Ezrin Is a Negative Regulator of Death Receptor-Induced Apoptosis

Oncogene (2010) 29, 1374–1383 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc ORIGINAL ARTICLE Ezrin is a negative regulator of death receptor-induced apoptosis

W-C Kuo1,2, K-T Yang2, S-L Hsieh1,3 and M-Z Lai1,2,4

1Institute of Microbiology and Immunology, National Yang-Ming University, Taipei, Taiwan; 2Institute of Molecular Biology, Academia Sinica, Taipei, Taiwan; 3Genomic Research Center, Academia Sinica, Taipei, Taiwan and 4Institute of Immunology, National Taiwan University, Taipei, Taiwan

Ezrin links cortical actin filaments with the cell mem- phosphatidylinositol 4,5-biphosphate levels lead to brane, and has a critical role in many membrane-initiated inactivation of ERM and their release from the cell events. Fas is directly associated with ezrin, but conflicting membrane (Hao et al., 2009). Ezrin and moesin are the results have been reported for the involvement of ezrin in major ERM proteins in T lymphocytes (Shcherbina Fas-induced cell death. In this study we show that ezrin et al., 1999), and participate in cell functions in either an was associated with Fas in T cells before stimulation and activated or an inactivated form. Activated ezrin and was released shortly after Fas ligand (FasL) engagement. moesin are required for the formation of the distal pole The knockdown of ezrin moderately increased Fas- complex (Allenspach et al., 2001) and for the movement triggered or tumor necrosis factor-related apoptosis- of a cytotoxic cell to target cells (Mrass et al., 2008). inducing ligand (TRAIL)-triggered cell death in normal Alternatively, lymphocyte activation and remolding T lymphocytes and in H9 cells, but had no effect on death involves ezrin/moesin inactivation and dissociation from receptor-induced apoptosis in type II cells, such as Jurkat cortical actin (Delon et al., 2001; Brown et al., 2003; and CEM. Expression of a dominant-negative form of Faure et al., 2004; Gupta et al., 2006). ezrin also led to an increased Fas-induced apoptosis in H9 Fas (CD95)-induced cell death starts from formation cells. Ezrin deficiency did not affect the internalization of of sodium dodecyl sulfate-resistant Fas microaggre- Fas after Fas ligation. Instead, an enhanced formation of gates, followed by the formation of death-inducing death-inducing signaling complex (DISC) was observed in signaling complexes (DISCs) containing Fas-associated H9 cells with ezrin knockdown, leading to accelerated protein with death domain (FADD), procaspase-8 and caspase-8 activation. Together, our results suggest that cellular-FLICE inhibitory protein (c-FLIP) (Nagata, ezrin has a negative role in the recruitment of Fas into 1997; Algeciras-Schimnich et al., 2002; Peter and signaling complexes in type I T cells. Loss of ezrin likely Krammer, 2003; Peter et al., 2007; Strasser et al., removes the constraint imposed by ezrin and facilitates the 2009). The processing of pro-caspase-8 to caspase-8 in assembly of death receptor complex in T cells. DISC leads to the activation of effector caspases and Oncogene (2010) 29, 1374–1383; doi:10.1038/onc.2009.417; eventual cell death. Depending on the amount of DISC published online 23 November 2009 formation, Fas-sensitive cells are divided into type I and type II (Scaffidi et al., 1998). In type I cells, Fas forms Keywords: Fas; ezrin; apoptosis; moesin; DISC sodium dodecyl sulfate-resistant high-order aggregates, followed by rapid internalization of Fas and DISC formation (Algeciras-Schimnich et al., 2002; Lee et al., 2006; Chakrabandhu et al., 2007; Feig et al., 2007). In Introduction type II cells, a very limited amount of DISC is formed and apoptotic signaling needs amplification through Ezrin, radixin and moesin (ERM) are proteins control- mitochondria (Scaffidi et al., 1998). ling the linkage between cortical actin filaments and Fas is directly associated with ezrin (Parlato et al., membrane proteins (Bretscher et al., 2002; Gautreau 2000; Lozupone et al., 2004; He´bert et al., 2008), but the et al., 2002; Niggli and Rossy, 2008). ERM are activated requirement for ezrin and actin filament in Fas-triggered by phosphorylation at the C-terminal threonine or by cell death remains controversial. A study by Kondo et al. the binding of phosphatidylinositol 4,5-biphosphate (1997) showed that dephosphorylation of ERM proteins (PIP2) (Bretscher et al., 2002; Niggli and Rossy, 2008; and translocation of ERM into cytoplasm is essential for Belkina et al., 2009). Activated ERM proteins bind Fas-induced apoptosis to proceed. Dephosphorylation directly to the cortical actin cytoskeleton. Dephospho- of ERM in activated T cells also confers a susceptibility rylation of the C-terminal threonine and reduction in to Fas-mediated apoptosis (Ramaswamy et al., 2007). Consistent with these observations, disruption of the Correspondence: Professor M-Z Lai, Institute of Molecular Biology, actin cytoskeleton enhances the clustering of Fas (Kulms Academia Sinica, 128 Academia Road, Section 2, Taipei 11529, et al., 2002). These results support the notion that the Taiwan. inactivation of ezrin and its release into the cytosol are E-mail: [email protected] Received 28 April 2009; revised 10 October 2009; accepted 24 October associated with Fas-induced apoptosis. Alternatively, the 2009; published online 23 November 2009 interaction of Fas with ezrin and actin cytoskeleton is Deficiency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1375 also suggested to be required for the initiation of Fas- Results mediated apoptosis. Fas is associated with ezrin in T cells that are sensitive to Fas-induced apoptosis, but not in T Dissociation of ezrin from death receptor after Fas ligand cells that are resistant to Fas-triggered death (Parlato (FasL) engagement et al., 2000). The downregulation of ezrin in CEM cells To analyse the function of ezrin in Fas-mediated abolishes their susceptibility to Fas-induced apoptosis apoptosis, we re-examined the interaction between ezrin (Parlato et al., 2000), and the knockdown of ezrin or and Fas. Similar to previous reports (Parlato et al., moesin in Jurkat cells decreases Fas-triggered apoptosis 2000; Lozupone et al., 2004), recombinant ezrin bound (He´bert et al., 2008). It has also been shown that Fas– directly to Fas in a glutathione S-transferase (GST) ezrin–actin linkage is involved in Fas endocytosis, and pull-down assay (Figure 1a). The association of ezrin Fas-induced apoptosis was attenuated in ezrin-knock- with Fas during the course of Fas-mediated apoptosis down L12.10 cells (Chakrabandhu et al., 2007, 2008). was also monitored in H9 cells, a type I cell line known Therefore, the exact role of ezrin in death receptor- for prominent DISC formation (Scaffidi et al., 1998; initiated apoptosis remains unsettled. Algeciras-Schimnich et al., 2002; Lee et al., 2006). Ezrin In this study, we analysed the role of ezrin in Fas- was associated with Fas in H9 cells before treatment induced apoptosis in T lymphocytes. We found that the (Figure 1b). Stimulation with FasL led to the recruit- knockdown of ezrin by small interfering RNA (siRNA) ment of FADD and pro-caspase-8 to Fas, as well as the did not interfere with Fas-triggered cell death in type II processing of pro-caspase-8. Interestingly, FasL ligation cells, but moderately enhanced Fas-induced apoptosis in also led to the dissociation of ezrin from Fas-activating normal T cells and type I cells. Ezrin was not required complex; despite a very small amount of ezrin remaining for the endocytosis of Fas in type I T cells. The increase detectable, the majority of Fas-associated ezrin was in Fas-mediated cell death in ezrin-deficient cells could released within 20 min (Figure 1b). We also monitored be partly attributed to increased DISC formation. Our the phosphorylation of ezrin after FasL ligation. results suggest that the removal of ezrin leads to Consistent with the reports of Kondo et al. (1997), enhanced Fas-induced DSIC formation and apoptosis ezrin phosphorylation was diminished after FasL in type I T lymphocytes. treatment (Figure 1c), suggesting that ezrin inactivation

Figure 1 Fas ligand (FasL) ligation induced ezrin dissociation and ezrin dephosphorylation. (a) Direct interaction between Fas-C and ezrin. A 3 mg sample of glutathione S-transferase (GST) or GST-FasC (a.a. 185–328 of Fas) was loaded on glutathione (GSH)-agarose, and incubatedwith3mg of His-ezrin. The His-ezrin proteins pulled down were detected by anti-His. Anti-GST and anti-His were used to assess input of GST/GST-FasC and His-ezrin, respectively. (b) Dissociation of ezrin from Fas after FasL stimulation. H9 cells (4 Â 107) were treated with FLAG-FasL (50 ng/ml) for the indicated time, lysed, and cell lysates were precipitated with FLAG-M2 beads. The associated ezrin, pro-caspase-8, Fas-associated protein with death domain (FADD) levels were determined using immunoblotting. Input indicates the level of the given protein in lysates. (c, d) Early dephosphorylation and late re-phosphorylation of ezrin after FasL treatment. H9 cells were treated with FLAG-FasL and cell lysates prepared at the indicated time points. The levels of ezrin, phospho-ezrin (Thr567) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were determined by western blots using the appropriate antibodies.

Oncogene Deficiency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1376 is accompanied with Fas DISC assembly. It should be Downregulation of ezrin moderately increases noted that prolonged FasL stimulation (>25 min) led to Fas-triggered apoptosis in normal T lymphocytes increased phosphorylation of ezrin–moesin (Figures 1c The finding that ezrin has a negative role in Fas-induced and d), in accordance with earlier reports (He´bert et al., cell death in H9 cells led us to further examine the effect of 2008). ezrin knockdown in Fas-mediated apoptosis in normal T cells. Peripheral blood T cells were activated by anti-CD3/ CD28 for 2 days, maintained in interleukin-2 for another Enhanced Fas- and tumor necrosis factor-related 3 days and then transfected with ezrin-specific siRNA. apoptosis-inducing ligand (TRAIL)-induced cell Figure 3a illustrates that ezrin levels were reduced by death in ezrin-deficient H9 cells more than 90% in normal T cells by specific hairpin Because both ezrin phosphorylation and ezrin associa- siRNA, but not by a control Cy3-labelled siRNA. In tion were reduced early on in the Fas-triggered normal T cells with ezrin-knockdown, apoptosis triggered apoptotic process, we analysed whether ezrin was by FasL or CH11 was also moderately increased relative required for Fas-mediated cell death in H9 cells. Ezrin to the control siRNA-transfected T cells (Figures 3b was effectively downregulated in H9 cells using a and c). The enhancement in Fas-induced cell death caused lentiviral vector carrying ezrin-specific short hairpin by ezrin downregulation in normal T cells was not as RNA (shRNA; Figure 2a). The knockdown of ezrin profound as in H9 cells, which may reflect some difference in H9 cells led to a nearly twofold increase in Fas- between the normal and transformed T cells. mediated apoptosis triggered by CH11, relative to the control H9 cells (Figure 2b). A similar extent of Enhanced apoptosis in H9 cells expressing ezrin increased cell death in ezrin-knockdown H9 cells was dominant-negative mutant found with FasL or Apo-1 treatment (data not shown). To ensure that the effect of ezrin knockdown was not due In addition, TRAIL-triggered cell death was also to the downregulation of some unknown off-target mech- substantially enhanced in ezrin-deficient H9 cells anism, we overexpressed a dominant-negative form of ezrin, (Figure 2c). The profound effect of ezrin knockdown ezrin (1–320), in H9 cells. This dominant-negative ezrin on death receptor-induced apoptosis in H9 cells was promotes the dissociation of ezrin from actin filaments, and specific for ezrin. The downregulation of moesin has been used to evaluate the requirement of ezrin in specific (Figure 2a) had no effect on CH11-triggered cell death biological events. Ezrin(1–320)-Myc was transduced into in H9 cells (Figure 2d). Therefore, ezrin is not required H9 cells by retroviral infection, and its expression confirmed for, but instead is a negative regulator of, death using anti-Myc antibodies (Figure 4a). Expression of the receptor-induced apoptosis in H9 cells. dominant-negative form of ezrin in H9 cells also increased

Figure 2 Deficiency of ezrin, but not moesin, increases Fas and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)- induced cell death in H9 cells. (a) Knockdown of ezrin and moesin in H9 cells. H9 cells were infected with lentivirus containing small interfering RNA (siRNA) specific for ezrin or moesin, infected cells were isolated by sorting based on green fluorescent protein (GFP) expression and the ezrin or moesin levels were determined using immunoblots. Anti-moesin antibodies (Clone 38, BD Biosciences) detected both ezrin and moesin. (b, c) Ezrin-deficient H9 cells were more sensitive to Fas- and TRAIL-mediated apoptosis. GFP control and ezrin-knockdown H9 cells were incubated with anti-Fas CH11 (a) or recombinant human TRAIL (c) for 4 h. Cell death was assessed by phycoerythrin (PE)-Annexin V staining. (d) Moesin knockdown did not affect Fas-induced cell death in H9 cells. Control, moesin-knockdown and ezrin-knockdown H9 cells were treated with CH11, and cell death was determined 4 h later. Data are the average of triplicates, independently repeated thrice. *Po0.05; **Po0.01; ***Po0.001 for paired t-test.

Oncogene Deﬁciency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1377

Figure 4 Enhanced Fas-triggered apoptosis in H9 cells expressing an ezrin dominant-negative mutant. (a) Expression of a dominant- Figure 3 Enhanced Fas-induced apoptosis by ezrin knockdown in negative ezrin in H9 cells. H9 cells were transduced with pGC-YFP normal human T lymphocytes. (a) Knockdown of ezrin in human T or pGC-YFP-ezrin (1–320)-Myc, and YFP þ cells isolated using cells by small interfering RNA (siRNA). Purified human T cells from sorting. The expression of ezrin (1–320)-Myc was confirmed by peripheral blood were activated by anti-CD3/anti-CD28 for 2 days, anti-Myc. (b) Increased Fas-induced cell death in ezrin (1–320)- and cultured in the presence of interleukin-2 (IL-2) for an additional expressing (DN) H9 cells. YFP control and ezrin (1–320)- 3 days. Activated T cells were transfected with control or ezrin- expressing H9 cells were treated with indicated doses of FasL, specific siRNA. Cell lysates were prepared 24 h later, and the level of and cell death was quantitated 4 h later using Annexin V staining. ezrin and moesin was determined using immunoblots. (b, c)Ezrin (c) Accelerated processing of pro-caspase-8 and pro-caspase-3 in knockdown increased CH11- and FasL-induced apoptosis in normal ezrin (1–320)-expressing H9 cells. Control and DN H9 cells (3 Â T cells. Ezrin-deficient and control normal human T cells were 106/ml) were treated with Fas ligand (FasL; 150 ng/ml), cell lysates treated with CH11 (b)orFasL(c) at the indicated dose, and were prepared at the indicated time points and the levels of pro- apoptosis was quantitated using hypodiploid fraction (b)or caspase-8, active caspase-8, pro-caspase-3 and glyceraldehyde-3- phycoerythrin (PE)-Annexin V staining at 4 h (c)or18h(b). Data phosphate dehydrogenase (GAPDH) were determined using are the average of triplicates ± s.d. in a single experiment, which has western blots. **Po0.01; ***Po0.001. been independently repeated four times. *Po0.05; ***Po0.001. their sensitivity to FasL-induced apoptosis (Figure 4b), observation in H9 and normal T cells (Figures 2–4). To accompanied with accelerated activation of caspase-8 and analyse whether ezrin may have opposite roles in type I caspase-3 (Figure 4c). Therefore, similar to ezrin deficiency, (H9) and type II (Jurkat and CEM) T cells, we also reduced anchoring of ezrin to actin by the dominant- studied the effect of ezrin knockdown in death receptor- negative ezrin led to enhanced Fas-induced apoptosis. As a mediated apoptosis in both Jurkat and CEM cells. Ezrin control to ezrin dominant-negative mutant, overexpression was knocked down in Jurkat by transfection with of wild-type (full-length) ezrin in H9 cells modestly inhibited hairpin siRNA specific for ezrin. Figure 5a illustrates FasL-induced apoptosis (Supplementary Figures 1a and b), more than 90% reduction in ezrin levels in Jurkat cells further supporting a negative role of ezrin in death receptor- up to 72 h after transfection with ezrin siRNA. FasL- initiated death in H9 cells. induced apoptosis was determined in Jurkat cells at 48 h after siRNA transfection, and the extent of cell death in ezrin-knockdown and control Jurkat cells was nearly Ezrin knockdown by small interfering RNA (siRNA) does identical (Figure 5b). CH11-induced or Apo1–3-induced not affect Fas- and tumor necrosis factor-related apoptosis- apoptosis was also similar in both ezrin-deficient and inducing ligand (TRAIL)-induced cell death in type II cells control Jurkat cells (Figures 5c and d). In addition, Ezrin downregulation has been reported to impair TRAIL-triggered cell death in Jurkat cells was not Fas-induced apoptosis in Jurkat and CEM cells (Parlato altered by ezrin deficiency (not shown). Ezrin was also et al., 2000; He´bert et al., 2008), in contrast to our knocked down in CEM cells by infection with pLL3.7-

Oncogene Deﬁciency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1378

Figure 5 Ezrin knockdown does not affect the sensitivity of Jurkat cells to Fas-induced cell death. (a) Knockdown of ezrin in Jurkat cells. Jurkat cells were transfected with control small interfering RNA (siRNA; pLL) or ezrin-specific siRNA (siEz) using electroporation, and cellular contents of ezrin at different time points were determined using immunoblots. (b) Fas-mediated apoptosis was not affected in ezrin-deficient Jurkat cells. Control and ezrin knockdown Jurkat cells, 48 h after transfection, were incubated with Fas ligand (FasL) for 18 h. Cell deaths were quantitated using propidium iodide staining, followed by flow cytometry for hypodiploid fraction quantitation. (c, d) Ezrin knockdown affected apoptosis induced by CH11 and Apo-1–1 in Jurkat cells. Control and ezrin knockdown Jurkat cells were treated with anti-Fas antibodies CH11 (c) or Apo-1–1 (d) at the doses indicated, and cell death was assessed 4 h later using phycoerythrin (PE)-Annexin V.

shEzrin (Supplementary Figure 2a). In this study, the T cells. The expression of Fas, pro-caspase-8 and downregulation of ezrin did not affect cell death in pro-caspase-3 was similar in control and ezrin-deficient CEM cells triggered by FasL or CH11 (Supplementary H9 cells (Figure 7a, 0 h). The expression of cellular Figure 2b, data not shown for CH11). Contrary to the FLICE inhibitory protein was also unaffected by reports that ezrin reduction impaired Fas-induced ezrin deficiency (Supplementary Figure 3). We then apoptosis (Parlato et al., 2000; He´bert et al., 2008), monitored the biochemical events after Fas ligation. In Fas-mediated apoptosis was not affected in Jurkat and ezrin-deficient H9 cells, the processing of pro-caspase-8 CEM cells that were deficient in ezrin. In a separated and the appearance of p43 caspase-8 intermediate were experiment, overexpression of wild-type ezrin in Jurkat detected earlier than in control H9 cells (Figure 7a). cells did not affect FasL-triggered apoptosis (Supple- The accelerated caspase-8 activation in ezrin-knock- mentary Figures 1c and d). Therefore, death receptor- down H9 cells was also confirmed by an increase in induced cell death in type II cells is ezrin independent. p18 active caspase-8. This was also accompanied by enhanced proteolysis of pro-caspase-3 and the appearance of active caspase-3 in ezrin-downregulated Normal Fas endocytosis in ezrin-deficient H9 cells after H9 cells (Figure 7a). A similar observation was made Fas engagement upon TRAIL treatment of ezrin-deficient H9 cells. Another reported function of ezrin is its involvement in TRAIL-triggered cleavage of pro-caspase-8, emergence the endocytosis of Fas that is essential for DISC of active caspase-8 and appearance of active caspase-3 formation and the activation of caspase-8 (Lee et al., were clearly accelerated in ezrin-knockdown H9 2006; Chakrabandhu et al., 2007). We analysed whether cells, relative to control H9 cells (Supplementary the internalization of Fas was affected by ezrin deficiency Figure 4). in H9 cells. The surface Fas levels in H9 cells after Fas Because caspase-8 activation is the immediate readout ligation were monitored. Figure 6a illustrates that anti- of DISC formation, we also measured the extent of Fas treatment led to a time-dependent disappearance of DISC formation after Fas ligation. FLAG-tagged FasL Fas from the cell surface in the control H9 cells. The was used to pull down DISC. In control H9 cells, Fas knockdown of ezrin did not affect either the extent or the engagement resulted in formation of the complex kinetics of Fas endocytosis (Figure 6b), suggesting that containing Fas, pro-caspase-8 and FADD. The assem- Fas internalization is not dependent on ezrin in H9 cells. bly of DISC was greatly enhanced in H9 cells expressing ezrin (1–320) (Figure 7b) or in H9 cells that were Accelerated caspase-8 cleavage and increased death- deficient in ezrin (Figure 7c). Therefore, either ezrin inducing signaling complex (DISC) formation in dissociation from cortical actin or ezrin knockdown ezrin-deficient H9 cells led to increased overall formation of DISC in H9 We examined the expression of components that were cells, resulting in accelerated activation of caspase-8 immediately downstream of Fas in ezrin-knockdown and caspase-3.

Oncogene Deﬁciency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1379

Figure 6 Ezrin knockdown does not affect endocytosis of Fas. H9 cells were incubated with anti-Fas Apo-1–3 (0.25 mg/ml) for 30 min at 4 1C, cells were warmed for the indicated times and internalization was then stopped by adding ice-cold medium with 0.5% sodium azide. Washed cells were maintained on ice, stained with anti-Fas (DX2), fixed with glutaraldehyde and the levels of surface Fas assessed using fluorescence-activated cell sorting (FACS) analysis (a). The kinetics of the reduction of cell-surface Fas are shown by plotting mean fluorescence intensity (MFI) of Fas versus time after Apo-1–3 ligation (b).

Discussion group of studies (Roumier et al., 2001; Ilani et al., 2007); ezrin was reported to be completely dispensable for IS In this study, we monitored the effect of ezrin deficiency formation in another study (Shaffer et al., 2009); and on Fas-mediated cell death. Increased Fas-induced in yet a third proposal, ezrin was also declared to apoptosis caused by ezrin knockdown was found in negatively regulate IS formation (Faure et al., 2004). normal lymphocytes and H9 cells, but not in Jurkat and Similarly, opposite effects of ERM have been documen- CEM cells. In contrast to reports that ezrin is essential ted for T-lymphocyte movement; T-cell migration could for Fas-initiated apoptosis (Parlato et al., 2000; Piaz- be enhanced by either ERM phosphorylation (Li et al., zolla et al., 2005; Chakrabandhu et al., 2007; He´bert 2007) or ERM dephosphorylation (Brown et al., 2003). et al., 2008), our results support an alternative view that The negative effect of ezrin in Fas-induced apoptosis ezrin inactivation and dissociation from actin enhance presented in this study, in contrast to the previously Fas-induced cell death (Kondo et al., 1997; Ramaswamy reported positive role for ezrin, highlights another et al., 2007). The opposite results found for the function controversy that is centered on the membrane events of ezrin were unlikely to be due to the use of different involving ERM. reagents; the ezrin-specific siRNA used in this study was Characterization of the exact involvement of ERM in identical to that used in a recent study illustrating the different cellular events is complicated by the fact that positive effect of ezrin in Fas-triggered apoptosis in ezrin is interchanged between inactivated and activated Jurkat cells (He´bert et al., 2008). In addition, the effect forms. Ezrin is present in a phosphorylated form in cells of ezrin knockdown in Fas-induced cell death could be before stimulation through Fas, antigen receptors or reproduced by the expression of a dominant-negative chemokine receptors, and is dephosphorylated after form of ezrin (Figure 4), excluding the possibility that these stimulations (Kondo et al., 1997; Delon et al., results observed with ezrin-specific siRNA were due to 2001; Brown et al., 2003; Faure et al., 2004; Gupta et al., some off-target effect. 2006; Hao et al., 2009). Our observation that ezrin was Ezrin, radixin and moesin (ERM) proteins link the dephosphorylated after FasL stimulation in T cells cortical actin cytoskeleton with membrane proteins. (Figure 1c) is in accordance with those reports. How ERM participates in several membrane-associated Regarding the observation that Fas engagement led to events in lymphocytes remains unclear. For example, an increased ezrin phosphorylation (He´bert et al., 2008), three different roles have been reported for the involve- we also found that in the later time points of FasL ment of ezrin–moesin in immunological synapse (IS) stimulation (>25 min), enhanced ezrin phosphorylation formation: ERM was found to be essential for the was observed (Figure 1d). Therefore, FasL treatment assembly of IS and the recruitment of ZAP70 in one leads to an early ezrin inactivation followed by ezrin

Oncogene Deficiency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1380 Fas treatment is likely to be irrelevant to the DISC formation and the subsequent apoptosis. Despite the sequence and structural homology between ezrin and moesin, there are distinct functional differences between these two proteins. One recent example of such differences is the formation of IS in T cells, which involves the opposing actions of ezrin and moesin, with the phosphorylation of ezrin recruiting ZAP70 and the dephosphorylation of moesin removing CD43 (Ilani et al., 2007). Our observations that knockdown of ezrin, but not of moesin, enhanced Fas-induced apoptosis in H9 cells (Figure 4) may represent another functional difference between ezrin and moesin. Whether ezrin and moesin confer distinct restraints on the formation of DISC, or whether Fas is preferentially associated with ezrin, is currently being analysed. Internalization of Fas is required for efficient DISC formation and caspase-8 activation (Lee et al., 2006; Chakrabandhu et al., 2007). Ezrin-actin association is reported to be required for endocytosis of Fas in mouse embryonic fibroblasts and L12.10 cells (Piazzolla et al., 2005; Chakrabandhu et al., 2007, 2008). We found that Fas internalization in H9 cells (Figure 6) was similar in efficiency to that documented elsewhere (Piazzolla et al., 2005; Lee et al., 2006; Chakrabandhu et al., 2007). However, deficiency in ezrin did not affect endocytosis of Fas in H9 cells, suggesting that the observed enhanced DISC formation (Figure 7) was not due to altered internalization of Fas. Discrepancies in the involvement of ezrin in Fas endocytosis between this study and others (Piazzolla et al., 2005; Chakrabandhu et al., 2007, 2008) may be due to differences between T cells and L cell/fibroblasts, because the function of ERM is highly cell type dependent (Bretscher et al., 2002; Niggli and Rossy, 2008). However, just why ezrin is required for Fas endocytosis in L cells/fibroblasts but not in T cells remains to be elucidated. Notably, ezrin was associated with Fas in H9 cells, and was disas- sociated from Fas after FasL treatment (Figure 1b). In contrast, Fas–ezrin interaction was increased after Fas Figure 7 Accelerated caspase-8 cleavage and death-inducing signal- engagement in L12.10 cells (Chakrabandhu et al., 2007). ing complex (DISC) formation in ezrin-deficient T cells. (a)Increased Ezrin-negative regulation was found in normal processing of pro-caspase-8 and pro-caspase-3 in ezrin-deficient lymphocytes and H9 cells, but not in type II cells T cells. Control and ezrin-knockdown H9 cells (4 Â 106/ml) were including Jurkat and CEM cells (Figures 2–5). Type I treated with Fas ligand (FasL; 50 ng/ml), cell lysates were prepared at the indicated time points and the levels of pro-caspase-8, pro-caspase- and type II cells are differentiated mainly by the 3, active caspase-8, active caspase-3, ezrin and glyceraldehyde-3- effectiveness of the DISC assembly (Algeciras-Schim- phosphate dehydrogenase (GAPDH) were determined using western nich et al., 2002; Lee et al., 2006; Feig et al., 2007). blots. (b, c) Increased DISC formation in T cells expressing ezrin (1– Consistent with the nature of type I cells, FasL 320) or with ezrin knockdown. Cell lysates of control and H9 cells treatment led to a high degree of DISC formation in expressing ezrin (1–320) (b), or of control and ezrin-knockdown H9 cells (c) were prepared before (0 min) and 25 min after FLAG-FasL H9 cells, and this was further increased by ezrin stimulation. DISCs in the lysate complexes were pulled down by knockout or by dominant-negative ezrin (Figure 7). FLAG-M2 beads. DISC components were eluted from M2 beads by Because ezrin was not required for internalization of Fas FLAG peptide, and analysed by immunoblots using antibodies in H9 cells, one possibility is that the association of Fas specific for Fas, pro-caspse-8, Fas-associated protein with death domain (FADD) and FLAG (for FasL). with ezrin–actin imposes constraints that prevent extensive recruitment of FADD and pro-caspase-8 to assemble DISC, and elimination of the ezrin binding facilitates the early death molecule clustering. On the reactivation, with sequential dephosphorylation and basis of such a scenario, in type II cells in which only phosphorylation events. However, as ezrin was mostly small amounts of DISCs are formed, the involvement of dissociated from Fas after 20 min (Figure 1b), the the ezrin–actin network in Fas-initiated apoptotic elevated ezrin phosphorylation at later time points after processes is expected to be minimal, as we have found

Oncogene Deficiency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1381 with Jurkat and CEM cells (Figure 5 and Supplemen- Death-inducing signaling complex (DISC) immunoprecipitation tary Figure 2). Whether ezrin-dependent inhibition Aliquots of 4 Â 107 H9 cells were incubated with FLAG-FasL represents another process that differentiates type I cells (50 ng/ml; Alexis) for the indicated time. Cells were lysed in DISC from type II cells also deserves further analysis. immunoprecipitation buffer (20 mM Tris-HCl, pH 7.4, 150 mM In summary, the results of this study clearly suggest sodium chloride, 2 mM EDTA, 1% NP-40, 0.25% sodium deoxycholate and protease inhibitors (Roche, Mannheim, Ger- that ezrin has a negative role in Fas- and TRAIL- many)). Cell extracts were immunoprecipitated with FLAG-M2 induced apoptosis in type I T cells, whereby removal of beads (Sigma) overnight at 4 1C. DISC complexes were eluted ezrin promotes the endocytosis-independent assembly of from M2 beads by FLAG peptide (0.5 mg/ml). The eluent was DISC. Future works are required to clarify the resolved using sodium dodecyl sulfate-polyacrylamide gel electro- differential role of ezrin and moesin in death receptor- phoresis and probed with anti-ezrin, anti-caspase-8 (BD), anti- mediated apoptosis in type I T cells, and to establish FADD, anti-tubulin and anti-FLAG antibodies (Sigma). whether the negative regulation of ezrin in Fas-induced cell death is applied to other type of cancer cell. This Ezrin and moesin knockdown in T leukemia cell lines re-definition of the role of ezrin and cortical actin in Ezrin and moesin were knocked down in T leukemia cell lines Fas-induced apoptosis should help our understanding of by two different methods. Jurkat cells (4 Â 106) were trans- the detailed molecular processes of DISC assembly, for fected with 500 pmol of ezrin-specific siRNA (Qiagen, 0 0 effective regulation of death receptor-triggered cell death. 5 -AACAGAAACAUUCUGGGCU-3 ) and/or the control siRNA (Silencer Cy3-labeled negative control no. 1 siRNA, Ambion, Austin, TX, USA) using electroporation (250 V, 400 mF, Bio-Rad, Hercules, CA, USA). The transfection Materials and methods efficiency, as determined by Cy3 fluorescence, was >85% in all cases. Alternatively, an ezrin-knockdown lentiviral con- Reagents struct was generated by subcloning ezrin-specific short hairpin CH11 and anti-tubulin antibodies were purchased from RNA sequence into pLentiLox vector (pLL3.7, gift of Dr Upstate (Lake Placid, NY, USA). Recombinant human Fas I-Chen Ho, Harvard Medical School, Boston, MA, USA). The ligand (FLAG-tagged), anti-Fas (Apo-1–3) and anti-cellular sequence of the ezrin-specific short hairpin RNA subcloned FLICE inhibitory protein (NF6) were purchased from Alexis was 50-GGACTGATTGAATTACGGA-30, and the sequence (San Diego, CA, USA). Recombinant human TRAIL was of the moesin-specific short hairpin RNA was 50-AGTTC obtained from R&D (Minneapolis, MN, USA). Phycoery- GAGGAACAGACATT-30. Lentiviruses were harvested from thrin-labeled anti-CD95 was purchased from eBioscience culture supernatant of 293FT cells transfected with 20 mg (San Diego, CA, USA). Anti-FADD, anti-poly (ADP-ribose) pLL3.7 or pLL3.7-PMLsiRNA, 15 mg psPAX2 and 6 mg polymerase, anti-ezrin (clone 18), anti-moesin (clone 38) and pMD2.G. CEM cells were infected with recombinant lenti- anti-caspase-8 antibodies were obtained from BD Bioscience virus, and green fluorescent protein-expressing cells were (Franklin Lakes, NJ, USA). Horseradish peroxidase-conju- isolated using fluorescence sorting 48 h later on a FACSVan- gated goat anti-rabbit immunoglobulin and horseradish tage SE (Becton Dickinson, Mountain View, CA, USA). The peroxidase-conjugated rabbit anti-mouse immunoglobulin levels of ezrin and moesin were confirmed using immunoblots. were obtained from Amersham Bioscience (Buckinghamshire, UK). Anti-caspase-3 (H-277) and anti-FADD (H-181) were Knockdown of ezrin in normal human T cells purchased from Santa Cruz Biotechnology (Santa Cruz, CA, Human T cells were isolated from peripheral blood using USA). Anti-phospho-ERMs (Thr567 of ezrin, Thr564 of RosetteSep Human T cell Enrichment cocktail (StemCell radixin and Thr558 of moesin), anti-caspase-8 (1C12), anti- Technologies, Vancouver, BC, USA) according to the sug- active caspase-8 (Asp391, 18C8) and anti-active caspase-3 gested protocol. The purity of CD3 þ T cells was over 97%, as (Asp175) were obtained from Cell Signaling (Beverly, MA, determined using fluorescence-activated cell sorting analysis. USA). Anti-b-tubulin (Clone AA2) was purchased from T cells were activated by plate-bound anti-CD3 (OKT3, 5 mg/ Upstate. ml) plus anti-CD28 (clone 37407, 2.5 mg/ml; R&D) for 2 days, and then incubated in the presence of interleukin-2 (10 U/ml; Glutathione S-transferase (GST) pull-down assay R&D) for an additional 3 days. Transfections of human T cells were performed using the human T cell nucleofector kit His-tagged ezrin was generated by subcloning the full-length 7 ezrin into pQE-16 vector, and the recombinant His-ezrin was (Amaxa, Koeln, Germany). Aliquots of 2 Â 10 human T cells purified on Ni-NTA agarose (both from Qiagen, Hilden, were resuspended in 100 ml of human T cell nucleofector Germany). The C-terminal region (a.a. 185–328) of Fas (Fas- solution, mixed with a total of 500 pmol siRNA, and C) was isolated using PCR and was subcloned into pGEX-4T2 electroporation conducted on a Nucleofector (Amaxa) using (GE Healthcare, Piscataway, NJ, USA) to produce GST program T20. T cells were immediately transferred to pre- fusion protein. Recombinant GST fusion proteins were then warmed medium and cultured in complete medium. Cells were purified on glutathione-agarose. A total of 3 mg of GST–Fas-C used 24 h after transfection for apoptosis induction or immunoblots analysis. The transfection efficiency, as deter- or GST in glutathione-binding buffer (50 mM potassium mined by Cy3 fluorescence, was >90% in all cases. phosphate buffer, pH 7.5, 150 mM potassium chloride, 1 mM magnesium chloride, 10% glycerol, 1% Triton X-100, 1% aprotinin and 1 mM phenylmethylsulphonyl fluoride) were Expression of dominant-negative ezrin loaded on to glutathione-agarose (20 ml, 50% slurry; Sigma, Ezrin(1–320)-Myc was sub-cloned into pGC-IRES-YFP (a gift St Louis, MO, USA), and incubated with 3 mg of His-ezrin for from Dr Gina Costa, Stanford University, Stanford, CA, 3 h at 4 1C. GST agarose beads were then washed five times USA) to generate pGC-ezrin(1–320)-IRES-YFP. Retroviruses with binding buffer, separated by 10% sodium dodecyl sulfate- were produced by transfection of Phoenix cells (gifts of Dr polyacrylamide gel electrophoresis, and detected using anti-His Garry P Nolan, Stanford University) with 10 mgofpGC-YFPor and anti-GST (both from Santa Cruz). pGC-ezrin(1-320)-YFP plasmids. Retrovirus-containing super-

Oncogene Deficiency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1382 natants were collected at 48 h after transfection, and were used Conflict of interest for spin infection of H9 T cells. At 48 h after infection, YFP- expressing H9 cells were then isolated by sorting on FACSVan- The authors declare no conflict of interest. tage SE (Becton Dickinson). Full-length ezrin-myc was similarly expressed in Jurkat by infection with pGC-F-ezrin. Expression of full-length ezrin-myc (in pcDNA4) in H9 cells was performed by electroporation with an MP-100 (MicroPorator, Digital Bio, Acknowledgements Korea) at 1300 V, 30 ms and 1 pulse. This work was supported by Grant NHRI-EX96–9527NI from National Health Research Institute, NSC 95–2320-B001–023 Abbreviations from National Science Council and an Academia Sinica Investigator Award from Academia Sinica, Taiwan, ROC. DISC, death-inducing signaling complex; ERM, ezrin, We thank Drs I-Chen Ho, Gina Costa and Garry Nolan for radixin and moesin; FasL, Fas ligand; IS, immunological critical reagents, Yamin Lin and FACS Core of Institute of synapse; TRAIL, tumor necrosis factor-related apoptosis- Molecular Biology, Academia Sinica for cell sorting and Dr inducing ligand. Harry Wilson for editing this paper.

References

Algeciras-Schimnich A, Shen L, Barnhart BC, Murmann AE, Hao JJ, Liu Y, Kruhlak M, Debell KE, Rellahan BL, Shaw S. Burkhardt JK, Peter ME. (2002). Molecular ordering of (2009). Phospholipase C-mediated hydrolysis of PIP2 releases the initial signaling events of CD95. Mol Cell Biol 22: ERM proteins from lymphocyte membrane. J Cell Biol 184: 207–220. 451–462. Allenspach EJ, Cullinan P, Tong J, Tang Q, Tesciuba AG, Cannon JL He´bert M, Potin S, Sebbagh M, Bertoglio J, Breárd J, Hamelin J. et al. (2001). ERM-dependent movement of CD43 defines a novel (2008). Rho-ROCK-dependent ezrin-radixin-moesin phosphoryla- protein complex distal to the immunological synapse. Immunity 15: tion regulates Fas-mediated apoptosis in Jurkat cells. J Immunol 739–750. 181: 5963–5973. Belkina NV, Liu Y, Hao JJ, Karasuyama H, Shaw S. (2009). LOK is a Kondo T, Takeuchi K, Doi Y, Yonemura S, Nagata S, Tsukita S. major ERM kinase in resting lymphocytes and regulates cytoske- (1997). ERM (ezrin/radixin/moesin)-based molecular mechanism of letal rearrangement through ERM phosphorylation. Proc Natl Acad microvillar breakdown at an early stage of apoptosis. J Cell Biol Sci USA 106: 4707–4712. 139: 749–758. Bretscher A, Edwards K, Feho RG. (2002). ERM proteins and Kulms D, Du¨ssmann H, Po¨ppelmann B, Sta¨nder S, Schwarz A, merlin: integrators at the cell cortex. Nat Rev Mol Cell Biol 3: Schwarz T. (2002). Apoptosis induced by disruption of the actin 586–599. cytoskeleton is mediated via activation of CD95 (Fas/APO-1). Cell Brown MJ, Nijhara R, Hallam JA, Gignac M, Yamada KM, Death Differ 9: 598–608. Erlandsen SL et al. (2003). Chemokine stimulation of human Ilani T, Khanna C, Zhou M, Veenstra TD, Bretscher A. (2007). peripheral blood T lymphocytes induces rapid dephosphorylation of Immune synapse formation requires ZAP-70 recruitment by ERM proteins, which facilitates loss of microvilli and polarization. ezrin and CD43 removal by moesin. J Cell Biol 179: Blood 102: 3890–3899. 733–746. Gautreau A, Louvard D, Arpin M. (2002). ERM proteins Lee KH, Feig C, Tchikov V, Schickel R, Hallas C, Schutze S et al. and NF2 tumor suppressor: the Yin and Yang of cortical actin (2006). The role of receptor internalization in CD95 signaling. organization and cell growth signaling. Curr Opin Cell Biol 14: EMBO J 25: 1009–1023. 104–109. Li Y, Harada T, Juang YT, Kyttaris VC, Wang Y, Zidanic M et al. Chakrabandhu K, Herincs Z, Huault S, Dost B, Peng L, Conchonaud (2007). Phosphorylated ERM is responsible for increased T cell F et al. (2007). Palmitoylation is required for efficient Fas cell death polarization, adhesion, and migration in patients with systemic signaling. EMBO J 26: 209–220. lupus erythematosus. J Immunol 178: 1938–1947. Chakrabandhu K, Huault S, Garmy N, Fantini J, Stebe E, Lozupone F, Lugini L, Matarrese P, Luciani F, Federici C, Iessi E Mailfert S et al. (2008). The extracellular glycosphingolipid-binding et al. (2004). Identification and relevance of the CD95-binding motif of Fas defines its internalization route, mode and outcome domain in the N-terminal region of ezrin. J Biol Chem 279: of signals upon activation by ligand. Cell Death Differ 15: 9199–9207. 1824–1837. Mrass P, Kinjyo I, Ng LG, Reiner SL, PureÉ, Weninger W. Delon J, Kaibuchi K, Germain RN. (2001). Exclusion of CD43 from (2008). CD44 mediates successful interstitial navigation by killer T the immunological synapse is mediated by phosphorylation- cells and enables efficient antitumor immunity. Immunity 29: regulated relocation of the cytoskeletal adaptor moesin. Immunity 971–985. 15: 691–701. Nagata S. (1997). Apoptosis by death factor. Cell 88: 355–365. Faure S, Salazar-Fontana LI, Semichon M, Tybulewicz VL, Bismuth Niggli V, Rossy J. (2008). Ezrin/radixin/moesin: versatile controllers of G, Trautmann A et al. (2004). ERM proteins regulate cytoskeleton signaling molecules and of the cortical cytoskeleton. Int J Biochem relaxation promoting T cell-APC conjugation. Nat Immunol 5: Cell Biol 40: 344–349. 272–279. Parlato S, Giammarioli AM, Logozzi M, Lozupone F, Matarrese P, Feig C, Tchikov V, Schutze S, Peter ME. (2007). Palmitoylation of Luciani F et al. (2000). CD95 (APO-1/Fas) linkage to the actin CD95 facilitates formation of SDS-stable receptor aggregates that cytoskeleton through ezrin in human T lymphocytes: a novel initiate apoptosis signaling. EMBO J 26: 221–231. regulatory mechanism of the CD95 apoptotic pathway. EMBO J 19: Gupta N, Wollscheid B, Watts JD, Scheer B, Aebersold R, DeFranco 5123–5134. AL. (2006). Quantitative proteomic analysis of B cell lipid rafts Peter ME, Budd RC, Desbarats J, Hedrick SM, Hueber AO, Newell reveals that ezrin regulates antigen receptor-mediated lipid raft MK et al. (2007). The CD95 receptor: apoptosis revisited. Cell 129: dynamics. Nat Immunol 7: 625–633. 447–450.

Oncogene Deficiency in ezrin enhances Fas-induced apoptosis W-C Kuo et al 1383 Peter ME, Krammer PH. (2003). The CD95 (APO-1/Fas) DISC and Scaffidi C, Fulda S, Srinivasan A, Li F, Tomaselli KJ, Debatin KM beyond. Cell Death Differ 10: 26–36. et al. (1998). Two CD95 (APO-1/Fas) signaling pathways. EMBO Piazzolla D, Meissl K, Kucerova L, Rubiolo C, Baccarini M. (2005). J 17: 1675–1687. Raf-1 sets the threshold of Fas sensitivity by modulating Rok-alpha Shaffer MH, Dupree RS, Zhu P, Saotome I, Schmidt RF, McClatchey signaling. J Cell Biol 171: 1013–1022. AI et al. (2009). Ezrin and moesin function together to promote T Ramaswamy M, Dumont C, Cruz AC, Muppidi JR, Gomez TS, cell activation. J Immunol 182: 1021–1032. Billadeau DD et al. (2007). Rac GTPases sensitize activated T cells Shcherbina A, Bretscher A, Kenney DM, Remold-O’Donnell E. to die via Fas. J Immunol 179: 6384–6388. (1999). Moesin, the major ERM protein of lymphocytes and Roumier A, Olivo-Marin JC, Arpin M, Michel F, Martin M, Mangeat platelets, differs from ezrin in its insensitivity to calpain. FEBS P et al. (2001). The membrane-microfilament linker ezrin is involved Lett 443: 31–36. in the formation of the immunological synapse and in T cell Strasser A, Jost PJ, Nagata S. (2009). The many roles of FAS receptor activation. Immunity 15: 715–728. signaling in the immune system. Immunity 30: 180–192.

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

Oncogene