A Novel Mitogen-Activated Protein Kinase Is Responsive to Raf and Mediates Growth Factor Specificity

Abstract

The proto-oncogene Raf is a major regulator of growth and differentiation. Previous studies from a number of laboratories indicate that Raf activates a signaling pathway that is independent of the classic MEK1,2-ERK1,2 cascade. However, no other signaling cascade downstream of Raf has been identified. We describe a new member of the mitogen-activated protein kinase family, p97, an ERK5-related kinase that is activated and Raf associated when cells are stimulated by Raf. Furthermore, p97 is selectively responsive to different growth factors, providing a mechanism for specificity in cellular signaling. Thus, p97 is activated by the neurogenic factor fibroblast growth factor (FGF) but not the mitogenic factor epidermal growth factor (EGF) in neuronal cells. Conversely, the related kinase ERK5 is activated by EGF but not FGF. p97 phosphorylates transcription factors such as Elk-1 and Ets-2 but not MEF2C at transactivating sites, whereas ERK5 phosphorylates MEF2C but not Elk-1 or Ets-2. Finally, p97 is expressed in a number of cell types including primary neural and NIH 3T3 cells. Taken together, these results identify a new signaling pathway that is distinct from the classic Raf-MEK1,2-ERK1,2 kinase cascade and can be selectively stimulated by growth factors that produce discrete biological outcomes.

A common mechanism by which cells respond to their extracellular environment is through the action of growth factor, hormone, or cytokine receptors that are linked to intracellular signaling cascades. These cascades utilize the reversible phosphorylation of component members to convert an extracellular signal into a coordinated intracellular response. A key mediator of intracellular signals is the mitogen-activated protein kinase (MAPK) (31, 41). The MAPK pathway is composed of a highly conserved three-component cascade containing a MAP kinase kinase kinase (MAPKKK) that, upon activation, phosphorylates a MAP kinase kinase (MAPKK) (reviewed in references 43 and 46). The dual-specificity MAPKK then phosphorylates the Thr-X-Tyr (TXY) motif within the activation loop of MAPK. Phosphorylation of both the tyrosine and threonine residues is necessary and sufficient for full activation of the MAPKs (4).

The most extensively studied of these MAPK pathways is the extracellular signal-regulated kinase (ERK) pathway in which the MAPKKK is Raf, the MAPKK is MEK(1,2), and the MAPK is ERK(1,2). Initiation of this pathway usually results from the binding of a ligand to a cell surface receptor leading to the activation of the small GTP-binding protein p21-Ras. Activated Ras then recruits Raf to the plasma membrane, where it becomes activated through a poorly defined mechanism. Activated Raf phosphorylates the dual-specificity kinase MEK(1,2) producing fully active MEK, which phosphorylates ERK(1,2). Once activated, ERK translocates to the nucleus where it regulates the activity of numerous transcription factors and leads to specific biological responses as diverse as proliferation, apoptosis, and differentiation (31, 41).

The ERK subfamily of MAPKs is characterized by a TEY motif within the activation loop. While there are currently seven enzymes that are designated ERKs in the literature, only four of these enzymes have the classic TEY sequence in the activation loop. The first-identified and most extensively characterized ERKs are ERK1 and ERK2 (5). ERK3, which shares about 43% overall sequence identity with ERK1 and ERK2, has a SEG activation motif (5), and ERK4 is an uncharacterized protein on an immunoblot (38). ERK5/BMK1 (30, 48) has been implicated in growth control, mediating epidermal growth factor (EGF)-induced proliferation in HeLa cells (25) as well as early gene expression by MEF2C phosphorylation (24). ERK5 is activated by MEK5, which in turn can be activated by MEKK3 (6). ERK6, which promotes differentiation of myoblasts to myotubes, has a TGY activation motif and thus is a member of the p38 family (29). ERK7, which we recently cloned, has the TEY activation motif but is regulated differently from other ERK family members (3). For example, ERK7 has constitutive kinase activity that is not further stimulated by common activators of other MAPKs. Furthermore, of the four TEY-type ERKs, only ERK7 is resistant to chemical and physiological inhibitors of the MEKs (11, 22), suggesting that most of the ERKs share a similar MEK-MAPK activation cascade.

Activation of the Raf-MEK1,2-ERK1,2 cascade occurs in response to numerous signals and has been associated with many integral cellular functions, including growth and differentiation. How activation of a common pathway can mediate conflicting cellular processes is often explained as being dependent on modulating factors such as signal kinetics, amplitude, or localization (35). One mechanism for regulating these processes has been provided by the recent discoveries of scaffolding proteins that tether signaling components into discrete signaling complexes (37). Thus, many kinases and most phosphatases are promiscuous when analyzed in vitro, suggesting that their activation could lead to significant cross talk with other pathways if left untethered in the cytosolic milieu of a cell. Anchoring these signaling complexes via receptors or other scaffolding proteins would greatly decrease the ability of individual signaling components to interact with elements of other pathways. However, while scaffolding can, at least partially, explain how an input signal can activate a specific pathway without cross-activating other related pathways, it may not fully explain how activation of a common signaling intermediate can lead to divergent responses. An alternative possibility is that different ligands activate other unique signaling components. Several lines of evidence suggest that Raf may activate an additional signaling cascade that differs from the classic MEK1,2/ERK1,2 pathway. For example, Raf-activated but MEK-independent signaling cascades have been described for muscle cell proliferation (40) and cardiac muscle gene expression (20). We have identified a similar cascade in a conditionally immortalized rat hippocampal cell line, H19-7, derived from E17 rat embryos (27). H19-7 cells differentiate in response to basic fibroblast growth factor (bFGF) or inducibly activated Raf, but not activated MEK, and proliferate in response to EGF (14, 27). Using this model system we have identified a Raf-dependent but MEK-independent differentiation signaling pathway that has several characteristics that distinguish it from the classic Raf/ERK1,2 signaling pathway (8, 27, 28). First, this pathway is activated by FGF or Raf but not EGF. Second, this pathway is insensitive to the MEK inhibitor PD98509 (11) at doses that inhibit ERK1,2 TEY phosphorylation and kinase activity but do not block differentiation. Third, this pathway is activated by Raf at levels that fail to significantly increase ERK1,2 TEY phosphorylation or kinase activity yet still lead to differentiation. Finally, stimulation by FGF or Raf leads to activation of an Elk-1 kinase that is insensitive to inhibition by the MEK inhibitor PD98059 and can be detected in the absence of significant ERK1,2 activation. These characteristics allowed us to search for components of a signaling pathway that satisfied these criteria.

We report here the affinity purification and analysis of p97, a novel ERK5-related member of the MAPK family. p97 is selectively responsive to the differentiating agent FGF but not the mitogenic agent EGF in H19-7 cells, forming a stable multimolecular complex with activated Raf. Like other ERK family members, p97 phosphorylates several transcription factors, suggesting a role in gene regulation. Although p97 shares limited antibody cross-reactivity with ERK5, both its activation and substrate profile are distinct from those of ERK5. Taken together, the results define a novel ERK pathway downstream of Raf that mediates growth factor-specific signaling.

MATERIALS AND METHODS

Materials.

Bovine serum albumin, myelin basic protein (MBP), peroxidase-conjugated goat anti-rabbit immunoglobulin G (IgG), and peroxidase-conjugated goat anti-mouse IgG were purchased from Sigma Chemical (St. Louis, Mo.). Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), trypsin, penicillin, streptomycin, histone H1, glutathione-Sepharose 4B, and MBP were purchased from Gibco/BRL (Grand Island, N.Y.). Protein A-Sepharose was purchased from Pharmacia Biotech (Piscataway, N.J.). Monoclonal antibody (12CA5) against the hemagglutinin (HA) epitope was purchased from BAbCO (Emeryville, Calif.). High-affinity rat monoclonal antibody (3F10) against the HA epitope and histone H2b were purchased from Boehringer Mannheim (Indianapolis, Ind.). Affinity-purified peroxidase-conjugated goat anti-rabbit IgG and goat anti-mouse IgG were purchased from Transduction Laboratories (Lexington, Ky.). Anti-phospho ERK1,2 polyclonal antibodies were purchased from Promega (Madison, Wis.) (anti-ACTIVE MAPK), BioSource (Camarillo, Calif.) [anti-pTEpY (A)], or New England BioLabs (Beverly, Mass.) [anti-pTEpY (B)]. Both anti-BMK1/ERK5 (24) and anti-ERK 283 (17) have been described previously. The antibody against pan-ERK was from QCB (Hopkinton, Mass.), as were the antibodies against phosphorylated ERK5, JNK, and p38 and the nonphosphospecific antibodies against these proteins. Enhanced chemiluminescence reagents and [γ-³²P]ATP (6,000 Ci/mmol) were purchased from DuPont/NEN Research Products (Boston, Mass.). Bio-Rad protein assay reagents and carboxymethyl (CM) cation exchange columns were purchased from Bio-Rad (Hercules, Calif.). Talon resin was purchased from Clontech (Palo Alto, Calif.). Transfection reagent Trans-It was purchased from PanVera (Madison, Wis.).

Treatment of H19-7 or ΔRaf-1:ER cells.

H19-7 and ER:Raf-expressing H19-7 cells were grown on poly-l-lysine-coated cell culture dishes to approximately 70% confluence in DMEM containing 10% FBS and antibiotics. Prior to preparation of cell extracts, medium was changed to DMEM without serum overnight. Cells were sometimes pretreated for 10 min with a 10 μM concentration of the MEK inhibitor PD98059 prior to treatment with either EGF (1 ng/ml) or bFGF (10 ng/ml) for 15 min or for 1 h with ethanol or 10 nM to 1 μM β-estradiol (E2) and lysed on ice in CLB buffer (10 mM Tris [pH 7.4], 1 mM EDTA, 150 mM NaCl, 50 mM NaF, 1% Triton X-100, and protease and phosphatase inhibitors) or RIPA buffer (15 mM Tris [pH 7.4], 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.25% sodium dodecyl sulfate [SDS], and protein and phosphatase inhibitors).

Preparation of beads and binding reaction.

ΔRaf-1:ER and MEK:ER beads were prepared by coating protein A-Sepharose beads with rabbit anti-rat IgG followed by washing five times in CLB. The IgG-coated beads were then incubated with the anti-ER antibodies H222 or D75 (16) for at least 2 h at 25°C and washed extensively in CLB. Inactive (ethanol-treated) and active (1 μM E2-treated) ΔRaf-1:ER and MEK:ER were immobilized on the antibody-coated Sepharose beads by incubating 1 mg of ΔRaf-1:ER and MEK:ER H19-7 cell extracts with 100 μl of antibody-coated beads and incubating overnight at 4°C followed by extensive washing in CLB. Beads were stored at 4°C in CLB containing sodium azide until use.

To isolate Raf-binding proteins, ΔRaf-1:ER beads (15 μl of a 1:1 slurry) were incubated with 100 μg of extracts from cells treated with various compounds. Following overnight binding, the complexes were washed several times in CLB. Binding proteins were dissociated by boiling in SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer, separated on SDS–10% PAGE gels, blotted onto nitrocellulose, and analyzed by immunoblotting.

To isolate MEK binding proteins, MEK:ER beads (20 μl of a 1:1 slurry) were incubated with 100 to 500 μg of extracts from untreated or 1 μM β-estradiol-treated ΔRaf-1:ER or MEK:ER cells and analyzed as described above.

Production of GST-ΔRaf and GST pull-downs.

The kinase domain of Raf (amino acids 348 to 692) was subcloned into the mammalian glutathione S-transferase (GST) expression vector pEBG. Ten micrograms of plasmid was transfected into COS cells and allowed to express for 36 h, after which the cells were lysed and lysates (100 μg/30 μl of beads) loaded onto GST-Sepharose beads. After extensive washing in RIPA buffer, the GST-ΔRaf beads were loaded with H19-7 cell lysate as described above.

Western blotting.

Affinity-purified or immunoprecipitated proteins were separated by SDS-PAGE through 10% gels, electroblotted onto nitrocellulose, and Western blotted with the indicated antibodies. Blots were blocked at 4°C with TBST plus 5% dry nonfat milk, and antibodies were diluted in this buffer as suggested by the manufacturers. Blots were incubated in primary antibodies from 2 h to overnight at 4°C, washed in Tris-buffered saline with 0.2% Tween 20 (TBST), and incubated with the appropriate horseradish peroxidase-conjugated secondary antibody diluted in TBST plus 5% milk for 2 h at room temperature. Immunoblots were visualized using a Renaissance kit (DuPont/NEN).

Expression and purification of baculovirus-expressed Flag–c-Raf-1.

Baculovirus expressing Flag-epitope-tagged c-Raf-1 was a generous gift of Deborah Morrison (National Institutes of Health, Frederick, Md.). A 50-ml aliquot of a 2 × 10⁶ cells/ml suspension culture of Sf9 cells was infected with 2 ml of a high-titer viral stock and incubated with stirring for 96 h at 27°C. Cells were pelleted and lysed in 4 ml of RIPA buffer, insoluble material was pelleted, and Flag-Raf was immunoprecipitated from the lysate with agarose-conjugated anti-Flag antibody M2 (Sigma). After extensive washing, the immunoprecipitated Flag-Raf was incubated with 250 μg of RIPA cell lysate from ΔRaf-1:ER cells that had been treated with 1 μM estradiol or ethanol vehicle. Weakly bound proteins were removed by extensive washing in RIPA, and Flag-Raf binding proteins were eluted by boiling in PAGE sample buffer.

Transfection of COS or ΔRaf-1:ER cells.

COS or H19-7 cells were plated at a density of 1 × 10⁶ cells per well in 100-mm-diameter dishes and allowed to grow overnight in complete medium. Cells were rinsed with phosphate-buffered saline and the medium was changed to Opti-MEM immediately prior to transfection with Trans-It. Eight microliters of transfection reagent was added to 100 μl of Opti-MEM and incubated at room temperature for 5 min, and then 4 μg of plasmid DNA was added and incubated for 5 min at room temperature. The reagent-DNA mixture was added to the cells, and the cells were incubated for 4.5 to 5 h at which time the medium was changed to complete medium and the cells were allowed to grow for 24 h. After 24 h, the medium was changed to serum-free DMEM and cells were incubated overnight and treated as described above.

Immunoprecipitations with anti-ER or anti-Raf antibodies.

One microliter of anti-ER D75 or anti-Raf 22-1658 was incubated for 3 h at 4°C with 100 to 500 mg of cell extracts prepared in RIPA. Twenty microliters of a 1:1 suspension of protein A-Sepharose was added and incubated from 4 h to overnight at 4°C with constant mixing. The beads were pelleted and washed extensively. Bound proteins were dissociated by boiling in 25 μl of 2× PAGE sample buffer, and the whole sample was separated on SDS–10% PAGE gels, electroblotted, and subjected to Western blotting with the anti-ACTIVE MAPK antibody.

Immunoprecipitation of ERK5/BMK1.

Immunobeads were prepared by incubating 1 μg of ERK5/big MAPK 1 (BMK1) antibody with 20 μl of protein A-Sepharose at room temperature for 3 to 6 h. Following this incubation, beads were washed extensively in CLB and either used immediately or stored as a 1:1 suspension in CLB containing sodium azide. Cell lysates were prepared from COS or H19-7 cells grown as for the ΔRaf-1:ER binding reactions and lysed in CLB, and 100 μg of cell lysate was added to 15 μl of BMK1 antibody beads. The beads were incubated overnight at 4°C with mixing and washed four times with 500 μl of CLB. The beads were dissociated by boiling for 2 min in 2× PAGE sample buffer, separated on 10% PAGE gels, transferred to nitrocellulose, and probed with the anti-HA (12CA5) or anti-ACTIVE MAPK antibodies.

In vitro kinase assay.

ΔRaf-1:ER affinity-purified proteins or anti-pERK5 antibody immunoprecipitates were utilized in in vitro kinase reactions. Fifteen microliters of washed beads was washed in 1 ml of kinase buffer (25 mM HEPES [pH 7], 10 mM MgCl₂, 2 mM MnCl₂, 1 mM dithiothreitol, and protease and phosphatase inhibitors) and then incubated for 30 min at 30°C in 20 μl of kinase buffer containing 25 mM ATP (200 μCi of [γ-³²P]ATP) and 1 μg of purified substrate. The substrates used were GST, GST-Elk1(307–428), GST-mElk1 (307–428 A383/A389), GST-c-Jun(1–93), GST-c-myc, GST-c-Fos(210–313), GST-Ets2 (T72), GST-mEts2 (A72), GST-ATF2(1–109), six-His-MEF2C, and GST-c-max. GST fusion proteins were isolated on glutathione (GSH)-Sepharose and eluted with free GSH using established protocols (Gibco/BRL), and six-His fusion proteins were isolated on Talon Ni-resin (Clontech). Protein levels were estimated by comparison to Coomassie-stained bovine serum albumin as a standard. Kinase reactions were terminated by adding 5 μl of 6× PAGE sample buffer and boiling. The incorporation of ³²P into proteins was analyzed by SDS-PAGE of the entire reaction mixture.

Fast-protein liquid chromatography.

Whole-cell lysates from ER:Raf H19-7 cells were dialyzed against 100 mM phosphate buffer (pH 7.2) and loaded onto a Bio-Rad CM column connected to a Pharmacia fast-performance liquid chromatography apparatus. Proteins were loaded at a flow rate of 1 ml/min and eluted with a linear gradient from 0.0 to 1.0 M NaCl. One-milliliter fractions were collected and analyzed as described in the figure legends.

RESULTS

Raf binds a protein that is recognized by an anti-phospho ERK antibody.

In order to identify novel downstream effectors of Raf, we generated a Raf affinity column. The Raf bound to the column was a fusion protein consisting of the kinase domain of c-Raf-1 and the estrogen-binding domain of the estrogen receptor (ER) (ΔRaf-1:ER) (42). The activated ΔRaf-1:ER was isolated from E2-treated H19-7 cells stably expressing ΔRaf-1:ER (ΔRaf-1:ER cells) (27) and bound to a rat anti-ER antibody. A rabbit anti-rat antibody was used to link the anti-ER antibody to protein A beads.

To isolate Raf-interacting proteins, the Raf affinity column was incubated with Triton X-100-solubilized extracts from ΔRaf-1:ER cells that had been stimulated for 1 h with either 1 μM or 10 nM E2. Activation by 10 nM E2 is sufficient to promote differentiation of ΔRaf-1:ER cells, but no significant induction of ERKs 1 or 2 is observed under these conditions (27). Binding proteins were eluted from the Raf affinity column by boiling and then were analyzed by Western blotting with a specific anti-phospho (pTEpY) ERK antibody. This antibody was generated against the dually phosphorylated TEY activation domain of ERKs 1 and 2, and it also recognizes activated ERK5 and ERK7 (3) (see below). As shown in Fig. 1, a strong immunoreactive band at 97 kDa, termed p97, was detected from the Raf column eluate.

FIG. 1.

The Raf affinity column binds a protein of 97 kDa (p97) that is recognized by an anti-phospho ERK antibody. (A) p97 binds to active but not inactive ΔRaf-1:ER. ΔRaf-1:ER was isolated from ΔRaf-1:ER cells that were serum starved for 16 h and then either treated with ethanol (inactive Raf beads) or 1 μM E2 (active Raf beads) for 60 min as described in Materials and Methods. Active and inactive Raf beads were then loaded with cell extracts from untreated or E2-treated ΔRaf-1:ER cells, and some cells were pretreated for 10 min with 10 μM MEK inhibitor PD98059 (PD) as indicated. Binding proteins were eluted by boiling in PAGE sample buffer and analyzed by Western blotting with an anti-phospho pTEpY ERK(A) antibody. (B) p97 does not bind to the ER directly or to the antibodies used to isolate ΔRaf-1:ER. H19-7 cells were treated with bFGF, lysed in RIPA, and immunoprecipitated with antibodies against rat IgG or the ER (H222). Immunoprecipitated proteins were Western blotted with the anti-phospho ERK(A) antibody. The GST-ER fusion protein (GST-ER) was prepared and incubated with ΔRaf-1:ER cell extracts, washed, and Western blotted with the anti-phospho ERK(A) antibody. (C) p97 binds through the kinase domain of Raf-1. GST-ΔRaf was isolated on GSH beads, and 25 μg of isolated protein was incubated with increasing amounts of ΔRaf-1:ER cell extracts, washed, and Western blotted as described above. (D) Differentiating but not mitogenic signals increase binding of p97 to ΔRaf-1:ER. Activated ΔRaf-1:ER beads were incubated with 100 μg of cell lysates prepared from untreated (UT), EGF-, or FGF-treated H19-7 cells and analyzed by Western blotting as described for panel A. Cells were lysed with buffers containing either SDS and sodium deoxycholate (RIPA) or Triton X-100 (TX) as described in Materials and Methods. (E) ΔRaf-1:ER binds with high affinity to p97. Extracts from ΔRaf-1:ER cells treated with 1 μM E2 were bound to ΔRaf-1:ER beads and then eluted with NaCl, Empiger-BB, urea, or boiling (Cont) at the indicated concentrations. The eluates were then analyzed by Western blotting with anti-phospho ERK(A) antibodies as described above.

Several controls indicated that p97 specifically binds to activated Raf. First, the amount of p97 bound to ΔRaf-1:ER was dependent upon the in vivo activation state of the Raf used to prepare the affinity column. When the Raf affinity column was prepared using inactive ΔRaf-1:ER isolated from serum-starved cells, very little p97 was bound (Fig. 1A). This difference in p97 binding to ΔRaf-1:ER was not due to differences in the amount of ΔRaf-1:ER bound to the column, since the anti-ER antibodies used in these experiments bound both liganded and unliganded ΔRaf-1:ER with comparable affinity (reference 16 and data not shown). Furthermore, no p97 binding was detected when H19-7 cell extracts were incubated with a GST-ER fusion protein immobilized on GSH-Sepharose beads (Fig. 1B). These results indicate that p97 does not bind to the ER domain of the ΔRaf-1:ER fusion protein. Finally, to eliminate the possibility that p97 was binding directly to the anti-Rat IgG or anti-ER antibodies used to immobilize ΔRaf-1:ER, immunoprecipitations using these antibodies in the absence of ΔRaf-1:ER were performed. As shown in Fig. 1B, neither immobilized ER nor protein A-immobilized antibodies against rat IgG or the ER were capable of binding significant levels of p97. In contrast, p97 could be readily detected when immobilized, activated ΔRaf-1:ER beads were used to isolate p97 from the same extracts. Further evidence that isolation of p97 occurs through the kinase domain of Raf was provided by GST-ΔRaf-1 pull-down assays. Isolation of constitutively active GST-Raf from COS cells was utilized as an affinity resin to isolate p97 from E2-treated but not untreated ΔRaf-1:ER H19-7 cells (Fig. 1C). These results indicate that p97 binds to the Raf kinase domain in an activation-dependent manner.

p97 can be isolated from FGF- but not EGF-stimulated cells.

p97 was also isolated from H19-7 cells treated with FGF, a differentiating factor that activates the MEK-ERK1,2 cascade. However, p97 could only be detected in extracts from FGF-treated H19-7 cells when these cells were lysed in a buffer containing the strong detergents sodium deoxycholate and SDS rather than Triton X-100 (Fig. 1D). This observation suggests that activation of p97 with FGF leads to localization of activated p97 complexes in a Triton X-100-resistant cell fraction. It should be noted that the anti-phospho ERK antibody used to detect p97 only recognizes the activated pTEpY phosphorylated form of the ERKs. To determine if p97 is always stimulated in response to growth factors, we treated H19-7 cells with a mitogenic stimulus, EGF. Like FGF, EGF also activates the MEK-ERK1,2 cascade in H19-7 cells. EGF-treated extracts were loaded onto activated ΔRaf-1:ER beads, and the binding proteins were analyzed by Western blotting with the anti-phospho ERK antibody. As shown in Fig. 1D, EGF treatment of H19-7 cells did not lead to the isolation of p97 bound to Raf. These results are consistent with the hypothesis that p97 is a downstream effector of FGF- or Raf-induced differentiation but not of EGF-induced mitogenesis in H19-7 cells.

p97 binds tightly in a complex with active Raf.

To analyze the relative affinity of the binding of p97 to Raf, we attempted to elute p97 from the Raf beads using several different elution buffers (Fig. 1E). Binding of p97 to the ΔRaf-1:ER beads was not disrupted by 3 M NaCl or urea levels up to 3 M. Urea levels above 3 M eluted p97 but caused leaching of significant amounts of IgG from the beads, suggesting that this level of urea leads to disruption of the antibody–ΔRaf-1:ER complex. The zwitterionic detergent Empigen-BB has been shown to reversibly disrupt the association of Raf with 14-3-3 proteins, and no 14-3-3 can be detected following washing with buffers containing 1% Empigen-BB (44). However, washing p97-bound ΔRaf-1:ER beads with cell lysis buffer containing concentrations of Empigen-BB as high as 3% failed to release any immunoreactive p97. Similar results were obtained using several other buffers containing detergents and chaotropic agents commonly used to disrupt affinity interactions; all failed to elute p97 at levels that maintained the ΔRaf-1:ER antibody interaction. Taken together, these results show that p97 binds with high affinity to a complex containing the activated kinase domain of Raf.

p97 binds wild-type c-Raf in vitro and in cells.

In the previous studies, p97 was isolated by binding to the activated kinase domain of Raf. To determine if p97 could bind full-length c-Raf-1 containing the N-terminal regulatory domain as well as the kinase domain, we infected Sf9 cells with baculovirus expressing Flag-tagged c-Raf-1. Flag–c-Raf-1 was isolated from Sf9 cell lysates by immunoprecipitation with the anti-Flag (M2) antibody and then incubated with cell extracts prepared from untreated or 1 μM E2-treated ΔRaf-1:ER cells. Analysis of the immunoprecipitated protein by immunoblotting with anti-phospho ERK(A) antibody showed that p97 could only be detected when Flag–c-Raf-1 was incubated with cell lysates prepared from the E2-treated ΔRaf-1:ER cells (Fig. 2A). Therefore, p97 binds to full-length c-Raf-1 in an activation-dependent manner. While this activation dependence is in part a reflection of the fact that the anti-phospho ERK antibody cross-reacts only with activated ERKs, we did not detect any nonstimulated p97 bound to Raf using antibodies that recognize nonphosphorylated regions of ERKs (see below).

FIG. 2.

p97 binds to full-length Raf. (A) p97 binds c-Raf-1 from baculovirus. Sf9 cells were infected with baculovirus containing full-length Flag-tagged c-Raf-1 as described in Materials and Methods. A 25-μl aliquot of Sf9 cell lysate was loaded onto agarose beads conjugated to anti-FLAG antibody M2. The Flag-Raf beads were then incubated with 100 μg of ΔRaf-1:ER cell lysates, and the bound proteins were analyzed by Western blotting. Expression of Flag-Raf was verified by Western blotting with an anti-Raf antibody, and p97 was detected with the anti-phospho ERK(A) antibody and a general anti-ERK antibody (pan-ERK). (B) p97 can be coimmunoprecipitated with c-Raf-1. RIPA lysates (1 mg) from ΔRaf-1:ER or parental H19-7 cells were immunoprecipitated with antibodies against the ER or c-Raf-1. Immunoprecipitated proteins were eluted by boiling in PAGE buffer and Western blotted with the anti-phospho ERK(A) antibody.

To determine whether the association of p97 and c-Raf occurs in vivo, we performed immunoprecipitations with H19-7 and ΔRaf-1:ER cell extracts using antibodies against c-Raf and the ER, respectively. The bound proteins were then immunoblotted with anti-phospho ERK(A) antibody. In both cases, no p97 was found associated with Raf when extracts from untreated cells were used. However, p97 could be immunoprecipitated with anti-Raf antibody when extracts from bFGF-treated H19-7 cells were used (Fig. 2B). Similarly, p97 coprecipitated with ΔRaf-1:ER from E2-treated ΔRaf-1:ER cell extracts when an anti-ER antibody was used (Fig. 2B). These results indicate that both endogenous c-Raf-1 and endogenous p97 associate in vivo, and this association occurs in a manner that is dependent upon cellular stimulation by FGF or activated Raf.

MEKs 1, 2, and 5 are not activators of p97.

Detection of p97 by immunoblotting with the anti-phospho ERK antibody generated against the activation domains of ERK1 and ERK2 raised the possibility that p97 phosphorylation and/or Raf binding might be regulated by MEK1 or MEK2. However, several lines of evidence suggest that none of the known ERK-associated MEKs are upstream of p97. First, when Raf-binding proteins from ΔRaf-1:ER cells treated with bFGF or E2 were probed with antibodies specific for MEK1 or MEK2, we failed to detect any cross-reactive MEK proteins in the bound complex (data not shown). Second, when H19-7 cells expressing an ER-MEK1 fusion construct (ER:MEK cells) were treated with 1 μM E2 and the cell extracts were incubated with activated ΔRaf-1:ER beads, no p97 bound to Raf was detected (Fig. 3A). Furthermore, when cell extracts from E2-stimulated ΔRaf-1:ER cells were incubated with a MEK affinity column consisting of the ER:MEK fusion protein substituted for ΔRaf-1:ER, no p97 bound to MEK was detected (Fig. 3A). Similar results were obtained when extracts from FGF-treated H19-7 cells were incubated with the MEK affinity column, whereas ERKs 1 and 2 were readily detectable bound to the MEK affinity column (data not shown). These results suggest that expression of activated MEK1 in cells does not induce activation of p97, and p97 does not bind directly to activated MEK1. Western blotting of ΔRaf-1:ER binding proteins with several antibodies directed against the known MEKs failed to cross-react (Janulis and Rosner, data not shown), suggesting that none of these MEKs are present. Finally, pretreatment of ΔRaf-1:ER cells with the MEK inhibitors PD98059 (10 μM) or U0126 (30 μM) for 10 min prior to addition of 10 nM E2 had no effect on the amount of p97 bound to Raf (Fig. 1A and data not shown). Further experiments using a range of concentrations of PD98059 or U0126 also failed to inhibit the association of active p97 with Raf (Fig. 3B). Since both of these inhibitors block MEK1, MEK2, and MEK5 activity (8, 11, 22), these results suggest that p97 activation and association with Raf are independent of the known MEKs. However, it is still possible that there is a MEK-level enzyme present in our Raf-p97 complexes.

FIG. 3.

(A) p97 does not bind to MEK and is not activated by MEK. One hundred micrograms of untreated or 1 μM E2-treated ΔRaf-1:ER cell lysate was purified by Raf affinity chromatography or MEK affinity chromatography as described in Materials and Methods and analyzed by Western blotting with the anti-pTEpY ERK(A) antibody. Alternatively, 100 μg of lysate prepared from untreated or 1 μM E2-treated MEK:ER was isolated by Raf affinity chromatography and analyzed with the same antibody. The blot shown is representative of results obtained in two separate experiments and a further experiment in which 500 μg of cell lysate was purified by Raf or MEK affinity chromatography with similar results. (B) p97 activation is not inhibited by the MEK inhibitors PD98059 or U0126. ΔRaf-1:ER cells were either untreated (UT, E2) or pretreated with increasing concentrations of U0126 (0.1, 0.5, 1.0 μM) or PD98059 (10, 25, 50 μM) followed by stimulation with ethanol (UT) or 1 μM E2 (all other lanes). Cells were lysed and analyzed by Western blotting with the anti-pTEpY ERK(B) antibody directly or following purification on ΔRaf-1:ER beads.

p97 cross-reacts with several ERK-specific MAPK antibodies.

The detection of p97 on Western blots with the anti-phospho ERK antibody suggests that this protein is a phosphorylated member of the MAPK superfamily. To investigate this possibility further, we performed Western blots on the ΔRaf-1:ER-bound proteins using several different MAPK-specific antibodies (Fig. 4A). Previous studies have shown that the anti-phospho (pTEpY) ERK antibody made against ERK1 and ERK2 recognizes ERK homologues such as ERK7, but only if the protein is dually phosphorylated at both the threonine and tyrosine sites of the activation loop (3). To confirm that the recognition is not an artifact of a specific antibody, ΔRaf-1:ER-binding proteins were immunoblotted with anti-phospho (pTEpY) ERK(B) antibodies made against ERK1 and ERK2 by an independent source and two different anti-phospho (pTEpY) ERK (anti-pERK5 A,B) antibodies made against the dually phosphorylated ERK5 activation domain. As shown in Fig. 4A, all the anti-phospho ERK antibodies exhibited strong immunoreactivity against p97. These results provide strong evidence that the recognition of p97 by the anti-phospho ERK antibodies is based upon a specific interaction.

FIG. 4.

p97 cross-reacts with anti-ERK antibodies. (A) Reactivity of p97 with anti-MAPK antibodies. p97 from untreated and 1 μM E2-treated ΔRaf-1:ER cells was isolated by Raf affinity chromatography as described in Materials and Methods and analyzed by Western blotting with a series of anti-MAPK antibodies. The antibodies included anti-phospho (pTEpY) ERK antibodies from two different sources (A, B), anti-phospho (pTEpY) ERK5 antibodies from two different sources (A, B), an antibody raised against a peptide at the C terminus of ERK5 (BMK1), a general anti-ERK antibody (pan-ERK), antibodies specific to ERK2 or ERK5 that were generated against peptides within the internal domains of these proteins, an anti-JNK antibody, an anti-p38 antibody, an anti-phospho (pTPpY) JNK antibody, and an anti-phospho (pTGpY) p38 antibody, as described in Materials and Methods. (B) Reactivity of antibodies to ERK1 or ERK2. Samples from ΔRaf-1:ER cells that were either untreated or stimulated with 10 nM E2, 10 nM EGF, or 10 ng of FGF/ml were directly immunoblotted with antibodies against phospho (pTEpY) ERK or ERK2 as described in Materials and Methods. (C) Reactivity of antibodies to p38. Samples from H19-7 cells treated with E2 or sorbitol were either directly immunoblotted with p38 or phospho (pTGpY) p38 as described in Materials and Methods. (D) Reactivity of antibodies to JNK. Samples from H19-7 cells treated with E2 or anisomycin were either directly immunoblotted with antibodies against JNK or phospho (pTPpY) JNK as described in Materials and Methods.

The previously described studies suggest that p97 is a member of the ERK family of MAPKs. To test this possibility further, we determined whether p97 cross-reacts with an antibody raised against a different, nonphosphorylated domain that is shared by known members of the ERK family. As shown in Fig. 4A, this anti-pan-ERK antibody also cross-reacts with p97 in immunoblots of proteins isolated by Raf affinity chromatography from E2-stimulated ΔRaf-1:ER cells. Interestingly, no p97 was detected when extracts from unstimulated cells were used, suggesting that p97 and Raf only associate when they are in an active conformation.

p97 appears to be most closely related to ERK5. An antibody raised against a peptide at the C terminus of ERK5 (α-BMK1) also cross-reacted with p97 isolated by Raf affinity chromatography from E2-stimulated ΔRaf-1:ER cells. p97 was readily immunoprecipitated from Raf-stimulated cells using anti-BMK1; however, this antibody did not recognize p97 in Western blots of the cell extracts (Fig. 5, α-BMK1). Interestingly, the BMK1 antibody did not appear to immunoprecipitate p97 from unstimulated cell extracts. In contrast, the anti-BMK1 antibody readily detected ERK5 (p110) in both unstimulated and stimulated cell extracts by immunoblotting as well as immunoprecipitation. In addition, we also identified anti-ERK5 antibodies made against different domains of ERK5 that recognize ERK5 but do not cross-react with p97 (Fig. 4A). These ERK5-specific antibodies demonstrated that p97 does not have a completely overlapping sequence with ERK5. Taken together, these results indicate that p97 and ERK5 are related but distinct proteins.

FIG. 5.

p97 is not a degradation product of ERK5. (A) Anti-ERK5 antibodies recognize p110 ERK5 and p97. Lysates from untreated and 1 μM E2-treated ΔRaf-1:ER cells were either isolated by Raf affinity chromatography or immunoprecipitated with 1 μg of anti-ERK5 antibody (anti-BMK1) generated against a C-terminal peptide as described in Materials and Methods. Samples from the cell lysate, the Raf affinity column, and the anti-ERK5 immunoprecipitates were analyzed by Western blotting with the anti-BMK1 antibody. (B) Expression of HA-ERK5 cDNA produces the p110 ERK5 protein in COS cells. Cells were transfected with 5 μg of an expression vector containing HA-tagged ERK5. Following transfection, cells were treated with vehicle or 10% FBS for 15 min. Cells were lysed in RIPA buffer and analyzed by Western blotting directly or following immunoprecipitation with anti-BMK1 or anti-HA antibodies. (C) Expression of HA-ERK5 cDNA produces the p110 ERK5 protein in H19-7 or ΔRaf-1:ER cells. Cells were transfected with 5 μg of an expression vector containing HA-tagged ERK5. Following transfection, cells were treated with vehicle, 1 μM E2 for 1 h (ΔRaf-1:ER cells), or 10 ng of bFGF/ml for 15 min (H19-7 cells).

Several control experiments indicated that the immunoreactivity observed toward p97 was the result of specific antibody recognition. First, other isozyme-specific anti-ERK antibodies also failed to detect p97. Thus, an anti-ERK2 antibody directed against a unique peptide from the ERK2 sequence (17) did not recognize p97 (Fig. 4A) but readily recognized ERKs 1 and 2 (Fig. 4B). Recognition also appeared to be specific for the ERK subfamily of MAPKs. Anti-JNK and anti-p38 antibodies failed to cross-react with p97 (Fig. 4A), although they readily detected JNK and p38 in the same extracts (Fig. 4B and C). Finally, the recognition of p97 by the anti-phospho TEY antibodies was not due to generic recognition of phosphorylated MAPK activation domains. Although ERKs, JNKs, and p38s share the TXY activation motif, both anti-phospho (pTPpY) JNK and anti-phospho (pTGpY) p38 antibodies failed to cross-react with p97 (Fig. 4). Taken together, the selective recognition of p97 by more than five independent anti-ERK antibodies but not by antibodies against JNK or p38 provides compelling evidence that p97 is a member of the ERK family of MAPKs.

p97 is not a degradation product of ERK5.

Given the similar molecular weight of p97 and its cross-reactivity with the anti-ERK5 antibodies, we investigated whether p97 is a cellular degradation product of ERK5. Therefore, HA-ERK5 that is tagged at the amino terminus was transiently expressed from a cytomegalovirus promoter in H19-7, ΔRaf-1:ER, or COS cells (Fig. 5B and C). Ectopic expression of HA-ERK5 produced only the 110-kDa form in COS, H19-7, and ΔRaf-1:ER cells when assayed with antibodies that detect either the amino or carboxyl terminus of HA-ERK5 (Fig. 5). In addition, we have identified anti-ERK5 antibodies made against internal domains of ERK5 that recognize ERK5 but do not cross-react with p97 (Fig. 4). These results demonstrate that p97 is not a degradation product of ERK5 formed either prior to or during its isolation.

Affinity purification of p97.

To purify p97, cell lysates from ΔRaf-1:ER cells treated for 1 h with 1 μM estradiol were loaded onto a CM cation exchange column, eluted with a linear gradient of NaCl from 0 to 1 M, and probed for the presence of ERKs 1, 2, 5, and 8. As shown in Fig. 6A, when CM column fractions were incubated with ΔRaf-1:ER beads and the bound proteins were immunoblotted with anti-phospho ERK(A) antibody or anti-phospho ERK5 antibody, p97 appears in the flowthrough of the CM column (fractions 1 to 5). When the same CM column fractions were immunoprecipitated directly with anti-ERK5 antibody and the immunoprecipitates probed with the same antibody, p97 was also detected in the flowthrough, while p110 ERK5 eluted much later (Fig. 6B). In contrast, ERKs 1 and 2 could be detected by the anti-phospho ERK (A) antibody in a straight Western blot of the CM column fractions as a distinct set of bands that eluted in overlapping fractions from the column (Fig. 6B). These results show that p97 is biochemically distinct from the previously characterized ERKs. Furthermore, since p97 is found in the flowthrough of the column while ERKs 1, 2, and 5 bind to the column, this step provides a source of p97 devoid of ERKs 1, 2, and 5 for subsequent analysis. Thus, immunoprecipitation of the CM column flowthrough with anti-ERK5 antibody yields purified p97.

FIG. 6.

Affinity purification of p97. (A) p97 elutes in the flowthrough of a CM cation exchange column. The first five fractions (1 ml each) eluted from the CM column were pooled, and 500 μl of the pooled sample was subjected to Raf affinity purification and analyzed by Western blotting with the anti-phospho ERK(A) or anti-BMK1 antibodies. (B) ERKs 1, 2, and 5 bind to the CM column and elute with NaCl. In the upper panel, 500-μl fractions from pooled fractions as used in panel A were immunoprecipitated with the anti-phospho ERK5 antibody (pERK5 Ab) and analyzed by Western blotting with the same antibody. In the lower panel, 50-μl aliquots of the pooled chromatography fractions were analyzed directly by Western blotting with the anti-phospho ERK antibody.

p97 is a kinase for Elk-1 and other transcription factors.

Previous studies in H19-7 cells demonstrated that FGF and Raf activate an Elk-1 kinase by a MEK-independent pathway (8). To determine whether p97 is an Elk-1 kinase, we assayed proteins bound to the Raf affinity column for phosphorylation of GST–Elk-1 in vitro. Increased phosphorylation of GST–Elk-1 was observed only in samples that contained significant levels of immunoreactive p97 (Fig. 7A). Importantly, neither GST–Elk-1 phosphorylation nor p97 binding to ΔRaf-1:ER was inhibited when cells were pretreated with the MEK inhibitor PD98059. These results show that ΔRaf-1:ER binds a MEK-independent Elk-1 kinase activity, and this activity correlates with the level of bound p97. To confirm that p97 was responsible for the Elk-1 kinase activity, p97 in samples eluting in the flowthrough of the CM column was affinity purified by immunoprecipitation with anti-ERK5 antibody and then incubated with GST–Elk-1 in an in vitro kinase reaction. As shown in Fig. 7B, significant levels of phosphorylated GST–Elk-1 were detected. When column fractions containing ERK1, ERK2, and ERK5 (fractions 31 to 35) were assayed directly, phosphorylation of GST–Elk-1 was also detected. However, when column fractions containing ERK1, ERK2, and ERK5 (fractions 31 to 35) were first immunoprecipitated with anti-BMK1 antibody and the ERK5-containing immunoprecipitates were then incubated with GST–Elk-1, little or no detectable Elk-1 kinase activity was detected. These data, taken together with the earlier Raf binding data, indicate that p97, like ERK1 and ERK2 but not ERK5, is an Elk-1 kinase.

FIG. 7.

p97 has kinase activity towards several transcription factors. (A) p97 phosphorylates Elk-1, and this phosphorylation cannot be inhibited by PD98059. (Top) Raf affinity-purified samples were prepared from untreated (UT) or 10 nM or 1 μM E2-treated ΔRaf-1:ER cells that had been untreated or pretreated for 10 min with the MEK inhibitor PD98059 (PD 1 μM E2) as indicated. (Bottom) H19-7 cells were left untreated or treated with 10 ng of bFGF/ml with and without pretreatment with PD98059 (PD FGF). ΔRaf-1:ER cells were lysed in RIPA buffer, while H19-7 cells were lysed in either Triton X-100 buffer (CLB) or RIPA lysis buffer (all other lanes) as described in Materials and Methods. p97 was affinity purified on a Raf affinity column and assayed for kinase activity with 1 μg of GST–Elk-1 as a substrate. The entire reaction mixture was loaded onto a 7.5% PAGE gel and analyzed by autoradiography. (B) p97 isolated from CM column flowthrough has Elk-1 kinase activity. ΔRaf-1:ER H19-7 cells treated with 1 μM E2 were lysed and subjected to CM column chromatography. p97 in the column flowthrough was purified by Raf affinity chromatography (left panel) and assayed in an in vitro kinase assay with GST–Elk-1 as a substrate. The samples were also Western blotted with the anti-phospho Erk(A) antibody to detect the presence of p97 (Fig. 6). Other column fractions were subjected to immunoprecipitation with the anti-BMK1 antibody, and the bound proteins were assayed for in vitro kinase activity using GST–Elk-1 as a substrate. These samples were also Western blotted with anti-pan-ERK antibodies to identify fractions containing the Elk-1 kinases ERK1 and ERK2 (Fig. 6).

To further define its substrate profile, p97 that was affinity purified from the CM column flowthrough was assayed in an in vitro kinase assay with an array of potential substrates. As shown in Fig. 8, p97 was capable of phosphorylating GST–Elk-1, GST–c-Myc, GST–c-Fos, GST–c-Max, GST–Ets-2, and MBP but failed to significantly phosphorylate GST, GST–c-Jun, or histones H1 or H2b. In addition, the phosphorylation of GST–Elk-1 and GST–Ets-2 occurred at transactivation sites, since no significant phosphorylation of GST–Elk-1 or GST–Ets-2 mutants that were missing key transactivation-dependent phosphorylation sites was observed. ERK5, which has a very limited substrate specificity, phosphorylates and activates MEF2C (24), and stress kinases from the JNK and p38 MAPK families readily phosphorylate ATF-2 (18, 21). In contrast, neither MEF2C (Fig. 9) nor ATF-2 (Janulis and Rosner, data not shown) are substrates of p97. These results indicate that the substrate specificity of p97 most closely resembles the broad substrate profile of ERK1 and ERK2 rather than that of ERK5 or the other MAPK superfamily members.

FIG. 8.

p97 phosphorylates a number of transcription factors. (A) Phosphorylation of GST, mutant GST–Elk-1 (A383,A389), GST–Elk-1, GST–c-Jun, GST–c-Myc, GST–c-Fos, and GST–c-Max. A 500μl aliquot of CM column flowthrough was immunoprecipitated with anti-BMK1 antibody and the in vitro kinase activity toward several substrates was determined. Approximately 1 μg of each substrate was incubated with immunoprecipitated p97, and incorporation of ³²P into the protein was determined by autoradiography. The arrows indicate the location of the different substrates stained by Coomassie blue. (B) Phosphorylation of MBP mutant Ets-2 (A72), Ets-2, histone 1, and histone 2. Samples were assayed as described above.

FIG. 9.

EGF but not FGF activates ERK5. (Top) EGF stimulates phosphorylation of MEF2C. Cells were treated for 15 min with 10 ng of FGF/ml, 10 nM EGF, 1 μM E2, or untreated, and the lysates (100 μg/ml) were immunoprecipitated with anti-phospho ERK5 antibody. The immunoprecipitates were assayed for in vitro kinase activity using 1 μg of His-MEF2C as a substrate. Lysates from untreated and E2-treated cells containing p97 isolated by Raf affinity binding were also assayed for kinase activity in an analogous manner. (Bottom) EGF stimulates phosphorylation of the TEY activation domain of ERK5. The same blot as in the top panel was probed with anti-phospho ERK5 antibody as previously described.

p97 has a unique activation profile.

p97 can also be distinguished from ERK5 on the basis of its activators. As shown in Fig. 9, a 15-min incubation with EGF but not FGF selectively stimulates TEY phosphorylation and ERK5 kinase activity in H19-7 cells. This pattern is in direct contrast to that of p97, which is activated by FGF but not EGF. Yet both factors activate ERKs 1 and 2 in H19-7 cells (27). Thus, p97 displays a more selective activation response than the classic ERKs, and its stimuli are distinct from those of ERK5 in these cells. Since EGF is a mitogen and FGF is a differentiating factor in these cells, the selective activation of p97 versus ERK5 provides a potential mechanism for achieving signaling specificity by two distinct but similar growth factors.

p97 can be isolated from other cell lines.

To determine if p97 expression is limited to H19-7 cells, p97 was isolated by Raf affinity binding from extracts of several cell lines that had been serum starved or treated with 10% serum for 15 min (Fig. 10A). We observed no immunoreactivity at 97 kDa using anti-phospho ERK antibody on Western blots of samples prepared from serum-starved or serum-stimulated RAT1 cells. Since this assay detects only activated, TEY-phosphorylated p97, it is possible that inactive p97 is present but undetected in Rat1 cells. In contrast, Western blotting of samples prepared from serum-stimulated NIH 3T3 cells showed some immunoreactive p97. We have also isolated p97 from macrophages, keratinocytes, and other cell lines (Janulis, Trakul, and Rosner, data not shown). Thus, p97 expression and activation is not restricted to cells of neuronal lineage.

FIG. 10.

p97 can be found in NIH 3T3 cells and in primary hippocampal neuronal cultures. (A) p97 can be detected in NIH 3T3 cells. Cell lysates were prepared from untreated or serum-treated RAT1 or NIH 3T3 cells, and p97 was isolated by Raf affinity purification. Samples were analyzed by Western blotting with the anti-phospho ERK(A) antibody. (B) p97 can be detected in FGF- but not EGF-treated primary hippocampal cultures. Hippocampi were removed from E17 rats and cultured as described in Materials and Methods. Cells cultured in serum-free N2 medium for 1 day were either untreated or treated with 10 ng of bFGF/ml or 10 nM EGF for 15 min. Cells were pretreated for 10 min with 10 μM PD98059 (PD) as indicated. Following treatment, cells were lysed in RIPA, incubated with ΔRaf-1:ER beads, and analyzed by Western blotting with anti-phospho ERK(A) antibodies.

p97 is selectively activated by FGF but not EGF in primary embryonal hippocampal neural cultures.

The previous studies were done primarily with a conditionally immortalized neuronal cell line (H19-7) that was derived from E17 rat hippocampal cultures. To determine whether the selective activation of p97 by FGF but not EGF was an artifact of this cell line or reflected signaling cascades in primary cells, we analyzed FGF and EGF signaling pathways in rat hippocampal E16 neural cultures. Like H19-7 cells, these cultures were composed primarily of nestin-expressing cells. Hippocampal cultures were stimulated for 15 min with EGF or FGF. p97 was then isolated by Raf affinity chromatography and assayed by immunoblotting with anti-phospho ERK antibody. As shown in Fig. 10B, activated p97 was isolated only in response to FGF, and this activation was not inhibited by pretreatment with PD98059. Thus, these results indicate that selective activation of p97 by growth factors occurs in primary neural cells as well as cultured cell lines.

DISCUSSION

Growth factors for tyrosine kinase receptors activate common signaling pathways but often elicit different biological outcomes, suggesting that there are other modulating factors. Previous studies have indicated that Raf, the upstream activator of the classic MAPK (ERK1,2) signaling cascade, also activates a distinct downstream target. We now describe the affinity purification and characterization of p97, a novel ERK5-related member of the MAPK superfamily that is Raf associated and activated upon stimulation by Raf. Upon brief exposure in neuronal H19-7 cells, p97 is selectively responsive to FGF, a differentiating factor, but not EGF, a mitogenic factor for these cells. In contrast, EGF but not FGF specifically activates ERK5, and p97 is recognized by generic anti-ERK antibodies and anti-ACTIVE ERK antibodies but not antibodies selectively directed against ERK1, ERK2, or members of the JNK and p38 MAPK families. The p97 protein can be separated both chromatographically and by size (97 kDa) from all other known members of the MAPK superfamily. In its active state, p97 tightly associates with Raf in vivo but, unlike ERKs 1, 2 and 5, p97 is not inhibited by the MEK inhibitors PD98059 or U0126. p97 phosphorylates Elk-1 and Ets-2 at critical transactivation sites (19, 33) as well as Fos, Myc, Max, and MBP; however, it does not significantly phosphorylate ATF-2, c-Jun, or MEF2C. Finally, the activation of p97 varies with stimulus and cell type. These results suggest that p97 is a component of a previously undefined signaling pathway capable of transmitting differentiation-specific signals from Raf.

The existence of another ERK responsive to Raf that has similar substrate specificity to ERK1 and ERK2 might seem redundant. However, previous studies have shown that Raf activation of the MEK-independent Elk-1 kinases occurs with different kinetics than that of the MEK-dependent (ERK1,2) kinases (8). Thus, it is likely that substrates that are common between p97 and other ERKs are phosphorylated by isozymes that differ in kinetics of activation, amplitude, or localization within the cell. Furthermore, we have only tested a small subset of potential substrates that appear to be common substrates for a number of MAPKs. There are undoubtedly other substrates that are uniquely phosphorylated by different ERK isozymes in addition to these common substrates. Finally, the combination of transcription factors phosphorylated by a particular ERK isozyme in response to different stimuli could lead to unique outcomes.

Although ERK5 is slightly larger than p97, several lines of evidence demonstrate that p97 is not a direct degradation product of ERK5. First, degradation during processing of the cells or during purification is usually independent of cell treatment before lysis, but we isolated p97 only from cells treated with bFGF or E2 but not EGF or buffer. Second, p97 is recognized by the anti-BMK1 antibody that cross-reacts with the C terminus of ERK5, arguing against C-terminal deletion by either degradation or alternative splicing. Consistent with this result, ectopically expressed ERK5 that was HA tagged at the N terminus was isolated by immunoprecipitation from H19-7 cells only as a full-length, 110-kDa band. Furthermore, in cells expressing ERK5 ectopically, no increase in the p97 band was detected with the anti-BMK1 antibody that should recognize ERK5 truncated at the N terminus. Finally, at least one antibody that recognizes ERK5 within the internal domain does not recognize p97 under denaturing conditions, indicating that limited degradation or splicing of either the N-terminal or C-terminal domains of ERK5 is insufficient to explain the data. Taken together, these results indicate that p97 is not generated by proteolytic degradation of ERK5 during workup and is not a minimal N- or C-terminally truncated splice variant of ERK5. However, we cannot exclude the possibility that p97 is an internal splice variant of ERK5. Given the distinct substrate profiles of both enzymes and their differential regulation by related growth factors, ERK5 and p97 are clearly biochemically and functionally distinct enzymes. Therefore, whether it is a splice variant or a closely related isozyme, p97 must have a unique physiological role and thus functionally corresponds to the eighth member of the ERK subfamily of MAPKs.

ERK5, also termed big MAPK 1 (BMK1) (30, 48), has only recently been recognized as a key mediator of growth, osmolarity, and other physiological processes. Initially, ERK5 was shown to be activated in response to redox signals (1). More recently, ERK5 was implicated as a direct mediator of growth. Thus, EGF activates ERK5 in HeLa cells, and disruption of the pathway inhibits DNA synthesis (25). Nerve growth factor has also been shown to activate ERK5 in PC12 cells, where it phosphorylates the Ets-domain transcription factor Sap1 and activates the serum response element (22). Finally, ERK5 is activated by at least one G protein-coupled receptor, the muscarinic receptor, whereupon it phosphorylates MEF2C and activates the c-jun promoter (24, 34). A similar cascade appears to be utilized by the oncogenic kinase cot (7). cot activates the c-jun promoter through JNK-dependent and -independent pathways, and the latter pathway also involves ERK5. The precise signaling pathway by which ERK5 is activated is still being elucidated. MEK5 is the upstream activator of ERK5 (13, 48). Although MEK5 can bind to Raf, and Raf may under certain conditions potentiate ERK5 activation, Raf is not a direct activator of ERK5 (12, 48). Instead, MEKK3 has recently been implicated as the upstream activator of ERK5 in response to growth factors (6). Finally, Ras and Src were identified as upstream activators of ERK5 in response to growth factors (25) and oxidative stress (2), respectively. Thus, it appears that ERK5 is activated by a cascade that is distinct from the Ras-Raf pathway, consistent with a role in different physiological endpoints.

Although ERK5 was the first member of the ERK family to be characterized as a big MAPK, ERK7 and p97 also have regions in addition to the kinase domain that could have both regulatory and scaffolding functions. Deletion of the C-terminal domain from ERK5 results in activation of the enzyme, suggesting that it has a negative regulatory function (48). This domain also has a putative cytoskeletal-binding motif and functions as a transcriptional activator in conjunction with MEF2D in lymphocytes (23). The C-terminal tail of ERK7 plays a key regulatory role and is required for constitutive activation of the kinase, nuclear localization, and growth inhibition (3). These extended ERK domains might also function as scaffolding proteins. For example, Pbst2p, a MAPKK in the Saccharomyces cerevisiae Sho1p-dependent high-osmolarity glycerol response pathway, has an N-terminal SH3-containing domain that acts as a scaffold for other members of the MAPK cascade (39). Similarly, it has been shown that the amino-terminal extension of JNKK1 interacts with upstream and downstream components of the cascade (47). The observation that p97 is also a BMAPK and binds so tightly to Raf suggests that this interaction might be mediated by a comparable domain. Alternatively, Raf may interact with p97 indirectly via separate proteins with scaffolding functions.

Surprisingly, we were unable to directly detect p97 isolated from unstimulated cells. Since anti-phospho-ERK antibodies were used originally to identify p97, we would not expect to detect inactive p97 with these reagents. Furthermore, it is possible that Raf may bind only the activated form of p97. However, at least two antibodies that recognized p97 from stimulated but not unstimulated cells were generated against epitopes that should be independent of the activation state of ERK. Although the reason for this lack of cross-reactivity is not clear, one possible explanation is that the epitopes recognized by these antibodies are inaccessible in the absence of a Raf stimulus.

Although the studies conducted here were primarily done in neuronal cells, p97 is likely to be a major intermediate in a number of other cell systems as well. Raf-activated signaling cascades independent of MEK have been identified in a variety of other tissues including cardiac muscle, differentiating macrophages, and proliferating skeletal muscle cells. It had been previously reported that activation of ΔRaf-1:ER in the macrophage cell line RAW leads to the differentiation of these cells (15). We have isolated p97 from stimulated RAW cells as well as primary macrophages, and the pattern of induction suggests that p97 may mediate macrophage differentiation (Janulis, Trakul, Ostrowski, and Rosner, unpublished data). The binding of p97 with Raf also correlated with the Raf-dependent but MEK-independent signaling pathway leading to differentiation in H19-7 cells. These results suggest that one of the ways proliferative and differentiation signals diverge at the level of Raf is through the formation of specific multimolecular complexes with unique downstream effectors. ERK5 and p97 are likely to be two of these effectors. The selective activation of p97 versus ERK5 provides a potential mechanism for achieving signaling specificity by two distinct but similar growth factors.

There are now several examples in the literature of specific cellular outcomes dependent upon activation of different isozymes within a particular family of enzymes. Thus, in H19-7 cells and primary rat embryonal hippocampal cultures, EGF selectively activates protein kinase Cζ (PKCζ) by a PDK-dependent pathway, whereas FGF activates PKCδ, and these PKCs are required for the differential activation of ERKs 1 and 2 and opposing growth phenotypes displayed by these factors (9, 10). Different PKC isozymes have been shown to lead to different endpoints in a variety of other systems as well. For example, studies using knockout mice have shown that PKCγ (32) is required for the neuropathic pain syndrome incurred after partial sciatic nerve sectioning, and PKCβ suppresses maximal interleukin-6 production in mast cells (36). Like PKCs, different phosphatidylinositol-3-kinase isozymes are responsible for mitogenesis versus migration in macrophages (45). Similarly, related phosphatidylinositol-4,5-bisphosphate kinases differ in subcellular localization, substrate specificity, and physiological function (26). The results presented here provide the first example of how different ERK isozymes mediate specificity in related signaling cascades.

The pattern of p97 activation provides compelling evidence for a second MAPK signaling pathway downstream of Raf. The physiological relevance of this work is supported by the fact that all of the p97 studies were done with endogenous enzyme isolated from both cell lines and primary cells. Taken together, these results identify a new signaling module that is distinct from the classic Raf-MEK1,2-ERK1,2 kinase cascade and can be selectively activated by growth factors with discrete biological outcomes.