Granulocyte Colony-stimulating Factor Induces Erk5 Activation, Which Is Differentially Regulated by Protein-tyrosine Kinases and Protein Kinase C REGULATION OF CELL PROLIFERATION AND SURVIVAL*

Granulocyte colony-stimulating factor (G-CSF) plays a major role in the regulation of granulopoiesis. Treatment of cells with G-CSF has been shown to activate multiple signal transduction pathways. We show here that Erk5, a novel member of the MAPK family, and its specific upstream activator MEK5 were activated in response to incubation of cells with G-CSF. Different from other members of the MAPK family including Erk1/2, JNK, and p38, maximal activation of Erk5 by G-CSF required the C-terminal region of the G-CSF receptor. Genistein, a specific inhibitor of protein-tyrosine kinases, blocked G-CSF-induced Erk5 activation. In contrast, inhibition of protein kinase C activity increased G-CSF-mediated activation of Erk5 and MEK5, whereas stimulation of protein kinase C activity inhibited activation of the two kinases by G-CSF. The proliferation of BAF3 cells in response to G-CSF was inhibited by expression of a dominant-negative MEK5 but potentiated by expression of a constitutively active MEK5. Expression of the constitutively active MEK5 also increased the survival of BAF3 cells cultured in the absence of or in low concentrations of G-CSF. Together, these data im-plicate

In addition to the Jak/Stat cascade, G-CSF has also been shown to activate the mitogen-activated protein kinases (MAPKs), which relay signals from the cell membrane to the nucleus (15,16). The MAPK module is composed of three kinases that establish a sequential activation cascade. A MAPK kinase kinase (MAPKKK or MEKK) activates a MAPK kinase (MAPKK or MEK), which in turn activates a MAPK via dual phosphorylation of a Thr-X-Tyr motif. The MAPK pathways are activated by a variety of extracellular stimuli via interaction with small GTPases including Ras and Rho family members and mediate distinct cellular responses. Members of MAPKs that are activated by G-CSF include Erk1/2, JNK, and p38 2 and activation of these kinases appears to be involved in G-CSFstimulated cell proliferation (8,(17)(18)(19)(20)(21).
Erk5, also known as BMK1, is a novel member of the MAPK family (22,23). Erk5 shares the Thr-X-Tyr sequence in the dual phosphorylation site with other MAPKs, but has a unique long C-terminal tail, suggesting a distinct regulatory mechanism of its activation. Similar to other members of the MAPK family, Erk5 has been shown to phosphorylate transcription factors such as c-Myc and members of MEF2 family of transcription factors (24 -27). Interestingly, the C-terminal region of Erk5 has been shown to contain a potent transcription activation domain that is required for coactivation of MEF2D by Erk5 (28). The direct upstream activator of Erk5 has been identified as MEK5 (23)(24)(25), which appears to be activated by MEKK3 and Cot (29,30). Although Erk5 was originally shown to be activated by stresses such as oxidants, hyperosmolarity, and fluid shear stress (24,31,32), recent studies have indicated that Erk5 is also activated by receptors with intrinsic tyrosine kinase activity such as the receptors for epidermal growth factor (EGF) and nerve growth factor (NGF) as well as by G-protein coupled receptors. (27,33,34). However, the role of Erk5 in cytokine receptor signal transduction has not been explored. In this study, we show that Erk5 is strongly activated by G-CSF stimulation of cells in a MEK5-dependent manner. Interestingly, unlike Erk1/2, JNK, and p38, maximal Erk5 activation requires the C-terminal region of the G-CSF receptor. We demonstrate that protein-tyrosine kinases and protein kinase C (PKC) exert differential effects on G-CSF-induced activation of Erk5 and MEK5. Furthermore, we provide evi-dence indicating that the MEK5/Erk5 pathway plays a positive role in the regulation of G-CSF-induced cell proliferation and survival.

MATERIALS AND METHODS
Cells-Murine BAF3 and 32D cells, stably transfected with cDNAs encoding either the wild-type or the truncated forms of the human G-CSF receptor, have been described (35,36). Murine L-G cell line (37) was kindly provided by Dr. T. Honjo (Kyoto University Faculty of Medicine). Cells were grown in RPMI 1640 medium supplemented with 10% fetal calf serum, 50 M 2-mercaptoethanol, gentamicin (50 g/ml), and 10% WEHI-3B cell-conditioned media. BAF3 cells expressing the wild-type G-CSF receptor (BAF/WT) were electroporated with cDNAs encoding MEK5 or dominant-active mutant MEK5DD that were cloned in pCEFL mammalian expression vector (30). Cells were selected in G418-containing medium, and individual clones were expanded and examined for expression of transfected proteins by Western blotting.
Reagents-Wortmannin was from Sigma. PD98059, SB 202190, PP1, genistein, and GF109203X were obtained from CalBiochem. U0126 was purchased from Promega. Rabbit anti-JNK and p38 antisera used for immunoprecipitation and kinase assays were raised against GST fusion proteins of murine JNK and p38. Anti-Erk2 antibody (TR10) was kindly provided by M. Weber (University of Virginia). Anti-Erk1/2 (Pan-Erk) antibody was from Transduction Laboratories. Anti-HA and -JNK1 monoclonal antibodies were obtained from PharMingen. Antibodies to Erk5 and MEK5 were purchased from StressGen. Anti-Erk5 antibody used for immunoprecipitation and anti-p38 antibody were from Santa Cruz Biotechnologies. Phosphospecific Erk1/2 antibody was obtained from New England BioLabs. [␥-32 P]ATP was purchased from Amersham Pharmacia Biotech.
Immunoprecipitation and in Vitro Kinase Assays-Cells were starved in the absence of serum and conditioned media for 4 h prior to stimulation with G-CSF for the times indicated. The preparation of whole cell extracts, immunoprecipitation, and in vitro kinase assays were essentially as described (38). GST⅐MEF2C fusion protein (amino acids 87-467) was used for assay of Erk5 activity, GST⅐ATF2(96) fusion protein for JNK and p38 activities, and myelin basic protein (MBP) for Erk2 activity. GST⅐Erk5 fusion protein used for assay of MEK5 activity has been described (34). The GST fusion proteins were prepared as described (27).
Expression Vectors and Transient Transfection-The expression vectors containing the HA-tagged Erk5, MEK5, and dominant-negative MEK5AA (Ser 311 and Thr 315 to Ala) have been previously described (27). BAF3 and 32D cells expressing the different G-CSF receptor forms were transfected by electroporation (Electro Square Porator; BTX Genetronics Inc.) (12). Sixteen hours after transfection, cells were deprived of serum for 4 h prior to treatment with G-CSF for the indicated times.
Thymidine Incorporation Assay-10 5 cells were incubated in triplicate in 100 l of serum-free medium in 96-well plates in the presence or absence of G-CSF (10 ng/ml) for 20 h. Cells were then pulsed with 1 Ci

Erk5
Is Activated by G-CSF Treatment of Cells-To investigate the involvement of Erk5 in G-CSF-dependent signaling pathway, we examined Erk5 activation in response to G-CSF in murine pro-B BAF3 cells expressing the wild-type G-CSF receptor (BAF/WT). Cells were transiently transfected with HAtagged Erk5 and were starved for 4 h prior to G-CSF stimulation for the indicated times. Whole cell extracts were prepared and subjected to immunoprecipitation with an anti-HA monoclonal antibody. In vitro kinase assays using MEF2C as a substrate revealed that G-CSF treatment of cells resulted in transient and strong (more than 8-fold) activation of Erk5 at 15 min of treatment (Fig. 1A). G-CSF also induced activation of endogenous Erk5 in BAF/WT cells, myeloid 32D cells stably transfected with the wild-type G-CSF receptor (32D/WT) and myeloid L-G cells that expressed the endogenous G-CSF receptor (Fig. 1B). Both 32D/WT and L-G cells were capable of differentiating into mature granulocytes in response to G-CSF (36,37). Activation of endogenous Erk5 appeared weaker than that seen when assaying the transfected protein because of the poor efficiency of the anti-Erk5 antibody to immunoprecipitate the protein (data not shown).

Maximal Activation of Erk5 by G-CSF Requires the C-terminal Region of the G-CSF Receptor-
To determine which cytoplasmic regions of the G-CSF receptor play a role in the regulation of Erk5 activation, we assayed Erk5 activity in BAF3 cells expressing the truncated forms of the G-CSF receptor ( Fig. 2A). As shown in Fig. 2B, G-CSF stimulation of BAF3 cells expressing the D715 receptor induced a weak increase in Erk5 activity. The kinetics of Erk5 activation was similar to that induced by the wild-type receptor (data not shown). In contrast, activation of endogenous or transfected Erk1/2, JNK, and p38 was strongly stimulated by the D715 receptor ( Fig. 2B and data not shown). No Erk5 activation was observed upon G-CSF stimulation of BAF3 cells expressing the D685 receptor over a period from 5 to 60 min ( Fig. 2B and data not shown). The D685 receptor was still able to mediate a weak activation of Erk1/2, JNK, and p38 and has previously been shown to activate Jak2, Stat5, and Akt (12,35,38). Together, these results indicated that the C-terminal 98 amino acids of the G-CSF receptor play an important role in regulating Erk5 activation and that the region between amino acids 686 and 715 of the G-CSF receptor is indispensable for mediating G-CSF-induced Erk5 activation.
MEK5 Is Activated by G-CSF and Is Required for G-CSFstimulated Activation of Erk5-MEK5 has been shown to be required for EGF-induced Erk5 activation (23)(24)(25)27). We examined whether MEK5 was activated by stimulation of cells with G-CSF. Whole cell extracts were prepared from BAF/WT cells stimulated with G-CSF for different times. Endogenous MEK5 was immunoprecipitated and its activity measured with in vitro kinase assay using GST⅐Erk5 fusion protein as a substrate. Incubation of cells with G-CSF stimulated MEK5 activity (Fig. 3A). Activation of MEK5 by G-CSF occurred earlier than that of Erk5, usually 5 min after addition of G-CSF, and the signal started to diminish after 20 min and disappeared 60 min after G-CSF stimulation (data not shown). It appeared that G-CSF-stimulated activation of MEK5 was slightly more sustained than Erk5 activation, suggesting that the two kinases are down-regulated by different mechanisms. Treatment and incubated with G-CSF (10 ng/ml) for the indicated times. Erk5 was immunoprecipitated with an anti-HA body, and Erk5 activity was measured by in vitro kinase assay using the GST⅐MEF2C fusion protein as a substrate (upper panel). Equal loading was confirmed by immunoblotting with an anti-Erk5 antibody (lower panel). B, activation of endogenous Erk5 by G-CSF in BAF3 and 32D cells transfected with the wild-type G-CSF receptor as well as in L-G cells expressing the endogenous G-CSF receptor. Cells were stimulated with G-CSF for 15 min, and Erk5 activity was determined by in vitro kinase assay following immunoprecipitation with the anti-Erk5 antibody.
of 32D/WT and L-G cells with G-CSF also induced activation of endogenous MEK5 (data not shown). Activation of MEK5 was further examined in BAF3 cells expressing the truncated forms of the G-CSF receptor. As shown in Fig. 3B, the D715 mutant mediated a weak activation of MEK5 whereas the D685 receptor failed to induce MEK5 activation. Thus, MEK5 activation by G-CSF correlated well with that of Erk5 activation in cells expressing the different forms of the G-CSF receptor (see Fig.  2). To determine the involvement of MEK5 in G-CSF-induced activation of Erk5, we cotransfected BAF/WT cells with HAtagged Erk5 and MEK5AA, a dominant-negative form of MEK5 (27). Expression of MEK5AA strongly inhibited Erk5 activation by G-CSF (Fig. 3C), but had no significant effect on G-CSFinduced Erk1/2 activation (data not shown). As shown in Fig.  3B, H 2 O 2 treatment of BAF3 cells stimulated Erk5 activation, which was also blocked by MEK5AA.
Protein-tyrosine Kinase Activity Is Required for G-CSF-induced Erk5 Activation Whereas PKC Down-regulates the Activity of the Erk5 Pathway-To investigate whether other signaling pathways have a role in regulating Erk5 activity, we tested the effects of different protein kinase inhibitors on G-CSFinduced Erk5 activation in BAF/WT cells (Fig. 4). Pretreatment  We further investigated whether stimulation of PKC activity would attenuate Erk5 activation by G-CSF. BAF/WT cells were treated with phorbol 12-myristate 13-acetate (PMA), a well known PKC activator, prior to stimulation with G-CSF. As shown in Fig. 5A, PMA alone did not affect Erk5 activity but inhibited G-CSF-induced Erk5 activation by more than 60% as determined by phosphorimager analysis. In contrast, PMA alone induced Erk1/2 activation and further potentiated the activation of Erk1/2 by G-CSF, presumably by stimulating Raf-1 activity (46,47). GF109203X alone induced Erk5 activation in BAF/WT cells and combination of GF109203X and G-CSF exerted a synergistic effect on Erk5 activation, leading to an ϳ4.5-fold greater increase in Erk5 activity as compared with that induced by G-CSF alone. It appeared that GF109203X slightly potentiated G-CSF-induced activation of Erk1/2 although its effect on Erk1/2 activation by G-CSF was significantly weaker than on Erk5 activation. Depletion of PKC through overnight treatment of cells with PMA also potentiated Erk5 activation by G-CSF. Comparable results were obtained in 32D/WT cells although GF109203X alone did not stimulate Erk5 activity in 32D cells (Fig. 5B).
We examined whether G-CSF-induced activation of MEK5 is affected by alteration of PKC activity. As shown in Fig. 5C, endogenous MEK5 activity was strongly induced by treatment of cells with GF109203X. The combination of G-CSF and GF109203X further dramatically enhanced the activity of MEK5. On the contrary, preincubation of cells with PMA inhibited G-CSF-induced activation of MEK5. These results suggest that PKC exerts the negative effect on G-CSF-induced Erk5 activation via inhibiting MEK5 activity either directly or indirectly.
Erk5 Signaling Pathway Plays a Positive Role in G-CSFinduced Cell Proliferation-To investigate whether the Erk5 pathway is involved in the regulation of cell proliferation, we transiently transfected BAF/WT cells with the dominant-negative mutant MEK5AA because we were unable to stably express MEK5AA in the cells. The transfection efficiency of BAF3 in several experiments was ϳ50% as determined by expression of a transfected green fluorescence protein (GFP) construct (data not shown). As shown in Fig. 6A, expression of MEK5AA inhibited G-CSF-stimulated cell proliferation by ϳ30% as measured by [ 3 H]thymidine uptake assay. In contrast, expression of MEK5 did not inhibit cell proliferation induced by G-CSF.
We stably expressed the constitutively active MEK5DD in BAF/WT cells to examine its effect on G-CSF biological responses. Expression of MEK5DD but not MEK5 stimulated Erk5 activity in BAF/WT cells in the absence of G-CSF (Fig.  6B), but did not affect the activities of Erk1/2, JNK, p38, or Akt (data not shown). Interestingly, G-CSF stimulation of cells expressing MEK5DD decreased Erk5 activity, possibly because of activation of a negative regulatory pathway. In the absence of G-CSF, MEK5DD weakly stimulated the uptake of [ 3 H]thymidine by the BAF/WT cells (about 3-fold increase as compared with cells not transfected or transfected with MEK5). Expression of MEK5DD potentiated cell proliferation stimulated by suboptimal concentrations of G-CSF (0.1-0.3 ng/ml, Fig. 6C). However, when stimulated with optimal concentration of G-CSF (10 ng/ml), the proliferation of cells expressing MEK5 or MEK5DD showed no significant differences. Together, these results indicated that Erk5 pathway is positively involved in regulating G-CSF stimulated-cell proliferation.
Erk5 Pathway Positively Regulates Cell Survival-We also examined the effect of expression of constitutively active MEK5DD on cell survival. Cells were washed and cultured in serum-free medium supplemented with no G-CSF or G-CSF at concentrations as indicated in Fig. 7. Cell viability was determined by exclusion of trypan blue staining after 30 and 54 h of incubation (Fig. 7, A and B). Expression of MEK5DD markedly enhanced the viability of cells cultured in the absence of G-CSF or in the suboptimal concentrations of G-CSF. The effect of MEK5DD was most significant at 54 h of culture at which time point MEK5DD reduced the concentration of G-CSF (as low as 0.03 ng/ml) required for maximal cell survival (Fig. 7B). MEK5DD did not further significantly increase the viability of cells that were cultured in G-CSF at optimal concentrations (10 ng/ml).

DISCUSSION
Treatment of cells with G-CSF has been shown to activate the Jak/Stat, MAPK (Erk1/2, JNK and p38), PI3-K, and Src family kinase cascades (2). Activation of these signaling cascades has been associated with G-CSF-stimulated cell proliferation although each of these signaling events may also be uniquely involved in eliciting a specific cellular response to G-CSF (11,12,48,49). In this paper, we show that a novel MAPK family member Erk5 and its upstream activator MEK5 are activated upon G-CSF treatment of cells. Furthermore, we demonstrate that the MEK5/Erk5 cascade positively regulates G-CSF-stimulated cell proliferation and survival. Although Erk5 has been shown to be activated by receptor-tyrosine kinases (33,34), our data provide the first example that Erk5 pathway is critically implicated in cytokine receptor signal transduction and plays a central role in the regulation of cell proliferation and survival that are stimulated by G-CSF Erk5 activation by G-CSF is inhibited by pretreatment of FIG. 5. PKC pathway negatively regulates the activation of Erk5 and MEK5 by G-CSF. A, BAF/WT cells transiently transfected with the HA-tagged Erk5 were pre-incubated either with nothing (Ctr), or with PMA for 5 min or overnight (o/n), or with GF109203X for 30 min. Cells were then stimulated with G-CSF for 15 min, and Erk5 activation was determined by in vitro kinase assay (upper panel). The membrane was probed for Erk5 (middle panel). Aliquots of whole cell extracts were examined by Western blotting for Erk1/2 phosphorylation using a phosphospecific antibody (lower panel). B, 32D cells expressing the wild-type G-CSF receptor were transiently transfected with the HA-tagged Erk5 and were then examined for the effects on PMA and GF109203X on G-CSF-induced Erk5 activation. C, BAF/WT cells were pretreated with PMA or GF109203X followed by stimulation with G-CSF as described above. MEK5 was immunoprecipitated and its activity was determined by in vitro kinase assay using GST⅐Erk5 fusion protein as a substrate. cells with genistein, implying that protein-tyrosine kinases are involved in this process. It has been shown that significant cross talk occurs between distinct signal transduction pathways that are activated by a given cytokine. For instance, activation of Raf/Erk1/2 kinase signaling by both growth factors and interferons requires Jak2 or Jak1 (50,51). We recently showed that G-CSF-induced activation of Akt requires the activity of Src family kinases whereas activation of Jaks appears to be not essential (38). The inhibitor of Src family kinases PP1 (44,52) has no significant or only a minimal effect on Erk5 activation induced by G-CSF, suggesting that Src family kinases may not be critically involved in this activation process. It remains to be determined whether Jaks are required for G-CSF-stimulated activation of Erk5.
In contrast to protein-tyrosine kinases, Erk5 activation in response to G-CSF is down-regulated by a PKC-dependent pathway. A recent report indicated that PKC activator PMA did not stimulate Erk5 activity, however, the role of PKC as a negative regulator of Erk5 activity was not examined (33). PKC stimulates the activities of Erk1/2, JNK, and p38 (46,(53)(54)(55). Consistent with these studies, PMA activates Erk1/2 and strongly potentiates G-CSF-induced Erk1/2 activation. These data indicate that the PKC pathway, which has been shown to be activated by G-CSF (56), has differential effects on G-CSFinduced activation of Erk5 and Erk1/2. It is still unknown which PKC isoform(s) and by what mechanism the involved PKC isoform down-regulates G-CSF-stimulated Erk5 activation. The fact that G-CSF-mediated activation of MEK5 is also down-regulated by the PKC pathway indicates that either MEK5 itself or a more upstream molecule of the Erk5 signaling cascade is the target of regulation by PKC.
The role of Erk5 in mediating cellular responses to extracellular stimuli is still unclear. Erk5 has been shown to phosphorylate and activate MEF2 family transcription factors including MEF2A, MEF2C, and MEF2D (24,27,28). We have also observed that G-CSF stimulated MEF2A-and MEF2C-dependent luciferase reporters (data not shown). It was shown recently that Erk5 was required for EGF-stimulated cell proliferation and for Cot-induced transformation of NIH3T3 cells (30,33). In transient transfection assays, we consistently observed that MEK5AA inhibited the proliferation of BAF/WT cells in response to G-CSF. Although the antiproliferative ef- fects of MEK5AA appear modest, they are likely to be more dramatic than appreciated because only about 50% of cells are transfected under the conditions used for the assay (data not shown). In addition, expression of constitutively active MEK5DD significantly potentiated cell proliferation that was stimulated by suboptimal concentrations of G-CSF. Together, these data suggest that the MEK5/Erk5 pathway is positively involved in G-CSF-stimulated cell proliferation. More interestingly, our data also indicate that the MEK5/Erk5 pathway is an important mediator of the survival signals transduced by the G-CSF receptor. Expression of MEK5DD in BAF/WT cell protected cells from apoptosis induced by cytokine withdrawal and significantly reduced the concentrations of G-CSF required for inducing maximal cell survival. Notably, maximal survival of cells expressing MEK5DD required a G-CSF concentration of only 0.03 ng/ml whereas G-CSF at 0.1 ng/ml induced less than 40% of maximal proliferation of the same cells. It thus appears that the Erk5 pathway has a more significant effect on cell survival than on cell proliferation.
Whether or not Erk5 is also implicated in the regulation of other biological activities of G-CSF remains to be investigated. Interestingly, the C-terminal region of the G-CSF receptor, a region required for induction of myeloid differentiation, plays an important role in the regulation of Erk5 activation in that truncation of this region markedly compromises the ability of the receptor to mediate Erk5 activation. In contrast, this carboxyl terminus does not play a major role in G-CSF-induced activation of Erk1/2, JNK, and p38. These results, together with the differential effects of PMA on Erk2 and Erk5 activation by G-CSF, suggest that distinct mechanisms are involved in regulating G-CSF-induced activation of Erk5 and other members of the MAPK family. The results reported here provide a framework to explore the roles of Ekr5 and its downstream substrates in the biological actions of G-CSF.