ERK1b, a 46-kDa ERK Isoform That Is Differentially Regulated by MEK*

We identified a 46-kDa ERK, whose kinetics of activation was similar to that of ERK1 and ERK2 in most cell lines and conditions, but showed higher fold activation in response to osmotic shock and epidermal growth factor treatments of Ras-transformed cells. We purified and cloned this novel ERK (ERK1b), which is an alternatively spliced form of ERK1 with a 26-amino acid insertion between residues 340 and 341 of ERK1. When expressed in COS7 cells, ERK1b exhibited kinetics of activation and kinase activity similar to those of ERK1. Unlike the uniform pattern of expression of ERK1 and ERK2, ERK1b was detected only in some of the tissues examined and seems to be abundant in the rat and human heart. Interestingly, in Ras-transformed Rat1 cells, there was a 7-fold higher expression of ERK1b, which was also more responsive than ERK1 and ERK2 to various extracellular treatments. Unlike ERK1 and ERK2, ERK1b failed to interact with MEK1 as judged from its nuclear localization in resting cells overexpressing ERK1b together with MEK1 or by lack of coimmunoprecipitation of the two proteins. Thus, ERK1b is a novel 46-kDa ERK isoform, which seems to be the major ERK isoform that responds to exogenous stimulation in Ras-transformed cells probably due to its differential regulation by MEK.

conditions the 105-kDa ERK5, the DP-ERK detected a band with a molecular mass of 46 kDa. This 46-kDa band might be related to a 46-kDa band previously detected with anti Cterminal domain of ERK-Ab and tentatively termed ERK4 (28). Herein we report the cloning of this 46-kDa ERK, that turned out to be an alternatively spliced form of ERK1 and therefore was termed ERK1b. This ERK isoform is abundant mainly in rat tissues, and lesser amounts were detected in other organisms, including human. Although ERK1b was readily activated by MEK, we found that its activation did not always parallel that of ERK1, possibly due to different subcellular localization of ERK1b, which could modify the mode of its regulation. This was apparent primarily in Ras-transformed Rat1 cells where ERK1b responded differently than ERK1 and ERK2 to exogenous stimulation, suggesting that ERK1b is particularly important for the transmission of signals via the ERK cascade under conditions where ERK1 and ERK2 are under tight phosphatase regulation.
Preparation of Cell Extracts and Western Blotting-Cells were grown to subconfluence and were then serum-starved for 18 h in Dulbecco's modified Eagle's medium containing 0.1% fetal calf serum. The cells were then exposed to various stimuli for variable amount of time. Then, the medium was removed, and the cells were rinsed twice with ice-cold PBS and once with ice-cold Buffer H. Cells were scraped into Buffer H (0.5 ml/plate) and disrupted by sonication (2 pulses for 7 s of 50 watts) on ice. The extracts were centrifuged (100,000 ϫ g, 15 min, 4°C), and the supernatants, which contained the cytosolic and nuclear proteins, were further kept at 4°C. The supernatants were then resolved by a 10% SDS-PAGE, transferred onto a nitrocellulose membrane, and probed with appropriate Ab. Abs binding was detected using alkaline phosphatase (Promega) or ECL (Amersham Pharmacia Biotech) according to the manufacturer's instructions. The Abs used were anti-C terminus of ERK-Ab (C-16, Santa-Cruz Biotechnology), anti-hemagglutinin (HA)-Ab (Santa Cruz Biotechnology), DP-ERK (Sigma, Rehovot, Israel), and Ab 4086 (see below).
Generation of a Polyclonal Anti-ERK1b Ab (Ab 4086)-The Ab was raised against the part of the ERK1b-specific peptide (VSRP-PAAGRGISVPSVRPVPYC) by the Ab Unit of The Weizmann Institute of Science. For immunization of rabbits, the peptide was conjugated to keyhole limpet hemocyanin using the Imject Maleimide Activated kit (Pierce).
Northern Blot Analysis-Northern blotting was performed on a Human Multiple Tissues Northern blot (CLONTECH). The probe used for detection was the ERK1b-specific 78-bp insert prelabeled with [␣-32 P]dCTP.
Total RNA was prepared using TRI reagent (Molecular Research Center Inc.) according to the manufacturer's instructions. For cloning of HA-ERK1b and HA-ERK1, the rat constructs obtained with RT-PCR were ligated to the 3Ј of HA and into the EcoRI and XhoI sites of pCDNA1 (Invitrogen).
Preparation of Extracts from EJ Tumors-EJ cells (1 ϫ 10 6 cells in 0.5 ml of PBS) were injected subcutaneously into 10 CD1-Nude mice obtained from the Animal Breeding Unit of The Weizmann Institute of Science. Two weeks later, the mice were sacrificed, and the tumors, which were 1-2 cm in diameter, were removed and placed in PBS at 4°C. The tumors were transferred to ice-cold buffer H (10 ml) homogenized by PCU (kinetica; three pulses of 20 s) and disrupted by sonica-tion (three pulses for 20 s at 50 watts). The homogenate was centrifuged (100,000 ϫ g, 30 min, 4°C), and the resulting supernatant was immediately applied to the anion exchange column at 4°C.
Anion Exchange Chromatography-Separations were performed using an AKTA system with a Resource Q column (20 ml; Amersham Pharmacia Biotech). After equilibrating with Buffer A, extracts (50 ml) were loaded at 1 ml/min. The bound proteins were then eluted (1 ml/min, 1-ml fractions) by increasing NaCl gradient (0 -30%). The flowthrough fractions containing ERK1b were collected and immediately applied to the affinity column at 4°C.
Affinity Chromatography-The affinity column consisted of ERK1-C16-Ab-agarose conjugate (0.75 ml; CS-93-AC, Santa Cruz Inc.). The flow-through of Resource Q column was loaded (two times, 1 ml/min) onto the column, which was then washed once with 5 ml of 0.1 M glycine, 0.15 M NaCl, pH 2, and once with 0.1 M triethylamine, 0.15 M NaCl, pH 11. The column eluted with 4 ml (four fractions of 1 ml) of 0.1 M triethylamine, 0.15 M NaCl, pH 12.5.
Immunoprecipitation-One day after transfection, cells were serumstarved (0.1% fetal calf serum) for additional 16 h and then stimulated with either VOOH (100 M Na 3 VO 4 and 200 M H 2 O 2 , 18 min, 37°C), EGF (50 ng/ml, Sigma, St. Louis) tetradecanoyl phorbol acetate (TPA, 250 ng/ml, Sigma, St. Louis), or PBS as control. After stimulation, the cells were washed, lysed, and the ERKs were incubated with a polyclonal anti-HA-Ab (Santa Cruz). For MEK1 coimmunoprecipitation studies, the beads were washed with low stringency buffer (20 mM HEPES, pH 8.0, 2 mM MgCl 2 2 mM EGTA) and then subjected to immunoblotting with monoclonal anti-MEK and anti-GFP antibodies as described previously (24,25). For determination of ERK activity, the beads were washed once with 0.5 M LiCl, twice with radioimmune precipitation buffer, and once with buffer A as described previously (29). The immunoprecipitates were subjected either to Western blotting or to myelin basic protein (MBP) phosphorylation assay as described below.
Determination of ERK Activity in Vivo-Immunoprecipitates of ERK1 and ERK1b proteins were mixed with either MBP (8.4 g), Elk1 (1 g, NEB), or RSK (immunoprecipitated with RSK Abs (Sigma, Rehovot, Israel)) from resting EJ cells) and Buffer R that contained 100 M [␥-32 P]ATP (1-2 cpm/fmol) in a final volume of 30 l. The phosphorylation reaction was allowed to proceed at 30°C for 15 min and terminated by sample buffer followed by boiling for 5 min. Phosphorylated proteins were assessed by SDS-PAGE and autoradiography or Western blotting with DP-ERK for phosphorylated ERKs. To separate between ERK1,2 and ERK1b activities we used anion exchange column as described previously (7). Briefly, cell extracts (0.75 ml, 0.5 mg) were loaded on DE52 columns (0.4 ml), and the flow-through fraction, containing ERK1b activity, was collected. After wash (1 ml ϫ 3) with Buffer A ϩ 0.02 M NaCl the ERK1,2 activity were eluted with 1 ml of Buffer A ϩ 0.22 M NaCl. The ERKs were then immunoprecipitated with ERK1-Ab (C-16) and subjected to in vitro phosphorylation as above.
In Vitro Activation of ERK-Extracts from nonstimulated transfected COS7 cells were subjected to immunoprecipitation with anti-HA-Ab. The HA-ERKs proteins, attached to the beads, were mixed with ⌬N-EE-MEK recombinant protein (Sigma, Rehovot, Israel) and Buffer R that contained 100 M [␥-32 P]ATP (1-2 cpm/fmol) at 30°C. After 15 min, MBP (8.4 g) was added (final volume of 30 l), and the phosphorylation was allowed to continue for an additional 15 min. The reaction was terminated and assessed as described above.

Phosphorylation of p46 ERK in Several Cell
Lines -Mitogenic signaling in Rat1 cells was examined using the DP-ERK, which specifically recognizes the dually phosphorylated TEY motif in the activation loop of ERKs. In Western blot analysis of cell extracts derived from serum-starved Rat1 cells, the Ab detected a faint band of 46 kDa in addition to the expected bands of 42-kDa ERK2 and 44-kDa ERK1 (Fig. 1A). Stimulation of the cells with either EGF or TPA resulted in an enhanced staining of the 46-kDa band, with kinetics similar to those of ERK1 and ERK2. The reactivity of all three bands with DP-ERK was reduced when the Rat1 cells were pretreated with the specific MEK inhibitor PD98059 prior to stimulation. When the antigenic peptide, used to generate the Ab, was added to the immunoblots, all three bands disappeared (data not shown), which confirmed that the staining of the 46-kDa band by DP-ERK was specific. The 46-kDa band was also recognized by Abs directed against the C and N terminus of ERK1, against subdomain XI of ERK1 (weaker staining), and against the C terminus of ERK2 (weaker staining; data not shown). In contrast, Ab directed against JNK1, which is a 46-kDa MAPK isoform, did not recognize this 46-kDa band (data not shown). Thus, the 46-kDa band appears to represent a genuine, TEYcontaining, ERK isoform, which is phosphorylated by MEKs. This band may be similar to the poorly characterized ERK4, which was previously detected in rat (28,31)-and human (32)-derived tissue culture cells.
A 46-kDa ERK appeared to be a major phosphorylated isoform of ERK in Rat1 cells, but its expression and activation in other cell lines had to be examined. Thus, in the rat-derived PC12 cell line, typical transient and sustained kinetics (33) of the 46-kDa ERK as well as of ERK1 and ERK2, were observed when the cells were stimulated with nerve growth factor and EGF (Fig. 1B). A sustained activation was observed for ERK1, ERK2, and the 46-kDa ERK also in Fc⑀RI-induced rat mucosaltype mast cells (RBL-2H3 line) and in heat-shocked rat glia cells (data not shown). In addition, a lesser amount of the 46-kDa ERK, which had kinetics similar to those of ERK1 and ERK2, was detected also in several other cell lines derived from mouse, calf, and guinea pig (data not shown), as well as in human breast cancer-derived MCF-7 cells where all three apparent ERKs (46 kDa, ERK1 and ERK2) were significantly phosphorylated upon stimulation with okadaic acid (an inhibitor of protein serine/threonine phosphatases 1 and 2a; Fig.  1C). These results indicate that p46 ERK is present mainly in rat-derived cells but can be found in slightly lower amounts also in cell lines from other species.
As shown above, the kinetics of TEY phosphorylation of the 46-kDa ERK is usually similar to that of ERK1 and ERK2. However, in several instances, the phosphorylation of the 46-kDa ERK varied significantly from its counterpart kinases. Thus, in Ras transformed Rat1 cells (EJ), the phosphorylation of ERK1 and ERK2 was reduced at 5-10 min and then increased at up to 60 min after the application of an osmotic shock to the cells (Fig. 2, A and B). On the other hand, the kinetics of the 46-kDa ERK phosphorylation increased shortly after the osmotic shock, peaked at 30 min, and declined thereafter. Similarly, the kinetics of the 46-kDa ERK's phosphorylation was different from that of ERK1 and ERK2 also upon EGF stimulation of the same EJ cells (Fig. 2, C and D), as well as in MAFA (34)-stimulated mast cells (data not shown). To confirm that the 46-kDa ERK is regulated differently from ERK1 and ERK2, we separated activated forms of ERK1 and ERK2 from the activated form of the 46-kDa ERK on a small bed of DE52 resin (7). While the 46-kDa ERK eluted in the flow-through, active ERK1 and ERK2 were retained on the resin and could be eluted with 0.2 M NaCl. As expected from the result with the total extracts above, staining of the flowthrough fractions with anti-active ERK antibodies revealed a 5-6-fold increase in the 46-kDa ERK staining after NaCl and EGF stimulation, without a change in the total amount of the 46-kDa protein (Fig. 2E, left). On the other hand, staining of the 0.2 M NaCl eluted fractions from the same columns showed that the staining of ERK1 and ERK2 was reduced after NaCl treatment and not significantly changed upon EGF stimulation, without a change in the total amount of proteins in these fractions (Fig. 2E, right). We then studied the catalytic activities of the ERKs by immunoprecipitating the ERK in the various fraction followed by an in vitro kinase assay using MBP as a substrate. The results obtained with this procedure demonstrated a direct correlation to the results obtained with the antibodies above (Fig. 2F), confirming the differential activation of the 46-kDa proteins in EJ cells. Therefore, these data indicate that although the mode of ERK1, ERK2, and the p46 Each of these blots was immunoblotted with Abs to dually phosphorylated ERK (DP) and anti-C terminus of ERK-Ab (C16). The position of ERK2, ERK1, and p46 ERK is indicated. Each of these experiments was reproduced at least three times.
ERK regulation is usually similar, a differential mode of p46 ERK regulation may also exist. This deferential regulation may be due to differences either in the phosphorylation or dephosphorylation processes, which may both culminate in the differential kinetics of phosphorylation observed under some conditions.
ERK1b; Cloning, Purification, and Preparation of Antibody-The presence of p46 ERK in a variety of cell types and its FIG. 2. Kinetics of ERK phosphorylation and activation upon osmotic shock or EGF treatment of EJ cells. EJ cells (7 ϫ 10 5 cells/6-cm plate) were serum-starved for 16 h and then stimulated with either 0.7M NaCl (A) or with 50 ng/ml EGF (C) for the indicated times. Extracts (50 g) were analyzed by immunoblotting with DP-ERK (DP) and anti-C terminus of ERK-Ab (C16). B and D, quantitative determination of the results in A and C. Phosphorylation of ERK2 (q), ERK1 (OE), and p46 ERK (f) was calculated as the intensity of staining with DP-ERK divided by the intensity of staining with C16. This is a repre- sentative experiment that was reproduced four times. E and F, EJ cells were serum-starved and then stimulated (ϩ) for 10 min with either 0.7 M NaCl (NaCl), 50 ng/ml EGF (EGF), or left untreated (Ϫ). ERK1 and ERK2 were separated from p46 ERK on an anion exchange column as described above. ERK activity in the different fractions was analyzed by immunoblotting with DP-ERK (DP in E) as compared with the staining by anti-C terminus of ERK-Ab (C16 in E) or by immunoprecipitation with ERK-Ab (C16) followed by kinase assay with MBP as a substrate (F). The results in E and F were reproduced four times. differential regulation in transformed cells, prompted the indepth study of this protein by molecular biology. Since p46 ERK was recognized by several Abs, directed to various regions of ERK1, sense and antisense oligonucleotide primers derived from the sequence of ERK1 were used for RT-PCR with an RNA template obtained from EJ cells. With one of these primer pairs, two amplified products were obtained (Fig. 3A): the expected ϳ300-bp band (ERK1) and another of ϳ400 bp. Both products were cloned and sequenced. The ϳ300-bp product corresponded to the sequence of ERK1; sequencing of the upper product revealed a unique 78-bp sequence flanked by sequences identical to those of the rat ERK1 (Fig. 3B). Using RT-PCR with oligonucleotides derived from the unique 78-bp insert, oligonucleotides derived from the 5Ј-and 3Ј-ends of the rat ERK1, and RNA from EJ cells as a template, we cloned the full-length cDNA of this ERK. This full-length cDNA contained the 78-bp inserted between bases 1020 and 1021 of ERK1, and its sequence, except for the insertion, was identical to that of ERK1 (Fig. 3C), which indicates that this is an alternative spliced form of ERK1, and was therefore termed ERK1b. The alternative splicing probably occurs at the end of exon 7 of the ERK1 gene as this exon ends with a sequence corresponding to the site of insertion (35). The 78-bp insertion encodes 26 amino acids (Fig. 3C), which are localized between Glu-340 and Pro-341 of the rat ERK1 in a large loop (L16) between two ␣ helices (␣i and ␣L16) just C-terminal to the conserved core of the kinase domain (Fig. 3B). The sequence of insertion did not exhibit significant homology with any known protein or cDNA in the data base, and no particular protein motif could be attributed to this sequence. Thus, in rat cells, besides the ERK1 and ERK1psi (5), an additional mRNA for ERK1b exists that encodes for a putative protein kinase with a calculated molecular mass of 45.6 kDa.
Whether the p46 ERK is indeed encoded by the ERK1b mRNA was determined by purifying the p46 ERK from EJ cells. EJ tumors, induced in nude mice, were excised and lysed in the presence of proteinase and phosphatase inhibitors. The cytosolic fraction obtained was subjected to anion exchange column, in which the p46 ERK did not bind and was eluted in the flow-through, whereas the ERK1 and ERK2 bound and eluted with 0.1 M NaCl (Fig. 4A). The fractions that contained the p46 ERK were applied to an affinity column that consisted of the anti-C terminus of ERK Ab bound to agarose. After the column was extensively washed, the bound proteins were eluted at high pH (Fig. 4B). This protocol resulted in ϳ4000fold purification of a 46-kDa protein that reacted with anti-ERK-Ab (Fig. 4C) and was about 70% pure. The purified 46-kDa protein was excised from the SDS gel, digested with trypsin, and the resulting peptides analyzed by mass spectroscopy and when necessary also Edman degradation. Eight peptides were recovered from the main protein in the band, seven of them were identical to those expected from ERK1, including a peptide containing the exact sequence around and including the TEY motif of ERK1. The other peptide that was obtained corresponded to the sequence GISVPSVR (part of the predicted protein sequence of the 78-bp insert of ERK1b), which indicated that the isolated 46-kDa band is, at least in part, the cloned ERK1b protein.
The identity of the 46-kDa band was further confirmed using a specific polyclonal Ab that was generated against the ERK1b insert. This Ab, termed 4086, recognized two bands in a cytosolic extract of EJ cells (Fig. 5), one of which migrated to the same location as the 46-kDa protein recognized by the anti-Cterminal of ERK-Ab (Fig. 5). The staining of the 46-kDa band, but not of the 30-kDa band, was competitively inhibited by the antigenic peptide (data not shown), which verified the specific-ity of the Ab. This Ab was used to assess the overexpression of the translation product of ERK1b in COS7 cells. A HA tag was fused to the N-terminal portions of ERK1 and ERK1b, and these constructs were transfected into COS7 cells, which were examined for ERK1b protein content using Ab 4086. In mocktransfected COS7 cells, no 46-kDa protein could be detected by Ab 4086. However, this Ab clearly recognized the expected 48-kDa band (ERK1b with HA tag) in the transfected cells, which was also recognized by an anti-HA-Ab and was retarded on SDS-PAGE upon VOOH stimulation of the COS7 cells. Thus, Ab 4086 recognizes both endogenous and recombinant ERK1b and therefore provides further evidence that ERK1b is indeed the p46 ERK.
Functional Characterization of ERK1b-The functional characteristics of ERK1b were studied by transfecting COS7 cells with HA-ERK1b, HA-ERK1, and an empty vector. Protein extracts obtained from serum-starved, transfected cells exposed to EGF, VOOH, or buffer were subjected to immunoprecipitation with anti-HA-Ab. After extensive washing to remove impurities, the precipitated proteins were subjected to Western blotting with DP-ERK and anti-HA-Ab or to a kinase assay with MBP as a substrate. Two days after transfection, the amount of HA-ERK1b produced in COS7 cells was usually less than that of ERK1 (Fig. 6A). However, the basal state of the specific activation (phosphorylation of the TEY motif per amount of HA-ERK) and specific activity (phosphorylated MBP per amount of HA-ERK) of HA-ERK1b in the transfected cells was somewhat higher than those of HA-ERK1 (Fig. 6B). ERK1b and ERK1, which were immunoprecipitated from EGFor VOOH-stimulated transfected COS7 cells, exhibited similar specific activation and specific activity.
The ability of ERK1b to be activated by MEK in vitro was also examined by subjecting HA-ERK1 and HA-ERK1b that were immunoprecipitated from resting COS7 cells to phosphorylation with the constitutively active ⌬N-EE-MEK1 (30). As with the in vivo ERK activation, both ERK1 and ERK1b were activated by the active MEK and exhibited similar specific activity and specific activation (Fig. 6, C and D). Moreover, ERK1b could phosphorylate both Elk1 and RSK in a rate that was just slightly lower than that of ERK1 when subjected to in vitro kinase assay (Fig. 6E). Therefore, our results indicate that the insertion of 26 amino acids in ERK1b affects the phosphorylation and activity of ERK1b in resting mammalian cells, but does not significantly change its ability to be activated by MEK in vitro or its catalytic activity toward MBP, RSK, and Elk.
Tissue and Organism Distribution of ERK1b-ERK1 and ERK2 were shown to be ubiquitously expressed in all organisms and tissues (5). Tissue distribution of ERK1b in the rat was assessed by performing quantitative RT-PCR on RNA obtained from several rat tissues and cell lines. A pair of oligonucleotide primers, that should yield a 303-bp product from ERK1 and a 381-bp product from ERK1b, were used. Indeed, two such DNA products were observed in many of the tissues examined (Fig. 7A). Whereas the amount of the ERK1 was roughly similar in all the tissues and cells type examined, the amount of ERK1b was always less and varied depending on the source. The maximal amount of ERK1b transcript was detected in the heart, significant amounts were also present in the kidney, lung, and brain, while only small amounts were observed in the liver and skeletal muscles. As with the actual expression of the ERK1b protein ( Fig. 2A), the amount of ERK1b RNA expressed in EJ cells was higher than in Rat1 cells. The distribution of ERK1b in human tissues was determined with 32 P-labeled probe prepared from the 78-bp insertion region of the ERK1b cDNA. In Northern blots, this ERK1b probe identified one major 2.2-kb transcript in human tissues (Fig. 7B). In human, as in rat, the ERK1b transcript was highest in the heart; the transcript was also detected in the brain and placenta, but not in the liver and lungs. Our results indicate that ERK1b RNA, although not restricted to expression in one organism, is not as ubiquitously expressed as ERK1 and ERK2 and seems to be most abundant in the heart. Differential Regulation of ERK1b in Ras-transformed Rat1 Cells-In contrast to the similar kinetics of ERK1 and ERK1b activation in most cells, the phosphorylation of ERK1b in Rastransformed EJ cells was different from that of ERK1 and ERK2 (Fig. 2). To verify the differential regulation in these cells we compared the expression and TPA activation of ERK1b to that of ERK1 in the EJ and Rat1 cells. Staining with the Ab directed to the C terminus of ERK1 that recognizes both active and inactive ERKs revealed that, despite a similar expression of ERK1 and ERK2 in EJ and Rat1 cells, the amount of ERK1b in EJ cells was significantly higher (6 -7-fold) than in Rat1 cells (C16, Fig. 8A). Similar elevation of expression of ERK1b was noticed also with other oncogenes and additional cell lines (data not shown). With DP-ERK, we observed that the basal TEY phosphorylation of ERK1 and ERK2 as well as ERK1b was slightly higher in EJ than in Rat1 cells (Fig. 8A). However, similarly to the stimulation by EGF and NaCl (Fig. 2), TPA stimulation caused significant differences in the extent of phosphorylation of the ERK1b as compared with ERK1 and ERK2 (Fig. 8, A and B). In EJ cells, stimulation of ERK1-TEY phosphorylation with TPA was moderate (up to 6-fold, Fig. 8B) and significantly lower than the stimulation of ERK1 phosphorylation in Rat1 cells (13.5-fold). The stimulation of ERK1b phosphorylation in EJ cells was higher than that of ERK1 and similar to that in Rat1 cells (12.5-fold). To verify the results obtained with the anti DP-ERK we transfected mammalian expression constructs with HA-ERK1 and HA-ERK1b to the rat1 and EJ cells. TPA-stimulated ERK activity was then measured by immunoprecipitation with anti HA antibodies, followed by an in vitro phosphorylation of MBP. The trend of the results obtained with these methods (Fig. 8, C and D) was similar to that obtained with the anti DP-ERK antibodies (Fig.  8, A and B). Most significantly, the activation of ERK1b with TPA was 2-3-fold higher than the activation of ERK1 in EJ, but not in Rat1 cells. Thus, our results clearly indicate a differential regulation of ERK1b in EJ cells and may suggest that in transformed cells, ERK1b is the major ERK isoform to respond to exogenous stimulation and thereby allow the physiological responses of transformed cells.
MEK-dependent Subcellular Localization of ERK1b-A possible cause for the differential regulation of ERK1b could be a different subcellular localization. Therefore the subcellular localization of ERK1b was examined in CHO cells that had been transfected with either HA-ERK1 or HA-ERK1b. Anti-HA-Ab revealed that before and after stimulation of the cells, most of the transfected proteins were localized in the nucleus without significant differences between them (data not shown). This observation is in agreement with the localization of ERK in the nucleus when overexpressed in various cells (24,25). Since overexpression of MEK results in cytoplasmic retention of ERK (24,25), both HA-ERK1 and HA-ERK1b were coexpressed in CHO cells, together with wild type MEK1. As reported previously for ERK2 (25), in resting cells, HA-ERK1 was localized primarily in the cytoplasm (ϳ90%), but upon stimulation with VOOH, a large amount of it was translocated into the nucleus (Fig. 9). In contrast, in resting cells HA-ERK1b was already localized primarily in the nucleus (70%) and upon stimulation became even more nuclear (80%, Fig. 9A). Therefore, the 26amino acid unique sequence of ERK1b might be involved in disruption of the MEK-dependent cytoplasmic localization of ERK1b in resting cells. A plausible explanation for this is that the 26-amino acid insertion changes the conformation of the neighboring MEK-dependent cytoplasmic localization sequence (residues 312-320 of ERK2) that was identified by our laboratory (25). To further verify the differential MEK-dependent localization of ERK1b, we examined the physical association of ERK1 and ERK1b with wild type MEK. As reported before for ERK2 and Xenopus ERK (8,25), ERK1 was associated with MEK1 in resting cells and the interaction was prevented upon stimulation (Fig. 9B). However, ERK1b did not significantly interact with MEK1 even in resting cells (Fig. 9B), supporting the idea that the 26-amino acid insertion interferes with the MEK-induced cytosolic anchoring of ERK1, causing the constitutive nuclear localization of ERK1b. This differential localization may cause a differential regulation of ERK1b by phosphatases and brings about its higher response to stimuli in transformed cells.

DISCUSSION
The ERK cascade is involved in diverse physiological processes that result in different outcomes. Thus, elucidation of the mechanism by which different signals can be transmitted by similar signaling cascades, but still evoke different downstream outcomes, is important. Several mechanisms for the specificity of the ERK cascade have been proposed. For example in PC12 cells the duration and strength of ERK activation seem to be important for specificity (33). In these cells, transient activation of ERK results in proliferation, whereas a sustained activation causes differentiation. Other proposed mechanisms for determining specificity are (i) compartmentalization, which can be achieved by association with either cytoskeletal elements (23) or scaffold proteins (36), and may direct signals to the correct destination; (ii) cross-talk with different signaling pathways (37), which modifies the signals in the ERK cascade; (iii) differential intensities of signals (38) FIG. 6. Expression and functional activity of ERK1b. COS7 cells were transfected with either HA-ERK1b, HA-ERK1, or pCDNA1 vector. A, the cells were stimulated with either EGF (50 ng/ml, 10 min), VOOH (20 min), or left untreated (Basal). After harvesting, proteins were immunoprecipitation with polyclonal anti-HA-Ab and subjected either to Western blotting with DP-ERK (DP) and monoclonal anti-HA-Ab (HA) or to an in vitro phosphorylation of MBP assay (Pho.). B, the results in A were quantitated, and specific phosphorylation on the TEY motif (designated D) was calculated as the intensity of staining by DP-ERK divided by the intensity of staining by HA-Ab. Specific ERK activity toward MBP (designated P) was calculated as the amount of phosphorylated MBP minus the phosphorylation in the plasmid control per the intensity of staining by HA. 1 stands for HA-ERK1, 1b for HA-ERK1b, B for Basal, E for EGF, and V for VOOH. C, immunoprecipitants from nonstimulated cells were subjected to an in vitro phosphorylation by constitutively active MEK (CA-MEK (30)) followed by either immunoblotting with DP-ERK (DP) and a polyclonal anti-HA-Ab that are simultaneously transmitted via additional signaling pathways; and (iv) multiplicity of isoforms that are differentially activated by distinct stimuli (5). In this manuscript, we describe the cloning and characterization of an alternatively spliced form of ERK1 (ERK1b), which appears to operate differently than ERK1 (Fig. 2), and therefore may alter signaling specificity.
Alternative splicing has been shown to influence the specificity of several kinases in various MAPK cascades. For example, in the human brain, 10 isoforms of JNK were identified that correspond to alternatively spliced isoforms derived from the JNK1, JNK2, and JNK3 genes (39). These isoforms differ mainly in their binding with ATF2, Elk-1, and Jun transcription factors, which accounts for the different specificities of the JNKs in vivo. In addition, MKK7 seems to have six spliced isoforms, which differ in their association with their substrate, JNK1 (40). The alternatively spliced form of MEK5 is thought to specifically associate with actin cytoskeleton and as result may cause a differential subcellular localization of the two isoforms of MEK5 (41). These examples demonstrate that alternative splicing of components of MAPK cascades mainly affects protein-protein interactions and localization, which is also observed with ERK1b and can account for its differential localization (Fig. 9).
Alternatively spliced forms (or pseudogenes) have been reported also for some components of the ERK cascade. For example, ERK1psi, which lacks the C-terminal domain of ERK1 and has some other modifications, was identified in rat (5); a p41 isoform of ERK2, which lacks the N terminus of ERK2 was identified in humans (42); and an inactive form of MEK1, termed MEK1b, which lacks 26 amino acids in subdomain V of MEK1, was identified in humans (43,44). However, all these isoforms except of ERK1b were identified based solely on RNA transcripts, and proteins with the appropriate molecular size or a particular distinct function have yet to be attributed to them. Thus, the ERK1b appears to be the only ERK isoform besides the p42 ERK2 and p44 ERK1 that is presented in appreciable amounts in vivo. Whether ERK1b is the only isoform recognized as a p46 ERK by Abs is not clear, since in EJ cells, an additional and slightly higher molecular weight band was sometimes recognized by the anti-C-terminal of ERK-Ab (data not shown). Moreover, the sequence of two peptides from the digested, purified p46 ERK differed slightly from the predicted sequence of ERK1. Therefore, an ERK isoform of 46 kDa, distinct from ERK1b, may also exist in transformed rat cells.
Ras transformation amplified the amount of ERK1b in a MEK-dependent manner ( Fig. 2A). Oncogenic transformation was previously shown to influence alternative splicing of several signaling proteins (e.g. Ref. 45). This effect of oncogenic transformation might be due to enhanced phosphorylation in the transformed cells, since phosphorylation is known to regulate splicing (46) by affecting several steps of the splicing proc- The amount of MEK was found to be similar in the cells examined (data not shown). This experiment was reproduced three times. B, EJ cells were transfected with HA-ERK1 (ERK1) or HA-ERK1b (ERK1b) together with MEK1 (1:1 ratio of ERK containing plasmid to MEK containing plasmid) and treated as described above. After stimulation the cells were harvested, lysed, and the HA-containing proteins were immunoprecipitated with polyclonal anti-HA Ab, washed briefly (as described under "Material and Methods"), and subjected to Western blotting with monoclonal anti-MEK antibody (B, upper panel). The amount of ERK in the different lysates was determined by a Western blot analysis with monoclonal anti HA antibody (B, bottom panel). esses and several splicing factors (47). Most notable are the SR proteins, which are involved in the assembly of spliceosomes, function as alternative RNA splicing factors, and need to be highly phosphorylated to induce splicing (48). In EJ cells, we found that the amount of ERK1b is reduced by overexpression of dominant negative MEK (data not shown) suggests that phosphorylation by a component of the ERK cascade is involved (directly or indirectly) in the regulation of alternative splicing upon malignant transformation. Moreover, in EJ (Fig. 8) as well as other transformed cells (49), constitutive activation of the upstream components of the ERK cascade was not accompanied by a comparable activation of ERK1 and ERK2. The latter suggests that these two isoforms (ERK1 and ERK2) might be under tight down-regulation, and thus in EJ cells their fold stimulation by extracellular stimuli is low. However, ERK1b appears to escape from the tight down-regulation in the EJ cells and appears to be the major responsive ERK isoform (Fig. 8).
The subcellular distribution of ERK1b differs from that of ERK1 and ERK2. When overexpressed in mammalian cells Xenopus ERK, ERK2, and ERK1 are localized to the nucleus of resting cells (Refs. 24 and 25 and data not shown). This subcellular localization changes when MEK1 is overexpressed together with the ERKs, which results in mostly cytosolic distribution of the ERKs (Refs. 24 and 25 and Fig. 9). In contrast, ERK1b was not retained in the cytosol, even upon high overexpression of MEK1 (Fig. 9). ERK1b also did not interact with the overexpressed MEK1. Recently, we identified a sequence (residues 312-320) of ERK2 that is responsible for its MEKinduced cytosolic retention (25). Changing these residues to alanines resulted in a nuclear distribution of ERK2, even in the presence of overexpressed MEK1. The sequence of residues 312-320 of ERK2 is almost identical to that of residues 332-340 of ERK1, in the end of which the ERK1b-specific 26-amino acid insertion is localized. Therefore, this insertion probably alters the structure of the cytosolic retention sequence of ERK1, thus causing a constitutive localization of ERK1b in the nucleus. Upon stimulation, ERK1 and ERK2 are transiently translocated into the nucleus, whereas ERK1b appears to be retained in the nucleus even after prolonged stimulation ( Fig. 9 and data not shown). Although the results were obtained with using overexpressed ERK1b and not the endogenous protein, these results support our previous findings (25) on the importance of the first portion of the C-terminal domain of ERK for subcellular localization.
The nuclear distribution of ERK1b does not seem to interfere with its activation upon mitogenic stimulation, since its kinetics of activation was similar to those of ERK1 and ERK2 under most condition. The activation of ERK in the nucleus could be explained by nuclear translocation of the upstream activator MEK. In fact, despite the usual cytosolic distribution of MEK1, this kinase may translocate into the nucleus upon mitogenic stimulation (30,50). The role for the nuclear translocation of MEK1, which was suggested to be important for cell proliferation and oncogenesis (51), may be to activate ERK1b. Although under most conditions, the stimulation of ERK1b was similar to those of ERK1 and ERK2, a few differences were observed in its stimulation-dependent kinetics of activation in Ras-transformed cells. These differences may be due to interaction of ERK1b with dual specificity MAPK phosphatases (MKPs). In fact, MKPs interacts with Asp-319 of ERK2, which is homologous to Asp-339 of ERK1, and this interaction is essential for their activity (52,53). The 26-amino acid insertion, at position 340, may account for a different interaction of ERK1b with MKPs and therefore effect its regulation, which may be mediated by phosphatases other than those regulate ERK1 and ERK2 (54).
In summary, we purified, cloned, and characterized the 46-kDa ERK1b, which is an alternatively spliced form of ERK1, with a 26-amino acid insertion between residues 340 and 341 of ERK1. Under most conditions, the kinetics of activation, activation by MEK, and kinase activity of ERK1b were similar to those of ERK1. However, unlike the ubiquitous expression of ERK1 and ERK2, ERK1b was only detected in some tissues of organisms examined; also the phosphorylation of ERK1b in Ras-transformed Rat1 cells differed from that of ERK1 and ERK2. Moreover, in MEK1-overexpressing cells, the distribution of ERK1 and ERK2 was mainly cytosolic, whereas that of ERK1b was primarily nuclear. This different subcellular localization is probably due to alteration of the conformation of the cytosolic retention region, which is near the site of the ERK1bspecific 26-amino acid insertion and may also affect the association of ERK1b with MKPs. Our findings are consistent with ERK1b being the major ERK isoform that responds to exogenous stimulation in Ras-transformed Rat1 cells.