Synergistic activation of signalling to extracellular signal-regulated kinases 1 and 2 by epidermal growth factor and 4b-phorbol 12-myristate 13-acetate Jorrit J. Hornberg 1 , Marloes R. Tijssen 1 and Jan Lankelma 1,2 1 Department of Molecular Cell Physiology, Institute of Molecular Cell Biology, Faculty of Earth and Life Sciences, Vrije Universiteit, Amsterdam, the Netherlands; 2 Department of Medical Oncology, VU Medical Center, Amsterdam, the Netherlands Signal transduction pathways are often embedded in com- plex networks, which result from interactions between pathways and feedback circuitry. In order to understand such networks, qualitative information on which inter- actions take place a nd quantitative data o n their strength become essential. Here, we have investigated how the mul- tiple interactions between the mitogen-activated protein kinase cascade and protein kinase C (PKC) affect the time profile o f extracellular signal-regulated kinase (ERK) phos- phorylation upon epidermal growth f actor (EGF) stimula- tion in normal rat kidney fibroblasts. This profile i s a major determinant for the cellular response that is evoked. We found that EGF s timulation leads to a biphasic ERK-PP pattern, consisting of an initial peak and a r elaxation to a low quasi-steady state-phase. Costimulation with the EGF and PKC activator, 4b-phorbol 12-myristate 13-acetate (PMA) resulted in a similar p attern, but the ERK-PP concentration in the quasi-steady state-phase was synergistically higher than afte r stimulation with either EGF o r PMA only . This resulted in prolonged signalling to ERK. PMA increased the EGF concentration sufficient to obtain half-maximum ERK phosphorylation. These data suggest that PKC amplifies EGF-induced signalling t o E RK, w ithout incre asing its sensitivity to l ow EGF c oncentrations. Furthermore, P KC inhibition did not affect the ERK-PP time profile upon EGF stimulation and a cellular phospholipase A2 (cPLA 2 ) inhibitor d id not decrease the synergistic effect of EGF and PMA. This indicates that the positive feedback loop from ERK to Raf via cPLA 2 and PKC does not contribute sig- nificantly to signalling from EGF to ERK i n normal rat kidney cells. Taken together, we provide a quantitative description of which reported interactions in this network affect the time p rofile of ERK phosphorylation. Keywords:EGF;MAPK;PMA;signalingnetwork;syner- gism. The increase in knowledge of the building blocks of living cells (genes, proteins) will stimulate t he development of integrative biology [1,2]. Cellular signalling provides an interesting platform for this integrative or Ôsystems approachÕ. Many signalling proteins have been identified and how they ÔcommunicateÕ with each other through signal transduction pathways has b een extensively researched. These pathways can interact at many levels (e.g. by direct interaction of the molecules or by regulation o f gene transcription), w hich gives rise to large signalling networks. In order to fully understand how such networks operate, it is necessary to integrate experimental d ata a nd to understand how (qualitatively) and to w hat extent (quantitatively) interactions in the network take place. By using biomathe- matical models, p redictions can be made about the beha- viour of si gnalling networks or, ultimately, of whole cells or organisms [1,3–8]. Among the most intensively studied signal transduction pathways are the mitogen-activated protein kinase (MAPK) cascades, which are involved in many cellular processes, such as proliferation, differentiation and apoptosis [9,10]. The mitotic MAPK pathway, via extracellular signal- regulated kinase (ERK), c an be act ivated by various extracellular stimuli, e.g. epidermal growth f actor (EGF), which bind to dedicated receptors. Upon EGF binding, its receptor (EGFR) dimerizes, leading to autophosphoryla- tion of tyrosine residues on the cytoplasmic domain of the receptor, thereby creating docking sites for adaptor pro- teins, such as Shc and G rb2. The latter protein recruits Sos to the plasma membrane, which causes the activation of Ras by exchanging GDP, bound to Ras, for GTP [11–13]. Ras-GTP can bind cytoplasmic Raf1 leading to its Correspondence to J. Lankelma, Department of Molecular Cell Physiology Faculty of Earth and Life Sciences Vrije Universiteit Amsterdam, De Boelelaan 1085 1081 HV Amsterdam, the Nether- lands. Fax: +31 20 4447229, Tel.: +31 20 4447248, E-mail: j.lankelma@vumc.nl Abbreviations: ATK, arachidonyl trifluoromethylketone; cPLA2, cel- lular phospholipase A2; DAG, diacylglycerol; DMEM, Dulbecco’s modified Eagle’s medium; EGF, epidermal growth factor; EGFR, EGF receptor; ERK, extracellular signal-regulated kinase; ERK-PP, doubly phosphorylated ERK; Grb2, growth factor receptor binding protein 2; IP3, inositol triphosphate; MAPK, mitogen-activated pro- tein kinase; MEK, MAPK/ERK kinase; MKP, MAPK phosphatase; NRK, normal rat kidney; PDGF, platelet-derived growth factor; PKC, protein kinase C; PMA, 4b-phorbol 12-myristate 13-acetate; PLC-c, phospholipase C-g; Shc, Src homology and collagen domain protein; TBS, Tris-buffered saline. Note: a website is available at h ttp://www.bio.vu.nl/vakgroepen/mcp/ (Received 2 2 June 2004, accepted 6 August 2004) Eur. J. Biochem. 271, 3905–3913 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04327.x phosphorylation and activation. Although R af1 c an be phosphorylated by many kinases, the exact mechanism by which it is activated after EGF stimulation is not entire ly clear [14]. Subsequently, Raf1 phosphorylates MAPK/ERK kinase (MEK) 1 and 2, which in turn phosphorylate ERK1 and ERK2. Phosphorylated ERK (ERK-PP) has several different cytoplasmic and nuclear targets. Transcription factors a ctivated by ERK-PP that indu ce expression of genes involved in cell cycle progression include Elk1, c-fos, c-Jun and c-myc [9,15]. The duration of ERK activation (transient or sustained) determines the repertoire of target genes expressed [16], and also affects the type of cellular response that is evoked [17,18]. Activation of ERK is required for proliferation of fibroblasts [19] and constitutive ERK activation frequently occurs in human primary tumours and tumour cell lines [20]. The latter is often caused by mutations in the genes encoding the constituents of the p athway, such as Ras21, rendering them over- activated. Signalling pathways are generally not simple linear chains, but have several feedback mechanisms and cross-reactivity with other signal t ransduction pathways [10], w hich may lead t o e mergent properties such as sustained oscillations and bistability [4,8,22]. We have investigated the interaction between the ERK cascade and protein kinase C (PKC). PKC is also involved in processes like proliferation, differentiation and cell death [23]. Several different interactions between these signal transduction modules have been reported. PKC can directly activate the MAPK pathway by phosphory- lating Raf [24–26]. It has also been implicated in a positive feedback loop of the MAPK pathway [4,8]. Therein, ERK-PP phosphorylates cytosolic phospholipase A 2 (cPLA 2 ) [27], causing the release of arachidonic acid, which, together with calcium and diacylglycerol (DAG), activates PKC (reviewed in [28]). Furthermore, PKC c an phosphorylate E GFR [ 29]. T his inhibits tyrosine kinase activity of the receptor and causes the decrease of EGF binding affinity [30–32]. It also results in diversion of the internalized EGFR from the regular degradative pathway to the recycling endosome [33]. EGFR is also capable of signalling to PKC, via phospholipase C -c (PLC -c) phosphorylation. PLC- c catalyses the production of inositol triphosphate (IP 3 ) a nd DAG. IP 3 brings about calcium release, which together with DAG activates PKC (reviewed in [ 28]). Taken together, all t hese interactions constitute a very complex signalling network (Fig. 1). We hypothesized that this network is cap able of quasi- intelligent behaviour, for instance by making the output (ERK phosphorylation) dependent on the integration of two signal inputs. Therefore, we measured signalling t o ERK after stimulation with EGF, PKC activator, 4b-phorbol 12-myristate 13-acetate (PMA) and a combi- nation of both. We show that, when both signal inputs were given simultaneously, ERK phosphorylation was synergistically activated, leading to a prolonged active quasi-steady state. Furthermore, we determined the relative qua si-steady state-concentrations of ERK-PP at different EGF concentrations and f ound that PKC not only affects the maximum level of the stimulus-response curve, but surprisingly also causes an increase in the EGF concentration sufficient for half-maximum ERK phos- phorylation. Experimental procedures Cell culture Normal rat k idney ( NRK) fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Biowhit- taker Europe), supplemented with 10% (v/v) foetal bovine serum (FBS, Gibco), 100 lgÆmL )1 penicillin and 100 lgÆmL )1 streptomycin in a humidified 5% (v/v) CO 2 incubator at 3 7 °C. For s erum-starvation, cells were washed once with 1· Hank’s buffered salt solution (Gibco) and cultured in DMEM, supplemented with 0.5% (w/v) BSA, (AppliChem), 1 00 lgÆmL )1 penicillin and 100 lgÆmL )1 streptomycin. Stimulation experiments Cells grown i n culture dishes (fo r Western blot analysis) or on glass cover slips (for immunocytochemistry) to subcon- fluency were serum-starved for 3 days in order to be arrested in the G 0 -phase of the cell cycle. Cells were stimulated with various concentrations of EGF (Becton Dickinson) and/or PMA (Calbiochem) f or different periods of time as indicated. PKC was inhibited by preincubation with 5 l M bisindolylmaleimide I (also referred to as GF109203X; Calbiochem) for 1 h [ 34] and c PLA 2 was inhibited by preincubation for 1 h with 10 l M arachidonyl Fig. 1. The complex str ucture of the signalling network to ERK. Depicted are t he MAPK and PKC sign alling modules (blue a nd yellow boxes, r esp ective ly). A c tivated EGFR signals via Ras through the MAPK cascade t o ERK, which leads to the activation of various transcription factors (TFs). EGFR can a lso activate PKC through PLCc. These two signalling modules communicate with each other via several m echanisms: (a) E RK activates cPLA 2 , which releases arachi- donic acid (AA) that, togethe r with calcium, can ac tivate PKC; (b) PKC d irectly phosphorylates Raf; (c) PKC also p hospho rylates EGFR. Internalized EGFR (iEGFR), although still capable of sig- nalling t o Ras, is n ormally degraded over t ime [57]. PKC-mediated phosphorylation of iEGFR causes it to recycle back to the cell su rface. In addition it cau ses a decreased EGF bin ding affinity and tyrosine kinase activity of the r eceptor. Also shown are the stimulators and inhibitors(depictedinred)usedinthisstudy.EGFwasusedtoactivate the EGFR, PMA to activate P KC. B isindolylmaleimide was used to block P KC activity an d ATK t o block cPLA 2 activity. 3906 J. J. Hornberg et al. (Eur. J. Biochem. 271) Ó FEBS 2004 trifluoromethylketone (ATK) [35] or 4-bromophenacyl bromide [36]. Western blot analysis After stimulation, cells were washed twice with i ce-cold phosphate-buffered saline (NaCl/P i ;17m M NaH 2 PO 4 , 38.5 m M Na 2 HPO 4 ,68m M NaCl, pH 7.4) and incubated on ic e with lysis buffer [10 m M Tris/HCl, pH 7.5, 150 m M NaCl, 0.1% (v/v) SDS, 0.1% (v/v) octylphenolpoly(ethylene glycolether) (Nonidet P40), 0.1% (w/v) sodium deoxycho- late, 50 m M NaF, 1 m M Na 3 VO 4 ,1· Complete protease inhibitor mix (Roche)] for 20 min. Cell l ysates were scraped in lysis buffer using a cell scrape r (25 cm/1.8 cm, Costar), collected, v ortexed for 10 s, frozen in liquid nitrogen and stored at )80 °C. Protein contents in the cell lysates were determined with the bicinchoninic a cid assay (Pierce). Proteins were separated by SDS/PAGE. For each sample, exactly 10 lg of total protein w as loaded on the gel in loading buffer (250 m M Tris/HCl, pH 7.6, 8 % (w/v) SDS, 40% (v/v) glycerol, 0.05% (w/v) b romophenol blue, 2 0 m M dithiothreitol). Proteins were electrotransferred to Immuno- Blot TM poly(vinylidene difluoride) membranes (Bio-Rad) using 400 mA overnight at 4 °C. Membranes were w ashed in Tris-buffered saline (TBS: 2 0 m M Tris/HCl, p H 7.6, 150 m M NaCl) s upplemented w ith 0 .05% (v/v) Tween-80 (TBS-T), preincubated for 1 h at r oom temperature w ith blocking buffer [5% (w/v) skimmed milk powder (Oxoid) in TBS-T], supplemented with 0.5 m M Na 3 VO 4 , and incu bated overnight at 4 °C with mo noclonal mouse anti-(phospho- p42/44 MAP kinase) Ig (Cell Signalling) in blocking buffer (1 : 2000), supplemented with 0.5 m M Na 3 VO 4 .After washing, membranes were incubated for 1 h at room temperature with horseradish peroxidase-conjugated goat anti-(mouse IgG) Ig (Bio-Rad) in blocking buffer (1 : 3000). Membranes w ere washed a gain a nd then incubated for 5 m in with Lumi-Light PLUS Western Blotting Substrate (Roche). Signals were detected with a FluorS TM MultiI- mager (Bio-Rad) and quantified using the MULTI - ANALIST software (Bio-Rad). All measurements were performed in the linear detection range of this method. All time c urves for Fig. 2 were m easured in i ndepend- ent experiments, and as we wanted to compare them to each other and calculate the stan dard errors for each time point, the individual time curves had to be scaled to each other. Scaling the curves with re spect to the m aximal ERK-PP concentration was not possible, as, due to the relatively dynamic nature of the curve around this time point, the maximal ERK-PP concentration measured for a certain time curve could not be exactly equal to the ÔrealÕ maximum that is reached in the cells. The maximal ERK-PP concentration measured for independent curves can therefore differ and scaling the whole curve to this time point would introduce errors at other time points. To avoid such problems, we chose to use all time points to scale all curves to one (arbitrarily chosen) ÔrepresentativeÕ curve. Therefore, we applied a multivariate least squares approximation [37,38] of the type Y ¼ X b +e,inwhich Y is the representative curve, X is the m atrix of all other curves (which were treated similarly), b represents the regression coefficients (one for each curve X)ande is the error vector (zero mean, common variance [37,38]). For the t hree different conditions [stimulated with (a) EG F (b) PMA or (c) EGF + (PMA)], the scaled time curves were drawn ( i.e. Xb) in Fig. 2 . The s tandard errors were calculated as described previously [39]. Immunocytochemistry After stimulation, cells were washed twice w ith ice-cold NaCl/P i , fixed by i ncubation for 30 min at 4 °Cwithice- cold 4% (v/v) paraformaldehyde in NaCl/P i and washed Fig. 2. Biphasic ERK-PP time profile induced by EGF or PMA alone and synergistic ERK phosphorylation induced by EGF and PMA together. Cells were serum-starved for three days and subsequently stimulated for the indicated t imes ( x-axis) with 10 ng ÆmL )1 EGF (n), 100 n M PMA (h) or both EGF and PMA (s). Cells were harvested a nd ERK-PP was m easured in the cell lysates by quantitative Western blotting. EGF or PMA stimulation leads to a biphasic time profile, with a high peak that decreases to a low quasi-steady state-level. EGF and PMA costimulation leads to synergistic ERK phosphorylation in this second phase. The curves shown are the result of five ind epende nt experiments, that were scaled to each other u sing a multivariate l east squares approximation (see E xperimental procedures). E rror bars represent the standard error o f the me an. Ó FEBS 2004 Synergistic ERK activation by EGF and PMA (Eur. J. Biochem. 271) 3907 with TBS-Triton (TBS, supplemented with 0.1% (v/v) Triton X-100). Cells were then incubated with 100% (v/v) methanol for 10 m in at )20 °C in order to permeabilize cellular membranes and w ashed. C ells were then incubated for 1 h a t room temperature with 5% FBS in TBS-Triton and subsequently incub ated overnight at 4 °C with mono- clonal mouse anti-(phospho-p42/44 MAP kinase) Ig (Cell Signalling) in 5% (w/v) BSA in TBS-Triton (1 : 400). C ells were washed for 15 min with TBS-Triton, for 15 min with 0.1% (w/v) BSA in TBS-Triton and incu bated for 2 h at room temperature with Cy 5 TM -labeled g oat anti-(mouse Ig) (Amersham) in 3% (w/v) BSA in TBS/Triton (1 : 400). Next, cells were washed with TBS/Triton and then with demi-water. The glass slides were air-dried, inversely p laced in Vectr ashield for fluorescence (Vector) on a microscope slide and stored in the d ark at 4 °C. Fluorescence was detected using a confocal scanning laser microscope (Leica TCS 4D). A krypton-argon laser line (647 nm) w as used for excitation of the Cy 5 TM -label, and a long pass filter (665 nm) was used f or detection o f the emitted light (with b eam splitter at 660 nm). Obtained images were quantified using the SCION IMAGE software (Scion Corporation). Results Biphasic time profile of ERK-PP by EGF or PMA stimulation We first determined the dynamic profile of phosphorylated ERK a fter stimulation w ith E GF in NRK fibroblasts by quantitative Western blotting. Cells were serum-starved and subsequently stimulated with 10 ngÆmL )1 EGF for various periods of time. W e observed a biphasic ERK-PP p rofile (Fig. 2 ). Upon EGF s timulation, the ERK-PP concentra- tion rose from a b ackground level t o a high peak concentration after about 4 min and then returned to the prestimulation level (after about 12 min) before increasing slightly again. This second in crease was f ollowed by a decrease to a relatively low level, after about 1 h of EGF stimulation. This ERK-PP profile suggests t he possibility of damped oscillatory behaviour, c onsistent with co mplex behaviour of the complex circuitry r egulating ERK-PP. We also determ ined the E RK phosphorylation d ynamics induced by PKC activation by addition of 100 n M PMA to serum-starved NRK cells. The profile resembled that induced by EGF s timulation (Fig. 2). After about 4 min, a peak c oncentration was reached, followed b y a rapid decline to a very low concentration that sustained for several hours. We refer t o this as a quas i steady-state, a s the ERK- PP concentration remains at approximately the same level for a relatively long period of time (compared to the time that was needed to attain this concentration). T he first peak concentration induced by PMA was always lower than t hat induced by EGF. EGF and PMA activate signalling to ERK synergistically In order to determine whether the different signal inputs t o ERK (via E GFR a nd via PKC) affect each other, we stimulated serum-starved NRK cells with both 10 ng ÆmL )1 EGF and 100 n M PMA and again determined the time profile of the ERK-PP concentration (Fig. 2). W e observed a biphasic pattern, with the first high peak being identical to that obtained during stimulation by EGF alone. After this first peak, the ERK-PP concentration reaches a quasi- steady state-concentration of 2–3· the sum of the concen- tration obtained after 1 h of stimulation with only EGF and the concentration obtained with PMA only. Appar- ently, EGF a nd PMA act synergistically on the quasi- steady state-phase of t he profile, but not on the initial peak. As the individual p eak shapes may have been lost during averaging of the c urves, we t itrated EGF a t a fixed time point of 60 min in order to measure accurately the synergistic effect. EGF concentration-dependency of synergistic activation To investigate the synergistic activation i n the second phase of the time profile further, w e measured t he ERK-PP concentration after 1 h stimulation with different EGF concentrations (ranging from 0 to 100 ngÆmL )1 ), both in the absence and presence of 100 n M PMA. The results (Fig. 3) show that the (quasi-stead y state) E RK-PP concentration depends on the EGF concentration used a nd reaches a maximal level. After stimulation with PMA alone, the Fig. 3. Stimulus–response curves of ERK-PP to EGF. ERK-PP was measured by quantitative Western blotting in cell lysates that were harvested after 1 h of EGF-stimulation with the indicated concentra- tions, in the absence ( n)orpresence(m) of PMA. T he data are averages of four independent experiments, the error bars represent the standard error of the mean. The drawn lines represent the curve fits that were ob tained using a Michaelis–Menten type equ ation (Eqn 1). The fitting parameters are shown in the table inset (th eir standard deviations are i ndicated betwe en brackets). [ERK-PP] basal :theERK- PP concentration without EGF present; [ERK-PP] max : the maximu m steady state ERK-PP concentration th at can be induced by EGF; K, EGF concentration needed to ob tain the half-maxim um ERK-PP concentration. In additio n, a representative image o f the immunob lots is depicted. 3908 J. J. Hornberg et al. (Eur. J. Biochem. 271) Ó FEBS 2004 ERK-PP concentr ation was similar to that obtained with high EGF concentrations, but when PMA and EGF were added simultaneously, i t reached a m uch higher level, again in an EGF concentration-dependent manner (Fig. 3). To draw a stimulus–response curve, we fitted these data points to the following equation (cf. the Michaelis–Menten equa- tion): ½ERK-PP steady state ¼½ERK-PP basal þ ½ERK-PP max ½EGF K þ½EGF ðEqn 1Þ [ERK-PP] basal is the ERK-PP concentration in s erum- starved cells befo re EGF addition , [ERK-PP] max is the maximum concentration of ERK-PP in the quasi-steady state-phase (after 1 h of stimulation) and K is the EGF concentration at which ERK-PP is 50% of its maximal level. Addition of PMA r esulted in a two- to threefold increase of the a dditive [ERK-PP] max , reflecting the synergistic activa- tion. Interestingly, K with respect to EGF w as also remark- ably higher when PMA was present. This indicates that, when PKC is activated, the signalling pathways to ERK still respond to EGF at higher growth factor concentrations while, without PMA, they are saturated at EG F concentra- tions above 1 ngÆmL )1 . The synergistic effect is only found in the concentration r ange above 1 ngÆmL )1 . Qualitative visualization of ERK-PP in fixed cells In addition to the quantitative measurement of ERK-PP by Western blotting, we qualitatively visualized ERK-PP in fixed NRK fibroblasts by immunocytochemistry. Cells were serum-starved a nd stimulated for 1 h with 100 ngÆmL )1 EGF, 100 n M PMA or both, after which ERK-PP was stained w ith a fluorescent labe l and visualized using laser scanning microscopy (Fig. 4A), as described in Experimen- tal procedures. The fluorescent signal per image was, after subtracting t he background, divided by the number of cells in the picture. T he ave rage signal intensities of four images showed that EGF or PMA stimulated cells were compar- able to untreated cells, whereas cells treated with both stimulators showed a considerably higher signal intensity (Fig. 4 B), which is consistent wi th the results discussed i n the previous section. Positive feedback circuit via cPLA 2 and PKC is not involved in EGF-mediated ERK phosphorylation in NRK cells nor in the synergistic activation According t o schemes available in t he literature, after E GF stimulation, PKC may be activated via P LC-c and, via the positive feedback loop, by cPLA 2 (Fig. 1). To monitor the effect of PKC on ERK phosphorylation, we stimulated Fig. 4. Qualitative visualization of synergistic ERK phosphorylation by EGF and PMA using immunofluorescence and quantification of the immunostaining. (A) ERK-PP was detected with a fluorescent label in fixed cells (for details see Experimental procedures) that were unstimulated (control) or s timulated f or 1 h with 100 ng ÆmL )1 EGF, 100 n M PMA or both and detected using a scanning laser microscope. Representative images of four independent experiments are shown. (B) Quantification o f the immunostainin g. The average fluorescent signal per cell in four independent e xperiments is depicted; the error bars represent t he s tandard e rror o f the mean . P lease not e that t he met hod app lie d prod uces a relatively high background, which hampers the quantific ation. The EGF and PMA costimulation produces a signal that sign ificantly emerges from this background, whereas stimulation with e ither EGF or PMA only d oes not. Ó FEBS 2004 Synergistic ERK activation by EGF and PMA (Eur. J. Biochem. 271) 3909 NRK cells with EGF, also in the presence of the PKC inhibitor bisindolylmaleimide I, and determined the ERK- PP concentration after 6 and after 60 min. In three independent experiments, we could not find an inhibitor- induced change in the E RK-PP concentration, neither in terms of the initial peak nor with respect to the quasi-steady state-phase (Fig. 5). Bisindolylmaleimide I completely abol- ished ERK phosphorylation after PMA s timulation, indi- cating that PM A does not have a PKC-independent effect on ERK-PP. The ERK-PP concentration induced by costimulation with both EGF and PMA was unaffected by PKC i nhibition in the early peak and w as reduced in the quasi-steady state-phase to a level comparab le to that after stimulation with EGF only. These r esults show that in our system, P KC had no s ignificant e ffect o n ERK phosphory- lation after E GF stimulation. This implies that t he positive feedback loop via cPLA 2 , which caused sustained ERK phosphorylation upon stimulation with p latelet-derived growth factor (PDGF) [8], was not activated by EGF stimulation in these cells. To investigate this further, we blocked cPLA 2 activity by the inhibitor ATK. The ERK-PP concentration again did not decline, even if the stimulation was carried out with EGF and PMA t ogether (Fig. 5). Similar results were obtained with 4-b romophenacyl bro- mide, which is a different, general PLA 2 inhibitor (results not shown). This shows that the positively regulating circuit is not involved in the s ynergistic ERK phosphorylation caused by EGF and PMA. Discussion The architecture of signal transduction networks i s often highly complex, due to the large number of participating protein co mplexes, cross interactions between pathways and the functioning of regulatory circuits. It is this complexity that makes th e understanding of cellular signalling a difficult task. For example, the features o f a whole network cannot be understood simply as the sum of features of its parts; the network as s uch may give rise t o s ystem o r Ôemergen tÕ properties [4,40]. To obtain reliable computer models that can calculate the outcome of signalling e vents, the interactions between signalling m odules nee d to be experimentally measured in a quantitative manner [3,7]. We have assessed the output of the EGF-activated MAPK pathway (ERK1/2) and its cross-talk with protein kinase C. We have measur ed the dynamic time profile of ERK phosphorylation a fter stimulation by EGF, and observed a biphasic pattern consisting of a first rapid peak and relatively l ow and a broad s econd peak developing into what we refer to as a quasi-steady state. The first peak has been described by o thers in m any cell types, but a biphasic pattern, that could be attributed to the existence of damped oscillations, seems to have escaped experimental resolution thus far. Sustained oscillations have been predicted in a theoretical study, in which they were explained by a combinatory effect of negative feedback and ultrasensitivity [22]. Both of these features have been demonstrated experimentally. Negative feedback is constituted by the phosphorylation of Sos by ERK, which c auses Sos to dissociate from the growth factor receptor complex. Thus, as the local Sos concentration at the inner s urface of the membrane decreases, Ras activation is impaired as well as subsequent activation of downstream signalling m olecules such as ERK [41]. Ultrasensitivity has been demonstrated in oocyte extracts, showing a steep stimulus/response curve for ERK-PP as a function of activated Raf (Hill coefficient: 4–5), which led to the suggestion that the pathway was equipped to filter out noise a nd behave as an Ôon/off-switchÕ [42]. The biphasic pattern we o bserve h ere may reflect sustained oscillations that are i nitially synchronized in all cells in the culture but that become desynchronized over time. Alternatively, damping may be c aused by MAPK phosphatases (MKPs) that are up-regulated within % 30 min after initial MAPK s ignalling [43]. Independent Fig. 5. The positive feedback loop via cPLA 2 and PKC plays no significant role in ERK phosphorylation. Cells were serum-starved for 3 d ays and t hen stimulated for either 6 or 60 min with EGF or PMA in the absence or presence of the PKC inhibitor Bisindolyl- maleimide I (5 l M ; 1 h preincubation) or the cPLA 2 inhibitor A TK (10 l M ;1hpreincu- bation). ERK-PP was measured i n the ce ll lysates by q uantitative Western blotting. The PKC inhibitor abolished PMA-induced ERK phosphorylation, whereas EGF-in duced ERK phosphorylation was unaffected. The ERK- PP concentration induced by EG F and PMA costimulation declined only in the quasi- steady st ate-phase. In hibition of cPLA 2 did not affect the peak o r the quasi-steady state- phase. S hown are the mean results o f three independent experiments, the erro r bars r ep- resent the s tandard error of the mean. 3910 J. J. Hornberg et al. (Eur. J. Biochem. 271) Ó FEBS 2004 of its precise mechanism, the transient oscillation is one aspect of complexity that is observed in the dynamics of this signal transduction chain. PKC was not found to be involved in the EGF-mediated ERK phosphorylation, bu t activation of PKC by PMA did result in a transient ERK-PP profile. Although phorbol ester receptors other than PKC have been reported [42,44], we have shown that these do not affect signalling to ERK in NRK cells, as PMA did not induce ERK phosphorylation if PKC had been inhibited by b isindolylmaleimide I. Simultaneous stimulatio n w ith PMA and E GF yielded a remarkably synergistic effect in the quasi-steady s tate phase of the time profile. In terestingly, in the presence of P MA, the ERK-PP quasi-steady state-concentration c ould still be elevated further by EGF concentrations above 1 ngÆmL )1 , whereas t his w as not the case in the absence of P MA (Fig. 3). Furthermore, the ratio betw een the ERK-PP concentration after stimulation with EGF and PMA together on the one hand, and the sum of the ERK-PP concentrations after stimulation with EGF and PMA separately on the other hand, was not constant, but increased with the EGF concentration, which confirmed synergistic activation. In osteoblastic cells, PMA did not affect EGF s ignalling [43,45], indicating that the response i s probably cell type-dependent. It h as been reported previ- ously that ERK1 was synergistically activa ted i n h amster fibroblasts after stimulation with serotonin and basic fibroblast growth f actor f or 120 min [46]. In platelets, pretreatment with thrombopoietin and stimulatio n with a-thrombin led to a 30% phosphorylation of ERK2 in the early phase (1–5 min), whereas thrombopoietin or a-thrombin alone activated E RK2, either, not at all or t o a very low degree, respectively [47]. In another study, simultaneous stimulation o f human embryon ic k idney cells with carbachol, which activates PKC via a G-protein coupled receptor, and EGF, did not exert additive effects on ERK activity a fter 10 min [48], which supports our finding that synergism i s not present in the initial peak (Fig. 2). We believe the latter is caused by s aturation of the phosphory- lation of all cellular ERK. Indeed, it h as been shown previously that in NRK c ells, this EGF concentration causes virtually all ERK to become doubly phosphorylated ([49], and our unpublished observation]. As to what molecular interactions underlie the observed synergism, we have obtained some indications. We have excluded the possibility that the synergism arises from a positive feedback loop via cPLA 2 andPKC.Infact,cPLA 2 inhibition did not alter the ERK-PP concentration upon EGF and PMA costimulation. Recently, this loop was found active after PDGF stimulation in NIH-3T3 cells, r esulting in prolonged ERK phosphorylation [8]. The synergism we find here might arise at a site where the two signal inputs converge, for instance at Raf. PKC phosphorylates Raf at Ser499, which was s uggested to cause Ser259 autophos- phorylation and activation [24]. Ser259 was also identified as a major Raf phosphorylation s ite upon growth factor stimulation [50]. The synergism might also originate up- stream of Sos, as ERK activation by P KC has been shown to depend both on Sos51 and on Ras-GTP-Raf complexes [52]. A different explanation could be that P KC has a n inhibitory effect on one of the down-regulating mechanism s of the pathway from EGF to ERK. One possibility could be that EGFR down-regulation is affected by PKC. EGFR phosphorylation on Thr654 by PKC has been shown to cause recycling of internalized EGFR to the cell s urface, instead of degradation [33], a nd EGF might then caus e a second, prolonged ERK phosphorylation phase. This could also explain why, during EGF and PMA costimulation, ERK-PP rises again after returning to a low level. On the other hand, the s ame phosphorylation of EGFR by PKC is known to cause decreased EGFR tyrosine kinase activity, its binding affinity for EGF [30–32] and mitogenic signalling [53]. In fact, we show in this study that activated PKC, although amplifying the E GF signal to ERK, also incr eases the EGF concentration that is needed for half-maximum ERK phosphorylation (Fig. 3). Clearly, con trol of PKC- mediated EGFR phosphorylation on signalling to ERK is distributed over these processes. Another, more speculative, candidate target could be a n MKP which is up-regulated about 30 min after ERK activation [43]. The activity of MKPs is known to be regulated by phosphorylation, both positively and negatively [54,55]. If PKC can inactivate an MKP that is up-regulated after stimulation with E GF, this may lead to sustained ERK phosphorylation. In conclusion, we have investigated the interaction between the MAPK and PKC s ignalling m odule s in a quantitative manner and found that they affect each other in multiple ways. EGF stimulation a nd PKC activation caused a synergistic a ctivation o f ERK in the quasi-steady state- phase. PKC here acts as a signal a mplifier for growth factor signalling. We found no evidence for the functioning of a positive feedback mechanism via cPLA 2 . The interactions between MAPK and PKC apparently serve to facilitate quasi-intelligent signal i ntegration, w hich m ay be necessary to assure that certain responses are induced only when more than one criterion needs to be met. DNA synthesis was shown previously to be synergistically activated by fibro- nectin and insulin [56]. As the duration of ERK signalling influences the repertoire of influenced target genes [ 16] and the cellular response [17,18], we h ypothesize that the synergistic ERK phosphorylation, which results in pro- longed signalling, has i mplications for the outcome of signalling. This may be of i mportance in the constitutive ERK activation often found in hu man tumour cells. Acknowledgements We thank W.P.H. de Boer and J.A. Ferreira for statistical advice. G.S.A.T. van Rossum is indebted for the kind gift of the cPLA 2 inhibitors ATK and 4-bromophenacyl bromide and advice on the manuscript. We are thankful to F.J. Bruggeman for stimulating discussions and advice on the manus cript. W e also thank K. Krab an d H. Dekker for excellent technical advice on curve fitting and immunofluorescence microscopy, r esp ectively. References 1. Palsson, B. (2000) The challenges of in silico biology. Na t. Bio- technol. 18, 1 147–1150. 2. Kitano, H. (2002) Systems biology: a brief ov erview. Science 295, 1662–1664. 3.Kholodenko,B.N.,Demin,O.V.,Moehren,G.&Hoek,J.B. (1999) Quantification o f short term signalling by t he epidermal growth f actor receptor. J. Biol. Chem. 274, 3016 9–30181. Ó FEBS 2004 Synergistic ERK activation by EGF and PMA (Eur. J. Biochem. 271) 3911 4. Bhalla, U.S. & Iyengar, R . (1999) Emergent properties of n et- works of b iological signalling pathways. Sc ience 283, 381–387. 5. Lauffenburger, D.A. (2000) Cell signalling pathways as control modules: complexity for simplicity? Proc. Natl Acad. Sci. USA 97, 5031–5033. 6.Bruggeman,F.,Westerhoff,H.,Hoek,J.&Kholodenko,B. (2002) Modular response analysis of cellular regulatory networks. J. Theor. Biol. 218, 507. 7. Schoeberl, B., Eichler-Jonsson, C., Gilles, E.D. & Muller, G. (2002) Com putational m odeling of the dynamic s of the MAP kinase cascade activated b y su rfac e and inte rnalized EG F receptors. Nat. Biotechnol. 20, 370–375. 8. Bhalla, U.S., Ram , P.T. & I yengar, R. (2002) MAP k inase phos- phatase as a locus of flexibility in a mitogen-activated protein kinase sign alling network. Science 297, 1018–1023. 9. Lewis, T.S., Shapiro, P.S. & Ahn, N.G. (1998) Signal transduction through MAP kinase cascad es. Adv. Cancer Res . 74 , 49–139. 10. Cobb, M.H. (1999) MAP kinase pathways. Prog. Biophys. Mol. Biol. 71, 479– 500. 11. Buday, L. & Downward, J. (1993) Epidermal g rowth factor reg- ulates p21ras through the formation of a complex of recep tor, Grb2 adapter prote in, an d Sos nucleotide exchange factor. Cell 73 , 611–620. 12. Schlessinger, J. (1993) How receptor tyrosine kinases activate Ras. Trends Bio chem. Sci. 18, 273–275. 13. Pawson, T. & Scott, J.D. (1997) Signalling through scaffold, anchoring, and adaptor p r oteins. Science 278, 2075–2080. 14. Weinstein-Oppenheimer, C.R., Blalock, W.L., Steelman, L.S., Chang, F. & M cCubrey, J.A. (2000) The Raf signal transduc tion cascade as a t arget for c hemotherap eutic i nterven tion in growt h factor-responsive tumors. Pharm acol. Ther. 88, 229–279. 15. Treisman, R. (1996) Regulation of transcription by M AP kinase cascades. Curr.Opin.CellBiol. 8, 205–215. 16. Cook, S.J., Aziz, N. & McMahon, M. (1999) The repertoire of fos and jun proteins expressed during the G1 phase o f t he cell cycle is determined by the duration of mitogen-activated protein k inase activation. Mol . Cell Biol. 19, 3 30–341. 17. Marshall, C.J. (1995) Specificity of receptor tyrosine kinase sig- nalling: transient versus s ustained extracellular signal-regulated kinase activation. Cell 80, 179–185. 18. Tombes, R.M., Auer, K.L., Mikkelsen, R., Valerie, K., Wymann, M.P., Marshall, C.J., McMahon, M. & Dent, P. (1998) The mitogen-activated prote in (MAP) kinase cascade can either sti- mulate or inhibit DNA synthesis in primary cultures of rat hepatocytes depending upon whethe r its activatio n is acute/p hasic or chronic. Biochem. J. 330, 1451–1460. 19. Page ` s, G., Lenormand, P., L’Allemain, G., Chambar d , J.C., Meloche, S. & Pouyssegur, J. (199 3) M itogen-activ ated protein kinases p42mapk and p44mapk are required for fibroblast pro- liferation. Proc. N atl Acad. Sc i. USA 90, 8 319–8323. 20. Hoshino, R., Chatani, Y., Yamori, T., Tsuruo, T., Oka, H., Yoshida, O., Shimada, Y., Ari-i. S., Wada, H., Fujimoto, J. & Kohno, M. (1999) Constitutive activation of the 41-/43-kDa mitogen-activated protein k inase s ignalling p athway in human tumors. Oncogene 18, 813–822. 21. Bos, J.L. (1989) ras oncogenes in human cancer: a review. Cancer Res. 49, 4 682–4689. 22. Kholodenko, B.N. (2000) Ne gative feedback and u ltrasensitivity can bring about oscillations in the mitogen-activated protein kinase cascades. Eur. J. Biochem. 267, 1583–1588. 23. Livneh, E. & Fishman, D.D . ( 1997) Linking protein kinase C to cell-cycle control. Eur. J. Biochem. 248 , 1–9. 24. Kolch, W., Heidecker, G., Koc hs, G., H ummel, R., V ahid i, H., Mischak, H., Finkenzeller, G., Marme, D. & Rapp, U.R. (1993) Protein kinase C alpha activates RAF-1 by direct phosphoryla- tion. Nature 364, 249–252. 25. Carroll, M.P. & May, W.S. (1994) P rotein kinase C-mediated serine phosphorylation directly activates R af- 1 in murine h ema- topoietic cells. J. Biol. Chem. 269, 124 9–1256. 26. Cai,H.,Smola,U.,Wixler,V.,Eisenmann-Tappe,I.,Diaz-Meco, M.T., Moscat, J., Rapp, U. & Cooper, G.M. (1997) Role of diacylglycerol-regulated protein kinase C isotypes in growth factor activation of the Raf-1 protein k inase. Mol. Cell. Biol. 17, 732–741. 27. Lin, L.L., W artmann, M., Lin, A.Y., Knopf, J.L., Seth, A. & Davis, R.J. (19 93) c PLA2 is phosphorylated and activated by MAP kinase. Cel l 72, 269 –278. 28. Nishizuka, Y . (1995) Protein kinase C and lipid signalling for sustained cellular responses. FASE B J. 9, 484–496. 29. Hunter, T ., Ling, N. & Cooper, J .A. (1984) Protein kinase C phosphorylation of the EGF receptor at a threonine residue close to the cyto plasmic face of the plasma m embrane. Nature 311, 480–483. 30. Cochet, C ., Gill, G.N., M eisenhelder, J., Cooper, J.A. & Hunter, T. (1984) C-kinase ph osphorylates the epidermal growth factor receptor and reduce s its epiderm al grow th factor-stimulated tyrosine protein kinase a ctivity. J. Bio l. Chem. 259, 2553–2558. 31. Downward, J., Waterfield, M.D. & Parker, P.J. (1 985) Auto- phosphorylation an d prote in k inase C phosphorylation of the epidermal growt h f actor r ecep tor. Effe ct o n t yro sine k inase activity and ligand b inding affi nity. J. Biol. Chem. 260, 14538– 14546. 32. Lee, L.S. & Weinstein, I.B. (1978) Tumor-promoting phorbol esters in hibit binding of e pidermal growth factor to cellular receptors. Science 202, 313–315. 33.Bao,J.,Alroy,I.,Waterman,H., Schejter, E.D., Brodie, C., Gruenberg, J. & Yarden, Y. (2000) Threonine phosphorylation diverts internalized epidermal growth factor receptors from a degradative pathway to the recycling en dosome. J. Biol. Ch em. 275, 26178–26186. 34. Toullec, D., Pianetti, P., Coste, H., Bellevergue, P ., Grand-P erret, T., Ajakane, M., Baudet, V., Boissin, P., Boursier, E., Loriolle, F., Duhamel, L., Charon, D. & Kirilovsky, J. (1991) The bisindo- lylmaleimide GF 109203X is a potent and selective inhibitor of protein kinase C. J. Biol. Chem. 266, 15771–15781. 35. Street, I.P., Lin, H .K., Laliberte, F., Ghomashchi, F ., Wang, Z., Perrier, H., Tremblay, N.M., Huang, Z., Weech, P.K. & Gelb, M.H. (1993) Slow- and tight-binding in hibito rs of the 85-kDa human phospholipase A2. Biochem ist ry 32, 5935–5940. 36. Chang, T.M., Chang, C.H., Wagner, D .R. & Chey, W.Y. ( 1999) Porcine pancreatic phosp holipase A2 stimulates secretin release from secretin-producing cells. J. Biol. Chem. 274 , 10758–10764. 37. Draper, N.H. & Smith, H. (1981) Applied Regression Analysis. Wiley, New Y ork. 38. Weisberg, S. (1980) Applied Linear Regression. Wiley, New York. 39. Jorgensen, B. (1993) The Theory of Linear Models. Chapman & Hall, New Y ork. 40. Hellingwerf, K .J., Postma, P.W., Tommassen, J. & Westerhoff, H.V. (1995) Signal transd uct ion in bacteria: phospho-neural net- work(s) in E sch erichia coli? FEMS Microbiol Rev. 16, 309–321. 41. Langlois, W.J., Sasaoka, T ., Saltiel, A.R. & Olefsky, J.M. (1995) Negative feedback regulation an d desensitization of insulin- and epidermal growth factor-stimulated p21ras activation. J. Biol. Chem. 270, 2532 0–25323. 42. Huang, C.Y. & Ferrell, J.E. Jr (1996) Ultrasensitivity in the mitogen-activated protein kinase cascade. Proc. Natl Acad. Sci. USA 93, 1 0078–10083. 43. Keyse, S.M. (2000) Protein phosphatases and the regulation of mitogen-activated protein kinase signalling. Curr. Opin. Cell Biol. 12, 186–192. 44. Ron, D. & Kazanietz, M .G. (1999) N ew insights i nto the re gula- tion of pro tein kinase C and novel phorbol ester re ceptors. FASE B J. 13, 1658–1676. 3912 J. J. Hornberg et al. (Eur. J. Biochem. 271) Ó FEBS 2004 45. Zhang, W., Dziak, R.M. & Aletta, J.M. (1995) EGF-mediated phosphorylation of extracellular signal-regulated kinases in osteoblastic cells. J. Cell Physiol. 162 , 348–358. 46. Meloche, S., Seuwen, K., Pages, G. & Pouyssegur, J. (1992) Biphasic and s ynergistic activation of p44mapk (ERK1) by growth factors: correlation between late phase activation and mitogenicity. Mol. Endoc rinol. 6, 845–854. 47. van Willigen, G., Gorter, G. & Akkerman, J.W. (2000) Throm- bopoietin increases platelet sensitivity to alpha-thrombin via activation of the ERK2-cPLA2 pathway. Thromb. Haemost. 83, 610–616. 48. Slack, B.E. (2000) The M3 muscarinic acetylcholine recep tor is coupled to mito gen-ac tivated protein kinase via protein kinase C and epidermal growth factor receptor kinase. Biochem. J. 348, 381–387. 49. Lahaye, D.H., Camps, M.G., Erp, P.E., Peters, P.H. & Zoelen, E.J. (1998) Epi dermal growth factor ( EGF) receptor density controls mitogenic activation of normal rat kidney (NRK) cells by EGF. J. Cell Physiol. 174, 9–17. 50. Morrison, P., Saltiel, A.R. & Rosner, M.R. (1996) Role of mito- gen-activated protein kinase kinase in regulation of the epidermal growth factor receptor by prote in kinase C. J Biol. Chem. 271, 12891–12896. 51. El-Shemerly, M.Y., Besser, D., Nagasawa, M. & N agamine, Y. (1997) 12-O-Tetradecanoylphorbol-13-acetate activates the Ras/ extracellular signal-regulated kinase (ERK) signalling pathway upstream of SOS involving serine phosphorylation of Shc in NIH3T3 cells. J. Biol. Chem. 272, 30599–30602. 52. Marais, R., Light, Y., Mason, C., Paterson, H., Olson, M.F. & Marshall, C.J. (1998) Requirement of Ras-GTP-Raf complexes for activation of Raf-1 by protein kinase C. Science 280 , 109–112. 53. Bowen, S., Stanley, K., Selva, E. & Davis, R.J. ( 1991) Constitutive phosphorylatio n of t he ep idermal g rowth f actor re - ceptor blocks mitogenic signal transduction. J . Biol. Che m. 266, 1162–1169. 54. Brondello, J.M., Pou yssegu r, J. & McKenzie, F.R. (1999 ) Reduced MAP kinase p hosphatase-1 d egradation after p42/ p44MAPK-dependent phosphorylation. Science 28 6, 2514–2517. 55. Saxena, M., Williams, S., Tasken, K. & Mustelin, T. (1999) Crosstalk between cAMP-dependent kinase and MAP kinase through a protein tyrosine phosphatase. Nat. Cell Biol. 1, 305–311. 56. Asthagiri, A.R., Reinhart, C.A., Horwitz, A.F. & Lauffenburger, D.A. (2000) The role of transient ERK2 signals in fibronectin- a nd insulin-mediated DNA s ynthesis. J. Cell Sci. 113 , 4499–4510. 57. Carpenter, G. (2000) The EGF receptor: a nexus for trafficking and signalling. Bioessay s 22, 697–707. Ó FEBS 2004 Synergistic ERK activation by EGF and PMA (Eur. J. Biochem. 271) 3913 . Synergistic activation of signalling to extracellular signal-regulated kinases 1 and 2 by epidermal growth factor and 4b-phorbol 12 -myristate 13 -acetate Jorrit. ttp://www.bio.vu.nl/vakgroepen/mcp/ (Received 2 2 June 20 04, accepted 6 August 20 04) Eur. J. Biochem. 2 71, 3905–3 913 (20 04) Ó FEBS 20 04 doi :10 .11 11/ j .14 32 -10 33 .20 04.04 327 .x phosphorylation and activation.