Advertisement

Identification of the Atypical MAPK Erk3 as a Novel Substrate for p21-activated Kinase (Pak) Activity*

  • Alina De La Mota-Peynado
    Affiliations
    From the Division of Biology and Molecular, Cellular, and Developmental Biology Program, Kansas State University, Manhattan, Kansas 66506 and
    Search for articles by this author
  • Jonathan Chernoff
    Affiliations
    the Cancer Biology Program, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111
    Search for articles by this author
  • Alexander Beeser
    Correspondence
    To whom correspondence should be addressed: Division of Biology, Kansas State University, 412-A Ackert Hall, Manhattan, KS 66506. Fax: 785-532-3616;
    Affiliations
    From the Division of Biology and Molecular, Cellular, and Developmental Biology Program, Kansas State University, Manhattan, Kansas 66506 and
    Search for articles by this author
  • Author Footnotes
    * This work was supported, in whole or in part, by National Institutes of Health Grant R01 CA58836 (to J. C.). This work was also supported by Terry C. Johnson Center for Basic Cancer Research (to A. B.).
    The on-line version of this article (available at http://www.jbc.org) contains supplemental Fig. 1.
Open AccessPublished:February 11, 2011DOI:https://doi.org/10.1074/jbc.M110.181743
      The class I p21-activated kinases (Pak1–3) regulate many essential biological processes, including cytoskeletal rearrangement, cell cycle progression, apoptosis, and cellular transformation. Although many Pak substrates, including elements of MAPK signaling cascades, have been identified, it is likely that additional substrates remain to be discovered. Identification of such substrates, and determination of the consequences of their phosphorylation, is essential for a better understanding of class I Pak activity. To identify novel class I Pak substrates, we used recombinant Pak2 to screen high density protein microarrays. This approach identified the atypical MAPK Erk3 as a potential Pak2 substrate. Solution-based in vitro kinase assays using recombinant Erk3 confirmed the protein microarray results, and phospho-specific antisera identified serine 189, within the Erk3 activation loop, as a site directly phosphorylated by Pak2 in vitro. Erk3 protein is known to shuttle between the cytoplasm and the nucleus, and we showed that selective inhibition of class I Pak kinase activity in cells promoted increased nuclear accumulation of Erk3. Pak inhibition in cells additionally reduced the extent of Ser189 phosphorylation and inhibited the formation of Erk3-Prak complexes. Collectively, our results identify the Erk3 protein as a novel class I Pak substrate and further suggest a role for Pak kinase activity in atypical MAPK signaling.

      Introduction

      The class I p21-activated kinases (Pak1–3) are established effectors of the small GTPases Rac1 and Cdc42 (
      • Cotteret S.
      • Chernoff J.
      ). Although initially discovered as regulators of the actin cytoskeleton, they have subsequently been implicated in a variety of different signaling pathways, including those that control proliferation, apoptosis, and transformation (
      • Kumar R.
      • Gururaj A.E.
      • Barnes C.J.
      ,
      • Szczepanowska J.
      ). Not surprisingly, misregulated Pak
      The abbreviations used are: Pak, p21-activated kinase; Prak, p38-regulated/activated kinase; KSI, kinase substrate identification; mRFP, monomeric red fluorescent protein; PID, protein inhibitory domain; MBP, myelin basic protein.
      kinase activity is associated with a variety of different pathological conditions (
      • Allen J.D.
      • Jaffer Z.M.
      • Park S.J.
      • Burgin S.
      • Hofmann C.
      • Sells M.A.
      • Chen S.
      • Derr-Yellin E.
      • Michels E.G.
      • McDaniel A.
      • Bessler W.K.
      • Ingram D.A.
      • Atkinson S.J.
      • Travers J.B.
      • Chernoff J.
      • Clapp D.W.
      ,
      • Hayashi M.L.
      • Rao B.S.
      • Seo J.S.
      • Choi H.S.
      • Dolan B.M.
      • Choi S.Y.
      • Chattarji S.
      • Tonegawa S.
      ,
      • Nguyen D.G.
      • Wolff K.C.
      • Yin H.
      • Caldwell J.S.
      • Kuhen K.L.
      ). As Pak kinase activity contributes to several different signaling pathways, a better understanding of the specific role for Paks in any biological response is greatly aided by the identification of the protein(s) they phosphorylate and the consequences of these phosphorylations. Although the current list of Pak kinases substrates is quite large (
      • Szczepanowska J.
      ,
      • Dummler B.
      • Ohshiro K.
      • Kumar R.
      • Field J.
      ), we are particularly interested in how Pak kinases influence MAPK signaling cascades.
      MAPK signaling pathways are among the most evolutionarily conserved signal transduction pathways and are critical for many biological responses (
      • Krishna M.
      • Narang H.
      ). Various cellular stimuli lead to the enzymatic activation of a family of serine/threonine kinases known as MAP3Ks. Activated MAP3Ks phosphorylate and activate specific MAP2Ks. MAP2Ks are dual specificity kinases that phosphorylate threonine and tyrosine residues within the activation loop of the MAPKs (the conserved TXY motif) to activate them; phosphorylation of both activation loop residues is required for MAPK activation. Activated MAPKs then transphosphorylate a variety of different proteins, including structural proteins, enzymes, and transcription factors that ultimately drive the appropriate cellular response to the initial signal (
      • Chen Z.
      • Gibson T.B.
      • Robinson F.
      • Silvestro L.
      • Pearson G.
      • Xu B.
      • Wright A.
      • Vanderbilt C.
      • Cobb M.H.
      ). For some signaling pathways, the MAPK substrates are themselves kinases and are defined as MAPK-activated kinases (MAPKAKs or MKs) extending the characteristic three-tiered architecture (
      • Roux P.P.
      • Blenis J.
      ).
      Although the central roles of the MAP3K, MAP2K, and MAPK isoforms to signaling pathways are incontrovertible, the intensity and duration of MAPK signaling are additionally regulated by the activity of several noncanonical proteins, including Paks. Pak kinase activity affects Erk1/2 signaling at two distinct points as follows: Pak phosphorylation of the MAP3K c-Raf on serine 338 cooperates with Src-dependent phosphorylation at tyrosine 341 for maximal c-Raf kinase activity (
      • Mason C.S.
      • Springer C.J.
      • Cooper R.G.
      • Superti-Furga G.
      • Marshall C.J.
      • Marais R.
      ), and Pak phosphorylation on serine 298 of Mek1 facilitates the interaction of Mek1 with Erk2 (
      • Eblen S.T.
      • Slack J.K.
      • Weber M.J.
      • Catling A.D.
      ).
      In addition to the classical MAPK families (ERK, p38, and JNK/SAPK), many cells also express atypical MAPK isoforms characterized by the Erk3, Erk4, Nlk (nemo-like kinase), and Erk7/8 proteins (
      • Coulombe P.
      • Meloche S.
      ). Erk3 and -4 are most closely related to Erk1, sharing 45 and 42% amino acid identity, respectively, to the catalytic domain of Erk1. Despite the extent of this similarity, Erk3 cannot phosphorylate validated Erk1/2 substrates in vitro (
      • Cheng M.
      • Boulton T.G.
      • Cobb M.H.
      ). A characteristic that distinguishes the atypical MAPKs from the classical MAPK isoforms is the lack of the characteristic “TXY” motif within their activation loops. In Erk3 and Erk4, the corresponding sequence is SEG, and for Nlk it is TQE (
      • Coulombe P.
      • Meloche S.
      ). Although Erk7 does contain a TXY sequence that is phosphorylated in vivo, this phosphorylation is not catalyzed by any known MAP2K but results from autophosphorylation (
      • Abe M.K.
      • Kuo W.L.
      • Hershenson M.B.
      • Rosner M.R.
      ). As such, Erk7 is also generally considered a member of the atypical MAPK family (
      • Coulombe P.
      • Meloche S.
      ).
      Unlike the classical MAPKs, dual phosphorylation of the Erk3/Erk4 activation loops is not possible. Mek2, but not Mek1, could poorly phosphorylate Erk3 Ser189 in vitro (
      • Robinson M.J.
      • Cheng M.
      • Khokhlatchev A.
      • Ebert D.
      • Ahn N.
      • Guan K.L.
      • Stein B.
      • Goldsmith E.
      • Cobb M.H.
      ). However, a cell fractionation approach to identify the kinases responsible for Erk3 Ser189 phosphorylation suggested that the cellular Ser189 kinase did not co-fractionate with Erk1/2 kinase activity, excluding Mek1/2 from consideration as cellular Erk3 kinases (
      • Cheng M.
      • Zhen E.
      • Robinson M.J.
      • Ebert D.
      • Goldsmith E.
      • Cobb M.H.
      ). Similar approaches excluded protein kinase C (PKC) as the cellular Ser189 kinase despite the ability of PKC to phosphorylate Erk3 in vitro (
      • Cheng M.
      • Zhen E.
      • Robinson M.J.
      • Ebert D.
      • Goldsmith E.
      • Cobb M.H.
      ). The initial observation of Erk3 Ser189 phosphorylation in vivo occurred more than a decade ago (
      • Cheng M.
      • Zhen E.
      • Robinson M.J.
      • Ebert D.
      • Goldsmith E.
      • Cobb M.H.
      ), but the kinase(s) responsible for this phosphorylation remain unidentified.
      Using high density protein microarrays, we identified the Erk3 atypical MAPK as a candidate Pak2 substrate. Pak2 kinase assays using recombinant full-length Erk3 protein in solution confirmed the protein microarray results and suggested that Erk3 is a direct p21-activated kinase substrate in vitro. We further demonstrated that Pak2 targets the Ser189 site, within the activation loop of Erk3, suggesting that p21-activated kinases may contribute to Erk3 activation and furthermore may represent the elusive Erk3 Ser189 kinase (referred to as Kinase X in Ref.
      • Déléris P.
      • Rousseau J.
      • Coulombe P.
      • Rodier G.
      • Tanguay P.L.
      • Meloche S.
      ). A variant of Erk3 lacking this phosphorylation site (S189A) displays an increased nuclear accumulation in fibroblasts, which can be phenocopied with wild type Erk3 by selective inhibition of class I Pak kinase activity. Furthermore, class I kinase inhibition reduced the levels of Erk3 Ser189 phosphorylation in vivo. As Ser189 phosphorylation stabilizes the interaction of Erk3 with its effector Prak, we further demonstrate class I inhibition restricts the formation of Erk3-Prak complexes in cells in a manner dependent on Ser189 phosphorylation. Collectively, our results identified Erk3 as a novel substrate for class I Pak kinase activity and identified Ser189, within the Erk3 activation loop, as a residue phosphorylated by class I kinase activity in vitro and in vivo suggesting a role for class I p21-activated kinases in regulating the subcellular localization and activity of the atypical MAPK Erk3.

      DISCUSSION

      Based on their similarity to the classical Erk1 and Erk2 MAPKs, the atypical MAPKs Erk3 and Erk4 were initially identified more than 15 years ago. The pathways regulated by these kinases, however, remain enigmatic with respect to both the cellular upstream-activating signals (functioning analogously to MAP2Ks) and to a slightly lesser extent in regards to downstream effectors. Models for Erk3/Erk4 function have so far been generally restricted to the following properties: (i) the ability of variants of Erk3/Erk4 to bind to Prak; (ii) the ability of Erk3/Erk4 variants to promote Prak Thr-182 phosphorylation and/or activate the kinase activity of Prak, and (iii) the ability of these variants to modulate their own subcellular localization or the subcellular localization of Prak.
      Using high density protein microarrays and recombinant Erk3 and Pak2 proteins, we found that Erk3 is a Pak2 substrate in vitro. Phospho-specific antibodies against residues within the Erk3 activation loop further indicated that serine 189 is a site phosphorylated by Pak2 in vitro. Additionally selective class I kinase inhibition led to a decrease in the extent of Ser189 phosphorylation in HEK293 cells (Fig. 4B) suggesting that this kinase substrate relationship between Pak and Erk3 is retained in vivo.
      As Erk3 Ser189 phosphorylation is required for the interaction with Prak, we further demonstrated that specific inhibition of class I kinase activity in cells decreases the interaction between His6-Erk3 and endogenous Prak in HEK293 cells (Fig. 5). This effect was specific to the phosphorylation status of Ser189 as Prak interacted poorly with the Erk3S189A variant, whereas Prak interacted strongly with Erk3 phosphomimetic variant (S189D). Importantly, the interaction between Prak and Erk3S189D was insensitive to PID expression indicating that the ability of Pak to stabilize Erk3-Prak complexes derives exclusively from the ability to promote Ser189 phosphorylation.
      Specific inhibition of class I kinase activity in fibroblasts led to an increased nuclear retention of GFP-Erk3, suggesting a role for Paks in regulating the subcellular localization of Erk3. This result was rather surprising as Meloche and co-workers (
      • Julien C.
      • Coulombe P.
      • Meloche S.
      ) had previously reported that nuclear export of Erk3 was independent of both activation loop phosphorylation and kinase activity in NIH3T3 cells, the same cell line used in our studies. The one obvious difference between the two approaches was the way in which Erk3 was detected; Meloche and co-workers (
      • Julien C.
      • Coulombe P.
      • Meloche S.
      ) detected the localization of Myc-tagged Erk3 by indirect immunofluorescence, whereas we used GFP-Erk3. We do not know whether this accounts for the observed differences, but GFP-tagged variants of Erk3 have been used by other groups that display very similar subcellular localization in HEK293 cells as we observe in NIH3T3 cells (
      • Kant S.
      • Schumacher S.
      • Singh M.K.
      • Kispert A.
      • Kotlyarov A.
      • Gaestel M.
      ). We also believe that the similarity in subcellular localization of the GFP-Erk3S189A variant to that seen for GFP-Erk3 in cells co-expressing mRFP-PID (Fig. 6, C and D) supports our conclusion that Ser189 phosphorylation modulates Erk3 subcellular localization. Collectively, our results demonstrate a relationship between the p21-activated kinases and Erk3 and further suggest a role for class I kinase activity in regulating atypical MAPK pathways by promoting Ser189 phosphorylation.
      An important question remains. How does Ser189 phosphorylation promote the interaction of Erk3 with Prak? The residues of Erk3 required for Prak binding map to a region distant from the activation loop characterized by the sequence 328FRIEDE333 (
      • Aberg E.
      • Torgersen K.M.
      • Johansen B.
      • Keyse S.M.
      • Perander M.
      • Seternes O.M.
      ). This interface is distinct from the common docking domain used by classical Erks to bind to both their respective MAP2Ks and their cognate substrates (
      • Bardwell A.J.
      • Frankson E.
      • Bardwell L.
      ). Molecular modeling suggests that Erk4 Ser186 phosphorylation (analogous the Erk3 Ser189) might promote a conformational change that increases the solvent accessibility of isoleucine 330 (within the conserved FRIEDE motif) allowing for efficient interaction with the C terminus of Prak (
      • Aberg E.
      • Torgersen K.M.
      • Johansen B.
      • Keyse S.M.
      • Perander M.
      • Seternes O.M.
      ).
      We believe that the identification of Paks as kinases responsible for Erk3 Ser189 phosphorylation represents an important step forward in our understanding of atypical MAPK signaling cascades. Their identification now affords the ability to modulate atypical MAPK signaling pathways in physiologically relevant settings by specific Pak inhibition. In our opinion, use of the PID affords several important advantages over other experimental approaches, such as pharmacological inhibition or siRNA approaches. These advantages include lack of isoform selectivity, high target specificity, lack of dominant negative effects, and the ability to inhibit endogenous class I activity in trans in tissue culture (
      • Beeser A.
      • Jaffer Z.M.
      • Hofmann C.
      • Chernoff J.
      ) and in transgenic animals (
      • Hayashi M.L.
      • Rao B.S.
      • Seo J.S.
      • Choi H.S.
      • Dolan B.M.
      • Choi S.Y.
      • Chattarji S.
      • Tonegawa S.
      ). Specific mutants of the PID (L107F) that no longer function as Pak inhibitors further provide an important control to evaluate the specific contribution of class I kinase activity.
      Even with the identification of Paks as Ser189 kinases, previous reports of the kinetics of Ser189 phosphorylation raise some important questions. Erk3 Ser189 phosphorylation does not respond to signals that activate the classical MAPKs (
      • Déléris P.
      • Rousseau J.
      • Coulombe P.
      • Rodier G.
      • Tanguay P.L.
      • Meloche S.
      ), and Ser189 phosphorylation persists in HEK293 cells rendered quiescent by serum starvation (
      • Déléris P.
      • Rousseau J.
      • Coulombe P.
      • Rodier G.
      • Tanguay P.L.
      • Meloche S.
      ). Although the signals that activate class I kinase activity differ from cell type to cell type, serum starvation is sufficient to reduce class I kinase activity in most cells as demonstrated by both Pak activation-specific antisera as well as in-gel kinase assays (
      • Deacon S.W.
      • Beeser A.
      • Fukui J.A.
      • Rennefahrt U.E.
      • Myers C.
      • Chernoff J.
      • Peterson J.R.
      ,
      • Beeser A.
      • Jaffer Z.M.
      • Hofmann C.
      • Chernoff J.
      ). It is important to note that although we commonly think of Pak kinase activity being “off” in quiescent cells, this activity is still readily detected in serum-starved cells, albeit at much lower levels than in cells stimulated with growth factors. As the relative levels of the inherently unstable Erk3 and class I Paks are unknown, basal class I kinase activity may still be sufficient to promote Ser189 phosphorylation in quiescent cells.
      Alternatively, Ser189 could be targeted by kinases other than Pak. Although entirely speculative, other candidate Ser189 kinases include the class II Paks (Pak4–6), which are constitutively active kinases with similar substrate specificity as the class I Paks (
      • Rennefahrt U.E.
      • Deacon S.W.
      • Parker S.A.
      • Devarajan K.
      • Beeser A.
      • Chernoff J.
      • Knapp S.
      • Turk B.E.
      • Peterson J.R.
      ) and/or the related mammalian sterile 20-like kinase (Mst1) as Erk3 was identified in a similar KSI protein microarray screen using recombinant Mst1.
      S. Jalan and J. Chernoff, unpublished observations.
      It is important to note that although these proteins are all members of the Ste20 superfamily, neither the class II Paks nor Mst1 is inhibited by expression of the PID. Irrespective of whether the ability to phosphorylate is unique to class I Paks, our results strongly suggest that class I kinase activity is critical for the formation of the Erk3-Prak complex in cells, which is one of the best metrics for Erk3 signaling, and suggest an important role for class I kinase activity in regulating atypical MAPK signaling.

      Acknowledgments

      We thank Dr. Sylvain Meloche and co-workers (University of Montreal) for providing the human Erk3 expression construct and for communicating unpublished results.

      REFERENCES

        • Cotteret S.
        • Chernoff J.
        Genome Biol. 2002; 3 (REVIEWS0002)
        • Kumar R.
        • Gururaj A.E.
        • Barnes C.J.
        Nat. Rev. Cancer. 2006; 6: 459-471
        • Szczepanowska J.
        Acta Biochim. Pol. 2009; 56: 225-234
        • Allen J.D.
        • Jaffer Z.M.
        • Park S.J.
        • Burgin S.
        • Hofmann C.
        • Sells M.A.
        • Chen S.
        • Derr-Yellin E.
        • Michels E.G.
        • McDaniel A.
        • Bessler W.K.
        • Ingram D.A.
        • Atkinson S.J.
        • Travers J.B.
        • Chernoff J.
        • Clapp D.W.
        Blood. 2009; 113: 2695-2705
        • Hayashi M.L.
        • Rao B.S.
        • Seo J.S.
        • Choi H.S.
        • Dolan B.M.
        • Choi S.Y.
        • Chattarji S.
        • Tonegawa S.
        Proc. Natl. Acad. Sci. U.S.A. 2007; 104: 11489-11494
        • Nguyen D.G.
        • Wolff K.C.
        • Yin H.
        • Caldwell J.S.
        • Kuhen K.L.
        J. Virol. 2006; 80: 130-137
        • Dummler B.
        • Ohshiro K.
        • Kumar R.
        • Field J.
        Cancer Metastasis Rev. 2009; 28: 51-63
        • Krishna M.
        • Narang H.
        Cell. Mol. Life Sci. 2008; 65: 3525-3544
        • Chen Z.
        • Gibson T.B.
        • Robinson F.
        • Silvestro L.
        • Pearson G.
        • Xu B.
        • Wright A.
        • Vanderbilt C.
        • Cobb M.H.
        Chem. Rev. 2001; 101: 2449-2476
        • Roux P.P.
        • Blenis J.
        Microbiol. Mol. Biol. Rev. 2004; 68: 320-344
        • Mason C.S.
        • Springer C.J.
        • Cooper R.G.
        • Superti-Furga G.
        • Marshall C.J.
        • Marais R.
        EMBO J. 1999; 18: 2137-2148
        • Eblen S.T.
        • Slack J.K.
        • Weber M.J.
        • Catling A.D.
        Mol. Cell. Biol. 2002; 22: 6023-6033
        • Coulombe P.
        • Meloche S.
        Biochim. Biophys. Acta. 2007; 1773: 1376-1387
        • Cheng M.
        • Boulton T.G.
        • Cobb M.H.
        J. Biol. Chem. 1996; 271: 8951-8958
        • Abe M.K.
        • Kuo W.L.
        • Hershenson M.B.
        • Rosner M.R.
        Mol. Cell. Biol. 1999; 19: 1301-1312
        • Robinson M.J.
        • Cheng M.
        • Khokhlatchev A.
        • Ebert D.
        • Ahn N.
        • Guan K.L.
        • Stein B.
        • Goldsmith E.
        • Cobb M.H.
        J. Biol. Chem. 1996; 271: 29734-29739
        • Cheng M.
        • Zhen E.
        • Robinson M.J.
        • Ebert D.
        • Goldsmith E.
        • Cobb M.H.
        J. Biol. Chem. 1996; 271: 12057-12062
        • Déléris P.
        • Rousseau J.
        • Coulombe P.
        • Rodier G.
        • Tanguay P.L.
        • Meloche S.
        J. Cell. Physiol. 2008; 217: 778-788
        • Deacon S.W.
        • Beeser A.
        • Fukui J.A.
        • Rennefahrt U.E.
        • Myers C.
        • Chernoff J.
        • Peterson J.R.
        Chem. Biol. 2008; 15: 322-331
        • Hofmann C.
        • Shepelev M.
        • Chernoff J.
        J. Cell. Sci. 2004; 117: 4343-4354
        • Schnack C.
        • Hengerer B.
        • Gillardon F.
        Proteomics. 2008; 8: 1980-1986
        • Feilner T.
        • Hultschig C.
        • Lee J.
        • Meyer S.
        • Immink R.G.
        • Koenig A.
        • Possling A.
        • Seitz H.
        • Beveridge A.
        • Scheel D.
        • Cahill D.J.
        • Lehrach H.
        • Kreutzberger J.
        • Kersten B.
        Mol. Cell. Proteomics. 2005; 4: 1558-1568
        • Xie X.
        • Chang S.W.
        • Tatsumoto T.
        • Chan A.M.
        • Miki T.
        Cell. Signal. 2005; 17: 461-471
        • Seoh M.L.
        • Ng C.H.
        • Yong J.
        • Lim L.
        • Leung T.
        FEBS Lett. 2003; 539: 131-137
        • Seternes O.M.
        • Mikalsen T.
        • Johansen B.
        • Michaelsen E.
        • Armstrong C.G.
        • Morrice N.A.
        • Turgeon B.
        • Meloche S.
        • Moens U.
        • Keyse S.M.
        EMBO J. 2004; 23: 4780-4791
        • Perander M.
        • Aberg E.
        • Johansen B.
        • Dreyer B.
        • Guldvik I.J.
        • Outzen H.
        • Keyse S.M.
        • Seternes O.M.
        Biochem. J. 2008; 411: 613-622
        • Jakobi R.
        • Huang Z.
        • Walter B.N.
        • Tuazon P.T.
        • Traugh J.A.
        Eur. J. Biochem. 2000; 267: 4414-4421
        • Rennefahrt U.E.
        • Deacon S.W.
        • Parker S.A.
        • Devarajan K.
        • Beeser A.
        • Chernoff J.
        • Knapp S.
        • Turk B.E.
        • Peterson J.R.
        J. Biol. Chem. 2007; 282: 15667-15678
        • Tanguay P.L.
        • Rodier G.
        • Meloche S.
        Biochem. J. 2010; 428: 103-111
        • Julien C.
        • Coulombe P.
        • Meloche S.
        J. Biol. Chem. 2003; 278: 42615-42624
        • Beeser A.
        • Jaffer Z.M.
        • Hofmann C.
        • Chernoff J.
        J. Biol. Chem. 2005; 280: 36609-36615
        • Kant S.
        • Schumacher S.
        • Singh M.K.
        • Kispert A.
        • Kotlyarov A.
        • Gaestel M.
        J. Biol. Chem. 2006; 281: 35511-35519
        • Aberg E.
        • Torgersen K.M.
        • Johansen B.
        • Keyse S.M.
        • Perander M.
        • Seternes O.M.
        J. Biol. Chem. 2009; 284: 19392-19401
        • Bardwell A.J.
        • Frankson E.
        • Bardwell L.
        J. Biol. Chem. 2009; 284: 13165-13173