MST, a physiological caspase substrate, highly sensitizes apoptosis both upstream and downstream of caspase activation.

The human serine/threonine kinase, mammalian STE20-like kinase (MST), is considerably homologous to the budding yeast kinases, SPS1 and STE20, throughout their kinase domains. The cellular function and physiological activation mechanism of MST is unknown except for the proteolytic cleavage-induced activation in apoptosis. In this study, we show that MST1 and MST2 are direct substrates of caspase-3 both in vivo and in vitro. cDNA cloning of MST homologues in mouse and nematode shows that caspase-cleaved sequences are evolutionarily conserved. Human MST1 has two caspase-cleavable sites, which generate biochemically distinct catalytic fragments. Staurosporine activates MST either caspase-dependently or independently, whereas Fas ligation activates it only caspase-dependently. Immunohistochemical analysis reveals that MST is localized in the cytoplasm. During Fas-mediated apoptosis, cleaved MST translocates into the nucleus before nuclear fragmentation is initiated, suggesting it functions in the nucleus. Transiently expressed MST1 induces striking morphological changes characteristic of apoptosis in both nucleus and cytoplasm, which is independent of caspase activation. Furthermore, when stably expressed in HeLa cells, MST highly sensitizes the cells to death receptor-mediated apoptosis by accelerating caspase-3 activation. These findings suggest that MST1 and MST2 play a role in apoptosis both upstream and downstream of caspase activation.

The human serine/threonine kinase, mammalian STE20-like kinase (MST), is considerably homologous to the budding yeast kinases, SPS1 and STE20, throughout their kinase domains. The cellular function and physiological activation mechanism of MST is unknown except for the proteolytic cleavage-induced activation in apoptosis. In this study, we show that MST1 and MST2 are direct substrates of caspase-3 both in vivo and in vitro. cDNA cloning of MST homologues in mouse and nematode shows that caspase-cleaved sequences are evolutionarily conserved. Human MST1 has two caspasecleavable sites, which generate biochemically distinct catalytic fragments. Staurosporine activates MST either caspase-dependently or independently, whereas Fas ligation activates it only caspase-dependently. Immunohistochemical analysis reveals that MST is localized in the cytoplasm. During Fas-mediated apoptosis, cleaved MST translocates into the nucleus before nuclear fragmentation is initiated, suggesting it functions in the nucleus. Transiently expressed MST1 induces striking morphological changes characteristic of apoptosis in both nucleus and cytoplasm, which is independent of caspase activation. Furthermore, when stably expressed in HeLa cells, MST highly sensitizes the cells to death receptor-mediated apoptosis by accelerating caspase-3 activation. These findings suggest that MST1 and MST2 play a role in apoptosis both upstream and downstream of caspase activation.
Apoptosis or programmed cell death is a normal cell suicide mechanism that is highly conserved from nematode to human (1,2). This regulated process plays a critical role during embryogenesis, tissue homeostasis, and remodeling and serves to remove unwanted or deleterious cells such as selfreactive lymphocytes, tumor cells, or virus-infected cells. Deregulation of apoptosis thus contributes to the pathogenesis of cancer, autoimmune disease, and sustained viral infection, whereas excessive apoptosis results in inappropriate cell loss and consequently degenerative disorders (3).
The Fas-mediated apoptosis has been extensively investigated as a model system of mammalian apoptotic cell death (4). Stimulation of Fas induces the translocation and recruitment of FADD together with a complex of pro-caspase-8 and FLASH to the cytoplasmic tail of Fas (5)(6)(7), forming the death-inducing signaling complex (8). Then pro-caspase-8 becomes proteolytically autoactivated by oligomerization, whereupon it stimulates other caspases, including caspase-3, caspase-6, and caspase-7, by cleaving them (9 -12). These downstream caspases cleave the death substrates that are central to apoptotic events such as morphological changes and DNA fragmentation (12,13). MST1 1 and MST2 belong to a mammalian SPS1/STE20-like kinase family, which is rapidly increasing in number (14 -18). However, the cellular function of MST as well as that of most other SPS1/STE20-like kinases is largely unknown. MST1 and MST2 have an N-terminal catalytic domain and C-terminal regulatory region, whereas other subfamily members such as p21-activated kinase (PAK) have a long N-terminal regulatory domain containing GTPase binding domain that confers the ability to bind to activated Cdc42 and/or Rac1 (17,19,20). Notably, PAK2 is proteolytically activated by caspase in Fasmediated apoptosis and has been reported to induce morphological changes of apoptosis (21,22). In contrast, the C-terminal regulatory region of MST does not have any known interaction motif, although this region is required for dimerization (23). Some of the MST subfamily is responsive to cellular stress; inflammatory cytokines such as tumor necrosis factor (TNF) ␣ can trigger activation of GCK, and SOK is responsive to oxidative stress (24 -26). However, MST1 is only activated by non-physiological stresses such as high temperature heat shock and high concentrations of sodium arsenite or staurosporine (16).
Previously we identified a protein kinase that is strongly activated in Fas-mediated apoptosis. Biochemical purification of its activity revealed the kinase to be a catalytic fragment of MST1 (27). Fas-mediated activation of MST resulted from specific cleavage at the caspase recognition site and was inhibited by caspase inhibitor, suggesting the direct cleavage of MST by caspase. These studies suggested that MST and other SPS1/ STE20 family kinases are involved in apoptosis. In the present study, we investigated the involvement of MST in apoptosis. We clearly show that both MST1 and MST2 are cleaved by caspase in vivo and in vitro. The proteolytic cleavage reveals the activation mechanism of MST and causes its cellular translocation. Most important, we show that MST can caspaseindependently induce morphological changes characteristic of apoptosis and enhances the sensitivity for death receptor-mediated apoptosis in epithelial cells.
Expression and Purification of His-tagged Proteins-His-tagged MST was expressed as a kinase-deficient protein because kinase-active MST was poorly expressed in bacteria. Human MST cDNA with a mutation at catalytic amino acid residue (MST1 K59R or MST2 K56R ) was subcloned in frame into His tag expression vector, pQE-30 (Qiagen). For C-terminal His-tagging, MST1 K59R,D326N was subcloned into pET29a (Novagen). To obtain active His-tagged caspase-3, cDNA of human caspase-3 lacking N-terminal 28 amino acid residues was amplified by PCR using 5Ј-AGGGATCCTCTGGAATATCCCTGGAC-3Ј and 5Ј-TGAGCCTTT-GACCATGCCCACAGA-3Ј as primers (underline indicates BamHI site). PCR product was double-digested with BamHI and PstI and subcloned into pQE30 vector. His-tagged proteins were expressed in BL21(DE3)lysS bacteria and purified using His-Trap affinity column as described previously (28).
Preparation of Anti-MST Monoclonal Antibody-For immunization, affinity-purified His-MST1 K59R and His-MST2 K56R were further loaded onto Mono-Q anionic exchange column (Amersham Pharmacia Biotech) and eluted with a linear NaCl gradient of 0.1 to 0.5 M as described (28). Purified fraction, dialyzed against 20 mM Tris⅐HCl (pH 7.5) containing 150 mM NaCl, was used for immunization of Balb/c mice. Establishment and cloning of hybridoma cells were performed as described previously (28). Monoclonal antibodies were purified from hybridoma culture with protein G-affinity column as described (Amersham Pharmacia Biotech) (28).
Cell Culture, Transfection, and Induction of Apoptosis-Human thymoma-derived HPB-ALL, human acute T cell leukemia-derived Jurkat, and murine lymphoma-derived WR19L12a cells were cultured in RPMI 1640 with 10% fetal calf serum, 20 mM Hepes (pH 7.3), 50 M ␤-mercaptoethanol, 50 units/ml penicillin, and 50 g/ml streptomycin. Murine fibroblast NIH 3T3, human adenocarcinoma-derived HeLa, and human embryonic kidney-derived 293T cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 100 g/ml kanamycin. Jurkat cells were transfected by electroporation at 300 V with capacitance of 960 microfarads. NIH 3T3 or HeLa cells were seeded on glass culture slides (Nalge Nunc) or 60-mm culture plates 24 h before transfection and then transfected with LipofectAMINE (Life Technologies, Inc.) in accordance with the manufacturer's suggestion. To establish stable HeLa cell lines, pME18S-Neo vector harboring neomycin-resistant gene was co-transfected with expression vectors encoding GFP, GFP-MST1, or MST1 K59R . The transfectants were grown and selected in the presence of 1 mg/ml G418. To induce apoptosis, cells were treated with anti-Fas antibody, CH-11 (0.1 g/ml) (27,30), TNF␣ (50 ng/ml), or staurosporine (0.5 M). Where indicated, cells were co-treated with cycloheximide (10 g/ml).
Preparation of Cell Lysate, Immunoprecipitation, and Immunoblot Analysis-Cells were washed once with PBS and suspended in cold lysis buffer (40 mM Hepes (pH 7.4) with 10% glycerol, 1% Triton X-100, 0.5% Nonidet P-40, 150 mM NaCl, 50 mM NaF, 20 mM ␤-glycerol phosphate, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, and 0.1 mM vanadate) with protease inhibitor mixture (1 g/ml aprotinin, leupeptin, and pepstatin). Cell lysates were cleared by centrifugation at 15,000 rpm for 20 min. For immunoprecipitation, cell lysate was incubated with 2 g of anti-FLAG mAb or anti-MST mAb for 2 h at 4°C and precipitated with protein G-Sepharose (Amersham Pharmacia Biotech). Cell lysate or immunoprecipitates were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membrane (Millipore). The membrane was blocked in TBST buffer (20 mM Tris⅐HCl (pH 7.5) containing 150 mM NaCl and 0.1% Tween 20) with 5% skim milk at room temperature for 1 h. The membrane was incubated with anti-FLAG mAb or anti-MST mAb for 1 h, washed in TBST, and then incubated for 1 h with horseradish peroxidase-conjugated anti-mouse IgG (Amersham Pharmacia Biotech). After further washing with TBST, peroxidase activity was detected on x-ray films using an enhanced chemiluminescence detection system (DuPont).
In Vitro Cleavage Assay and Cleavage Site Mapping-Purified His-MST1 K59R or His-MST2 K56R was incubated with the purified recombinant active caspase-3 at 30°C for 30 min. The reaction was stopped by adding Laemmli's sample buffer, resolved on 10% SDS-polyacrylamide gel, and immunoblotted with anti-MST mAb. For purification of Cterminal His-tagged protein, MST1 K59R,D326N was subcloned into pET29a (Novagen) and transformed into BL21(DE3)lysS bacteria. Jurkat cells were exposed to CH-11 for 4 h, and lysate was prepared in lysate buffer containing protease inhibitor mixture. Cell lysate (10 7 cell equivalents) was incubated with 50 g of purified MST1 K59R,D326N -His at 30°C for 1 h, and the reaction was terminated by adding 50 M Z-VAD-fmk. The sample was mixed with nickel-nitrilotriacetic acidagarose (Qiagen) and incubated at 4°C for 2 h. The resin was washed with 10 volumes of lysis buffer containing 5 mM imidazole. Laemmli's sample buffer was directly added to the resin, and cleaved proteins were separated on 15% SDS-polyacrylamide gel and transferred to polyvinylidene difluoride membrane. The membrane was immunoblotted or stained with Coomassie Brilliant Blue. The fragment was cut and analyzed by N-terminal peptide sequencing (27,28).
Cloning of Murine and C. elegans MST Homologues-MST homologues of mouse and C. elegans were identified by homology search against public data bases and cloned by PCR-based methods as follows. BLAST search identified overlapping EST murine cDNA clones homologous to 3Ј regions of human MST1 gene. To obtain a complete cDNA sequence, 5Ј-rapid amplification of cDNA ends PCR was carried out using a cDNA library of mouse lymphoma WR19L12a cells as template. For cloning of murine MST2, primers were designed based on the reported nucleotide sequence (GenBank TM accession number U28726), 5Ј-CAGGATCCGCCATGGAGCAGCCGC-3Ј and 5Ј-AACTGCAGTCA-GAAATTCTGCTGCCTCC-3Ј. The cDNA was amplified from the WR19L12a cDNA library. The C. elegans homologue was found by a BLAST search against EST or genome data base (ACeDB). C. elegans MST was obtained from a strain N2 cDNA library by PCR using 5Јatgccaccgtctacagacag-3Ј and 5Ј-cgaagccgtcattgaagtcg-3Ј as primes. The cloned cDNA was compared with a genome data base (ACeDB) and matched completely with the predicted sequences by Gene Finder, a gene structure prediction program. The cDNAs of murine and C. elegans MST were subcloned pME18S-FL and pCMX-SAH/Y145F vector, respectively.
MST Kinase Assay-Endogenous MST or various forms of FLAGtagged MST were immunoprecipitated with 2 g of antibody, and the same amounts of immunoprecipitates as analyzed by immunoblot were used for in-gel phosphorylation assay (27) or immune complex kinase assay. For immune complex kinase assay, immunoprecipitates were incubated with 2 g of histone H2B in 20 l of kinase reaction buffer ␤-Galactosidase Staining-NIH 3T3 or HeLa cells were cultured on 6-well plates and transfected with 1 g of MST kinase expression vector and 0.2 g of a ␤-galactosidase expression vector (pLacZ). After 48 h, the cells were washed once with PBS and fixed with 2% formaldehyde and 0.2% glutaraldehyde in PBS for 5 min at 4°C. After being further washed with PBS, the cells were overlaid with X-gal staining solution (5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 1 mM MgCl 2 , and 1 mg/ml X-gal in PBS) for 1 h at 37°C and photographed.
Fluorescence Microscopy of Cultured Cells-NIH 3T3 or HeLa cells on glass culture slides were washed with PBS, fixed in 3.7% formaldehyde in PBS containing 0.1% Triton X-100 for 10 min, and blocked in 2% bovine serum albumin in PBS for 10 min. Anti-FLAG mAb (1 g/ml) was treated for 1 h and then washed in PBS. The secondary antibody, fluorescein isothiocyanate-conjugated anti-mouse antibody (Cappel), was used at a 1:500 dilution for 45 min. Cells were washed with PBS, incubated with Hoechst 33342 (0.2 g/ml), and washed again in PBS. For fluorescence microscopy of live cells, Hoechst 33342 (0.2 g/ml) was added in culture medium, and the cells were incubated for 30 min at 37°C before microscopic observation. Cells were washed with PBS and observed under a 40ϫ water immersion lens (Carl Zeiss). Samples were examined, and images were collected using Axioplan2 microscope (Carl Zeiss) connected to a CCD camera. The figures were prepared using Photoshop software (Adobe).
Fluorometric Caspase Activity Assay-Cell lysate (100 l) was incubated with 50 M DEVD-7-amido-4-trifluoromethylcoumarin in reaction buffer, 10 mM Tris⅐HCl (pH 7.5), and 1 mM EDTA and incubated at 37°C for 30 min. The reaction was terminated by adding an equal volume of 50 mM glycine HCl (pH 2.3), and fluorescence emission at 460 nm (excitation at 370 nm) was measured.

RESULTS
Proteolytic Cleavage and Activation of Both MST1 and MST2 by Caspase in Fas-mediated Apoptosis-We established several hybridoma cell lines producing anti-MST mAbs, and purified mAbs were characterized by immunoblot analyses for HeLa cells transfected with FLAG-MST1 or FLAG-MST2 (Fig. 1A). Monoclonal antibody G2B equally recognized both MST1 and MST2, whereas A8C selectively recognized only MST1. When apoptosis was induced, G2B and A8C could detect a cleaved 34-kDa fragment of MST. On the contrary, J7B detected the C-terminal 21-kDa fragment of MST2. Other mAbs, DIH, F5C, and H3B, recognized N-terminal 34-kDa fragment of MST1 and MST2. 2 Immunoprecipitation analysis showed that A8C and D1H selectively immunoprecipitated FLAG-MST1, whereas J7B immunoprecipitated only FLAG-MST2 (Fig. 1B). A8C could not immunoprecipitate N-terminal 34-kDa fragment although A8C recognized it in immunoblot. 2 Both FLAG-MST1 and FLAG-MST2 were efficiently immunoprecipitated by G2B or H3B. With G2B and A8C anti-MST mAbs, we examined the cleavage of endogenous MST1 and MST2 during Fas-mediated apoptosis in HPB-ALL cells, which expressed both MST1 and MST2. G2B recognized a 34-kDa fragment of MST at 2 h after the stimulation of Fas (Fig. 1C). This fragment was also confirmed by immunoprecipitation ( Fig. 2A). The cleavage of MST was inhibited when apoptosis was blocked by the pretreatment with Z-VAD-fmk, a broad spectrum caspase inhibitor ( Fig. 1C and  2A). The molecular weight of cleaved MST coincided with that of the previously purified protein kinase from apoptotic HPB-ALL lysate, confirming our result that MST is specifically . Immunoblot analysis was performed using G2B (A) or A8C (B). As a control, purified MST from apoptotic HPB-ALL cells (27) was also analyzed and indicated as purified 34-kDa MST (A and B). An aliquot of immunoprecipitate was subjected to in vitro kinase assay using histone H2B as substrate (C). Cell lysates were analyzed for DNA fragmentation as a marker of apoptosis (D). E, bacterially expressed His-MST1 (1 g) and His-MST2 (1 g) were incubated with (ϩ) or without (Ϫ) recombinant active caspase-3 (50 ng) in the presence (ϩ) or absence (Ϫ) of 25 M DEVD-CHO at 30°C for 30 min. The samples were resolved on SDS-PAGE gel and analyzed by immunoblotting with G2B.
cleaved to a 34-kDa catalytic fragment in apoptosis ( Fig. 2A) (27). A8C efficiently precipitated MST1 when apoptosis was not induced or Z-VAD-fmk was pretreated. However, A8C could not immunoprecipitate MST1 when apoptosis was induced (Fig.  2B). These results suggest that MST1 was completely cleaved to a fragment that could not be precipitated by A8C because A8C could immunoprecipitate only uncleaved MST1. 2 Next, we examined whether the cleavage of MST induces its activation by immune complex kinase assay using histone H2B as substrate. As depicted in Fig. 2, A-D, the phosphorylation of H2B was substantially increased by Fas ligation with approximately the same kinetics as processing of MST and DNA ladder formation. Preincubation with Z-VAD-fmk effectively prevented MST activation for up to 3 h after Fas ligation. Bacterially expressed His-tagged MST1 and MST2 were purified and incubated with active caspase-3 in vitro. His-tagged MST1 and MST2 were completely cleaved to the 34-kDa fragment, and the cleavage was blocked by Z-VAD-fmk (Fig. 2E). These results suggest that both human MST1 and MST2 are physiological substrates of caspase-3 and are proteolytically activated by caspase in Fas-mediated apoptosis.
Evolutionary Conservation of the Cleavage Site in MST-To address the functional importance of cleavage sequences in human MST1 and MST2 (hMST1 and hMST2), we cloned murine and C. elegans homologues (mMST and cMST, respectively) and compared the polypeptide sequences of hMST1, hMST2, mMST1, mMST2, rat MST2 (rMST2, GenBank TM accession number AJ001529), and cMST (Fig. 3). Murine MST1 and MST2, which show 75% identity to each other, had 97 and 96% homology to human MST1 and MST2, respectively (Fig.  3A). Previously reported cDNA (GenBank TM accession number U28726) that encodes a murine MST2-like polypeptide deleted the C-terminal region, which is caused by single nucleotide deletion. However, we could not obtain MST2 cDNA with nucleotide deletion or alternative splicing from a murine lymphoma WR19L12a cDNA library (28). We obtained only one MST homologue in C. elegans, which showed similar homology (ϳ50%) to mammalian MST1 and MST2. The kinase domain was most conserved, with 95-100% homology between mammalian homologues. Even the kinase domain of cMST showed 75% homology to mammalian MST1 or MST2. The C-terminal 60 amino acids required for dimerization were also highly conserved in all homologues (23) (Fig. 3B). The putative caspase-cleavable sites (Fig. 3B, indicated by arrow) were con- served in all MST homologues, and these homologues were actually cleaved when apoptosis was induced. 2 Thus, the functional cleavage sequence for caspase is evolutionarily conserved in MST homologues.
Two Activation Mechanisms of MST1 following Treatment with Staurosporine-Previously, Taylor et al. (16) identified a protein kinase responsive to stress (KRS), which was activated by treatment with staurosporine but not by Fas ligation. Subsequent biochemical purification revealed it as MST (16). However, since our result showed that Fas ligation activated MST (27) (Fig. 2), we compared the MST activation by staurosporine treatment and that by Fas ligation in Jurkat cells that undergo apoptosis following these stimulations. Both staurosporine treatment and Fas ligation effectively induced the cleavage of MST1 to 34 kDa (Fig. 4). Unexpectedly, MST1 D326N in which the site of cleavage (Asp 326 ) by caspase-3 is mutated to Asn (27) was also processed to a slower migrating fragment of 40 kDa. In addition, the mobility of the 34-kDa fragment was slightly but significantly different in staurosporine-and Fas-induced apoptosis. The reason for the different mobility might be that staurosporine treatment caused a change of phosphorylation state in MST1, which was not caused by Fas stimulation. We investigated the kinase activity of each MST1 fragment by in-gel phosphorylation assay using histone as substrate. Phosphorylation at 55 kDa corresponding to full-length MST1 was greatly increased at 0.5 h by staurosporine treatment and peaked at 1 h (Fig. 4, lanes 1-4). In contrast, Fas ligation could not induce phosphorylation of full-length MST1 up to 3 h (Fig.  4, lanes 5 and 6). Phosphorylation at 40 kDa was also rapidly induced at 0.5 h by staurosporine, although this fragment was not detected in immunoblot analysis. Fas stimulation did not induce a 40-kDa fragment in immunoblot or kinase activity of 40-kDa in-gel kinase assay. The 34-kDa fragment with histone kinase activity was generated from FLAG-MST1 (Fig. 4, lanes  1-6) but not from FLAG-MST1 D326N within 3 h by staurosporine and Fas ligation (Fig. 4, lanes 7-12). These results indicate that staurosporine can activate MST by two mechanisms, caspase-independent early activation and caspase-dependent late activation, whereas Fas ligation activates MST only in a caspase-dependent pathway.
Cleavage of MST1 D326N to a 40-kDa fragment indicated that another cleavage site exists in the C-terminal region of Asp 326 , and this cleavage is also the biochemical event involved in apoptosis. His-tagged MST1 K59R,D326N was effectively cleaved to 40 kDa in apoptotic Jurkat cell lysate. 2 To determine the site of this cleavage, we purified a C-terminal 15-kDa fragment derived from MST1 K59R,D326N and performed N-terminal peptide sequencing. It was revealed that Asp 349 was the cleavage site, and the peptide sequence surrounding Asp 349 (TMTD 349 G) was similar to the identified caspase recognition sequence of MST homologue, particularly that of cMST (Fig. 3B). We mutated both Asp 326 and Asp 349 to Asn (MST1 D326N,D349N ) and tested whether MST1 D326N,D349N is resistant to proteolytic cleavage during Fas-mediated apoptosis. MST1 D326N,D349N was cleaved to neither the 40-nor to the 34-kDa fragment (Fig. 5). MST2 D322N was not cleaved to a smaller fragment indicating that MST2 is cleaved only at Asp 322 during apoptosis. In contrast, wild type MST and MST1 D326N were cleaved to the 34-and 40-kDa fragment, respectively (Fig. 5). Therefore, MST1 is cleaved at two different sites resulting in a 34-or 40-kDa fragment, and the 40-kDa fragment is further cleaved to 34 kDa.
Apoptotic Morphological Changes Induced by MST1-To investigate the possible function of MST in apoptosis, we transfected NIH 3T3 cells with FLAG-tagged MST constructs (Fig.  6A). Expression of the various MST1 constructs was confirmed by immunoblotting with anti-FLAG mAb (Fig. 6B, top). Overexpression did not induce MST1 cleavage. The expression level of MST1⌬327-487, with a C-terminal truncation corresponding to the active MST1 fragment generated by caspase, was very low. In comparison with MST1, MST1⌬327-487 showed enhanced histone phosphorylation activity but little autophosphorylation activity (Fig. 6B, bottom). A similar result was obtained when FLAG-tagged MST2 was expressed in NIH 3T3 cells. 2 Cells expressing active MST1 showed marked changes resulting in round morphology (Fig. 6C). The morphological change induced by MST1 was indistinguishable from that induced by overexpression of caspase-8, an initiator caspase. However, the round morphology induced by MST1 was not inhibited by the treatment with Z-VAD-fmk, although that by caspase-8 was inhibited by Z-VAD-fmk. 2 We compared the morphological change induced by MST1 with that caused by PAK2 (21). Similar change was observed when PAK2 was transfected into NIH 3T3 cells. Neither kinase-negative PAK2 K299R , MST1 K59R , nor MST1⌬1-326 deleting N-terminal kinase domain evoked changes in cell morphology indicating that kinase activity of MST1 or PAK2 is required for this morphological change. MST-induced change of morphology was also observed in HeLa cells (Fig. 7). MST-expressing cells were rounded and shrunken with associated destruction of cytoskeletal structure (Fig. 7A). Hyperactive MST1⌬327-487 showed more dramatic change of morphology and resulted in the loss of MST1⌬327-487-expressing cells by detachment from substrate. This result may elucidate why the percentage of the round cells in MST1⌬327-487-transfected cells was reduced in ␤-galactosidase assay (Fig. 6C). The enhanced morphological change induced by MST1⌬327-487 may be the result from the increased kinase activity because kinase-negative MST1 K59R ⌬327-487 or MST1 K59R did not induce any morphological changes.
We investigated MST-induced morphological changes in detail. In MST1-expressing HeLa cells at 16 h post-transfection, the cytoplasm was remarkably reduced compared with untransfected cells or transfected cells with MST1 K59R (Fig. 7B,  top and middle). The morphological change of chromosome was not observed, although the nuclei were slightly reduced and rounded. MST1 and MST1 K59R localized in the cytoplasm. At 48 h post-transfection, the cytoplasmic space was completely lost, and cells were eventually detached from the plate (Fig. 7B,  bottom). In addition, condensation and fragmentation of chromosome was obvious by Hoechst staining. MST1 appeared to localize in the nucleic area but not overlapped with chromosome. However, these cells were not positive for terminal deoxynucleotidyltransferase-mediated dUTP nick end-labeling staining, and the characteristic DNA ladder in apoptosis was not detected on agarose gel electrophoresis. 2 These observations indicate that chromosome was not cleaved to much smaller fragments generated by caspase-activated nuclease, such as CAD. We conclude that part of apoptosis-related morphological changes are induced by overexpression of MST, which mimics the caspase-dependent activation of MST in apoptosis.
Nuclear Translocation of Caspase-cleaved MST-MST1 and After 48 h of post-transfection, apoptosis was induced with of CH-11 (0.1 g/ml) and cycloheximide (10 g/ml). Immunoblot analysis (20 g) was performed with anti-FLAG mAb. WT, wild type. MST1 K59R were exclusively localized in the cytoplasm when expressed as FLAG-tagged or GFP fusion forms ( Fig. 7 and 8A). MST1⌬1-326 deleting N-terminal kinase domain was also localized in the cytoplasm, but MST1⌬327-487 was distributed in both cytoplasm and nucleus. To test whether the cytoplasmic localization of MST1 is controlled by active nuclear export, HeLa cells expressing MST1 or MST1⌬1-326 were treated with leptomycin, an inhibitor of CRM1/exportin (31,32). MST1 or MST1⌬1-326 rapidly lost cytoplasmic localization (Fig. 8A). These results indicate that cytoplasmic localization of MST1 is controlled by a C-terminal regulatory region through active nuclear export by the CRM1/exportin (33)(34)(35).
To test whether the localization of MST is changed in Fasmediated apoptosis, we established stable HeLa cell lines expressing caspase-cleavable and uncleavable MST1. MST1 was fused at the C-terminal end of GFP for rapid detection of morphological change and translocation. Established cell lines expressing wild type MST did not show significant morphological differences with control cell lines, although transiently expressed MST induced morphological change (Fig. 7). In the stable cell lines, cleavable and uncleavable MST1 showed a similar cytoplasmic localization before induction of apoptosis. 2 However, in early apoptotic cells that showed blebbing of membrane but retained their nuclear structure, cleavable MST1 and MST1 K59R translocated into the nucleus although MST1 and MST1 K59R were still detected in the cytoplasm (Fig., 8B, top and middle rows). In contrast, uncleavable MST1 K59R ,D326N,D349N (Fig. 5) remained in the cytoplasm although the apoptosis was apparently occurring as judged by the apoptotic blebbing of membrane (Fig. 8B, bottom row). In later apoptotic cells with nuclear fragmentation, even MST1 K59R,D326N,D349N completely lost cytoplasmic localization and was detected in the fragmented nuclei, 2 suggesting the translocation after the breakdown of the nuclear envelope. We conclude that MST1 is cytoplasmic protein and the cleaved MST1 translocates into the nucleus before the nuclear fragmentation is initiated.
MST Sensitizes Fas-and TNF␣-mediated Apoptosis-To address further the function of MST in apoptosis, we investigated the sensitivity of HeLa cell lines stably expressing MST1 in Fas-and TNF␣-mediated apoptosis. The expression of GFP-MST1 and GFP-MST1 K59R was confirmed by immunoblotting with G2B or with anti-PARP antibody as a control for loading and transfer (Fig. 9A). All sub-cell lines expressing GFP-MST1 were more sensitive to Fas-or TNF␣Ϫmediated apoptosis than parental HeLa cells or GFP-MST1 K59R -expressing cells (Fig.  9B). GFP-MST1 K59R did not accelerate or inhibit Fas-or TNF␣Ϫmediated apoptosis.
Interestingly, GFP-MST1 greatly facilitated the activation of caspase-3-like protease in Fas-mediated apoptosis (Fig. 9C) suggesting that MST1 can accelerate caspase-3 activation. To test this possibility, we investigated the cleavage kinetics of caspase-3 in detail. As shown in Fig. 9D, the p17 subunit of caspase-3 was detected at 1.5-2 h in MST1-expressing cell lines (HL-WT1 and HL-WT2) but p17 was detected after 3 h in control cells. Furthermore, the cleavage kinetics of GFP-MST1 was correlated with that of caspase-3 supporting the accelerated activation of caspase-3, because MST1 was shown to be cleaved by activated caspase-3 (Fig. 2). GFP-MST1 did not accelerate the onset of other apoptosis induced by pharmaceutical drugs such as ceramide, staurosporine, and etoposide. 2 We conclude that stably expressed MST1 sensitizes death receptormediated apoptosis through the accelerated onset of caspase-3 activation.

DISCUSSION
Cleavage and Activation of MST in Apoptosis-We previously reported that during Fas-mediated apoptosis MST1 and MST2 are proteolytically activated by specific cleavage. However, it remained unclear how MST1 and MST2 are involved in apoptosis. In this study, we clearly show that MST1 and MST2 are physiologically activated by caspase-3-like protease and involved in apoptosis. Caspase cleavage sequences of MST1 (Asp-Glu-Met-Asp 326 -Ser) and MST2 (Asp-Glu-Leu-Asp 322 -Ser) are more conserved and optimal than those of other protein kinases known to be activated proteolytically by caspase-3-like proteases including caspase-6 and caspase-7 (27,36). This probably explains why only MST was purified as a major protein kinase that was activated in Fas-mediated apoptosis (27). Caspase cleavage sequences in MST1 and MST2 are conserved not only in their mammalian but in the nematode homologue, strongly suggesting their biological significance including a role in apoptotic cell death signaling. The cleavage of MST may be a common molecular event in apoptosis since various apoptotic signals induce caspase-dependent MST cleavage, including signals from Fas and TNF␣ receptor, and chemical apoptotic inducers such as staurosporine, etoposide, and ceramide. Recently bisphosphonates, pharmacological drugs for osteoporosis, and anti-tumor drugs such as cytotrienin A have been reported to induce caspase-dependent cleavage of MST in osteoclast and leukemia cells (37,38).
The cleavage of MST is closely coupled with its activation. Although MST has basal kinase activity in non-apoptotic cells, caspase-dependent MST cleavage induces significant and strong activation of the catalytic fragment. Furthermore, truncated MST, MST1⌬327-487, has powerful kinase activity supporting the activation by caspase-mediated cleavage. In human MST1, another cleavage site (TMTDG 349 ) was found in this report, which generates a catalytic fragment of 40 kDa. The 40-kDa fragment is detected early in the in-gel kinase assay, were treated with CH-11 (0.1 g/ml) together with cycloheximide (10 g/ml) for 4 h, and then cell lysate was prepared. Caspase-3-like activity was measured using 50 M of DEVD-7-amido-4-trifluoromethylcoumarin as substrate. D, two MST1-expressing cell lines (HL-WT1 and HL-WT2) and control HL-Neo cell lines were treated with CH-11 (0.1 g/ml) and cycloheximide (10 g/ml) for the indicated times. Cell lysate (50 g) was analyzed by immunoblot with anti-caspase-3 (top) or with G2B (bottom). Full-length and cleavage products are indicated by arrows and arrowheads, respectively. Other stable cell lines showed essentially similar results.
whereas the 34-kDa catalytic fragment is detected only at a late time point by staurosporine treatment or Fas stimulation (Fig. 4). Thus, cleavage at Asp 349 precedes that at Asp 326 and suggests rapid turnover of the 40-kDa fragment to the 34-kDa form. Based on the cleavage kinetics, we conclude that human MST1 is sequentially cleaved at Asp 326 and Asp 349 in staurosporine-induced apoptosis. However, it is not likely that the cleavage at Asp 349 is a prerequisite for MST1 activation because Asp 349 is not conserved in murine MST1. In addition, MST1 D349N is processed to the 34-kDa fragment with strong kinase activity. Nevertheless, a specific function of the 40-kDa fragment, particularly in early apoptotic events, cannot be ruled out because it shows distinct catalytic activity from the 34-kDa fragment.
Apoptotic Cleavage of MST Induces Nuclear Translocation-We show that MST is localized exclusively in the cytoplasm. Deletion analysis shows that the C-terminal region of MST is sufficient for cytoplasmic localization that is lost on treatment with leptomycin. Cytoplasmic localization of MST may be regulated by nuclear export signal although the possibility remains that other nuclear export signal-containing proteins or cytoplasmic anchor protein regulates MST localization. MST cleavage by caspase results in the subcellular relocalization of catalytic fragment. In Fas-mediated apoptosis, cleaved MST1 translocates into the nucleus before nuclear fragmentation is initiated. This result suggests additional nuclear functions of cleaved MST1 in apoptosis. Histone H2B is selectively phosphorylated in mammalian apoptotic cells and is associated with chromatin condensation in apoptosis (39). It is of interest to test whether cleaved MST phosphorylates H2B in apoptosis because H2B is an excellent substrate in vitro ( Fig. 2 and 6). Caspase-dependent cleavage might affect other biochemical functions of MST. MST1 and MST2 were reported to form homo-or heterodimers through the C-terminal region (23), and cleavage of MST naturally leads to the dissociation of N-terminal catalytic fragments. Therefore, cleavage of MST during apoptosis results in at least three coupled events, activation, monomerization, and translocation of catalytic fragment into the nucleus. These representative regulatory mechanisms, observed in many signaling molecules, may be involved in the apoptotic function of MST.
MST Is Involved in Apoptotic Events-Overexpression of MST or truncated MST induces morphological changes in cells. These changes are at least partly related to the characteristic morphology of apoptotic cells such as cell rounding, detachment from substratum, and the fragmentation and condensation of nucleus. However, we could not detect other apoptotic characteristics such as externalization of phosphatidylserine, DNA laddering, or formation of apoptotic bodies. These morphological changes induced by MST are similar to those induced by PAK2 (21,22). Overexpression of each kinase-negative, caspase-resistant form of MST1 and MST2 does not inhibit Fas-mediated apoptosis. 2 This suggests that other SPS1/ STE20 family members are also involved in apoptosis. Recently, STE20-related kinase, SLK, was reported to be cleaved by caspase-3 and to induce cytoskeletal rearrangement (40 -42). Caspase-dependent cleavage and effects on apoptotic morphology of MST, PAK2, and SLK suggest that SPS1/STE20 family kinases are closely involved in apoptosis, probably targeting similar molecules.
Previously it was suggested that overexpression of MST1 activates caspase through the JNK/SAPK kinase pathway (43). However, others (37) could not detect the SAPK/JNK activation. We detected neither the JNK activation nor the caspase activation by overexpression of MST, and PARP, caspase substrate, was not cleaved in our system. 2 In addition, the MST1-induced morphological change of the cell was unaffected by preincubation with caspase inhibitor, Z-VAD-fmk (Fig. 6, C and  D). Caspase-resistant MST1 D326N,D349N still induced similar morphological changes in the presence of caspase inhibitor. Thus, we conclude that morphological changes induced by transient overexpression of MST do not require the activation of caspase, and SAPK/JNK might not be involved in the morphological change.
When stably expressed in the GFP fusion form, MST1 highly sensitizes HeLa cells to Fas-or TNF␣ receptor-mediated apoptosis. Surprisingly, susceptibility to apoptosis was enhanced by accelerated activation of caspase. This result contradicts the finding of transient expression (Fig. 6) that indicates caspase activation is not required for the MST-induced apoptosis-like morphological changes. One possible explanation is that MST has two distinctive functions, direct induction of morphological changes and sensitization to death receptor-mediated apoptosis. The limited apoptotic change induced by transiently expressed MST may reflect MST function downstream of caspase. This explanation is supported by the fact that MST1 is a substrate of caspase, and truncated MST induces morphological changes that are more dramatic. When stably expressed, MST may accelerate caspase activation through the phosphorylation of pro-or anti-apoptotic regulators or caspases. It cannot be ruled out that MST regulates the expression of apoptotic regulators or caspases. Although it is largely unknown how MST is involved in apoptosis, our result clearly shows that it acts as both an upstream activator and a downstream effector of caspase; MST regulates death receptor-induced caspase activation and amplifies caspase-dependent morphological change.
Our result provides some clue to the regulation and function of MST in non-apoptotic cell signaling. Although no physiological activators are known, the rapid activation of MST by staurosporine indicates that there exists a caspase-independent mechanism of activation, which is regulated by upstream kinase(s) or phosphatase(s). Another critical clue to understanding the function of MST may be obtained from the analysis of subcellular localization since some protein kinases are tightly regulated in intracellular spaces (44 -46). We propose the regulation of subcellular localization as a possible mechanism regulating cellular function of MST. The identification of physiological substrate will undoubtedly lead to a better understanding of MST function in apoptosis or other cell signaling.