Regulation of Nuclear Localization during Signaling

Abstract

nuclear localization sequence nuclear export sequence nuclear factor of activated T cells mitogen-activated protein kinase glycogen synthase kinase casein kinase I IκB kinase small ubiquitin-like modifier extracellular signal-regulated kinase MAPK kinase stress-activated protein kinase c-Jun/AP1 kinase MAPK phosphatase-3 protein-tyrosine phosphatase kinase interaction motif Signaling pathways consist of a chain of biochemical events that form the intracellular equivalent of a fire bucket brigade. An initial change or chemical signal outside the cell is sensed at the cell surface by a receptor, which then transduces the signal to the cytosol. Over the last few years, investigations have focused increasingly on the spatiotemporal aspects of signaling, and it has become clear that the localization of key signaling components is highly regulated during signal transduction. Most signal transduction pathways cause specific changes in gene expression. Thus, an extracellular signal must be transduced across the plasma membrane and subsequently across the nuclear envelope to propagate the signal from the cytosol to the nucleus. Many signaling responses, therefore, rapidly effect the nuclear localization of transcription factors or alternatively of kinases that, once translocated, phosphorylate and activate transcription factors in the nucleus. Proteins travel into and out of the nucleus exclusively through the nuclear pore complex, an elaborate constellation of at least 30 distinct components embedded in the nuclear envelope (1Rout M.P. Aitchison J.D. J. Biol. Chem. 2001; 276: 16593-16596Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar). The mechanism by which proteins travel through the nuclear pore is the subject of much investigation (1Rout M.P. Aitchison J.D. J. Biol. Chem. 2001; 276: 16593-16596Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar); however in the past few years several aspects of nuclear transport have been clarified. Small molecules (less than 50,000 daltons) diffuse freely in and out of the nucleus through nuclear pores. To gain access to the nucleus, larger proteins require a nuclear localization sequence (NLS),1 a series of basic residues that mediates binding to a protein called either a karyopherin or importin. The karyopherin-cargo complex is actively transported through the nuclear pores. Similarly, large proteins containing a nuclear export sequence (NES) bind to a different karyopherin protein, also called an exportin, and exit the nucleus. Other regulatory proteins, including the small GTPase RAN1, are required for trafficking in and out of the nucleus and ensure the directionality of transport (1Rout M.P. Aitchison J.D. J. Biol. Chem. 2001; 276: 16593-16596Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar). Signaling can induce the rapid redistribution of a protein from the cytosol to the nucleus by a number of different mechanisms. Two major factors that affect protein partitioning between the cytosol and nucleus are interactions with the nuclear transport machinery and interactions with anchor proteins that reside stably in the nucleus or cytosol. In the first case, signaling may directly regulate the association of a protein with nuclear import and/or export factors. Thus, a protein, the distribution of which is cytosolic and exhibits a low rate of nuclear import relative to its rate of nuclear export, will rapidly translocate to the nucleus in response to a regulated increase in its association with an importin, a regulated decrease in its association with an exportin, or a combination of these two effects. YAP1 and nuclear factor of activated T cells (NFAT) exhibit this type of regulation. Alternatively a protein may contain an NLS but fail to enter the nucleus at a significant rate because of a high affinity interaction with a cytosolic anchor protein that does not enter the nucleus and may even occlude the NLS of its binding partner. Similarly, a protein may reside stably in the nucleus even though it contains an NES if it interacts tightly with a nuclear protein. In either case, the intracellular distribution of such a protein can be regulated by signal-mediated changes in its affinity for or the availability of its anchor(s). NF-κB and MAPK illustrate this type of control. Yap1p is a transcription factor found in the budding yeast,Saccharomyces cerevisiae, that allows these cells to respond to oxidative stress. Yap1p is related to c-Fos and c-Jun (AP1) in mammalian cells and the homologous transcription factors, PAP1 in fission yeast (Schizosaccharomyes pombe) and CAP1 inCandida albicans. In S. cerevisiae, Yap1p is found primarily in the cytosol, but after exposure of cells to a number of different oxidizing agents, such as diamide and diethylmaleate, the protein rapidly accumulates in the nucleus (2Kuge S. Jones N. Nomoto A. EMBO J. 1997; 16: 1710-1720Crossref PubMed Scopus (347) Google Scholar). Yap1p localization is regulated primarily at the level of nuclear export. In the absence of the exportin Xpo1p, the S. cerevisiae homologue of the mammalian Crm1 exportin, the protein is constitutively nuclear and transcribes its target genes in the absence of oxidizing agents (3Kuge S. Toda T. Iizuka N. Nomoto A. Genes Cells. 1998; 3: 521-532Crossref PubMed Scopus (136) Google Scholar, 4Yan C. Lee L.H. Davis L.I. EMBO J. 1998; 17: 7416-7429Crossref PubMed Scopus (205) Google Scholar). The NES of Yap1p is localized to a highly conserved carboxyl-terminal domain, termed the cysteine-rich domain, that is both necessary and sufficient for oxidation-induced nuclear translocation (3Kuge S. Toda T. Iizuka N. Nomoto A. Genes Cells. 1998; 3: 521-532Crossref PubMed Scopus (136) Google Scholar, 4Yan C. Lee L.H. Davis L.I. EMBO J. 1998; 17: 7416-7429Crossref PubMed Scopus (205) Google Scholar). Yap1p binds to Xpo1p in vivo and in vitro, and oxidation-induced relocalization of Yap1p seems to occur by direct inactivation of its NES. A group of conserved cysteine residues adjacent to the NES is required for regulated localization of Yap1p, and oxidation of these cysteines disrupts binding of Yap1p to Xpo1pin vitro (4Yan C. Lee L.H. Davis L.I. EMBO J. 1998; 17: 7416-7429Crossref PubMed Scopus (205) Google Scholar). In S. pombe and C. albicans, the respective transcription factors PAP1 and CAP1 are similarly regulated, although PAP1 also requires the activity of the Spc1/Sty1 MAPK, an upstream component of the oxidative stress response, for its nuclear localization (5Kudo N. Taoka H. Toda T. Yoshida M. Horinouchi S. J. Biol. Chem. 1999; 274: 15151-15158Abstract Full Text Full Text PDF PubMed Scopus (114) Google Scholar, 6Toone W.M. Kuge S. Samuels M. Morgan B.A. Toda T. Jones N. Genes Dev. 1998; 12: 1453-1463Crossref PubMed Scopus (265) Google Scholar, 7Zhang X. Micheli M.D. Coleman S.T. Sanglard D. Moye-Rowley W.S. Mol. Microbiol. 2000; 36: 618-629Crossref PubMed Scopus (123) Google Scholar). Together the YAP1 family of transcription factors illustrates an elegant and direct method by which a change in environmental conditions leads to nuclear localization of a critical transcription factor through a regulated decrease in its rate of nuclear export. The NFAT family of mammalian transcription factors is regulated by Ca2+ and plays a key role in activating gene expression in T lymphocytes during antigenic stimulation (reviewed in Ref. 8Rao A. Luo C. Hogan P.G. Annu. Rev. Immunol. 1997; 15: 707-747Crossref PubMed Scopus (2223) Google Scholar). More recently, NFAT has also been shown to play an important role in the function and development of neuronal, cardiac, and other cells (9Graef I.A. Mermelstein P.G. Stankunas K. Neilson J.R. Deisseroth K. Tsien R.W. Crabtree G.R. Nature. 1999; 401: 703-708Crossref PubMed Scopus (456) Google Scholar, 10de la Pompa J. Timmerman L. Takimoto H. Yoshida H. Elia A. Samper E. Potter J. Wakeham A. Marengere L. Langille B. Crabtree G. Mak T. Nature. 1998; 392: 182-186Crossref PubMed Scopus (547) Google Scholar, 11Molkentin J. Lu J. Antos C. Markham B. Richardson J. Robbins J. Grant S. Olson E. Cell. 1998; 93: 215-228Abstract Full Text Full Text PDF PubMed Scopus (2205) Google Scholar). The activity of NFAT transcription factors is controlled primarily through its localization. In unstimulated cells, these proteins reside in the cytosol. However, when, in response to the appropriate stimulation, cellular Ca2+ levels rise, the Ca2+/calmodulin-activated phosphatase, calcineurin, dephosphorylates NFAT and causes its rapid translocation to the nucleus (12Flanagan W. Corthesy B. Bram R. Crabtree G. Nature. 1991; 352: 803-807Crossref PubMed Scopus (954) Google Scholar, 13Jain J. McCaffrey P.G. Miner Z. Kerppola T.K. Lambert J.N. Verdine G.L. Curran T. Rao A. Nature. 1993; 365: 352-355Crossref PubMed Scopus (679) Google Scholar, 14Shaw K. Ho A. Raghavan A. Kim J. Jian J. Park J. Sharma S. Rao A. Hogan P. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 11205-11209Crossref PubMed Scopus (318) Google Scholar, 15Beals C.R. Clipstone N.A. Ho S.N. Crabtree G.R. Genes Dev. 1997; 11: 824-834Crossref PubMed Scopus (342) Google Scholar). In T-cells, NFAT acts together in the nucleus with AP-1 transcription factors to turn on gene expression (16Jain J. McCaffrey P.G. Valge-Archer V.E. Rao A. Nature. 1992; 356: 801-804Crossref PubMed Scopus (428) Google Scholar). This nuclear localization requires the sustained activation of calcineurin. Hence, as Ca2+ levels drop or calcineurin is inhibited with the immunosuppressive drugs FK506 or cyclosporin A, NFAT is rapidly rephosphorylated and exported from the nucleus (17Timmerman L.A. Clipstone N.A. Ho S.N. Northrop J.P. Crabtree G.R. Nature. 1996; 383: 837-840Crossref PubMed Scopus (473) Google Scholar, 18Loh C. Shaw K.T.Y. Carew J. Viola J.P.B. Lou C. Perrino B.A. Rao A. J. Biol. Chem. 1996; 271: 10884-10891Abstract Full Text Full Text PDF PubMed Scopus (266) Google Scholar, 19Shibasaki F. Price E.R. Milan D. McKeon F. Nature. 1996; 382: 370-373Crossref PubMed Scopus (434) Google Scholar). A number of different kinases have been implicated in the negative regulation of NFATs including GSK3, CK I, MEKK, and p38 MAPK (20Beals C.R. Sheridan C.M. Turck C.W. Gardner P. Crabtree G.R. Science. 1997; 275: 1930-1933Crossref PubMed Scopus (635) Google Scholar, 21Zhu J. Shibasaki F. Price R. Guillemot J.C. Yano T. Dotsch V. Wagner G. Ferrara P. McKeon F. Cell. 1998; 93: 851-861Abstract Full Text Full Text PDF PubMed Scopus (275) Google Scholar, 22Porter C.M. Havens M.A. Clipstone N.A. J. Biol. Chem. 2000; 275: 3543-3551Abstract Full Text Full Text PDF PubMed Scopus (136) Google Scholar, 23Gomez del Arco P. Martinez-Martinez S. Maldonado J.L. Ortega-Perez I. Redondo J.M. J. Biol. Chem. 2000; 275: 13872-13878Abstract Full Text Full Text PDF PubMed Scopus (133) Google Scholar). The phosphorylation sites for these kinases are not conserved between family members, thus the kinase(s) that oppose calcineurin may vary with cell type. There is evidence that calcineurin regulates both the nuclear import and export of NFAT. NFATc (also called NFAT2), contains two NLSs that are thought to be inaccessible when the protein is phosphorylated because of intramolecular interactions with two different conserved regions: the serine-rich region and the SP repeats (15Beals C.R. Clipstone N.A. Ho S.N. Crabtree G.R. Genes Dev. 1997; 11: 824-834Crossref PubMed Scopus (342) Google Scholar). A similar “masking” domain has been defined for another NFAT family member, NFAT4 (21Zhu J. Shibasaki F. Price R. Guillemot J.C. Yano T. Dotsch V. Wagner G. Ferrara P. McKeon F. Cell. 1998; 93: 851-861Abstract Full Text Full Text PDF PubMed Scopus (275) Google Scholar). Dephosphorylation is required for NFAT to undergo nuclear import, but mutant proteins in which potential phosphorylation sites in the masking regions have been replaced with alanines localize to the nucleus constitutively (15Beals C.R. Clipstone N.A. Ho S.N. Crabtree G.R. Genes Dev. 1997; 11: 824-834Crossref PubMed Scopus (342) Google Scholar). Efficient dephosphorylation of NFAT by calcineurin is promoted by the interaction of these two proteins via calcineurin docking sites in NFAT (24Luo C. Shaw K.T. Raghavan A. Aramburu J. Garcia-Cozar F. Perrino B.A. Hogan P.G. Rao A. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 8907-8912Crossref PubMed Scopus (157) Google Scholar, 25Wesselborg S. Fruman D.A. Sagoo J.K. Bierer B.E. Burakoff S.J. J. Biol. Chem. 1995; 271: 1274-1277Abstract Full Text Full Text PDF Scopus (90) Google Scholar, 26Park S. Uesugi M. Verdine G.L. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 7130-7135Crossref PubMed Scopus (84) Google Scholar). Mutant NFATs that lack a calcineurin binding site fail to translocate to the nucleus in response to Ca2+ signaling unless calcineurin is overexpressed (21Zhu J. Shibasaki F. Price R. Guillemot J.C. Yano T. Dotsch V. Wagner G. Ferrara P. McKeon F. Cell. 1998; 93: 851-861Abstract Full Text Full Text PDF PubMed Scopus (275) Google Scholar). Other studies suggest that calcineurin also down-regulates the nuclear export of NFAT. NFAT4 binds to the exportin Crm1 in vitro, and overexpression of this exportin inhibits the nuclear accumulation of NFAT4 in vivo (27Zhu J. McKeon F. Nature. 1999; 398: 256-260Crossref PubMed Scopus (161) Google Scholar). Surprisingly the regions of NFAT4 that are required for export and Crm1 binding overlap with the calcineurin binding site. Thus, Zhu and McKeon (27Zhu J. McKeon F. Nature. 1999; 398: 256-260Crossref PubMed Scopus (161) Google Scholar) have proposed that calcineurin inhibits nuclear export of NFAT4 by dephosphorylating it and also by directly binding to NFAT4 and competing with Crm1. In support of this model, they find that overexpression of catalytically inactive calcineurin is able to inhibit the nuclear export of NFAT4. Thus, NFAT localization is apparently controlled by the accessibility of its NLS and NES to the nuclear transport machinery, and signaling coordinately regulates its rate of nuclear import and export. Calcineurin's role as a key mediator of Ca2+-regulated gene expression has been highly conserved during evolution. In S. cerevisiae, this phosphatase regulates the localization of Crz1p/Tcn1p, a zinc finger-containing transcription factor that is not related to NFAT. In response to increased intracellular Ca2+, calcineurin dephosphorylates Crz1p, causing its rapid relocalization from the cytosol to the nucleus (67Stathopoulos-Gerontides A. Guo J.J. Cyert M.S. Genes Dev. 1999; 13: 798-803Crossref PubMed Scopus (200) Google Scholar). This change is effected through regulation of both Crz1p nuclear import and export. Crz1p gains entry to the nucleus via the importin Nmd5p, and the dephosphorylated form of Crz1p binds preferentially to Nmd5p in vitro. 2M. S. Cyert and R. Polizotto, manuscript in preparation. Nuclear export of Crz1p requires the exportin, Msn5p, and is more active for the phosphorylated form of the protein. 3M. S. Cyert and L. M. Boustany, manuscript in preparation. A family of related transcription factors, NF-κB, play a major role in activating genes required for inflammation and the immune response in mammals (reviewed in Ref. 30Karin M. Ben-Neriah Y. Annu. Rev. Immunol. 2000; 18: 621-663Crossref PubMed Scopus (4085) Google Scholar). In Drosophila, NF-κB homologues function in immunity as well as in dorsal-ventral patterning. NF-κB proteins bind DNA as homo- or heterodimers and also interact with a key regulatory protein, IκB, that serves as a cytosolic anchor for NF-κB. NF-κB binds to ankyrin repeats contained within IκB, and this interaction both masks the NF-κB NLS and inhibits its binding to DNA (reviewed in Refs. 31Ghosh S. May M.J. Kopp E.B. Annu. Rev. Immunol. 1998; 16: 225-260Crossref PubMed Scopus (4611) Google Scholar and 32Baldwin A.S. Annu. Rev. Immunol. 1996; 14: 649-681Crossref PubMed Scopus (5578) Google Scholar). In a stereotypical NF-κB-activating signaling pathway, a stimulus, including proinflammatory cytokines, viruses, bacteria, and other antigens, leads to phosphorylation of IκB by IκB kinase (IKK) (33Mercurio F. Zhu H. Murray B.W. Shevchenko A. Bennett B.L. Li J. Young D.B. Barbosa M. Mann M. Manning A. Rao A. Science. 1997; 278: 860-866Crossref PubMed Scopus (1853) Google Scholar,34DiDonato J.A. Hayakawa M. Rothwarf D.M. Zandi E. Karin M. Nature. 1997; 388: 548-554Crossref PubMed Scopus (1913) Google Scholar). Once phosphorylated, IκB is polyubiquitinated by a specific ubiquitin ligase (E3RSIKB) and is immediately degraded by the 26 S proteosome (35Winston J.T. Strack P. Beer-Romero P. Chu C.Y. Elledge S.J. Harper J.W. Genes Dev. 1999; 13: 270-283Crossref PubMed Scopus (811) Google Scholar, 36Yaron A. Hatzubai A. Davis M. Lavon I. Amit S. Manning A.M. Anderson J.S. Mann M. Mercurio F. Ben-Neriah Y. Nature. 1998; 396: 590-594Crossref PubMed Scopus (571) Google Scholar, 37Spencer E. Jiang J. Chen Z.J. Genes Dev. 1999; 13: 284-294Crossref PubMed Scopus (374) Google Scholar). Once its cytosolic anchor is eliminated, NF-κB rapidly enters the nucleus and initiates transcription of its target genes. The two major forms of IκB in most mammalian cells are IκBα and IκBβ. IκBα shuttles between the nucleus and the cytosol and accumulates in the nucleus of cells treated with leptomycin B to inhibit Crm1-dependent nuclear export (38Arenzana-Seisdedos F. Turpin P. Rodriguez M. Thomas D. Hay R.T. Virelizier J.L. Dargemont C. J. Cell Sci. 1997; 110: 369-378Crossref PubMed Google Scholar, 39Johnson C. Van Antwerp D. Hope T.J. EMBO J. 1999; 18: 6682-6693Crossref PubMed Google Scholar, 40Tam W.F. Lee L.H. Davis L. Sen R. Mol. Cell. Biol. 2000; 20: 2269-2284Crossref PubMed Scopus (133) Google Scholar). Furthermore, IκBα contains an NES in its amino terminus which, when deleted, also results in its accumulation in the nucleus (39Johnson C. Van Antwerp D. Hope T.J. EMBO J. 1999; 18: 6682-6693Crossref PubMed Google Scholar, 40Tam W.F. Lee L.H. Davis L. Sen R. Mol. Cell. Biol. 2000; 20: 2269-2284Crossref PubMed Scopus (133) Google Scholar). Thus, rather than serving as a static cytosolic anchoring protein, IκBα is a dynamic protein that binds NF-κB in the nucleus, disrupts its association with DNA, and causes its export to the cytosol. Additional mechanisms seem to regulate NF-κB nuclear translocation, including phosphorylation and modification by SUMO, a ubiquitin-like protein (41Drier E.A. Huang L.H. Steward R. Genes Dev. 1999; 13: 556-568Crossref PubMed Scopus (80) Google Scholar, 42Bhaskar V. Valentine S.A. Courey A.J. J. Biol. Chem. 2000; 275: 4033-4040Abstract Full Text Full Text PDF PubMed Scopus (87) Google Scholar). Nonetheless, studies of NF-κB and IκB provide a clear example of how protein localization can be regulated by varying the abundance of an anchoring protein. Numerous responses in eukaryotic cells are mediated by a conserved module of protein kinases: a MAP kinase kinase kinase (MAPKKK), a MAP kinase kinase (MAPKK), and a MAP kinase (MAPK). The three kinases form a cascade with the MAPKKK, once activated, phosphorylating and activating the MAPKK (also called MEK), which in turn activates the MAPK by phosphorylating it on both a conserved threonine and tyrosine residue. MAPKs fall into three distinct but related families: the extracellular signal-regulated (ERK) kinases, stress-activated (SAPK), and c-Jun/AP1 (JNK) kinases (reviewed in Ref. 43English J. Pearson G. Wilsbacher J. Swantek J. Karandikar M. Xu S. Cobb M.H. Exp. Cell Res. 1999; 253: 255-270Crossref PubMed Scopus (377) Google Scholar). All of these MAPKs phosphorylate a variety of substrates, some of which are cytosolic and some of which, including transcription factors, are nuclear. The studies discussed here will focus on MAPKs from three different organisms (see Table I). The closely related mammalian kinases, ERK1 and ERK2 (also called p44 and p42), are activated by the MEK1/2 kinases in response to mitogenic signals. Spc1/Sty1 from S. pombe and Hog1p from S. cerevisiae are members of the SAPK class. Several different types of environmental stress including high osmolarity, oxidative stress, heat, exposure to UV, and nutritional limitation lead to activation of the Spc1 MAPK by its upstream kinases, Wis4 (MKKK) and Wis1 (MKK). In contrast, the homologue of Spc1 in budding yeast, Hog1p, is activated only by high osmolarity and not by other types of stress. The HOG1 signaling pathway consists of distinct osmosensors in the plasma membrane (SLN1 and SHO1) and several different downstream components that lead to activation of Ste11p (MKKK), Pbs2p (MKK), and Hog1p (SAPK) (reviewed in Ref. 44Posas F. Takekawa M. Saito H. Curr. Opin. Microbiol. 1998; 1: 175-181Crossref PubMed Scopus (139) Google Scholar).Table IComponents of MAP kinase signaling cascades referred to in textS. pombeS. cerevisiaeMammalsMAPKSpc1/Sty1Hog1pERK1/2MAPKKWis1Pbs2pMEKTranscription factorsAtf1Msn2pELK1Pcr1Msn4pPhosphatasesPyp1Ptp2pMKP3Pyp2Ptp3pPTP-SL Open table in a new tab A similar pattern of localization has been observed for ERK1/2, Spc1, and Hog1p. In unstimulated cells, the MAPK localizes primarily to the cytosol but is not excluded from the nucleus. After stimulation, MAPK rapidly and markedly accumulates in the nucleus. This nuclear localization is transient, however, and MAPK redistributes to the cytosol upon cessation of signaling (45Lenormand P. Sardet C. Pages G. L'Allemain G. Brunet A. Pouyssegur J. J. Cell Biol. 1993; 122: 1079-1088Crossref PubMed Scopus (585) Google Scholar, 46Gaits F. Degols G. Shiozaki K. Russell P. Genes Dev. 1998; 12: 1464-1473Crossref PubMed Scopus (133) Google Scholar, 47Ferrigno P. Posas F. Koepp D. Saito H. Silver P.A. EMBO J. 1998; 17: 5606-5614Crossref PubMed Scopus (348) Google Scholar, 48Reiser V. Ruis H. Ammerer G. Mol. Biol. Cell. 1999; 10: 1147-1161Crossref PubMed Scopus (177) Google Scholar). For Hog1p, the resumption of cytosolic localization postsignaling is unperturbed in cells treated with protein synthesis inhibitors and thus is thought to occur via nuclear export rather than by protein degradation and subsequent resynthesis (47Ferrigno P. Posas F. Koepp D. Saito H. Silver P.A. EMBO J. 1998; 17: 5606-5614Crossref PubMed Scopus (348) Google Scholar). Regulation of MAPK localization is critical for signaling and is required for activation of nuclear targets. In mammalian cells, when ERK1/2 is prevented experimentally from entering the nucleus, phosphorylation of the transcription factor, ELK1, in the nucleus is disrupted, and reinitiation of DNA synthesis is blocked, whereas activation of cytosolic substrates is unperturbed (49Brunet A. Roux D. Lenormand P. Dowd S. Keyse S. Pouyssegur J. EMBO J. 1999; 18: 664-674Crossref PubMed Scopus (517) Google Scholar). The mechanism by which dynamic localization of MAPK is achieved is complex. Taken together, studies from yeast and mammalian cells suggest that changes in MAPK localization are caused by changes in its interaction with a host of different nuclear and cytosolic proteins rather than through a direct regulation of its rate of nuclear import and/or export. The small size of a MAPK monomer (40–50 kDa) suggests that MAPKs could enter the nuclear pore through diffusion. Although this may happen to some extent (50Adachi M. Fukuda M. Nishida E. EMBO J. 1999; 18: 5347-5358Crossref PubMed Scopus (241) Google Scholar) components of active nuclear transport are clearly required for correct localization of several MAPKs in vivo. The nuclear localization of Spc1 is blocked in a mutant strain defective for Pim1, the guanine nucleotide exchange factor for Ran in S. pombe (51Gaits F. Russell P. Mol. Biol. Cell. 1999; 10: 1395-1407Crossref PubMed Scopus (46) Google Scholar). Similarly, the accumulation of Hog1p in the nucleus requires both Gsp1, the S. cerevisiae homolog of Ran, and the karyopherin, Nmd5p (47Ferrigno P. Posas F. Koepp D. Saito H. Silver P.A. EMBO J. 1998; 17: 5606-5614Crossref PubMed Scopus (348) Google Scholar). Furthermore, nuclear export machinery, i.e. Crm1-related exportins, are required for the relocalization of Hog1p and Spc1 to the cytosol upon termination of signaling (47Ferrigno P. Posas F. Koepp D. Saito H. Silver P.A. EMBO J. 1998; 17: 5606-5614Crossref PubMed Scopus (348) Google Scholar, 51Gaits F. Russell P. Mol. Biol. Cell. 1999; 10: 1395-1407Crossref PubMed Scopus (46) Google Scholar). Although the activity of Ran and karyopherins clearly affect MAPK localization in vivo, no MAPK has been shown to contain a bona fide NLS or NES sequence or to bind directly to an importin or exportin in vitro using purified proteins. In vivo, MAPKs exist in large signaling complexes and are found in association with many different substrate and regulatory proteins (reviewed in Ref. 43English J. Pearson G. Wilsbacher J. Swantek J. Karandikar M. Xu S. Cobb M.H. Exp. Cell Res. 1999; 253: 255-270Crossref PubMed Scopus (377) Google Scholar). Thus, MAPKs likely travel in and out of the nucleus by associating with NLS- and NES-containing proteins. MAPK localization is tightly coupled to its activation state, and studies in yeast have demonstrated that a MAPK must be phosphorylated by its MKK to accumulate in the nucleus. Fission yeast lacking the Spc1-activating kinase, Wis1, fails to display nuclear enrichment of Spc1 following exposure of cells to stress. Furthermore, mutant Spc1 proteins that cannot be phosphorylated by Wis1 because of substitution of the conserved threonine or tyrosine residue with alanine do not accumulate in the nucleus following stress. These observations were made using strains in which endogenous Spc1 was replaced with unphosphorylatable mutant proteins expressed at physiological levels. Because cells expressing only these mutant Spc1 proteins are also compromised in the signaling pathway, it was possible that the observed defect in MAPK nuclear localization resulted from defects in events occurring downstream of MAPK activation. However, unphosphorylated Spc1 species also fail to accumulate in the nucleus in diploid cells that contain an additional fully functional copy ofspc1 + and are not sensitive to stress (46Gaits F. Degols G. Shiozaki K. Russell P. Genes Dev. 1998; 12: 1464-1473Crossref PubMed Scopus (133) Google Scholar). Similar findings were made in S. cerevisiae for Hog1p, and additional studies using mutant Hog1p that is catalytically inactive established that Hog1p kinase activity is not required for its nuclear translocation (47Ferrigno P. Posas F. Koepp D. Saito H. Silver P.A. EMBO J. 1998; 17: 5606-5614Crossref PubMed Scopus (348) Google Scholar, 48Reiser V. Ruis H. Ammerer G. Mol. Biol. Cell. 1999; 10: 1147-1161Crossref PubMed Scopus (177) Google Scholar). Studies of ERK1/2 mutant proteins have yielded somewhat mixed results. Activation by MEK1/2 was shown to be necessary and sufficient for nuclear localization of ERK1/2 (52Lenormand P. Brondello J.-M. Brunet A. Pouyssegur J. J. Cell Biol. 1998; 142: 625-626Crossref PubMed Scopus (193) Google Scholar). However, other studies demonstrated that unphosphorylatable derivatives of ERK1/2 do accumulate in the nucleus of transfected cells following exposure to mitogens (45Lenormand P. Sardet C. Pages G. L'Allemain G. Brunet A. Pouyssegur J. J. Cell Biol. 1993; 122: 1079-1088Crossref PubMed Scopus (585) Google Scholar). A major caveat with the latter studies is that the mutant ERK1/2 derivatives were expressed well above physiological levels. In fact, ERK1/2 was constitutively nuclear in transfected cells expressing the highest levels of the kinase. One proposed explanation for these findings is that the nuclear localization of unphosphorylatable ERK1/2 occurs via formation of heterodimers with endogenous (phosphorylatable) ERK (53Khokhlatchev A.V. Canagarajah B. Wilsbacher J. Robinson M. Atkinson M. Goldsmith E. H. C.M. Cell. 1998; 93: 605-615Abstract Full Text Full Text PDF PubMed Scopus (578) Google Scholar). Whether ERK1/2 actually forms dimers in vivo, however, is unclear. Alternatively expression of mutant ERKs at high levels leads to an altered pattern of localization because it perturbs the balance of MAPK association with other proteins (see below). Overexpression of Hog1p in budding yeast also leads to increased nuclear localization of this kinase in unstimulated cells (54Mattison C.P. Ota I.M. Genes Dev. 2000; 14: 1229-1235PubMed Google Scholar). The major factor that determines MAPK localization is its interaction with other proteins. In contrast to NF-κB, for MAPK there does not appear to be one major anchoring protein that determines its localization. Instead MAPK distribution seems to reflect the sum of its interactions with a large number of cellular proteins, many of which are probably still unidentified. Why does MAPK localization change after stimulation? There are several factors that influence its distribution. First, the affinity of MAPKs for at least some of their binding partners changes after phosphorylation by MKK. Second, as a result of signaling-induced changes in gene expression, the abundance of some cytosolic and/or nuclear MAPK-interacting proteins changes. Third, as a result of signaling the localization of some proteins with which MAPK interacts changes. These points are illustrated by studies of several different MAPK-interacting proteins. Studies in yeast and mammalian cells have examined the effect of interaction with MKK on MAPK localization. In each case the MAPK binds to its activating MKK, which is cytosolic. Analysis of MEK1/2 in mammalian cells revealed that these kinases shuttle in and out of the nucleus. MEK1/2 export from the nucleus is mediated by an amino-terminal NES and is sensitive to the inhibitor leptomycin B (55Adachi M. Fukuda M. Nishida E. J. Cell Biol. 2000; 148: 849-856Crossref PubMed Scopus (174) Google Scholar,56Fukuda M. Gotoh Y. Nishida E. EMBO J. 1997; 16: 1901-1908Crossref PubMed Scopus (331) Google Scholar). ERK1/2 binds tightly to MEK1/2 in its unphosphorylated but not in its activated state. Overexpression of the ERK-binding domain of MEK1/2 prevents mitogen-induced nuclear accumulation of ERK1/2, and this cytosolic retention requires the MEK1/2 NES (56Fukuda M. Gotoh Y. Nishida E. EMBO J. 1997; 16: 1901-1908Crossref PubMed Scopus (331) Google Scholar). Thus, these studies suggest that MEK1/2 shuttles in and out of the nucleus and preferentially catalyzes the export of unphosphorylated ERK from the nucleus. However, MAPKs interact with many additional types of proteins, and studies in yeast indicate that MKKs do not uniquely serve as cytosolic anchors for MAPK. In S. pombe, deletion of Wis1, the MKK for Spc1, does not result in constitutive nuclear localization of Spc1. Similarly in S. cerevisiae, mutants that lack Pbs2p still retain Hog1p in the cytosol. Furthermore in most cells a MAPK is more abundant than its MKK, again suggesting that MKK alone is not sufficient to tether MAPK in the cytosol of unstimulated cells. Because the return of MAPK to the cytosol correlates so closely with its dephosphorylation, it was proposed by several investigators that MAPK dephosphorylation is required for its nuclear export. This appears not to be the case for Hog1p, although interaction of Hog1p with phosphatases clearly influences its distribution. The budding yeast proteins Ptp2p and Ptp3p account for the bulk of tyrosine dephosphorylation of Hog1p (57Jacoby T. Flanagan H. Faykin A. Seto A.G. Mattison C. Ota I. J. Biol. Chem. 1997; 272: 17749-17755Abstract Full Text Full Text PDF PubMed Scopus (132) Google Scholar). Surprisingly, under high osmotic conditions, cells lacking the nuclear phosphatase, Ptp2p, exhibit a more rapid return of Hog1p to the cytosol than do wild type cells, despite the fact that Hog1p remains phosphorylated significantly longer in ptp2 mutants (54Mattison C.P. Ota I.M. Genes Dev. 2000; 14: 1229-1235PubMed Google Scholar). In contrast, cells lacking the cytosolic phosphatase, Ptp3p, show prolonged accumulation of Hog1p in the nucleus (54Mattison C.P. Ota I.M. Genes Dev. 2000; 14: 1229-1235PubMed Google Scholar). These findings suggest that in addition to dephosphorylating Hog1p, Ptp2p and Ptp3p might act as a nuclear and cytosolic tether, respectively, for this MAPK. Consistent with this idea, Hog1p forms a stable interaction with Ptp2p and Ptp3p in extracts. Also overexpression of PTP2causes nuclear accumulation of Hog1p, and overexpression ofPTP3 inhibits nuclear accumulation of Hog1p (54Mattison C.P. Ota I.M. Genes Dev. 2000; 14: 1229-1235PubMed Google Scholar). Ptp2p and Ptp3p are both less abundant than Hog1p, and cells lacking both of these proteins still exhibit signal-dependent changes in Hog1p localization. Therefore, as discussed for MEK, Ptp2p and Ptp3p likely represent two of many possible tethers for Hog1p. After exposure to osmotic stress, expression of PTP2 and PTP3increases in a Hog1p-dependent manner (57Jacoby T. Flanagan H. Faykin A. Seto A.G. Mattison C. Ota I. J. Biol. Chem. 1997; 272: 17749-17755Abstract Full Text Full Text PDF PubMed Scopus (132) Google Scholar). Thus, the impact of these phosphatases on signaling is complex: as their abundance increases, they both negatively regulate Hog1p through dephosphorylation and influence its localization. In mammalian cells, a family of dual specificity phosphatases is implicated in negative feedback control of MAPK. Several of these phosphatases are localized to the nucleus (MKP-1, MKP-2, PAC-1), whereas others localize to the cytosol (MKP-3/Pyst-1). Overexpression of a catalytically inactive mutant of MKP-3 causes retention of ERK1/2 in the cytosol, suggesting that tethering of MAPKs by their cognate phosphatases may be a general phenomenon (49Brunet A. Roux D. Lenormand P. Dowd S. Keyse S. Pouyssegur J. EMBO J. 1999; 18: 664-674Crossref PubMed Scopus (517) Google Scholar). A tyrosine phosphatase, PTP-SL, also seems to act as a cytosolic anchor for ERK1/2 in mammalian cells. PTP-SL, an integral plasma membrane protein, dephosphorylates and inactivates ERK2 and is also a substrate of activated ERK2 (58Zuniga A. Torres J. Ubeda J. Pulido R. J. Biol. Chem. 1999; 274: 21900-21907Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). A conserved 16-amino acid region of PTP-SL, designated KIM, for kinase interaction motif, mediates its interaction with ERK2. This KIM domain seems to serve two different functions. It promotes catalytic interactions between PTP-SL and serves as a cytosolic anchoring site for ERK2 (58Zuniga A. Torres J. Ubeda J. Pulido R. J. Biol. Chem. 1999; 274: 21900-21907Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar). Phosphorylation of residues in KIM by protein kinase A decreases ERK binding in vitro and decreases PTP-SL-dependent cytosolic retention of ERKin vivo (59Blanco-Aparicio C. Torres J. Pulido R. J. Cell Biol. 1999; 147: 1129-1135Crossref PubMed Scopus (143) Google Scholar). Thus, other signaling pathways may potentiate changes in ERK localization by modulating its interaction with anchoring proteins. In fission yeast, Spc1 phosphorylates and activates the nuclear transcription factor, Atf1 (60Wilkinson M.G. Samuels M. Takeda T. Toone W.M. Shieh J.-C. Toda T. Millar J.B.A. Jones N. Genes Dev. 1996; 18: 2289-2301Crossref Scopus (312) Google Scholar, 61Shiozaki K. Russel P. Genes Dev. 1996; 10: 2276-2288Crossref PubMed Scopus (365) Google Scholar). In cells lacking Atf1 or its transcriptional co-activator Pcr1, Spc1 fails to accumulate in the nucleus during signaling, and Atf1 binds to Spc1 in vitro, suggesting that Atf1 acts as a nuclear tether for Spc1 (46Gaits F. Degols G. Shiozaki K. Russell P. Genes Dev. 1998; 12: 1464-1473Crossref PubMed Scopus (133) Google Scholar). However, mutants lacking Atf1 are defective for the expression of many stress-induced gene products, and some of these may also impact Spc1 localization. Although the interaction between Atf1 and Spc1 is insensitive to the phosphorylation state of Spc1, Atf1 mRNA and protein levels increase following exposure to stress (60Wilkinson M.G. Samuels M. Takeda T. Toone W.M. Shieh J.-C. Toda T. Millar J.B.A. Jones N. Genes Dev. 1996; 18: 2289-2301Crossref Scopus (312) Google Scholar, 61Shiozaki K. Russel P. Genes Dev. 1996; 10: 2276-2288Crossref PubMed Scopus (365) Google Scholar). Thus, increased abundance of Atf1 during signaling likely contributes to nuclear accumulation of Spc1. Similarly, in budding yeast, cells lacking Msn2p and Msn4p, two stress-activated transcription factors, show decreased nuclear accumulation of Hog1p following osmotic stress (48Reiser V. Ruis H. Ammerer G. Mol. Biol. Cell. 1999; 10: 1147-1161Crossref PubMed Scopus (177) Google Scholar). In mammalian cells, protein synthesis is required for nuclear accumulation following serum stimulation, suggesting that in these cells the number of high affinity binding sites for ERK1/2 may also increase as a result of signal-induced changes in gene expression (52Lenormand P. Brondello J.-M. Brunet A. Pouyssegur J. J. Cell Biol. 1998; 142: 625-626Crossref PubMed Scopus (193) Google Scholar). MAPKAP-2, a kinase that is phosphorylated in the nucleus by p38, the human homologue of HOG1, also influences p38 localization. Once phosphorylated, MAPKAP-2 redistributes to the cytosol through unmasking of a leptomycin B-sensitive NES and acts as a cytosolic tether for p38 (62Engel K. Kotlyarov A. Gaestel M. EMBO J. 1998; 17: 3363-3371Crossref PubMed Scopus (236) Google Scholar, 63Ben-Levy R. Hooper S. Wilson R. Paterson H.F. Marshall C.J. Curr. Biol. 1998; 8: 1049-1057Abstract Full Text Full Text PDF PubMed Scopus (291) Google Scholar). An activated form of MAPKAP-2 in which its phosphorylation sites have been mutated to glutamate is constitutively cytosolic and when co-expressed with p38 causes this MAPK to also remain in the cytosol. Several factors contribute to the signaling-induced redistribution of MAPK. Before signaling, MAPK is unphosphorylated and inactive, and for the MAPKs discussed here, localization is primarily cytosolic. This cytosolic distribution results from high affinity interactions between unphosphorylated MAPK and key cytosolic proteins (such as MKK) and from the low abundance of key nuclear MAPK-interacting proteins (such as Atf1 in S. pombe and Ptp2p in S. cerevisiae). In response to a signal, MAPK becomes phosphorylated and activated. Its affinity for some cytosolic proteins goes down, its affinity for some nuclear proteins may go up, and the abundance of nuclear proteins for which MAPK has high affinity may increase as a result of signaling-induced changes in gene expression. Together these factors cause an accumulation of MAPK in the nucleus that persists as long as the signaling pathway remains active. The cessation of signaling coincides with MAPK dephosphorylation and decreased expression of signaling-induced genes. Thus, the cell returns to its basal state, and dephosphorylated MAPK resumes its cytosolic distribution because of increased interaction with cytosolic proteins in combination with a loss of high affinity nuclear binding sites. Recent studies, such as those described above, have established that protein function is often regulated at the level of localization during signaling. In this review we focused on mechanisms that lead to increased nuclear localization during signaling. In other cases, proteins are retained in the nucleus to ensure their inactivity prior to signaling and are released from the nucleus to allow their activation in the cytosol or at the surface of intracellular membranes (28Shou W. Seol J.H. Shevchenko A. Baskerville C. Moazed D. Chen Z.W. Jang J. Shevchenko A. Charbonneau H. Deshaies R.J. Cell. 1999; 97: 233-244Abstract Full Text Full Text PDF PubMed Scopus (597) Google Scholar, 64Shimada Y. Gulli M.P. Peter M. Nat. Cell Biol. 2000; 2: 117-124Crossref PubMed Scopus (151) Google Scholar, 65Nern A. Arkowitz R.A. J. Cell Biol. 2000; 148: 1115-1122Crossref PubMed Scopus (94) Google Scholar, 66Visintin R. Hwang E.S. Amon A. Nature. 1999; 398: 818-823Crossref PubMed Scopus (488) Google Scholar). In yet other scenarios proteins must shuttle both in and out of the nucleus to function, and disrupting this trafficking interferes with activity (29Mahanty S.K. Wang Y.M. Farley F.W. Elion E.A. Cell. 1999; 98: 501-512Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar). As our understanding of the many mechanisms governing protein function in vivo increases, so does our appreciation of the myriad of possible modes for their regulation.