Role of the Kinase MST2 in Suppression of Apoptosis by the Proto-Oncogene Product Raf-1

See allHide authors and affiliations

Science  24 Dec 2004:
Vol. 306, Issue 5705, pp. 2267-2270
DOI: 10.1126/science.1103233


The ablation of the protein kinase Raf-1 renders cells hypersensitive to apoptosis despite normal regulation of extracellular signal–regulated kinases, which suggests that apoptosis protection is mediated by a distinct pathway. We used proteomic analysis of Raf-1 signaling complexes to show that Raf-1 counteracts apoptosis by suppressing the activation of mammalian sterile 20–like kinase (MST2). Raf-1 prevents dimerization and phosphorylation of the activation loop of MST2 independently of its protein kinase activity. Depletion of MST2 from Raf-1–/– mouse or human cells abrogated sensitivity to apoptosis, whereas overexpression of MST2 induced apoptosis. Conversely, depletion of Raf-1 from Raf-1+/+ mouse or human cells led to MST2 activation and apoptosis. The concomitant depletion of both Raf-1 and MST2 prevented apoptosis.

Mitogen-activated protein kinase (MAPK) pathways are primordial signaling systems that enable cells to respond to external cues. In metazoans, the proteins Ras, Raf, and MEK act sequentially to activate the MAPK extracellular signal–regulated kinase (ERK). This pathway has a crucial role in the control of cell proliferation, differentiation, and survival (13), which is demonstrated by its frequent hyperactivation in human tumors, most notably those caused by active mutations in Ras (4) or B-Raf (5). The Raf family of serine-threonine kinases comprises three members: Raf-1, B-Raf, and A-Raf. All Raf isoforms are activated by binding to the guanosine triphosphate (GTP)–bound form of Ras, and they share MEK as the only commonly recognized substrate (13). However, studies in knock-out mice revealed distinct physiological functions of the Raf isozymes (6), which suggest the existence of other effectors.

Ablation of the Raf-1 gene causes widespread apoptosis and embryonic lethality despite normal regulation of ERK through B-Raf (7, 8). Raf-1–/– fibroblasts are hypersensitive to apoptosis that is induced by selected stimuli, including serum withdrawal and stimulation of the death receptor Fas (7, 8). This hypersensitivity suggests that Raf-1 can protect fibroblasts from apoptosis independently of B-Raf and ERK. When knocked back in, a Raf-1 mutant, Raf-1YY340-341FF, which cannot be activated, fully rescued the Raf-1–/– phenotype, which resulted in viable mice (7). This result indicates that full activity of Raf-1 may not be required for suppression of apoptosis.

To search for new partners in Raf-1 signaling, we immunopurified proteins associated with Flag-tagged Raf-1 that were expressed in COS-1 cells, and identified them by mass spectrometry (Fig. 1A). Proteins known to interact with Raf-1, including 14-3-3, heat shock protein 50 (Hsp50), and Hsp90 (3), associated with Raf-1 under conditions of both serum starvation and stimulation. A 55-kD band, which preferentially coprecipitated with Raf-1 from serum-starved cells, contained the mammalian sterile 20–like kinase (MST2). In untransfected serum-deprived cells, endogenous MST2 coimmunoprecipitated with endogenous Raf-1, and Raf-1 was detected in MST2 immunoprecipitates (Fig. 1B). In addition, Raf-1YY340/341FF and Raf-1 K375M, a catalytically inactive mutant, could interact with MST2 in transfected cells (Fig. 1C). Pulldown assays with bacterially expressed Raf-1 deletion proteins (fig. S1) showed that MST2 bound to amino acids 151 and 303 of Raf-1. This region diverges between Raf isoforms and is thought to mediate isozyme-specific interactions with other proteins (9). MST2 did not bind B-Raf (fig. S2), which suggests that MST2 is part of a Raf-1–specific signaling pathway.

Fig. 1.

MST2 is a component of Raf-1 signaling complexes. (A) Flag–Raf-1 was immunoprecipitated from COS-1 cells kept in 10% (+) or 0.1% (-) serum. Flag–Raf-1 immunoprecipitates were separated by SDS-polyacrylamide gel electrophoresis and silver stained. Associated proteins were identified by mass spectrometry. (B) Immunoprecipitation (IP) of endogenous MST2 and endogenous Raf-1 from COS-1 cells kept in 10 or 0.1% serum. Raf-1 and MST2 were detected by Western blotting. (C) Immunoprecipitation of endogenous MST2 from COS-1 cells expressing Flag–Raf-1, K375M (catalytically inactive mutant), or YY340-341FF (non-activatable Raf-1 mutant). Cells were maintained in 0.1% serum and proteins were detected by Western blotting. α, antibody; vec, vector.

MST2 was identified as a kinase that is activated by the pro-apoptotic agents staurosporine and Fas ligand (10, 11). Treatment of COS-1 cells with staurosporine or antibody to Fas reduced the amount of MST2 associated with Raf-1 (Fig. 2A). Mitogenic and pro-apoptotic signaling are often linked in order to prevent unwanted proliferation (12). Therefore, we compared the functional consequences of disrupting the endogenous MST2–Raf-1 complex in response to mitogens or stress signals (Fig. 2B). Staurosporine caused activation of MST2, but not of the Akt kinase, and induced cleavage of poly(ADP-ribose) polymerase (PARP), which indicates activation of caspase and promotion of apoptosis. In contrast, serum growth factors and oncogenic RasV12 induced Akt activation, which is characteristic of survival pathways that promote cell survival. However, they did not stimulate MST2 activity or PARP cleavage. Serum deprivation did not activate MST2, Akt, or caspases, suggesting that serum withdrawal may allow apoptosis by decreasing survival signaling rather than by actively promoting death signaling.

Fig. 2.

Regulation of the MST2–Raf-1 complex. (A) Association of Raf-1 with endogenous MST2 immunoprecipitated from COS-1 cells exposed to the indicated treatments. STR, 100 nM staurosporine; α-Fas, 50 nM antibody to Fas (CD95/apo1 Roche) for 2 hours. Endogenous Raf-1 was detected by Western blotting. (B) Immunoprecipitation of MST2 from cells transfected with RasV12 or treated as indicated. MST2 kinase activity was determined by an in-gel kinase assay. Amounts of MST2 in the immunoprecipitates were determined by Western blotting. PARP cleavage and activation of Akt were assayed in the same lysates by Western blotting as indicated. Full-length PARP is marked by an arrow; cleaved PARP is marked by an asterisk. (C) Immunoprecipitation of Flag-MST2 from Raf-1–/– cells expressing myc-MST2 and Flag-MST2 in the presence of 0.1% serum. MST2 dimerization was detected by Western blotting. Flag-MST2 immunoprecipitates were associated with Myc-MST2 protein. Myc-Raf-1 was transfected at concentrations of 0, 0.1, 0.5, and 1.0 μg of plasmid DNA. (D) Immunoprecipitation of MST2 from serum-starved parental Raf-1+/+ (+/+) and Raf-1–/– (–/–) fibroblasts transfected with Flag-Raf-1, Flag-Raf-1-K375M, or vector control. Endogenous MST2 was immunoprecipitated, and phosphorylated MST2 was detected by Western blotting with an antibody to phosphothreonine (25, 26). Subsequently, blots were washed and MST2 was visualized with antibody to MST2. (E) Purified recombinant MST2 protein (Upstate) was in vitro phosphorylated and activated, then incubated with Flag immunoprecipitates from COS-1 cells expressing Flag-Raf-1, Flag-K375M, or vector control. Phosphatase inhibitor (1mM okadaic acid) was added as indicated. Western blots were developed with the phosphospecific antibody to MST2, stripped, and incubated with a pan-MST2 antibody.

We found no evidence that MST2 regulates Raf-1 (13). However, Raf-1 interfered with MST2 activation on several levels. MST2 is activated through homodimerization followed by transphosphorylation of a critical threonine in the activation loop (14). MST2 formed homodimers in Raf-1–/– cells. However, reconstituting Raf-1–/– cells with increasing amounts of exogenous Raf-1 caused disassembly of the MST2 dimers in a dose-dependent manner (Fig. 2C). MST2 was constitutively phosphorylated at the activating threonine residue in Raf-1–/– cells but not in Raf-1+/+ cells. Reconstitution of the Raf-1–/– cells with Raf-1 or catalytically inactive Raf-1 completely abrogated both MST2 phosphorylation and kinase activity of MST2 (Fig. 2D), which indicated that the effects of Raf-1 did not depend on its kinase activity. Because Raf-1 associates with phosphatases, including PP2A (15, 16), and because the PP2A inhibitor okadaic acid can cause activation of MST2 (14), we examined whether Raf-1 might recruit a phosphatase that dephosphorylates MST2. Recombinant MST2 was activated by autophosphorylation (14) and incubated with Raf-1 or catalytically inactive Raf-1 immunoprecipitated from COS-1 cells. Both Raf-1 immunoprecipitates readily caused dephosphorylation of MST2, which was prevented by okadaic acid (Fig. 2E). These results indicate that Raf-1 restricts MST2 activity by preventing dimerization and recruiting a phosphatase to dephosphorylate the activation site of MST2.

MST2 was constitutively activated in Raf-1–/– cells, and this activity was enhanced by Fas ligation. In contrast, Fas ligation was inefficient in activating MST2 in Raf-1+/+ cells (Fig. 3A). Staurosporine activated MST2 in Raf-1–/– cells better than in Raf-1+/+ cells (Fig. 3B). Furthermore, Raf-1–/– cells succumb to apoptosis more rapidly than Raf-1+/+ cells in response to Fas or staurosporine (Fig. 4A), but not tumor necrosis factor–α (TNF-α) (8). TNF-α failed to stimulate MST2 activity in Raf-1–/– cells (Fig. 3C). Thus, the pattern of MST2 activation reflects the pattern of apoptosis sensitivity of Raf-1–/– cells (7), suggesting that MST2 could be a critical apoptosis effector controlled by Raf-1.

Fig. 3.

Inhibition of threonine phosphorylation and activation of MST2 by Raf-1. (A) Raf-1+/+ and Raf-1–/– cells were treated with mouse-specific antibody to Fas (JO2, 50 nM) plus cycloheximide (CHX, 5 μg/ml) for 1 hour, and MST2 activity was measured in an in-gel kinase assay. (B) Serum-starved Raf-1+/+ and Raf-1–/– fibroblasts were treated with 100 nM STR for the indicated time points. MST2 was immuno-precipitated and its activity was determined in an in-gel kinase assay (11). (C) Serum-starved Raf-1–/– cells were stimulated with murine TNF-α (10ng/ml) or antibody to FasJO2(50nM) for the indicated times. MST2 immunoprecipitates were examined in an in-gel kinase assay, and portions were immunoblotted with antibody to MST2.

Fig. 4.

Depletion of MST2 prevents enhanced apoptosis in cells lacking Raf-1. (A) Apoptosis in Raf-1+/+ (solid bars) and Raf-1–/– (open bars) fibroblasts grown in 10% serum (control) were treated with 100 nM STR or placed in 0.1% serum. Apoptosis was measured 8 and 16 hours after treatment. (B) Absence of MST2 in Raf-1+/+ and Raf-1–/– fibroblasts transfected with one of two siRNA nucleotides directed against MST2 (MST2A or MST2B). Control siRNA was to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Western blots show MST2 protein in total cell extracts. (C) Raf-1+/+ and Raf-1–/– fibroblasts were transfected with the indicated siRNAs, and treated with 50 nM antibody to Fas JO2 and CHX (5 μg/ml) or placed in 0.1% serum. After 16 hours, apoptosis was quantified by measuring DNA fragmentation by fluorescence-activated cell sorting. (D) Raf-1+/+ (solid bars) and Raf-1–/– (open bars) fibroblasts were transfected with Flag-MST2 or kinase-dead Flag-MST2-KD, and assayed for apoptosis as in (C). Overexpressed MST2 was detected by immunoblotting with antibody to Flag. (E) Raf-1+/+ fibroblasts were transfected with siRNA directed against Raf-1 (open bars) or GADPH (hatched bars) or left untreated (solid bars). Cells were placed in 0.1% serum and assayed 16 hours later. Raf-1, ERK, or MST2 was detected by Western blotting of cell lysates. MST2 activity of MST2 immunoprecipitates was detected in an ingel kinase assay. Apoptosis was determined by measuring DNA fragmentation. (F) Lovo and MCF7 cells were transfected with siRNA oligos against Raf-1(50 nM; Ambion), MST2 (50 nM; Ambion), or GAPDH as control, and apoptosis in response to antibody to Fas was measured as above.

Therefore, decreasing the levels of MST2 protein should reduce the sensitivity of Raf-1–/– cells to Fas ligation and serum withdrawal. Decreasing MST2 expression in Raf-1–/– cells using two different small interferening RNAs (siRNAs) (Fig. 4B) completely protected Raf-1–/– cells against apoptosis induced by serum withdrawal or Fas ligation (Fig. 4C). Conversely, the overexpression of MST2 increased the percentage of Raf-1–/– cells that underwent apoptosis when deprived of serum. A catalytically inactive MST2 mutant had no effect on apoptosis (Fig. 4D). Increasing MST2 expression enhanced apoptosis in the Raf-1–/– cells in a dose-dependent manner, whereas Raf-1+/+ cells were unaffected except at the highest dose tested (fig. S3). We propose that Raf-1 may control MST2 by sequestering it into an inactive complex, which can be abolished by overexpression of MST2 or by disruption of the complex through stress signals. Decreasing Raf-1 expression in Raf-1+/+ parental cells by siRNA also led to MST2 activation and apoptosis (Fig. 4E). We also explored MST2 regulation in Lovo (a colon cancer cell line) and MCF7 cells (a mammary cell line). Both MST2 activity and apoptosis increased upon down-regulation of Raf-1 expression by siRNA (fig S4). Fas stimulated both MST2 activity and apoptosis, and both activities were further increased upon Raf-1 down-regulation. The concomitant down-regulation of MST2 and Raf-1 completely prevented Fas-induced apoptosis in Lovo and MCF-7 cells (Fig. 4F).

These results provide evidence for Raf-1 as a physiological regulator of MST2 in antiapoptotic signaling. This role is independent of MEK activation because kinase-negative Raf-1 also could inhibit MST2 activation (Fig. 2D) and apoptosis (fig. S5). Immunodepletion experiments showed that MST2 was quantitatively associated with Raf-1 (fig. S6), which suggests that Raf-1 sequesters MST2 in an inactive state. Raf-1 can counteract apoptosis by different mechanisms, which may be cell type–dependent (1). For instance, heart dilatation and cardiomyocyte apoptosis caused by cardiac-specific Raf-1 ablation is prevented by ablation of the proapoptotic kinase ASK1 (17). Raf-1 also can regulate Fas membrane expression, and the heterozygous inactivation of Fas rescues both the embryonic lethality of Raf-1–/– mice and the hypersensitivity of fibroblasts to Fas-induced apoptosis (18). In the fruit fly Drosophila melanogaster, mutations in the MST2 homolog Hippo cause excessive proliferation and survival of cells in imaginal discs (1923). Hippo forms a complex with Warts and Salvador, whose respective mammalian orthologs, Lats and hWW45, are candidate tumor suppressor genes (24). The existence of similar complexes in mammalian cells has yet to be verified, but would link Raf-1 through MST2 with an important tumor suppressor pathway.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


References and Notes

Stay Connected to Science

Navigate This Article