Characterization of a protein kinase that phosphorylates serine 189 of the mitogen-activated protein kinase homolog ERK3.

A novel protein kinase activity present in nuclear and cytosolic extracts has been identified and partially purified as a consequence of its tight binding to and phosphorylation of the extracellular signal-regulated protein kinase (ERK) 3. This novel protein kinase is inactivated by treatment with phosphoprotein phosphatase 2A. The ERK3 protein kinase was immunologically distinct from mitogen-activated protein (MAP) kinase/ERK kinases (MEK) 1 and 2 which phosphorylate the ERK3-related MAP kinases ERK1 and ERK2. This ERK3 kinase phosphorylated a single site on ERK3, Ser, comparable to Thr, one of the two activating phosphorylation sites of ERK2. To test the specificity of the ERK3 kinase, mutants of ERK3 and ERK2 were made in which the phosphorylated residues were exchanged. The double mutant S189T,G191Y ERK3, in which the phosphorylated residues from ERK2 replaced the comparable residues in ERK3, was phosphorylated by the ERK3 kinase but only on threonine. The ERK3 kinase did not phosphorylate ERK2 or ERK2 mutants. These findings indicate that although the ERK3 kinase is highly specific for ERK3, it does not recognize tyrosine, a feature that distinguishes it from MEKs that phosphorylate other ERK/MAP kinase family members.

A novel protein kinase activity present in nuclear and cytosolic extracts has been identified and partially purified as a consequence of its tight binding to and phosphorylation of the extracellular signal-regulated protein kinase (ERK) 3. This novel protein kinase is inactivated by treatment with phosphoprotein phosphatase 2A. The ERK3 protein kinase was immunologically distinct from mitogen-activated protein (MAP) kinase/ ERK kinases (MEK) 1 and 2 which phosphorylate the ERK3-related MAP kinases ERK1 and ERK2. This ERK3 kinase phosphorylated a single site on ERK3, Ser 189 , comparable to Thr 183 , one of the two activating phosphorylation sites of ERK2. To test the specificity of the ERK3 kinase, mutants of ERK3 and ERK2 were made in which the phosphorylated residues were exchanged. The double mutant S189T,G191Y ERK3, in which the phosphorylated residues from ERK2 replaced the comparable residues in ERK3, was phosphorylated by the ERK3 kinase but only on threonine. The ERK3 kinase did not phosphorylate ERK2 or ERK2 mutants. These findings indicate that although the ERK3 kinase is highly specific for ERK3, it does not recognize tyrosine, a feature that distinguishes it from MEKs that phosphorylate other ERK/MAP kinase family members.
The ERK/MAP 1 kinase pathway is stimulated by numerous hormones and growth factors and its activation is associated with increased proliferative and differentiated functions of cells (1)(2)(3)(4)(5). The importance of intracellular processes thought to be regulated by the MAP kinases has focused attention on understanding the control of this pathway. The MAP kinase kinases, also known as MAP/ERK kinases or MEK1 and MEK2, originally discovered by Ahn and Krebs, are dual-specificity protein kinases known to activate the MAP kinases ERK1 and ERK2 in a highly selective manner (6 -8). The MAP kinases, on the other hand, are pleiotropic, phosphorylating many substrates throughout the cell (reviewed in Ref. 3). Kinase cascades containing a MEK and an ERK/MAP kinase are present in multiple pathways in yeast and have been reiterated in mammalian cells (1,9). Although mechanisms regulating the similar, but parallel mammalian pathways are less well characterized, the activation of a multipotential ERK/MAP kinase by a highly specific MEK is the common feature of all the related cascades.
ERK1 and ERK2 are phosphorylated on two sites separated by a single residue in the phosphorylation lip at the mouth of their active sites (10,11). Phosphorylation of both Tyr 185 and Thr 183 on ERK2 and comparable residues on ERK1, catalyzed by the dual specificity protein kinases, MEK1 and MEK2, is required for high activity (10,(12)(13)(14)(15). Because of their exquisite specificity, MEK1 and MEK2 are not able to phosphorylate other MAP kinase-related enzymes such as Jun-N-terminal kinase/stress-activated protein kinase (JNK/SAPK) or p38 MAP kinase, even though the phosphorylation sites are in comparable positions in the sequence (1).
Much less is known about the protein kinase ERK3. It was cloned in the same cDNA library screen as ERK1 and ERK2 (16) and has greater sequence identity to ERK1 and ERK2 than do the JNK/SAPKs or p38 MAP kinase. However, three important features distinguish ERK3 from the other family members. First, it lacks the tyrosine phosphorylation site that is absolutely conserved among those other related kinases. Second, ERK3 is a constitutively nuclear protein kinase (17). Third, it apparently has a very restricted substrate specificity, because it does not phosphorylate any of the known MAP kinase substrates. As no ERK3 substrates are known, its regulation has been difficult to define.
To understand more about the regulation of ERK3, we have examined the phosphorylation of ERK3 by MEK family members, and find that ERK3 is a poor substrate for MEK1, MEK2, MKK4, and MEK5 (18). We have identified a novel protein kinase activity in nuclear and cytosolic extracts that binds very tightly to the catalytic domain of ERK3 and phosphorylates it selectively. This ERK3 kinase phosphorylates a single site on ERK3, Ser 189 , which is comparable to Thr 183 , one of the activating phosphorylation sites of ERK2.

MATERIALS AND METHODS
Cell Culture-PC12 and 293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 2 mM glutamine. Prior to mitogen stimulation, PC12 and 293 cells were maintained in Dulbecco's modified Eagle's medium without serum for 4 h and 18 -20 h, respectively, then treated with nerve growth factor (NGF, 100 nM, 15 min), epidermal growth factor (100 ng/ml, 5 min), or phorbol ester (100 nM, 20 min, Sigma) as indicated.
Subcellular Fractionation-Extracts from rabbit muscle and from NGF-stimulated PC12 cells were prepared according to the method of Seger et al. (14) for analysis of ERK1 and ERK2 phosphorylating activities. PC12 cells and 293 cells were fractionated into cytosolic and nuclear fractions as described by Dignam et al. (19) with modifications described previously (17). ERK3 was highly enriched in the nuclear fraction.
Purification of the Kinase Activity That Phosphorylates ERK3-Extracts were fractionated by chromatography on Q Sepharose, S Sepharose, Mono Q, or Mono S. Protein kinase activities in fractions eluted with a 0 -0.4 M NaCl gradient in MEK purification buffer (14) were assayed with 10 g/ml ERK2 or ERK3 as substrate as described below. For affinity chromatography, GST-ERK3 or GST-ERK3⌬Ct bound to glutathione-agarose beads was incubated with either crude cell extracts or fractions of the ERK3 kinase, partially purified from rabbit muscle or NGF-treated PC12 cells, for 2 h at 4°C with shaking. The glutathioneagarose beads were washed 5 ϫ with 1 ml of 1 M NaCl, 20 mM HEPES, pH 7.5, 0.05% Triton X-100, 1 mM EDTA and once with 1 ml of 50 mM HEPES pH 8.0. The ability of bound kinase to phosphorylate ERK3 was measured by in vitro phosphorylation assays (see below).
Treatment of the ERK3 Kinase with Phosphoprotein Phosphatase 2A (PP2A)-The ERK3 kinase adsorbed to GST-ERK3⌬Ct on glutathione agarose beads was washed once with 1 ml of phosphatase assay buffer (50 mM HEPES, pH 7.5, 0.5% bovine serum albumin, 1 mM dithiothreitol). Aliquots of beads were then mixed with 2.5 or 5 g/ml of PP2A in phosphatase assay buffer either without or with okadaic acid (5 M, Moana Bioproducts) at room temperature with shaking for 45 min. The reactions were stopped with 10 mM sodium phosphate and 5 M okadaic acid and the beads were washed twice with 1 ml of 50 mM HEPES, pH 8.0. The ERK3 kinase activity on the beads was measured as described above.
Other Methods and Reagents-Antibodies to ERK3 and isoform selective antibodies recognizing MEK1 and MEK2 were reported previously (17,21). Phosphopeptide mapping and phosphoamino acid analysis were described before (22). The catalytic subunit of PP2A was purified from bovine heart (23). Protein kinase C was partially purified from rat brain as described previously (24) which yields a mixture of isoforms. Recombinant protein kinase C␣ purified from Sf9 cells infected with the recombinant baculovirus was kindly provided by B. Singer (University of Texas Southwestern).

Chromatographic Analysis of Activities That Phosphorylate
ERK3-Rabbit muscle extracts, a source of activated MEK1 and MEK2 (25), were fractionated on Q Sepharose and the resulting fractions were assayed for their ability to phosphorylate ERK3 and ERK2 (Fig. 1A). The predominant peak of kinase activity phosphorylating ERK3 was found in fractions 42-52, and lacked activity toward ERK2. An activity with similar specificity and chromatographic behavior was also detected in Mono Q fractions 12-17 from extracts of NGF-stimulated PC12 cells (Fig. 1B). Q Sepharose fractions 47-53 containing ERK3 phosphorylating activity from rabbit muscle extracts, referred to hereafter as the ERK3 kinase, were pooled and the kinase was further purified on Mono S (Fig. 1C). A single peak of activity was detected. Q Sepharose fractions 20 -42, which contained the major peak of activity phosphorylating ERK2 from rabbit muscle extracts, also displayed a small amount of activity toward ERK3. Other studies indicated that MEK1 and MEK2 were both contained in the major peak of activity phosphorylating ERK2 (26). 2 However, these enzymes contributed little of the ERK3 phosphorylating activity because MEK1 and MEK2 immunoprecipitated from extracts of NGFstimulated PC12 cells phosphorylated ERK2 but not ERK3 FIG. 1. Chromatographic analysis of activities that phosphorylate ERK3. A, rabbit muscle extracts were fractionated on Q Sepharose and assayed for ERK2 (q) and ERK3 (ç) phosphorylating activities as described under "Material and Methods." B, extracts of NGF-stimulated PC12 cells were fractionated on Mono Q and assayed as in A. C, fractions 47-53 containing the ERK3 kinase activity from the Q Sepharose profile in A were pooled and fractionated on Mono S, and the resulting fractions were assayed as in A.
(data not shown). The activities in Q Sepharose fractions 32-40 from rabbit muscle extracts that phosphorylated ERK3 were not further characterized, but may have been due to protein kinase C, because protein kinase C eluted in this region of the gradient and was found to phosphorylate ERK3 in vitro (data not shown).
The ERK3 Kinase Binds Tightly to the ERK3 Catalytic Domain-The ERK3 kinase could be distinguished from other protein kinases based on its tight binding to the catalytic domain of ERK3 ( Fig. 2A). GST-ERK3 or GST-ERK3⌬Ct bound to glutathione-agarose beads was incubated with the ERK3 kinase activity from rabbit muscle that had first been partially purified on Q Sepharose and S Sepharose. The bound protein kinase activity was measured by its ability to phosphorylate GST-ERK3 or GST-ERK3⌬Ct on the beads (Fig. 2A). The ERK3 kinase activity was not eluted with concentrations of NaCl up to 1 M, with 1% Triton X-100 or with 1 M MgCl 2 . The ERK3 kinase did not bind to GST-ERK2. The tight association of the ERK3 kinase with ERK3 has a parallel in the binding of JNK/ SAPK to c-Jun. The binding requires only the kinase domain of ERK3 as deleting the C-terminal domain did not eliminate binding to the ERK3 kinase. Because protein kinase C was able to phosphorylate ERK3 in vitro, we tested its capacity to bind to GST-ERK3. Neither rat brain protein kinase C nor recombinant protein kinase C␣ bound to GST-ERK3 (data not shown), indicating that protein kinase C is not the ERK3 kinase.
Because it was difficult to elute the ERK3 kinase from the GST-ERK3 on glutathione-agarose beads, we ascertained if the activity that was bound to GST-ERK3 on beads would phosphorylate exogenously added ERK3. As shown in Fig. 2B, the ERK3 kinase bound to GST-ERK3⌬Ct on beads phosphorylated not only bound GST-ERK3⌬Ct but also added His 10 -ERK3⌬Ct, which is different in size from GST-ERK3⌬Ct. It seemed unlikely that the ERK3 kinase was ERK3 itself because ERK3 autophosphorylation is intramolecular not intermolecular (17). However, it was possible that the protein bound to ERK3 was not an ERK3 kinase but an activator that accelerated ERK3 autophosphorylation. To demonstrate that the ERK3 kinase was not ERK3 or an activator of ERK3 autophosphorylation, a catalytically defective mutant, D171A ERK3, that neither autophosphorylates nor is phosphorylated by wild type ERK3 in vitro (17) was tested as a substrate for the ERK3 kinase. The ERK3 kinase bound tightly to GST-D171A ERK3⌬Ct and it phosphorylated GST-D171A ERK3⌬Ct or added His 10 -D171A ERK3⌬Ct as well as the wild type protein (Fig. 2, A and B), indicating that the protein bound to GST-ERK3 is an ERK3 protein kinase. Phosphoamino acid analysis showed that the ERK3 kinase phosphorylated ERK3 on serine (Fig. 2C).
Subcellular Localization of the Kinase That Phosphorylates ERK3-Because ERK3 is primarily in the nucleus (17), the subcellular distribution of the ERK3 kinase was examined. Cytosolic and nuclear extracts from multiple cell types were tested for the ERK3 kinase activity by binding to GST-ERK3⌬Ct on beads and assay of the bound material by phosphorylation of GST-ERK3⌬Ct. Activity that bound tightly to ERK3⌬Ct and phosphorylated it was found in both cytosolic and nuclear extracts of PC12 and 293 cells (Fig. 3), and in extracts of other cell lines such as NIH3T3, Cos, and Jurkat T cells (data not shown). The distribution of the activity between cytosolic and nuclear fractions was not changed by extracellular stimuli including epidermal growth factor, NGF, or phorbol ester. These agents activate ERK1 and ERK2 and cause their translocation to the nucleus (27), but may not be physiological regulators of ERK3.
Regulation of the ERK3 Kinase-To test the possibility that the ERK3 kinase, like MEK1 and MEK2, is regulated by phosphorylation, its activity was measured before and after treatment with PP2A. The initial rate of ERK3 phosphorylation by the ERK3 kinase was reduced 85-90% by a 45-min treatment with PP2A (Fig. 4). Preincubation of PP2A with okadaic acid blocked the inactivation of the ERK3 kinase, demonstrating that loss of activity is due to dephosphorylation of the ERK3 kinase preparation.
Sites Phosphorylated on ERK3 by the ERK3 Kinase-We determined previously that ERK3 autophosphorylated in vitro and was phosphorylated in intact cells on Ser 189 (17), the residue comparable to Thr 183 , one of the two activating phosphorylation sites in ERK2 (Fig. 5A). The stoichiometry of phosphorylation of ERK3 by the ERK3 kinase was 0.7 mol phosphate/mol ERK3, consistent with a single site of phosphorylation. In comparison, incorporation due to autophosphorylation was never greater than 0.04 mol of phosphate/ mol of ERK3 even after overnight incubation. To determine if Ser 189 was the site phosphorylated by the ERK3 kinase, this residue was mutated to alanine (S189A ERK3) or glutamic acid (S189E ERK3) (Fig. 5A). The ERK3 kinase bound to GST-S189A ERK3 and GST-S189E ERK3 on beads as determined by its ability to phosphorylate added His 10 -ERK3⌬Ct (data not shown). However, it no longer phosphorylated GST-S189A ERK3 or GST-S189E ERK3 to which it was bound (Fig. 5, B  and C). The ERK3 kinase also did not phosphorylate added His 10 -S189A or S189E ERK3⌬Ct (data not shown). These data FIG. 3. The ERK3 kinase activity is present in both cytosolic and nuclear fractions of PC12 and 293 cells. The ERK3 kinase activity from nuclear (N) or cytosolic (S) fractions was adsorbed to GST-ERK3⌬Ct on glutathione-agarose beads as described under "Material and Methods." The mobilities of GST-ERK3⌬Ct and standard markers resolved as in Fig. 2 are indicated.   FIG. 4. Treatment of the ERK3 kinase with PP2A decreases its protein kinase activity. The ERK3 kinase bound to GST-ERK3⌬Ct on glutathione-agarose beads was untreated (Ϫ) or treated with PP2A (ϩ, 2.5 g/ml; ϩϩ, 5 g/ml), or with PP2A plus 5 M okadaic acid (OA), and its ERK3 phosphorylating activity was then measured. Top, GST-ERK3⌬Ct was resolved by SDS-PAGE and an autoradiogram is shown. The position of GST-ERK3⌬Ct is indicated. Bottom, bar graph quantitating the rate of phosphorylation of GST-ERK3⌬Ct by the bound ERK3 kinase before and after treatment with PP2A.
FIG. 5. The ERK3 kinase phosphorylates Ser 189 of ERK3. A, comparison of the phosphorylation lips of ERK2 and ERK3. The ERK2 and ERK3 mutants are indicated above and below the lip sequences. The sites phosphorylated to activate ERK2 are marked with asterisks. Identical residues between ERK2 and ERK3 are indicated with vertical bars. B, GST-ERK3⌬Ct, GST-S189A ERK3⌬Ct, and GST-S189E ERK3⌬Ct bound to glutathione-agarose beads were incubated with Mono S fractions containing the ERK3 kinase activity from rabbit muscle. After the beads were washed as described under "Material and Methods," the bound ERK3 kinase activity was measured by its ability to phosphorylate GST-ERK3⌬Ct, and mutants are as described in Fig.  2. Top, an autoradiogram showing 32 P incorporation into GST-ERK3⌬Ct and mutants. Bottom, Coomassie Blue stain of GST-ERK3⌬Ct and mutants. C, phosphoamino acid analysis of phosphorylated GST-ERK3⌬Ct and GST-S189A ERK3⌬Ct. Spots close to the origin were partially hydrolyzed phosphorylated products. The positions of the phosphoamino acid standards are indicated. support the conclusion that Ser 189 is the site phosphorylated by the ERK3 kinase. To confirm that the ERK3 kinase phosphorylated the same site on ERK3 that was phosphorylated in intact cells, tryptic phosphopeptide maps of ERK3 and ERK3⌬Ct phosphorylated by the ERK3 kinase were compared to a map of ERK3 phosphorylated in intact cells. Each map revealed a major phosphopeptide (Fig. 6A-C) that migrated as a single spot if tryptic phosphopeptides from ERK3 phosphorylated in vitro were mixed with those from ERK3 phosphorylated in intact cells (Fig. 6D). This major phosphopeptide was absent from S189A ERK3 phosphorylated by the ERK3 kinase (Fig. 6E). The incorporation into S189A ERK3 was about 1-2% of that incorporated into wild type ERK3. The addition of tryptic phosphopeptides from phosphorylated wild type ERK3 to those from phosphorylated S189A ERK3 restored the major phosphopeptide (Fig. 6F). These data indicate that Ser 189 of ERK3, the site phosphorylated in intact cells, is the major site phosphorylated by the ERK3 kinase.
Specificity of the ERK3 Kinase-The specificity of the ERK3 kinase was characterized. For these experiments, the ERK3 kinase was partially purified from rabbit muscle and then affinity purified on GST-ERK3⌬Ct bound to glutathione-agarose beads. Similar results were obtained with the ERK3 kinase prior to binding to GST-ERK3⌬Ct on beads. His 10 -ERK3⌬Ct was phosphorylated by the ERK3 kinase but less efficiently than bound GST-ERK3⌬Ct (Fig. 7A). In ERK2, Thr 183 and Tyr 185 are the activating phosphorylation sites. These residues were interchanged with Ser 189 and Gly 191 , the comparable residues in ERK3. The double mutant His 10 -S189T, G191Y ERK3⌬Ct was phosphorylated by the ERK3 kinase primarily on threonine and to a lesser extent on serine but not on tyrosine (Fig. 7, A and B). Thus, unlike MEK1 and MEK2, the ERK3 kinase did not phosphorylate tyrosine in the phosphorylation lip at a position equivalent to Tyr 185 of ERK2. Further, the ERK3 kinase did not phosphorylate His 6 -K52R ERK2 or the ERK3-like mutants T183S ERK2, Y185G ERK2, and T183S,Y185G ERK2 (Fig. 7A). DISCUSSION A concept that has developed from studies in yeast and mammalian cells is that of the MAP kinase module (1,9,28,29). A MAP kinase module is a three-kinase cascade including a MAP kinase or ERK, a MEK, and an activator of MEK, MEK kinase or MEKK. Thus far, studies indicate that the MEK component has the greatest substrate specificity of enzymes in the cascade (1, 14, 18). The known MEK family members se-FIG. 6. Tryptic phosphopeptide mapping of phosphorylated ERK3. Autoradiograms of tryptic phosphopeptide maps of A, ERK3 phosphorylated by the ERK3 kinase; B, ERK3⌬Ct phosphorylated by the ERK3 kinase; C, ERK3 phosphorylated in intact cells; D, mixture of ERK3 phosphorylated by the ERK3 kinase and in intact cells; E, S189A ERK3 phosphorylated by the ERK3 kinase; F, mixture of ERK3 and S189A ERK3 phosphorylated by the ERK3 kinase. Equal counts/min were loaded onto each plate for mapping. FIG. 7. Specificity of the ERK3 kinase. A, the ERK3 kinase bound to GST-ERK3⌬Ct on glutathione-agarose beads phosphorylated both GST-ERK3⌬Ct and added His 10 -ERK3⌬Ct or His 10 -S189T, G191Y ERK3⌬Ct, but not added His 6 -ERK2 mutants (all at 30 g/ml). An autoradiogram is shown. The mobilities of GST-ERK3⌬Ct, His 10 -ERK3⌬Ct and His 6 -ERK2 are indicated. Added His 6 -ERK2 mutants T183S ERK2, Y185G ERK2, and S183T,Y185G ERK2 displayed autophosphorylation rates higher than ERK3 and different from each other. K52R ERK2 lacked the ability to autophosphorylate. B, phosphoamino acid analysis of phosphorylated S189T,G191Y ERK3⌬Ct. Spots near the origin were partially hydrolyzed phosphorylated products. The phosphoamino acid standards are indicated. lectively activate their designated MAP kinase family members, by phosphorylating a threonine and a tyrosine that are arranged with a single intervening residue.
The three-dimensional structure of the MAP kinase ERK2 contains the two-domain organization characteristeric of the protein kinases (11, 30 -32). The active site is formed at the interface of these two domains. The two regulatory phosphorylation sites in ERK2, Tyr 185 and Thr 183 , are in a surface loop known as the phosphorylation lip, that lies at the mouth of the active site. The phosphorylation lip is an important but highly variable regulatory element of the protein kinase family. Structural and biochemical studies indicate that mutation of Tyr 185 in ERK2 changes the conformation of the phosphorylation lip and dramatically decreases ERK2 activity (33); thus Tyr 185 is essential for correct folding of this lip in both low and high activity forms. In ERK2, Tyr 185 faces the active site and can be partially autophosphorylated. Unlike any other ERK/MAP kinase homologs, ERK3 lacks this tyrosine residue, in spite of the significant similarities of the ERK3 phosphorylation lip in sequence and length to the phosphorylation lip of ERK2. A single residue, Ser 189 comparable to Thr 183 of ERK2, is phosphorylated on ERK3 in intact cells (17). The essential nature of Tyr 185 in ERK2 indicates that major differences in folding of the lip may occur in ERK2 and ERK3. Replacement of Gly 191 of ERK3 with tyrosine changes the autophosphorylated residue from serine to tyrosine (17). This suggests that tyrosine may also face the active site in this ERK3 mutant. This mutant is a poor MEK substrate, however, suggesting that portions of the protein that lie outside of the phosphorylation lip are important determinants of MEK-ERK recognition.
A protein kinase that phosphorylates ERK3 and may serve as an activator or MEK for ERK3 has been partially purified and is characterized by its ability to bind to the catalytic domain of ERK3. The ERK3 kinase is found in both the cytoplasm and nucleus of several cell types, unlike MEK1 and MEK2 which are reported to be exclusively in the cytoplasm (34). Like known MEKs, this ERK3 kinase is inactivated by dephosphorylation and is highly specific as demonstrated by its inability to phosphorylate ERK1, ERK2, or ERK2 mutants that more closely resemble ERK3 in the phosphorylation lip. Unlike known MEKs, the ERK3 kinase will not phosphorylate tyrosine when tyrosine is introduced into the appropriate position of the phosphorylation lip of ERK3. Importantly, this ERK3 kinase phosphorylates Ser 189 of ERK3, the site phosphorylated in intact cells. Thus, the ERK3 kinase identified here may be the upstream regulator of ERK3. From a primarily cytosolic location when inactive, ERK1 and ERK2 are translocated in part to the nucleus upon activation, while the activating MEKs are believed to remain cytosolic (27,34). In contrast, ERK3 is found primarily in the nucleus and the ERK3 kinase is present in both cytosolic and nuclear extracts. This suggests a regulatory mechanism in which the ERK3 kinase may receive signals from membrane bound or cytoplasmic cues and shuttle into the nucleus to phosphorylate ERK3 (Fig. 8).