are potent inhibitors of programmed cell death, inhibiting the activation of caspases in cells by a direct interaction (Thornberry and Lazebnick, 1998). Bcl-2 and Bcl-x production were analysed by Western blotting and Bcl-x de novo synthesis was monitored using radiolabeled amino acids on each of the 7 days of cultures. Addition of FGF2 had no significant effect on Bcl-2 production over the 7 days of cultures (data not shown). In contrast, exogenous FGF2 induced a sustained and a two to three times increase in Bcl-xl production (Figure 6a) and sustained de novo synthesis (Figure 6b). Immunoneutralization of endogenous FGF1 in FGF2-stimulated cells induced after 1 day of culture a large and sustained inhibition in Bcl-xl production and de novo synthesis over the 7 days of cultures (data not shown). The inhibition of ERK activation led also to the inhibition of FGF2-stimulated de novo synthesis and production of Bcl-xl (Figure 6c,d), indicating that the sustained de novo Bcl-xl synthesis stimulated by FGF2 was controlled by the ERK pathway. We tested whether Bcl-x was a key component in the control of FGF2-stimulated RPE cell survival, by depleting Bcl-x from FGF2-stimulated cells by an antisense ODNs approach. Treatment with Bcl-x antisense ODNs specifically inhibited Bcl-x synthesis over the 7-day period of culture (Figure 7a). The depletion of BcL-x from FGF2-stimulated RPE cells induced a dramatic increase in cell death, and FGF2 treatment could not overcome Bcl-x depletion (Figure 7b). Bcl-x antisense ODNs induced a 4.3 and 3.8 times increase in RPE cell death on days 5 and 7 respectively, relative to Bcl-x sense ODN-treated and FGF2-stimulated cells. Bcl-x sense ODNs treatment had no significant effect on FGF2 survival activity. Thus, after serum withdrawal and FGF2 treatment, stimulation of Bcl-x by ERK2 activation is a key part of the protective mechanism promoting RPE cell survival.

Discussion

Anti-apoptotic effect of FGF2 is mediated by endogenous and excreted FGF1

Adult RPE cells are terminally differentiated and are mostly post-mitotic by the end of the second week after birth. Very recently, RPE cell apoptosis was shown to be the suicide mechanism involved in human age-related macular degeneration and in in vivo and in vitro experimental models of RPE cell death (Hinton et al., 1998). We found that RPE cell apoptosis due to serum deprivation was mediated by the activation of a caspase 3-like pathway and not by a caspase 1-like pathway. These results confirm that caspase 1 is apparently not involved in the direct transmission of apoptotic signals, but is involved in cytokine activation. In contrast, caspase 3 is an executioner pro-apoptotic factor, the activation of which leads to apoptotic collapse and the demise of the cell. It has been shown that the production of p35, a baculovirus analog of the Inhibitor of Apoptosis (IAP) family, which inhibits CED3/caspase 3 activity, prevents blindness in Drosophilia mutants with retinal degeneration (Davidson and Steller, 1998). Thus, it would be of value to analyse the presence of IAP proteins and analogs in RPE cells during serum deprivation.

Previous studies have suggested that endogenous FGFs act as protection factors in vivo by showing a dramatic upregulation of FGF gene expression in response to cell death in various retinal degeneration models (Portera-Gailleau et al., 1994; Gao and Hollyfield, 1996). We have confirmed that this is the case in retina, by demonstrating that inhibition of the binding of endogenous FGFs to their receptors accelerates and increases retinal cell apoptosis (Guillonneau et al., 1998a). More recently, inhibition of FGF1 by antisense ODNs in embryonic chick retinal cells in vitro and in vivo has been shown to inhibit neuron survival (Désiré et al., 1998). In none of these studies, was information provided about the signaling and the proteins involved in the survival activity of FGFs. In this report, using FGF1 antisense ODNs and FGF1 immunoneutralization strategies we showed that endogenous FGF1 was a major component of the FGF2 survival activity. In addition, we demonstrated that endogenous FGF1 has to be excreted to mediate FGF2 anti-apoptotic activity, contrasting with the intracrine FGF1 survival activity in PC12 cells (Renaud et al., 1996). One of the striking findings reported in this study is that FGF2-stimulated FGF1 excretion induces sustained activation of FGFR1 and ERK2 over the 7-day period of culture. Furthermore, specific inhibition of the MED1/2 pathway led to an increase in apoptosis, suggesting that the survival signaling pathway of excreted FGF1 may be specifically controlled by ERK1/2 activation. This seems to be confirmed by previous studies showing that cell survival is correlated with the overexpression of ERK2 protein during the aging of RPE cells (Guillonneau et al., 1998b).

The stimulation of de novo BcL-x synthesis is a major pathway in FGF2 survival signaling

Intraocular injections of FGF2 inhibit experimental retinal degeneration (Faktorovich et al., 1990), demonstrating that exogenous FGF2 may act as a survival-promoting trophic factor in retina in vivo. The trophic/protective effects of exogenous FGF have been demonstrated in human phase II trials with FGF2 in stroke patients (Onal and Fisher, 1997), but little is known about the signaling and proteins involved in the survival activity of FGF2. Here, we have shown that FGF2 rescued RPE cells from caspase 3-dependent cell apoptosis. FGF2 stimulated FGF1 production and excretion, inducing a paracrine loop of FGFR1 and ERK2 activation. However, the FGF1 paracrine loop of activation was not the only signaling pathway involved in FGF2 survival activity. Members of the BcL-2 family of proteins are important regulators of apoptosis in many cellular systems. The first member of this family, BcL-2 was identified at the breakpoint site of a t (14;18) translocation present in many human B-cell lymphomas (Cleary et al., 1986). Increased production of BcL-2 as a result of t (14;18) translocation contributes to neoplastic B-cell expansion by preventing B-cell death. BcL-2 and its homologue, BcL-x inhibit apoptosis induced with a wide variety of stimuli by mechanism that remain unidentified. Despite distinct functions, BcL-2 and BcL-x have both been found to inhibit the release of cytochrome c, which is associated with the activation of caspase 3 from mitochondria. In transgenic mice expressing the Bcl-2 gene with light-damage, and in degenerating rd mice, there is a partial and temporary preservation of the retina, suggesting a role for anti-apoptotic members of the Bcl-2 protein family in the survival of retinal neurons (Chen et al., 1996). Thus, these two anti-apoptotic factors seems to be good candidates for mediating FGF2 survival activity in RPE cells. In serum-deprived RPE cells, FGF2 induced sustained de novo synthesis of BcL-x1. In these conditions, BcL-x may inhibit the caspase 3 activity we detected in RPE cells. This is consistent with very recent data showing that BcL-x is cleaved directly by caspase 3 (Clem et al., 1998; Fujita et al., 1998). In RPE cells, the depletion of BcL-x blocked the FGF2-mediated rescue of apoptotic death and caused a large increase in cell death despite the production of FGF1. This strongly indicates that a major survival pathway induced by FGF2 involved the de novo synthesis of BcL-x. The pivotal role of BcL-x was recently demonstrated in eosinophil cell cultures, in which granulocyte-macrophage colony-stimulating factor rescued spontaneous apoptosis by upregulating BcL-x1 protein levels, whereas no BcL-2 stimulation was observed (König et al., 1997). In the present study, the addition of FGF2 had no effect on BcL-2 production, indicating that in RPE cells, BcL-2 does not play a pivotal role in FGF2-mediated cell survival. This contrasts with recent data suggesting a link between the endogenous FGF1 signaling pathway and BcL-2 in neuronal survival modulation (Raguenez et al., 1999). In this model, the de novo synthesis of Bcl-x was not studied. Thus, we hypothesize that depending on the cell types or on the excretion of FGF1, the signaling involved in the survival activity of FGFs may implicate different pathways. Regulation of Bcl-2 expression or function by growth factors mediated survival signals is also controversial. In some systems, survival factors up-regulate the transcription of Bcl-2 (Deng et al., 1993). However, cells can die without down-regulation of Bcl-2 expression and growth factors can inhibit apoptosis in the absence of RNA or protein synthesis. RPE cells produced low levels of Bcl-2. This is consistent with previous data showing that bcl-2 is not normally expressed to high levels in the retina (Papermaster, 1997). The role of Bcl-2 in retina is unclear. Chen et al. (1996) showed that in transgenic mice, expression of bcl-2 under the control of an opsin promoter resulted in partial, temporary preservation of photoreceptors in three types of retinal degenerations, whereas other studies have shown no significant reduction in the rate or extent of apoptosis in photoreceptors (Papermaster, 1997). Furthermore, both increases and decreases in bcl-2 levels have been described after various injuries (Montpied et al., 1993; Gillardon et al., 1995). In contrast to bcl-2, bcl-x expression, predominantly the production of its anti-apoptotic long form, Bcl-xl, is very abundant in adult retina, consistent with the concept that bcl-xl is expressed in long-lived postmitotic cells whereas bcl-xs and bcl-2 are expressed predominantly in proliferating cells in adult animals.

Activation of ERK2 is a key step required for Bcl-x-dependent inhibition of the apoptosis mediated by FGF2

Our study showed that the FGF2-stimulated increase in RPE cell survival was directly dependent on ERK2 activation. This, also shown by Gardner and Johnson (1996) who found that FGF2-suppression of TNF-mediated apoptosis required the activation of ERK pathway. However, although we are beginning to understand the involvement of the ERK pathway in cell death, the biochemical and molecular changes underlying this phenomenon are largely unknown. This study shows for the first time that sustained activation of ERK2 over a 7-day period of culture was required to protect RPE cells from apoptosis. We have proposed a scheme to illustrate the manner in which FGF2 predominantly activates ERK2 and stimulates sustained FGF1 and Bcl-x production (Figure 8). The importance of the long-term activation of ERK2 for cell survival is consistent with studies showing that acute activation of the ERK cascade by growth factors potentiates proliferation whereas a chronic increase in ERK activity is cytoprotective (Carter et al., 1998). In addition, it was shown that overproduction of ERKs in human breast cancer was required for cell survival (Sivaraman et al., 1997) and that cell survival is correlated with the overproduction of ERK2 during aging of RPE cells (Guillonneau et al., 1998b). The importance of both activation and production of proteins of the ERK pathway for cell survival was recently confirmed (Mishima et al., 1998; Yazlovitskaya et al., 1999; Sahl et al., 1999). We never detected significant activation of the member of the MAP kinase group, JNK1 after FGF2 stimulation of RPE cells (our unpublished results). More recently, it has been demonstrated that both the PI3-K and ERK pathways necessary for the IGF survival effect on adipocytes are associated with the inhibition of Bcl-xs production (Navarro et al., 1998), confirming a role for the ERK pathway in the activation of members of the Bcl-2 family. It is believed that the balance between death agonists and antagonists from the Bcl-2 family may regulate apoptosis. Overproduction of Bcl-x inhibits caspase 3-induced PC12 cell apoptosis after serum deprivation (Lindenboim et al., 1998), suggesting that Bcl-x acts upstream from caspase activation. This is consistent with the sustained production of Bcl-x and the rescue from caspase 3-dependent apoptosis in FGF2 stimulated RPE cells. Little is known about bcl-x expression in the retina during cell apoptosis and the results that have been obtained are conflicting. Bcl-xl expression decreases shortly after retinal ganglion cell apoptosis after axotomy (Levin et al., 1997) whereas, Bcl-x production increases and then decreases in ganglion cells after optic nerve lesion (Iseman et al., 1997). The exact mechanism of the regulation of Bcl-x expression is still unknown. It has been shown that the reduction in apoptosis of keratinocytes induced by EGF activation is mediated by Bcl-xl (Stoll et al., 1998). In this study, ERK activation was not examined. In contrast, in the present report, we clearly demonstrated the pivotal role of Bcl-x in the survival activity of FGF2 mediated by ERK2, confirming the relationship between the ERK pathway and Bcl-x which has been suggested in hematopoietic cells (Garland and Rudin 1998). Recently, it has been shown that the expression of the anti-apoptotic MCL1 gene product is regulated by the ERK pathway (Townsend et al., 1998). Thus, it would be of value to analyse the production of MCL1 protein in RPE cells during serum deprivation. In conclusion, our results provide insight into the apoptotic signaling implicated in RPE cell death and the molecular changes that may reduce apoptosis in FGF2-stimulated RPE cells. In addition, as cell apoptosis can be regulated by FGF1 and 2, a better understanding of the molecular mechanism and signaling induced by these growth factors has important implications for anti-degenerative and cancer therapy.

Materials and methods

Cell culture and treatment of cells

Bovine RPE cells were isolated as previously described (Guillonneau et al., 1997). Primary RPE cell cultures were grown in Dulbecco's modified essential medium (DMEM, GIBCO/BRL, New York, USA) containing 10% fetal calf serum (FCS, GIBCO/BRL, New York, USA), 2.5 g/ml fungizone, 50 g/ml gentamycin, 2 mM L-glutamine. Human recombinant FGF2 (18 kDa form) (Carlo Erba) was added to confluent, quiescent cells at a final concentration of 20 ng/ml and cells were incubated for a further 7 days. RPE cells were protected against apoptosis by adding two caspase inhibitors: asp-glu-val-asp-aldehyde (DEVD-CHO) and tyr-val-ala-asp-aldehyde (YVAD-CHO) (Calbiochem), to a final concentration of 10 M, every 48 h for 7 days. For each compound, we tested the inhibitory or stimulating activity and we checked that the vehicle alone was not cytotoxic. In some experiments, anti-FGF1 neutralizing polyclonal antibody (R & D Systems) and blocking anti-FGFR monoclonal antibody (clone VBS1, Chemicon International) were added on days 1 and 3 to final concentrations of 200 g/ml (anti-FGF1) and 20 g/ml (anti-FGFR1) of culture medium. The specificity of the anti-FGF1 and anti-FGFR1 has been previously verified (Guillonneau et al., 1998b). The MEK inhibitors, PD098059 (Calbiochem, Meudon, France) and UO126 (Promega, Charbonniere, France), were added on days 1 and 3 to final concentrations of 10- and 4 M respectively. There were no non-specific cytotoxic effects of the MEK inhibitors, as previously reported (Guillonneau et al., 1998b).

For labeling experiments, RPE cells were incubated in serum-free DMEM for 7 days. Each day, 200 Ci/ml of a mixture of [³⁵S]cysteine and [³⁵S]methionine (Amersham, SA: 37 TBq Ci/mole) was added in complete DMEM and incubated with the cells for 24 h as previously described (Guillonneau et al., 1998b). Each day, cells were lysed in Triton X-100 lysis buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 50 mM NaF, 5 mM EDTA, 40 mM -glycerophos-phate, 0.2 mM sodium orthovanadate, 1 g/ml leupeptin, 1 M pepstatin A and 1% Triton X-100). Cell lysates (equal amount of protein) were then immunoprecipitated overnight at 4°C with 0.1 g of polyclonal anti-ERK2 antibody, and with 1 g of polyclonal anti-FGF1 antibodies (Sigma, Saint Quentin Fallavier, France). Protein G-Sepharose beads (50 l v/v) were added and incubated with the mixture for 1 h at 4°C. Immune complexes were collected by centrifugation at 12 000 g. The immunoprecipitated proteins were separated by SDS - PAGE (12% polyacrylamide gel for ERK2 and 15% for BcL-x and FGF1). The protein bands detected by autoradiography were quantified using an LKB Ultrascan XL laser densitometer (Pharmacia).

Oligonucleotides and oligonucleotide treatment of cells

Phosphodiester sense and antisense ODNs directed against FGF1 were designed based on the published sequence of the bovine FGF1 gene (Alterio et al., 1998). The 16-mer ODN was targeted against the first donor-acceptor site of FGF1 at codon 57 of bFGF1. The antisense ODN to bFGF1 was 5'-GCTGAATGTGCTGGTC-3' (referred to as AS-F1) and the corresponding sense ODN was 5'-GACCAGCACATTCAGC-3' (referred to as S-F1). It has been demonstrated that the addition of AS-F1 to adenocarcinoma HSY cells, inhibits endogenous FGF1-dependent cell proliferation (Myoken et al., 1996). Two phosphodiester sense and antisense ODNs directed against Bcl-x were designed based on the published sequence of the human Bcl-x (hBcl-x) gene (Boise et al., 1993). The 16-mer ODN was based directed against the start colon (ATG) plus the 13 bases immediately downstream in the hBcl-x as previously published by Amarante-Mendes et al. (1998). The antisense ODN was 5'-CCGGTTGCTCTGAGACAT-3' (referred to as AS-Bcl-x) and the corresponding sense ODN was 5'-ATGTCTCAGCAACCCGG-3' (referred to as S-Bcl-x). 3' exonucleases are thought to be more active than 5' exonucleases, so an amine group was added at the 3'-end, this modification having been previously shown to increase the stability of classical phosphodiester antisense ODNs to levels similar to those for phosphothioate ODNs and the stability of the ODNs-target complex. In addition, these 3'-modified ODNs are not toxic and do not compete with FGF1 and 2 for their binding sites (heparin-like activity), in contrast to phosphorothioate ODNs (Fennwald and Rando, 1995; Gukova et al., 1995). We used lipofectin (GIBCO/BRL, Cergy Pontoise, France) as a cationic lipid to deliver the ODNs, because this method results in high levels of uptake and stability of phosphodiester ODNs in the intracellular compartment without affecting their final nuclear location (Clarenc et al., 1993; Lezoulac'h et al., 1995) after endocytosis and release from the endocytic compartment. All ODNs were synthesized commercially (Eurogentec, Seraing, Belgium) and purified by HPLC. Lipofectin-ODN complexes were produced according to the manufacturer's instructions. Preliminary experiments were performed with 2.5, 5, 7, 10 and 20 g/ml of lipofectin and various concentrations of ODNs (between 2 and 40 M). All results presented here were obtained with 15 M ODNs for FGF1 and 10 M ODNs for Bcl-x, and with respectively 5 g/ml and 7 g/ml of lipofectin for cells incubated in the absence and presence of serum. Cells were incubated for 3 days in the presence of 10% heat-treated calf serum. Then, the typically 70 - 80% confluent RPE cells were treated with ODNs/lipofectin for 24 h, washed twice with DMEM serum-free medium and fresh serum-free medium was added and the cells cultured for 7 days. On day 2 of serum-free culture, the appropriate concentration of ODN was added in the presence of lipofectin. For 3 days beginning on day 4 of incubation, cells were treated with the appropriate concentration of ODN without lipofectin.

Diffusion of ODNs in RPE cells was studied with AS-F1, S-F1, AS-Bcl-x and S-Bcl-x coupled to cyanin 3 groups at the 5'-terminus. These fluorescent ODNs were observed with an Aristoplan microscope (Leica, Rueil-Malmaison, France) and by confocal microscopy with a Bio-Rad microscope (Ivry sur Seine, France). Cellular uptake of ODNs was observed as early as 5 min after ODN/lipofectin treatment and 99% of the cells were labeled after 6 h of exposure to ODN. The nuclear and perinuclear location of ODN was checked by confocal microscopy.

Cell proliferation and cell death assays

The status of proliferating and quiescent culture cells was determined during the 7 days of culture as previously described (Guillonneau et al., 1998b). The proliferation of RPE cells was assayed daily by counting the number of cells and by determining [³H]thymidine (Amersham, SA:0.92 TBq/mmole) incorporation. The number of dead cells was determined by counting the cells remaining on the culture dish after staining with Trypan blue and MTT (3(4,5-dimethylthiazol-,yl)2,5 diphenyltetrazolium bromide). Cells undergoing programmed cell death (PCD) were detected daily and counted by two methods: (1) terminal d'UTP nick end labeling (TUNEL), (PCD kit, Boehringer) and (2) by staining of nuclei with Hoechst 33258 (0.5 g/ml) after fixation of the cells with 4% paraformaldehyde and permeabilization with ice-cold ethanol. Labeled, fragmented and condensed cells were scored as apoptotic and were counted in three different fields, in three culture wells at each time point. The positive cells were counted within an ocular grid using a 25´ objective. The grid was placed on the culture well and a minimum of 200 cells were counted per field.

ERK2 activity assay

Quiescent confluent RPE cells in 6-well plates were incubated in serum-free medium for 7 days. Each day, cells were lysed in Triton X-100 lysis buffer. Cell lysates (equal amount of protein) were then immunoprecipitated overnight at 4°C with 0.1 g of polyclonal antibody against ERK2 (Santa Cruz, USA). Protein G-Sepharose beads (50 l v/v) were added for 1 h at 4°C. Immune complexes were collected by centrifugation as 12 000´g and washed three times in lysis buffer and once in kinase buffer (20 mM HEPES, pH 7.4, 10 mM MgCl₂, 1 mM dithiothreitol and 10 mM p-nitrophenylphosphate). They were then suspended in 40 l kinase buffer containing 10 g of myelin basic protein (MBP), 50 M unlabelled ATP and 3 Ci of [-³²P]ATP (Amersham, 5000 c.p.m./pmol) per sample. The mixture was incubated for 10 min at 30°C and the reaction was stopped by adding 40 l of 2´Laemmli's sample buffer. Samples were subjected to SDS - PAGE on a 12% polyacrylamide gel. We checked that equal amounts of protein were loaded in each lane, by systematically staining the gels with Coomassie brilliant blue so that the protein bands were visible and could be scanned by densitometry. In addition, blots were reprobed with anti-actin antibody to check that the 42 kD band was similarly intense in each case. The specificity and lack of cross-reaction of the antibodies with specific control peptides were tested.

Western blot analysis

RPE cells were incubated in serum-free medium for 7 days. They were washed twice in PBS and lysed in ice-cold Triton X-100 lysis buffer and centrifuged at 4°C for 10 min at 10 000 g. Monoclonal antibody directed against -actin was used as an internal standard for checking protein loading. For ERK2 and FGFR1 analysis, Triton X-100 cell lysate (30 and 100 l respectively), was mixed with 5´ Laemmli's buffer and heated for 5 min at 95°C. The soluble proteins of the cell lysates were separated by SDS - PAGE (12 and 7% polyacrylamide gel for ERK2 and FGFR1 analysis respectively), transferred by electroblotting onto nitrocellulose filters and probed with polyclonal antibodies raised against p42 ERK2 and FGFR1 (Santa Cruz, USA). The primary antibodies were detected using a horseradish peroxidase-conjugated goat anti-rabbit secondary antibody. ECL-chemiluminescence substrates were used to detect positive bands, according to the manufacturer's instructions, and the membrane was placed against Hyperfilm TM ECL (Amersham, France). The protein bands detected on the autoluminogram were quantified using an LKB Ultrascan XL laser densitometer (Pharmacia).

Statistics

Each figure shows the results from experiments repeated at least three times. All data are expressed as the average±s.e.m. Statistical comparisons were made using the two-tailed Student's t-test (Gaussian population with equal s.d.) and the Wilcoxon or Mann and Whitney test (non-parametric tests).

Note added in proof

While this paper was being written, another study addressing the stimulation of Bcl-2 expression through ERK2 activation during cell survival was published (Liu et al., 1999).

Acknowledgements

The authors wish to thank Dr JC Jeanny and Dr L Désiré for helpful suggestions. The excellent technical assistance of F Regnier-Ricard is gratefully acknowledged. This work was supported by the Association pour la Recherche sur le Cancer (contract numbers 1369 and 9697, F Mascarelli), the Ligue Nationale contre le Cancer (Comité de Paris, M Bryckaer), the Ministère de la Recherche et de l'Enseignement Supérieur (X Guillonneau) and the Universite René Descartes, Faculté de Médecine Paris-Ouest (F Mascarelli).

Abraham JA, Whang JL, Tumolo A, Mergia A, Friedman J, Gospodarowicz D and Fiddes JC. (1986). EMBO J. 5, 2523. MEDLINE

Alterio J, Halley C, Brou C, Soussi, Courtois Y and Laurent M. (1988). FEBS Lett. 242, 41-46. MEDLINE

Amarente-Mendes G, McGahon A, Nishioka W, Afar D, Witte O and Green D. (1998). Oncogene 16, 1383-1390. MEDLINE

Boise L, Gonzales-Garcia M, Postela C, Ding L, Lindsten T, Turka L, Mao X, Nunez G and Thompson C. (1993). Cell 74, 597-608. MEDLINE

Burgess WH and Maciag T. (1989). Ann. Rev. Biochem. 58, 575-606. MEDLINE

Burgess WH, Dionne CA, Kaplow J, Mudd R, Friesel R, Zilberstein A, Schlessinger J and Jaye M. (1990). Mol. Cell. Biol. 10, 4770-4777. MEDLINE

Carter S, Auer K, Reardon D, Birrer M, Fisher P, Valerie K, Schmidt-Ullrich R, Mikkelsen R and Dent P. (1998). Oncogene 16, 2787-2796. MEDLINE

Chen J, Flannery JG, LaVail MM and Steinbert RH. (1996). Proc. Natl. Acad. Sci. 93, 7042-7047. MEDLINE

Clarenc J, Lebleu B and Leonetti J. (1993). J. Biol. Chem. 268, 5600-5604. MEDLINE

Clem R, Cheng E, Karp C, Kirsch D, Ueno K, Takahashi A, Kastan M, Griffin D, Earnshaw W, Veliuona M and Hard-wick M. (1998). Proc. Natl. Acad. Sci. USA 95, 554-559. Article MEDLINE

Cleary ML, Smith SD and Sklar J. (1986). Cell 47, 19-28. MEDLINE

Cobb MH, Bourton TG and Robbins DJ. (1991). Cell Regul. 2, 965-978. MEDLINE

Courlier F, Pontarotti P, Roubin G, Goldfarb M and Birnbaum D. (1997). J. Mol. Evol. 44, 43-56. MEDLINE

Davidson F and Steller H. (1998). Nature 391, 587-590. Article MEDLINE

Deng G and Podack ER. (1993). Proc. Natl. Acad. Sci. USA 90, 2189-2193. MEDLINE

Désiré L, Head M, Fayen N, Courtois Y and Jeanny JC. (1998). Dev. Dynamics 212, 63-74.

Enzmann V, Faude F, Kohen L and Wiedmann P. (1998). Ophthalmic Res. 30, 189-194. MEDLINE

Faktorovich E, Steinberg R, Yasumara D and La Vail M. (1990). Nature 347, 83-86. MEDLINE

Fennwald S and Rando R. (1995). J. Biol. Chem. 270, 21718-21721. MEDLINE

Fujita N, Nagashi A, Nagashima K, Rokudai S and Tsuruo T. (1998). Oncogene 17, 1295-1304. MEDLINE

Funato N, Moriyama K, Shimokawa H and Kuroda T. (1997). Biochem. Biophys. Res. Commun. 240, 21-26. Article MEDLINE

Gao H and Hollyfield JG. (1996). Exp. Eye. Res. 62, 181-189. MEDLINE

Gardner AM and Johnson GL. (1996). J. Biol. Chem. 14560-14571.

Garland J and Rudin C. (1998). Blood 4, 1235-1246.

Gillardon F, Barrle J, Wickert H and Zimmerman M. (1995). J. Neurosci. Res. 5, 708-715.

Givol D and Yayon A. (1992). FASEB J. 6, 3362-3369. MEDLINE

Guillonneau X, Regnier-Ricard F, Dupuis C, Courtois Y and Mascarelli F. (1997). Exp. Cell Res. 233, 198-206. MEDLINE

Guillonneau X, Laplace O, Regnier-Ricard F, Jonet L, Brickaert MC, Courtois Y and Mascarelli F. (1998a). Mol. Biol. Cell 9, 2785-2802.

Guillonneau X, Brickaert MC, Launay-Longo C, Courtois Y and Mascarelli F. (1998b). J. Biol. Chem. 273, 22367-22373.

Gukova M, Yakubov L, Vlodavsky I, Tonkinson J and Stein C. (1995). J. Biol. Chem. 270, 2620-2627. MEDLINE

Hinton D, He S and Lopez F. (1998). Arch. Ophthalmol. 116, 203-209. MEDLINE

Iseman S, Wahl C, Krajjewski S, Reed J and Bahr M. (1997). Eur. J. Neurosci. 9, 1763-1772. MEDLINE

Jaye M, Burgess R, Ricca W, Chiu GA, Ravera IM, O'Brien MW, Modi SJ, Maciag T and Drohan WN. (1986). Science 233, 241-244.

Jaye M, Schlessinger J and Dionne CA. (1992). Biochem. Biophys. Acta 1135, 185-199.

König A, Menzel T, Lynen S, Wrazel L, Rosén A, Al-Katib A, Raveche E and Gabrilove JL. (1997). Leukemia 11, 258-265. MEDLINE

Levin L, Sclamp C, Spieldoch R, Gesvain K and Nickells R. (1997). Invest. Ophthalmol. Vis. Sci. 38, 2545-2553. MEDLINE

Lezoualc'h F, Seugnet I, Monnier AL, Ghysdael J, Behr J and Demeinex B. (1995)). J. Biol Chem. 270, 12100-12108. MEDLINE

Lindenboim L, Haviv R and Stein R. (1998). Neurosci. Lett. 253, 37-40. MEDLINE

Liu J, Issad T, Chevet E, Ledoux D, Courty J, Caruelle J, Barritault D, Crepin M and Bertin B. (1998). Eur. J. Biochem. 258, 271-276. MEDLINE

Liu YZ, Boxer L.M and Latchman D.S. (1999). Nucleic Acid. Res. 27, 2086-2090. MEDLINE

Malecaze F, Mascarelli F, Bugra K, Fuhrmann G, Courtois Y and Hicks D. (1993). J. Cell Physiol. 8, 631-642.

Mishima K, Yamada E, Masui K, Shimokawara T, Sugimura M and Ichijima K. (1998). Mod. Pathol. 11, 886-891. MEDLINE

Montpied R, Weller M and Paul S. (1993). Biochem. Biophys. Res. Commun. 195, 623-629. MEDLINE

Myoken Y, Myoken Y, Okamoto T, Kan M, McKeehan W, Sato D and Takada K. (1996). Int. J. Cancer 65, 650-657. MEDLINE

Navarro P, Valverde A, Benito M and Lorenzo M. (1998). Exp. Cell Res. 243, 213-221. Article MEDLINE

Onal M and Fisher M. (1997). Eur. Neurol. 38, 141-154. MEDLINE

Papermaster D. (1997). Cell Death Different. 4, 21-28.

Partanen J, Vainakka S, Korhonen J, Armstrong E and Alitalo K. (1992). Prog. Growth Factor Res. 4, 69-83. MEDLINE

Portera-Gailleau C, Sung C, Nathans J and Adler R. (1994). Proc. Natl. Acad. Sci. USA 91, 974-978. MEDLINE

Raguenez G, Désiré L, Lantrua V and Courtois Y. (1999). Biochem. Biophys. Res. Commun. 258, 745-751. MEDLINE

Renaud F, Desset S, Oliver L, Gimenez-Galleco G, Van Obberghen E, Courtois Y and Laurent M. (1996). J. Biol. Chem. 271, 2801-2811. MEDLINE

Rifkin D and Moscatelli D. (1989). J. Cell Biol. 109, 1-6. MEDLINE

Sahl B, Marotta A, Matthewson C, Ahluwalia M, Flint J, Owen D and Pelech S. (1999). Anticancer Res. 19, 731-740. MEDLINE

Sato Y and Rifkin DB. (1988). J. Cell Biol. 107, 1129-1205.

Shin JT, Barbeito L, MacMillan-Crow LA, Beckman JS and Thompson JA. (1996). Arch. Biochem. Biophys. 335, 32-41. MEDLINE

Sivaraman VS, Wang H, Nuovo GJ and Malbon CC. (1997). J. Clin. Invest. 99, 1478-1483. MEDLINE

Spear N, Estevez AG, Johnson GV, Bredesen DE, Thompson JA and Beckman JS. (1998). Arch. Biochem. Biophys. 356, 41-45. MEDLINE

Stoll S, Benedict M, Mitra R, Hiniker A, Elder J and Nunez G. (1998). Oncogene 16, 1493-1499. MEDLINE

Thornberry N and Lazebnick Y. (1998). Sciences 281, 1312-1316.

Townsend KJ, Trusty JL, Traupman MA, Eastman A and Craig RW. (1998). Oncogene 17, 1223-1234. MEDLINE

Vlodavski I, Arias Y, Atzman R and Fuks Z. (1982). Exp. Cell. Res. 140, 149-159. MEDLINE

Wang Q, Maloof P, Wang H, Faenig E, Stein D, Nichols G, Denny TN, Yahalom J and Wieder R. (1998). Exp. Cell. Res. 238, 177-187. Article MEDLINE

Yazlovitskaya EM, Pelling JC and Person DI. (1999). Mol. Carcinog. 25, 14-20. MEDLINE

Yokoiyama Y, Ozawa S, Seyama Y, Namiki Y, Kaji K, Shirama K, Shioda M and Kano K. (1997). J. Neurochem. 5, 2212-2215.

Figure 1 Effects of caspase inhibitors and FGF2 on programmed cell death in serum-depleted, long-term cultures of RPE cells. RPE cells were cultured until confluence, then after 3 days of culture (day 0) the medium was removed, and cells were washed four times with PBS and cultured in serum-free conditions for 7 days in the absence or presence of caspase 1 and 3 inhibitors (a) as described in Materials and methods. (b) At day 0, half the cultures received a single addition of FGF2 and the other half remained in serum-free and FGF2-free medium. The number of dead cells was determined by counting the cells remaining on the culture dish after staining with Trypan blue and MTT. Cells undergoing programmed cell death (PCD) were detected daily and counted by two methods: (1) the terminal d'UTP nick end labeling (TUNEL) technique, (PCD kit, Boehringer) and (2) by nuclei staining (blue) with Hoechst 33258. Values are means of four experiments and differences between means were analysed by the Mann and Whitney test, **P<0.01, ***P<0.005

Figure 2 Effect of FGF1 depletion in FGF2-stimulated long-term cultures of RPE cells. RPE cells were cultured in serum-free conditions for 7 days as described in Figure 1. (a) For detection of FGF1 by Western blotting, lysate (150 g of total protein) from FGF2, FGF2 plus S-F1 and FGF2 plus AS-F1 treated cells was subjected to SDS - PAGE, transferred to nitrocellulose and FGF1 protein was detected with a specific polyclonal FGF1 antibody. (b) Cells were treated with FGF1 sense and antisense ODNs using lipofectin and were stimulated with FGF2 as described in Materials and methods. PCD was detected daily and quantitated by TUNEL. The data are representative of four independent experiments which gave similar results. Values are means±s.d. and differences between means were analysed by the Mann and Whitney test, **P<0.01

Figure 3 Effects of FGF2 on FGFR1 autophosphorylation and production in serum-depleted long-term cultures of RPE cells. Cells were cultured in the presence and absence of FGF2 for 5 days as described in Figure 1. (a,b) Each day, cells were incubated with 200 Ci/ml of Na₃ ³²PO₄ to study FGFR1 autophosphorylation. Neutralizing anti-FGF1 antibody was added on days 1 and 3. Equal volumes of cell lysates containing 200 g protein were incubated with protein A-Sepharose that had previously been treated with anti-phosphotyrosine antibody. Phosphotyrosine-containing protein was eluted with 0.1% Triton X-100 buffer and anti-FGFR1 antibody coupled to protein A-Sepharose was added. The adsorbed material was analysed by SDS - PAGE and autoradiography. The exposure time for autoradiographs was 4 days. (c) Cells were lysed on days 1, 3 and 5 and equal volumes of supernatant containing 200 g protein were analysed by Western blotting with anti-FGFR1 antibody. These experiments were repeated three times with similar results obtained each time

Figure 4 Effects of exogenous FGF2 on ERK2 phosphorylation and activity. Cell cultures were incubated in serum-free conditions for 7 days as described in the legend to Figure 1. Cells were lysed and equal amounts of protein were reduced and subjected to SDS - PAGE and Western blotting performed using an anti-active ERK1/2 antibody (a) and an anti-ERK2 antibody (b) as described in Materials and methods. (c) Cell lysates were immunoprecipitated with the anti-ERK2 antibody and ERK2 activity was measured. Similar results were obtained in three independent experiments

Figure 5 Effects of the inhibition of FGF1 and MEK1 activities on ERK2 phosphorylation and programmed cell death in FGF2-stimulated cultures. RPE cells were incubated in serum-free medium for 7 days in the presence or in the absence of neutralizing anti-FGF1 (a) and blocking anti-FGFR1 (b) antibodies, and MEK1 inhibitors, PD098059 (c) and UO126 (d). Phosphorylation of ERK2 (a - d) was analysed as described in the legend to Figure 4. (e) PCD was studied on days 1, 3, 5 and 7 as described in Materials and methods. Similar results were obtained in four independent experiments. Values are means±s.d. and differences between means were analysed by the Mann and Whitney test, ***P<0.005

Figure 6 Effects of FGF2 on de novo BcL-x synthesis and production. Effects of FGF1 neutralization and MEK1 inhibition. Cells were incubated in serum-free conditions for 7 days in the presence or absence of FGF2. We analysed production of Bcl-x (a,c), de novo Bcl-x synthesis (b,d) were analysed on days 1, 3, 5 and 7 by culturing serum-depleted RPE cells in the presence or absence of MEK1 inhibitor (10 M) added on days 1 and 3 of culture. Bcl-x production was analysed by Western blotting and de novo Bcl-x synthesis was performed by daily adding 150 Ci/ml of a mixture of [³⁵S]cysteine and [³⁵S]methionine in complete DMEM for 24 h. Samples, each containing 200 g protein, were prepared daily with RIPA buffer and Bcl-x was immunoprecipitated with the anti-Bcl-x antibody. Autoradiographs were exposed for 7 days. The data are representative of four independent experiments which gave similar results

Figure 7 Effect of Bcl-x depletion in FGF2-stimulated long-term cultures of RPE cells. RPE cells were cultured in serum-free conditions for 7 days as described in Figure 1. (a) For detection of Bcl-x by Western blotting, lysate (150 g of total protein) isolated daily from FGF2, FGF2 plus S-Bcl-x and FGF2 plus AS-Bcl-x treated cells was subjected to electrophoresis in 18% SDS-polyacrylamide gels, transferred to nitrocellulose and Bcl-x protein was detected with a specific polyclonal Bcl-x antibody. (b) Cells were treated with Bcl-x sense and antisense ODNs using lipofectin and were stimulated with FGF2 as described in Materials and methods. PCD was detected daily and quantitated by TUNEL. The data are representative of three independent experiments which gave similar results. Values are means±s.d. and differences between means were analysed by the Mann and Whitney test, **P<0.01

Figure 8 Proposed scheme for activation of ERK2 and stimulation of BcL-x in response to FGF2 during RPE cell survival. FGF2 activates ERK2 via FGFR1 and stimulates sustained FGF1 production and excretion. Paracrine loop of excreted FGF1 also activates FGFR1 and ERK2, induces sustained BcL-x de novo synthesis and production, and rescues RPE cells from caspase 3-dependent apoptosis. Neutralization of excreted FGF1 leads to inactivation of ERK2 and inhibition of BcL-x production and apoptosis in FGF2-stimulated RPE cells. Inhibition of MEK by PD098059 and UO126 also leads to ERK2 inactivation, inhibition of BcL-x production and RPE cell apoptosis