Molecular cloning and functional characterization of a novel mitogen-activated protein kinase phosphatase, MKP-4.

Extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase/stress-activated protein kinase (JNK/SAPK), and p38/RK/CSBP (p38) mitogen-activated protein (MAP) kinases are target enzymes activated by a wide range of cell-surface stimuli. Recently, a distinct class of dual specificity phosphatase has been shown to reverse activation of MAP kinases by dephosphorylating critical tyrosine and threonine residues. By searching the expressed sequence tag data base (dbEST) for homologues of known dual specificity phosphatases, we identified a novel partial human sequence for which we isolated a full-length cDNA (termed MKP-4). The deduced amino acid sequence of MKP-4 is most similar to MKP-X/PYST2 (61% identity) and MKP-3/PYST1 (57% identity), includes two N-terminal CH2 domains homologous to the cell cycle regulator Cdc25 phosphatase, and contains the extended active site sequence motif VXVHCXAGXSRSXTX3AYLM (where X is any amino acid) conserved in dual specificity phosphatases. MKP-4 produced in Escherichia coli catalyzes vanadate-sensitive breakdown of p-nitrophenyl phosphate as well as in vitro inactivation of purified ERK2. When expressed in COS-7 cells, MKP-4 blocks activation of MAP kinases with the selectivity ERK > p38 = JNK/SAPK. This cellular specificity is similar to MKP-3/PYST1, although distinct from hVH-5/M3-6 (JNK/SAPK = p38 >>> ERK). Northern analysis reveals a highly restricted tissue distribution with a single MKP-4 mRNA species of approximately 2.5 kilobases detected only in placenta, kidney, and embryonic liver. Immunocytochemical analysis showed MKP-4 to be present within cytosol although punctate nuclear staining co-localizing with promyelocytic protein was also observed in a subpopulation (10-20%) of cells. Chromosomal localization by analysis of DNAs from human/rodent somatic cell hybrids and a panel of radiation hybrids assign the human gene for MKP-4 to Xq28. The identification and characterization of MKP-4 highlights the emergence of an expanding family of structurally homologous dual specificity phosphatases possessing distinct MAP kinase specificity and subcellular localization as well as diverse patterns of tissue expression.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) Y08302.
¶ To whom correspondence should be addressed. 3/PYST1 is also the first dual specificity phosphatase to display an exclusively cytosolic rather than nuclear localization (43,45). Interestingly, several dual specificity phosphatases undergo powerful regulated expression following cell stimulation by growth factors and/or exposure to stresses, suggesting one major mechanism for control of MAP kinases (29, 32, 34, 36 -40, 42-45). One critical unanswered question is whether the existence of multiple ERK, JNK/SAPK, and p38 kinase genes FIG. 1. Nucleotide and encoded amino acid sequences of human MKP-4 cDNA. Amino acids are indicated using single-letter code. Nucleotides and amino acids are numbered at the end of each sequence line, starting at the first nucleotide of the cloned cDNA and at the initiator methionine, respectively. The extended active-site motif together with a critical catalytic aspartate conserved in dual specificity phosphatases are indicated by the solid boxes. Residues within two (CH2) domains conserved with the Cdc25 phosphatase are in shaded boxes. Underlined nucleotides within the 3Ј untranslated region correspond to the StyI fragment obtained from the EST clone (accession number R51175) and used to screen the human placental cDNA library. Also indicated within the 3Ј tail is the predicted poly(A) signal sequence (open box) preceding the 3Ј poly(A) tail. and splice variants (1-8, 13, 26, 27) demands a similarly diverse range of dual specificity phosphatases to allow specific and potentially highly compartmentalized control of MAP kinase signaling.

Materials-Restriction
Identification and Cloning of MKP-4 -By screening the expressed sequence tag data base (dbEST) for sequences similar to dual specificity phosphatases, we identified a partial human cDNA (accession number R51175) highly homologous to the C terminus of MKP-3 (43). This clone (ID 38872) was obtained from Research Genetics, Inc. (Huntsville, AL) I.M.A.G.E. Consortium (Lawrence Livermore National Laboratory (LLNL), Livermore, CA) and used to prepare a random [ 32 P]dCTPradiolabeled 428-bp StyI fragment for screening an oligo(dT)-primed human placenta gt10 cDNA library (provided by Dr. G. Buell, Geneva Biomedical Research Institute), which was performed as described previously (43). Three positive clones were isolated and EcoRI inserts subcloned into pBluescript SK(Ϫ) (Stratagene) for sequencing. One clone was 2,303 kilobase pairs in length and contained an open reading frame of 1152-bp with a poly(A) signal sequence preceding a 3Ј poly(A) tail. The nucleotide sequence is 94.5% identical with the original 474-bp EST clone (R51175). This clone was termed MKP-4.
MKP-4 in Vitro Transcription and Translation-Full-length MKP-4 in pBluescript SK(Ϫ) was translated in vitro using a TNT T7 RNA polymerase-coupled reticulocyte lysate system (Promega) in 25-l incu-bations for 2 h at 30°C in the presence of [ 35 S]methionine as described by the manufacturer. Reactions were stopped by addition of Laemmli sample buffer (47) and resolved using a 12% SDS-polyacrylamide gel, which was then soaked in Amplify™ (Amersham) to reveal labeled protein.
MKP-4 Expression in Escherichia coli-For bacterial expression the MKP-4 open reading frame was subcloned into pGEX4T3 (Pharmacia) as follows. A StyI 1152-bp fragment was first isolated, Klenow-treated, and ligated to a SmaI-digested pBluescript SK(Ϫ). A construct with the correct orientation was then identified, digested with BamHI and NcoI, Klenow-treated, and religated. The MKP-4 insert was then subcloned as a BamHI-XhoI fragment in frame with the GST coding sequence. pGEX/MKP-4 was used to transform E. coli BLR (Novagen). Cells were grown overnight to saturation in LB medium containing 100 g/ml ampicillin, after which growth was resumed by diluting the culture 1:50 and incubating at 37°C for 1 h. Following transfer to 20°C for 1 h, IPTG was then added to a final concentration of 100 M and cells cultured for another 9 h. Cells were harvested, resuspended in phosphate-buffered saline containing 1% (v/v) Triton X-100, 5 mM dithiothreitol, 2 mM EDTA, 5 mM benzamidine, and 1 mM Pefabloc™ (Boehringer Mannheim) and lysed by passing three times through a French press at 1000 p.s.i. The extract was then centrifuged at 10,000 ϫ g for 15 min at 4°C, and the supernatant incubated with glutathione-Sepharose (Pharmacia, Uppsala), washed, and eluted as described by the manufacturer.
Phosphatase Activity-GST/MKP-4 fusion protein was assayed for intrinsic phosphatase activity as described (29,45) with minor modifications. Briefly, 8 -40 g of GST/MKP4 was incubated for 1 h at 37°C in a reaction volume of 800 l containing 20 mM p-nitrophenyl phosphate (pNPP), 50 mM imidazole (pH 7.5), and 5 mM dithiothreitol. Reactions were stopped by addition of 1 M NaOH and pNPP hydrolysis measured by absorbance at 410 nm.
In Vitro ERK2 Inactivation by MKP-4 -Constitutively activated rabbit MAP kinase kinase (MEK1 S218E,S222E) and mouse ERK2 were generated as GST fusion proteins in E. coli and purified using glutathione-Sepharose and nickel-agarose columns as described (48,49). Human stathmin (50) expressed in E. coli was purified using a Q-Sepharose column 2 and biotinylated using sulfo-NHS-LC-biotin (Pierce) at 4°C for 2 h according to the manufacturers instructions. Biotinylated stathmin was then dialyzed against phosphate-buffered saline and 20 mM Tris-HCl (pH 7.5). ERK2 enzymatic activity was based on measurement of stathmin phosphorylation by SPA in 96-well sample plates. 2  0.1% Triton X-100, 5 mM EDTA, and 50 M ATP and left to incubate for 1 h at room temperature. Plates were then centrifuged at 1800 ϫ g for 5 min and counted in a 1450 MicrobetaPlus liquid scintillation counter (Wallac).

MKP-4 Mammalian Expression
Plasmid--MKP-4 was tagged at the C terminus with the Myc epitope and subcloned into pMT-SM (43,51) as follows. A 1152-bp StyI fragment containing the entire open reading frame of MKP-4 was blunt-ended using mung bean nuclease and ligated with two sets of double-stranded oligonucleotide adaptors, which simultaneously added SalI and EcoRI restriction sites together with sequence encoding the Myc epitope EQKLISEEDLN followed by a stop codon at the C terminus. The SalI-EcoRI fragment was then subcloned into the corresponding sites of pMT-SM.
Cell Culture, Transfection, and Stimulation-COS-7 cells were grown under 7.5% CO 2 in Dulbecco's modified Eagles's medium con-taining 10% (v/v) fetal calf serum and 2 mM glutamine. Cells were grown in 6-well plates (35 mm diameter) to 80% confluence and transfected using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. Transfections were performed using the following plasmid combinations: 1.0 g of pcDNA1-HA-p44 ERK1, pMT2T-HA-p54-SAPK␤, or pcDNA3-HA-p38 together with 0.01, 0.05, 0.1, 0.5, or 1.0 g of pMT-SM-Myc-MKP4. Total plasmid concentration was maintained constant by supplementing with pMT-SM vector. Following 6 h of incubation with LipofectAMINE and plasmid DNA, cells were washed and grown for 40 h before starvation by incubation in serum-free medium for 2 h. Cells were then exposed to EGF (10 nM), anisomycin (10 g/ml), or H 2 O 2 (0.5 mM) at 37°C for 10 -30 min. Where indicated, cells were triple-transfected with 0.25 g of pEXV3-Myc-p21 ras (G12V), 1.0 g of pcDNA1-HA-p44 ERK1, together with 0.01-1.0 g of pMT-SM-Myc-MKP4; under these conditions, starvation was for Immunoprecipitation and Immune Complex Kinase Assays-Following stimulation cells were washed twice in 2 ml of ice-cold phosphatebuffered saline and scraped into Eppendorf tubes with 300 l of buffer TP (50 mM Tris-HCl, pH 7.5, containing 150 mM NaCl, 1% (v/v) Nonidet P-40, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 10 mM NaF, 1 mM sodium pyrophosphate, 1 mM sodium vanadate, 1 mM EDTA, 10 nM calyculin, and 25 mM ␤-glycerophosphate). Cells were then homogenized using a sonicator probe at full power for 3 s on ice. Aliquots (150 l) of the COS-7 cell homogenate were mixed with 850 l of buffer TP and rotary mixed for 1 h at 4°C, after which time they were centrifuged at 100,000 ϫ g for 20 min at 4°C. The supernatant (800 l) was then mixed by rotary mixing for 2 h at 4°C with 75 l of a preformed immunoprecipitating complex

FIG. 4. In vitro transcription and translation of MKP-4 protein.
Full-length MKP-4 subcloned in pBluescript SK(Ϫ) was translated in vitro using a RNA polymerase-coupled reticulocyte lysate in the presence of [ 35 S]methionine. Synthesized proteins were visualized following separation in a 12% gel. MKP-4 can be seen as a major band running at ϳ44 kDa when using T7 polymerase (T 7 ) but not in control incubations with T3 polymerase (T 3 ). (100 l of HA-epitope specific HA.11 monoclonal antibody preincubated with 900 l of 50% (v/v) protein A-Sepharose beads in 10 mM Tris-HCl, pH 7.5, for 2 h at 4°C). Beads were then sedimented by centrifugation at 10,000 ϫ g for 3 min and washed twice in 1.0 ml of ice-cold buffer TP and once in 1.0 ml of buffer K (50 mM HEPES, pH 7.4, containing 20 mM MgCl 2 , 200 M sodium vanadate, 2 mM dithiothreitol, and 10 mM ␤-glycerophosphate) followed by final resuspension in 50 l of buffer K. Immune complex assays were performed by mixing 10 l of bead suspension with 10 l of 6 M [␥-32 P]ATP (ϳ300,000 dpm/pmol), 10 l of substrate protein (15 g of MBP, 10 g of GST-ATF-2, or 10 g of GST-MAPKAP kinase 2 ⌬3), and 30 l of buffer K followed by incubation for 30 min at 30°C. Reactions were terminated by adding 15 l of 10 ϫ Laemmli sample buffer (47) and heating for 5 min at 95°C. Following centrifugation at 10,000 ϫ g for 5 min, supernatants were analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography as described previously (43).
Western Blotting-For immunodetection of MAP kinases precipitated for immune complex assay, 10 l of immunocomplex beads (see above) were diluted with 20 l of 10 ϫ Lammli sample buffer and heated for 5 min at 95°C, followed by centrifugation at 10,000 ϫ g for 5 min. Western analysis was then performed using supernatant fractions (20 l) by SDS-polyacrylamide gel electrophoresis and electrotransfer to nitrocellulose membranes as described (43). Immunoprecipitated HA-ERK1, HA-p54-SAPK␤, and HA-p38 MAP kinases were detected using biotinylated HA-epitope-specific monoclonal antibody 12CA5 (Boehringer Mannheim) together with avidin-horseradish peroxidase conjugate and enhanced chemiluminescence. To detect levels of Myc-MKP-4 and Myc-p21 ras (G12V) expression in transfected COS-7 cells, Western analysis was performed on cell homogenates (20 g of protein) using anti-Myc epitope monoclonal antibody 9E10 and goat anti-mouse IgG horseradish peroxidase conjugate with chemiluminescence.
Subcellular Localization of MKP-4 -Rat sympathetic neurons from superior cervical ganglia of newborn rats were prepared, cultured, and microinjected with pMT-SM-Myc-MKP4 or pMT-SM-Myc-MKP-3 as described (43). COS-7 or NIH 3T3 cells were transfected with the same MKP-4 plasmid using LipofectAMINE as described above. Immunocytochemistry was performed using monoclonal antibody 9E10 with FITC-conjugated goat anti-mouse second antibody. In double labeling experiments, nuclear bodies were labeled with a rabbit polyclonal antibody detecting promyelocytic protein (PML), which was detected using a combination of biotinylated anti-rabbit antibody and avidin-Cy3 conjugate. Microscopic analysis of cells was performed using ultraviolet fluorescence.
Somatic Cell Hybrid Analysis-PCR primers corresponding to the sequences 5Ј-CAACGATGCCTATGACCTGG-3Ј (sense; nucleotides 1052-1071, Fig. 1) and 5Ј-GAAGGCGCCATCACTGGTGG-3Ј (antisense; nucleotides 1231-1250, Fig. 1) were used to amplify a 200-bp fragment of the 3Ј coding sequence of MKP-4 from a panel of human/ rodent somatic cell hybrid DNAs (obtained from the UK Human Genome Mapping Project Resource Center). DNA amplification was performed with 32 cycles of denaturation (94°C, 30 s), annealing (52°C, 30 s), and extension (72°C, 1 min), with a final extension (72°C, 5 min). The products were analyzed by electrophoresis in 4% Metaphor (FMC Bioproducts) agarose gels. Chromosomes retained by the hybrids are summarized schematically in Fig. 11. The same primers and PCR conditions were used to amplify MKP-4 from a panel of radiation hybrid DNAs (52).
Southern Blot Analysis-Human male and female genomic DNAs (10 g) were digested overnight with EcoRI or HindIII and subjected to agarose gel electrophoresis followed by transfer to Hybond N ϩ (Amersham). Membranes were hybridized with a random [ 32 P]dCTP-radiolabeled 428-bp StyI fragment corresponding to part of the 3Ј-untranslated region of MKP-4. Hybridization was performed as described (53).

RESULTS AND DISCUSSION
Identification and Cloning of MKP-4 -To identify novel dual specificity phosphatases, we performed BLAST computer searches of the expressed sequence tag data base (dbEST) for partial cDNAs showing similarity with MKP-3 (43). One human sequence (accession number R51175) was found to be highly homologous to the C terminus of MKP-3 and, moreover, encoded the extended active site sequence motif VXVHCXAGX-SRSXTXXXAYLM conserved in dual specificity phosphatases ( and used to prepare a probe to screen a human placental cDNA library (see "Experimental Procedures"). One of the positive clones contained an insert of 2,303 bp, and nucleotide sequence analysis revealed 113 bp of 5Ј-untranslated sequence upstream of a translation start site CGCCCATGG compatible with the Kozak consensus sequence (Fig. 1). This is followed by an open reading frame of 1152 bp and a poly(A) signal sequence preceding a 3Ј poly(A) tail (Fig. 1). Over the relevant region, the nucleotide sequence displays 94.5% identity with the original 474-bp EST clone (R51175), indicating that we had cloned the corresponding full-length cDNA, which we called MKP-4.
The MKP-4 open reading frame encodes a protein of 384 amino acids, and comparison of this sequence with the Gen-Bank™/EMBL Data Bank revealed greatest homology with other members of the dual specificity phosphatase family (see Introduction), with identities ranging from 61% for MKP-X/ PYST2 down to 35% for hVH-5/M3-6. Direct comparison between MKP-4 and its closest full-length homologue, MKP-3/ PYST1 (57% identity), indicates that regions of greatest similarity include, and fall either side of, the extended active site sequence motif VXVHCXAGXSRSXTX 3 AY(L/I)M (where X is any amino acid) (Fig. 2, solid boxes). This is also true for all other members of the dual specificity phosphatase family, with which MKP-4 displays significantly greater sequence identity within the catalytic C-terminal half of the molecule (Fig. 3). Importantly, this region in MKP-4 includes Asp-259, Cys-290, and Ser-300 (Fig. 1), which are cognate to Asp-92, Cys-124, and Ser-131 of the dual specificity phosphatase VHR and are likely to represent critical residues underlying enzymatic activity (54,55). According to a model of VHR catalysis, Cys-290 of MKP-4 may function as an active-site nucleophile forming a covalent thiol-phosphate intermediate, while Asp-259 acts as a general acid to donate a proton to the leaving group (54,55). Despite lower homology within the N-terminal half of the dual specificity phosphatases (Figs. 2 and 3), this region of MKP-4 contains two stretches containing residues conserved with two segments flanking the active site of the Cdc25 phosphatase (Fig. 1, shaded boxes). These regions of homology are termed CH2 domains and are also found in other dual specificity phosphatases, although their functional significance is currently unknown (56,57).
MKP-4 Expression, Phosphatase Activity, and MAP Kinase Inactivation-MKP-4 is predicted to encode a protein of 41.8 kDa, and this is in agreement with the size of a single 35 Slabeled protein band generated upon in vitro transcription and translation using a T7 RNA-polymerase-coupled reticulocyte lysate system (Fig. 4). To establish whether MKP-4 possesses endogenous catalytic activity, we overexpressed full-length MKP-4 as a GST fusion protein in E. coli and tested the purified protein using pNPP as substrate. Increasing concentrations of GST/MKP-4 resulted in a linear rise in pNPP hydrolysis (up to 40 g of protein), and this catalytic activity was effectively inhibited by the protein-tyrosine phosphatase inhibitor sodium vanadate (Fig. 5A). Given the close homology between MKP-4 and MKP-3/ PYST1, which is highly selective for inactivating ERK family MAP kinases (45,46), we next established whether GST/ MKP-4 is able to inactivate ERK2 in vitro. Using a combination of purified recombinantly expressed proteins, ERK2 enzymatic activity was assessed by measuring phosphorylation of the MAP kinase substrate stathmin (50). Phosphorylation of stathmin is dependent upon ERK2 co-incubation with the constitutively active MAP kinase kinase MEK1 (S217E,S221E) (19). This phosphorylation reflects ERK2 activation as MEK1 (S217E,S221E) is alone unable to phosphorylate stathmin directly (data not shown). Inclusion of GST/MKP-4 in the reaction mix results in efficient blocking of ERK2 enzymatic activity (Fig. 5B). Importantly, GST/MKP-4 is unable to dephosphorylate stathmin directly (control, Fig. 5B), indicating highly effective ERK2 inactivation under these assay conditions.
To establish whether MKP-4 displays similar enzymatic activity within intact cells, we next co-transfected the HA-tagged MAP kinases p44 ERK1, p54-SAPK␤, or p38 together with Myc-tagged MKP-4 in COS-7 cells. To obtain a clear impression of the relative effectiveness of MKP-4 to inactivate each MAP kinase, cells were transfected using a range of plasmid concentrations (0.01-1.0 g/well). These concentrations were chosen to give a reproducible dose-dependent increase in the levels of immunodetectable MKP-4 protein (at ϳ44 kDa), while expression of each MAP kinase is unaltered (Fig. 6). Thus, although with higher plasmid concentrations levels of MKP-4 protein are likely to represent a considerable overexpression, these conditions allow a direct comparison of MKP-4-dependent inactivation of different MAP kinase family members. Using this approach MKP-4 displays moderate selectivity for p44 ERK1 when compared with p54 SAPK␤ and p38 MAP kinases. Hence, EGF-stimulated p44 ERK1 activation is inhibited by ϳ50% when cells were transfected with only 0.01 g of plasmid, conditions under which MKP-4 protein was not detected by Western analysis (Fig. 6). This level of MKP-4 expression had little effect on anisomycin-stimulated p54 SAPK␤ or H 2 O 2 -dependent activation of p38 MAP kinase (Fig. 6). Inhibition of p44 ERK1 activation is increased to ϳ80% following tranfection with 0.05 g of plasmid, while under identical conditions stress-induced activation of p54 SAPK␤ and p38 MAP kinases was suppressed by Ͻ50% (Fig. 6). Full blockade of p54 SAPK␤ and p38 MAP kinases required maximal expression of MKP-4 using 1.0 g of plasmid (Fig. 6). Together, these studies confirm that as observed in vitro (Fig. 5B), MKP-4 is able to inactivate ERK family MAP kinases when expressed in mammalian cells. Importantly, these results also indicate that MKP-4 selectivity (ERK Ͼ p38 ϭ JNK/SAPK) is similar to its close homologue MKP-3/PYST1, which is significantly more effective against ERK family isoforms (ERK Ͼ Ͼ JNK/SAPK ϳ p38) (45,46). In contrast to this, MKP-4 enzymatic specificity is quite different from hVH-5/M3-6 (JNK/SAPK ϳ p38 Ͼ ϾϾ ERK) (46), MKP-2/ hVH-2/TYP-1 (ERK ϭ JNK/SAPK Ͼ p38) (58), and PAC-1 (ERK ϭ p38 Ͼ JNK/SAPK) (58). It is of note that, despite these observations, we cannot exclude the possibility that additional cellular proteins, possibly unrecognized MAP kinases, also represent targets for MKP-4 action particularly at more physiological levels of expression. Notwithstanding this caveat, the existence of multiple dual specificity phosphatases with clear selectivity for inactivation of known MAP kinases strongly suggests specific functional roles for different family members.
Mutated constitutively activated p21 ras (G12V) stimulates ERK MAP kinases, and this may underlie mitogenesis and cellular transformation induced by this oncogene (19,59,60). To test whether MKP-4 is able to block oncogenic p21 ras -de-pendent MAP kinase activation, COS-7 cells were triple-transfected with constitutively active Myc-tagged p21 ras (G12V), p44 HA-ERK1, and varying concentrations of Myc-MKP-4 plasmid. This experiment reveals that, as with acute exposure to EGF (Fig. 6), MKP-4 blocked completely p44 ERK1 activation by p21 ras (G12V), although this required higher concentrations of plasmid and expressed protein (Fig. 7). MKP-4-dependent blockade of growth factor and oncogene-stimulated MAP kinase activation as observed here appears functionally equivalent to previous observations of ERK inhibition following expression of a dominant negative mutant form of MAP kinase kinase MEK1 (19), or in cells treated with the MEK1 inhibitor PD098059 (61). Since inhibition of ERK activity using these approaches also blocks growth factor-stimulated proliferation and even reverts oncogene-driven transformation (19,61), one important function for MKP-4 could be to inhibit cellular proliferation and possibly act as a tumor suppressor.
MKP-4 Tissue Distribution-To establish which cells and tissues may be major sites for MKP-4 action, we performed Northern blot analysis on poly(A) ϩ RNA isolated from a range of human cell and tissue types. Using a probe from the 3Јuntranslated sequence of MKP-4, we detected a single band at 2.5 kilobases expressed only in placenta and kidney (Fig. 8) and fetal liver (not shown). We failed to detect any MKP-4 mRNA in the following adult cells and tissues: heart, lung, liver, skeletal muscle, pancreas, adrenal medulla, adrenal cortex, testis, ovary, thyroid, colon, intestine, stomach, appendix, thymus, spleen, lymph node, leukocytes, bone marrow, prostate, or any of 15 brain regions tested. This distribution pattern may be similar to the human dual specificity phosphatase hVH-2/ TYP-1, which is also enriched in placenta while being expressed in other tissues at low or undetectable levels (36,38). In contrast, MKP-4 expression differs substantially from PYST1, CL100, and hVH-3 (also human genes), which although detected in placenta, are also found in a number of other cell and tissue types at similar or higher levels (39,40,45,57). Three other human dual specificity phosphatases, PAC-1, PYST2, and hVH-5, display a highly restricted expression pattern, although these are distinct from MKP-4 insofar that they are enriched in hematopoietic tissues, liver, or brain and skeletal muscle tissue, respectively (34,42,45). Together, these studies highlight significant differences in the distribution of MKP-4 and other dual specificity phosphatase family members, indicating important roles regulating MAP kinases in a limited complement of cell and tissue types. High level MKP-4 expression in placenta, for instance, appears to parallel the distribution ERK1 and ERK2 outside of the nervous system (1) and supports the notion of regulatory interaction under physiological conditions. MKP-4 Subcellular Localization-Sustained activation of MAP kinases has been reported to trigger their translocation to the nucleus (62,63). This relocalization places MAP kinases in close proximity with several nuclear target proteins, including the transcription factors Elk-1, c-Jun, ATF-2, and CHOP (11)(12)(13)(14)(15)(16). ERK isoforms are also associated with microtubules (64), and this could indicate sites of specific subcellular anchorage as demonstrated for MAP kinases in Saccharomyces cerevisiae, which bind the scaffold protein Ste5 (65). Such an emerging picture of MAP kinase compartmentalization could suggest that dual specificity phosphatases responsible for their regulation will be localized to overlapping subcellular sites. For instance, recently, we and others have reported that MKP-3/ PYST1 is exclusively cytosolic, indicating a role inactivating MAP kinases in non nuclear compartments (43,45). This contrasts with MKP-1/CL100, PAC-1, hVH-2/MKP-2/TYP-1, and hVH-3/B23, which are localized entirely within the nucleus (34,36,39), while another dual specificity phosphatase family member, hVH-5/M3-6, is cytosolic or nuclear depending on the cellular environment (41). We have studied the subcellular localization of Myc-tagged MKP-4 following expression in rat sympathetic neurons as well as in COS-7 and NIH 3T3 cell lines. In sympathetic neurons both MKP-3 and MKP-4 were immunodetectable within the cell body cytoplasm, while MKP-4 protein was also clearly detectable within neurites (Fig.  9, A and B). Interestingly, in some neurons, MKP-4 protein appeared to concentrate in regions resembling synaptic swellings (Fig. 9B, indicated by arrows). By electron microscopy these regions possess a high content of vesicular membranes characteristic of premature synaptic structures. 3 MKP-4 was also localized within the cytosol of NIH 3T3 (Fig. 9, C and D) and COS-7 cells (Fig. 9, E and F), although a subset of transfected cells (10 -20%) also displayed punctate staining within the nucleus (Fig. 9, D and F). Confocal microscopy confirms an intra-nuclear localization for MKP-4 (data not shown). Although we have not yet been able to test whether localization of heterologously expressed protein reflects exactly compartmentation of endogenous MKP-4, this pattern of subcellular localization is clearly different from other dual specificity phosphatases also expressed as epitope-tagged proteins (34,36,39,41,43,45).

MKP-4 Co-localization with Promyelocytic Leukemia (PML)
Protein-Punctate nuclear staining as seen for MKP-4 has also been reported for PAC-1 (34), although not for CL100, hVH-2, or hVH-3 (36,39), and may be a distinctive property of a subset of dual specificity phosphatases. Punctate nuclear localization is also reminiscent of the PML protein, which concentrates in discrete subnuclear regions known as nuclear bodies or PML oncogenic domains (67). To test directly for co-localization, we used a rabbit polyclonal antibody to PML and performed double immunocytochemistry on MKP-4-transfected COS-7 cells. This approach revealed that MKP-4 and PML display patterns of nuclear staining that are exactly overlapping (Fig. 10, A-C). Since PML has been reported as a phosphoprotein (68), this nuclear staining pattern indicates a potential role for MKP-4 as a direct or indirect regulator of kinases phosphorylating PML. Such a regulatory interaction is currently under investigation in our laboratories.
MKP-4 Chromosomal Localization-To determine the chromosomal localization of MKP-4, DNAs from human/rodent somatic cell hybrids were analyzed for the presence of the human MKP-4 gene by PCR. An MKP-4-specific human fragment of 200 bp was distinguishable from rodent-specific fragments. This human fragment was present in hybrids GM07299, MCP6BRA, 1aA9607ϩ, GM10478, THYB1.3, and HORL9, all of which retain the long arm of human chromosome X (Fig. 11). PCR analysis of DNAs from a panel of radiation hybrids (52) confirmed this, assigning the gene for MKP-4 to Xq28 (data not shown). Southern blot analysis of human male and female genomic DNAs using a fragment from the 3Ј-untranslated region of MKP-4 as a probe identified a single band (Fig. 12). A stronger signal was detected in the lanes containing female DNA, again indicating that the gene for MKP-4 is X-linked (Fig. 12). No other dual specificity phosphatase genes have been found to map to the Xq28 gene (69,70), 4 and, based on tissue expression in adult, there are no obvious candidate human diseases assigned to this region that may be caused by mutations in MKP-4.
In this paper we report the identification and initial characterization of MKP-4 as a novel dual specificity phosphatase. Although structurally homologous to other family members, including an extended active-site consensus as well as two N-terminal CH2 domains, MKP-4 is unique in terms of both function and distribution. Our studies indicate that MKP-4 inactivates MAP kinases with the selectivity ERK Ͼ p38 ϭ JNK/SAPK. This enzymatic specificity is similar to its close homologue MKP-3/PYST1 (ERK Ͼ Ͼ JNK/SAPK ϭ p38) (45,46), although distinct from hVH-5/M3-6 (JNK/SAPK ϭ p38 ϾϾϾ ERK) (46), MKP-2/hVH-2/TYP-1 (ERK ϭ JNK/SAPK Ͼ p38) (58), and PAC-1 (ERK ϭ p38 Ͼ JNK/SAPK) (58). In addition to blockade of acute MAP kinase activation by MKP-4, inhibition of chronic ERK activation by oncogenic p21 ras (G12V) could indicate a tumor suppressor function. MKP-4 displays a distinctive distribution insofar that, among a large range of cell and tissue types, mRNA was detected only in placenta, kidney, and embryonic liver. Its subcellular localization is also striking in that in three distinct cell types MKP-4 is clearly cytosolic with additional punctate nuclear staining observed in a subset of cells. These observations highlight the emergence of a growing family of dual specificity phosphatase possessing many distinctive properties indicative of a gene family performing specific functional roles in different cellular and physiological contexts.