Feedback Regulation of Raf-1 and Mitogen-activated Protein Kinase (MAP) Kinase Kinases 1 and 2 by MAP Kinase Phosphatase-1 (MKP-1)*

Inactivation of growth factor-regulated mitogen-activated protein (MAP) kinases (ERK1 and ERK2) has been proposed to occur in part through dephosphorylation by the dual specificity MAP kinase phosphatase-1 (MKP-1), an immediate early gene that is induced by mitogenic signaling. In this study, we examined the effect of MKP-1 on signaling components upstream of ERK1 and ERK2. Coexpression of MKK1 or MKK2 with MKP-1 resulted in 7–10-fold activation of mitogen-activated protein kinase kinase (MKK), which required the presence of regulatory serine phosphorylation sites. Endogenous MKK1 and MKK2 were also activated upon MKP-1 expression. Raf-1, a direct regulator of MKK1 and MKK2, was activated under these conditions, and a synergistic activation of MKK was observed upon coexpression of Raf-1 and MKP-1. This effect did not appear to involve synthesis of autocrine growth factors or the inhibition of basal extracellular signal-regulated kinase (ERK) activity but was inhibited by a dominant negative Ras mutant, indicating that MKP-1 enhances Ras-dependent activation of Raf-1 in a cell autonomous manner. This study demonstrates positive feedback regulation of Raf-1 and MKK by the MKP-1 immediate early gene and a potential mechanism for activating Raf-1/MKK signaling pathways alternative to those involving ERK.

The mitogen-activated protein (MAP) 1 kinase cascade has emerged as a key signaling pathway regulating factor-dependent cell growth and differentiation through intracellular phosphorylation (1,2). Growth factor regulation of this pathway involves the phosphorylation and activation of MAP kinases, ERK1 and ERK2, by MAP kinase kinases, MKK1 and MKK2 (3)(4)(5)(6). ERKs 1 and 2 phosphorylate various targets including upstream and downstream protein kinases, cell surface receptors, and nuclear transcription factors (1,7). MKKs 1 and 2, in turn, can be activated through phosphorylation by members of the Raf protein kinase family, including Raf-1, which is ubiq-uitously expressed (8,9). Receptor-dependent activation of Raf-1 involves its interaction with p21 Ras, through a mechanism that is not completely defined, but appears to involve Raf-1 dimerization and phosphorylation by heterologous protein kinases (10 -13).
The product of the immediate early gene, MAP kinase phosphatase (MKP-1), is able to dephosphorylate phosphoserine/ threonine as well as phosphotyrosine residues, and shows selectivity for ERKs 1 and 2 in vitro, with lower activity toward other MAP kinases such as JNK and p38 MAP kinase (14,15). MKP-1 inactivates ERK following growth factor stimulation in intact cells and also suppresses signaling downstream of ERK at the level of gene transcription and proliferation (15)(16)(17), most likely through its inhibitory effects on MAP kinase. Interestingly, MKP-1 is transcriptionally induced by stress-regulated pathways such as those occurring in response to UV treatment or activation of stress-regulated protein kinases (18,19), suggesting that MKP-1 may function as a mediator of cross-regulatory pathways between mammalian protein kinase cascades. A similar cross-regulation is observed between osmotic stress-and pheromone-activated protein kinase cascades in yeast involving transcriptional induction of protein-tyrosine phosphatases (20 -23).
To obtain a more complete understanding of how MKP-1 regulates cell signaling, we examined the effects of this enzyme on other components in the MAP kinase pathway in NIH 3T3 cells. Our studies demonstrate the novel finding that MKP-1 positively regulates Raf-1 and MKK components upstream of ERK in an ERK-independent manner, providing a means by which different pathways downstream of Raf-1 might be differentially controlled.
Transfections-Cells were seeded at 2 ϫ 10 5 cells per well into six-well plates (35 mm diameter) and transfected at 50 -80% confluence using 5 l of LipofectAMINE (Life Technologies, Inc.) according to manufacturer's instructions and 1 g of cDNA unless otherwise noted. Transfection efficiency was estimated as 20 -30%, based on fluorescence of cells transfected in parallel with a construct (pK7-GFP) expressing green fluorescent protein (a gift of Dr. Ian Macara). The cDNA constructs for expression of wild type and mutant hemagglutinin (HA)tagged MKKs 1 and 2 in pMCL have been described previously (24 -26). Constructs for expression of Myc-tagged wild type MKP-1 and MKP-1 (C258S) in a pCEP4 vector were a gift of Drs. Hong Sun and Nicholas Tonks (15). A construct for expression of Flag-tagged Raf-1 was a gift of Dr. Roger Davis (27). Dominant negative Raf-1 (S621A) construct was a gift of Dr. Deborah Morrison (28). Dominant negative H-Ras (S17N) construct was a gift of Dr. Melanie Cobb.
[ 35 S]Methionine Labeling of Cells-To confirm MKP-1 expression, cells were transfected with MKP-1 cDNA and incubated for 40 h in DMEM containing 10% FBS. Cells were removed to methionine-free DMEM containing 4% FBS for 1 h, then incubated with the same medium containing 75 Ci/ml [ 35 S]methionine for 4 h. Extracts were prepared, and MKP-1 was immunoprecipitated using monoclonal antibody 9E10 and resolved by SDS-PAGE followed by autoradiography.

MKP-1 Expression Activates MKK1 and MKK2-Cells
were transiently transfected with wild type MKK1 or MKK2 and MKP-1, and the activity of expressed MKK was measured. As shown in Fig. 1A, coexpression with wild type MKP-1 resulted in 7-10-fold enhancement of the activity of expressed MKK1 or MKK2. Controls from one experiment showed comparable MKK protein levels present in immunoprecipitates (Fig. 1B), demonstrating enhanced specific activity of wild type MKK. This effect was not observed with MKP-1 (C258S), indicating that the effect requires MKP-1 activity. Under conditions where MKK was activated by MKP-1, endogenous ERK2 activity was low ( Fig. 1, C and D), as was coexpressed wild type ERK2 (data not shown).
In parallel experiments, endogenous MKK1 and MKK2 activities were enhanced 2-fold upon expression of MKP-1, demonstrating that these effects were not limited to overexpressed MKK (data not shown). This activation is submaximal, since the transfection efficiency of these cells was approximately 20 -30%. In comparison, PDGF (25 ng/ml) treatment of cells resulted in 8 -10-fold activation of endogenous MKK1 (data not shown). The activation of expressed MKK by MKP-1 was also observed in cells that were serum-depleted for 18 h prior to cell lysis ( Fig. 2A). Under serum-starved conditions, ERK activation was not detected in the absence of growth factor stimulation (Fig. 2B), suggesting that the effect of MKP-1 on MKK was not due to removal of ERK activity. In three separate serum starvation experiments, the average MKK1 activation by MKP-1 was approximately 10-fold (Fig. 2C). Therefore, these effects of MKP-1 were observed with expressed as well as endogenous MKK under serum-starved and serum-fed conditions.
Expression of MKP-1 was confirmed by metabolically labeling cells with [ 35 S]methionine and immunoprecipitating the Myc-tagged MKP-1. A 40-kDa protein corresponding to the reported size of MKP-1 or the MKP1 (C258S) mutant was present in transfected cells (Fig. 3A). As expected, MKP-1 suppressed ERK activation in response to constitutively active MKK2-KWϩ71, evident by the disappearance of the gel-retarded form of ERK2 (Fig. 3B). Similarly, MKP-1 suppressed ERK2 activation by constitutively active MKK1-G1C, PDGF, or serum (data not shown). Inhibition of signaling downstream of ERK was tested by coexpressing an AP1 promoter-driven CAT reporter construct with constitutively active MKK1, MKK2, or v-Raf in the presence versus absence of MKP-1. As shown in Fig. 3C, transcription from the AP-1 promoter was inhibited in the presence of MKP-1.
Involvement of Raf-1 in MKK1 Activation in Cells Expressing MKP-1-Raf-1 has been shown to be a major upstream activator of MKK1 and MKK2 (8,9). The dependence of Raf-1 activity on MKP-1 was explored by transfecting cells with HA-tagged MKK1 and varying amounts of Raf-1 cDNA in the presence or absence of MKP-1. A small amount of Raf-1 expression significantly enhanced the MKK activation by MKP-1 (Fig. 5A). Increasing the amount of overexpressed Raf-1 did not further augment MKK1 activation by MKP-1 (Fig. 5A). Western blots probing the Flag epitope showed increased expressed Raf-1 levels in transfected cells, but no variation due to MKP-1 (data not shown). Endogenous MKK activity was also elevated with increasing Raf-1 levels in the presence of MKP-1 compared with MKK activity observed in the absence of MKP-1 (data not shown).
Raf-1 immunoprecipitated from the same cell extracts showed corresponding activation by MKP-1 (Fig. 5B). In the absence of expressed Raf-1, MKP-1 caused a 7-fold activation of endogenous Raf-1. Expression of Raf-1 resulted in up to 50-fold greater Raf activity in immunoprecipitates, which was further augmented by 3-5-fold by coexpression of MKP-1. When the same cells were treated with PDGF, Raf-1 activity was elevated by 10 -20-fold, comparable to the activities seen in the presence of MKP-1. Thus, MKP-1 activates Raf-1 in unstimulated cells to levels comparable to PDGF-treated conditions. Raf-1 protein from PDGF-treated cells showed a characteristic gel mobility retardation due to phosphorylation, whereas Raf-1 from MKP-1-expressing cells showed a migration similar to Raf-1 from unstimulated cells (data not shown). This suggests that Raf-1 regulation by MKP-1 may occur by a mechanism distinct from PDGF-induced Raf-1 stimulation. Furthermore, MKP-1 failed to activate the mutant, Raf-1 (S621A) (data not shown). Although expression of Raf-1 (S621A) did not inhibit the activation of expressed MKK by MKP-1, mutant Raf-1 led to no further enhancement of MKK-1 activation by MKP-1 (data not shown). Taken together, these data provide evidence that the activation of MKK by MKP-1 occurs through elevation of Raf-1 catalytic activity.
Involvement of Ras in MKK and Raf-1 Activation by MKP-1-Involvement of Ras was further investigated using a dominant negative H-Ras mutant (S17N), which inhibits growth factor-dependent activation of ERK. Coexpression of Ras (S17N) with MKP-1 and wild type MKK1 showed partial inhibition of MKK activation by MKP-1 (Fig. 6A) under conditions of comparable loading of MKK immunoprecipitates. Parallel controls confirmed the dominant interfering effect of Ras (S17N) on Ras signaling by showing a complete inhibition of PDGF-stimulated ERK activity (Fig. 6B). The data strongly suggest that Raf-1 activation by MKP-1 occurs through a Rasdependent mechanism.
A key mechanism for Raf-1 activation involves interaction of Raf-1 with Ras⅐GTP following GDP/GTP exchange catalyzed by the SOS exchange factor (10,31,32). Phosphorylation of SOS by ERK-and MKK-dependent pathways reportedly prevents Ras/Raf-1 association and Raf-1 activation under conditions correlating with a reduced mobility of SOS on SDS-PAGE (33)(34)(35). We examined SOS phosphorylation under conditions where MKK was activated by MKP-1 and endogenous ERK phosphorylation was inhibited by MKP-1. Under basal conditions, no gel mobility shift of SOS was observed (data not shown). Growth factor treatment of cells results in retardation of SOS gel mobility previously attributed to Ser/Thr phosphorylation and correlated with SOS inactivation (33,34). In addition, no discernible difference in the gel retardation was observed in cells transfected with MKP-1, suggesting that SOS phosphorylation response to PDGF is not altered by MKP-1 (data not shown).
To test whether the activation of Raf-1 and MKK by MKP-1 occurred through MKP-1 induction of secreted growth factors, conditioned medium was removed from cells transfected with MKP-1 and added onto cells transfected only with HA-MKK1. Under these conditions, no activation of expressed MKK1 was observed (data not shown). Furthermore, cells were transfected with MKK1 in the presence or absence of MKP-1 and treated 24 h later with the translation inhibitors, cycloheximide and puromycin. After 48 h, no inhibition of MKK activation by MKP-1 was observed (data not shown). These data indicate that the activation of MKK by MKP-1 occurs through a cell autonomous mechanism. DISCUSSION The growth factor-stimulated MAP kinase cascade has been shown to be inactivated at the level of ERK1 and ERK2 by the dual specificity phosphatase, MKP1 (14 -17, 36 -38). Previous studies proposed that MKP-1 might function as a feedback regulator of ERK signaling. Originally, MKP-1 was identified through its transcriptional induction following growth factor stimulation of cells and was described as an immediate early gene whose message is rapidly induced in a manner independent of protein synthesis (39). The kinetics of MKP-1 induction are thus consistent with its potential role in catalyzing ERK dephosphorylation, which occurs during a phase of inactivation observed in response to many types of stimuli. However, the induction of MKP-1 and dephosphorylation of ERK are uncorrelated in some studies, suggesting that this may not occur in all cells. For example, the inactivation of ERK in PC12 cells that occurs several minutes following epidermal growth factordependent activation is unaffected by cycloheximide, conditions that suppress MKP-1 induction (40,41). These findings suggest that alternative functions of MKP-1 exist that have yet to be described.
In this study, we show for the first time that MKP-1 positively regulates enzymes upstream of ERK, namely MKK1, MKK2, and Raf-1, thus providing an additional pathway for feedback regulation of Raf-1 by immediate early gene expression. This effect appears to occur through a cell autonomous mechanism and is partially blocked by dominant negative Ras, indicating that Ras-mediated activation of Raf-1 is most likely involved.
Several mechanisms for negative feedback regulation of upstream signaling targets by ERK, including cell surface receptors (epidermal growth factor receptor), Ras guanine nucleotide exchange factor (SOS), or Raf-1 have been proposed (33-35, 42, 43). In particular, SOS phosphorylation by ERK has been shown to down-regulate signaling through dissociation of Grb2-SOS (33)(34)(35). Furthermore, others have reported Raf-1 activation in unstimulated cells treated with the MKK1 inhibitor, PD98059 (60). It is possible that part of the stimulatory effect we observe with MKP-1 can be accounted for by relief of signal repression by ERK. However, it is unlikely that ERK inhibition accounts for all of the effects because our data were taken in cells treated under basal conditions and reproduced under serum-starved conditions where ERK activity is low or undetectable. Consistent with this, under basal or serum-starved conditions, we found no evidence for retardation in SOS gel mobility, which has been attributed to feedback phosphorylation by ERK-or MKK-dependent kinases (33,34,44). In addition, PDGF-induced SOS phosphorylation and gel mobility retardation were unaffected by MKP-1 expression, suggesting that altered SOS activity due to phosphorylation is not involved. The available data suggest the possibility that MKP-1 activates Raf-1 through Ras, involving a mechanism that may not involve inactivation of ERK.
The incomplete effect of Ras (S17N) on MKP-1 activation of MKK suggests that Ras-independent mechanisms for Raf-1 activation might also be regulated by MKP-1. Ras-independent means of Raf-1 activation might occur through Raf dimerization or through direct phosphorylation by tyrosine kinases, protein kinase C, ceramide-activated protein kinase, or kinase suppressor of ras, KSR (11)(12)(13)(45)(46)(47)(48)(49)(50)(51)(52)(53). Raf-1 inhibition may involve phosphorylation of Ser-43 or Ser-621 by protein kinase A (54 -56). Alternatively, Raf-1 may be regulated by interactions with other signaling components such as 14-3-3 proteins (57)(58)(59). It is conceivable that alternative mechanisms for Raf-1 activation may be regulated indirectly through dephosphorylation of cellular targets by MKP-1. A significant implication of our study is that MKP-1 may alter the flux through different Raf-1-dependent pathways, thus providing a means of uncoupling different regulatory pathways downstream of Ras. For example, Raf-1 has been shown to phosphorylate other proteins, including IkB and the proapoptotic Bad protein (61)(62)(63), suggesting that it targets other substrates and downstream pathways in addition to MKK1 and MKK2 and the ERK pathway. Under some conditions, signaling through Raf-1 may also activate p70 S6 kinase through an ERK-independent mechanism (64). Our results would predict that induction of MKP-1 (e.g. through immediate early gene expression or activation of stress-activated protein kinase pathways) might enhance signaling through alternative Raf-1 targets under conditions where ERK signaling is suppressed. Another possibility that has yet to be demonstrated is that MKK targets other cellular FIG. 6. MKK activation by MKP-1 is inhibited by dominant negative Ras. Cells were transfected with HA-wild type MKK1 in the presence or absence of MKP-1, Ras (S17N), or HA-ERK. A, MKK1 was immunoprecipitated using 12CA5 antibody, and activities were measured by phosphorylation of ERK2(K52R). Western blots of immunoprecipitates from a representative experiment show equal amounts of MKK1 protein in lysates. The number in parentheses indicates the number of separate experiments in which the mean and standard errors were derived. B, Ras (S17N) inhibited expressed HA-ERK2 activation by PDGF, as measured by phosphorylation of myelin basic protein by ERK2 immunoprecipitated using 12CA5 antibody.
substrates in addition to ERK, in which case phosphorylation of these substrates might also be enhanced by MKP-1.
In summary, this study reveals the ability of MKP-1 to serve as a positive regulator of Raf-1 signaling and presents an important consideration for studies utilizing MKP-1 as an inhibitor of ERK signaling. The results suggest a novel approach to address Raf-1 regulation by both Ras-independent and dependent mechanisms, and provide the groundwork for future studies examining the control of multiple signaling pathways downstream of Raf-1.