Biochemical characterization of a family of serine/threonine protein kinases regulated by tyrosine and serine/threonine phosphorylations.

Mitogen-activated protein kinase (p42mapk) becomes transiently activated after treatment of serum-starved murine Swiss 3T3 cells or EL4 thymocytes with a diversity of mitogens. Similarly, a meiosis-activated protein kinase (p44mpk) becomes stimulated during maturation of sea star oocytes induced by 1-methyladenine. Both p42mapk and p44mpk have been identified as protein-serine/threonine kinases that are activated as a consequence of their phosphorylation. Because homologous protein kinases may play essential roles in both mitogenesis and oogenesis, we have compared in detail the biochemical properties of these two kinases. We find that these kinases are highly related based on their in vitro substrate specificities, sensitivity to inhibitors, and immunological cross-reactivity. However, they differ in apparent molecular weight and can be separated chromatographically, indicating that the two enzymes are distinct. Furthermore, in the course of this investigation, we have identified a 44-kDa protein kinase in mitogen-stimulated Swiss mouse 3T3 cells and EL4 thymocytes that co-purifies with p44mpk and thus appears to be a closer homolog of the sea star enzyme. Analysis of these protein kinases clarifies the relationships between a set of tyrosine-phosphorylated 41-45-kDa proteins present in mitogen-stimulated cells (Martinez, R., Nakamura., K. D., and Weber, M. J. (1982) Mol. Cell. Biol. 2, 653-655; Cooper, J. A., and Hunter, T. (1984) Mol. Cell. Biol. 4, 30-37), two myelin basic protein kinases identified in epidermal growth factor-treated Swiss mouse 3T3 cells (Ahn, N. G., Weiel, J. E., Chan, C. P., and Krebs, E. G. (1990) J. Biol. Chem. 265, 11487-11494), and p42mapk. Our work points to the existence of a group of related serine/threonine protein kinases, regulated by tyrosine phosphorylation and functioning at different stages of the cell cycle.

Re-entry of quiescent cells into the cell cycle from Go can be stimulated by binding of peptide growth factors to their cognate receptors. Many of these receptors are ligand-stimulated protein-tyrosine kinases whose enzymatic activity is essential for mitogenesis (1,2). Signals from these receptors are most likely propagated in part by protein phosphorylation cascades. Presumably, the receptor kinase phosphorylates cellular substrates on tyrosine, initiating a cascade of events resulting in cell division. Thus, identifying the in uiuo targets for these mitogen-stimulated tyrosine kinases has been a recognized goal for understanding the G,/G1 transition.
Two potentially important tyrosine kinase substrates of approximately 42 and 44 kDa were initially identified in murine fibroblasts stimulated with many diverse mitogenic agents (3)(4)(5)(6)(7). These proteins become phosphorylated nearly stoichiometrically on tyrosine and also on threonine within 10 min after mitogen stimulation. Furthermore, phorbol esters, which are thought to act exclusively through protein kinase C, a serine/threonine protein kinase, also elicit the phosphorylation of p42 and p44 on tyrosine as well as threonine (8)(9)(10). Based on two-dimensional gel electrophoresis, phosphorylated p42 (pp42) was found to exist in two phosphoforms, pp42A and pp42B (11). Rossomando et al. (12) have shown that pp42A corresponds to the activated form of mitogen-activated protein kinase (MAP kinase, termed herein p42'""pk),' a serine/threonine protein kinase, with a preference for phosphorylating MAP-2 or MBP (13,14). Activated ~4 2~~ is phosphorylated on both tyrosine and threonine (13), and both phosphorylations are necessary for kinase activity (15).
Tyrosine-phosphorylated proteins of 40-45 kDa have also been identified in Xenopus laeuis eggs. Using two-dimensional electrophoresis, Lohka et al. (16) found that a pair of 42-kDa proteins contained phosphotyrosine and phosphothreonine only at M-phase in maturing frog oocytes, and these were rapidly dephosphorylated following fertilization or parthenogenetic activation of Xenopus eggs. Cooper (17) provided further evidence that these phosphoproteins were related to mammalian p42 and p44 by partial phosphopeptide mapping. The activation of MBP kinases in maturing Xenopus oocytes has also been observed by column chromatography (18). Recently, Gotoh et al. (19) purified a tyrosine-and serinephosphorylated Xp42 to homogeneity from unfertilized Xen-The abbreviations used are: ~4 2 -~, 42-kDa mitogen-activated protein kinase (MAP kinase); ~4 4 ' "~~, murine 44 kDa MAP-kinase; ~4 4 ' "~~, meiosis-activated protein kinase; MAP-2, microtubule-associated protein-2; MBP, myelin basic protein; ERK 1, extracellular signal-regulated kinase 1. opus eggs and demonstrated that it possessed many properties in common with an epidermal growth factor-activated MAP-2 kinase from rat 3Y1 cells.
Pelech et al. (20) have previously described the marked stimulation of an MBP kinase near the time of germinal vesicle breakdown in maturing sea star oocytes. Purified sea star MBP kinase (designated ~4 4~~ for meiosis-activated protein kinase) migrates as a 44-kDa protein on sodium dodecyl sulfate-polyacrylamide gels (21). The following evidence indicated that murine ~4 2~~ and sea star ~4 4 " '~~ might be related enzymes: in addition to their similar sizes, both enzymes are cytosolic, tyrosine-phosphorylated, require phosphorylation for activity, and exhibit a preference for MAP-2 or MBP as substrates (13,15,(21)(22)(23).
To determine whether murine ~4 2 " ' "~~ and sea star ~4 4~~ are species homologs or related isoforms, we have compared their biochemical properties in detail. Based on these studies, we conclude that the sea star kinase is closely related but not identical to ~4 2~~. Moreover, we have identified a 44-kDa MAP-2/MBP kinase in mammalian cells which appears to be more closely related to the sea star ~4 4 " '~~. This murine ~4 4~~ appears to correspond to p45, a phosphotyrosine-containing protein first identified on two-dimensional gels by Cooper et al. (4). We also provide evidence that ~4 2 " '~~ and ~4 4~~ correspond to the epidermal growth factor-activated protein kinases described by Ahn et al. (24). MBP kinase 1 (peak E3) and MBP kinase I1 (peak E4), respectively. The data provide biochemical evidence of a gene family for the 42-44-kDa MAP-2/MBP kinases regulated by tyrosine phosphorylation. This conclusion is in agreement with the recent identification by Boulton et al. (25) of at least four potential genes related to ERK 1, a MAP-2 kinase.

Identification of the MAP K i m e
Family-Evidence for a family of 42-45-kDa tyrosine-phosphorylated proteins was provided by investigators studying proteins that become phosphorylated in response to mitogenic stimulation. The expectation has been that a subset of these proteins would play crucial roles in the transition from Go into the cell cycle (3)(4)(5)(6)(7). Two proteins, variously estimated to be 41-42 kDa (pp42) and 44-45 kDa (pp44), were found to be transiently phosphorylated on tyrosine in response to many mitogenic agents, including peptide growth factors and phorbol esters (3)(4)(5)(6)(7)(8)(9)(10)(11).
We previously identified pp42 as mitogen-activated protein kinase (~4 2 " "~~) (12). Since pp42 and pp44 had similar onedimensional phosphopeptide maps (4,5,7), it was reasonable to expect pp44 to be a similarly activated protein kinase. Evidence presented here is consistent with this expectation. We have identified a 44-kDa protein which becomes phosphorylated on tyrosine in response to mitogenic stimulation in both murine fibroblasts and thymocytes. Using two-dimensional gel electrophoresis, this 44-kDa phosphoprotein migrates to a position similar to that observed by Cooper and Kohno for pp44 (4,5,7). In addition, the 44-kDa phosphoprotein was found to be a protein kinase, a conclusion based in part on its co-purification with MBP kinase activity over five chromatographic steps. Both kinases phosphorylated MBP at a single site (Thr-97), implying that they possess very similar substrate specificities. Finally, both pp42 and pp44 could be recognized by antisera raised against a sea star 44-kDa protein kinase (~~4 4 " '~~) .
Thus, pp42 (~4 2~~) and pp44 are structurally (4,5,7) and immunologically related serine/threonine protein kinases, with as yet indistinguishable substrate specificities. We therefore will refer to pp44 as Activation of ~4 2 " '~~ requires the phosphorylation of two closely spaced tyrosine and threonine residues (15) that were found to occur on a single tryptic peptide (27). ~4 4~ also contained a tryptic peptide with equal amounts of phosphotyrosine and phosphothreonine, and an electrophoretic mobility similar to that of ~4 2 " '~~. Similarly, the kinase activity of p U W k is also reduced by both the protein-tyrosine phosphatase CD45 and the protein-serine/threonine phosphatase 2A, thereby demonstrating the importance of both types of phosphorylation in its activation (data not shown).
Although ~4 2~~ and ~4 4 ' "~ are highly related, they can be distinguished by electrophoretic mobility and chromatographic procedures, and even though ~4 2~~ and ~4 4~ each contained a tryptic phosphopeptide with similar electrophoretic mobilities and equal amounts of phosphotyrosine and phosphothreonine, the two phosphopeptides can be separated by TLC: the ~4 2~~ phosphopeptide is less hydrophobic than that of ~4 4 " '~~. The recent determination that the tyrosine and threonine phosphorylation sites of ~4 2~~ occur within the middle of the protein (27) strongly argues against p42"'@ being a proteolytic fragment of ~4 4~~. If ~4 2~~ was a fragment of pUWk, then a single phosphopeptide common to both proteins should have been identified.
Sea star ~4 4 " '~~ also resembled the murine ~4 2~ with respect to substrate specificity, but differed in size and chromatographic behavior. In marked contrast, the sea star ~4 4 " '~~ and the murine ~4 4 " "~~ were indistinguishable by these criteria. Particularly striking was the co-purification of murine ~4 4~~ with the sea star enzyme over polylysine-agarose, phosphocellulose, Mono-Q (data not shown), and phenyl-Superose, the latter two being capable of separating ~4 2 from ~4 4~~. However, in uiuo labeling studies with ["PI orthophosphate revealed that, unlike pp42 and pp44, the activated sea star ~4 4 " '~~ was not phosphorylated on threonine (data not shown). Thus, although it is clear that the sea star ~4 4 ' "~~ is most closely related to ~4 4 -~, i t is uncertain whether these enzymes are truly homologous. to p4ZWk. We also find that Mono-Q resolves two MBP kinase peaks, but have observed the best and most consistent resolution of the kinases could be obtained with phenyl-Superose fast protein liquid chromatography. By utilizing phenyl-Superose combined with sequential Mono-& fast protein liquid chromatography, we have confirmed and extended the work of Ahn et al. (24) by identifying MBP kinase I (peak E3) and MBP kinase I1 (peak E4) as ~4 2 " ' "~~ and p44"' "pk, respectively.
Boulton et al. have cloned a rat brain cDNA (ERK 1) encoding a MAP-2 kinase which is a member of the kinase family described here (25). We have cloned and sequenced a murine cDNA corresponding to ~4 2 " '~~, and found it to encode a protein with a predicted amino acid sequence 85% identical pp44-k.
to that of ERK 1 (32). This corresponds to the ERK 2 sequence of Boulton et al. (33). Microprotein sequencing of sea star ~4 4 " '~~ revealed approximately 70% amino acid identity with the predicted sequence of ERK 1 : but a different sea star cDNA has been obtained which corresponds more closely to the sequence of ERK l.5 Thus, ERK 1 and p44"pk appear to be different members of this kinase family. It is not clear at present how many members the MAP kinase family has, but the best current estimate is a lower bound of four, based on the analyses of Boulton et al. (25,33).
Regulation of the MAP Kinase Family-Some evidence exists that ~4 2~~ and p44"@+ are differentially regulated. Acute treatment of fibroblasts with mitogens such as epidermal growth factor of phorbol esters stimulates a greater increase in ~4 2 " '~~ kinase activity than of ~4 4 "~ kinase activity. Similarly, we have found the relative amount of pp42 phosphorylation to be 4-6-fold greater than that of pp44, as determined by immunoblotting with antiphosphotyrosine antibodies, even though blotting with the anti-GLAYIGE-GAYGMV peptide antibody indicated that there were equal amounts of each protein in the sample (data not shown). Further work will be required to determine the circumstances and significance of differential activation of these highly related enzymes.
phenyl-Superose.   in a single tryptic peptide (27). To determine whether this is also true for pp44, p 4 e and pP44 were isolated from ["P]onhophosphate-labeled cells, and were Subjected to proteolylic cleavage with trypsin followed by 2-dimensional HVEI'TLC. AS previously reported (27). p42" contained a single tryptic phosphopeptide (arrow in Fig. SA). Murine pp44 also contained a single tryptic phosphopeptide with an electrophoretic mobility nearly Identical to that of the p42"' phosphopeptide. The phosphopeptldes from both enzymes contained similar amounts of phosphotyrosine and phosphothreonine (data not shown). In contrast to the phosphopeptide from ~42"~'. however, the pp44 phosphopeptide was sllghtly more hydrophobic as mdlcated by its higher mobility in the solvent chromatography dtrection (arrow in Fig. 60). A mixing experiment (Flg. 6C) confirmed their similar but non.ldentical properties (see Discussion). A.

C. Mix
The relative positions of pp44 and of p42" on these 2-D gels. and the fact that the tyrosine phosphorylation of p44 is mitogen-induced. suggest thls kmase may be pp45. one of the 42-45 kDa tyrosine kinase substrates first identified by Cooper et a/. (4) (see Discussion).
phenyl-Superose chromatography. Each phenyl-Superose fraction was monitored for EGF-stimulated Swiss 3T3 homogenate was applied to sequential DE52-cellulose and phosphotransferase activity using MAP-2 as Substrate (0). Two peaks of kinase activity were the 44 kDa kinase, as shown by immunoblotting with the anti-phosphotyrosine antiserum (inset. identified (peaks II and I). Peak I contained ~4 2 "~' and the broad wash fraction peak contained labeled arrows). Unstimulated cell extract was also subjected to the same fractionation procedures and no MBP kinase activity was identified (b) nor was any phosphoprotein detected (data not shown).
phosphoryiating activity in EGF-stimulated Swiss 3T3 fibroblasts and phorbol ester-treated EL4 Krebs and co-workers utilized Mono Q Chromatography to identity two peaks of MBP cells (24,29). To compare the kinases described by these workers to the kinases reported here. p 4 T " and p44 were isolated from PDB-treated EL4 cells, mixed together and then separated by Mono 0 chromatography (Flg. 9). Two peaks of MBP klnase activity were identified. as reponed determined by anti-phosphotyrosine immunoblonmg; the second peak at 100 mM NaCl (11) by Ahn et a/. (24). The first kinase peak (I) eluted at 80 mM NaCl and contalned p4T". as contained pp44 (inset. Fig. 9).  Panel A. "P-labeled ~4 2 "~' purified by phenyl8uperose chromatography was electrophoresed on a 2-D polyacrylamide gel. dried and exposed to film. The acidic portion of the gel is on the left. pp42A , the actlve form of ~42"~'. is identified by an arrow. Panel Chromatographic comparisons of p4* and pp44 Imm EGF-srimulated mufine 373 fibroblasts. chromatography. using MAP-2 as a substrate. The cells were Stimulated for 10 min with EGF.
Murine pp44 was a: = identified in EGF-treated Swiss 3T3 cells by phenyl-Superose followed by Sequential chromatography on DE52-cellulose and phenyl-Superose. A peak of MAP-2 phosphorylating activity eluted at the predlcted posltion of p 4 T " at 37% ethylene glycol and 100 mM NaCl (peak I, Fig. 8 ) (12). The presence of p42-was verified by immunoblotting this peak with the ant!-phosphotyrosine antibody (lane I. m e t Fig. 8). In addition to the ~4 2 "~ fractions (peak II. Fig. E). These fractions were pooled and pp44 was detected by immunoblonmg peak. a broad peak of MAP-2 phosphorylating activity eluted in the phenylduperose wash with the same antibody (lane II. inset Fig. 8). Unstlmulated cell extract. put through the same purification procedures, had no detectable MAP-2 phosphorylating activity (Fig. 8) nor was any phosphoproteln detected (data not shown). Therefore, pp44 is present in EGF-stimulated Swiss wash as does pp44 from PDB-treated EL4 cells. mouse 3T3 cells. and it phosphorylates MAP-2 like ~42"". but it elutes in the phenyl-Superose PDB-treated EL4 homogenate was Subjected to sequential chromatography on DE52-cellulose and phenyl-Superose. Both p 4 F and pp44 were identifled. pooled (data not shown), the Salt concentratron was diluted to 25 mM. and the protein sample applied to Mono Q. The phosphotransferase activity was eluted from Mono 0, using the shallow Salt gradient of Ahn @! a, (24) and was monitored (0) with MBP as substrate. A major activity peak eluted at EO mM contalned the 42 kDa phosphoprotein while peak II contained the 44 kDa phosphoprotein (inset, NaCl followed by a smaller kinase peak at 100 mM NaCl (peaks I and I N. respectively). Peak I lane I and II. respectively).