MST4, a New Ste20-Related Kinase That Mediates Cell Growth and Transformation Via Modulating ERK Pathway

MST4, a new Ste20-related kinase that mediates cell growth and transformation via modulating ERK pathway

Jei-Liang Lin1,2, Hua-Chien Chen1, Hsin-I Fang1, Dan Robinson3, Hsing-Jien Kung3 and Hsiu-Ming Shih*,1,2

1Division of Molecular and Genomic Medicine, National Health Research Institutes, 128, Sec2, Yen-Chiu-Yuan RD, Taipei 11529, Taiwan; 2Graduate Institute of Life Sciences, National Defense Medical Center, Taipei, Taiwan; 3University of California at Davis Cancer Center, Sacramento, California, CA 95817, USA

In this study, we report the cloning and characterization kinase) (Feng et al., 1998; Lee and Elion, 1999), Ste7 of a novel human Ste20-related kinase that we (MAPK kinase) (Leberer et al., 1996) and FUS3/KSS1 designated MST4 (accession number AF231012). The (MAPK) (Feng et al., 1998; Schrick et al., 1997) in 416 amino acid full-length MST4 contains an amino- yeast. Several serine/threonine protein kinases related terminal kinase domain, which is highly homologous to to the Ste20 protein kinase have been recently MST3 and SOK, and a unique carboxy-terminal identi®ed in mammalian cells. Based on the relative domain. Northern blot analysis indicated that MST4 is location of the kinase catalytic domain and the ability highly expressed in placenta, thymus, and peripheral to bind small G-proteins, these mammalian Ste20 blood leukocytes. Wild-type but not kinase-dead MST4 related protein kinases can be broadly divided into can phosphorylate myelin basic protein in an in vitro two subfamilies. kinase assay. MST4 speci®cally activates ERK but not The ®rst subfamily contains a C-terminal catalytic JNK or p38 MAPK in transient transfected cells or in domain and an N-terminal Cdc42/Rac1 interactive stable cell lines. Overexpression of dominant negative binding (CRIB) motif (Osada et al., 1997). The yeast MEK1 or treatment with PD98059 abolishes MST4- Ste20 kinase and its mammalian homolog, p21- induced ERK activity, whereas dominant-negative Ras or activated kinase 1 (PAK1), belong to this class (Brown c-Raf-1 mutants failed to do so, indicating MST4 et al., 1996). The PAK kinases have been implicated in activates MEK1/ERK via a Ras/Raf-1 independent activating the stress-activated protein kinase (SAPK)/ pathway. HeLa and Phoenix cell lines overexpressing Jun N-terminal kinase (JNK) pathway as well as in wild-type, but not kinase-dead, MST4 exhibit increased regulating the actin cytoskeleton and cell mobility growth rate and form aggressive soft-agar colonies. (Frost et al., 1996; Pombo et al., 1996). The second These phenotypes can be inhibited by PD98059. These subfamily of Ste20s is represented by the germinal results provide the ®rst evidence that MST4 is center kinase (GCK) which contains an N-terminal biologically active in the activation of MEK/ERK catalytic domain with a C-terminal regulatory domain pathway and in mediating cell growth and transforma- (Dan et al., 2000; Fu et al., 1999). This class of kinases tion. Oncogene (2001) 20, 6559 ± 6569. lacks a recognizable CRIB motif and thus cannot be activated by Cdc42/Rac1. Like, PAK, many of the Keywords: Ste20 kinase; cell transformation; kinase GCK class members activate JNK. For instance, GCK, signaling GCK-related (GCKR) (Shi et al., 2000), hematopoietic progenitor kinase 1 (HPK1) (Hu et al., 1996; Oehrl et al., 1998; Zhou et al., 1999), kinase homologous to Introduction Ste20/Sps1p kinase (KHS) (Oehrl et al., 1998; Tung and Blenis, 1997), Mammalian Sterile Twenty-like The Sterile-20 (Ste20) protein kinase was ®rst identi®ed (MST1) (Graves et al., 1998), Nck-interacting kinase as a crucial factor involved in pheromone signaling (NIK) (Su et al., 1997), HPK/GCK-like kinase (HGK) pathway in Saccharomyces cerevisiae (Leberer et al., (Yao et al., 1999), Traf2- and Nck-interacting kinase 1992). Genetic epistasis analysis has placed Ste20 (TNIK) (Fu et al., 1999), Ste20-like kinase (SLK) kinase between the heterotrimeric G protein (Gilbreth (Ellinger-Ziegelbauer et al., 2000), Prostate-derived et al., 1996; Mosch et al., 1996), which is activated by Ste20-like Kinase (PSK) (Moore et al., 2000), Kinase pheromone exposure, and the MAPK cassette (Mosch From Chicken (KFC) (Yustein et al., 2000), Missha- et al., 1996; Raitt et al., 2000). Ste20 activates the pen/NIKs-related kinase (MINK) (Dan et al., 2000) MAPK cascade which includes Ste11 (MAPK kinase and NIK-like embryo speci®c kinase (NESK) (Nakano et al., 2000) are all known activators of JNK. By contrast, overexpression of the JNK-inhibitory kinase *Correspondence: H-M Shih; E-mail: [email protected] (JIK) (Tassi et al., 1999) leads to an inhibition of JNK Received 9 May 2001; revised 5 July 2001; accepted 9 July 2001 activation. Not all GCK family members are involved Induction of cell transformation by MST4 J-L Lin et al 6560 in modulating the activity of JNK/SAPK. MST3 and several tissues and is highly expressed in thymus, STE20/SPS1-related proline alanine-rich kinase placenta and peripheral blood leukocytes. We present (SPAK), for instance, respectively activate p42/44 evidence that MST4 speci®cally activates ERK, but not ERK and p38 MAPK, but not JNK (Johnston et al., JNK and p38 MAPK. This activation is via MEK1 but 2000; Zhou et al., 2000). There are also GCK members independent of Ras/Raf. Importantly, we provide including Ste20/oxidant stress response kinase-1 (SOK- evidence that MST4 is able to stimulate cell growth 1) (Pombo et al., 1996, 1997), lymphocyte-oriented and mediate anchorage-independent transformation of kinase (LOK) (Endo et al., 2000; Kuramochi et al., human epithelial cell lines. 1997), and Proline- and alanine-rich Ste20-related kinase (PASK) (Tsutsumi et al., 2000), which do not activate any of the known MAPK pathways. These Results observations suggest that Ste20 family of kinases have diverse functions and most, but not all, are involved in Cloning of MST4 the modulation of MAPK family kinases. While the signal pathways following kinase activa- The full-length MST4 cDNA was cloned by digital tions have been elucidated for many members of the cloning strategy we previously developed (Chen et al., Ste20 family kinases, we know much less about their 1998). Brie¯y, to search novel protein kinases, six 15- biological eects on cells. Several recent reports mer peptide sequences derived from human PAK1 implicate Ste20 family kinases in the apoptosis kinase, CDK kinase, ERK1 kinase, YAK1 kinase, pathway (for a review see Dan et al., 2001). This is PKA kinase and PKC kinase were designed and used perhaps understandable, since JNK and p38MAPK, as peptide probes to search the EST data base two kinases involved in apoptosis are downstream maintained in the National Center for Biotechnology targets of many of these Ste20 kinases (for a review Information. Several EST clones corresponding to this see Chen and Tan, 1999). It was shown that MST1, gene were identi®ed. RT ± PCR and DNA sequencing which activates JNK and p38MAPK, induces apop- analysis were performed to con®rm the existing cDNA tosis when overexpressed in cells or when cells were for the novel kinase. The missing 5' and 3' portions treated with Fas or staurosporine (Graves et al., 1998; were obtained by RACE. We used the total RNA from Lee et al., 1998). The activity of MST1 is enhanced by HR cells (a gastric cancer cell line) as a template to caspase 3 cleavage, and, in turn, the active MST1 is obtain the full-length MST4 clone since HR cells capable of activating caspase 3. Likewise, HPK1 and express a relatively high level of MST4 transcript (data SLK have also been reported to be activated by not shown). Sequencing of the MST4 RACE clone caspase 3 cleavage during Fas mediated apoptosis revealed a protein of 416 amino acid residues with a (Chen et al., 1999; Sabourin and Rudnicki, 1999; molecular mass of approximately 46 kDa (Figure 1a). Sabourin et al., 2000). The resulting kinase domain Blast search of GeneBank and EMBL database has an enhanced potential to activate JNK pathway identi®ed a serine/threonine kinase domain (Hanks (Chen et al., 1999; Sabourin and Rudnicki, 1999; and Lindberg, 1991; Hanks and Quinn, 1991) at the Sabourin et al., 2000). These observations also extend amino-terminal region (amino acid residues 18 ± 300) to the PAK subfamily of kinases; PAK2 is activated with homology to members of the Ste20-like kinase by caspase cleavage during Fas- and TNFa-induced family. The highest homology is with MST3 (Figure apoptosis, and the cleaved kinase domain induces 1b): MST4 shares 88% identity in the kinase domain formation of apoptotic bodies (Rudel and Bokoch, and 66% overall identity. Figure 1c showed a 1997). PAK1, on the other hand, has the potential to dendrogram illustrating the evolutionary relationship prevent apoptosis by phosphorylating Bad (Schur- between MST4 and other members of the Ste20-like mann et al., 2000). The above ®ndings suggest a role kinases. The genomic sequence comparison suggests of Ste20 family kinases in the modulation of apoptosis that the MST4 gene is located on chromosome Xq25- pathway. Relatively little is known about the involve- 26.3 region with at least 11 exons. ment of Ste20 kinases in cellular growth and transformation. Recently, Vadlamudi et al. (2000) Expression pattern of MST4 mRNA reported that activated PAK1 induces anchorage- independent growth of human epithelial breast cancer The expression pattern of MST4 transcript was cells and Yustein et al. (2000) showed chicken embryo examined by human multi-tissue Northern blot. As ®broblasts overexpressing KFC, a GCK-subfamily shown in Figure 2, a major band of 3.6 kilobase pairs kinase, exhibit enhanced growth rate. In both cases, was detected. MST4 is ubiquitously expressed, with the activation of MAPK family kinases accompanies higher levels of message detected in placenta, thymus the growth stimulation. and peripheral blood leukocytes. By contrast, the In the present study, we cloned and characterized a expression level of MST4 in brain, heart, skeletal human protein kinase representing a new member of muscle and colon is low. MST4 expression was also the Ste20-related kinase family. Database searches detected in cancer cell lines including HeLa (cervical revealed that this novel kinase has the highest cancer cell line), HR (gastric cancer cell line), SK-Hep1 homology to MST3 kinase. Thus, we designate this (hepatoma cell line) and Huh7 (Hepatoma cell line) cell kinase, MST4. MST4 mRNA can be detected in lines via RT ± PCR (data not shown).

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6561

Figure 2 Northern blot analysis of MST4 transcript in human tissues. Top panel, MST4 mRNA was detected as a 3.6 kb message by Northern blot analysis (see Materials and methods). Bottom panel, the same ®lter was also subjected to Northern blot analysis with g-actin as a probe which demonstrated equal loading of RNA from dierent tissues

the antibody was tested by Western blot analysis of MST4 immunoprecipitated with the T7 tagged antibody. MST4 anti-serum, but not preimmune serum, recognized a speci®c band (*46 kDa) which comi- grated with a band detected by anti-T7 antibody (Figure 3a, lanes 2 and 4), attesting to the speci®city of the antibody. Using this antibody, we examined the protein expression pattern of MST4 among various cancer cell lines. As shown in Figure 3b, MST4 is highly expressed in SK-Hep1, Jurkat and HR cells. Lower expression was seen in HeLa and no expression was detected in Phoenix (a 293 T derived cell line), and hepatoma cell lines, Tong, HA22T and PLC5. We next performed immunostaining to examine the subcellular localization of MST4 protein. As illustrated in Figure 4 (left column) overexpressed T7-tagged MST4 protein was detected in cytoplasm of Phoenix cells by MST4 antibody. Note that parental, untrans- fected cell did not give any signal in this staining, indicating the speci®city of MST4 antibody. Since HR cells express a higher level of MST4, we also examined the distribution of endogenous MST4. As shown in Figure 4 (right column) the majority of the MST4 immunostaining is again localized in cytoplasm with Figure 1 Amino acid sequence comparison of MST4. (a) MST4 some weak staining in the nuclei. Thus, MST4 seems to is a 46 kDa protein composed by 416 putative amino acids and contains serine/threonine kinase conserved catalytic domain be primarily localized in the cytoplasm. (18*300 amino acids), underline. (b) MST4 amino acids We then sought to determine whether MST4 has sequences share 88% identity, bold letter, with MST3 protein in intrinsic kinase activity. Endogenous MST4 was kinase catalytic domain (18*299 amino acids). (c) Phylogenetic immunoprecipitated from HeLa cells with MST4 tree of Ste20-related kinases. The phylogenetic tree was generated antibody and subjected to an in vitro kinase assay by the Clustal W program and MST4, bold letter, might be classi®ed into GCK subfamily with MST3 and SOK using myelin basic protein (MBP) as a substrate. Cell lysates immunoprecipitated by MST4 antibody, but not by preimmune serum, were able to phosphorylate MBP (Figure 5a), indicating that Identification and characterization of the human MST4 MST4 is associated with a kinase activity. To more kinase de®nitely demonstrate that the kinase activity comes To generate antibody against MST4, we used GST from MST4 itself and not from an associated kinase, fusion protein containing the carboxyl region of MST4 a kinase-dead mutant T7-MST4 KR was con- as an antigen to immunize rabbits. The speci®city of structed. T7-MST4 KR has a T7 tag at its amino

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6562

Figure 3 Western blot analysis of MST4 expression. (a) speci®city of anti-MST4 antiserum. T7-MST4 expression vector Figure 4 Subcellular localization of MST4. Immuno¯uorescence was transfected to Phoenix cells. Following 48 h, 500 mg protein analysis was performed with T7-MST4 transfected Phoenix cells from each cell lysate was subjected to immunoprecipitation (IP) (left column) and non-transfected HR cell line (right column). assay and Western blot (WB) analysis with antibodies or serum as Phoenix cells were transfected with T7-MST4 expression vector. indicated (PI: preimmune serum). Lane 1 represents non- Following 48 h, cells were ®xed, then incubated with MST4 transfected Phoenix cell lysates. Lanes 2 ± 4 represent Phoenix antibody, following by adding the secondary antibody conjuga- cell lysate overexpressing T7-MST4. Arrow indicates the position tion with TRITC. TRITC (red color) indicates the distribution of of T7-MST4. (b) endogenous MST4 expression analysis. Twenty- T7-MST4 and endogenous MST4. Cell morphology is shown by ®ve mg of total cell lysate from dierent cancer cell lines, Jurkat, phase image. DAPI staining (blue) shows nuclei HeLa, hepatoma (Tong, HA22T, PLC5, and SK-Hep1) and HR (gastric cancer), was subjected to Western blot analysis with MST4 antibody. To show the comparable amount of protein input in each lane, the cell lysates were also immunoblotted with tion, indicating that the carboxyl domain is not anti-a-actin antibody critical for its enzymatic activity.

MST4 modulates ERK kinases activity terminus and is defective in ATP binding due to the conversion of lysine residue 53 to arginine. Cells To explore whether MST4 is involved in regulating transfected with this kinase-dead mutant were known MAPK pathways, we tested the eect of subjected to the immunoprecipitated kinase assay as overexpressed MST4 on the activation of JNK, p38, before. Wild-type T7-tagged MST4 was used as a and ERK pathways. As shown in Figure 6a, over- positive control. Cell lysates carrying wild-type T7- expressed MST4 failed to activate JNK activity, based tagged MST4 eectively phosphorylated MBP in on in vitro kinase assay of HA-JNK using GST-c-jun vitro, whereas no phosphorylation of MBP was as a substrate. Positive control was provided by an observed with cell extracts containing the kinase- constitutively active form of MEKK1. Similarly, MST4 dead mutant (Figure 5c (bottom panel) lanes 2 and overexpression showed no signi®cant eect on the 4). These ®ndings provide strong evidence that the activation of p38 MAPK (Figure 6b). By contrast, activity detected from the wild-type MST4 is derived MST4 signi®cantly increased the ERK activity, from its intrinsic kinase and not to a co-precipitated although the level is less than that conferred by kinase. To further examine whether the carboxyl constitutively active form of MEK1 (Figure 6c, lanes region of MST4 aects its kinase activity, a deletion 2 and 3). The activation of ERK by MST4 is kinase mutant T7-MST4DC (deletion of 301 ± 416 amino activity-dependent, since no activation was observed by acids) was tested in the in vitro kinase assay. As the MST4 KR mutant. These results suggest that illustrated in Figure 5c (lane 3), deletion of this overexpressed MST4 can speci®cally activate ERK region did not signi®cantly alter MBP phosphoryla- kinase, but not JNK or p38 MAPK.

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6563

Figure 6 Activation of ERK MAPK by MST4. Phoenix cells were transfected with wild-type MST4 (WT) (lane 3) or kinase dead mutant (KR) (lane 4) and expression vector of HA-JNK1 (a), Flag-p38 (b) and His-ERK (c). The cell lysates were separated by SDS ± PAGE and immunoblotted with anti-HA, anti-Flag, and anti-ERK antibodies (middle panel). MST4 expression level indicated by anti-T7 mAb (lower panel). Activated JNK1 was indicated by in vitro kinase assay with immunoprecipitated HA- Figure 5 Kinase activity of MST4. (a) kinase activity of JNK using GST-c-jun as a substrate (a, top panel). Activated p38 endogenous MST4. HeLa cell lysate was precipitated by either (b) and ERKs (c) were indicated by immunoblot with anti- preimmune serum or MST4 antibody. The immunoprecipitated phospho-p38, or anti-phospho-ERKs polyclonal antibodies (b and complex was divided into two portions. One was for Western blot c, top panels). As a positive control for activation of JNK1, p38, analysis with anti-MST4 antiserum (top panel). The other portion and ERK pathways, Phoenix cells were co-transfected with was subjected to in vitro kinase assay using [g-32P]ATP and MBP expression vector of MEKK1, MEK3, and MEK1, respectively as a substrate (Materials and methods). The reaction products (lane 2) were separated by SDS ± PAGE and subjected to autoradiography (bottom panel). (b) a schematic diagram of MST4 and its mutants. MST4 wild-type (WT) and its various mutants including c-terminus deletion (WTDC), kinase dead mutant (K53R), and Phoenix cell lines stably overexpressing MST4 or its K53R with c-terminus deletion (KRDC) were fused to a T7-tag kinase-dead KR mutant. Consistent with the above expression vector. K53R is a kinase inactive mutant, in which the results, overexpressed MST4 conferred sixfold activa- critical lysine (amino acid 53) required for ATP-binding has been changed to an arginine. The kinase domain is draw as black box tion of the reporter gene, whereas MST4 KR mutant and the T7 tag are indicated by dot box. (c) kinase activity of only yielded a twofold induction (Figure 7a, lanes 2, MST4 and its mutants. Phoenix cells were transfected with 7 and 11). This activation was abolished by various MST4 constructions and cell lysate were subjected to PD98059, an inhibitor of MEK1, in a dose-dependent immunoprecipitation assay and Western blot analysis with anti- T7 antibody (upper panel). After immunoprecipitation, in vitro fashion, but not by SB202190, a p38 inhibitor (lanes kinase assay was performed (lower panel) using MBP as a 7 ± 10). These ®ndings suggest that MST4-ERK substrate. The resulting reactions were subjected to SDS ± PAGE pathway is mediated by MEK1. To further sub- and autoradiography stantiate this result, a dominant negative (DN) MEK1 was transfected into MST4 stable cells and the ERK-activation of ELK1 was measured. In accordance with the result of PD98059 treatment, MST4 activates ERK via MEK1 but not Ras/Raf MEK1-DN can block the ERK activity induced by To explore the pathway by which MST4 activates MST4 (Figure 7b, lanes 10 and 13). To study ERK, a reporter gene assay of ERK dependent Elk1 whether MEK1/ERK activation by MST4 is activation was used. A reporter gene driven by Elk1- mediated through Ras and Raf, dominant negative dependent promoter was transiently transfected into mutant of Ras or Raf was included in this assay;

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6564 that MST4 activates MEK1-ERK signaling cascade in a Ras/Raf-independent fashion.

Overexpression of MST4 induces cellular growth To explore the biological eect of MST4 on cell growth and transformation, we established stable clones of Phoenix and HeLa cells overexpressing MST4 wild-type or KR mutant. We chose Phoenix and HeLa cells for studies because of their undetect- able or low MST4 expression level. To avoid clonal eect of stable cells, studies were carried out with a pool of more than 20 single clones, shown to express T7-tagged MST4 wild-type or KR mutant (data not shown). Immunocomplex kinase assay revealed that the precipitated MST4 from cells expressing wild-type kinase has the ability to phosphorylate MBP in vitro (Figure 8a, lane 3), consistent with that of the transient expression studies. In addition, the kinase activity of wild-type MST4 could be further enhanced upon serum stimulation (lane 7), indicating that MST4 may mediate serum-induced cellular processes such as cellular growth. To explore this possibility, the growth of these cells was compared to that of the parental cell line or cell lines harboring the kinase-dead mutant. As shown in Figure 8b, overexpression of the wild-type MST4 leads to enhanced growth of the Phoenix cells, detectable by days 3 and 5. The KR mutants failed to do so. This is especially evident, when cells were cultured in low serum media (1%). This observation is interesting, as most of the other Ste-20 family kinases characterized prior to the present study do not induce cell growth. Our observation is however consistent with the ®nding that constitutive activation of MEK1 results in cell proliferation (Greulich and Erikson, 1998).

Overexpression of MST4 confers cellular transformation Figure 7 MST4 activates Erk MAPK in a Ras/Raf-independent We next explored whether MST4 can potentiate cell 5 manner. (a) 3.5610 cells/well of each Phoenix stable cell line transformation. MST4 wild-type and KR mutant including pcDNA-T7 empty vector, T7-MST4 wild-type and KR mutant were transfected with CMV-b-galactosidase and GAL- stable clones were subjected to anchorage-independent luciferase reporter plasmid and various expression vectors as growth (AIG) assay. Remarkably, overexpressed MST4 indicated by lipofectamine. After 24 h, cell lysates were collected, in Phoenix cells resulted in increased number as well as and the luciferase assay was performed (see Materials and the size of the colonies in soft agar (Figure 9a). The methods). The relative activity was normalized to relative luciferase activity of vector stable cell transfected with Gal4 colony number was counted and summarized in Figure DBD (lane 1). ELK1 represents the Gal4-ELK1 fusion expression 9b. The MST4 induced colony formation in AIG assay vector. Vector stable clone, wild-type MST4, and KR mutant can be blocked by PD98059, but not SB20219 (Figure stable clone were treated with MEK1 speci®c inhibitor PD98059 9b). These ®ndings reinforce the view that MST4 exerts (2 or 20 mM) (lanes 4, 5, 8, 9, 12 and 13) or p38 MAPK inhibitor its eect through the activation of ERK signaling SB20219 (10 mM) (lanes 6, 10 and 14). As a positive control for ERK activation, vector stable clone was co-transfected with pathway. The similar ®ndings were obtained in HeLa MEK1 (100 ng) (lane 3). B, MST4 stable clones were co- MST4 stable clones (Figure 9c), indicating that the transfected with 1 mg of dominant negative Ras (Ras-DN) (lanes biological eect of MST4 is not cell type speci®c. 5 and 11), Raf-DN (lanes 6 and 12), and MEK1-DN (lanes 7 and 13). The reporter assays were performed after EGF (10 ng/ml) treated for 12 h (lanes 4 ± 7). The experiments have been repeated at least four times Discussion

By digital cloning, we have identi®ed a novel Ste20- both failed to eectively block the MST4 mediated related kinase, which, when overexpressed, can activate ERK activation. At the same time, both mutants ERK pathway, resulting in cell growth and transforma- diminished the EGF-induced ERK activation (Figure tion. We also showed that MST4 speci®cally activates 7b, lanes 4 ± 6). Taken together, these ®ndings suggest ERK, but not JNK and p38 (Figure 10). This

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6565 stimulation in Jurkat T cells and several proteins (Nck, LAT, and Crk) may be involved in linking HPK1 activation to T-cell (Ling et al., 2001). MST4, likewise, may link a cell surface receptor to MEK1/ERK. To explore this possibility, we have surveyed several growth factors that might stimulate MST4 kinase activity. Preliminary data indicate that one of the agonists which activate MST4 is EGF (unpublished data). In addition to the activation of ERK pathway, we showed that overexpression of wild-type MST4 also promotes cell proliferation and anchorage-independent growth. While kinase-dead MST4 failed to completely block these eects, it diminished the ELK-mediated reporter gene expression and anchorage-independent growth of cells (Figures 7a and 9). We do not know why MST4 kinase-dead mutant does not fully reverse the transformation phenotypes, it is possible the biological functions of MST4 are not all kinase- dependent. One pertinent precedent is TNIK kinase (Fu et al., 1999), whose kinase-dead mutant is still capable of activating downstream signaling pathway. We found that MST4 is capable of dimerizing with itself (unpublished data) and it is possible that kinase dead mutant can activate endogenous MST4 or downstream signaling molecules through protein ± Figure 8 Overexpressed MST4 enhances cell growth. (a) Serum induces the kinase activity of MST4 stable clones. Cell lysates protein interactions. from serum-starvation for 24 h or serum-stimulation for 30 min Previous studies by others showed that expression of of Phoenix cells, empty vector stable clone, MST4 wild-type and a constitutively activated MEK1 induce in NIH3T3 KR mutant stable clones were subjected to immunoprecipitation cells morphological changes, anchorage-independent assay with anti-T7 antibody and followed by in vitro kinase assay. Cell lysates were also separated by SDS ± PAGE and immuno- growth and accelerated S phase entry (Blalock et al., blotted with anti-T7 antibody (bottom panel). (b) Each stable 2000; Brunet et al., 1994; Greulich and Erikson, 1998; clone was seeded 500 cells into 6-cm dish in culture media Ljungdahl et al., 1998; Rimerman et al., 2000; containing dierent serum concentrations (1 or 10%) as indicated. Schramek et al., 1997), phenotypes similar to cells Cells were harvested on days 1, 3 and 5 followed by Trypan blue overexpressing MST4. Likewise, constitutively active staining for viable counting. The cell count is based on four independent experiments ERK was shown to be sucient to elicit cell transformation (Greulich et al., 1996; Huang et al., 1999; Plattner et al., 1999; Verheijen et al., 1999). activation requires MEK1, as both PD98059 and These ®ndings are in good agreement with our results MEK1 DN mutant block the ERK activation induced that MEK1/ERK activation by MST4 is the underlying by MST4. The molecular mechanism whereby MST4 cause of cellular transformation induced by MST4. The transduces the signal to MEK1 is currently unknown. downstream target genes mediating the cell cycle Based on the studies using Ras or Raf DN mutants, we progression and transformation are currently un- know that Ras and Raf are not involved. Also not known, although activation of MEK/ERK pathway is clear is the upstream signals that may modulate MST4 linked to cell cycle progression by transcriptional activity. In this study, we used overexpressed MST4, elevation of cyclin D1 levels in G1 phase (Cheng et which appears to exhibit kinase activity without al., 1998; Greulich and Erikson, 1998; Lavoie et al., additional ligand activation. It is possible that 1996). endogenous MST4 may require an upstream signal Our ®ndings that overexpressed MST4 can reduce for activation. Indeed, the kinase activity of MST4 can the serum requirement for cell growth and anchorage be further increased upon serum stimulation. Accumu- independent growth may shed some lights on the lating evidence indicate that Ste20 kinases may be biological roles of MST4 in cell proliferation and involved in transmitting signals from cell surface cancer formation. MST4 is highly expressed in poorly receptors to the MAPK cascades. For instance, GCK dierentiated hepatoma cell SK-Hep1 but not ex- functions as a molecular bridge linking TRAF2 and pressed in relatively dierentiated hepatoma cells TNF type-1 receptor signaling complex to MEKK1- (Tong, HA22T and PLC5). While not conclusive, these SEK1-JNK signaling pathway (Nakano et al., 2000). ®ndings suggest that MST4 expression may be related Nck interacting kinase (NIK) has been shown to to the dierentiation state of the cancer cells. Further couple Eph receptors to JNK pathway via Nck (Becker investigation is required to support this notion. To et al., 2000). Recent study demonstrated that kinase explore the correlation between MST4 expression and activity of HPK1 can be induced upon TCR/CD3 cancer formation, we have examined its expression

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6566

Figure 9 Overexpressed MST4 potentiates anchorage independent growth. (a). Phoenix cells, empty vector stable clone, MST4 wild-type and KR mutant stable clones were subjected to anchorage-independent growth assay. Ten thousand cells were embedded into medium containing 0.3% agarose and cultured in regular medium. The culture medium was changed every 3 days. After 3 weeks, anchorage independent growth colonies were observed under microscope (top panel). Bar length represents 100 mm. The culture plates from each stable clone were further stained with 0.1% crystal violet dye (bottom panel) and then counted for larger than 0.1 mm by b and c, during the anchorage independent growth assay of Phoenix (b) or HeLa (c) stable clones, the culture cells were treated with PD98059 (2 mM) or SB20219 (2 mM) when culture medium was changed. After 3 weeks, the amount of colonies was counted and plotted to the barograph. The colony number is based on three independent experiments

level in several gastric tumor tissues and colon cancer activity is by caspase cleavage. MST1, MST2, HPK1, tissues (cancer versus adjacent normal tissues from the SLK and PAK2 have been reported to be cleaved by same patient). The preliminary results indicate that caspases (Arnold et al., 2001; Graves et al., 1998; Lee MST4 has elevated expression in cancer tissue et al., 1998, 2001; Sabourin et al., 2000). The caspase (unpublished data). cleavage leads to increased kinase activity with the Among the Ste20 family kinases, MST3 shares the consequent activation of JNK (Arnold et al., 2001; highest homology with MST4. While the biological Graves et al., 1998, 2001) or p38 MAPK (Graves et al., functions of MST3 remains to be de®ned, MST3 was 1998; Kakeya et al., 1998) to promote apoptosis. shown to activate ERK in HEK293 cells, consistent Inspection of amino acid sequences of MST4 reveals a with the data reported here for MST4 (Zhou et al., putative caspase 3-like protease cleavage site 2000). It should be noted that while the kinase domain (DESD305S) which is adjacent to the carboxyl end of between MST3 and 4 share a high degree of homology, kinase domain. At least in vitro, we failed to the carboxyl terminal domains for these two kinases demonstrate cleavage of MST4 by caspase 3, 6 or 7 are entirely dierent. For GCK class of Ste20 kinases, (unpublished data). The mode of regulation of MST4 the carboxyl-terminal domain often serves a regulatory thus remains to be determined. function in kinase activity. Thus, removal of the After completion of this manuscript, we saw an in carboxyl-terminal domain of KFC and MST1 un- press report of Qian et al. (2001) who described the leashes the kinase activity (Creasy et al., 1996; Yustein identi®cation and biochemical characterizations of et al., 2000). This however does not seem to be the case MST4. Their results are generally in good agreement for MST4, as the C-terminal truncated MST4 displays with ours. Our report, however diers from theirs by at a similar level of kinase activity as the wild-type. least two important aspects. First, Qian et al. (2001) Another common means of regulating the Ste20 kinase failed to detect the activation of the MEK-ERK

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6567 peptide sequences derived from human PAK1 kinase, CDK kinase, ERK1 kinase, YAK1 kinase, PKA kinase and PKC kinase were designed and used as peptide probes to search the EST data base maintained in the National Center for Biotechnology Information. Several EST clones which indicate the same gene by their overlapped nucleotide sequences were assembled. RT ± PCR and DNA sequencing analysis were performed to con®rm the existing cDNA for the novel kinase. Missing 5' and 3' portions were obtained by RACE. We used the total RNA from HR cells to get the full- length MST4 since HR cells conferred a higher content of MST4 messenger. Total RNA from HR cells was isolated using Trizol (Life Technologies, Inc.) and followed by reverse-transcribed to cDNA using high ®delity Taq poly- merase (Roche) primed with oligo(dT)17. PCR was performed using two primers 5'-GCGGTACCATGGCTA- GCATGACTGGTGGACAGCAAATGGGTTCCGAGGG- CTCTGATTCGGAATCTACC and 5'-GGGGTACCGGG- GATCTAGATCTTAAAAGCAAAACGGTG. The full- length MST4 cDNA was subcloned into pcDNA 3.0 expression vector with T7-tag at its N-terminal. T7-tag MST4 DC mutant (1 ± 300 amino acids containing kinase domain) and MST4 K53R kinase dead mutant were generated by PCR and site directed mutagenesis, respectively.

Northern blot analysis cDNAs of MST4 (nucleotides 900 ± 1251) and the human g- Figure 10 Summary of MST4 signaling pathway. A scheme 32 diagram is provided to illustrate the MST4 in activation of actin were labeled with a- P-dCTP using a random priming MEK1/ERK pathway, resulting in growth induction and cellular kit (Amersham). The labeled probes were used in Northern transformation. The activation of MEK-1/ERK by MST4 is Ras/ blot analysis with commercial human multiple tissue North- Raf independent. MST4 fails to activate either JNK or p38 ern blot (Clontech). Hybridization was performed as MAPK pathway recommended by the manufacturer. After prehybridization, hybridization and highly stringent washing, the membrane was exposed to phosphoimager and analysed by Bio-imaging pathway by MST4. This is most likely due to the Analyser BAS1500 (Fuji). dierent cell lines used, although in our hands, two dierent cell lines understudy showed ERK activation In vitro kinase assay by MST4. Second, we ascribe the ®rst biological The expression vectors of T7 MST4 wild-type (WT), KR, function for MST4 and show its involvement in cell WTDC, and KRDC were transfected into Phoenix cells by growth and transformation. lipofectamine (GIFCO). After 48 h, cells were lysed in lysis buer containing 20 mM Tris-HCl (pH 8.0), 0.5% Nonidet P- 40, 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 2 mM DTT, 0.5 mM phenylmethylsulfony¯uoride (PMSF), 20 mM b- glycerolphosphate, 1 m sodium orthovanadate and 1 mg/ Materials and methods M ml leupeptin. Cell lysates were precleared with protein A- Sepharose (Calbiochem) for 1 h and then incubated with Cell cultures anti-T7 monoclonal antibody (Novagene) for 2 h at 48C. The Phoenix, HeLa, Jurkat, and several hepatoma cell lines immune complexes were precipitated with protein A-sephar- (Tong, PLC5, HA22T, and SK-Hep1) were cultured in ose, washed three times with lysis buer, and then washed Dulbecco's modi®ed Eagle's medium (DMEM) (Life Tech- three times with kinase buer containing 20 mM HEPES nologies, Inc.). HR cells, a gastric cancer cell line, were pH 7.5, 20 mM b-glycerophosphate, 10 mM MgCl2,1mM cultured in RPMI 1640 culture medium (Life Technologies, NaF, 1 mM NaVO4, 0.5 mM dithiothreitol. To perform the Inc.). Both media were supplemented with 10% FBS, kinase reaction, the immunoprecipitates were incubated with 16nonessential amino acids, 2 mM glutamine, 100 U/ml 50 mM cold ATP and [g-32P]ATP with substrate, myelin basic penicillin and 100 mg/ml streptomycin. Phoenix cells were protein (MBP) for 30 min at 258C. The reaction products also transfected with dierent human MST4 constructs and were subjected to 10% SDS-polyacrylamide gel electrophor- stable transfectants were isolated by selection in G418 (1 mg/ esis. The gels were ®xed, dried, and autoradiographed. ml) and regularly maintained in selected medium containing G418 and replaced medium every 3 ± 4 days. Preparation of polyclonal antibody The GST-MST4-C plasmid was constructed by inserting a Cloning of MST4- and plasmid constructs MST4 DNA fragment coding amino residues from 302 ± 416 The full-length MST4 cDNA was cloned by digital cloning into the pGEX4T-3 expression vector (Pharmacia). The GST- strategy which has been described (Chen et al., 1998). For fusion proteins was expressed in Escherichia coli BL21 in the novel EST clones homologous to protein kinase, six 15-mer presence of 0.5 mM isopropyl-1-thio-b-galactopyranoside,

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6568 puri®ed and used to immunize rabbits. Polyclonal antibody (New England Bio-Labs.), following by Western blot analysis was anity-puri®ed using GST-MST4-C proteins bound to with p42/44 MAPK antibody. CNBr-activated Sepharose 4B (Pharmacia). The antibody was used in detecting MST4 in Western blotting and Soft agar colony formation assay immunostaining analysis. Phoenix cells, HeLa cells, and cells stably expressing MST4 wild-type or kinase dead proteins were trypsinized, diluted in Immunohistochemical staining 0.3% of top agar and plated onto 60-mm dishes with 0.5% of To examine the subcellular localization of MST4, the bottom agar. Ten thousand cells were seeded in each plate. polyclonal antibody against GST-MST4C was utilized in Plates were incubated at 378C for 3 weeks. Colonies were the immunostaining of the T7-MST4-transfected cells and visualized by staining with 0.05% p-iodonitrotetrazolium non-transfected HR cells (for endogenous MST4). In brief, violet dye and then colonies larger than 0.1 mm were counted T7-MST4 transfected Phoenix cells or HR cells were plated by AIS image analysis system (Imaging Res. Inc.). Each on cover slide glasses and cultured in medium. After 48 h experiment was repeated four times. The average and the transfection, cells were ®xed for 10 min in 4% paraformalde- standard error of the mean are shown in a barograph. hyde and then incubated with MST4 polyclonal antibody, followed by adding the secondary antibody conjugation with Luciferase reporter assay TRITC. DAPI was used to stain nucleus. The results were photographed by a ¯uorescence microscope. Phoenix and various MST4 stable cells were transfected with CMV-b-galactosidase, 56gal-luciferase reporter gene and expression vector of Gal4-ELK1, Gal4-DBD, constitutively Assays of the JNK/SAPK, p38 and p42/44 MAP kinases active MEKK1, MEK1 and MEK3 (from Stratagene) as activation indicated in Figures. Cells were transfected for 20 h and then To examine the MAPK activation in response to the serum starvation for 24 h followed by harvested cell lysates overexpression of MST4, Phoenix cells were cotransfected with lysis buer as described above. The lysates were with either wild-type or kinase dead mutant T7-MST4 along collected and the relative luciferase activity (®re¯y luciferase with dierent MAPKs (HA-JNK, Flag-p38, His-ERK). The for reporter and b-galactosidase activity for normalization of JNK activity was analysed by in vitro kinase assay. Brie¯y, transfection eciency) was measured as manufacturer's cell lysates containing HA-JNK were immunoprecipitated instructions (LucLite, Packard). Each experiment was with anti-HA antibody (BAbCO) and followed by in vitro repeated at least three times. kinase assay with GST-jun (New England Bio-lab) as a substrate. For p38 MAPK activation, cell lysates containing Flag-p38 were immunoprecipitated with anti-Flag antibody Acknowledgments (M2) and subjected into Western analysis. The activated p38 We thank Dr M-Z Lai for pFlag-p38 and pHis-ERK was detected by a speci®c phospho-p38 MAPK antibody plasmid constructs, Dr Y-S Jou for all the hepatoma cell which can recognize the phosphorylated threonine 180 and lines and Dr W-C Lin for HR cell. This work was tyrosine 182 of p38 MAPK (New England Bio-Labs.). For supported by the National Health Research Institutes (to p42/44 MAPK activation, cell lysates were immunoprecipi- H-M Shih) and NIH grants (to HJ Kung). Unpublished tated with a speci®c antibody against phospho-p42/44 MAPK data: J-L Lin, H-I Fang and H-M Shih.

References

Arnold R, Liou J, Drexler HC, Weiss A and Kiefer F. (2001). Dan I, Watanabe NM, Kobayashi T, Yamashita-Suzuki K, J. Biol. Chem., 29, 29. Fukagaya Y, Kajikawa E, Kimura WK, Nakashima TM, Becker E, Huynh-Do U, Holland S, Pawson T, Daniel TO Matsumoto K, Ninomiya-Tsuji J and Kusumi A. (2000). and Skolnik EY. (2000). Mol. Cell. Biol., 20, 1537 ± 1545. FEBS Lett., 469, 19 ± 23. Blalock WL, Pearce M, Steelman LS, Franklin RA, Dan I, Watanabe NM and Kusumi A. (2001). Trends Cell McCarthy SA, Cherwinski H, McMahon M and McCu- Biol., 11, 220 ± 230. brey JA. (2000). Oncogene, 19, 526 ± 536. Ellinger-Ziegelbauer H, Karasuyama H, Yamada E, Tsuji- Brown JL, Stowers L, Baer M, Trejo J, Coughlin S and kawa K, Todokoro K and Nishida E. (2000). Genes Cells, Chant J. (1996). Curr. Biol., 6, 598 ± 605. 5, 491 ± 498. Brunet A, Pages G and Pouyssegur J. (1994). Oncogene, 9, Endo J, Toyama-Sorimachi N, Taya C, Kuramochi- 3379 ± 3387. Miyagawa S, Nagata K, Kuida K, Takashi T, Yonekawa Chen HC, Kung HJ and Robinson D. (1998). J. Biomed. Sci., H, Yoshizawa Y, Miyasaka N and Karasuyama H. (2000). 5, 86 ± 92. FEBS Lett., 468, 234 ± 238. Chen YR, Meyer CF, Ahmed B, Yao Z and Tan TH. (1999). Feng Y, Song LY, Kincaid E, Mahanty SK and Elion EA. Oncogene, 18, 7370 ± 7377. (1998). Curr. Biol., 8, 267 ± 278. Chen YR and Tan TH. (1999). Gene Ther. Molec. Biol., 4, Frost JA, Xu S, Hutchison MR, Marcus S and Cobb MH. 83 ± 98. (1996). Mol. Cell. Biol., 16, 3707 ± 3713. ChengM,SexlV,SherrCJandRousselMF.(1998).Proc. Fu CA, Shen M, Huang BC, Lasaga J, Payan DG and Luo Y. Natl. Acad. Sci. USA, 95, 1091 ± 1096. (1999). J. Biol. Chem., 274, 30729 ± 30737. Creasy CL, Ambrose DM and Cherno J. (1996). J. Biol. Chem., 271, 21049 ± 21053.

Oncogene Induction of cell transformation by MST4 J-L Lin et al 6569 Gilbreth M, Yang P, Wang D, Frost J, Polverino A, Cobb Oehrl W, Kardinal C, Ruf S, Adermann K, Groen J, Feng MH and Marcus S. (1996). Proc. Natl. Acad. Sci. USA, 93, GS, Blenis J, Tan TH and Feller SM. (1998). Oncogene, 17, 13802 ± 13807. 1893 ± 1901. Graves JD, Draves KE, Gotoh Y, Krebs EG and Clark EA. Osada S, Izawa M, Koyama T, Hirai S and Ohno S. (1997). (2001). J. Biol. Chem., 13, 13. FEBS Lett, 404, 227 ± 233. Graves JD, Gotoh Y, Draves KE, Ambrose D, Han DK, Plattner R, Gupta S, Khosravi-Far R, Sato KY, Perucho M, Wright M, Cherno J, Clark EA and Krebs EG. (1998). Der CJ and Stanbridge EJ. (1999). Oncogene, 18, 1807 ± EMBO J., 17, 2224 ± 2234. 1817. Greulich H and Erikson RL. (1998). J. Biol. Chem., 273, Pombo CM, Bonvetre JV, Molnar A, Kyriakis J and Force 13280 ± 13288. T. (1996). EMBO J., 15, 4537 ± 4546. Greulich H, Reichman C and Hanafusa H. (1996). Oncogene, Pombo CM, Tsujita T, Kyriakis JM, Bonventre JV and 12, 1689 ± 1695. Force T. (1997). J. Biol. Chem., 272, 29372 ± 29379. Hanks SK and Lindberg RA. (1991). Methods Enzymol., 200, Qian Z, Lin C, Espinosa R, LeBeau MM and Rosner MR. 525 ± 532. (2001). J. Biol. Chem., 16, 16. Hanks SK and Quinn AM. (1991). Methods Enzymol., 200, Raitt DC, Posas F and Saito H. (2000). EMBO J., 19, 4623 ± 38 ± 62. 4631. Hu MC, Qiu WR, Wang X, Meyer CF and Tan TH. (1996). Rimerman RA, Gellert-Randleman A and Diehl JA. (2000). Genes Dev., 10, 2251 ± 2264. J. Biol. Chem., 275, 14736 ± 14742. Huang C, Ma WY, Li J, Goranson A and Dong Z. (1999). J. Rudel T and Bokoch GM. (1997). Science, 276, 1571 ± 1574. Biol. Chem., 274, 14595 ± 14601. Sabourin LA and Rudnicki MA. (1999). Oncogene, 18, Johnston AM, Naselli G, Gonez LJ, Martin RM, Harrison 7566 ± 7575. LC and DeAizpurua HJ. (2000). Oncogene, 19, 4290 ± Sabourin LA, Tamai K, Seale P, Wagner J and Rudnicki 4297. MA. (2000). Mol. Cell. Biol., 20, 684 ± 696. Kakeya H, Onose R and Osada H. (1998). Cancer Res., 58, Schramek H, Feifel E, Healy E and Pollack V. (1997). J. Biol. 4888 ± 4894. Chem., 272, 11426 ± 11433. KuramochiS,MoriguchiT,KuidaK,EndoJ,SembaK, Schrick K, Garvik B and Hartwell LH. (1997). Genetics, 147, Nishida E and Karasuyama H. (1997). J. Biol. Chem., 272, 19 ± 32. 22679 ± 22684. Schurmann A, Mooney AF, Sanders LC, Sells MA, Wang Lavoie JN, L'Allemain G, Brunet A, Muller R and HG, Reed JC and Bokoch GM. (2000). Mol. Cell. Biol., Pouyssegur J. (1996). J. Biol. Chem., 271, 20608 ± 20616. 20, 453 ± 461. Leberer E, Dignard D, Harcus D, Thomas DY and Shi CS, Tuscano J and Kehrl JH. (2000). Blood, 95, 776 ± 782. Whiteway M. (1992). EMBO J., 11, 4815 ± 4824. Su YC, Han J, Xu S, Cobb M and Skolnik EY. (1997). Leberer E, Harcus D, Broadbent ID, Clark KL, Dignard D, EMBO J., 16, 1279 ± 1290. Ziegelbauer K, Schmidt A, Gow NA, Brown AJ and Tassi E, Biesova Z, Di Fiore PP, Gutkind JS and Wong WT. Thomas DY. (1996). Proc. Natl. Acad. Sci. USA, 93, (1999). J. Biol. Chem., 274, 33287 ± 33295. 13217 ± 13222. Tsutsumi T, Ushiro H, Kosaka T, Kayahara T and Nakano Lee BN and Elion EA. (1999). Proc. Natl. Acad. Sci. USA, K. (2000). J. Biol. Chem., 275, 9157 ± 9162. 96, 12679 ± 12684. Tung RM and Blenis J. (1997). Oncogene, 14, 653 ± 659. Lee KK, Murakawa M, Nishida E, Tsubuki S, Kawashima S, Vadlamudi RK, Adam L, Wang RA, Mandal M, Nguyen D, Sakamaki K and Yonehara S. (1998). Oncogene, 16, Sahin A, Cherno J, Hung MC and Kumar R. (2000). J. 3029 ± 3037. Biol. Chem., 275, 36238 ± 36244. Lee KK, Ohyama T, Yajima N, Tsubuki S and Yonehara S. Verheijen MH, Wolthuis RM, Bos JL and De®ze LH. (1999). (2001). J. Biol. Chem., 7, 7. J. Biol. Chem., 274, 1487 ± 1494. Ling P, Meyer CF, Redmond LP, Shui JW, Davis B, Rich YaoZ,ZhouG,WangXS,BrownA,DienerK,GanHand RR, Wange RL and Tan TH. (2001). J. Biol. Chem., 13, Tan TH. (1999). J. Biol. Chem., 274, 2118 ± 2125. 13. Yustein JT, Li D, Robinson D and Kung HJ. (2000). LjungdahlS,LinderS,FranzenB,BinetruyB,AuerGand Oncogene, 19, 710 ± 718. Shoshan MC. (1998). Cell Growth Dier., 9, 565 ± 573. Zhou G, Lee SC, Yao Z and Tan TH. (1999). J. Biol. Chem., Moore TM, Garg R, Johnson C, Coptcoat MJ, Ridley AJ 274, 13133 ± 13138. and Morris JD. (2000). J. Biol. Chem., 275, 4311 ± 4322. ZhouTH,LingK,GuoJ,ZhouH,WuYL,JingQ,MaLand Mosch HU, Roberts RL and Fink GR. (1996). Proc. Natl. Pei G. (2000). J. Biol. Chem., 275, 2513 ± 2519. Acad. Sci. USA, 93, 5352 ± 5356. Nakano K, Yamauchi J, Nakagawa K, Itoh H and Kitamura N. (2000). J. Biol. Chem., 275, 20533 ± 20539.

Oncogene