Advertisement

Phosphorylation of MEK1 by cdk5/p35 Down-regulates the Mitogen-activated Protein Kinase Pathway*

Open AccessPublished:October 29, 2001DOI:https://doi.org/10.1074/jbc.M109324200
      Cyclin-dependent protein kinase 5 (cdk5), a member of the cdk family, is active mainly in postmitotic cells and plays important roles in neuronal development and migration, neurite outgrowth, and synaptic transmission. In this study we investigated the relationship between cdk5 activity and regulation of the mitogen-activated protein (MAP) kinase pathway. We report that cdk5 phosphorylates the MAP kinase kinase-1 (MEK1) in vivo as well as the Ras-activated MEK1 in vitro. The phosphorylation of MEK1 by cdk5 resulted in inhibition of MEK1 catalytic activity and the phosphorylation of extracellular signal-regulated kinase (ERK) 1/2. In p35 (cdk5 activator) −/− mice, which lack appreciable cdk5 activity, we observed an increase in the phosphorylation of NF-M subunit of neurofilament proteins that correlated with an up-regulation of MEK1 and ERK1/2 activity. The activity of a constitutively active MEK1 with threonine 286 mutated to alanine (within a TPXK cdk5 phosphorylation motif in the proline-rich domain) was not affected by cdk5 phosphorylation, suggesting that Thr286 might be the cdk5/p35 phosphorylation-dependent regulatory site. These findings support the hypothesis that cdk5 and the MAP kinase pathway cross-talk in the regulation of neuronal functions. Moreover, these data and the recent studies of Harada et al. (Harada, T., Morooka, T., Ogawa, S., and Nishida, E. (2001) Nat. Cell Biol. 3, 453–459) have prompted us to propose a model for feedback down-regulation of the MAP kinase signal cascade by cdk5 inactivation of MEK1.
      cdk5
      cyclin-dependent protein kinase 5
      MAP
      mitogen-activated protein
      MEK1
      MAP kinase kinase 1
      CA-MEK1
      constitutively active MEK1
      ERK
      extracellular signal-regulated kinase
      GST
      glutathioneS-transferase
      PRD
      proline-rich domain
      HA
      hemagglutinin
      DN
      dominant negative
      NGF
      nerve growth factor
      cdk51 is a member of the cyclin-dependent protein kinase family (cdc2, CDC28, and other generically cyclin-dependent CDKs). Although cdk5 binds to cyclin D, its activity is not regulated by cyclins and there is little evidence that cdk5 is involved in the progression of the cell cycle (for review see Ref.
      • Morgan D.O.
      ; see also Refs.
      • Guidato S.
      • McLoughlin D.M.
      • Grierson A.J.
      • Miller C.C.
      and
      • Xiong Y.
      • Zhang H.
      • Beach D.
      ). cdk5 is active mainly in post-mitotic cells such as neurons (
      • Lew J.
      • Huang Q.Q.
      • Qi Z.
      • Winkfein R.J.
      • Aebersold R.
      • Hunt T.
      • Wang J.H.
      ,
      • Tsai L.H.
      • Delalle I.
      • Caviness Jr., V.S.
      • Chae T.
      • Harlow E.
      ), retinal cells (
      • Hirooka K.
      • Tomizawa K.
      • Matsui H.
      • Tokuda M.
      • Itano T.
      • Hasegawa E.
      • Wang J.H.
      • Hatase O.
      ), and muscle cells (
      • Philpott A.
      • Porro E.B.
      • Kirschner M.W.
      • Tsai L.H.
      ), where its activators p35 (or its truncated form p25) (
      • Lew J.
      • Huang Q.Q.
      • Qi Z.
      • Winkfein R.J.
      • Aebersold R.
      • Hunt T.
      • Wang J.H.
      ,
      • Tsai L.H.
      • Delalle I.
      • Caviness Jr., V.S.
      • Chae T.
      • Harlow E.
      ) and p39 (
      • Honjyo Y.
      • Kawamoto Y.
      • Nakamura S.
      • Nakano S.
      • Akiguchi I.
      ,
      • Humbert S.
      • Dhavan R.
      • Tsai L.
      ,
      • Wu D.C.
      • Yu Y.P.
      • Lee N.T.
      • Yu A.C.
      • Wang J.H.
      • Han Y.F.
      ,
      • Zheng M.
      • Leung C.L.
      • Liem R.K.
      ) are specifically expressed. cdk5 has been suggested to play important roles in neurite outgrowth (
      • Nikolic M.
      • Dudek H.
      • Kwon Y.T.
      • Ramos Y.F.
      • Tsai L.H.
      ,
      • Sharma M.
      • Sharma P.
      • Pant H.C.
      ), neuronal migration (
      • Chae T.
      • Kwon Y.T.
      • Bronson R.
      • Dikkes P.
      • Li E.
      • Tsai L.H.
      ,
      • Ohshima T.
      • Gilmore E.C.
      • Longenecker G.
      • Jacobowitz D.M.
      • Brady R.O.
      • Herrup K.
      • Kulkarni A.B.
      ,
      • Ohshima T.
      • Ward J.M.
      • Huh C.G.
      • Longenecker G.
      • Veeranna
      • Pant H.C.
      • Brady R.O.
      • Martin L.J.
      • Kulkarni A.B.
      ), dopamine signaling in the striatum (
      • Bibb J.A.
      • Snyder G.L.
      • Nishi A.
      • Yan Z.
      • Meijer L.
      • Fienberg A.A.
      • Tsai L.H.
      • Kwon Y.T.
      • Girault J.A.
      • Czernik A.J.
      • Huganir R.L.
      • Hemmings Jr., H.C.
      • Nairn A.C.
      • Greengard P.
      ), exocytosis (
      • Fletcher A.I.
      • Shuang R.
      • Giovannucci D.R.
      • Zhang L.
      • Bittner M.A.
      • Stuenkel E.L.
      ,
      • Matsubara M.
      • Kusubata M.
      • Ishiguro K.
      • Uchida T.
      • Titani K.
      • Taniguchi H.
      ,
      • Rosales J.L.
      • Nodwell M.J.
      • Johnston R.N.
      • Lee K.Y.
      ,
      • Shuang R.
      • Zhang L.
      • Fletcher A.
      • Groblewski G.E.
      • Pevsner J.
      • Stuenkel E.L.
      ), differentiation of muscle cells (
      • Philpott A.
      • Porro E.B.
      • Kirschner M.W.
      • Tsai L.H.
      ), and organization of acetylcholine receptors at the neuromuscular junction (
      • Fu A.K.
      • Fu W.Y.
      • Cheung J.
      • Tsim K.W.
      • Ip F.C.
      • Wang J.H.
      • Ip N.Y.
      ). Although neuronal cytoskeletal proteins were initially identified as the major target substrates (
      • Lew J.
      • Huang Q.Q.
      • Qi Z.
      • Winkfein R.J.
      • Aebersold R.
      • Hunt T.
      • Wang J.H.
      ,
      • Paudel H.K.
      • Lew J.
      • Ali Z.
      • Wang J.H.
      ,
      • Shetty K.T.
      • Link W.T.
      • Pant H.C.
      ), the number of cdk5 substrates has expanded considerably (see Table I in Ref.
      • Grant P.
      • Sharma P.
      • Pant H.C.
      ). These include DARPP-32, a dopamine and cyclic AMP-regulated phosphoprotein involved in dopamine signaling (
      • Bibb J.A.
      • Snyder G.L.
      • Nishi A.
      • Yan Z.
      • Meijer L.
      • Fienberg A.A.
      • Tsai L.H.
      • Kwon Y.T.
      • Girault J.A.
      • Czernik A.J.
      • Huganir R.L.
      • Hemmings Jr., H.C.
      • Nairn A.C.
      • Greengard P.
      ), NUDEL (a murine homolog of theAspergillus nidulans nuclear migration mutant NudE), a protein involved in neuronal migration and axon transport (
      • Sasaki S.
      • Shionoya A.
      • Ishida M.
      • Gambello M.J.
      • Yingling J.
      • Wynshaw-Boris A.
      • Hirotsune S.
      ), and other proteins involved in cross-talk between protein kinases and phosphatases (
      • Bibb J.A.
      • Nishi A.
      • O'Callaghan J.P.
      • Ule J.
      • Lan M.
      • Snyder G.L.
      • Horiuchi A.
      • Saito T.
      • Hisanaga S.
      • Czernik A.J.
      • Nairn A.C.
      • Greengard P.
      ). cdk5 also modulates protein kinase reactions such as the small GTPase-Rac dependent phosphorylation of p21-activated kinase, which results in modification of the actin cytoskeleton (
      • Nikolic M.
      • Chou M.M.
      • Lu W.
      • Mayer B.J.
      • Tsai L.H.
      ). By virtue of phosphorylating these diverse substrates, cdk5 plays a multifunctional role in the nervous system.
      It has been demonstrated that the absence of cdk5 in cdk5 −/− mice results in embryonic lethality (
      • Ohshima T.
      • Ward J.M.
      • Huh C.G.
      • Longenecker G.
      • Veeranna
      • Pant H.C.
      • Brady R.O.
      • Martin L.J.
      • Kulkarni A.B.
      ). Although the p35 knockout mice survive longer (
      • Chae T.
      • Kwon Y.T.
      • Bronson R.
      • Dikkes P.
      • Li E.
      • Tsai L.H.
      ), both cdk5 −/− and p35 −/− mice exhibit similar defects in cortical neuronal migration and affect the development of the nervous system (
      • Chae T.
      • Kwon Y.T.
      • Bronson R.
      • Dikkes P.
      • Li E.
      • Tsai L.H.
      ,
      • Ohshima T.
      • Gilmore E.C.
      • Longenecker G.
      • Jacobowitz D.M.
      • Brady R.O.
      • Herrup K.
      • Kulkarni A.B.
      ,
      • Ohshima T.
      • Ward J.M.
      • Huh C.G.
      • Longenecker G.
      • Veeranna
      • Pant H.C.
      • Brady R.O.
      • Martin L.J.
      • Kulkarni A.B.
      ). We observed that in cdk5 −/− mice brain stem neurons showed ballooning and hyperphosphorylation of cytoskeletal proteins as detected by the SMI31 antibody (see Fig. 1). Similar observations were obtained from p35 (−/−) mice.
      I. Vincent, personal communication.
      2I. Vincent, personal communication.
      The antibody cross-reacts with phosphorylated Lys-Ser-Pro (KSP) motifs in neurofilament proteins, tau, and MAPs (
      • Sternberger L.A.
      • Sternberger N.H.
      ), sites that are specifically targeted by proline-directed kinases such as cdk5 and MAP kinases (
      • Shetty K.T.
      • Link W.T.
      • Pant H.C.
      ,
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ). The data suggested that in the absence of cdk5 activity, other proline-directed protein kinases were up-regulated. The findings that KSP motifs in rat NF proteins (particularly, NF-M) are preferentially phosphorylated by ERK1/2 (
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ) prompted us to examine the relationship between cdk5/p35 and MAP kinase activities in vitro andin vivo.
      Figure thumbnail gr1
      Figure 1A, immunohistochemical staining of 18-day-old embryonic brain stem sections from cdk5 −/− and cdk5 +/+ mice with SMI31 antibody. Note the intense immunostaining of hyperphosphorylated cytoskeletal proteins in large brain stem neurons of cdk5 −/− mice (arrows). These neurons in cdk5 +/+ mice were not immunostained (arrowheads). B, immunoblot analysis of phosphorylated cytoskeletal protein preparations from 4-week-old p35 −/− and wild type mice brains using phosphoepitope-specific SMI-31 antibody. The SMI-31 immunoreactivity of NF-M, particularly from the cortex and cerebella in p35 −/− mice, was severalfold higher than in p35 +/+ mice.
      The MAP kinases mediate a wide range of cellular functions via a variety of signal transduction pathways (
      • Pearson G.
      • Robinson F.
      • Beers Gibson T.
      • Xu B.
      • Karandikar M.
      • Berman K.
      • Cobb M.H.
      ,
      • Schaeffer H.J.
      • Weber M.J.
      ). In one well studied pathway, the binding of GTP to Ras protein initiates a phosphorylation cascade through Raf-1 and MEK1/2 (MAPK kinase), which results in stimulation of the MAP kinases, ERK1/2. Upon stimulation, ERKs are known to phosphorylate a variety of cytosolic substrates and are also translocated into the nucleus where they initiate the transcription of immediate early genes (
      • Cowley S.
      • Paterson H.
      • Kemp P.
      • Marshall C.J.
      ). The Ras-Raf-MEK-ERK pathway is stimulated by various growth factors and extracellular stimuli and plays important roles in cell survival, differentiation, and proliferation. This pathway interacts (cross-talks) with other signal transduction cascades, either because of overlapping substrate specificity, shared regulatory sites (
      • Pearson G.
      • Robinson F.
      • Beers Gibson T.
      • Xu B.
      • Karandikar M.
      • Berman K.
      • Cobb M.H.
      ), and/or associations with shared scaffolding proteins (
      • Whitmarsh A.J.
      • Davis R.J.
      ).
      To explore the nature of interactions between cdk5 and the MAP kinase signaling cascade, we studied the effect of cdk5 on MEK1 activityin vitro and in vivo. In this report we provide evidence that cdk5 regulates the MAP kinase pathway in a negative manner via phosphorylation of MEK1.

      EXPERIMENTAL PROCEDURES

      Materials

      All fine chemicals were purchased from Sigma unless indicated. [γ-32P]ATP and [32P]orthophosphate were purchased from Amersham Pharmacia Biotech. The glutathione-Sepharose beads were a product of Sigma Life Sciences. Roscovitine was a product of BioMol.

      Plasmids and Expressed Proteins

      A constitutively active mutant (CA-MEK1) (engineered by deleting residues 32–51 from the N terminus of MEK1 and by mutating its Ser218 and Ser222 to Glu and Asp, respectively (
      • Mansour S.J.
      • Matten W.T.
      • Hermann A.S.
      • Candia J.M.
      • Rong S.
      • Fukasawa K.
      • Vande Woude G.F.
      • Ahn N.G.
      )) or a T286A mutant was used for cell transfection (as HA tag) as well as for bacterial protein expression (His6 and/or GST-tagged) purposes. The mutant (T286A) was created in a plasmid encoding CA-MEK1 using a Quick Change site-directed mutagenesis kit (Stratagene). CA-MEK1(K97M) was engineered by mutating Lys97 to Met in the CA-MEK1 plasmid. This was used as a template for making T286A(K97M) by mutating Thr286 to Ala. CA-MEK1 and its variant proteins (T286A, CA-MEK1(K97M), and T286A(K97M)) were bacterially expressed with His6 tag (
      • Mansour S.J.
      • Matten W.T.
      • Hermann A.S.
      • Candia J.M.
      • Rong S.
      • Fukasawa K.
      • Vande Woude G.F.
      • Ahn N.G.
      ), and cdk5 and p35 were expressed as GST fusion proteins as described previously (
      • Lew J.
      • Huang Q.Q.
      • Qi Z.
      • Winkfein R.J.
      • Aebersold R.
      • Hunt T.
      • Wang J.H.
      ). cdk5 and cdk5 dominant negative (DN) constructs for cell transfections were in pcDNA3.1His vector and were expressed as His6 tag proteins (gift from Dr. Li-Huei Tsai, Harvard Medical School). The CMV-p35 plasmid was a gift from Dr. Li Tsai (Harvard Medical School). Raf-activated MEK1 (GST fused at the N terminus and His6 fused at the C terminus) was purchased from Upstate Biotech Industries.

      Cell Culture, Transfection, Metabolic Labeling, and Immunoblotting

      Cortices from 18-day-old rat embryos were dissected, and the cortical neuronal cell cultures were grown on polylysine-treated 6-well cell culture dishes. The cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum for 7–8 days before drug treatment.
      PC12 cells were cultured in Dulbecco's modified Eagle's medium containing 12.5% fetal horse serum and 2.5% fetal bovine serum, and NIH 3T3 cells were cultured in 10% fetal bovine serum as described earlier (
      • Cowley S.
      • Paterson H.
      • Kemp P.
      • Marshall C.J.
      ). The cells were serum-starved by culturing in medium containing 1% fetal bovine serum for 16 h prior to both transfection and roscovitine inhibition experiments.
      cDNA encoding HA-tagged CA-MEK1 or its variant, T286A, His6-tagged cdk5, or its kinase defective mutant cdk5(DN) and CMV-p35 (
      • Lew J.
      • Huang Q.Q.
      • Qi Z.
      • Winkfein R.J.
      • Aebersold R.
      • Hunt T.
      • Wang J.H.
      ) were transfected in NIH 3T3 or PC12 cells using LipofectAMINE PLUS (Invitrogen) following the manufacturer's instructions. Briefly, 2 μg of each plasmid DNA was used in 35-mm collagen-coated dishes to perform transfections for 60 h. In some experiments cells were treated with NGF (50 ng/ml) 12 h after transfection for every 24 h. Subsequently, the cells were lysed and used for immunoblotting after normalizing the protein using anti-phospho-ERK1/2 or anti-ERK1/2 antibodies (New England Biolabs). The phosphorylated form of ERK1/2 is indicated as pp-ERK1/2 in the figures, whereas ERK1/2 indicates total amount of ERK1/2. For in vivo labeling experiments, the cells were incubated in phosphate-deficient Dulbecco's modified Eagle's medium 2 h prior to incubation in [32P]orthophosphoric acid (0.2 mCi/ml) for 3 h. The cell lysates were prepared in a buffer containing 50 mm Tris, pH 7.5, 1 mm EDTA, 0.1% Nonidet P-40, 50 μm β-glycerophosphate, 50 μm sodium fluoride, 0.1 μm sodium vanadate, and protease inhibitor mixture (Roche Molecular Biochemicals). An enhanced chemiluminescence (Amersham Biosciences, Inc. or Pierce) method was used for immunoblotting following manufacturer's protocol in all experiments. Anti-HA antibody (Roche Molecular Biochemicals) was used to immunoprecipitate CA-MEK1 or T286A as described earlier (
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ).

      Kinase Assay and Metabolic Labeling

      cdk5 kinase assays were performed in a total volume of 50 μl by incubating a preformed complex of bacterially expressed GST-cdk5 and GST-p25, a truncated form of p35 (
      • Lew J.
      • Huang Q.Q.
      • Qi Z.
      • Winkfein R.J.
      • Aebersold R.
      • Hunt T.
      • Wang J.H.
      ,
      • Tsai L.H.
      • Delalle I.
      • Caviness Jr., V.S.
      • Chae T.
      • Harlow E.
      ) and 1 μg of either Raf-phosphorylated or unphosphorylated GST-MEK1-His6 (N terminus tagged with GST and C terminus tagged with His6) (Upstate Biotech Industries) or bacterially expressed CA-MEK1 or its variants, in a buffer containing 20 mm Tris, pH 7.4, 1 mmEDTA, 10 mm MgCl2, 10 μm sodium fluoride, 10 μm β-glycerophosphate, 1 μmsodium vanadate, protease inhibitor mixture (Roche Molecular Biochemicals), 100 μm [γ-32P]ATP for 60–90 min at 30 °C. The reaction was stopped by boiling the samples in Laemmli's sample buffer. The phosphate incorporation was detected by autoradiography of the protein gels. A similar procedure was used for assessing ERK2 activity using myelin basic protein or a synthetic KSPXK peptide derived from NF-H (VKSPAKEKAKSPEK) (
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ) as the substrate. The reaction mixture was spotted on phospho-cellulose paper (Whatman), and the phosphate incorporation was measured by scintillation counting as described previously.
      To examine the effect of cdk5/p25 phosphorylation on MEK1 activity, similar kinase assays were performed using unlabeled ATP (1 mm). GST-Sepharose was used to concentrate MEK1 because it was GST-fused, and then the fusion protein-coupled Sepharose beads were used to phosphorylate 1 μg of bacterially expressed GST-ERK2 as described above for the cdk5 assays. The reaction mixture was immunoblotted using anti-phospho-ERK1/2 antibody (New England Biolabs) to assess the MEK1 activity. When radiolabeled [γ-32P]ATP was used, the phosphate incorporation was observed by autoradiography of the protein gels or scintillation counting as described above.

      Analysis of NF Proteins, cdk5, and MEK1 Activity in p35 −/− Mice

      The p35 −/− mice were created as described earlier (
      • Chae T.
      • Kwon Y.T.
      • Bronson R.
      • Dikkes P.
      • Li E.
      • Tsai L.H.
      ,
      • Ohshima T.
      • Ogawa M.
      • Veeranna
      • Hirasawa M.
      • Longenecker G.
      • Ishiguro K.
      • Pant H.C.
      • Brady R.O.
      • Kulkarni A.B.
      • Mikoshiba K.
      ). Lysates were prepared from the cerebral cortex and cerebellar tissues of 3–4-week-old p35 −/− mice as described above for PC12 cells. Cytoskeletal extracts containing NF proteins were prepared according to previously published procedures (
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ). The protein levels were normalized, and cdk5 and MEK1 were immunoprecipitated by using anti-cdk5 (C-8, Santa Cruz) and anti-MEK1/2 (New England Biolabs) antibodies, respectively. Bacterially expressed GST-ERK2 was used as the substrate for MEK1 assays, whereas VKSPAKEKAKSPEK, a synthetic KSPXK peptide derived from the sequence of neurofilament-H, was used for cdk5/p35 assays as described previously (
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ). The lysates were immunoblotted using phospho-ERK1/2 or ERK1/2 antibodies.

      RESULTS

      Neurofilament Protein Hyperphosphorylation and MAP Kinase Activities in p35 −/− Mice

      It has been observed that neurofilament and microtubule-associated proteins are hyperphosphorylated in neurons of cdk5 −/− mice (
      • Ohshima T.
      • Ward J.M.
      • Huh C.G.
      • Longenecker G.
      • Veeranna
      • Pant H.C.
      • Brady R.O.
      • Martin L.J.
      • Kulkarni A.B.
      ). Because cdk5 activity is dependent on p35, we examined whether p35 −/− mice would show hyperphosphorylation of neurofilament proteins. Unlike cdk5 −/− mice, the p35 −/− mice are viable after birth, and therefore, the phosphorylation levels of neurofilament proteins were easily followed by SMI31 antibody. Fig. 1Ashows the ballooning and accumulation of hyperphosphorylated anti-SMI31 epitope immunoreactive proteins in the brain stem neurons of cdk5 −/− mice. To further verify this observation, the cytoskeletal protein fraction from the cortex and cerebella of 3–4-week-old p35 −/− and +/+ wild type mice were analyzed by immunoblotting with SMI-31 antibody, which specifically recognizes phosphorylated KSP sites on neurofilament and microtubule-associated proteins. As shown in Fig.1B, the immunoreactivity of NF-M to SMI31 in the cortex of p35 −/− mice was severalfold higher than in the wild type mice. However, in contrast to the observations in the cortex, the NF-M from the control and p35 −/− mice cerebella showed fewer significant differences in immunoreactivity to SMI-31. However, the intensity of immunoreactivity of NF-H to SMI-31 in wild type and knockout mice was very similar. It should be noted that the rodent NF-M is a preferred substrate for ERK1/2 phosphorylation as compared with cdk5 (
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ), therefore suggesting that in p35 −/− mice the absence of cdk5 activity might have up-regulated ERK1/2. It has been reported earlier that the brain extracts from p35 −/− mice exhibited insignificant levels of cdk5 activity (
      • Chae T.
      • Kwon Y.T.
      • Bronson R.
      • Dikkes P.
      • Li E.
      • Tsai L.H.
      ). It is possible that the cerebellum might contain higher levels of p39, a cdk5 activator present in both wild type and mutant mice that could possibly compensate for the absence of p35 (
      • Wu D.C.
      • Yu Y.P.
      • Lee N.T.
      • Yu A.C.
      • Wang J.H.
      • Han Y.F.
      ,
      • Zheng M.
      • Leung C.L.
      • Liem R.K.
      ). These data suggested that cdk5 activity in +/+ mouse cortex might inhibit the activity of other proline-directed kinases like ERK1/2 that are known to preferentially phosphorylate NF-M.
      Because MEK1 is a key regulator in the MAP kinase pathway, we immunoprecipitated MEK1 from the cerebral cortex of p35 −/− and +/+ mice to examine whether these two preparations had a differential effect on ERK2 phosphorylation and activity. The animals used for these studies were 3–4 weeks old because MEK1 is expressed at significant levels mainly in the adult brain (
      • Alessandrini A.
      • Brott B.K.
      • Erikson R.L.
      ). The MEK1 activity as measured by ERK1/2 phosphorylation was 60–75% higher in the brain extract from the p35 −/− mice as compared with that observed in p35 +/+ mice (Fig.2, B and C). This increase in MEK1 activity correlated with the observed decrease in cdk5 activity in p35 −/− mice (Fig. 2A). The levels of total MEK1 were the same in p35 −/− and +/+ mice as measured by immunoblotting (data not shown). Interestingly, not only did the level of ERK1/2 phosphorylation increase in the p35 −/− mice (Fig.2B), but the amount of phosphorylated ERK1/2 also increased, although the amount of total ERK1/2 remained unchanged (Fig.2D). These data prompted the idea that in vivocdk5/p35 and MEK1 cross-talk might result in regulation of the MAP kinase pathway.
      Figure thumbnail gr2
      Figure 2p35 −/− mice showed elevated levels of phospho-ERK2 and enhanced MEK1 activity.A, the p35 −/− mice showed basal levels of cdk5 activity. cdk5 was immunoprecipitated from the cerebral cortex of age-matched (3–4 weeks) p35 +/+ and p35 −/− mice, and the kinase activity was determined by using a cdk5-specific peptide substrate (
      • Pearson G.
      • Robinson F.
      • Beers Gibson T.
      • Xu B.
      • Karandikar M.
      • Berman K.
      • Cobb M.H.
      ). B, MEK1 activity in p35 −/− mice was increased over the wild type. MEK1 was immunoprecipitated from cortex extracts of 3–4-week-old p35 −/− or +/+ mice by using MEK1-specific antibody, and the MEK1 immunoprecipitates were then used to phosphorylate bacterially expressed ERK2. The autoradiogram shows the increased phosphorylation of ERK2 in p35 −/− mice. C, MEK1 activity was quantitated from three separate ERK2 phosphorylation experiments described forB. D, the blots from 3–4-week-old p35 −/− and +/+ mice were immunostained with anti-phospho-ERK1/2 (pp-ERK1/2) or anti-ERK1/2. Results representative of four different experiments are shown here. p35 −/− mice showed elevated levels of phospho-ERK2, but the levels of total ERK1/2 were not affected.

      cdk5 Phosphorylates and Inhibits Activated MEK1

      To further explore the relationship between cdk5 and MEK1, we compared thein vitro phosphorylation of bacterially expressed MEK1 (inactive), Raf-phosphorylated MEK1 (active), and constitutively active MEK1 (CA-MEK1) by cdk5/p25. Although inactive (unphosphorylated) MEK1 was not phosphorylated by cdk5/p25 (Fig.3A, lane 5), Raf-phosphorylated MEK1 (Fig. 3A, lane 2) and CA-MEK1 (not shown here) were good substrates of cdk5/p25. These data suggested that the Raf-activated MEK1 served as a substrate for cdk5/p25.
      Figure thumbnail gr3
      Figure 3A, cdk5/p25 phosphorylated Raf-activated MEK1 in vitro. Autoradiogram showing bacterially expressed GST-MEK1-His6, previously phosphorylated by Raf, that was incubated alone (lane 1) or with cdk5/p25 (lane 2) for 60 min and then subjected to SDS-PAGE and autoradiography. The right panel (lanes 4 and 5) shows a similar experiment done with unphosphorylated (inactive) MEK1 and cdk5/p25. Note the high level of MEK1 phosphorylation in lane 2. The data presented are representative of four experiments.B, the phosphorylation of active MEK1 by cdk5/p25 resulted in inhibition of its ability to phosphorylate ERK2. Left panel, Raf-phosphorylated MEK1 was incubated in vitrowith (lanes 1) or without cdk5/p25 (lanes 3) for 2 h in the presence of unlabeled ATP. Lane 2 is the control without MEK1. The modified MEK1 was used to phosphorylate equal amounts of GST-ERK2. The level of phospho-ERK 2 was reduced after cdk5 phosphorylation of MEK1 as shown in the Western blot using phospho-ERK2 antibody (lane 1 compared with lane 3).Right panel, ERK 2 phosphorylation by active MEK1 was performed by incubating it with cdk5/p25 and [γ-32P]ATP in the assay mix for 1 h, and the phosphate incorporation was detected by autoradiography of protein gels. Note that a similar decrease in MEK1 activity by cdk5-mediated phosphorylation was observed (compare lanes 1 and 2). C, activation of ERK2 by CA-MEK1 is suppressed by cdk5/p25 phosphorylation.Upper panel, in vitro kinase assays for ERK2 activity were performed using a synthetic peptide derived from NF-H as the substrate (
      • Pearson G.
      • Robinson F.
      • Beers Gibson T.
      • Xu B.
      • Karandikar M.
      • Berman K.
      • Cobb M.H.
      ) using 0.3 μg of CA-MEK1 and 1 μg of ERK2 in the presence or absence of cdk5/p35 as indicated. Lower panel, ERK2 was phosphorylated by MEK1 in the presence or absence of cdk5/p25 for 1 h, and the [32P]phosphate incorporation was measured from the autoradiograms. The data are representative of three separate experiments.
      The effect of cdk5/p25-mediated phosphorylation on MEK1 catalytic activity was then tested using expressed ERK2 as its substrate. In experiments described here the Raf-modified MEK1 with or without cdk5/p35 phosphorylation was used to phosphorylate ERK2. Immunoblot analyses using a phospho-ERK1/2-specific antibody that detects the phosphorylation at the regulatory T and Y residues in the activation loop of ERK2 showed a significant decrease in MEK1 activity (Fig.3B, left panel, lanes 1 and3). Similarly, when the phosphorylation of ERK2 was followed by 32P incorporation, cdk5/p35-mediated phosphorylation of Raf-1, phosphorylated MEK1 showed a decreased phosphorylation of ERK2 (Fig. 3B, right panel, lane 1 versus lane 2). A quantitative measurement of [32P]phosphate incorporation into ERK2 suggested a 75% decrease in MEK1 activity as a result of cdk5/p25 phosphorylation (Fig. 3C, lower panel). Similarly, in another experiment, when an NF-M peptide containing the KSP motif KAKSPVPKSPVEEVKP, a preferred substrate for ERK (
      • Veeranna
      • Amin N.D.
      • Ahn N.G.
      • Jaffe H.
      • Winters C.A.
      • Grant P.
      • Pant H.C.
      ), was incubated in an assay mixture containing ERK2 and CA-MEK1 with and without cdk5/p25, the phosphorylation of the peptide was reduced by ∼40% in the presence of cdk5/p25 (Fig. 3C,upper panel). These experiments supported the idea that the activation of ERK2 by CA-MEK1 is inhibited by cdk5/p35-mediated phosphorylation of CA-MEK1.

      cdk5 Inhibits the MAP Kinase Pathway in PC12 Cells and Cortical Neurons

      NGF stimulates the Ras-Raf-MEK-ERK (MAP kinase) pathway in PC12 cells, which results in neuronal differentiation (
      • Cowley S.
      • Paterson H.
      • Kemp P.
      • Marshall C.J.
      ). Also, cdk5 is active in PC12 cells because p35 is endogenously expressed in these cells (
      • Yan G.Z.
      • Ziff E.B.
      ). To examine the effect of cdk5 on the MAP kinase pathway, PC12 cells were treated with roscovitine, a specific cdk5 inhibitor shown to inhibit endogenous cdk5 activity in cultured cells (
      • Bibb J.A.
      • Snyder G.L.
      • Nishi A.
      • Yan Z.
      • Meijer L.
      • Fienberg A.A.
      • Tsai L.H.
      • Kwon Y.T.
      • Girault J.A.
      • Czernik A.J.
      • Huganir R.L.
      • Hemmings Jr., H.C.
      • Nairn A.C.
      • Greengard P.
      ,
      • Patrick G.N.
      • Zukerberg L.
      • Nikolic M.
      • de la Monte S.
      • Dikkes P.
      • Tsai L.H.
      ). Subsequent treatment of these cells for 25 min with NGF stimulated the MAP kinase pathway as indicated by enhanced phosphorylation of ERK1/2 (Fig.4A, lane 2). Interestingly, when the cells were stimulated with NGF in the presence of roscovitine, the increase in ERK phosphorylation was about 3-fold higher (Fig. 4A, lane 3). The effect of cdk5 on the kinetics of MAP kinase activation was also tested (Fig.4B). PC12 cells were treated with NGF for different times in the presence or absence of roscovitine. In the absence of roscovitine, NGF stimulated the MEK1-dependent MAP kinase pathway in a manner reported by several groups. The MEK activity (as judged by phosphorylated ERK1/2 levels) was near maximal at 20 min after NGF treatment. In the presence of roscovitine, however, there was a slight increase in phospho-ERK1/2 levels after 15 min of NGF treatment followed by a significant increase in phosphorylation of ERK1/2 between 15 and 25 min, further suggesting that inhibition of cdk5 activity enhances the activation of the MAP kinase pathway. Interestingly, the maximal effect of roscovitine was observed when the MAP kinase pathway or MEK1 was substantially activated and is consistent with the data presented in Fig. 3A. A similar increase in ERK1/2 phosphorylation was also observed when rat cortical neurons were treated with 50 μm roscovitine, (Fig. 4C,lane 2), suggesting that cdk5 inhibits the MAP kinase pathway in primary neuron cultures.
      Figure thumbnail gr4
      Figure 4cdk5/p35-mediated phosphorylation of MEK1inhibits the stimulation of the MAP kinase pathway.A, PC12 cells were incubated with dimethyl sulfoxide (lane 2) or 50 μm roscovitine (lane 3) for 30 min prior to 25 min of stimulation with NGF. Cell lysates were prepared in 2% SDS and immunoblotted with phospho-ERK1/2 antibodies, and the densitometric quantitation is expressed in optical density units (ODU). A significant increase in phospho-ERK1/2 level was observed upon roscovitine treatment in the absence of any change in non-phospho-ERK 1/2. B, PC12 cells were treated with roscovitine or only with dimethyl sulfoxide for 30 min prior to stimulation of the cells with NGF for indicated times. The activation of the MAP kinase pathway or MEK activity was estimated by quantitation of the phospho-ERK1/2 bands from the immunoblots. Activation in the presence of roscovitine was most evident after 15 min of NGF treatment. This experiment was repeated two times. C, cortical neurons from 18-day-old embryonic rats were treated with dimethyl sulfoxide alone (lane 1) or with roscovitine (50 μm) for 20 min. Immunoblotting was performed on cell extracts using phospho-ERK1/2 antibody. A significant increase in phospho-ERK1/2 level was observed upon roscovitine treatment. D, PC12 cells were transfected with plasmids encoding HA-CA-MEK1, CMV-p35, His6-cdk5, and His6-cdk5(DN), a kinase defective mutant as indicated by the labels. Immunoblots of cell lysates with anti-HA antibody showed no decrease in transfected CA-MEK1 (lane 1). The expression level, however, of phospho-ERK 1/2 was significantly lower in the presence of cdk5/p35 (lane 2) than in either the control (lane 1) or in the presence of DNcdk5 (lane 3). For the experiments in the right panel His6-cdk5 and CMV-p35 followed treatment with NGF (50 ng/ml) every 24 h starting 12 h after transfection. The cell lysates were prepared after 60 h and immunoblotted with anti-phospho-ERK1/2 or anti-ERK1/2 antibodies. The data in Aand C are representative of three independent experiments, and the densitometric quantitation is expressed in optical density units (ODU). DMSO, dimethyl sulfoxide.
      To investigate whether this cdk5-mediated down-regulation of the MAP kinase pathway was also due to inhibition of MEK1 in cultured cells, cdk5 and p35 plasmids were co-transfected with CA-MEK1 in PC12 cells in the absence of NGF treatment (Fig. 4D, left panel). CA-MEK1 transfection caused ERK1/2 phosphorylation of the activation loop (Fig. 4D, lane 1). Overexpression of cdk5/p35 along with CA-MEK1 resulted in a 4-fold decrease in phosphorylation of ERK1/2 (Fig. 4D, lane 2). Overexpression of a mutant of cdk5 (cdk5DN) with only 10% of cdk5 activity (
      • Tsai L.H.
      • Delalle I.
      • Caviness Jr., V.S.
      • Chae T.
      • Harlow E.
      ,
      • Nikolic M.
      • Chou M.M.
      • Lu W.
      • Mayer B.J.
      • Tsai L.H.
      ), together with p35, did not produce a similar reduction in phosphorylation of ERK1/2 (Fig. 4D, lane 3). This suggested that the down-regulation of the MAP kinase pathway was due to cdk5 catalytic activity.
      The long term NGF-mediated activation of endogenous MEK1 and ERK1/2 was also inhibited by cdk5/p35 overexpression in the PC12 cells (Fig.4D, right panel). Although the cell lysates were analyzed at 60 h, long after the early induction of high levels of endogenous ERK and MEK1 by NGF, the inhibition by cdk5/p35 was evident.

      Thr286 of MEK1 Is a Putative cdk5 Target Site

      To determine whether cdk5/p25 phosphorylates MEK1 at a threonine residue, cdk5/p25-phosphorylated MEK1 was immunoblotted with a phosphothreonine-specific antibody. As shown in Fig.5A (lane 1), cdk5 phosphorylated threonine residues on MEK1. Significantly, the cdk5 consensus motifs are absent in MEK2, and MEK2 was not phosphorylated by cdk5/p25 (data not shown).
      Figure thumbnail gr5
      Figure 5Thr286 of MEK1 is a putative site for cdk5/p35 phosphorylation.A, MEK1 is phosphorylated at a threonine residue. Raf-activated MEK1 was phosphorylated by cdk5/p25 (lane 1) in vitro as described for Fig.A. The reaction mixture was immunoblotted using a phosphothreonine antibody. Lane 2 and 3 represent controls with Raf-phosphorylated MEK1 alone and cdk5/p25 alone, respectively. B, schematic diagram showing the domain structure of MEK1. The regulatory sites in the activation loop as well as the putative cdk5 consensus sites in the PRD are indicated inbold type. C, NIH 3T3 cells were co-transfected with plasmids encoding HA-CA-MEK1 or its variant (HA-CA-T286A) along with plasmids encoding for cdk5 and p35. The cell lysates were prepared 60 h after transfection, and immunoblotting was performed using anti-phospho-ERK2 antibody.
      Only two threonine residues (Thr286 and Thr292) are localized within the cdk5 consensus motifs (TPXK) on MEK1. The sites of Raf phosphorylation (Ser218 and Ser222) in the activation loop and a proline-rich domain (PRD) at its C-terminal region, which contains the two threonine residues located within the cdk5 consensus motifs, are shown in Fig.5B. It has been reported that although ERK2 phosphorylated Thr292 of MEK1, this phosphorylation did not inhibit MEK1 activity (
      • Gardner A.M.
      • Vaillancourt R.R.
      • Lange-Carter C.A.
      • Johnson G.L.
      ). However, a related cyclin-dependent kinase, p34cdc2, found in mitotically active cells, phosphorylated MEK1 at these two threonine sites and inactivated its enzymatic activity (
      • Rossomando A.J.
      • Dent P.
      • Sturgill T.W.
      • Marshak D.R.
      ). Therefore, it was reasonable to assume that phosphorylation of Thr286 and/or Thr292 by neuronal-specific cdk5/p35 could also inhibit MEK1 activity.
      To ascertain which threonine residue in MEK1 is the putative site for cdk5/p35 phosphorylation, NIH3T3 cells were co-transfected with plasmids encoding HA-CA-MEK1 or HA-CA-MEK1 (Thr 286A) together with cdk5/p35 (Fig. 5C). A significant decrease in MEK1 activity (as judged by reduced phospho-ERK levels) resulted upon co-transfection of CA-MEK1 with cdk5 and p35 (Fig. 5C, lane 1compared with lane 2). On the other hand, co-transfection of CA-MEK1 (T286A) with cdk5/p35 did not show any significant change in phospho-ERK levels compared with the control (Fig. 5C,lane 3), suggesting that Thr286 in MEK1 is a site of cdk5/p35 phosphorylation that inhibits MEK1 activity. The levels of total ERK1/2 were not affected in these experiments (data not shown). These data do not preclude the possibility that Thr292 was also phosphorylated by cdk5/p35. The fact, however, that phosphorylation of Ser292 alone by ERK had no effect on MEK1 (
      • Gardner A.M.
      • Vaillancourt R.R.
      • Lange-Carter C.A.
      • Johnson G.L.
      ) implies that phosphorylation of Thr286is necessary and may be sufficient to inhibit MEK1 activity.

      DISCUSSION

      MEK1 occupies a central position in the network of interactive signaling cascades in all cells. Its target specificity, extent of activation, and localization in cells are controlled by complex formation with other kinases and non-kinase scaffolding proteins (
      • Whitmarsh A.J.
      • Davis R.J.
      ,). Signal cascades may be regulated positively or negatively by a variety of factors including cross-talk interactions between components of specific signaling pathways (
      • Hucho F.
      • Buchner K.
      ,
      • Xia Z.
      • Dickens M.
      • Raingeaud J.
      • Davis R.J.
      • Greenberg M.E.
      ). For example, the Rho family of G-proteins may cooperate with Raf-1 to activate the Erk pathway (
      • Frost J.A.
      • Steen H.
      • Shapiro P.
      • Lewis T.
      • Ahn N.
      • Shaw P.E.
      • Cobb M.H.
      ). Our results, on the other hand, show that a neuronal-specific cdk5/p35 complex phosphorylated MEK1 in vitro and in vivo, which resulted in a reduction of MEK1 activity. The cdk5/p35-mediated decrease in MEK1 activity down-regulated the MAP kinase pathwayin vivo. The data also suggest that for cdk5 to down-regulate MEK1, the latter must be in an activated state, phosphorylated at Ser218 and Ser222 in the T-loop by Raf. It implies that cdk5 regulation occurs only after the MAP kinase cascade has been stimulated by cellular signals that interact with diverse surface receptors (
      • Pearson G.
      • Robinson F.
      • Beers Gibson T.
      • Xu B.
      • Karandikar M.
      • Berman K.
      • Cobb M.H.
      ,
      • Widmann C.
      • Gibson S.
      • Jarpe M.B.
      • Johnson G.L.
      ). Furthermore, phosphorylation of the Thr 286 residue in the PRD of MEK1 inhibited MEK1 and ERK activity, suggesting it as the putative site of cdk5/p35 phosphorylation. This has led us to propose that a conformational change induced by the Raf activation of MEK1 may be required for phosphorylation of Thr286 by cdk5. It is possible that the conformational change in MEK1 caused by cdk5 phosphorylation of Thr 286 is also unfavorable for MEK1 interaction with ERK1/2, thereby inhibiting the phosphorylation and activation of the latter. This, in turn, would switch off the MAP kinase signaling cascade in stimulated cells.
      A related cyclin-activated kinase, p34cdc2, active during the cell division cycle, also inactivates MEK1 by phosphorylation in vivo and in vitro at sites Thr286 and Thr292 in the PRD (
      • Rossomando A.J.
      • Dent P.
      • Sturgill T.W.
      • Marshak D.R.
      ). It was suggested that this phosphorylation might act as a feedback regulator to shut down the cell cycle. It appears, therefore, that the PRD may be a critical domain for the regulation of MEK1 catalytic activity in both proliferating and terminally differentiated cells such as neurons.
      The PRD of MEK1 seems to be principally involved in modulating the efficient activation of ERK1/2 in the MAP kinase cascade (
      • Dang A.
      • Frost J.A.
      • Cobb M.H.
      ). On the one hand, deletion of the PRD domain residues (265) has no effect on MEK1 binding to Raf or its activity in vitro (
      • Dang A.
      • Frost J.A.
      • Cobb M.H.
      ). On the other hand, a similar PRD deletion in MEK1 (residues 270–307) blocked MEK1-Raf binding and decreased MEK1 activity in response to growth factors (
      • Catling A.D.
      • Schaeffer H.J.
      • Reuter C.W.
      • Reddy G.R.
      • Weber M.J.
      ). The contradictory results could be attributed to a difference in the residues, i.e. residues between 301 and 307 may be essential for PRD activity.
      Formation of the Raf-MEK1 complex seems to be essential for downstream signaling, and complex formation is modulated by phosphorylation of sites in the PRD domain. For example, phosphorylation of Ser298 by p21-activated kinase 1 enhances activation of MEK1 by promoting MEK1-Raf binding (
      • Frost J.A.
      • Steen H.
      • Shapiro P.
      • Lewis T.
      • Ahn N.
      • Shaw P.E.
      • Cobb M.H.
      ). This is consistent with the observation that mutation of sites Ser298 and Thr292 to Ala inhibited MEK-Raf binding. Evidently, conformational changes induced by the additional negativity of the phosphate groups favor Raf binding and MEK1 activation. Our results suggest that phosphorylation of Thr286 by neuronal-specific cdk5/p35 inhibits MEK activity. It may do so by virtue of a conformational change in MEK1 that may affect binding of the activated Raf-MEK1 complex to other proteins essential for downstream activation of ERK1/2. Apparently, different conformational changes may be induced upon phosphorylation of different residues in the PRD domain.
      Our model of cdk5/p35 down-regulation of the MAP kinase signaling cascade is shown in Fig. 6 on left side of the diagram. The cdk5 cross-talk inhibition of the cascade is targeted at Raf-activated MEK1, an event occurring shortly after receptor activation. We suggest that transient increases in activated MEK1 are modulated by cdk5 phosphorylation of MEK1 in the Raf-MEK1 complex. Because cdk5 activity depends, in part, on its regulator, p35, the extent of cdk5 inhibition is limited by the availability of p35. This model is consistent with recent data showing the activation of cdk5/p35 by ERK in NGF-stimulated PC12 cells (
      • Harada T.
      • Morooka T.
      • Ogawa S.
      • Nishida E.
      ) (NGF to ERK1/2 in Fig. 6). It has been well established that NGF stimulates the MAP kinase cascade with the peak of ERK and MEK1 activity attained rapidly, within 10–20 min (see Fig. 1C) (
      • Harada T.
      • Morooka T.
      • Ogawa S.
      • Nishida E.
      ). They have also shown that ERK activation induces a transcription factor, EGR-1, that initiates p35 transcription and activation of cdk5 within 1–2 h (Fig.1) (
      • Harada T.
      • Morooka T.
      • Ogawa S.
      • Nishida E.
      ). What is most striking is that the increasing cdk5/p35 activity correlates directly with the subsequent decline in ERK and MEK1 activation as if cdk5/p35 is acting as a feedback regulator or switch to shut down the signaling cascade by phosphorylating and inactivating the Raf-MEK1 complex. It is significant that the data in Fig.1B (
      • Harada T.
      • Morooka T.
      • Ogawa S.
      • Nishida E.
      ) showing an early decrease in activated MEK1 preceding the decline in ERK activation are consistent with our model.
      Figure thumbnail gr6
      Figure 6A model of cdk5/p35 feedback regulation of the MAP kinase cascade in PC12 cells based on our data and the data reported in Ref.
      • Harada T.
      • Morooka T.
      • Ogawa S.
      • Nishida E.
      on the transcription of p35 in PC12 cells by NGF-induced ERK activation of the transcription factor Egr1. The cdk5/p35 feedback inhibitory loop is shown on the left sideof the figure in bolder arrows, targeted at activated MEK-1. Inhibition is limited by the availability of cdk5/p35, which is in turn dependent on levels of p35. As seen in Fig. B (
      • Harada T.
      • Morooka T.
      • Ogawa S.
      • Nishida E.
      ), within 5–10 min, NGF induces a rapid activation of the ERK 1/2 pathway, which induces the active transcription factor EGR-1 followed by transcription and up-regulation of p35. This persists for about an hour until phospho-ERK 1/2 and phospho-MEK1 begin to decline. The timing of this decline (1–3 h) coincides with increased expression of p35 and cdk5 activity. We suggest that cdk5/p35 phosphorylation and inhibition of MEK1 activity is a feedback switch responsible for down-regulating the MAP kinase cascade.

      Acknowledgments

      We are thankful to Drs. R. W. Albers and H. Gainer for critical discussions and suggestions with the manuscript. DNA sequencing support by Jim Neagle (NINDS DNA sequencing facility at the National Institutes of Health) is appreciated. We would also like to thank Drs. T. Oshima and T. Tanaka for providing p35(−/−) mice when they were at NIDCR, NIH, Bethesda, MD.

      REFERENCES

        • Morgan D.O.
        Annu. Rev. Cell Dev. Biol. 1997; 13: 261-291
        • Guidato S.
        • McLoughlin D.M.
        • Grierson A.J.
        • Miller C.C.
        J. Neurochem. 1998; 70: 335-340
        • Xiong Y.
        • Zhang H.
        • Beach D.
        Cell. 1992; 71: 505-514
        • Lew J.
        • Huang Q.Q.
        • Qi Z.
        • Winkfein R.J.
        • Aebersold R.
        • Hunt T.
        • Wang J.H.
        Nature. 1994; 371: 423-426
        • Tsai L.H.
        • Delalle I.
        • Caviness Jr., V.S.
        • Chae T.
        • Harlow E.
        Nature. 1994; 371: 419-423
        • Hirooka K.
        • Tomizawa K.
        • Matsui H.
        • Tokuda M.
        • Itano T.
        • Hasegawa E.
        • Wang J.H.
        • Hatase O.
        J. Neurochem. 1996; 67: 2478-2483
        • Philpott A.
        • Porro E.B.
        • Kirschner M.W.
        • Tsai L.H.
        Genes Dev. 1997; 11: 1409-1421
        • Honjyo Y.
        • Kawamoto Y.
        • Nakamura S.
        • Nakano S.
        • Akiguchi I.
        Neuroreport. 1999; 10: 3375-3379
        • Humbert S.
        • Dhavan R.
        • Tsai L.
        J. Cell Sci. 2000; 113: 975-983
        • Wu D.C.
        • Yu Y.P.
        • Lee N.T.
        • Yu A.C.
        • Wang J.H.
        • Han Y.F.
        Neurochem. Res. 2000; 25: 923-929
        • Zheng M.
        • Leung C.L.
        • Liem R.K.
        J. Neurobiol. 1998; 35: 141-159
        • Nikolic M.
        • Dudek H.
        • Kwon Y.T.
        • Ramos Y.F.
        • Tsai L.H.
        Genes Dev. 1996; 10: 816-825
        • Sharma M.
        • Sharma P.
        • Pant H.C.
        J. Neurochem. 1999; 73: 79-86
        • Chae T.
        • Kwon Y.T.
        • Bronson R.
        • Dikkes P.
        • Li E.
        • Tsai L.H.
        Neuron. 1997; 18: 29-42
        • Ohshima T.
        • Gilmore E.C.
        • Longenecker G.
        • Jacobowitz D.M.
        • Brady R.O.
        • Herrup K.
        • Kulkarni A.B.
        J. Neurosci. 1999; 19: 6017-6026
        • Ohshima T.
        • Ward J.M.
        • Huh C.G.
        • Longenecker G.
        • Veeranna
        • Pant H.C.
        • Brady R.O.
        • Martin L.J.
        • Kulkarni A.B.
        Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 11173-11178
        • Bibb J.A.
        • Snyder G.L.
        • Nishi A.
        • Yan Z.
        • Meijer L.
        • Fienberg A.A.
        • Tsai L.H.
        • Kwon Y.T.
        • Girault J.A.
        • Czernik A.J.
        • Huganir R.L.
        • Hemmings Jr., H.C.
        • Nairn A.C.
        • Greengard P.
        Nature. 1999; 402: 669-671
        • Fletcher A.I.
        • Shuang R.
        • Giovannucci D.R.
        • Zhang L.
        • Bittner M.A.
        • Stuenkel E.L.
        J. Biol. Chem. 1999; 274: 4027-4035
        • Matsubara M.
        • Kusubata M.
        • Ishiguro K.
        • Uchida T.
        • Titani K.
        • Taniguchi H.
        J. Biol. Chem. 1996; 271: 21108-21113
        • Rosales J.L.
        • Nodwell M.J.
        • Johnston R.N.
        • Lee K.Y.
        J. Cell. Biochem. 2000; 78: 151-159
        • Shuang R.
        • Zhang L.
        • Fletcher A.
        • Groblewski G.E.
        • Pevsner J.
        • Stuenkel E.L.
        J. Biol. Chem. 1998; 273: 4957-4966
        • Fu A.K.
        • Fu W.Y.
        • Cheung J.
        • Tsim K.W.
        • Ip F.C.
        • Wang J.H.
        • Ip N.Y.
        Nat. Neurosci. 2001; 4: 374-381
        • Paudel H.K.
        • Lew J.
        • Ali Z.
        • Wang J.H.
        J. Biol. Chem. 1993; 268: 23512-23518
        • Shetty K.T.
        • Link W.T.
        • Pant H.C.
        Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 6844-6848
        • Grant P.
        • Sharma P.
        • Pant H.C.
        Eur. J. Biochem. 2001; 268: 1534-1546
        • Sasaki S.
        • Shionoya A.
        • Ishida M.
        • Gambello M.J.
        • Yingling J.
        • Wynshaw-Boris A.
        • Hirotsune S.
        Neuron. 2000; 28: 681-696
        • Bibb J.A.
        • Nishi A.
        • O'Callaghan J.P.
        • Ule J.
        • Lan M.
        • Snyder G.L.
        • Horiuchi A.
        • Saito T.
        • Hisanaga S.
        • Czernik A.J.
        • Nairn A.C.
        • Greengard P.
        J. Biol. Chem. 2001; 276: 14490-14497
        • Nikolic M.
        • Chou M.M.
        • Lu W.
        • Mayer B.J.
        • Tsai L.H.
        Nature. 1998; 395: 194-198
        • Sternberger L.A.
        • Sternberger N.H.
        Proc. Natl. Acad. Sci. U. S. A. 1983; 80: 6126-6130
        • Veeranna
        • Amin N.D.
        • Ahn N.G.
        • Jaffe H.
        • Winters C.A.
        • Grant P.
        • Pant H.C.
        J. Neurosci. 1998; 18: 4008-4021
        • Pearson G.
        • Robinson F.
        • Beers Gibson T.
        • Xu B.
        • Karandikar M.
        • Berman K.
        • Cobb M.H.
        Endocr. Rev. 2001; 22: 153-183
        • Schaeffer H.J.
        • Weber M.J.
        Mol. Cell. Biol. 1999; 19: 2435-2444
        • Cowley S.
        • Paterson H.
        • Kemp P.
        • Marshall C.J.
        Cell. 1994; 77: 841-852
        • Whitmarsh A.J.
        • Davis R.J.
        Trends Biochem. Sci. 1998; 23: 481-485
        • Mansour S.J.
        • Matten W.T.
        • Hermann A.S.
        • Candia J.M.
        • Rong S.
        • Fukasawa K.
        • Vande Woude G.F.
        • Ahn N.G.
        Science. 1994; 265: 966-970
        • Ohshima T.
        • Ogawa M.
        • Veeranna
        • Hirasawa M.
        • Longenecker G.
        • Ishiguro K.
        • Pant H.C.
        • Brady R.O.
        • Kulkarni A.B.
        • Mikoshiba K.
        Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 2764-2769
        • Alessandrini A.
        • Brott B.K.
        • Erikson R.L.
        Cell Growth Differ. 1997; 8: 505-511
        • Yan G.Z.
        • Ziff E.B.
        J. Neurosci. 1995; 15: 6200-6212
        • Patrick G.N.
        • Zukerberg L.
        • Nikolic M.
        • de la Monte S.
        • Dikkes P.
        • Tsai L.H.
        Nature. 1999; 402: 615-622
        • Gardner A.M.
        • Vaillancourt R.R.
        • Lange-Carter C.A.
        • Johnson G.L.
        Mol. Biol. Cell. 1994; 5: 193-201
        • Rossomando A.J.
        • Dent P.
        • Sturgill T.W.
        • Marshak D.R.
        Mol. Cell. Biol. 1994; 14: 1594-1602
        • Hunter T.
        Cell. 2000; 100: 113-127
        • Hucho F.
        • Buchner K.
        Naturwissenschaften. 1997; 84: 281-290
        • Xia Z.
        • Dickens M.
        • Raingeaud J.
        • Davis R.J.
        • Greenberg M.E.
        Science. 1995; 270: 1326-1331
        • Frost J.A.
        • Steen H.
        • Shapiro P.
        • Lewis T.
        • Ahn N.
        • Shaw P.E.
        • Cobb M.H.
        EMBO J. 1997; 16: 6426-6438
        • Widmann C.
        • Gibson S.
        • Jarpe M.B.
        • Johnson G.L.
        Physiol. Rev. 1999; 79: 143-180
        • Dang A.
        • Frost J.A.
        • Cobb M.H.
        J. Biol. Chem. 1998; 273: 19909-19913
        • Catling A.D.
        • Schaeffer H.J.
        • Reuter C.W.
        • Reddy G.R.
        • Weber M.J.
        Mol. Cell. Biol. 1995; 15: 5214-5225
        • Harada T.
        • Morooka T.
        • Ogawa S.
        • Nishida E.
        Nat. Cell Biol. 2001; 3: 453-459