Upon engagement by its ligand, the Fas receptor (CD95/APO-1) is oligomerized in a manner dependent on F-actin. It has been shown that ezrin, a member of the ERM (ezrin-radixin-moesin) protein family can link Fas to the actin cytoskeleton. We show herein that in Jurkat cells, not only ezrin but also moesin can associate with Fas. The same observation was made in activated human peripheral blood T cells. Fas/ezrin or moesin (E/M) association increases in Jurkat cells following Fas triggering and occurs concomitantly with the formation of SDS- and 2-ME-stable high molecular mass Fas aggregates. Ezrin and moesin have to be present together for the formation of Fas aggregates since down-regulation of either ezrin or moesin expression with small interfering RNAs completely inhibits Fas aggregate formation. Although FADD (Fas-associated death domain protein) and caspase-8 associate with Fas in the absence of E/M, subsequent events such as caspase-8 activation and sensitivity to apoptosis are decreased. During the course of Fas stimulation, ezrin and moesin become phosphorylated, respectively, on T567 and on T558. This phosphorylation is mediated by the kinase ROCK (Rho-associated coiled coil-containing protein kinase) I subsequently to Rho activation. Indeed, inhibition of either Rho or ROCK prevents ezrin and moesin phosphorylation, abrogates the formation of Fas aggregates, and interferes with caspase-8 activation. Thus, phosphorylation of E/M by ROCK is involved in the early steps of apoptotic signaling following Fas triggering and regulates apoptosis induction.
Fas (CD95/APO-1), a transmembrane protein belonging to the TNF receptor family, is a key player in apoptosis induction, essential in the control of lymphocyte homeostasis. Patients with inherited loss-of-function mutations of Fas develop a chronic lymphoproliferation termed autoimmune lymphoproliferative syndrome (1). Morever, in some instances, defects in the Fas signaling pathway without Fas gene mutation were found to predispose to development of autoimmune disease and cancer (2). Therefore, the signaling pathways activated after Fas ligation are under intensive investigation.
Fas receptors are preassociated trimers present constitutively at the membrane (3, 4). Engagement of Fas by agonist Abs or by the cognate ligand, FasL, induces its oligomerization. Oligomerized Fas then recruit the adapter molecule FADD (Fas-associated death domain protein)4 and the initiator caspase-8 to form the death-inducing signaling complex (DISC) (5). Two pathways for Fas signaling have been described (6). In type I cells, caspase-8 is recruited to the DISC, resulting in release of active caspase-8 in quantities sufficient to directly activate caspase-3. In type II cells such as Jurkat cells, formation of the DISC is a limiting factor for apoptotic signal, which is then amplified through a mitochondrial loop that involves the proapoptotic member of the Bcl-2-family, Bid, that acts on the mitochondria to promote release of cytochrome c (7, 8, 9).
Although the signaling pathway after formation of the DISC is relatively well understood, the events leading to Fas oligomerization remain less well defined. This oligomerization proceeds in several stages that require the presence of polymerized actin (10). Recently, it has been shown that ezrin, a member of the ERM (ezrin-radixin-moesin) protein family, can link Fas to the actin cytoskeleton (11, 12) and might therefore be a good candidate for the understanding of initial events leading to DISC formation.
ERM proteins have the property to link various membrane proteins (13) with F-actin in microvilli, filopodia, membrane ruffles, and at cell-cell junctions (14, 15, 16, 17, 18, 19). ERMs possess two conserved domains that have been termed N- and C- ERM association domains, or ERMADs (20, 21). The N-ERMAD associates with several membrane proteins (22, 23, 24, 25), whereas the C-ERMAD contains the F-actin binding site (26). These binding sites are masked in cytoplasmic, inactive ERMs due to an intramolecular N-ERMAD/C-ERMAD interaction.
ERMs activation proceeds in two steps: transition from the folded inactive form to the open form dependent on phosphatidylinositol 4,5-bisphosphate (PIP2) (27), and phosphorylation on serine/threonine residues. Phosphorylation of ezrin on threonine 567 and of moesin on threonine 558 have been found to be critical for maintaining the active, open form competent for membrane localization and actin binding (28). To date, the protein serine/threonine kinases that are responsible in vivo for these phosphorylations have not been formally identified (29). Candidate kinases include protein kinase C (30), PIP2-dependent kinase (29), and ROCK (Rho-associated coiled coil-containing protein kinase), an effector of the small GTPase Rho (28).
Fas receptor internalization is necessary for initiation of apoptotic signaling in type I cells (10), and ezrin recruitment to Fas has been shown to be critical for this internalization (31). Such a receptor internalization does not occur in type II cells (10), and therefore the role of ezrin in this setting is not clearly defined. We show herein that after Fas ligation, not only ezrin but also moesin bind to this receptor in type II Jurkat cells. Interestingly, moesin was also found to associate with Fas in activated human T cells that are type I cells. Moreover, we show in Jurkat cells that ezrin and moesin are directly phosphorylated on their activating threonine by the serine/threonine kinase ROCK in a Rho-dependent manner, which facilitates Fas receptor aggregation as well as activation of caspase-8 and apoptosis induction.
Materials and Methods
Cells and reagents
The following human cell lines were used: Jurkat (human acute T cell leukemia cell line), H9 (human T cell line), and Jurkat clones expressing myc-ROCKΔ, VSVG (vesicular stomatitis virus glycoprotein)-tagged ezrin mutants T567A or T567D. ROCKΔ is a constitutive active fragment of ROCK I, truncated at aspartate 1113, corresponding to the fragment generated by caspase-3 cleavage (32). These cell lines were routinely maintained in RPMI 1640 (Invitrogen) supplemented with 10% FCS, 1% sodium pyruvate, 0.1 mg/ml streptomycin, and 100 U/ml penicillin at 37°C in a 5% CO2 humidified atmosphere.
Mononuclear cells were isolated from peripheral blood by density gradient centrifugation, and monocytes were eliminated on anti-CD14-coated beads. Resulting PBLs were activated by OKT3 mAb in RPMI 1640 (Invitrogen) supplemented with 10% FCS, 1% sodium pyruvate at 37°C in a 5% CO2 humidified atmosphere. At day 3, dead cells were removed by density gradient centrifugation. T cells were maintained in 7.5 μg/ml of human recombinant IL-2 until day 9. For the induction of activation-induced cell death (AICD), activated T cells were stimulated with PMA (50 ng/ml) and ionomycin (500 ng/ml) for 4, 12, and 24 h. In Fas-mediated apoptosis studies, the agonist anti-Fas Ab clone 7C11 (Immunotech) was used at 20 ng/ml for Jurkat cells and at 120 ng/ml for activated T cells.
Y-27632 (Calbiochem), a selective ROCK inhibitor, was used at 20 μM with a 2-h preincubation time. TAT-C3, a cell-permeable protein inhibitor of Rho, was prepared in the laboratory as described (32) and was used at 40 μg/ml with a 2-h preincubation time. Ionomycin and PMA were purchased from Sigma-Aldrich.
The following Abs were used: anti-ERM (Cell Signaling Technology), anti-phospho-ezrin (Thr567)/moesin (Thr558)/radixin (Thr564) (Cell Signaling Technology), anti-Fas APO-1–3 biotin, anti-caspase-8 (both from Alexis Biochemicals), anti-Fas clone B10, anti-FADD (Transduction Laboratories), anti-ROCK I (H85), anti-Hsc70 (both from Santa Cruz Biotechnology), and anti-RhoA (clone 26C4, produced in the laboratory; see Ref. 33).
Apoptotic cells were quantified by flow cytometry with propidium iodide (Sigma-Aldrich). Cells were washed twice in PBS, incubated overnight in 70% ethanol at −20°C, and stained with propidium iodide (50 μg/ml in PBS). Data acquisition was performed on a FACSCalibur flow cytometer and analysis was performed with CellQuest software (BD Biosciences). Pseudo-hypodiploid apoptotic cells display sub-G1 DNA content.
Resting or stimulated cells were pelleted and washed in cold PBS. For detection of Fas (monomers and aggregates), caspase-8, ERM, and phospho-ERM cells were lysed in Laemmli buffer (60 mM Tris (pH 6.8), 10% glycerol, and 2% SDS) and lysates were sonicated gently on ice. All lysates were clarified by centrifugation at 15,000 rpm for 15 min at 4°C. Protein concentration was assessed using a BCA (bicinchoninic acid) kit (Pierce). For SDS-PAGE, proteins were loaded at 60 μg (for caspase-8), 30 μg (for ERM and phospho-ERM), and 100 μg (for Fas monomers and aggregates) and were transferred to a polyvinylidene difluoride (PVDF) membrane (GE Healthcare) overnight at a constant voltage of 40 V. After blocking for at least 90 min in TBS, 5% BSA, and 0.1% Tween 20, blots were probed with the appropriate Abs and proteins were visualized by ECL (GE Healthcare).
The following lysis buffers were used: 1% Nonidet P-40 buffer (30 mM Tris (pH 7.5), 150 mM NaCl, 1% Nonidet P-40, 10% glycerol, 1 mM PMSF, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 1 mM Na3VO4) for Fas IP; and Brij buffer (25 mM Tris (pH 7.5), 50 mM NaCl, 5 mM MgCl2, 2 mM EGTA, 1% Brij 58, 1 mM tetrasodium pyrophosphate, 1 mM PMSF) for ROCK IP. After clarification, equal quantities of proteins were incubated with appropriate Ab overnight at 4°C. Immune complexes were collected with streptavidin or protein G-Sepharose beads (Sigma-Aldrich). Beads were washed three times in lysis buffer, boiled in sample buffer, and proteins were separated by SDS-PAGE. As above, blots were probed with the appropriate Abs and proteins visualized by ECL (Amersham). Densitometric analysis of the blots was performed using the Quantity One software (Bio-Rad Laboratories).
Small interfering RNA (siRNA) synthesis
Human serum (HS) ezrin, HS moesin HP validated siRNA, and All Star negative control siRNA were from Qiagen. The siRNA sequence used for silencing of ezrin are: sense, r(GGA CUG AUU GAA UUA CGG A)dTdT and antisense, r(UCC GUA AUU CAA UCA GUC C)dTdG; of moesin: sense, r(GGG AUG UCA ACU GAC CUA A)dTdT and antisense, r(UUA GGU CAG UUG ACA UCC C)dTdG; and of control siRNA: sense, r(UUC UCC GAA CGU GUC ACG U)dTdT and antisense: r(ACG UGA CAC GUU CGG AGA A)dTdT.
For electroporation, a Gene Pulser Xcell electroporation system (Bio-Rad) and preoptimized programs for Jurkat cells (160 V, pulse length 15 ms, 1 pulse, cuvette 2 mm) were used. The delivery of siRNA in activated PBLs was done with the electroporation method described by Prechtel et al. (34). Briefly, cells were washed in RPMI 1640 without FCS and then in PBS, and resuspended at 40 × 106 cells/ml. One hundred microliters of this suspension was used per cuvette with 5 μg of the corresponding siRNA. The parameters for each cuvette was: 300 V, 150 μF, Ω = ∞, cuvette 4 mm. Seventy-two hours after transfection cells were used for apoptosis and biochemistry studies.
RhoA-GTP pull-down assay
The Rho-binding domain of rhotekin (RBD) was expressed in Escherichia coli with the use of pGEX-2T (Pharmacia) encoding GST-fusion protein. Cells transformed with pGEX were grown to midexponential phase, induced for 4 h with 1 mM isopropyl β-d-thiogalactopyranoside (IPTG), and lysed in PBS supplemented with 10 mM DTT, 1 mM PMSF, 2 mg/ml lysozyme, 1% Triton X-100, and 10 mM MgCl2. GST-RBD was purified from bacterial lysates by incubation with glutathione-Sepharose beads (Pharmacia).
Cells were washed in PBS and lysed in RIPA buffer (50 mM Tris (pH 7.2), 1% Triton X-100, 0.5% deoxycholic acid, 0.1% SDS, 500 mM NaCl, 20 mM MgCl2, 1 mM PMSF, and 10 μg/ml each leupeptin and aprotinin). Cell lysates were clarified by centrifugation at 15,000 rpm for 15 min at 4°C and equal quantities of proteins were incubated with GST-RBD beads at 4°C for 45 min. The beads were washed four times with lysis buffer (with 150 mM NaCl) and proteins associated with GST-RBD bound to beads were separated by SDS-PAGE. Immunoblotting was conducted using a specific anti-RhoA mAb.
Expression, mutagenesis, and purification of GST fusion proteins
Full-lengh ezrin cDNA was kindly provided by Dr. Monique Arpin. Ezrin cDNA was cloned into pGEX-2T vector to generate GST-ezrin wild type (WT). Full-length human moesin cDNA, intracytoplasmic domain of Fas were amplified from Jurkat cell RNAs using specific primers containing EcoRI and SalI restriction sites. The resulting PCR products were digested with EcoRI and SalI and cloned into the same sites of pGEX-4T2 vector to generate GST-moesin WT and GST-Fas. To substitute Thr with Ala in ezrin and moesin, PCR reactions were performed with oligonucleotides in which the codon ACG of ezrin (Thr567) and the codon ACC of moesin (Thr558) were replaced by GCG (Ala567) and by GCC (Ala558), respectively. The sequence was then verified. GST fusion proteins were expressed in bacteria as described above, purified with glutathione-Sepharose 4B beads, eluted in buffer (5 mM Tris (pH 8), 20 mM NaCl, 1 mM EDTA, 20 mM DTT, and 100 mM reduced glutathione). GST-ezrin WT, GST-moesin WT, GST-ezrin T567A, and GST-moesin T558A were then dialyzed against kinase reaction buffer (50 mM Tris (pH 7.4), 10 mM MgCl2, 2 mM MnCl2, 1 mM DTT, 1 mM EDTA, and 1 mM EGTA).
Jurkat cells expressing active Myc-tagged ROCKΔ were lysed in buffer (25 mM HEPES (pH 7.3), 300 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.5% Triton X-100, 5% glycerol, 20 mM glycerophosphate, and inhibitors). Cell lysates were incubated overnight at 4°C with anti-Myc Ab and then incubated for 1 h with protein G-Sepharose beads. Immunocomplexes were washed three times with lysis buffer and three times with kinase reaction buffer. Kinase reactions were performed with the immunoprecipitated extracts at 37°C for 30 min in kinase reaction buffer with 10 μM ATP, and GST, GST-ezrin WT, GST-moesin WT (with or without Y27632 compound), GST-ezrin T567A, and GST-moesin T558A as substrate. Reactions were stopped by addition of sample buffer. Proteins were separated by SDS-PAGE and immunoblotted with anti-phospho-ERM, anti-total ERM, and anti-Myc Abs.
Ezrin or moesin depletion by siRNAs decrease sensitivity of Jurkat cells to Fas-mediated apoptosis
To test whether E/M could be involved in Fas-mediated apoptosis, siRNAs were used to knockdown either ezrin (si-ezrin) or moesin (si-moesin) in Jurkat cells. Efficiency and specificity of targeted siRNA sequences were confirmed by ERM immunoblot on total lysates 72 h posttransfection. Thus, E/M expression was significantly and specifically reduced (∼60–70%) by their respective siRNAs as compared with cells transfected with control siRNAs (Fig. 1,A). Jurkat cells depleted of E/M were stimulated with the agonist anti-Fas Ab (clone 7C11) and the level of apoptotic cells was measured, in kinetics, by flow cytometry quantifying the hypodiploid cell population. As shown in Fig. 1,B, E/M down-regulation led to a significant decrease of apoptosis since almost 50% less apoptotic cells were detected as compared with cells transfected with control siRNAs, and this occurred even 8 h after stimulation. Since ezrin and moesin down-regulation each led to the same decrease in apoptosis, we investigated the effects of targeting both ezrin and moesin together to determine whether their effect could be additive. Fig. 1 B (white bar) shows that the down-regulation of expression of both ezrin and moesin has no additional effect on apoptosis compared with single ezrin or moesin depletion. Thus, this result indicates that even thought ezrin and moesin share high structural homology, they cannot substitute each other, at least for their inductive effect on Fas-mediated apoptosis. The decrease in apoptosis sensitivity induced by ezrin or moesin silencing does not seem to be a general effect on apoptosis signaling since it does not modify significantly the apoptotic level observed in cells treated by an intrinsic apoptosis inducer such as etoposide (data not shown). These results demonstrate that apoptosis mediated by Fas receptors is sensitive to the level of E/M expression.
Inducible association of E/M and Fas receptor in Jurkat cells
To understand how E/M affects Fas signalization, and since ezrin has already been described to interact with Fas receptor (12), we first tested whether ezrin associates with Fas receptor in Jurkat cells. To this aim, Jurkat cells were stimulated with the 7C11 Ab and lysates were prepared followed by an anti-Fas (anti-Fas APO-1–3 biotin) IP. As shown in Fig. 2, immunoblotting with anti-ERM Ab demonstrated a co-IP of Fas receptor with ezrin, as expected, but also with moesin. However, in unstimulated cells basal association of E/M and Fas was observed and Fas stimulation increased it up to 2 h (Fig. 2).
Fas activation has been shown to induce the formation of SDS- and 2-ME-resistant Fas aggregates, which can be detected as high molecular mass complexes (>200 kDa) (5, 35). Reprobing the blot with an anti-Fas (clone B10) Ab indicated that formation of Fas aggregates, which was accompanied by a concomitant decrease of Fas monomer, occurred with a kinetic parallel to Fas-E/M association and DISC formation as assessed by the co-IP of the adapter protein FADD (Fig. 2). Altogether, these results show that in Jurkat cells, Fas receptor appears constitutively associated with ezrin and moesin and that Fas ligation increases this association.
Ezrin and moesin are required for Fas aggregate formation after Fas ligation
Since inhibition of ezrin and/or moesin expression decreased the percentage of cells undergoing apoptosis and since there is an inducible interaction between Fas and E/M concomitantly to Fas aggregates and DISC formation, we investigated whether E/M could be involved in those events. The appearance of Fas aggregates was thus studied in ezrin- or moesin-depleted Jurkat cells stimulated with the 7C11 Ab. As shown in Fig. 3 A, although Fas aggregates, as well as reduction in Fas monomers, could be easily detected after 2 h of stimulation in control cells, neither ezrin- nor moesin-depleted cells exhibited any Fas aggregates or Fas monomer variations. Thus, it appeared that formation of Fas aggregates requires the presence of both ezrin and moesin.
Since Fas complexes initiate the apoptotic pathway by recruitment of the adapter molecule FADD and the initiator caspase-8 to form the DISC, we investigated whether E/M down-regulation could have consequences on caspase-8 activation. To this aim, extracts prepared in the same conditions as above were subjected to caspase-8 immunoblotting. Pro-caspase-8 observed at 53/55 kDa appears to have decreased during the kinetics concomitantly with the appearance of its active fragments visualized as p41/43 and p18 in control conditions (Fig. 3,A). In cells depleted of either ezrin or moesin, a reduction in the appearance of these fragments was noticeable and was 50% when quantified for the p18 (Fig. 3 A).
We then studied the effect of ezrin and moesin down-regulation on their interaction with Fas. To this aim, Fas was immunoprecipitated in control Jurkat cells and in cells where either ezrin or moesin expression was inhibited. As shown in Fig. 3,B, neither loss of ezrin or moesin leads to an increased association of the other protein (moesin or ezrin, respectively) with Fas. The down-regulation of ezrin or moesin does not affect the recruitment of FADD and pro-caspase-8 to the DISC but it affects the activation of caspase-8 as shown in Fig. 3 A.
Moreover, pull-down assay with Fas receptor intracellular domain fused to GST in control Jurkat lysates or cells silenced for E/M confirmed that loss of one of these proteins did not affect the binding of the other, demonstrating independent Fas receptor binding properties for each of them (Fig. 3 C).
Taken together, these data clearly indicate that ezrin and moesin take part in the early events that follow Fas ligation. First, these two proteins do not compete for Fas binding and both must be present to allow the formation of resistance to SDS and 2-ME Fas aggregates. Second, preventing Fas aggregate formation reduces apoptotic signaling such as caspase-8 activation.
Ezrin and moesin are phosphorylated after Fas ligation in a Rho- and ROCK-dependent manner
ERM interaction with membrane receptors requires their activation, which involves their opening by PIP2, usually a Rho-dependent process, followed by their phosphorylation on a conserved Thr567 for ezrin or Thr558 for moesin (29). We analyzed the phosphorylation state of E/M after Fas receptor ligation by immunoblot with an anti-phospho-ERM Ab. As shown in Fig. 4 A, resting cells exhibit constitutive E/M phosphorylation. The phospho-ERM signal decreases, as soon as 2 min of Fas stimulation, and then increases again after 5 min of stimulation.
As ezrin phosphorylation on Thr567 has been described as being potentially Rho-dependent (36), we then evaluated Rho activation by GST-RBD pull-down assay in Jurkat cells stimulated for Fas signaling. As shown in Fig. 4 B, the amount of GTP-RhoA increased rapidly 2 min after Fas engagement, and this activation remained stable for at least 45 min.
To explore whether Rho is involved in E/M phosphorylation, we analyzed the consequences of its inhibition by the use of a cell-permeant form of Clostridium botulinum C3 toxin, a specific inhibitor of Rho (32). Jurkat cells were incubated 2 h with the C3 toxin before Fas stimulation and then E/M phosphorylation was analyzed. Fig. 4 C shows that Rho inhibition almost completely abolishes ezrin and moesin phosphorylation, indicating that Rho activity is required in this process.
Among Rho effectors, the serine/threonine kinase ROCK appeared as a good candidate to mediate the Rho effect on E/M phosphorylation. To assess this, we preincubated Jurkat cells with a specific ROCK inhibitor, the Y-27632 compound, and analyzed E/M phosphorylation upon Fas signaling (Fig. 4 D). Thus, it appeared that, like C3 toxin, ROCK inhibition prevented E/M phosphorylation. The efficiency of ROCK inhibition by the Y-27632 compound was determined by the lack of phosphorylation of the myosin L chain.
The sustained E/M phosphorylaion after Fas ligation implies that the corresponding kinase is at close proximity. Most interestingly, we found that ROCK is present in anti-Fas immunoprecipitates (Fig. 4 E). The faster migrating band revealed at late time points (2 h) corresponds to the cleavage fragment generated by caspase-3 (32).
ROCK binds to and phosphorylates ezrin and moesin
As shown above, inhibition of Rho or ROCK alters ezrin and moesin phosphorylation. We then investigated whether ezrin and moesin could interact with ROCK. Lysates from Fas-stimulated Jurkat cells were subjected to IP with anti-ROCK I Ab followed by immunoblotting with anti-ERM Ab. As shown in Fig. 5,A, ezrin and moesin coimmunoprecipitated with ROCK I, and this association was not modified by Fas stimulation. Then, we determined whether ROCK was able to phosphorylate ezrin and moesin. To this aim, in vitro ROCK kinase assays were performed using active ROCK I immunoprecipitated from Jurkat cells stably expressing a Myc-ROCKΔ construct and GST-ezrin or GST-moesin as exogenous substrate. As shown in Fig. 5 B, ROCKΔ efficiently phosphorylated WT GST-ezrin as well as WT GST-moesin, whereas no phosphorylation was detected in WT GST proteins in the presence of Y27632 compound or in GST proteins bearing T567A or T558A mutations. Collectively, these results indicate that ROCK can bind ezrin and moesin and induce their phosphorylation at the Thr567 or Thr558 position, respectively.
Ezrin and moesin phosphorylation is involved in the formation of Fas receptor aggregates and apoptotic signaling
To define whether ezrin and moesin phosphorylation contributes to their role in Fas signaling, we evaluated the effect of ROCK inhibition on the formation of Fas aggregates. To this end, Jurkat cells were incubated with the ROCK inhibitor Y27632 before Fas stimulation. As shown in Fig. 6,A, ROCK inhibition prevented SDS- and 2-ME-resistant Fas aggregate formation as well as caspase-8 activation (Fig. 6,B) and reduced the apoptosis level (Fig. 6 C). Therefore, ROCK inhibition, and thus prevention of ezrin and moesin phosphorylation, interferes with Fas signaling by affecting Fas aggregate formation and subsequent events such as caspase-8 activation and thus apoptosis level.
Ezrin and moesin phosphorylation sensitizes Jurkat cells to Fas-induced apoptosis
Since ROCK inhibition reduced the sensitivity of Jurkat cells to Fas-induced apoptosis, we determined whether an increase in ROCK activity could increase this sensitivity. To this end, we used a Jurkat clone expressing ROCKΔ. As shown in Fig. 7,A, cells expressing ROCKΔ exhibited constitutive phosphorylation of E/M without variation due to Fas stimulation. Parallel analysis of the induced apoptosis level showed that Jurkat cells with a deregulated ROCK activity were more sensitive to Fas-induced apoptosis (Fig. 7,B) and have an increased rate of caspase-8 activation compared with control cells (Fig. 7 C).
We then generated clones expressing the phosphomimetic T567D ezrin mutant or the nonphosphorylable T567A ezrin mutant. Isolated clones were tested for their expression level of ezrin and Fas receptor and in both cases no significant difference was found (data not shown). The formation of Fas aggregates was studied in these clones after Fas stimulation. As shown in Fig. 8,A, the expression of the T567D ezrin mutant increased the rate of Fas aggregate formation compared with control Jurkat cells, whereas the nonphosphorylable ezrin mutant (T567A) substantially interfered with Fas aggregate formation. That the T567A mutant does not totally abrogate Fas aggregates and apoptotic signaling is probably due to a contribution of the endogenous ezrin. As expected, these effects on Fas aggregates correlated with the efficiency of Fas signaling since cells expressing T567D ezrin appeared more sensitive whereas ezrin T567A clones were less sensitive to Fas-mediated apoptosis compared with Jurkat control cells (Fig. 8 B). Altogether, these results demonstrate that the level of phosphorylated ezrin and moesin induced by ROCK activity or by mimetic mutant correlates with sensitivity to Fas-induced apoptosis, an effect that may be linked to Fas aggregate formation.
Association of Fas, ezrin, moesin, and ROCK in peripheral blood T cells
To strengthen these observations, we investigated whether Fas, ezrin, moesin, and ROCK also associate in normal human T cells. Resting T cells are known to be resistant to Fas-induced apoptosis, whereas activated T cells become sensitive (37). Therefore, we stimulated naive T cells with anti-CD3 Ab for 3 days, followed by 1 wk of incubation with IL-2. Cells were then subjected to the anti-Fas Ab 7C11, and Fas IP was performed at several time points. As shown in Fig. 9, no DISC was immunoprecipitated in resting T cells. In contrast, caspase-8 was found to be already cleaved in activated T cells before the addition of the 7C11 Ab. Association of active caspase-8 with lipid rafts has been described in TCR-activated T cells and described to be involved in cell proliferation (38). However, this does not seem to be the case in our study, as active caspase-8 was found to be associated with Fas after IP at time 0. We detected an apoptosis rate of ∼10% at this time, probably due to the expression of FasL in activated T cells. Peripheral blood T cells are type I cells with very efficient caspase-8 recruitment and processing compared with type II cells such as Jurkat cell line. Therefore, 10% death might be sufficient to generate the level of active caspase-8 and the presence of Fas aggregates that we observed. Interestingly, ezrin and moesin as well as native and cleaved ROCK I are immunoprecipitated with Fas. Ezrin and moesin were not found associated with Fas at time 0 in another type I cell, the H9 T cell line (Fig. 9,B), confirming the idea that the association found in peripheral blood T cells is due to apoptosis. Indeed, the fact that cleaved ROCK I is present in the complex indicates that caspase-3 is active at time 0 in activated T cells (the cleavage of caspase-3 was verified; data not shown). Stimulation of Fas led to a more complete caspase-8 cleavage as shown by the appearance of the 18-kDa fully mature fragment. This fragment is not found in IP as it dissociates from the DISC. Intriguingly, IP of the DISC appears to be less efficient at 12 h of incubation with Fas. This might be due to the fact that in type I cells the complex is internalized (10) and thus less accessible for IP. In any case, these results show that in primary T cells ezrin and moesin are found associated with Fas. We thus silenced their expression to investigate whether this association was involved in apoptosis signaling. As shown in Fig. 9,C, we were more efficient in inhibiting ezrin expression compared with moesin. This may be due to the fact that siRNAs are much more difficult to use in primary cells and that basal moesin expression is higher. However, the decrease of ezrin or moesin expression has an effect since the siRNA-transfected cells are less sensitive to Fas-mediated apoptosis (Fig. 9 D).
Upon repeated TCR triggering, activated T cells undergo AICD, which is mediated via the Fas/FasL system (39). The second TCR activation can be mimicked by a combination of PMA and ionomycin (40). We therefore investigated whether ezrin and moesin could be involved in this form of Fas-dependent apoptosis. As shown in Fig. 9 E, silencing ezrin or moesin was efficient in inhibiting death by AICD of activated T cells. Inasmuch as in this setting apoptosis is triggered by endogenous FasL, these results suggest that ezrin and moesin may be involved in AICD in vivo.
Collectively, these data indicate that in peripheral blood T cells, ezrin and moesin participate in Fas signaling, as is the case in the Jurkat cell line.
Constitutive linkage of Fas to the cytoskeleton via ezrin has been described in CEM cells and activated T lymphocytes (12). Our report is in agreement with this study, as we find a basal association between Fas and ezrin in unstimulated Jurkat cells. However, we also established that Fas is complexed with moesin, the other ERM protein present in Jurkat cells (radixin is not expressed in these cells). We also show, in contrast to the study of Parlato et al. (12), that moesin is associated with Fas in activated human peripheral blood T cells and that both ezrin and moesin are involved in Fas signaling in this cell type.
The site of interaction of ezrin with Fas has been identified (11), and this site is not conserved in moesin. Thus, ezrin and moesin probably do not share the same binding domain within the Fas intracellular domain. Moreover, we show that cross-linking of Fas with agonist Abs increases the association of both ezrin and moesin with Fas. These results suggest either that a pool of Fas receptor, free of E/M in resting cells, recruits new E/M molecules or that E/M binding sites for E/M are unmasked on Fas upon stimulation.
We further demonstrated that ezrin and moesin are involved in Fas oligomerization, as down-regulation of their expression with the corresponding siRNAs prevented the formation of SDS- and 2-ME-resistant Fas aggregates, indicating that both proteins are necessary. In the absence of E/M, caspase-8 activation was diminished and the percentage of apoptotic cells was reduced. We could show by IP that down-regulation of E/M did not interfere with DISC assembly, as shown by an equal recruitment of FADD and caspase-8, but that it reduced the level of activated caspase-8. We have no definite explanation for the decreased caspase-8 activation in the absence of E/M but suggest that the tight proximity between several pro-caspase-8 molecules that is needed for their autoactivation is not achieved due to a less efficient Fas clustering. Simultaneous inhibition of both ezrin and moesin expression did not lead to any further inhibition compared with inhibition of either ezrin or moesin alone. Importantly, knocking down ezrin or moesin expression while totally abrogating the formation of SDS- and β2-ME-resistant aggregates diminishes the amount of apoptotic cells only by half. Thus, aggregate formation would solely amplify the death signal.
Activation of ERM has been described as dependent on threonine phosphorylations, and we found that ezrin and moesin are constitutively phosphorylated in Jurkat cells. Fas ligation induces a rapid dephosphorylation that is followed by a sustained rephosphorylation. This phosphorylation is dependent on the GTPase Rho and its effector ROCK. Indeed, we show that Fas cross-linking leads to a sustained Rho activation and that inhibition of either Rho or ROCK abrogates ezrin and moesin phosphorylation. Preventing ezrin and moesin phosphorylation by ROCK inhibition interfered with the formation of Fas aggregates, diminished caspase-8 activation, and reduced the percentage of apoptotic cells. Thus the Rho/ROCK cascade, through the activation of ezrin and moesin, is involved in Fas-induced apoptotic signaling. These results are confirmed by the fact that Jurkat cells expressing an active ROCK variant, and in which ezrin and moesin remain phosphorylated, are more prone to apoptosis induction after Fas ligation. Moreover, Jurkat cells expressing a phosphomimetc T567D ezrin mutant are sensitized to aggregate formation and apoptosis compared with their wild-type counterpart. Conversely, the expression of a T567A ezrin mutant is inhibitory.
Our results showing that Fas aggregation is ROCK-dependent are consistent with recently published work showing that Fas capping is a ROCK-regulated process (41). However, these authors also reported that Fas capping is not implicated in apoptotic signaling. In a reciprocal manner, Strauss et al. (42) found by immunofluorescence microscopy that Fas receptors were capped in the absence of cell death in long-term-activated T cells. Moreover, these cells were resistant to Fas triggering. Most likely, the aggregates that we detect biochemically and the observation of Fas capping by imaging techniques do not relate to the same phenomenon. Additionally, the very high amount of anti-Fas Ab used in the above study (41), 10-fold higher than ours, prevents a clear comparison between the two sets of results.
It is generally assumed, based on the in vitro data of Matsui et al. (28), that ROCK phosphorylates ERM directly. However, in a subsequent publication, the same authors concluded that ROCK is not responsible for threonine phosphorylation of ERM in vivo in Hela cells transiently tranfected with HA-PI4P5K (29). Our data are consistent with the notion that during the course of Fas stimulation, ROCK directly phosphorylates ezrin and moesin based on the following results: ezrin and moesin are not phosphorylated in presence of the Y-27632 ROCK inhibitor, and the expression of active ROCKΔ leads to constitutive ezrin and moesin phosphorylation. Morever, ezrin and moesin are immunoprecipitated with ROCK I in cell lysates. Our in vitro data are also in favor of a direct phosphorylation of ERM by ROCK. Finally, and most interestingly, ROCK is present in the complex immunoprecipitated with anti-Fas Abs, suggesting that it is maintained in close proximity with its substrates ezrin and moesin during the course of Fas activation.
We have previously reported that during apoptosis, ROCK is cleaved by caspase-3, generating a constitutive active fragment, ROCKΔ (32). Theoretically, this fragment could mediate the phosphorylation of ezrin and moesin that we observe. However, this is likely not to be the case given that ezrin and moesin are phosphorylated before the activation of caspases. Moreover, this phosphorylation is abrogated by Rho inhibition, indicating that it is due to a classical Rho/ROCK pathway. Indeed, we found that Rho is activated in a sustained manner after Fas ligation. The notion that Rho can increase Fas-induced apoptotic signaling has previously been inferred by Subauste et al. (43), who showed that microinjection of a constitutive active Rho mutant sensitizes CHO cells to Fas-induced cell death. The molecular mechanisms by which Fas triggering can lead to Rho activation are currently under investigation in our laboratory.
We consistently observed that a very rapid and transient ERM dephosphorylation occurred 2 min after Fas triggering. Such a dephosphorylation has also been described in T cells after TCR stimulation and in B cells after BCR triggering (44, 45, 46). The release of ERMs from the cortical actin cytoskeleton has been shown to decrease membrane rigidity in T cells (47). This transient disanchoring might allow for ezrin reorganization and formation of new linkages at the membrane after Fas triggering.
In conclusion, we have provided evidence that ezrin and moesin phosphorylation through the Rho-ROCK pathway increases their association with Fas and facilitates Fas aggregation, caspase-8 activation, and apoptotic signaling. Inasmuch as ERM proteins link membrane receptors with cortical actin, our results confirm the notion that the actin cytoskeleton participates in the Fas-activated death signal.
We thank Dr. M. Arpin for providing pcB6-ezrin constructs.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by Institut National de la Santé et de la Recherche Médicale, Ligue Nationale contre le Cancer (équipe labellisée 2006–2009) and Agence Nationale de la Recherche Grant BLAN06-2_135169.
Abbreviations used in this paper: FADD, Fas-associated death domain protein; AICD, activation-induced cell death; DISC, death-inducing signaling complex; E/M, ezrin or moesin; ERM, ezrin-radixin-moesin; IP, immunoprecipitation; PIP2, phosphatidylinositol 4,5-bisphosphate; RBD, Rho-binding domain of rhotekin; ROCK, Rho-associated coiled coil-containing protein kinase; si, small interfering; WT, wild type.