The Mitogenic Signaling Pathway of Fibroblast Growth Factor Is Not ...

THEJOURNAL OF BIOLOGICAL CHEMISTRY Q 1986 by The American Society of Biological Chemists, Inc.

Vol. 261, No. 36, Issue of December 25, pp. 16916-16922.1986 Printed in U.S.A.

The Mitogenic Signaling Pathwayof Fibroblast Growth FactorIs Not Mediated through Polyphosphoinositide Hydrolysisand ProteinKinase C Activation in Hamster Fibroblasts* (Received for publication, August 6, 1986)

Isabelle Magnaldo, Gilles L’Allemain, Jean Claude Chambard, Michel MoennerS,Denis BarritaultS, and Jacques Pouyssegur From the Centre de Bwchimie-Centre National de la Recherche Scientifique, Universite de Nice, Parc Valrose, 06034 Nice, France and SUniuersite Paris XII, Laboratoire de Bwtechnologie des Cellules Eucaryotes, Avenue General de Gaulle, 94010, Creteil, France

Basic or acidic fibroblast growth factor (FGF), alone, was found to be as potent as a-thrombin to reinitiate DNA synthesis in Go-arrested Chinese hamster lung fibroblasts (CCL39). Basic FGF at 50 ng/ml or thrombin at 1 unit/ml rapidly initiated early eventssuch as cytoplasmic alkalinization (0.2-0.3 pH units), rise in cytoplasmic Ca2+, phosphorylation of ribosomal protein S6 and increasedc-myc expression, followed by a 30-40-fold increase in labeled nuclei. Whereas thrombin is a potent activator of phospholipase C as judged by the rapid releaseof inositol trisphosphate, inositol bisphosphate and by the massive accumulation of total inositol phosphate (IP) in the presence of 20 mM Li+, FGF failed to induce the breakdown of polyphosphoinositides in quiescent CCL39 cells. Indeed, no inositol trisphosphate nor inositol bisphosphate could be detected in response to FGF; in presence of Li+ the total IP release neverexceeded 8% of the IP released by the action of thrombin. Two additional findings indicated that FGF and thrombin activate different signaling pathways. First, we found that, in contrast to thrombin, the FGF-induced rise in thecytoplasmic free Ca2+ concentration measured by quin-2 fluorescence, is strictly dependent upon the presence of Ca2+ in the external medium. Second, we found that FGF failed to activate protein kinase Cas judged by the epidermal growth factor-receptor binding assay. Treatment of the cells with either thrombin or phorbol esters, rapidly inhibited ‘261-labeled epidermalgrowthfactor binding (50-60%). Basic or acidic FGF had no effect. We conclude that: 1) the FGF-receptor signaling pathway is not coupled to phospholipase C activation, and 2)early mitogenic events and reinitiation of DNA synthesis can be initiated independently of inositol lipid breakdown and protein kinase C activation.

cellular signals rapidly induce the breakdown of phosphatidylinositol 4,5-bisphosphate into 1,2-diacylglycerol and inositol 1,4,5-trisphosphate (IP3).lVery strong evidence supports the contention that these two inositol lipid intermediates serve as second messengers, one by activating protein kinase C (2-4), the second by inducing a transient rise in [Ca2+Ii(1,

5). Although little is known about the mechanisms of activation and the intermediate steps required to set in motion the progression of Go-arrested cells into S phase, it is of great interest that many growth factors such asPDGF (1, 6), bombesin (7, 8), bradykinin, vasopressin (9), and a-thrombin (10-12) are potent activators of polyphosphoinositide breakdown.However, it remains to be established whether the activation of this signaling pathway is sufficient or even required to trigger the full mitogenic response. Indeed, it is clearly emerging that the reinitiation of DNA synthesis can be accomplished via mechanisms independent of inositol lipid breakdown. This is exemplified with EGF or the more potent mitogenic combination, EGF/insulin (13, 14). In thispaper, we first show that bovine brain FGF, purified to homogeneity (15, 16), is, alone, as potent as a-thrombin to reinitiate DNA synthesis inGo-arrested Chinese hamster lung fibroblasts (CCL39 cell line). Second, we demonstrate, that in contrast to a-thrombin, FGF: 1) fails to induce detectable levels of IP, and IPp,2) fails to stimulate protein kinase C as judged by an EGF-receptor binding assay (8, 17-19), and 3) fails to induce a rise in [Ca2+Iiin a Ca2+-freemedium. These findings strongly indicate that themitogenic action of FGF is not mediated through the activation of phospholipase C. This conclusion reinforces the notion that additional growth-signaling pathways must exist but also questions the role played by inositol lipid breakdown in growth control. EXPERIMENTALPROCEDURES

Receptor-mediated hydrolysis of polyphosphoinositides is now a widely accepted transmembrane signaling pathway for a variety of externalstimuli including neurotransmitters, hormones, growth factors, antigens, etc. (1, 2). These extra-

* These studieswere supported by grants from the Centre National de la Recherche Scientifique (LP 7300, ATP 136, and ASP 394) the Institut National de la Santk et de la Recherche Mkdicale (CRE 842015), the Fondation pour la Recherche Mbdicale, the Association pour la Recherche contre le Cancer, and the Ligue Nationale FranGaise contre le Cancer. 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 solelyto indicate this fact.

CeU Culture and DNA Synthesis-The Chinese hamster lung fibroblast line CCL39 (American Type Culture Collection) was cultivated in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) containing 5% fetal calf serum and antibiotics as previously described (20). For ‘The abbreviations used are: IPI, inositol trisphosphate; FGF, fibroblast growth factor; EGF, epidermal growth factor; PDGF, platelet-derived growth factor; PMA, phorbol12-myristate 13-acetate; pHi, cytoplasmic pH; [Ca2+]i,cytoplasmic free Ca2+concentration; quinEGTA, [ethylenebis(oxy2/AM, quin-2-tetra-(acetoxymethyl)ester; ethylenenitri1o)ltetraacetic acid; IP, inositol phosphate; IPI, inositol monophosphate; IP2, inositol bisphosphate; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; Mops, 3-(N-morpholino)propanesulfonic acid; DMEM, Dulbecco’s modified Eagle’s medium; HBS, Hepes-buffered saline; BEL cells, bovine epithelial lens cells; SDS, sodium dodecyl sulfate.

16916

Signaling Mitogenic measurements of DNA synthesis, cells grown to confluence in 24well culture plates orin 35-mm dishes were arrested in GOby a 24-h incubation in serum-free medium (1:l ratio of DMEM Ham's F-12). These conditions of Goarrest by serum starvation have been used for all experimenb described in thispaper. Go-arrestedcell cultures were initiated by incubation for 24 h with the mitogen to be tested along with [3H]thymidine (4.5 PM a t 1 pCi/ml). Thymidine incorporated into trichloroacetic acid-precipitable material after the 24-h stimulation period was assayed by liquid scintillation spectrometry. For determination of labeled nuclei, conditions were the same except that [3HJthymidinewas used at 1.5 pM, 5 pCi/ml; cell monolayers were rinsed, fixed with trichloroacetic acid, and processed for autoradiography as described (21). Three independent fields containing between 300 and 400 cells were usually counted for each reported value. Measurement of IntraceUular pH-Confluent cultures in 35-mm dishes were arrested in Go and incubated for 15 min at 37 "C with a Hepes-buffered saline solution (HBS) consisting of 130 mM NaCl, 5 mM KCl, 2 mM CaC12,1mM M&12, 5 mM glucose, and 20mM Hepes/ NaOH, pH = 7.4. This solution was changed with the same medium pre-equilibrated at 37 "C and containing 1 pCi/ml ["Clbenzoic acid and either a-thrombin,FGF, or bovine serum albumin. Intracellular pH was calculated from the equilibrium distribution of this weak acid, as previously described (22). Phosphorylation of Ribosomal Protein S6-Cells grown to confluence in 12-well plates and arrested in Go were first equilibrated for 30 min in HBS. This medium was then replaced by the same medium containing 250 pCi/ml 3zP04(carrier-free), and the different mitogens. Cells were then incubated for 30 min at 37 "C. Acid-soluble proteins were extracted according to Glover (23) and separated by SDS-polyacrylamide gel electrophoresis as reported (24). We had previously shown by two-dimensional gel electrophoresis that 33-kDa stimulated phosphoprotein indeed corresponds to authentic ribosomal protein S6 (25). Preparation of RNA and Northern Hybridization Anulysis-Cells grown to confluence in 15-cm culture dishes and arrested in Go were stimulated by each individual growth factor for 90 min in fresh serumfree culturemedium used for Go arrest. Subsequently, cell monolayers were rinsed twice with phosphate-buffered saline a t room temperature, and total RNA were extracted by the guanidinium isothiocyanate technique as described (24). Poly(A)+ RNA were then separated on 1.2% formaldehyde-agarose gel, transferred to nitrocellulose filters, andhybridized with two 32P-labelednick-translated probes (24). The plasmid DNA probes used in this study were pSVc-myc-1,which carries a rearranged form of the mouse c-myc gene (26), and pRGAPDH13, containing rat glyceraldehyde-3-phosphatedehydrogenase cDNA (27). Measurement of Phsphoinositide Breakdown-Confluent cultures in 35-mm dishes were labeled with [3H]inositol (2 pCi/ml) over the 24-h incubation in serum-free DMEM required for Goarrest. Equilibrium was attained after7 h, as judged by the steady-state radioactivity incorporated in inositol phosphates. Unless otherwise specified, the cells were incubated for 10 min in HBS containing 20 mM LiQ and then stimulated with either a-thrombin or FGF. Incubations were stopped by quickly aspirating the medium, and adding 0.5 ml of 10% (w/v) HC104.Separation of free inositol, glycerophosphoinositol, IPl, IP2, and IP3 was carried out essentially as previously described (12, 28). Measurement of Diacylgycerol Formution-Confluent cultures in 35-mm dishes were labeled with [2-3H]glycerol(2 pCi/ml) over the 24-h incubation in serum-free DMEM. Labeling for 48 h (24 h during growth phase followed by 24 h during the serum-starvation period) gave identical results. Cell monolayers were rinsed with HBS and stimulatedin HBS at 37 "C with either a-thrombin or FGF. At different times, incubation was stopped by quickly aspirating the medium and adding 0.5 ml of ice-cold methanol. Cells were scraped with a rubber policeman, and the dishes rinsed with 0.8 ml of H20. Radioactive lipids were extracted by the method of Bligh and Dyer (29) and diacylglycerol was separated by thin layer chromatography (30) on Silica Gel G plates (Merck)with the solvent system of hexane, diethyl ether, formic acid (60401, v/v/v). Measurement of CytoplasmicFree ea2+Concentratiorts-Levels and variations of [Ca2+liwere measured as described by Moolenaar et al. (31). Briefly, CCL39-arrested cells grown on rectangular glass coverslips (3 X 1cm) were loaded with quin-2 by incubating them in Hepesbuffered DMEM containing 50 p~ quin-a/AM for 30 min a t 37 "C. After washing, the quin-2-loaded monolayers incubated in HBSwere inserted into a thermostated cuvette (37 "C) in a Perkin-Elmer 3000 spectrofluorometer. Fluorescence was continuously recorded at an

Pathway of FGF

16917

excitation wavelength of 339 nm and emission wavelength of 492 nm. Values of cytoplasmic free Ca" concentrations were calculated from fluorescence signals as described by Tsien et ai. (32,33). lz6I-EGFBiding-Cells grown to confluence in 24-well plates and arrested inGo wererinsed with the binding medium (DMEM buffered with 20mM Mops, pH 7.4, and containing 0.1% crystalline bovine serum albumin). Cell monolayers were then incubated for different times a t 20 "C with 0.5 ml of binding medium containing 1 ng/ml 'T-EGF (100-250 X lo3 dpm/ml) and the different effectors tested (a-thrombin, FGF, or PMA). Nonspecific binding was measured in the presence of 1pg/ml unlabeled EGF and represented less than 5% of total binding. Materids-Highly purified human a-thrombin (2660 NIH units/ mg) was generously provided by Dr. J. W. Fenton I1 (New YorkState Department of Health, Albany, NY). EGF and basic and acidic forms of bovine brain FGF were purified to homogeneity in our laboratory according to theprocedures published by Savage and Cohen for EGF (34), by Gospodarowiczet al. (15) for basic FGF, and by Bohlen et al. (16) for acidic FGF. Stock solutions of a-thrombin and FGF were made respectively a t 100units/ml and 10 pg/ml in phosphate-buffered saline containing 1 mg/ml crystalline bovine serum albumin. PMA, crystalline bovine insulin, and crystalline bovine serum albumin were purchased from Sigma. [32P]dCTP,[32P]orthophosphate (carrierfree), [2-3H]glycerol,and T - E G F were products of the Radiochemical Centre (Amersham, France); [methyl-3H]thymidine, [7-14C]benzoic acid, and my0-[2-~H]inositolwere from New England Nuclear. RESULTS

FGF Alone Is a Potent Mitogen for (&-arrested CCL39 Cells-The basic form of FGF eluted from heparin-Sepharose a t 2 M NaCl and electrophoresed in SDS-polyacrylamide gel was resolved as a single band with apparent molecular weight 16,000-17,000 (35). The preparations we have used in this study were estimated to be at least 95% homogenous as judged by silver staining of an overloaded gel (not shown). Fig. 1 shows that FGF alone is capable of reinitiating DNA synthesis in Go-arrested CCL39 cells (30-40-fold stimulation) as measured by [3H]thymidine incorporation or by labeled nuclei. This biological response is dose-dependent with a maximal effect around 50-100 ng/ml and is strongly potentiated by insulin (70% labeled nuclei). Moreover, Fig. 1 shows that the FGF mitogenic response is equivalent or even slightly higher than that obtained with maximal doses of thrombin. Finally, similar results were obtained with the acidic form of FGF, with the maximal mitogenic response noted with FGF concentrations varying between 50 and 300 ng/ml, depending on the preparation. Early Euents Stimulated by FGF and Thrombin in Go-

c

loo

2 d

3

75 -

50-

ii

P

25-

C

0-

FIG. 1. Comparison of FCF and thrombin for reinitiation of DNA synthesis in Go-arrestedCCL39 cells. Confluent cultures of CCL39 werearrested in GOby serum deprivation as indicated under "Experimental Procedures." The percentage of cell-replicating DNA was measured by counting labeled nuclei after 24-h stimulation by different growth factors: C, control (no addition); FGF, 100 ng/ml; FGF, EGF, INS, at 100 ng/ml, 50 ng/ml, and 10 pg/ml, respectively; THR, 1 unit/ml; and THR, INS, 1 unit/ml, 10 pg/ml. THR, athrombin; INS, insulin.

Mitogenic Signaling Pathway of FGF

16918

arrested CCL39 Cells-Fig. 2, A-Cy depicts three early events which are known to accompany the initial stages of the Go/ GI to S phase transition: cytoplasmic alkalinization, ribosomal protein S6 phosphorylation, and increased c-rnyc expression. The three early responses were found to be activated by basic FGF. Cytoplasmic alkalinization triggered by FGF or thrombin is amiloride-sensitive, reflecting the activation of the Na+/H+ exchange (22). However, when compared at maximumbiological concentrations, the pHi rise induced with 100 ng/mlFGF was alwayssmaller in amplitude and often transientas shown in Fig. 2A. Stimulation of ribosomal protein S6 phosphorylation (Fig. 2B)and increased c-rnyc mRNA (Fig. 2C) occurred similarly in FGF and thrombin-stimulated arrested cells. Second Messengers(IP3 and Didcylglycerol) Produced by the Action of Thrombin and FGF on Go-arrested CCD9 CellsWe have reported (11, 12) that addition of thrombin to Goarrested CCL39cells rapidly activates the breakdown of polyphosphoinositides.IP, formation can be detected as early as 5 s after thrombin stimulation followed by a subsequent rise in IP2andIPI.Incontrast, Fig. 3A shows that no detectable levels of IP3 and IP2 are released in cells stimulated for 1min with basic FGF.A lack of response was also observed by varying FGF concentrations from 10to 250 ng/ml and the time of stimulation from 10 s to 30 min (not shown). Even the highest mitogenic combination (FGF, EGF, insulin), capable of reinitiating DNA synthesis in more than 80% of the cells (Fig. I), failed to stimulate significant levels of IPS and IP2 (5% of the levels obtained with thrombin, Fig. 3A). To increase the sensitivity of detection of phospholipase C activation, we also compared the levels of total inositol phosphate formed in the presence of 20 mM LiC1, an inhibitor of inositol

monophosphatase (36). Addition of Li+ to quiescent CCL39 cells had no effect on the levels of IP over 60 min of incubation, indicating that the basal activity of phospholipase C in Go-arrested CCL39 cells is below the limit of detection (12). In the presence of Li+, thrombin (at 1 unit/ml) stimulated the level of IP 10-15-fold within 30 min of incubation, whereas FGF, at the maximal biological concentration, displayed a very weak response. IP released by basic FGF (Fig. 3B) or acidic FGF (notshown) did not exceed 5-8% of those released by thrombin. All these experiments with thrombin and FGF havealways been performed in parallel with thetest of mitogenicityshown in Fig. 1. The same results were also obtained by using two independent commercial sources of bovine brain basic FGF and by performing the inositol release assay within the same medium used for reinitiation of DNA synthesis. The simplest interpretation of this finding is that FGF is not an agonist of the polyphosphoinositidebreakdown pathway in Chinese hamster lung fibroblasts. As a consequence, we therefore did not expect that thelevel of diacylglycerol,the otherimmediate metabolite of this pathway, would change in response to FGF. Surprisingly, Fig. 3C shows that FGF at 10 or 100 ng/ml is as potent as thrombin in stimulating the level of total diacylglycerol in quiescent CCL39 cells. The increase of diacylglycerol content was dependent on the dose of the mitogen = l ng/mlfor FGF and lo-’ units/ml for thrombin); maximal stimulation (5080% above basal level) was obtained within 1 min after stimulation, remained elevated for 10 min, and thendeclined thereafter to return to thebasal level in 60 min (not shown). The relative diacylglycerolincrease was identical, whether the cells were labeled with [3H]glycerol for 1day or 2, and occurred also in response to EGFor PMA,two growth-promoting

B

MW.~O-~

m“3u

45-

A

S627K-

30 -

n -

THR C FGF p

28s

‘/,

I

10

20

TIME, rnin

-

T

30 “ -

F G F THR

C

FIG.2. Early responses initiatedby FGF and thrombin in Go-arrested CCL39 cells. A, Activation of the Na+/H+ antiporter measured by variations of intracellular pH. Cytoplasmic pH was measured by the distribution of [“Clbenzoic acid as described under “ExperimentalProcedures.” Serum-deprived CCL39 cells were stimulated at pH, = 7.4 with either 10 ng/ml FGF (A); 100 ng/ml FGF (A)or 1 unit/ml cy-thrombin (THR)(0). Results are expressed as the increase in pHi over unstimulated cells, and each point is the average of duplicate values. B, Phosphorylation of ribosomal protein S6. CCL39-arrested cells were stimulated for 30 min in the presence of 32P(250 pCi/ml) either with 1unit/ml a-thrombin, 100 ng/ml FGF,or no addition (C).Acid-soluble proteins were separated by SDS-polyacrylamide gel electrophoresis and phosphoproteins revealed by autoradiography as indicated under “Experimental Procedures.” C, c-myc mRNA expression. CCL39-arrested cells were stimulated by FGF (100ng/ml) or cy-thrombin (1unit/ml) for 90 min. Northern blots of the poly(A)+ RNA from total cell extract were hybridized with the mixture of two specific probes, one for c-myc (2.4kilobases) and one for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1.4 kilobases) as indicated under “Experimental Procedures.”

Mitogenic SignalingPathway of FGF

7

16919

A 1.5 -

C

1

-

FGFFGF THR EGF INS

B

1.25

-

-

1

1J-

0 C

-KIF EGF INS

nm

C

10FGF 100

TW,

FIG. 3. Polyphosphoinositide breakdown and diacylglycerol formation in Go-arrested CCL39 cells in response to FGF and thrombin. A , Release of inositol trisphosphate and bisphosphate after 1 min of stimulation. CCL39-arrested cells prelabeled at equilibrium with ny0-[2-~H]inositolwere stimulated either with (1unit/ml) or the combination FGF, EGF, insulin (INS) (250 ng/ml, FGF (250 ng/ml) or with a-thrombin (THR) 50 ng/ml, 10 pg/ml) or no addition (C).Inositol trisphosphate and inositol bisphosphate levels formed after 1min of stimulation in the absence of LiCl are given as the percent value of thrombin stimulation. B, Release of total inositol phosphate after 30 min of stimulation in the presence of LiCl. CCL39-arrested cells were stimulated with growth factors for 30 min in the presence of 20 mM LiCI. Total inositol phosphate released (IPl, IP,, IP3) is represented as thepercent value of thrombin stimulation. Concentrations of the growth factors were the same as in A except that FGF was at 50 ng/ml. Each value is the mean of duplicate experiments. C, Increase of diacylglycerol content. CCL39-arrested cells prelabeled at equilibrium with [3H]glycerol werestimulated for 10 min with 1unit/ ml a-thrombin or 10 ng/ml FGF (10) or 100 ng/ml FGF (100). Diacylglycerol (DAG) was measured as described under "Experimental Procedures," and relative increase to the control value C (no addition) is reported. The data presented are themean of duplicate determinations and arerepresentative of results obtained in three independent experiments. Values were normalized to the totalamount of phospholipids.

A FIG. 4. Effects of FGF and thrombin on [Caa+Jlin CCL39-arrested cells. Quin-2loading of Go-arrested CCL39 cells and fluorescence monitoring were performed as described under "Experimental Procedures." Traces represent typical responses of quin-2 fluorescence to addition of 1 unit/ml athrombin or 100 ng/ml FGF in Ca2+containing medium (A, C)or in Ca2+-free medium (B, D ) . For the experiments in Caz+-freemedium, CaZ+was reduced to 1 mM and at the time indicated 2 m M EGTA was added to themedium.

E5

1.8mM Ca2:

2or'

-

B

ca2,+

free

I

nMCa2

'

nMca2+ -100

-48 lot 0

0

"1 OL

20

FGF

-

lmin

-250

c " (

D

min

1 EGTA

-227

"4,

1rnin

agents not coupled to phospholipase C activation (12, 13). FGF and Thrombin Induce a Rapid Rise in [Ca2+Ii in Goarrested CCD9 Cells-Fig. 4,A and C, shows that addition of thrombin or FGF to Go-arrested CCL39 cells rapidly stimulated a rise in [Caz+],. Theapparent resting [Ca2+Iilevel estimated from quin-2 cellular fluorescence was -240 nM. This level wasimmediately raised to -400 nM upon thrombin addition (Fig. 4A)before declining to 260 nM over 4 min. The [Ca2+]iincreased by FGF was always moderate in amplitude (30%increase instead of 70-90% in thecase of thrombin) but the rise was sustained for a much longer period. Although the amplitude of the response varied from experiment to experiment, the Ca2+signal kinetics were always characteristic of

either a thrombin effect (transient response) or of an FGF effect (lasting response). This difference in Ca2+signal became more striking when the responses of these two growth factors were compared in Ca2+-freemedium (Fig. 4, B and 0). Addition of EGTA to complex external Ca2+rapidly induced a progressive decrease in [Ca2+]i,but did not preventthe thrombin-induced transient rise in [Ca2+Ii(Fig. 4B). In contrast, the FGF-induced rise in [Ca2+Iiwas lost in theabsence of Ca2+in theexternal medium (Fig. 40). FGF Does Not Down-regulate lz5I-EGFBinding in CCL39 Cells-Hydrolysis of polyphosphoinositides is one of the major pathways leading to activation of protein kinase C. Because

16920

Signalin

Mitogenic

.g Pathway of FGF

that IP3induces the release of Ca'+ from an intracellular store believed to be a component of the endoplasmic reticulum (5, 38). The thrombin-induced [Ca2+IL rise is rapid, transient, and independent of external Ca'+, a result similar to that reported for serum and PDGF (31),two effectors provoking the rapid formation of IP,. In addition, the transient formationof IP3, in particular the inositol 1,4,5-trisphosphate isomer (39) appears to correlate with the transient[Ca2+]i riseobserved. In contrast, the FGF-induced [Ca2+ILrise is of longer duration and is strictly dependent upon externalCa'+, a result consistent with the lack of IPS production in response to FGF. FGF must therefore activate some Ca'+ entry at the plasma membrane level, a mechanism proposed for EGF in mouse 3T3 cells (40) and A431 cells (41) andfor antibodies inT lymphocytes (42). These results indicate that the polyphosphoinositide cycle is not involved in the mitogenic signaling pathway of FGF. We therefore did not expect protein kinase C to be activated (2), unless FGF was capable of stimulating theproduction of diacylglycerol via a different route. All the growth-promoting agents we tested,whethertheyactivate phospholipaseC (thrombin, serum) or not (EGF, FGF, PMA), stimulated by DISCUSSION 1.5-2-fold the levels of diacylglycerol. These findings The findings presented here demonstrate that the addition prompted us toanalyze more directly whether protein kinase of highly purified FGF to quiescent Chinese hamster lung C is activated by FGF. EGF-receptor has been shown to be an in vivo target of proteinkinase C, itsphosphorylation fibroblasts leads to reinitiation of DNA synthesis without activation of polyphosphoinositide breakdown. This conclu- leading to a pronounced decrease in EGF binding capacity sion isbased on the observation that little, if any, IP, and IP2 (17-19, 37). A decrease in EGF binding has been observed is formed whencells are treated with maximal concentrationswhen intact cells are treated with PMA (43, 44), synthetic of basic or acidic FGF. This negative result is not due to a diacylglycerol (19), vasopressin (45), bombesin (7, 8), or failure of the assay or toa peculiarity of the cells used, since PDGF (46). This effect is rapid and is due primarily to a a loss under the same conditions, a-thrombin, an equally potent decrease in the affinity for EGF binding rather than to mitogen, rapidly induced the formation of IPS andIP,. More- of receptors from the surface (8). Treatment of Go-arrested over, the thrombin-inducedpolyphosphoinositide breakdown CCL39 cells with PMA or thrombin rapidly reduced the EGF highaffinity occurred at the same rate in cells stimulated together with binding capacity(50-60%), inparticularthe binding sites (not shown). Interestingly, within the samecells, FGF and thrombin. It is only under the extremely sensitive conditionsobtained by the use of the Li' blocker of the acidic or basic FGF had no effect on EGF binding. Another inositol lipid cycle (36) that a very weak formation of IP, (less in vivo assay for protein kinase C is based on its inhibitory be detected with FGF. effects on phospholipase C activation. We have recently rethan 8%of the thrombin response) can The variations of [Ca2+Ii thatwe observed in response to ported a rapid desensitization of thrombin-induced phosphoeitherFGF or thrombinareentirelyconsistent with the lipase C activity in CCL39 that we have interpreted, in part, above-presented results.Indeed, it is now well established as a feedback inhibition resulting from activation of protein kinase C (12). In this assay, PMA inhibited by 60-70% the rate of thrombin-induced I P release (12),whereas, here again, -100 FGF had no effect (data not shown). The simplest interpretation of these two independent andconverging results is that FGF does not activate protein kinase C in CCL39 cells. At thesametime,asthis conclusion pointsout, changes in 6 1 diacylglycerol levels should be interpreted with caution. An increase in diacylglycerol does not necessarily mean activation - 50 of protein kinase C. For example, stimulation of triglyceride hydrolysis by the action of lipoprotein lipase will produce a diacylglycerol of 2,3-sn configuration which does not activate protein kinase C (2). We are currently analyzing the specific changes of 1,2-sn-diacylglycerol in response toFGFand thrombin (47). -0 Our results and theconclusion we have drawn are in disaPMA FGF C THR greement with those of Takai's group (48). They have shown lu/ml100ng/ml10100ng/ml that bovine pituitaryFGF inducesCa2+mobilization and protein kinaseC activation in Swiss 3T3 cells presumably FIG. 5. Effects of thrombin, FGF, and phorbol ester on "'1EGF binding to CCL39 cells. CCL39-arrested cells grown in 24- through inositol lipid hydrolysis. Protein kinase C activation well plates were incubated for 1 h at 20 "C in the presence of 0.5 ml was measured by the degree of phosphorylation of the 80-kDa of binding medium containing 1 ng/ml lZ5I-EGFand either 1 unit/ml protein (7, 49). Although the phosphorylation of this protein a-thrombin, or 100 ng/ml PMA, or 10 ng/ml FGF, or 100 ng/ml FGF, or no addition. Specific EGF binding was expressed as percent of the is correlated with proteinkinase C activation, it isalso phosvalue obtained with no effector. The data presented are the mean of phorylated by other pathways;' therefore, increased phospho-

this pathway was clearly not activated by FGF in CCL39, we asked whether FGF was capable of activating protein kinase C by an independent route. We have chosen the now classical EGF-receptor binding assay to monitor the in vivo kinase C activation (8,19). Indeed, elegant studies have recently established that PMA-induced down-regulation of EGF binding results from phosphorylation of the EGF-receptor by kinase C (37). As reported for a variety of cell systems, we found that additionof PMA to quiescentCCL39 cells reduced EGF specific binding by60% after 1 h. This inhibition of EGF binding was rapid (60% inhibition was already obtained within 5 min) and affectedprimarilyhighaffinity binding sites as judged by Scatchard analyses (not shown). As expected, we obtained a similar inhibition(50-60%) with thrombin. This inhibitory effect is rapid (no lag), dose-dependent (maximum at 1unit/ml, ICso at lo-' units/ml), and partially reversed by pertussis toxin, an inhibitor of thrombin-induced inositol lipid breakdown (11).In contrast, neither the basic nor acidic forms of FGF affected the kinetic parameters of EGF binding. A brief account of these results is summarized in Fig. 5 .

duplicate determinations (less than 10% variation) and are representative of results obtained in three independent experiments.

I. Magnaldo and J. Pouysskgur, unpublished observations.

Mitogenic Signaling Pathway of FGF rylation of the 80-kDa proteindoes not provide absolute proof for protein kinase C activation. As far as the stimulation of inositol lipid turnover is concerned, the authorswere not able to detect an increase in IP3 in response to FGF, but they reported a slight stimulation of 32Pincorporation into phosphatidylinositol and a moderate increase in cellular diacylglycerol in comparison to PDGF (48). However, these two assays areof questionable value to constitute a direct evidence in favor of an activation of phospholipase C by FGF. In an attempt toclarify these discrepancies, we have analyzed FGF effects on the rate of polyphosphoinositide hydrolysis in Swiss 3T3 andbovine epithelial lens cells (BEL). Go-arrested BEL cells, like CCL39 cells, can be mitogenically stimulated with FGF with no activation of polyphosphoinositide hydrolysis, whereas parallel mitogenic stimulation of these cells with prostaglandin Fla induces a rapid and important release of IP.3 In Swiss 3T3 cells however, we found, in the presence of Li', a significant and reproducible release of IP in response to high concentrations of basic and acidic FGF (100-200 ng/ ml). IP release was not detectable within 5 min but developed slowly with time, contrasting with the immediate release obtained with serum in these cells. We favor the idea that FGF must have a unique signal-transducing mechanism, independent of the natureof the cell expressing FGF-receptors. Our interpretation of the differences seen between Swiss 3T3 and CCL39/BEL cells is as follows: 1)Swiss 3T3 cells appear to be more "permissive" to a variety of growth-promoting agents than are CCL39 and BEL cells, perhaps because they express a much higher variety of receptors. For example, among the growth factors (FGF, platelet growth factor 2a, bombesin, bradykinin, vasopressin) known to stimulate DNA synthesis in Swiss 3T3 cells, only FGF stimulates CCL39 cells and only FGF and platelet growth factor 2a stimulate BEL cells. 2) Some other growth-promoting activity might copurify with the FGF preparation(that we estimate to be about 95% homogenous) and might be responsible for the weak stimulation of inositol lipid breakdown in themore permissive Swiss 3T3 cells. This point iscurrently under study with very highly purified FGF. In summary, we conclude from the results presented in this paper that the receptor(s) of the basic and acidic forms of FGF (35,50) is (are) not coupled to phospholipase C-mediated inositol lipid breakdown. Although no evidence, so far, has been reported in favor of an FGF-induced autophosphorylation of tyrosine residues (51), this receptor remains a likely candidate for being a protein kinase. We have also clearly shown that early mitogenic events such as Na+/H+ exchange activation, increased cytoplasmic Ca2+,S6 phosphorylation, c-myc expression, and reinitiation of DNA synthesis can be triggered via a transducing mechanism which does not involve inositol lipid breakdown and protein kinase C activation. This finding strengthens a similar conclusion attained from the study of the mechanism of EGF or EGF/insulin mitogenic action (13,14). It remains, now, to be established whether the inositol lipid signaling pathway stimulated by potent mitogens such as PDGF, thrombin, or bombesin plays an essential role in the commitment of quiescent cells to replicate DNA synthesis. Acknowledgments-We are grateful to Dr. Wayne Anderson for critical reading of the manuscript and to Joyce Sharrar for expert secreterial assistance. M. Moenner, I. Magnaldo, G. L'Allemain, D. Barritault, and J. Pouyssegur, manuscript in preparation.

1. 2. 3. 4. 5. 6.

7. 8. 9. 10. 11. 12. 13.

16921

REFERENCES Berridge, M. (1985) Sci. Am. 253, 142-152 Nishizuka, Y.(1986) Science 233, 305-312 Takai, Y.,Kishimoto, A., Inoue, M., and Nishizuka, Y. (1977) J. Bwl. Chem. 252,7603-7609 Kaibuchi, K.,Sano, K., Hoshijima, M., Takai, Y., and Nishizuka, Y.(1982) Cell Calcium 3,323-335 Streb, H., Irvine, R., Berridge, M., and Schulz, I. (1983) Nature 306,67-69 Habenicht, A. J. R., Glomset, J. A., King, W. C., Nist, C., Mitchell, C. D., and Ross, R. (1981) J. Biol. Chem. 256,12329-12335 Zachary, I., Sinnett-Smith, J., and Rozengurt, E. (1986) J. Cell BWl. 102,2211-2222 Brown, K., Blay, J., Irvine, R., Heslop, J., and Berridge, M. (1984) Bwchem. Bwphys. Res. Commun. 123,377-384 Coughlin, S., Lee, W., Williams, P., Giels, G., and Williams, L. (1985) Cell 4 3 , 243-251 Carney, D., Scott, D., Gordon, E., and LaBelle, E.(1985) Cell 4 2 , 479-488 Paris, S., and Pouyssegur, J. (1986) EMBO J. 5,55-60 L'Allemain, G., Paris, S., Magnaldo, I., and Pouyssegur, J. (1986) J. Celt. Physiol., in press L'Allemain, G., and Pouysdgur,J. (1986) FEBS Lett. 197,344348 ."

14. Besterman, J. M., Watson, S. P., and Cuatrecasas, P. (1986) J. Bwl. Chem. 261,723-727 15. Gospodarowicz, D., Cheng, J., Lui, G.-M., Baird, A., and Bijhlen, P. (1984) Proc. Natl. Acad. Sci. U.S. A. 81,6963-6967 16. Bohlen, P., Esch, F., Baird, A., and Gospodarowicz, D. (1985) EMBO J. 4 , 1951-1956 17. Cochet, C., Gill, G.N., Meisenhelder, J., Cooper, J. A,, and Hunter, T. (1984) J. Bbl. Chem. 259,2553-2558 18. Davis, R. J., and Czech, M. P. (1984) J. Biol. Chem. 259,85458549 19. Sinnett-Smith, J., and Rozengurt, E. (1985) J. Cell. Physiol. 1 2 4 , 81-86 20. Pouyssbgur, J., Chambard, J. C., Franchi, A., Paris, S., and Van Obberghen-Schilling, E. (1982) Proc. Natl. Acad. Sci. U. S. A. 79,3935-3939 21. Pirez-Rodriguez, R., Chambard, J. C., Van Obberghen-Schilling, E., Franchi, A., and Pouysskgur, J. (1981) J. Cell. Physwl. 1 0 9 , 387-396 22. L'Allemain, G., Paris, S., and Pouyssegur, J. (1984) J. Biol. Chem. 259,5809-5815 23. Glover, C. (1982) Proc. Natl. Acad. Sci. U. S. A. 79, 1781-1785 24. Van Obberghen-Schilling, E., Chamhard, J. C., Paris, S., L'Allemain, G., and Pouyssegur, J. (1985) EMBO J. 4,2927-2932 25. Chambard, J.-C., Franchi, A., Le Cam, A., and Pouyssegur, J. (1983) J. Bwl. Chem. 253, 1706-1713 26. Land, H.,Parada, L., and Weinberg, R. (1983) Nature 304,596602 27. Fort, P., Marty, L., Piechaczyk, M., El Sabouty, S., Dani, C., Jeanteur, P., and Blanchard, J. M. (1985) Nucleic Acid Res. 13,1431-1442 28. Bone, E., Fretten, P., Palmer, S., Kirk, C., and Michell, R. (1984) Bwchem. J. 221,803-811 29. Bligh, E., and Dyer, W. (1959) Can. J. Bwchem. Physwl. 3 7 , 911-917 30. Rittenhouse-Simmons, S. (1979) J. Clin Invest. 63,580-587 31. Moolenaar, W. H., Tertoolen, L. G. J., and de Laat, S. W. (1984) J. Bwl. Chem. 259,8066-8069 32. Tsien, R., Pozzan, T., and Rink, T.(1982) Nature 295,68-71 33. Tsien, R., Pozzan, T., and Rink, T.(1982) J. Cell Bwl. 9 4 , 325334 34. Savage, C. R., Jr., and Cohen, S. (1972) J. Biol. Chem. 2 4 7 , 7609-7611 35. Moenner, M., Chevallier, B., Badet, J., and Barritault, D. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,5024-5028 36. Hallcher, L. M.,and Sherman, W. R. (1980) J. Bwl. Chem. 2 5 5 , 10896-10901 37. Lin, C., Chen, W., Lazar, C., Carpenter, D., Gill, G., Evans, R., and Rosenfeld, M. (1986) Cell 4 4 , 839-848 38. Burgess, G., Godfrey, P. P., McKinney, J., Berridge, M., Irvine, R., and Putney, J. (1984) Nature 309,63-66 39. Burgess, G., McKinney, J., Irvine, R., and Putney, J. (1985) Bwchem. J. 2 3 2 , 237-243 40. Hesketh, T., Moore, J., Morris, J., Taylor, M., Rogers, J., Smith,

16922

Mitogenic Signaling Pathway of FGF

G., and Metcalfe, J. (1985)Nature 313, 481-484 41. Moolenaar, W. H.,Aerts, R. J., Tertoolen, L. G. J., and de Laat, S. W. (1986)J. Bwl. Chem. 261,279-284 42. Oettgen, H.,Terhorst, C., Cantley, L., and Rosoff, P. (1985)Cell 40,583-590 48. 43. Lee, L., and Weinstein, I. (1979)P m . N d . A d . Sci. u. S. A. 76,5168-5172 44. Brown, K., Dicker, p-, and bzengud, E. (1979)X ~ C h t m Bio. phys. Res. Commun. 86,1037-1043 5631-5637 45. Rozengurt, E.,Brown, K.D., and Pettican, P. (1981)J. Biol. 13860-13868 Chem. 266, 716-722

46. Collins, M. K. L., Sinnett-Smith, J. W., and Rozengurt, E.(1983) J.Biol. Chem. 258,11689-11693 47. Preiss, J., Loomis, C. R., Bishop, W. R., Stein, R., Niedel, J. E., and Bell, R. M. (1986)J. Bwl. Chern. 261,8597-8600 Tsuda, T., Kaibuchi, K., Kawahara, Y., Fukuzaki, H., and Takai, Y.(1985)FEBS Lett. 191,205-210 49. R ~E., Rodriguez-pena, ~ ~ A*, and~Smith, K.~ (1983)pm, , Natl. Acad. Sci. U. S. A . 80,7244-7248 50. Neufeld, G., and Gospo&rowicz, D. (1986)J. Bbl. Chm. 261, 51. Neufeld, G., and Gospodarowicz, D. (1985)J. B i d . Chem. 260,