Oncogene (2011) 30, 4208–4218 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE Oligomeric -I is an essential intermediate for p53 to activate MST1 kinase and apoptosis

A Morinaka, Y Funato, K Uesugi and H Miki

Laboratory of Intracellular Signaling, Institute for Research, Osaka University, Osaka, Japan

Mammalian Ste20-like kinase-1 (MST1) kinase mediates molecular mechanisms by which is H2O2-induced cell death by anticancer drugs such as relayed inside cells remain poorly characterized. Protein cisplatin in a p53-dependent manner. However, the kinases generally have key roles in mechanism underlying MST1 activation by H2O2 remains driven by various stimuli (Karin and Hunter, 1995). For unknown. Here we show that peroxiredoxin-I (PRX-I) is example, mammalian Ste20-like kinase-1 (MST1) is a an essential intermediate in H2O2-induced MST1 activa- Ser/Thr kinase that is activated by apoptosis-inducing tion and cisplatin-induced cell death through p53. Cell stimuli (de Souza and Lindsay, 2004). Lehtinen et al. stimulation with H2O2 resulted in PRX-I oxidation to (2006) reported that MST1 is activated by H2O2 and form homo-oligomers and interaction with MST1, leading induces cell death by phosphorylating the transcription to MST1 autophosphorylation and augmentation of factor FOXO in primary mammalian neurons. Further- kinase activity. In addition, RNA interference knockdown more, Ste20 mediates H2O2-induced cell death by experiments indicated that endogenous PRX-I is required phosphorylating histone H2B in Saccharomyces cerevi- for H2O2-induced MST1 activation. Live-cell imaging siae (Ahn et al., 2005). These studies implicate MST1 as showed H2O2 generation by cisplatin treatment, which a crucial kinase in H2O2-induced cell death, but MST1 likewise caused PRX-I oligomer formation, MST1 activation mechanisms by H2O2 have yet to be clarified. activation and cell death. Cisplatin-induced PRX-I MST1 is also known as a tumor-suppressor protein oligomer formation was not observed in embryonic involved in the death of cells treated with fibroblasts obtained from p53-knockout mice, confirming anticancer drugs such as cisplatin, a platinum-based the importance of p53. Indeed, ectopic expression of p53 DNA-damaging agent (O’Neill et al., 2005; Ren et al., induced PRX-I oligomer formation and cell death, both of 2008). It is widely accepted that p53 is a key determinant which were cancelled by the antioxidant NAC. Moreover, for cell death in response to anticancer drugs (Lowe we succeeded in reconstituting H2O2-induced MST1 et al., 1993, 1994). By using Drosophila genetics, activation in vitro, using purified PRX-I and MST1 Colombani et al. (2006) discovered that Dmp53 . Collectively, our results show a novel PRX-I (Drosophila melanogaster p53) activates Hippo, the function to cause cell death in response to high levels of Drosophila MST1 homolog, to induce cell death oxidative stress by activating MST1, which underlies the responses elicited by DNA-damaging ionizing radiation, p53-dependent cytotoxicity caused by anticancer agents. suggesting a functional link between p53 and MST1. Oncogene (2011) 30, 4208–4218; doi:10.1038/onc.2011.139; Moreover, it has been reported that p53-dependent published online 25 April 2011 apoptosis occurs through increasing oxidative stress (Johnson et al., 1996; Polyak et al., 1997; Li et al., 1999). Keywords: MST1; PRX-I; reactive oxygen species Indeed, expression of p53 induces the generation of (ROS); p53; cisplatin mitochondrial reactive oxygen species (ROS), likely causing MST1 activation. However, the precise mechan- isms of p53-induced cell death through oxidative stress remain unknown. Introduction Recent studies have uncovered a role for H2O2 as a signaling molecule under various physiological/patho- Oxidative stress greatly affects diverse biological phe- logical settings (Rhee, 2006; Veal et al., 2007). Several nomena, including cell death, aging and various studies have found many H2O2-responsive proteins, disorders ranging from to neurodegenerative among which peroxiredeoxins (PRXs) are the most diseases (Finkel and Holbrook, 2000). However, the extensively characterized (Rhee et al., 2005). The main function of PRXs generally has been thought to be H2O2 removal by catalysis of H2O2 reduction to H2O. Correspondence: Dr H Miki, Laboratory of Intracellular Signaling, However, Jang et al. (2004) reported quite unexpectedly Institute for Protein Research, Osaka University, 3-2 Yamadaoka, that high levels of H2O2 stimulate PRX-I and PRX-II to Suita, Osaka 565-0871, Japan. E-mail: [email protected] form homo-oligomers with chaperone-like activities. In Received 2 December 2010; revised and accepted 22 March 2011; addition, Tpx1, the yeast PRX-I homolog, mediates published online 25 April 2011 H2O2-induced activation of the p38/JNK homolog Sty1, PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4209 by forming a transient, intermolecular disulfide bond anti-PRX-SO2/SO3 antibody and found that the hyper- (Veal et al., 2004). Therefore, PRX-I is now regarded oxidation status of WT PRX-I correlated well with not only as a scavenger of H2O2, but also as an impor- oligomer formation. However, we did not observe so tant intermediate activating H2O2-responsive signaling clear positive signals in the case of C173S PRX-I, and pathways. thus hyper-oxidation of PRX-I may not be required for Here, we investigated the possible involvement of oligomer formation. PRX-I in H2O2- and cisplatin-induced MST1 activation Having confirmed PRX-I oligomerization, we exam- and cell death. In response to H2O2, PRX-I formed ined possible PRX-I interaction with MST1. PRX-I oligomers that specifically associated with MST1. Over- (WT and the Cys mutants) and MST1 were coexpressed expression and knockdown analyses indicated an and cell lysates were subjected to co-immunoprecipita- essential role of PRX-I in MST1 activation by H2O2 tion. The results clearly indicated that MST1 associates in cells. Live-cell imaging analyses clearly showed that with WT PRX-I depending on H2O2 stimulation. cisplatin treatment induced massive H2O2 generation However, MST1 associated constitutively with C173S also resulting in PRX-I oligomer formation. This PRX-I (Figure 1b), correlating well with the oligomeric occurred in a p53-dependent manner because primary status of PRX-I (Figure 1a). There are six different murine embryonic fibroblasts (MEFs) obtained from encoding PRX isoforms in mammalian genomes; p53À/À mice did not form PRX-I oligomers. Moreover, PRX-II is also reported to form oligomers (Jang et al., p53À/À MEFs and PRX-I-knockdown cancer cells were 2004; Moon et al., 2005). Therefore, we also examined unable to activate MST1 in response to cisplatin and whether MST1 associates with PRX-II, but observed no resisted cell death. Further, we successfully reconstituted positive signal (Figure 1c), indicating that MST1 MST1 activation by H2O2 and PRX-I in vitro, using interacted specifically with PRX-I. In support of this purified recombinant proteins. result, PRX-II barely formed oligomers compared with PRX-I (Supplementary Figure 1). We then performed co-immunoprecipitation analyses Results against endogenous MST1/2 and PRX-I proteins in U2OS cells, and confirmed that they form complexes Specific association of MST1 with oligomeric PRX-I in vivo when cells are treated with H2O2 (Figure 1d). To H2O2 induces both MST1 activation and PRX-I examine the direct interaction between these proteins, oligomerization, prompting us to speculate that MST1 we expressed and purified GST-MST1 and PRX-I may be activated by associating with oligomeric PRX-I. recombinant proteins. The purified proteins were We first checked whether PRX-I forms oligomers as incubated in the presence of H2O2 or the reducing agent reported previously. We ectopically expressed PRX-I dithiothreitol (DTT) and then subjected to a pull-down (wild-type (WT) and three different Cys mutants, C51S, assay using glutathione beads. The results indicated a C173S and C83S) in COS-7 cells, which were subse- clear positive signal in the presence of H2O2 (Figure 1e), quently stimulated with H2O2. Cys51 and Cys173, both indicating their direct interaction. We also confirmed of which are known to function as the catalytic center that recombinant PRX-I formed oligomers in vitro for the activity, have an important but similar to those in cells. opposite role in oligomer formation (Jang et al., 2004; Moon et al., 2005; Rhee et al., 2005). Yeast C47S (C51S in mammals) PRX-I does not form oligomers at all even PRX-I mediates H2O2-induced MST1 activation in the presence of H2O2, and yeast C170S (C173S in As H2O2 stimulation is known to activate MST1, we mammals) PRX-I constitutively forms oligomers even in next investigated the possible importance of PRX-I in the absence of H2O2 (Jang et al., 2004). Cys83 is also MST1 activation. First, we coexpressed MST1 with important for oligomerization because C83S PRX-I PRX-I (WT and the Cys mutants) in COS-7 cells and barely forms oligomers in comparison with WT PRX-I examined MST1 autophosphorylation, which reflects (Lee et al., 2007). Cells were then harvested and cell MST1 activation (Praskova et al., 2004). As shown in lysates were subjected to native PAGE without sodium Figure 2a, expression of WT PRX-I significantly dodecyl sulfate (SDS) or reducing agents. As shown in augmented MST1 autophosphorylation. In addition, Figure 1a, a ladder-like signal of WT PRX-I appeared C173S PRX-I, which constitutively associates with above the main band when cells were treated with H2O2. MST1 (Figure 1b), strongly induced MST1 autopho- 2-Cys PRXs, including PRX-I, is known to form sphorylation even without H2O2 stimulation. We also obligate homodimers (Wood et al., 2003). The main examined the effect of PRX-II and found that it did not band and the ladder bands should correspond to the stimulate MST1 autophosphorylation (Supplementary dimer and oligomer forms of PRX-I, respectively, Figure 2). Therefore, there was a very clear correlation because a very similar pattern was reported previously between MST1 autophosphorylartion and interaction in the case of plant 2-Cys PRX (Ko¨nig et al., 2002). By with oligomeric PRX-I. c-Jun N-terminal kinase-1 contrast, C173S PRX-I constitutively formed oligomers (JNK1) is also known to be activated by H2O2 and but C51S or C83S PRX-I did not form oligomers even in has a crucial role in H2O2-induced signaling (Guyton the presence of H2O2. These results agree with previous et al., 1996). Therefore, we examined the possible effect studies (Jang et al., 2004; Moon et al., 2005; Lee et al., of PRX-I expression on JNK1 activity. In contrast to 2007). We also performed immunoblot analyses using an the case of MST1, we did not observe any stimulatory

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4210

Figure 1 Specific association of MST1 with oligomeric PRX-I. (a) COS-7 cells were transfected with the indicated constructs and treated with 100 mM H2O2 for 30 min. Cell lysates were treated with IAA and immunoblotted under native (without SDS or reducing agents) or denaturing (with SDS and reducing agents) conditions. (b, c) COS-7 cells were transfected with the indicated constructs and treated with 100 mM H2O2 for 30 min. Cell lysates were immunoprecipitated (IP) with an anti-FLAG antibody and analyzed by immunoblotting using the indicated primary antibodies. (d) Lysates of U2OS osteosarcoma cells treated with 500 mM H2O2 for 30 min were immunoprecipitated with anti-MST1/2 antibodies and immunoblotted with the indicated primary antibodies. (e) GST-MST1 pull-down assays of recombinant PRX-I in the presence of 10 mM DTT or 10 mM H2O2. Proteins were subjected to SDS–PAGE and Coomassie Brilliant Blue (CBB) staining. The immunoblot of recombinant PRX-I subjected to native PAGE is also shown (left). DTT, dithiothreitol; IAA, iodoacetamide; GST, glutathione-S-transferase; MST1, mammalian Ste20-like kinase-1; PRX-I, peroxiredoxin-I; SDS–PAGE, sodium dodecyl sulfate–PAGE.

effect of PRX-I, although H2O2-induced activation of MST1 (Figure 1b) even in the absence of H2O2, which JNK1 was clearly observed (Supplementary Figure 3). presumably resulted in weak activation of MST1. As mentioned above, MST1 autophosphorylation is a Because MST1 activation has been linked to apoptosis, useful marker for estimating MST1 activation, but we next examined the effect of MST1 coexpression with whether MST1 kinase activity was actually promoted PRX-I on apoptosis as a biological measure for MST1 was still unknown. Therefore, we aimed to measure activation. When MST1 was ectopically expressed alone MST1 kinase activity directly by collecting MST1 in U2OS cells, B15% of cells underwent apoptosis protein by immunoprecipitation and performing kinase showing typical chromatin condensation (Figure 2c). A assays in vitro. First, we used myelin basic protein as slight but significant increase in the apoptotic rate was substrate for assaying MST1 kinase activity, but found observed by coexpression of WT PRX-I, whereas C173S only a weak signal (data not shown). Therefore, we used PRX-I further enhanced the apoptotic rate to B35%, a GST-fusion MOBKL1B protein, which was reported which is consistent with the stronger ability of C173S to be a good MST1 substrate (Praskova et al., 2008). We PRX-I to activate MST1 (Figures 2a and b). basically found similar results to those in the autopho- To examine the requirement of endogenous PRX-I for sphorylation analyses (Figure 2b), thus confirming H2O2-induced activation of endogenous MST1/2, short MST1 activation by PRX-I. We also noticed that interfering RNA (siRNA) against PRX-I was intro- expression of WT PRX-I could induce statistically duced into U2OS cells to reduce endogenous PRX-I significant MST1 activation even in the absence of expression. As shown in Supplementary Figure 4, H2O2. The results shown in Figure 1 show very weak but treatment with PRX-I siRNA specifically suppressed significant positive signals of WT PRX-I oligomer the expression of PRX-I, without any significant effects formation (Figure 1a) and complex formation with on PRX-II expression. Cells were then stimulated with

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4211

Figure 2 PRX-I mediates H2O2-induced activation of MST1. (a) COS-7 cells were transfected with the indicated constructs and treated with 100 mM H2O2 for 30 min. Cell lysates were analyzed by immunoblotting using the indicated primary antibodies. (b) COS-7 cells were transfected with the indicated constructs and treated with 100 mM H2O2 for 30 min. Cell lysates were immunoprecipitated with an anti-FLAG antibody and subjected to in vitro kinase assays using [g-32P]ATP and recombinant MOBKL1B as substrate. The phosphorylation level of MOBKL1B was analyzed by autoradiography and the relative radioactivity is indicated. The data are the mean±s.e.m. for n ¼ 3. *Po0.05, **Po0.01 indicates significant difference by one-way analysis of variance (P ¼ 0.0001) followed by Dunnett’s multiple comparison test. (c) U2OS cells were transfected with the indicated constructs together with GFP-expressing plasmids. The cells were fixed and DAPI-stained to visualize chromatin condensation. GFP-positive cells were examined for apoptosis. The data are the mean±s.e.m. for n ¼ 3. For each experiment, more than 100 cells were examined. *Po0.05 indicates significant difference by one-way analysis of variance (P ¼ 0.004) followed by Dunnett’s multiple comparison test among the Myc-MST1- transfected groups. (d) U2OS cells were transfected with PRX-I siRNA and treated with 100 mM H2O2 for 30 min. Cell lysates were analyzed using the indicated primary antibodies. DAPI, 40,6-diamidino-2-phenylindole; GFP, green fluorescent protein; MST1, mammalian Ste20-like kinase-1; PRX-I, peroxiredoxin-I; siRNA, short interfering RNA.

H2O2 and MST1/2 autophosphorylation was examined H2O2 generation, we next examined PRX-I oligomer by immunoblotting. As shown in Figure 2d, H2O2 formation and found that cisplatin treatment could stimulation resulted in significant activation of endogen- induce oligomers of both ectopically expressed PRX-I in ous MST1/2 in control cells, but it was clearly suppressed COS-7 cells and endogenous PRX-I in U2OS cells in PRX-I-knockdown cells. Collectively, these results (Figures 3b and c). indicate that PRX-I is an essential intermediate linking We noticed that p53 levels significantly increased in H2O2 stimulation to MST1 activation. U2OS cells upon cisplatin treatment (Figure 3c) as reported previously (Praskova et al., 2008). Therefore, Cisplatin induces PRX-I oligomer formation through p53 we next examined whether p53 has any important role in It has been reported that MST1 is important for PRX-I oligomer formation induced by cisplatin treat- anticancer drugs to induce cell death (Ren et al., ment. As shown in Figure 3d, p53 ectopic expression by 2008), prompting us to examine the possible roles of itself could induce PRX-I oligomerization, which was H2O2 and PRX-I in this process. First, we confirmed the abrogated by treatment with the antioxidant N-acetyl requirement of MST1/2 for cisplatin-induced apoptosis cysteine (NAC). We also examined the apoptotic rate of in U2OS cells (Supplementary Figure 5). We then the cells and found that p53 expression raised the performed H2O2-imaging analysis using GFP-HyPer, number of apoptotic cells with condensed chromatin which specifically responds to H2O2 by increasing its (from 4.3% in control cells to 22.6% in p53-expressing fluorescent signal intensity (Belousov et al., 2006). As cells; Figure 3e). This cell death was again inhibited by shown in Figure 3a, treatment of COS-7 cells with NAC treatment, and thus, PRX-I oligomer formation cisplatin gradually increased the GFP-HyPer signal, correlated well with p53 apoptosis-inducing ability, which was evident after 4 h. By contrast, we did not suggesting a functional link between p53 and PRX-I. observe any significant increase in GFP-HyPer-expres- To indicate a definite endogenous p53 requirement for sing cells not treated with cisplatin or control GFP- cisplatin-induced H2O2 generation and PRX-I oligomer- expressing cells treated with cisplatin. Having confirmed ization, we isolated MEFs from WT (p53 þ / þ ) or p53-

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4212 À/À þ / þ À/À knockout (p53 ) mice and examined H2O2 generation in p53 but not in p53 MEFs (Supplementary and endogenous PRX-I oligomer formation. The results Figure 6 and Figure 3f). This PRX-I oligomer formation clearly showed that cisplatin treatment could induce was abrogated by NAC treatment (Supplementary H2O2 generation and PRX-I oligomer formation Figure 7). We further investigated possible relationships

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4213 with human cancers, which often lack functional p53, by cell-death responses induced by cisplatin treatment. performing similar experiments using human cancer- We also examined the importance of p53 in cisplatin- derived cell lines such as U2OS (osteosarcoma, p53- induced MST1 activation. MEFs obtained from p53 þ / þ positive), SaOS-2 (osteosarcoma, p53-negative), MCF-7 or p53À/À mice were treated with cisplatin and subse- (breast cancer, p53-positive) and MDA-MB-231 (breast quently, levels of phosphorylated endogenous MST1/2 cancer, p53-mutated). As shown in Figure 3g, we could were examined by immunoblotting. As shown in confirm that cisplatin-induced PRX-I oligomer forma- Figure 4e, cisplatin-induced phosphorylation of MST1/ tion occurred in U2OS and MCF-7 cells but not in 2 was very weak in p53À/À MEFs, which is consistent SaOS-2 and MDA-MB-231 cells, which is consistent with the p53-dependent PRX-I oligomer formation with the functional status of p53. (Figure 3f). We also performed similar experiments by using human cancer cells and confirmed that p53- PRX-I mediates cisplatin-induced MST1 activation and positive U2OS and MCF-7 cells activated MST1/2 in cell death response to cisplatin, but SaOS-2 (p53-negative) and It has been reported that cisplatin treatment induces MDA-MB-231 (p53-mutated) cells did not (Figure 4f). MST1 activation (Ren et al., 2008). The results shown in Figure 3 clearly show that cisplatin induces PRX-I Mechanism of MST1 activation by PRX-I and H2O2 oligomer formation through p53. Therefore, we specu- Finally, we tried to clarify the molecular mechanisms of lated that PRX-I may mediate cisplatin-induced MST1 MST1 activation by PRX-I and H2O2. For this purpose, activation. To test this hypothesis, we first examined the we purified various preparations of recombinant MST1 complex formation status by ectopically expressing and PRX-I proteins (Figure 5a). We first incubated full- MST1 and PRX-I in COS-7 cells, and confirmed that length MST1 with or without PRX-I under a reducing cisplatin treatment significantly augmented the co- (10 mM DTT) or oxidizing (10 mM H2O2) condition, and immunoprecipitation of the two proteins (Figure 4a). then examined MST1 autophosphorylation. As shown We next examined complex formation between endo- in Figure 5b, the levels of phosphorylated MST1 genous MST1 and PRX-I in U2OS cells. When cells specifically increased when PRX-I was with H2O2. were not treated with cisplatin, only very weak co- Moreover, the results of kinase assays using MOBKL1B immunoprecipitation signal was observed, but cisplatin as substrate also confirmed MST1 activation by PRX-I treatment significantly increased this signal (Figure 4b). and H2O2 (Supplementary Figure 8). Therefore, we To directly examine the importance of PRX-I in successfully reconstituted H2O2-induced MST1 activa- cisplatin-induced MST1 activation, we next performed tion by PRX-I, indicating a crucial role for PRX-I in PRX-I-knockdown analysis and found that MST1 MST1 activation. activation was severely impaired in cells treated with We then further explored molecular mechanisms PRX-I siRNA (Figure 4c). We also examined the PRX- underlying this activation. The NH2-terminal half of I-knockdown effect on the apoptotic rate. We observed MST1 contains a kinase domain with the autopho- a slight increase in the apoptotic rate by PRX-I sphorylation site and the COOH-terminus includes an knockdown itself (from 4.7% in control cells to 12.3% inhibitory domain for the kinase activity (Figure 5a) in PRX-I-knockdown cells; Figure 4d). As PRX-I has (Creasy et al., 1996; Praskova et al., 2004). Thus, we an important role in scavenging H2O2, this increase is created two glutathione-S-transferase (GST)-fusion probably caused by augmented oxidative stress. Cispla- MST1 fragments, lacking either the NH2- or the tin treatment induced a significant increase in the COOH-terminal region (GST-MST1 C or GST-MST1 apoptotic rate in control cells. However, PRX-I-knock- N, respectively) (Figure 5a), and examined their binding down cells relatively resisted against cisplatin and their ability to recombinant PRX-I by pull-down assays. As apoptotic rate was significantly lower than in control shown in Figure 5c, GST-MST1 C, but not GST-MST1 cells, thus uncovering an unexpected role for PRX-I in N, could specifically precipitate PRX-I in the presence

Figure 3 Cisplatin induces oligomer formation of PRX-I through p53. (a) COS-7 cells were transfected with GFP-HyPer-expressing plasmids and then treated with 25 mM cisplatin for the indicated durations. The relative intensity of the GFP fluorescence in cells was analyzed. The data are the mean±s.e.m. for n ¼ 3. For each experiment, more than 50 cells were examined. *Po0.05, **Po0.01 (against the ‘HyPer’ group) and wPo0.05 (against the ‘GFP þ Cisplatin’ group) indicate significant difference by one-way analysis of variance (Po0.05) followed by Bonferroni’s multiple comparison test. (b) COS-7 cells transfected with the indicated constructs were treated with cisplatin for the indicated durations. Cell lysates were treated with IAA and immunoblotted under native or denaturing conditions. (c) U2OS cells were treated with cisplatin for the indicated durations. Cell lysates were treated with IAA and immunoblotted under native or denaturing conditions. (d) U2OS cells were transfected with the indicated constructs and treated with NAC. Cell lysates were treated with IAA and immunoblotted under native or denaturing conditions. (e) U2OS cells were transfected with the indicated constructs together with GFP-expressing plasmids and treated with NAC. The cells were fixed and DAPI-stained to visualize chromatin condensation. GFP-positive cells were examined for apoptosis. The data are the mean±s.e.m. for n ¼ 3. For each experiment, more than 100 cells were examined. **Po0.01 indicates significant difference by one-way analysis of variance (P ¼ 0.0009) followed by Bonferroni’s multiple comparison test. (f) MEFs from p53 þ / þ or p53À/À mice were treated with cisplatin. Cell lysates were treated with IAA and immunoblotted under native or denaturing conditions. (g) Human cancer-derived cell lines such as U2OS, SaOS- 2, MCF-7 and MDA-MB-231 were treated with cisplatin. Cell lysates were treated with IAA and immunoblotted under native or denaturing conditions. DAPI, 40,6-diamidino-2-phenylindole; GFP, green fluorescent protein; IAA, iodoacetamide; MEF, murine embryonic fibroblast; NAC, N-acteyl cysteine; PRX-I, peroxiredoxin-I.

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4214

Figure 4 PRX-I mediates cisplatin-induced MST1 activation and cell death. (a) COS-7 cells were transfected with the indicated constructs and treated with cisplatin for the indicated durations. Cell lysates were immunoprecipitated with anti-FLAG antibodies and analyzed by immunoblotting using the indicated primary antibodies. (b) Lysates of U2OS cells treated with cisplatin were immunoprecipitated with anti-MST1/2 antibodies and immunoblotted with the indicated primary antibodies. (c) U2OS cells were transfected with PRX-I siRNA and treated with cisplatin. Cell lysates were analyzed using the indicated primary antibodies. (d) U2OS cells were transfected with PRX-I siRNA together with GFP-expressing plasmids and treated with cisplatin. The cells were fixed and DAPI-stained to visualize chromatin condensation. Apoptosis was assessed in GFP-positive cells. The data are the mean±s.e.m. for n ¼ 3. For each experiment, more than 100 cells were examined. *Po0.05, **Po0.01 indicate significant difference by one-way analysis of variance (P ¼ 0.0029) followed by Bonferroni’s multiple comparison test. (e) MEFs from p53 þ / þ or p53À/À mice were treated with cisplatin. Cell lysates were analyzed with the indicated primary antibodies. (f) Human cancer-derived cell lines such as U2OS, SaOS-2, MCF-7 and MDA-MB-231 were treated with cisplatin. Cell lysates were analyzed with the indicated primary antibodies. DAPI, 40,6-diamidino-2-phenylindole; GFP, green fluorescent protein; MEF, murine embryonic fibroblast; MST1, mammalian Ste20-like kinase-1; PRX-I, peroxiredoxin-I; siRNA, short interfering RNA.

of H2O2. As MST1 N has been shown to possess Discussion increased kinase activity compared with full-length MST1 in similar kinase assays in vitro (Creasy et al., Here we showed a quite unexpected MST1 activation 1996), we assumed that the direct interaction of the mechanism by H2O2, mediated by oligomeric PRX-I. COOH-terminal region with the NH2-terminal kinase We elucidated novel PRX-I functions in signal trans- domain inhibits the kinase activity and that H2O2- duction promoting appropriate responses to oxidative induced association with PRX-I abrogates this interac- stress. We also showed the important role of PRX-I in tion, resulting in the opening of the kinase domain. the p53-dependent cell death triggered by the anti- Indeed, interaction of recombinant MST1 C with GST- cancer drug cisplatin. Recent studies have suggested MST1 N was confirmed (Figure 5d). Furthermore, this a potential approach to treat cancers by manipulat- interaction was significantly weakened by addition of ing their redox environment. In these studies, it has PRX-I with H2O2 but not with DTT. been indicated that cancer cells are very sensitive We next assessed the effect of PRX-I oligomers on the to artificial ROS elevation because most cancer kinase activity of MST1. Purified MST1 N showed an cells are chronically subjected to oxidative stress autophosphorylation signal and addition of MST1 C and an additional increase in ROS levels brings the suppressed it (Figure 5e), thus supporting the auto- level of oxidative stress to a fatal degree (reviewed by inhibitory regulation of the MST1 kinase activity. We Trachootham et al., 2009). For example, b-phenylethyl then added PRX-I to the mixture of MST1 N and MST1 isothiocyanate, a natural compound that promotes C, and found that PRX-I could clearly negate the ROS generation, is reported to induce apoptosis suppression imposed by MST1 C when incubation was selectively in cancer cells (Trachootham et al., 2006). under an H2O2-driven oxidative condition. Based on Although cell death was attributed to aberrant regula- these results, we concluded that oligomeric PRX-I tion of H-Ras, JNK and nuclear factor-kB, PRX-I- activates MST1 by dissociating the inhibitory COOH- mediated MST1 activation probably also takes part in terminal region of MST1 from the NH2-terminal kinase this process, and further examinations of these pathways domain, thus mediating the H2O2-induced MST1 could facilitate the development of better chemother- activation. apeutic agents.

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4215

Figure 5 Mechanism of MST1 activation by PRX-I and H2O2.(a) Purified recombinant proteins expressed in E. coli or Sf9 cells were analyzed by SDS–PAGE and Coomassie Brilliant Blue (CBB) staining. The schematics illustrate MST1 constructs. (b) In vitro kinase assays were performed by mixing the indicated recombinant proteins in the presence of 10 mM DTT or 10 mM H2O2. Proteins were analyzed using the indicated primary antibodies. (c) GST-MST1 pull-down assays for PRX-I. Proteins were subjected to SDS–PAGE and CBB staining. (d) GST-MST1 pull-down assays for MST1 C in the presence of PRX-I. Proteins were subjected to SDS–PAGE and CBB staining. (e) In vitro kinase assays were performed using the indicated recombinant proteins in the presence of 10 mM H2O2 or 10 mM DTT. Proteins were analyzed using the indicated primary antibodies or by CBB staining. DTT, dithiothreitol; GST, glutathione-S-transferase; MST1, mammalian Ste20-like kinase-1; PRX-I, peroxiredoxin-I; SDS–PAGE, sodium dodecyl sulfate–PAGE.

Recent studies have indicated the importance of a in mammals, and the human genes for several pathway mammalian signaling pathway equivalent to the Droso- components have been found to be mutated in cancers phila Hippo pathway in tumor suppression (Harvey and (O’Neill et al., 2005; Harvey and Tapon, 2007; Zeng and Tapon, 2007; Zeng and Hong, 2008). In Drosophila, Hong, 2008). PRX-I is also suggested to suppress cancer Hippo activates Warts (LATS1/2 in mammals) together development because mice lacking the PRX-I have with Salvador (WW45 in mammals). Activated Warts a short life span owing to frequent development of then excludes an oncogenic Yorkie several cancers (Neumann et al., 2003; Egler et al., 2005; (YAP in mammals) from the nucleus and induces Graves et al., 2009). In these reports, oncogenesis has apoptosis. The Hippo pathway appears to be conserved been attributed to loss of PRX-I peroxidase activity.

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4216 However, mice lacking the PRX-II gene, the closest mutagenesis was performed using the Quick-Change Mutagenesis homolog of PRX-I, do not show cancers in any cell type kit (Stratagene, Santa Clara, CA, USA). MST1 lacking the or tissue (Choi et al., 2005; Lee et al., 2007). PRX-II is COOH-terminus (MST1 N, 1–308) and NH2-terminus (MST1 expressed in a wide variety of tissues and has a more C, 302–488) were generated by PCR. The cDNA fragments potent peroxidase activity than PRX-I (Rhee et al., were inserted into pEF-BOS, pCAGGS (Clontech Labora- tories, Mountain View, CA, USA), pGEX-2T (Amersham, 2005; Lee et al., 2007). Indeed, PRX-IIÀ/À mice show a Piscataway, NJ, USA) and pFastBac1 (Invitrogen, Carlsbad, massive increase in cellular ROS levels (Choi et al., CA, USA) plasmid vectors. The siRNA duplex oligonucleo- 2005). Therefore, loss of peroxidase activity alone tides against human PRX-I were purchased from Invitrogen. does not account for the cancer-forming phenotype in The control siRNA had the shuffled sequence of the human PRX-IÀ/À mice. It should be noted that PRX-I, but not PRX-I siRNA and was designed as a non-silencing siRNA PRX-II, is able to bind to and activate MST1 (Figure 1c that does not correspond to any known mammalian mRNA and Supplementary Figure 2), which may explain sequence. The target mRNA sequences are as follows: PRX-I, the phenotypic differences between PRX-IÀ/À and 50-ATGTTTGTCAGTGAACTGGAAGGCC-30;MST1,50-AA 0 0 PRX-IIÀ/À mice. TGATATCAGATACAGAACCAGCC-3 ; MST2, 5 -ATTAT GTTCAATCATGGTCTGGGCC-30; Control, 50-GAGTCTC PRX-I and PRX-II are highly homologous proteins 0 (91% homology and 78% identity in human) and thus GTGGAATCGAACGTTAAA-3 . expected to have similar roles intracellularly. However, our study clearly shows a difference in their role related Antibodies and reagents to MST1 (Figure 1c and Supplementary Figure 2). One Mouse anti-FLAG (M2) (Sigma, St Louis, MO, USA), rabbit possible reason may be in their oligomerization anti-Myc (A-14) (Santa Cruz Biotechnology, Santa Cruz, CA, tendency because PRX-I promptly forms oligomers in USA), rabbit anti-MST1/2 (Bethyl Laboratories, Montgomery, TX, USA), rabbit anti-PRX-I, rabbit anti-PRX-SO2/SO3 (Abcam, response to low H2O2 levels, whereas PRX-II is Cambridge, MA, USA), rabbit anti-phospho-MST1/2 (Thr183), comparatively resistant to H2O2-induced oligomeriza- rabbit anti-JNK1/2, rabbit anti-phospho-JNK1/2 (Thr183/ tion (Lee et al., 2007). This difference can be attributed Tyr185), mouse anti-p53 (1C12) (Cell Signaling Technology, to the presence of PRX-I Cys83, which is absent in Beverley, MA, USA, for detection of mouse p53), mouse PRX-II. PRX-I Cys83 reportedly forms a disulfide anti-p53 (Ab-6, for detection of human p53) (Calbiochem, linkage with another PRX-I Cys83, thus promoting Darmstadt, Germany) and mouse anti-actin (Millipore, Billerica, homo-oligomer formation. Consistently, our study also MA, USA) were used. Cisplatin (Yakult Honsha, Tokyo, Japan) confirmed that C83S PRX-I did not efficiently form and NAC (Sigma) were from commercial sources. oligomers (Figure 1a), only weakly binding to and activating MST1 (Figures 1b and 2a). Recombinant proteins PRX-I is a well-known antioxidant enzyme thought to Recombinant proteins of MST1, PRX-I and MOBKL1B were protect cells against oxidative stress by reducing H2O2 expressed as GST fusions in Escherichia coli (PRX-I and (Rhee et al.,2005).Indeed,thereareseveralreports MOBKL1B) or in Sf9 cells (MST1, MST1 N and MST1 C). indicating that PRX-I inactivation results in the elevation They were purified using glutathione–sepharose beads (GE of ROS levels and oxidative DNA damage (Neumann Healthcare, Little Chalfont, UK) and eluted with excess glutathione. The GST tag was removed by incubating GST- et al., 2003; Egler et al., 2005; Rhee et al., 2005; Graves fusion proteins bound to the glutathione–sepharose beads with et al.,2009).However,weshowedhereaquiteunexpected thrombin (GE Healthcare) for PRX-I or with PreScission PRX-I function, linking ROS to cell death. This difference protease (GE Healthcare) for MOBKL1B, MST1, MST1 N in PRX-I function apparently depends on the degree of and MST1 C for 4 h at 4 1C. Thrombin was removed by using oxidative stress. Indeed, the extent of PRX-I oligomeriza- benzamidine–sepharose beads (Sigma). tion depends on the concentration of H2O2 and correlates well with cell death (Supplementary Figure 9). PRX-I Native PAGE reduces H2O2 and protects cells from oxidative stress when Cell lysates in native PAGE sample buffer (50 mM Tris–HCl H2O2 level is within a manageable range. However, when (pH 6.8), 10% glycerol, 1% deoxycholate and 15 mM cells are exposed to excessive H2O2 levels, PRX-I is iodoacetamide (IAA)) were subjected to native PAGE. The inactivated as a peroxidase and forms oligomers (Jang cathode buffer contained 0.2% deoxycholate (Iwamura et al., et al., 2004; Moon et al., 2005), thus resulting in MST1 2001). activation. In this case, we postulate that PRX-I contributes to protection of the whole organism by Cell culture and transfection removing cells severely damaged by excess ROS, avoiding COS-7, U2OS, SaOS-2 and p53 þ / þ or p53À/À MEFs were cancer development. cultured in Dulbecco’s Modified Eagle’s Medium with 10% fetal bovine serum, and MCF-7 and MDA-MB-231 cells were cultured in RPMI-1640 with 10% fetal bovine serum. SaOS-2 and MCF-7 cells were obtained from the Japanese Collection Materials and methods of Research Bioresources, and MDA-MB-231 cells were from American Type Culture Collection. COS-7 and U2OS cells Expression constructs were generous gifts from Dr Tadaomi Takenawa (Kobe The cDNAs for mouse PRX-I, mouse PRX-II, human MST1, University) and Kiyoko Fukami (Tokyo University of human JNK1 and human MOBKL1B were generated by Pharmacy and Life Sciences), respectively. The cancer cells reverse transcription–PCR. The GFP-HyPer-expressing plasmid are routinely authenticated through cell morphology monitor- was purchased from Evrogen (Moscow, Russia). Site-directed ing and growth curve analysis, and we confirmed that p53

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4217 status agrees with the information on the International Agency amounts of GST were added to adjust the total protein for Research on Cancer TP53 database (http://www-p53. concentration. The kinase reaction was initiated by addition of iarc.fr/). Primary MEFs were derived from 13.5-day embryos 1 mCi [g-32P]ATP and then incubated for 5 min at 30 1C. The and genotyped by PCR. p53 þ /À mice were purchased from reaction was then stopped by adding the SDS sample buffer. RIKEN BioResource Center (BRC) (Tsukada et al., 1993). The samples were subjected to SDS–PAGE and the radio- p53À/À embryos were derived from p53 þ /À crosses maintained activity of MOBKL1B was determined by autoradiography on a C57BL/6C background. Cells were seeded in 36- or 60- (BAS2000; FUJIFILM Corporation, Tokyo, Japan). mm dishes, cultured overnight and transfected with Lipofecta- mine 2000 (Invitrogen). The cells were harvested with lysis Real-time PCR analyses buffer (20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, Real-time PCR experiments were performed using MiniOpti- 0.5% Triton X-100) 1 or 2 days after transfection. con (Bio-Rad, Hercules, CA, USA) using the iQ SYBR green Supermix (Bio-Rad). The quality of the final PCR product was H2O2-imaging analysis checked by agarose gel electrophoresis and it was confirmed Cells were transfected with a GFP-HyPer-expressing plasmid. that there were no obvious non-specifically amplified DNAs. The cells were cultured for 24 h and then treated with 25 mM The primers used are as follows: PRX-I (50-GACCCATGAA 0 0 0 cisplatin or 100 mM H2O2. The relative intensities of fluores- CATTCCTTTG-3 and 5 -AGGCTTGATGGTATCACTGC-3 ), cence emitted from the cells were analyzed by real-time PRX-II (50-GTCCGTGCGTCTAGCCTTTG-30 and 50-TCCC microscopy (the signal intensity just before stimulation was TTTGTAGTCCGACAGC-30) and GAPDH (50-AGGTGAA set to be 1.0). We performed time-lapse imaging analyses using GGTCGGAGTCAACG-30 and 50-AGGGGTCATTGATGG an Olympus IX81 microscope equipped with an Olympus CAACA-30). DP30BW camera. Fluorescence observation was performed using a conventional filter set for green fluorescent protein Animal experiments (GFP). We quantified image intensities with the DP Controller We appropriately treated mice according to the guidelines for software (Olympus, Tokyo, Japan) and the ImageJ software proper conduct of animal experiments (issued by the Science (National Institutes of Health, Bethesda, MD, USA). For each Council of Japan), and we received approval of this experi- experiment, more than 50 cells were examined. ment from Osaka University.

Apoptosis assay Cells were transfected with various expression plasmids or siRNAs together with GFP-expressing plasmids. The cells Conflict of interest were cultured for 24 h and then treated with 25 mM cisplatin for 48 h. The cells were fixed with formaldehyde and then The authors declare no conflict of interest. stained with 40,6-diamidino-2-phenylindole (DAPI) to visua- lize nuclei. GFP-positive cells were assessed for their chromatin condensation. For each experiment, more than Acknowledgements 100 cells were measured. We thank Dr Yoshiaki Kise (Flanders Institute for Biotech- In vitro kinase assay nology, Belgium) for technical assistance in preparing the The kinase reaction was assessed in a kinase buffer (20 mM expression constructs. This study was supported in part by a Tris–HCl (pH 7.5), 15 mM MgCl2) containing either 10 mM Grant-in-Aid for Scientific Research from the Japan Society DTT or 10 mM H2O2. Recombinant MOBKL1B was used as for the Promotion of Science (JSPS) and Ministry of Education, the substrate. In the absence of PRX-I or MST1 C, equal Culture, Sports, Science and Technology (MEXT)-Japan.

References

Ahn SH, Cheung WL, Hsu JY, Diaz RL, Smith MM, Allis CD. Egler RA, Fernandes E, Rothermund K, Sereika S, de Souza-Pinto N, (2005). Sterile 20 kinase phosphorylates histone H2B at serine 10 Jaruga P et al. (2005). Regulation of reactive oxygen species, DNA during hydrogen peroxide-induced apoptosis in S. cerevisiae. Cell damage, and c-Myc function by peroxiredoxin 1. Oncogene 24: 120: 25–36. 8038–8050. Belousov VV, Fradkov AF, Lukyanov KA, Staroverov DB, Shakh- Finkel T, Holbrook NJ. (2000). Oxidants, oxidative stress and the bazov KS, Terskikh AV et al. (2006). Genetically encoded biology of ageing. Nature 408: 239–247. fluorescent indicator for intracellular hydrogen peroxide. Nat Graves JA, Metukuri M, Scott D, Rothermund K, Prochownik EV. Methods 3: 281–286. (2009). Regulation of reactive oxygen species homeostasis by Choi MH, Lee IK, Kim GW, Kim BU, Han YH, Yu DY et al. (2005). and c-Myc. J Biol Chem 284: 6520–6529. Regulation of PDGF signalling and vascular remodelling by Guyton KZ, Liu Y, Gorospe M, Xu Q, Holbrook NJ. (1996). peroxiredoxin II. Nature 435: 347–353. Activation of mitogen-activated by H2O2. Role Colombani J, Polesello C, Josue´F, Tapon N. (2006). Dmp53 activates in cell survival following oxidant injury. J Biol Chem 271: the Hippo pathway to promote cell death in response to DNA 4138–4142. damage. Curr Biol 16: 1453–1458. Harvey K, Tapon N. (2007). The Salvador–Warts–Hippo pathway— Creasy CL, Ambrose DM, Chernoff J. (1996). The Ste20-like protein an emerging tumour-suppressor network. Nat Rev Cancer 7: kinase, Mst1, dimerizes and contains an inhibitory domain. J Biol 182–191. Chem 271: 21049–21053. Iwamura T, Yoneyama M, Yamaguchi K, Suhara W, Mori W, Shiota de Souza PM, Lindsay MA. (2004). Mammalian Sterile20-like K et al. (2001). Induction of IRF-3/-7 kinase and NF-kappaB in kinase 1 and the regulation of apoptosis. Biochem Soc Trans 32: response to double-stranded RNA and virus infection: common and 485–488. unique pathways. Genes Cells 6: 375–388.

Oncogene PRX-I mediates H2O2-induced activation of MST1 A Morinaka et al 4218 Jang HH, Lee KO, Chi YH, Jung BG, Park SK, Park JH et al. (2004). O’Neill EE, Matallanas D, Kolch W. (2005). Mammalian sterile 20- Two enzymes in one; two yeast peroxiredoxins display oxidative like kinases in tumor suppression: an emerging pathway. Cancer Res stress-dependent switching from a peroxidase to a molecular 65: 5485–5487. chaperone function. Cell 117: 625–635. Polyak K, Xia Y, Zweier JL, Kinzler KW, Vogelstein B. (1997). Johnson TM, Yu ZX, Ferrans VJ, Lowenstein RA, Finkel T. (1996). A model for p53-induced apoptosis. Nature 389: 300–305. Reactive oxygen species are downstream mediators of p53- Praskova M, Khoklatchev A, Ortiz-Vega S, Avruch J. (2004). dependent apoptosis. Proc Natl Acad Sci USA 93: 11848–11852. Regulation of the MST1 kinase by autophosphorylation, by the Karin M, Hunter T. (1995). Transcriptional control by protein growth inhibitory proteins, RASSF1 and NORE1, and by Ras. phosphorylation: signal transmission from the cell surface to the Biochem J 381: 453–462. nucleus. Curr Biol 5: 747–757. Praskova M, Xia F, Avruch J. (2008). MOBKL1A/MOBKL1B Ko¨nig J, Baier M, Horling F, Kahmann U, Harris G, Schu¨rmann P phosphorylation by MST1 and MST2 inhibits cell proliferation. et al. (2002). The plant-specific function of 2-Cys peroxiredoxin- Curr Biol 18: 311–321. mediated detoxification of peroxides in the redox-hierarchy Ren A, Yan G, You B, Sun J. (2008). Downregulation of mammalian of photosynthetic electron flux. Proc Natl Acad Sci USA 99: sterile 20-like kinase 1 by heat shock protein 70 mediates cisplatin 5738–5743. resistance in prostate cancer cells. Cancer Res 68: 2266–2274. Lee W, Choi KS, Riddell J, Ip C, Ghosh D, Park JH et al. Rhee SG, Chae HZ, Kim K. (2005). Peroxiredoxins: a historical overview (2007). Human peroxiredoxin 1 and 2 are not duplicate proteins: and speculative preview of novel mechanisms and emerging concepts the unique presence of CYS83 in Prx1 underscores the structural in cell signaling. Free Radic Biol Med 38: 1543–1552. and functional differences between Prx1 and Prx2. J Biol Chem 282: Rhee SG. (2006). Cell signaling. H2O2, a necessary evil for cell 22011–22022. signaling. Science 312: 1882–1883. Lehtinen MK, Yuan Z, Boag PR, Yang Y, Ville´n J, Becker EB et al. Trachootham D, Alexandre J, Huang P. (2009). Targeting cancer cells (2006). A conserved MST–FOXO signaling pathway mediates by ROS-mediated mechanisms: a radical therapeutic approach? Nat oxidative-stress responses and extends life span. Cell 125: 987–1001. Rev Drug Discov 8: 579–591. Li PF, Dietz R, von Harsdorf R. (1999). p53 regulates mitochondrial Trachootham D, Zhou Y, Zhang H, Demizu Y, Chen Z, Pelicano H membrane potential through reactive oxygen species and induces et al. (2006). Selective killing of oncogenically transformed cells cytochrome c-independent apoptosis blocked by Bcl-2. EMBO J 18: through a ROS-mediated mechanism by beta-phenylethyl isothio- 6027–6036. cyanate. Cancer Cell 10: 241–252. Lowe SW, Bodis S, McClatchey A, Remington L, Ruley HE, Fisher Tsukada T, Tomooka Y, Takai S, Ueda Y, Nishikawa S, Yagi T et al. DE et al. (1994). p53 status and the efficacy of cancer therapy (1993). Enhanced proliferative potential in culture of cells from p53- in vivo. Science 266: 807–810. deficient mice. Oncogene 8: 3313–3322. Lowe SW, Ruley HE, Jacks T, Housman DE. (1993). p53-dependent Veal EA, Day AM, Morgan BA. (2007). Hydrogen peroxide sensing apoptosis modulates the cytotoxicity of anticancer agents. Cell 74: and signaling. Mol Cell 26: 1–14. 957–967. Veal EA, Findlay VJ, Day AM, Bozonet SM, Evans JM, Quinn J et al. Moon JC, Hah YS, Kim WY, Jung BG, Jang HH, Lee JR et al. (2005). (2004). A 2-Cys peroxiredoxin regulates peroxide-induced oxida- Oxidative stress-dependent structural and functional switching tion and activation of a stress-activated MAP kinase. Mol Cell 15: of a human 2-Cys peroxiredoxin isotype II that enhances HeLa 129–139. cell resistance to H2O2-induced cell death. J Biol Chem 280: Wood ZA, Schro¨der E, Robin Harris J, Poole LB. (2003). Structure, 28775–28784. mechanism and regulation of peroxiredoxins. Trends Biochem Sci Neumann CA, Krause DS, Carman CV, Das S, Dubey DP, Abraham 28: 32–40. JL et al. (2003). Essential role for the peroxiredoxin Prdx1 in Zeng Q, Hong W. (2008). The emerging role of the hippo pathway in erythrocyte antioxidant defence and tumour suppression. Nature cell contact inhibition, organ size control, and cancer development 424: 561–565. in mammals. Cancer Cell 13: 188–192.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene