MEK1 Mediates a Positive Feedback on Raf-1 Activity Independently of Ras and Src

MEK1 mediates a positive feedback on Raf-1 activity independently of Ras and Src

Sven Zimmermann, Christian Rommel, Algirdas Ziogas, Josip Lovric1, Karin Moelling and Gerald Radziwill

Institute of Medical Virology, University of Zurich, Gloriastrasse 30, CH-8028 Zurich, Switzerland

Growth factor stimulated receptor tyrosine kinases directly interacts with two binding sites located within activate a protein kinase cascade via the serine/threonine CR1 of Raf-1 (Brtva et al., 1995) and thereby recruits protein kinase Raf-1. Direct upstream activators of Raf- the inactive cytoplasmic Raf-1 to the plasma mem- 1 are Ras and Src. This study shows that MEK1, the brane (Leevers et al., 1994; Stokoe et al., 1994). There direct downstream eector of Raf-1, can also stimulate Raf-1 gets fully activated by additional events such as Raf-1 kinase activity by a positive feedback loop. phosphorylation and interaction with lipids (Marais et Activated MEK1 mediates hyperphosphorylation of the al., 1995; Dent et al., 1995). Moreover, Ras amino terminal regulatory as well as of the carboxy independent mechanisms and pathways that can terminal catalytic domain of Raf-1. The hyperpho- regulate Raf activity have also been described (Dent sphorylation of Raf-1 correlates with a change in the et al., 1995; Michaud et al., 1995). The 14-3-3 family of tryptic phosphopeptide pattern only at the carboxy proteins also appears to play a role in regulation of the terminus of Raf-1 and an increase in Raf-1 kinase Raf-1 activity (Aitken, 1995; Suen et al., 1995). activity. MEK1-mediated Raf-1 activation is inhibited by Activated Ras interferes with the binding of 14-3-3 to co-expression of the MAPK speci®c phosphatase MKP-1 the N-terminus but not to the C-terminus of Raf-1 indicating that the MEK1 eect is exerted through a (Rommel et al., 1996). Mutations in Raf-1 which MAPK dependent pathway. Stimulation of Raf-1 prevent binding of 14-3-3 to the N-terminus lead to an activity by MEK1 is independent of Ras, Src and increase of the Raf kinase activity in vitro as well as in tyrosine phosphorylation of Raf-1. MEK1 can however vivo as shown in a dierentiation system, the synergize with Ras and leads to further increase of the development of the compound eye of Drosophila Raf-1 kinase activity. Thus, MEK1 can mediate melanogaster (Rommel et al., 1997). Raf-1 derived activation of Raf-1 by a novel positive feedback peptides that prevent binding of activated Ras or 14-3- mechanism which allows fast signal ampli®cation and 3 to Raf-1 can inhibit activation of Raf-1 and MAPK could prolong activation of Raf-1. (Muslin et al., 1996; Radziwill et al., 1996). Recently, dimerisation of Raf-1 has been shown to be involved in Keywords: Feedback loop; mitogen-activated protein activation of Raf-1 kinase activity (Farrar et al.,1996; kinase cascade; MEK1; Raf-1 activation Luo et al., 1996). Activated Raf-1 binds to and activates the dual speci®city protein kinase MEK1 (MAPK/ERK kinase). In turn, MEK1 stimulates the extracellular signal-regulated kinases, ERK1 and Introduction ERK2, which are members of the MAPK family (Kyriakis et al., 1992). ERKs regulate the activity of Raf-1 is a serine/threonine protein kinase (Moelling et cytoplasmic targets such as the ribosomal protein S6 al., 1984) which plays a central role in signal transduction kinase and nuclear targets such as the transcripion pathways connecting the upstream receptor and non factors c-Jun, c-Myc and Elk-1 (reviewed in Seger and receptor kinases to downstream protein kinases Krebs, 1995). Recently, we have shown that Raf-1 can (reviewed in Daum et al., 1994). Growth factor phosphorylate c-Jun in vitro and can physically interact stimulated receptor tyrosine kinases (RTK) activate the with c-Jun in avian cells indicating a bypass of the small GTP binding protein Ras via the adaptor molecule MAPK dependent Jun phosphorylation (Radziwill et Grb2 and the guanine nucleotide exchange factor mSos. al., 1995). Raf-1 can also bind and phosphorylate IkB, This is followed by the sequential stimulation of several the inhibitor of NFkB, further indicating that Raf-1 cytoplasmic protein kinases headed by Raf-1 and can be directly involved in regulation of transcription collectively known as the mitogen-activated protein (Li and Sedivy, 1993). kinase (MAPK) signaling cascade. Besides the association of Raf-1 with proteins such Raf-1 is one member of the Raf family (Raf-1, B- as Ras and 14-3-3 also phosphorylation of Raf-1 is Raf, A-Raf) characterized by three conserved regions involved in regulation of Raf-1 activity. Growth factor CR1, CR2 and CR3 (reviewed in Daum et al., 1994; stimulation of cells results in an increase of Raf-1 see Figure 1a). Several proteins have been described phosphorylation mainly on serine residues (e.g. Ser259) that physically interact with Raf-1 and function as and stimulation of Raf-1 kinase activity (Morrison et regulators or substrates for Raf-1. Activated Ras al., 1993). It has been shown that protein kinase C can activate Raf-1 by phosphorylation of Ser499 (Kolch et al., 1993) whereas phosphorylation of Raf-1 by protein Correspondence: G Radziwill 1Present address: GSF Klinische Molekularbiologue, Marchioninistr, kinase A at Ser43 and Ser621 decreases Raf-1 kinase 25, 81377 MuÈ nchen, Germany. activity (Haefner et al., 1994; Mischak et al., 1996). In Received 7 February 1997; revised 2 June 1997; accepted 3 June 1997 addition, stimulation of Raf-1 activity by cytokines MEK1-mediated Raf-1 activation S Zimmermann et al 1504 involves phosphorylation on tyrosine residues (Turner Src mutant protein SrcF527 (Figure 1b, lanes 3 and 6). et al., 1991). It has been shown that Src and Src-like These data indicate that the Raf-1 substrate MEK can kinases can bind to Raf-1 and activate it by stimulate hyperphosphorylation of Raf-1. To analyse phosphorylation of Tyr340 and Tyr341 (Cleghon and this eect in more detail the N-terminus (RafNT) and Morrison, 1994). Moreover, Ras and Src can synergize C-terminus (RafCT) of Raf-1 were expressed sepa- in phosphorylation and activation of Raf-1 (Williams rately. Co-expression of RafNT and MEKEE217/221 led et al., 1992; Marais et al., 1995; Jelinek et al., 1996). to a signi®cant hyperphosphorylation of RafNT Thus, phosphorylation is an important regulation (Figure 1c, compare lane 1 and 3). The RasV12 mechanism for Raf-1 kinase activity. Moreover, mediated phosphorylation of RafNT cannot be extracellular signals are not only mediated by linear further increased by co-expression of MEKEE217/221 pathways from the plasma membrane to the cytoplasm (Figure 1c, lanes 2 and 4). RafNT phosphorylation and nucleus but by a complex network with cross-talks was not in¯uenced by co-expression of MEKwt and feedback mechanisms. For example, ERKs can (compare lane 1 and 5) indicating that the presence also phosphorylate upstream elements such as mSos of activated MEK is essential for inducing the RafNT and thereby downregulate the Ras activity (Cherniack hyperphosphorylation. In contrast, MEKwt as well as et al., 1995; Waters et al., 1995). MEKEE217/221 were able to trigger hyperphosphorylation In the present study the interaction between Raf-1 of RafCT and thereby reducing the electrophoretic and its substrate MEK1 was analysed by transient mobility because in this case MEKwt can be activated expression studies in dierent cell lines. We show that by RafCT (Figure 1d, lanes 3 and 4). In order to test the downstream eector MEK1 can mediate hyper- whether MEK1 stimulated hyperphosphorylation of phosphorylation of its upstream activator Raf-1 which Raf-1 correlates with new phosphorylation sites tryptic corresponds to an increase in the Raf-1 kinase phosphopeptide analyses were performed (Figure 2). activity. MEK1-mediated activation of Raf-1 corre- MEK1-mediated phosphorylation of RafCT resulted lates with a change in the phosphopeptide pattern of in one additional tryptic phosphopeptide and stronger the C-terminus of Raf-1. MEK1 stimulated Raf-1 phosphorylation of another one (Figure 2b, marked by activation can be inhibited by co-expression of arrowheads; peptides 2 and 3 in the scheme). This is phosphatase MKP-1 which dephosphorylates and shown by comparison with the phosphopeptide inactivates MAPKs. This indicates that MAPKs are pattern obtained by transfection of RafCT alone involved in the MEK1-mediated Raf-1 activation. The (Figure 2a). The major spot can be attributed to MEK1 triggered Raf-1 stimulation does not involve Ser621 (Morrison et al., 1993; Mischak et al., 1996) the Ras signalling pathway nor does it depend on Src (Figure 2, peptide 1 in the scheme). To analyse or tyrosine phosphorylation of Raf-1. MEK1 can whether the change in the phosphorylation pattern synergize with Ras on Raf-1 activation similar to the induced by MEK1 depends on autophosphorylation of known synergism of Src and Ras. Thus, a positive RafCT or another kinase, the kinase-defective mutant feedback loop from activated MEK1 to Raf-1 allows RafCTE375, in which lysine 375 is changed to a fast signal ampli®cation in addition to the activation glutamate, was tested. Comparison of the tryptic of Raf-1 by Ras and Src. phosphopeptide patterns of RafCTE375 obtained from cells cotransfected without or with MEKEE217/221 revealed that also under these conditions activated MEK can induce phosphorylation of peptide 2 and Results stronger phosphorylation of peptide 3 (Figure 2c and d). This demonstrates that the MEK1-induced MEK1 mediates hyperphosphorylation of Raf-1 proteins phosphorylation of RafCT at peptide 2 and increased In order to study the interaction between Raf-1 and phosphorylation of peptide 3 do not depend on its substrate MEK1 and to investigate a possible autophosphorylation indicating phosphorylation of feedback mechanism from MEK1 to Raf-1 , we used RafCT by another kinase. Moreover, stimulation of the transient transfection system in 293 cells. FLAG- cells with EGF also induced phosphorylation of epitope tagged Raf-1 was co-transfected with MEK1 RafCT at peptide 2, suggesting that this site is also wild-type (MEKwt) and constitutively activated MEK normally targeted during growth factor stimultion (MEKEE217/221) in which the activating phosphorylation (data not shown). In addition, tryptic phosphopeptide sites Ser217 and Ser221 were changed to the negatively maps of the N-terminal fragment of Raf-1 expressed charged residue glutamate (Alessi et al., 1994). Cells alone or co-expressed with MEKEE217/221 were com- were metabolically labelled with 32P-orthophosphate pared. In this case the hyperphosphorylation does not and Raf-1 proteins were analysed by immunoprecipi- result in novel tryptic phosphopeptides (data not tation using an antibody directed against the FLAG- shown). epitope. Co-expression of Raf-1 with MEKwt as well as Taken together, these data indicate that MEK1 MEKEE217/221 increased the phosphorylation of Raf-1 mediates hyperphosphorylation of the N- and C- which correlates with a decrease of the electrophoretic terminus of Raf-1. The appearance of a new mobility (Figure 1b, compare lane 1 with lanes 4 and phosphorylation site and the intensi®ed phosphoryla- 5). Co-transfection of Raf-1 with activated RasV12 tion of another one is independent of autopho- results in a stronger increase in phosphorylation sphorylation and must be attributed to another (lane 2). A triple transfection with plasmids expres- kinase phosphorylating the RafCT. This suggests that sing Raf-1, constitutively activated RasV12 and MEK1 could be involved in the regulation of the Raf-1 MEKEE217/221 resulted in an additional increase in Raf- activity by inducing hyperphosphorylation at the N- 1 hyperphosphorylation similar to that obtained by terminal regulatory domain as well as the C-terminal co-expression of Raf-1 with RasV12 and the activated catalytic domain. MEK1-mediated Raf-1 activation S Zimmermann et al 1505

b a

EE217/221 F527

+Src +MEK Wt EE217/221 V12 V12 V12 Raf+Ras Raf Raf+Ras Raf+MEK Raf+Ras M Raf+MEK

Raf-1 69 —

1 2 3 4 5 6

d EE217/221 +MEK Wt EE217/221 EE21/221 Wt V12 V12

M RafCT RafCT+SrcF527 RafCT+MEK RafCT+MEK RafNT+MEK M RafNT RafNT+Ras RafNT+Ras RafNT+MEK

46 —

RafCT RafNT 30 — 46 —

1 2 3 4 5 1 2 3 4

Figure 1 MEK1 mediates hyperphosphorylation of Raf-1 proteins. (a) Scheme of the domain structure of Raf-1. The conserved regions CR1, CR2, CR3 and the mutated amino acid residues analysed in this study are indicated. (b), (c) and (d) 293 cells were transfected with plasmids coding for FLAG-Raf full-length (b), FLAG-RafNT (c) and FLAG-RafCT (d) either alone or co- transfected with MEKwt, MEKEE217/221, RasV12 and SrcF527 as indicated on top of each lane. The cells were labelled with 32P- orthophosphate and the FLAG-epitope tagged Raf-1 proteins were immunoprecipitated with an aFLAG antibody, subjected to SDS ± PAGE and visualized by a PhosphorImager

because of the truncation of the negatively regulating MEK1 mediated hyperphosphorylation of Raf-1 N-terminal domain. However, it has been shown that correlates with increased Raf-1 kinase activity RafCT can still be regulated and overexpression of In most cases hyperphosphorylation of Raf-1 correlates activated Src eciently stimulates its Raf kinase with an increase in its kinase activity. To study whether activity (Haefner et al., 1994; Marais et al., 1995). As co-expression of MEK1 which stimulate hyperpho- shown in Figure 3a RafCT kinase activity could be sphorylation of Raf-1 can in¯uence Raf-1 kinase increased by MEKwt as well as MEKEE217/221 (14- and activity immunopuri®ed Raf-1 was analysed in an in 11-fold, respectively), whereas activated SrcF527 resulted vitro kinase assay with the kinase inactive GST- in a 12.5-fold induction of RafCT kinase activity. MEKK97R protein as substrate. Similar amounts of Treatment of the Raf-1 immunoprecipitates from Raf-1 proteins were immunoprecipitated as controlled MEK1 overexpressing cells with the phosphatase by immunoblotting with aFlag antibody (data not PP2A prior to in vitro kinase assay resulted in a loss shown). of Raf kinase activity (Figure 3a). Thus, MEK1 First, RafCT was tested which by itself has an mediated activation of RafCT depends on phosphor- increased kinase activity compared to Raf-1 wild-type ylation of Raf-1. MEKwt and MEKEE217/221 also MEK1-mediated Raf-1 activation S Zimmermann et al 1506 a b a

RafCT RafCT MEKEE217/221

RafCTE375 RafCTE375 MEKEE217/221

Figure 3 MEK1-mediated hyperphosphorylation of Raf-1 correlates with an increase of Raf-1 kinase activity. Co- Figure 2 MEK1-mediated hyperphosphorylation of RafCT expression of FLAG-RafCT (a) and FLAG-Raf full length (b) results in change of the phosphorylation pattern. 293 cells with MEKwt, MEKEE217/221, RasV12 or SrcF527 was performed as expressing Flag-RafCT or the kinase-defective mutant Flag- described in Materials and methods. Additionally, serum starved E375 RafCT either alone (a and c) or together with activated cells overexpressing FLAG-Raf full length were stimulated with 32 MEKEE217/221 (b and d) were metabolically labelled with P- EGF (100 ng/ml, 7 min). The kinase activity of immunoprecipi- orthophosphate. RafCT proteins were immunoprecipitated, tated FLAG-RafCT (a) and FLAG-Raf full length (b) was resolved by SDS ± PAGE and subjected to two-dimensional measured by in vitro kinase assays using recombinant kinase- tryptic phosphopeptide analysis. Phosphopeptides obtained after inactive GST-MEK as a substrate. The phosphorylated reaction tryptic digestion were spotted on thin layer chromatography products were separated by SDS ± PAGE and analysed using a plates and separated by electrophoresis (horizontal direction) and PhosphorImager. To show that activation of Raf-1 depends on subsequent chromatography (vertical direction). The origin is phosphorylation RafCT immunocomplexes were treated with the indicated by a dot. Identical amounts of radioactivity were phosphatase PP2A prior to in vitro kinase assay. The results are loaded. In the scheme peptide 1 represents Ser621. Peptides 2 and shown as the relative kinase activity compared to the basal kinase EE217/ 3 (dark spots) are in¯uenced by cotransfection of MEK activity of FLAG-RafCT or FLAG-Raf-1. Indicated are the 221 and marked by arrowheads in the panels b and d. means and standard deviations from at least three independent experiments

stimulated the kinase activity of the wild-type Raf-1 protein but to less extent (two- and sixfold, respec- MEK1 stimulated Raf-1 kinase activity is independent of tively; Figure 3b). This is comparable to the Ras stimulation of serum starved cells by EGF (fourfold) but is much less than the stimulation observed by co- Activation of Raf-1 in the receptor tyrosine kinase expression of activated RasV12 (25-fold). Interestingly, pathway depends on recruitment of Raf-1 to the the MEKwt and MEKEE217/221 induced stimulation plasma membrane by activated Ras. Therefore, synergized with RasV12 and resulted in a maximal mutant proteins interfering with the Ras signalling activation of Raf-1 (47- and 44-fold) comparable to the pathway were tested for their in¯uence on the MEK1- eect seen by co-expression of Raf-1 with RasV12 and mediated Raf-1 activation. First, the mutant Raf-1 SrcF527 (50-fold). protein RafL89, which cannot interact with Ras, was These data provide evidence that MEK1 can stimulate tested (Figure 4). MEKEE217/221 stimulated the kinase Raf-1 kinase activity presumably by a positive feedback activity of Raf-1 wild-type as well as RafL89 to a similar mechanism. Activation of RafCT by MEK1 is similar to extent. In contrast, RasV12 induced Raf-1 activation was that of SrcF527 and co-expression of both activators strongly reduced in the case of RafL89, since this mutant together does not further increase RafCT kinase activity fails to physically interact with Ras (Fabian et al., (data not shown). On the other hand, MEK1-induced 1994; Niehof et al., 1994). Secondly, the eect of the Raf-1 kinase activity can be drastically increased by dominant negative mutant RasN17 was tested (Figure 4). RasV12 indicating a synergism between MEK1 and Ras MEKEE217/221-mediated stimulation of Raf-1 kinase which is comparable to the known cooperativity between activity was not inhibited by co-expression of Ras and Src on Raf-1 activation. RasN17, in contrast to EGF stimulated Raf-1 kinase MEK1-mediated Raf-1 activation S Zimmermann et al 1507 activity which was reduced by RasN17. Both experi- membrane fraction and from the cytoplasm. Taken ments together clearly demonstrate that the MEK- together, these data indicate that Src and phosphoryla- mediated Raf1 activation is independent of Ras tion of Raf-1 on tyrosine residues by other tyrosine activity. kinases are not involved in MEK-mediated hyperphosphorylation and stimulation of Raf-1 kinase activity. MEK1-induced RafCT kinase activity is independent of tyrosine phosphorylation MKP-1 inhibits MEK1 mediated Raf-1 activation Recently it has been shown that Src and Src-like MEK1 functions as an upstream activator of the kinases can phosphorylate two adjacent tyrosine MAPKs ERK1 and ERK2. The kinase activity of residues at the C-terminus of Raf-1 (residues 340 and ERK can be downregulated by the dual speci®city 341) and thereby can stimulate Raf-1 kinase activity phosphatase MKP-1 which dephosphorylates the (Fabian et al., 1993). This raised the question of threonine and tyrosine residue phosphorylated by whether MEK1-mediated activation of Raf-1 depends MEK1 (reviewed in Nebrada, 1994). To test whether on phosphorylation of Raf-1 on tyrosine residues. Therefore, the Raf-1 mutant RafFF340/341 in which the two known tyrosine phosphorylation sites were a replaced by phenylalanine was cotransfected with MEKEE217/221 (Figure 5a). Activated MEK1 can still activate RafFF340/341 while SrcF527 failed to activate this mutant. Substitution of the tyrosine residues 340 and 341 to phenylalanine reduced the Ras-dependent activation of Raf-1 to the level obtained by MEKEE217/221 indicating that in some way these two tyrosine residues are involved in Ras-dependent signalling. Moreover, coexpression of SrcM297,a dominant negative mutant of Src, did not interfere with the MEK-mediated activation of Raf-1 (Figure 5a). To test whether MEK1 can mediate phosphorylation of Raf-1 on tyrosine residues at all, phosphoamino acid analyses were performed (Figure 5b). 293 cells expressing RafCT were labelled metabolically with 32P- orthophosphate. Cells lysates were fractionated in a membrane (P100) and a cytoplasmic fraction (S100) and RafCT was immunoprecipitated from both b fractions. Phosphoamino acid analysis showed that RafCT co-expressed with MEKEE217/221 was phosphorylated mainly on serine and to some extent on EE217/221 EE217/221 threonine. In contrast, RafCT co-expressed with Src F527 F527 showed in addition phosphorylation on tyrosine. Same results were obtained for RafCT isolated from the RafCT+MEK RafCT+Src RafCT+MEK RafCT+Src P-Ser

P-Thr

p-Tyr

S100 P100 Figure 5 MEK1-induced activation of Raf-1 is independent of Src and tyrosine phosphorylation. (a) The kinase activity of Raf-1 and the mutant RafFF340/341 was tested after expressing these proteins alone or together with SrcF527, RasV12, MEKEE217/221 or the dominant negative SrcM297 as indicated. (b) Phosphoamino acid analysis. Cells were co-transfected with plasmids coding for Figure 4 MEK1-mediated activation of Raf-1 is independent of FLAG-RafCT and MEKEE217/221 or FLAG-RafCT and SrcF527 Ras. Cells were co-transfected with plasmids coding for FLAG- and metabolically labelled. Subsequently the cell lysate was Raf or the mutant FLAG-RafL89 alone or with RasV12, MEKEE217/ fractionated into a cytoplasmic (S100) and a membrane (P100) 221 and dominant negative RasN17 as indicated. The kinase activity fraction, for details see Materials and methods. FLAG-RafCT of Raf-1 and RafL89 was assayed as described in Figure 3. was immunoprecipitated from both fractions and used for Dominant negative RasN17 was able to reduce the EGF mediated phosphoamino acid analysis. The labelled phosphoamino acids (100 ng/ml, 7 min) Raf-1 activity but not the MEKEE217/221 were detected with a PhosphorImager. Phosphoamino acids dependent activity of Raf standards were stained with ninhydrin and are indicated by circles MEK1-mediated Raf-1 activation S Zimmermann et al 1508 MEK1-mediated stimulation of Raf-1 depends on kinase activity. Thus, the Raf-1 substrate MEK1 is ERK activity the eect of MKP-1 was analysed. involved in a positive feedback mechanism. This Expression of MKP-1 inhibits the MEKEE217/221 MEK1-triggered activation of Raf-1 is independent of mediated stimulation of Raf-1 to the level obtained the known upstream activators Ras and Src. The by expression of Raf-1 alone as can be seen by the downstream eectors of MEK1, ERK1 and ERK2, reduction of GST-MEK phosphorylation (Figure 6, must play a role since the phosphatase MKP-1, which upper panel). For control of MKP-1 activity the eect can inactivate MAPKs, also downregulates the MEK1 of MKP-1 on MAPKs was analysed by an immunoblot mediated Raf-1 activation. A model for the MEK1- speci®c for ERK2 (Figure 6, middle panel) and by a mediated positive feedback loop is shown in Figure 7. MAPK activity test in which myelin basic protein Recently, it has been shown that MEK1 and ERK (MBP) was used as a substrate for MAPKs (Figure 6, can also initiate negative feedback mechanisms which middle and lower panel). Co-expression of MEKEE217/221 downregulate Ras activity and subsequently Raf-1 and Raf-1 resulted in hyperphosphorylation of ERK2 activity (Cherniak et al., 1995; Waters et al., 1995). as evidenced by the appearance of a protein doublet in In this case, the guanine nucleotide exchange factor SDS ± PAGE and in phosphorylation of MBP in mSos is phosphorylated by ERK or by an ERK- cellular lysates. MKP-1 prevented ERK2 hyperpho- independent manner through MEK1 (Cherniak et al., sphorylation and also reduced phosphorylation of 1995; Ueki et al., 1994; Holt et al., 1996). Hyperpho- MBP indicating that ERK2 and other MBP phosphor- sphorylation of mSos results in the dissociation of the ylating kinases were inactivated. Thus, expression of Grb2-mSos complex that is formed after growth factor MKP-1 inhibits MEK1-mediated activation of ERK stimulation of cells and is essential for Ras activation. and Raf-1. This indicates that the MEK1-induced The positive feedback described here is independent of positive feedback on Raf-1 may act through ERK, a Ras activity. This is demonstrated by the mutant common substrate for MEK1 and MKP-1. proteins RasN17 and RafL89 which interfere with Ras signalling but do not aect MEK1-mediated Raf-1 activation. Moreover, transient expression of activated Discussion MEK1 does not result in a hyperphosphorylation of endogenous mSos (data not shown). Interestingly, the We demonstrate that MEK1 mediates hyperphosphor- MEK-triggered Raf-1 stimulation is not only indepen- ylation of Raf-1 and leads to stimulation of the Raf-1 dent of Ras but also of tyrosine phosphorylation. Tyrosine phosphorylation and activation of Raf-1 has been described for Src and Src-like kinases (Cleghon and Morrison, 1994) and is involved in stimulation of +MKP-1 EE 217/221 EE 217/221

Raf Raf+MEK Raf+MEK Raf-1 kinase activity

GST-MEK

ERK2 Immunoblot — ERK2-P — ERK2

MAP kinase activity

MBP

1 2 3 Figure 6 The MEK1-mediated increase in Raf-1 kinase activity is reverted by co-expression of MKP-1. Raf-1 was expressed alone (lane 1), together with MEKEE217/221 (lane 2) or with MEKEE217/221 Figure 7 Model of feedback mechanisms acting on Raf-1. and MKP-1 (lane 3). The Raf-1 kinase activity was measured by Pathways which have previously been described to be involved in vitro kinase assays with GST-MEK as a substrate (upper in the downregulation of Raf-1 kinase activity are indicated to the panel). MEKEE217/221-mediated Raf-1 activation (sixfold) was left. These negative feedback events emerge from either ERK or reduced by co-expression of MKP-1 (1.5-fold). Cell lysates were MEK1 and act via mSos and Ras. In this study, however, a either directly subjected to SDS ± PAGE and immunoblotted with positive feedback is described which is mediated by MEK1 and is an aErk-2 antibody (middle panel) or were assayed for MAP independent of Ras and Src. The role of ERK is proven by the kinase activity using MBP as a substrate (lower panel). MEK1- fact that the MAPK-speci®c phosphatase, MKP-1, can antagonize mediated MBP-phosphorylation (ninefold) was reduced by co- the MEK1-induced stimulation of Raf-1. Presumably an expression of MKP-1 (1.5-fold). additional unknown factor(s) (X) may play a role MEK1-mediated Raf-1 activation S Zimmermann et al 1509 cells with cytokines (Turner et al., 1991). However, as secreted factors lead to an autocrine loop involved in shown here a dominant negative Src mutant cannot MEK1-mediated Raf-1 activation. In order to rule out inhibit MEK1-mediated Raf-1 activation. Also the that E1A or E1B proteins, which are expressed in the Raf-1 mutant RafFF340/341, which can no longer be Adenovirus-transformed 293 cell-line used here, are activated by Src and Src-like kinases, can still be responsible for the MEK1-induced Raf-1 activation, activated by MEK1. In addition, hyperphosphorylation we performed similar experiments in COS cells. of RafCT stimulated by MEK1 does not result in Expression of activated MEK in COS cells also phosphorylation on tyrosine as shown by phosphoa- induced the positive feedback loop on Raf-1 as was mino acid analysis. Thus, the positive feedback on Raf- the case in 293 cells (data not shown). 1 is independent of Ras and Src whereas the negative In the positive feedback described here the ®rst step feedback acts through Ras on Raf-1 (see Figure 7). But is the activation of MEK1. This can either be achieved both, the positive and negative feedback mechanisms, by using the activated mutant MEKEE217/221 or by using depend on MEK1 and ERK. Therefore, the time RafCT or the mutant RafA259 (data not shown) both of course of the stimulus as well as dierent factors might which exhibit a moderate but constitutive kinase be important to determine which feedback loop is activity (Morrison et al., 1993). A weak activation activated. signal on MEK1 might be sucient for further MEK1 mediates hyperphosphorylation of the N- as activation of Raf-1 by the positive feedback loop well as the C-terminus of Raf-1 and hyperphosphor- which could prolong the duration of Raf-1 activitation. ylation of RafCT led to the appearance of a new The MEK1 stimulated Raf-1 kinase activity can be phosphopeptide and to stronger phosphorylation of further increased by Ras. This means that although another one. Since coexpression of activated MEK1 MEK1 can stimulate Raf-1 independently of Ras, it did not result in dierences between the tryptic can act synergistically with activated Ras. This is phosphopeptide pattern of RafCT and the kinase- similar to the known synergism of Ras and Src on Raf- defective RafCTE375, the autophosphorylation activity 1 activation. It is of interest that besides Raf-1 there of RafCT is not responsible for these changes. The also exist other activators for MEK1 such as Mos additional new phosphorylation site was also detectable (Posada et al., 1993) or MEKK (Lange-Carter et al., upon EGF stimulation indicating that it is normally 1993). Therefore, under certain conditions MEK1 targeted during growth factor stimulation. The might be activated through pathways which do not increased phosphorylation of one of the peptides involve Raf-1. These could still lead to stimulation of could be speci®c for the positive feedback induced by Raf-1 by the positive feedback mechanism described in MEK1. A MEK1-induced phosphorylation of the this study. We are currently trying to identify further serine residues 621 and 624 leading to a mobility shift factors essential for the MEK1-mediated positive of Raf-1 has recently been described by Ferrier et al. feedback loop. Our results point to a cross-talk and (1997), however, activation of Raf-1 by MEK1 was not an ampli®cation mechanism between various signalling addressed. pathways which are under control of dierent stimuli. Previously, it has been shown that ERK can phosphorylate Raf-1 in vitro (Anderson et al., 1991; Kyriakis et al., 1993) and in a cell-line overexpressing Materials and methods ERK the stimulation of hyperphosphorylation of Raf- 1 was enhanced (Ueki et al., 1994), whereby an Plasmids in¯uence on Raf-1 kinase activity was not detectable. The plasmids pcDNA3FLAG-Raf, pcDNA3FLAG- Recently, Alessandrini et al. showed that in some RafNT, pcDNA3FLAG-RafCT and pcDNA3FLAG- clonal cell lines stably expressing constitutively RafCTE375 have been described recently (Rommel et al., activated MEK the kinase activity of Raf-1 and 1996). pcDNA3FLAG-RafL89 (exchange of Arg89 to Leu) ERK can be increased (Alessandrini et al., 1996). and pcDNA3FLAG-RafFF340/341 (exchange of Tyr340/341 to The data presented here demonstrate that the MEK1 Phe) were generated using the QuickChange site-directed mediated Raf-1 activation is antagonized by expres- mutagenesis kit (Stratagene). The cDNAs coding for sion of the phosphatase MKP-1 indicating that ERK oncogenic RasV12 (exchange of Glu12 to Val) and for plays a role in the positive feedback loop. Therefore, dominant negative RasN17 (exchange of Ser17 to Asn) were we postulate that there must be an additional a gift from Dr A Wittinghofer (Feig et al., 1986; John et kinase(s) between ERK and Raf-1 which contribute al., 1988) and were subcloned into pCis2 (Rommel et al., 1996). cDNAs for activated SrcF527 (exchange of Tyr527 to to phosphorylation and activation of Raf-1. Because Phe) and for dominant negative SrcM297 (exchange of ERK can regulate transcription by phosphorylation of Lys297 to Met) were obtained from Drs T Hunter and M transcription factors such as Elk-1 and Myc, it might Broome (de Hertog et al 1994) and were subcloned into be possible that ERK induces an autocrine loop which pcDNA3 (Invitrogen). The rabbit wild-type MEK1 and the results in activation of Raf-1. In the MEK1-mediated constitutively active mutant MEKEE217/221 (exchange of positive feedback this autocrine mechanism would be Ser217/221 to Glu) inserted in the vector pBABEpuro independent of Ras, Src and tyrosine phosphorylation were kindly provided by Dr C J Marshall (Alessi et al., of Raf-1. Therefore, a possible autocrine loop would 1994) and subcloned into pcDNA3. The cDNA for the not involve growth factor and cytokine stimulated human MAPK phosphatase MKP-1 (CL100) was a gift pathways because these are dependent on Ras and Src from Dr S Keyse (Keyse and Emslie, 1992) and was subcloned into pcDNA3. or Src-like kinases, respectively. Incubation of Raf-1 transfected cells with medium collected from Raf-1 and MEK1 overexpressing cells did not increase Raf-1 Cell culture and transfection kinase activity, even with medium collected 24 h after Adenovirus 5 E1A/B transformed human embryonic transfection (data not shown). Thus, it is unlikely that kidney cells (293 cells) were grown at 378Cand10%CO2 MEK1-mediated Raf-1 activation S Zimmermann et al 1510 in Dulbecco's modi®ed Eagles medium (DMEM/GIBCO- In vitro kinase assay BRL). The medium was supplemented with 4mM gluta- Cells were starved for at least 8 h and in some cases mine, 100 units/ml penicillin, 100 mg/ml streptomycin and stimulated with EGF (Sigma) at a concentration of 100 ng/ 10% fetal calf serum (FCS). 293 cells were transiently ml for 7 min. Lysis of the cells and immunoprecipitation transfected at a con¯uency of 80% using the Lipofectamine were performed as described above. Except, after reagent transfection system (GIBCO-BRL) as described immunoprecipitation the beads were washed 3 times in (Rommel et al., 1996). The transfection medium was RIPA buer and two times in kinase buer (50 mM Tris- replaced after 8 h by DMEM containing 10 % FCS to HCl, pH 7.5, 50 mM NaCl, 7.5 mM MnCl ,2mM DTT, allow another 12 ± 14 h expression of the transfected 2 25 mM b-glycerophosphate). For in vitro dephosphoryla- plasmids. tion Raf-immunocomplexes were washed once in phosphatase dilution buer (20 mM MOPS,pH7.5,60mM 2- 32 Metabolic labelling with P-orthophosphate mercaptoethanol, 1 mM MgCl2, 0.1 mg/ml serum albumin) and incubated in 30 ml dilution buer with 0.5 Units of In vivo labelling of the cells was performed as described Protein Phosphatase Type-2A (PP2A) (UBI, Lake Placid). previously (Radziwill et al., 1995). Transfected cells were After 15 min at 308C PP2A treated immunocomplexes were rinsed once with phosphate-free medium and were then washed two times with kinase buer. To suppress the incubated for 2 h in phosphate-free DMEM supplemented autophosphorylation of Raf-1 the precipitates were with 4mM glutamine and 10% dialysed FCS. Subsequently, preincubated with 50 ml kinase buer containing 20 mM 32 the cells were labelled for 3 h with 0.3 mCi/ml P- ATP for 8 min at room temperature. In vitro kinase assays orthophosphate in the presence of 10% dialysed FCS. were performed by incubating the immunocomplexes in Afterwards, the plates were washed with 3 ml of ice-cold 30 ml kinase buer containing 10 mCi [g-32P]ATP (Amer- phosphate-buered saline solution (PBS) containing sham) and 1 mg of recombinant kinase-inactive GST- 1mMDTT, 1 mM PMSF, 200 KIE/ml Trasylol (Bayer), MEKK97A for 30 min at 308C. The reaction was stopped 40 mM b-glycerophosphate, 1 mM Na-orthovanadate, by adding 10 ml46SDS-sample buer and heating the 25 mM NaF, 1 mM benzamidine, 1 mg/ml pepstatin and samples at 958C for 5 min. After SDS ± PAGE the 1 mg/ml leupeptine and scraped in 0.8 ml PBS. The cells quanti®cation of radioactivity incorporated into GST- were harvested by centrifugation (8 min, 1500 r.p.m.) and MEKK97A was carried out with a PhosphorImager. The lysed in 0.75 ml RIPA buer (20 mM Tris-HCl, pH 7.5, puri®cation of the recombinant kinase-inactive GST- 150 mM NaCl, 0.1% SDS, 0.5% DOC, 1% Triton X-100, MEKK97A as a substrate for Raf proteins has been 2mM EDTA, 10% Glycerol and inhibitors as above). described elsewhere (Lovric' et al., 1996). Precleared lysates were incubated with aFLAG antibody (M2, IBI/ Kodak) and rotated for 2 h at 48C. The immunoprecipitates were washed four times in RIPA ERK2 immunoblot and MAP kinase activity assay buer and heated for 5 min in SDS sample buer. After Cells were transfected as described above and starved for SDS-polyacrylamide gel electrophoresis the proteins were 6 h. Prior to lysis, cells were washed with 2 ml ice-cold PBS analysed by autoradiography using a PhosphorImager. and resuspended in 300 ml MAP kinase buer (20 mM Tris-

HCl, pH 7.5, 1 mM EGTA, 10 mM MgCl2,2mM MnCl2, Two-dimensional tryptic phosphopeptide mapping and phos- 0.1% NP-40, 1 mM PMSF, 1 mM pepstatin, 200 KIE/ml phoamino acid analysis Trasylol). Cells were broken by passing them 10 times through a 25-gauge needle. The lysates were cleared by The FLAG-RafCT was immunoprecipitated from 32P- centrifugation and aliquots were resolved by SDS ± PAGE. orthophosphate labelled 293 cells, resolved by SDS ± After electrophoresis the proteins were blotted onto PAGE and subjected to two-dimensional tryptic phospho- nitrocellulose membranes. The ®lter was probed with a peptide analysis as described previously (Schramm et al., polyclonal aERK2 antibody (UBI) overnight at 48Cand 1994). For phosphoamino acid analysis the cells were the ERK2 proteins were visualised with the ECL detection labeled as above, washed and lysed in hypotonic lysis system (Amersham). For measurement of MAP kinase buer (HLP) (10 mM Tris-HCl, pH 7.5, 10 mM NaCl, activity cellular extracts containing 30 mg protein were 3mM MgCl2,0.2mM EDTA, 1 mM DTT, 1 mM PMSF, incubated with 20 mg myelin basic protein (Sigma), 10 mM 200 KIE/ml Trasylol (Bayer), 40 mM b-glycerophosphate, ATP and 2 mCi of [g-32P]ATP for 20 min at 308C. The 0.2 mM Na-orthovanadate, 25 mM NaF, 1 mM benzami- reaction was stopped with 46 SDS-sample buer, boiled dine and 1 mg/ml leupeptine). Cells were broken by 50 for 5 min and the samples were resolved by 15% SDS ± strokes with a Dounce homogenizer and centrifuged PAGE. MBP-phosphorylation was evaluated with a (15 min, 3000 r.p.m.) to pellet nuclei. The supernatant PhosphorImager. was ultracentrifuged at 100 000 g for 45 min to separate the lysate into a membrane fraction (P100) and a cytosolic fraction (S100). The pellet was extracted with modi®ed HLP, containing 1% NP-40 and 100 mM NaCl, whereas Acknowledgements the cytosolic fraction was adjusted to 1% NP-40 and We would like to thank Chris J Marshall, Stephen M 100 mM NaCl. Both fractions were precleared with Keyse, Alfred Wittinghofer, and Tony Hunter for cDNAs sepharose beads and subsequently immunoprecipitated. of MEK1, MKP-1 (CL100), ras and src, respectively. The The precipitates were subjected to SDS ± PAGE. FLAG- technical help of Yvonne Schae¯i is gratefully acknowl- RafCT was excised from the gel and subsequently treated edged. We thank Jovan Pavlovic for critical reading of the according to the protocol of Boyle et al., 1991. The manuscript. This work was supported by the Swiss radiolabeled tryptic phosphopeptides and phosphoamino National Fond, the ZuÈ richer Krebsliga and the Hartmann acids were visualized by a PhosphorImager. MuÈ ller-Stiftung.

References

Aitken A. (1995). TIBS, 20, 95 ± 97. Alessi DR, Saito Y, Campell DG, Cohen P, Sithanandam G, Alessandrini A, Greulich H, Huang WD and Erikson RL. Rapp U, Ashworth A, Marshall CJ and Cowley S. (1994). (1996). J. Biol. Chem., 271, 31612 ± 31618. EMBO J., 13, 1610 ± 1619. MEK1-mediated Raf-1 activation S Zimmermann et al 1511 Anderson N.G, Li P, Marsden LA, Williams N, Roberts TM Li S and Sedivy JM. (1993). Proc. Natl. Acad. Sci. USA, 90, and Sturgill T. (1991). Biochem. J., 277, 573 ± 576. 9247 ± 9251. Boyle B, van der Greer P and Hunter T. (1991). Methods in Lovric J and Moelling K. (1996). Oncogene, 12, 1109 ± 1116. Enzymology, 201, 110 ± 149. Luo ZJ, Tzivion G, Belshaw PJ, Vavvas D, Marshall M and Brtva TR, Drugan JK, Ghosh S, Terrell RS, Campbell-Burk Avruch J. (1996). Nature, 383, 181 ± 185. S, Bell RM and Der CJ. (1995). J. Biol. Chem., 270, 9809 ± Marais R, Light Y, Paterson HF and Marshall CJ. (1995). 9812. EMBO J., 14, 3136 ± 3145. Cherniack AD, Klarlund JK, Conway BR and Czech MP. MichaudN,FabianJR,MathesKDandMorrisonDK. (1995). J. Biol. Chem., 270, 1485 ± 1488. (1995). Mol. Cell. Biol., 15, 3390 ± 3397. Cleghon V and Morrison DK. (1994). J. Biol. Chem., 269, Mischak H, Seitz T, Janosch P, Eulitz M, Steen H, Schellerer 17749 ± 17755. M, Phillip A and Kolch W. (1996). Mol. Cell. Biol., 16, Daum G, Eisenmann-Tappe I, Fries HW, Troppmair J and 5409 ± 5418. Rapp U. (1994). TIBS, 19, 474 ± 479. Moelling K, Heimann B, Beimling P, Rapp U and Sander T. Dent P, Reardon DB, Morrison DK and Sturgill TW. (1995). (1984). Nature, 312, 558 ± 561. Mol. Cell. Biol., 15, 4125 ± 4135. Morrison DK, Heidecker G, Rapp U and Copeland TD. Fabian JR, Daar IO and Morrison DK. (1993). Mol. Cell. (1993). J. Biol. Chem., 268, 17309 ± 17316. Biol., 13, 7170 ± 7179. Muslin AJ, Tanner JW, Allen PM and Shaw AS. (1996). Cell, Fabian JR, Vojtek AB, Cooper JA and Morrison DK. 84, 889 ± 897. (1994). Proc. Natl. Acad. Sci. USA, 91, 5982 ± 5986. Nebreda AR. (1994). TIBS, 19, 1±2. Farrar MA, Alberola-Ila J and Perlmutter RM. (1996). Niehof M, Radziwill G, Klauser S and Moelling K. (1994). Nature, 383, 178 ± 181. Bioch. Bioph. Res. Commun., 206, 46 ± 50. Feig LA, Pan B-T, Robberts T and Cooper GM. (1986). Posada J, Yew N, Ahn NG, Vande Woude GF and Cooper Proc. Natl. Acad. Sci. USA, 83, 4607 ± 4611. JA. (1993). Mol. Cell. Biol., 13, 2546 ± 2553. Ferrier AF, Lee M, Anderson WB, Benvenuto G, Morrison Radziwill G, Niehof M, Rommel C and Moelling K. (1995). DK, Lowy DR and DeClue JE. (1997) J. Biol. Chem., 272, Proc.Natl.Acad.Sci.USA,92, 1421 ± 1425. 2136 ± 2142. Radziwill G, Steinhusen U, Aitken A and Moelling K. Haefner S, Adler HS, Mischak H, Janosch P, Heidecker G, (1996). Biochem. Bioph. Res. Commun., 227, 20 ± 26. Wolfman A, Pippig S, Lohse M, Ueng M and Kolch W. Rommel C, Radziwill G, Lovric' J, Noeldecke J, Heinicke T, (1994). Mol. Cell. Biol., 14, 6696 ± 6703. Jones D, Aitken A and Moelling K. (1996). Oncogene, 12, den Hertog J, Tracy S and Hunter T. (1994). EMBO J., 13, 609 ± 619. 3020 ± 3032. Rommel C, Radziwill G, Moelling K and Hafen E. (1997). Holt KH, Kasson BG and Pessin JE. (1996). Mol. Cell. Biol., Mech. of Dev., 64, 35 ± 105. 16, 577 ± 583. Schramm K, Niehof M, Radziwill G, Rommel C and Jelinek T, Dent P, Sturgill TW and Weber M. (1996). Mol. Moelling K. (1994). Bioch. Bioph. Res. Commun., 201, Cell. Biol., 16, 1027 ± 1034. 740 ± 747. John J, Frech M and Wittinghofer A. (1988). J. Biol. Chem., Seger R and Krebs E G. (1995). FASEB J., 9, 726 ± 735. 263, 11792 ± 11799. Stokoe D, Macdonald SG, Cadwallader K, Symons M and Keyse SM and Emslie EA. (1992). Nature, 359, 644 ± 647. Hancock JF. (1994). Science, 264, 1463 ± 1467. Kolch W, Heidecker G, Kochs G, Hummel R, Vahidi H, Suen KL, Bustelo XR and Barbacid M. (1995). Science, 264, Mischak H, Finkenzeller G, MarmeÂ D and Rapp U. 1463 ± 1467. (1993). Protein kinase C alpha activates Raf-1 by direct Turner BC, Tonks NK, Rapp U and Reed J. (1991). Proc. phosphorylation. Nature, 364, 249 ± 252. Natl. Acad. Sci. USA, 90, 5544 ± 5548. Kyriakis JM, App H, Zhang XF, Banerjee P, Brautigan DL, Ueki K, Matsuda S, Tobe K, Gotho Y, Tamemoto H, Yachi Rapp UR and Avruch J. (1992). Raf-1 activates MAP M, Akanuma Y, Yazaki Y, Nishida E and Kadowaki T. kinase-kinase. Nature, 358, 417 ± 421. (1994). J. Biol. Chem, 269, 15756 ± 15761. Kyriakis JM, Forces TL, Rapp U, Bonventre JV and Avruch Waters SB, Yamauchi K and Pessin JE. (1995) Mol. Cell. J. (1993). J. Biol. Chem., 268, 16009 ± 16019. Biol., 15, 2791 ± 2799. Lange-Carter CA, Pleiman CM, Gardner AM, Blumer KJ Williams NG, Roberts TM and Li P. (1992). Proc. Natl. and Johnson GL. (1993). Science, 260, 315 ± 319. Acad. Sci. USA, 89, 2922 ± 2926. Leevers SJ, Paterson HF and Marshall CJ. (1994). Nature, 369, 411 ± 414.