ERM (Ezrin/Radixin/Moesin)-based Molecular Mechanism of Microvillar Breakdown at an Early Stage of Apoptosis

Mouse L929 cells expressing recombinant human Fas antigen (LHF) and mouse MTD-1A cells were cultured in DME supplemented with 10% FCS. HL-60 cells were cultured in RPMI1640 supplemented with 10% FCS.

Ezrin-, radixin-, and moesin-specific mAbs (M11, R21, and M22) were raised in rats using recombinant ezrin, radixin, and moesin, respectively, as antigens (Hirao et al., 1996). Two pAbs, TK89 and TK90, were raised in rabbits against synthesized peptides corresponding to the mouse radixin and ezrin sequences (amino acids 551–570 and 480–489), respectively. TK89 detected all ERM proteins, whereas T90 was specific for ezrin.

LHF cells were detached from the dishes by trypsinization and washed with culture medium. They were then cultured on 2% agar-coated 12-well plates in 1 ml of culture medium at the concentration of 5 × 10⁵ cells/ml. To induce apoptosis, FasL was added to the culture medium at the final concentration of 1 μg/ml (Suda et al., 1993). In some experiments, calyculin A (Research Biochemicals International, Natick, MA) was added to the culture medium with FasL at the final concentration of 300 nM (Morana et al., 1996).

To induce apoptosis, MTD-1A cells were incubated with 0.5 μM staurosporine for 16 h, and HL-60 cells were cultured for 16 h in the presence of 50 μM C2 ceramide, 0.5 μM staurosporine, or 5 μg/ml actinomycin D.

LHF cells were collected by centrifugation, washed with PBS, and then placed on poly-l-lysine–coated coverslips for 15 min for attachment. LHF cells on coverslips or cultured MTD-1A cells were fixed with 3% formaldehyde in PBS for 10 min at room temperature and treated with 0.1% Triton X-100 in PBS for 5 min. After soaking in PBS containing 1% BSA for 10 min, samples were incubated with the first antibody, washed with PBS, and then incubated with the second antibody (FITC-conjugated goat anti– rat IgG or FITC-conjugated goat anti–rabbit IgG [Tago, Inc., Burlingame, CA]) in the presence of 4′,6′-diamidino-2-phenylindole (DAPI). Samples were then washed with PBS and examined using a fluorescence microscope (Axiophoto photomicroscope; Carl Zeiss, Oberkochen, Germany).

Collected LHF cells or HL-60 cells were resuspended and homogenized by weak sonication in a solution containing 130 mM KCl, 20 mM NaCl, 2 mM EDTA, 1 mM EGTA, 50 mM Tris-HCl, pH 7.4, 1 mM p-amidino PMSF, 10 μg/ml leupeptin, and 10 μg/ml aprotinin at 4°C. The homogenates were centrifuged at 100,000 g for 10 min at 4°C to recover the soluble and insoluble fractions in the supernatant and pellet, respectively. Equivalent amounts of supernatant and pellet were applied to SDS–polyacrylamide gels, electrophoresed, and subsequently subjected to immunoblotting with pAb TK89, mAb M11, mAb R21, or mAb M22.

LHF cells were placed in 2% agar-coated 12-well plates in phosphate-free DME with 10% FCS. Phosphate-free FCS was prepared by dialyzing against 0.9% NaCl in 10 mM Hepes buffer, pH 7.4. The cells were cultured for 4 h in the same medium containing 0.5 mCi/ml [³²P]orthophosphate (Phosphorous-32; NEN Life Science Products, Boston, MA).

The labeled cells were lysed and incubated for 5 min in 0.1 ml of solubilization buffer (1% SDS, 10 mM Tris-HCl, pH 7.4, 10 mM Na₃VO₄, 1 mM Na₂MoO₄, and 10 mM p-amidinoPMSF). After addition of 0.9 ml of IP buffer (10 mM Tris-HCl, pH 7.4, 1% Triton X-100, 1% NaDOC, 150 mM NaCl, 2 mM EDTA, 1 mM EGTA, 5 mM NaF, 1 mM Na₃VO₄, 0.1 mM Na₂MoO₄, 1 mM p-amidinoPMSF, and 1 μg/ml leupeptin), the lysate stood on ice for an additional 10 min and was centrifuged at 50,000 g for 20 min. The supernatant was immunoprecipitated with 20 μl of protein G–Sepharose 4B (Zymed Laboratories, Inc., South San Francisco, CA) conjugated with pAb TK89, pAb T90, or mAb M22. Sepharose 4B–bound immune complexes were washed five times with IP buffer containing 0.1% SDS. Immune complexes were then eluted by boiling in SDS-PAGE sample buffer and resolved by SDS-PAGE. After transferring from gels onto nitrocellulose sheets, the ³²P signals were analyzed by autoradiography (Fujix Bioimage Analyzer Bas 200 System; Fuji Photofilm Corp., Tokyo, Japan), and ERM proteins were detected by immunoblotting with pAb TK89.

Antiezrin pAb (TK90) or antimoesin mAb (M22) immunoprecipitates from ³²P-labeled LHF cells with or without 1-h FasL treatment were resolved by SDS-PAGE and transferred onto polyvinyl difluoride membranes (Immobilon; Millipore Corp., Bedford, MA). Phosphoamino acids were analyzed based on the method of Boyle et al. (1991) with minor modifications. ³²P-labeled phosphorylated ezrin or moesin bands were excised from membranes and hydrolyzed in 200 μl of 6 M HCl at 110°C for 60 min. The hydrolysates were lyophilized using a Speed Vac Concentrator (Savant Instruments, Hicksville, NY) and resuspended in 6 μl of pH 3.5 buffer (5% glacial acetic acid, 0.5% pyridine) containing cold phosphoamino acid standards. The samples were then spotted onto silica gel– coated thin-layer chromatography plates and resolved by electrophoresis at 1,300 V for 1.3 h using pH 3.5 buffer at 4°C (model Multiphor II; Pharmacia Biotech AB, Uppsala, Sweden). The positions of ³²P-labeled phosphoamino acids were determined by autoradiography, and cold phosphoamino acid standards were visualized by ninhydrin staining.

LHF cells were labeled with [³²P]orthophosphate for 3 h, and then Ac-YVAD-cho or Ac-DEVD-cho (Takara Shuzo Co., Ltd., Kyoto, Japan) was added at the final concentration of 300 μM. After 1 h incubation, 1 μg/ml of FasL was added. At 3 h after the FasL stimulation, the phosphorylation level of ERM proteins was examined as described above.

One-dimensional SDS-PAGE (7.5%) was performed, based on the method of Laemmli (1970). After electrophoresis, proteins were electrophoretically transferred from gels onto nitrocellulose membranes, which were then incubated with the first antibody. Bound antibodies were visualized with alkaline phosphatase–conjugated goat anti–rabbit IgG and the appropriate substrates as described by the manufacturer (Amersham International, Buckinghamshire, UK).

Cells (1 × 10⁶) were washed with PBS and incubated in 15 μl of 10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10 mM EDTA, 0.5% SDS, and 1 μg/ml proteinase K at 50°C for 30 min. Then, 3 μl of 1 mg/ml RNase A was added and incubated for an additional 1 h. The digested samples were added to the wells of a 2% agarose gel. After electrophoresis, the DNA was visualized by UV illumination after ethidium bromide staining.

LHF cells were cultured in 2% agar-coat plates and vitally labeled for 1.5 h with DiO (170 μg/ml) (Hong and Hume, 1986). The FasL-treated (1 h) or nontreated cells were collected by centrifugation, washed with PBS, and placed on poly-l-lysine–coated coverslips for 15 min for attachment. Then, cells were subjected to jet streams of 5–10 ml of buffer containing 10 mM Hepes, pH 7.4, 1 mM MgCl₂, and 50 mM KCl using a 10-ml syringe with a 23-gauge needle (Tsukita et al., 1984). The isolated plasma membranes on the glass were fixed and stained with rhodamine phalloidin or with pAb TK89 followed by FITC-conjugated goat anti–rat IgG (Tago, Inc., Burlingame, CA). Samples were then washed with PBS and examined using a fluorescence microscope (Axiophoto photomicroscope; Carl Zeiss).

To examine whether ERM proteins are involved in the microvillar disappearance commonly observed in various types of apoptosis, mouse L929 fibroblast transfectants expressing human Fas (LHF) were cultured on agar-coated dishes (Yonehara et al., 1989; Itoh et al., 1991; Nagata and Golstein, 1995), and the apoptosis was initiated by addition of recombinant human FasL (Suda et al., 1993; Nagata and Golstein, 1995). On agar-coated dishes, LHF cells were not spread out but rounded, and they were killed by apoptosis within 3 h as determined by DNA fragmentation ladder formation, retaining their spherical shape (Fig. 1). The fairly quick and synchronized induction of apoptosis in this system allowed us to analyze the cellular events at the early stage of apoptosis in detail. Furthermore, since LHF cells retained their spherical shape without attachment to the dish, this system was advantageous for morphological analysis of the apoptotic cellular changes as well as for biochemical analyses.

Using this system, we first examined the behavior of radixin at the early stage of Fas-mediated apoptosis by immunofluorescence microscopy. In LHF cells before FasL stimulation, radixin was concentrated at the plasma membranes, especially in microvilli, and were also distributed rather diffusely in the cytoplasm (Fig. 2). Within 1 h after FasL stimulation, marked deformation of plasma membranes such as bleb formation was observed. Concomitantly, the concentration of radixin on the plasma membranes faded. At 2–3 h after FasL stimulation, radixin was distributed evenly in the cytoplasm, and DAPI staining identified condensed and fragmented nuclei. As shown in Fig. 3, ezrin and moesin showed the same behavior as radixin at the early stage of Fas-mediated apoptosis of LHF cells.

The cytoplasmic translocation of ERM proteins shown by immunofluorescence microscopy was also detected biochemically (Fig. 4). At various times after FasL stimulation, LHF cells were homogenized in physiological solution and divided into soluble and insoluble fractions by centrifugation. Equivalent amounts of the soluble and insoluble fractions were resolved by SDS-PAGE, and the amount of ERM proteins in each fraction was estimated by immunoblotting with pAb TK89 that recognized all ERM proteins. As shown in Fig. 4,a, in the absence of FasL, ∼50% of ERM proteins were recovered in the insoluble fraction. FasL stimulation gradually decreased the amount of insoluble ERM proteins, and at 1, 2, and 3 h after FasL stimulation, ∼35, 15, and 5% of total ERM proteins were recovered in the insoluble fraction, respectively. The cytoplasmic translocation of ERM proteins was also detected by immunoblotting with ezrin-, radixin-, or moesin-specific antibodies (Fig. 4 b). We then concluded that the microvillar ERM proteins translocate from the plasma membrane to the cytoplasm at an early stage of Fas-mediated apoptosis.

The apoptosis-associated cytoplasmic translocation of ERM proteins was also observed in cultured mouse epithelial cells (MTD-1A) and human promyelocytic leukemic cells (HL-60), although the fairly slow and nonsynchronized induction of apoptosis in these cells prevented analysis of the early events of apoptosis in detail. As shown in Fig. 5,A, when apoptosis was induced in MTD-1A cells by 16 h incubation with staurosporine, radixin as well as ezrin and moesin were translocated from apical microvilli and lateral cell–cell contact sites to the cytoplasm. The 16-h incubation with C2 ceramide, staurosporine, or actinomycin D induced apoptosis in HL-60 cells, and in all cases the cytoplasmic translocation of ERM proteins was clearly detected (Fig. 5 B).

We next examined what types of modification of ERM proteins occur at the time of their cytoplasmic translocation at the early stage of the Fas-mediated apoptosis. The possible degradation of ERM proteins was first postulated since ICE proteases are known to be commonly activated as an early event of apoptosis. As shown in Fig. 4, however, ERM translocation was not associated with ERM degradation. We next examined the phosphorylation levels of ERM proteins since some reports suggested that phosphorylated ERM proteins are concentrated at the plasma membrane. LHF cells were metabolically labeled with [³²P]orthophosphate and stimulated with FasL. At various times after FasL stimulation, ERM proteins were recovered by immunoprecipitation with pAb TK89 that recognized all ERM proteins and analyzed by autoradiography (Fig. 6 a). Before stimulation, ERM proteins were fairly highly phosphorylated, and FasL markedly decreased their phosphorylation level within 1 h. This downregulation continued up to the completion of apoptosis.

To determine which types of phosphoamino acid residues were dephosphorylated, we analyzed phosphoamino acids of ezrin and moesin after 0 and 1 h of incubation with FasL using LHF cells that were metabolically labeled with [³²P]orthophosphate. Ezrin and moesin were recovered by immunoprecipitation with their specific antibodies, pAb TK90 and mAb M22, respectively. Radixin, however, was not purified from LHF cells by immunoprecipitation, partly because the available radixin-specific mAb R21 was not potent for immunoprecipitation, and partly because the expression level of radixin was fairly low in LHF cells. Recovered ezrin and moesin were phosphorylated and dephosphorylated before and 1 h after FasL stimulation, respectively (Fig. 6,b). As shown in Fig. 6 c, without FasL stimulation, serine/threonine as well as tyrosine residues of ezrin and moesin were phosphorylated. Among these phosphoamino acids, phosphothreonine appeared to be preferentially dephosphorylated after 1 h of incubation with FasL, although the levels of phosphorylation of serine and tyrosine residues were also partially decreased. These observations suggested that dephosphorylation on threonine residues of ERM proteins is primarily accompanied by their cytoplasmic translocation during Fas-mediated apoptosis.

To clarify whether the dephosphorylation of ERM proteins is coupled with their cytoplasmic translocation, and, if so, which is located upstream in the FasL-triggered signaling pathway, we stimulated LHF cells with FasL in the presence of 300 nM calyculin A, a potent serine/threonine phosphatase inhibitor (Morana et al., 1996). As shown in Fig. 7,a, calyculin A suppressed the FasL-induced dephosphorylation of ERM proteins at 1 h of incubation, conversely elevating the phosphorylation level of ERM proteins. This suppression continued up to 3 h after FasL stimulation, and the phosphorylation level of ERM proteins remained at the high level. Then, we examined the effects of calyculin A on the FasL-induced cytoplasmic translocation of ERM proteins. Immunoblotting of soluble and insoluble fractions from FasL/calyculin A–incubated LHF cells revealed that ∼90% of ERM proteins were recovered in the insoluble fraction at 1 h of incubation, which continued for at least 4 h, indicating that the FasL-induced cytoplasmic translocation of ERM proteins was suppressed by calyculin A (Fig. 7,b). In agreement with these observations, when these FasL/calyculin A–incubated cells were examined by immunofluorescence microscopy using anti-ERM antibodies, ERM proteins were shown to be concentrated at the plasma membrane (Fig. 7 c). These findings indicated that FasL-induced dephosphorylation of ERM proteins is coupled with their cytoplasmic translocation and that the former is located upstream of the latter.

We next examined whether the dephosphorylation of ERM proteins is located upstream or downstream of the ICE protease cascade. As previously reported, in the presence of 300 μM Ac-YVAD-cho or Ac-DEVD-cho (ICE-like or CPP32-like protease-specific inhibitors, respectively), LHF cells resisted FasL-induced apoptosis (data not shown) (Nicholson et al., 1995). Under these conditions, i.e., when the ICE protease cascade was completely suppressed, the phosphorylation level of ERM proteins was analyzed after FasL stimulation. LHF cells were metabolically labeled with [³²P]orthophosphate and then stimulated with FasL in the presence of 300 μM Ac-YVAD-cho or Ac-DEVD-cho. After incubation for 3 h, ERM proteins were recovered by immunoprecipitation, and their phosphorylation level was examined by autoradiography. As shown in Fig. 8, the FasL-induced dephosphorylation of ERM proteins was mostly suppressed by either Ac-YVAD-cho or Ac-DEVD-cho, indicating that the dephosphorylation of ERM proteins is located downstream of the FasL-induced activation of ICE proteases.

The question then arose as to what type of cellular event is evoked by the cytoplasmic translocation of ERM proteins, i.e., the disappearance of microvilli, in apoptosis. Considering that ERM proteins function as general cross-linkers between plasma membranes and actin filaments, it was expected that gross dissociation of the actin-based cytoskeleton and plasma membranes occur as a result of the cytoplasmic translocation of ERM proteins. To evaluate this expectation, we prepared DiO-labeled single-layered plasma membranes from FasL-treated and nontreated LHF cells (Tsukita et al., 1984; Hong and Hume, 1986). Briefly, free-floating LHF cells were incubated with DiO to vitally label their plasma membrane proper, stimulated with FasL, and then attached to poly-l-lysine–coated coverslips. In control experiments, without FasL stimulation, cells were attached to poly-l-lysine–coated coverslips. At 1 h of incubation, cells were ruptured with a jet stream of hypotonic solution, and the remnant single-layered plasma membranes with the cytoplasmic side upward on the glass were stained with anti-ERM antibodies. As shown in Fig. 9, a and b, when the single-layered plasma membranes were prepared from nontreated cells, they were double positive with DiO and anti-ERM antibodies. In contrast, FasL-treated cells left various sizes of DiO-positive plasma membranes on the cover glasses, almost free of ERM proteins (Fig. 9, c and d). This finding confirmed the FasL-induced cytoplasmic translocation of ERM proteins. Also, in this system the suppression of the cytoplasmic translocation of ERM proteins by calyculin A was observed (Fig. 9, e and f).

Next, the single-layered plasma membranes were stained with rhodamine phalloidin to visualize actin filaments. As shown in Fig. 10, the membranes prepared from the nonstimulated cells were associated with dense actin filaments, whereas those from the FasL-stimulated cells were almost free of actin filaments. Furthermore, calyculin A suppressed this FasL-induced dissociation of actin filaments from plasma membranes. We then concluded that FasL stimulation induced the gross dissociation of the actin-based cytoskeleton from plasma membranes through the cytoplasmic translocation of ERM proteins.

Microvillar disappearance is one of the early common events in various types of apoptosis. In this study, using FasL-induced apoptosis in LHF cells, we found that plasma membrane–bound ERM proteins, ubiquitously expressed microvillar proteins, are translocated to the cytoplasm concomitantly with microvillar disappearance (within 1 h of incubation with FasL). Considering that the depletion of ERM proteins from living cells by antisense oglionucleotide treatment resulted in the complete disappearance of microvilli from the cell surface (Takeuchi et al., 1994b), we conclude that in FasL-induced apoptosis the cytoplasmic translocation of ERM proteins, i.e., the depletion of ERM proteins from plasma membranes, is directly responsible for microvillar disappearance. Furthermore, in this study we analyzed the upstream and downstream events of the microvillar disappearance, i.e., the cytoplasmic translocation of ERM proteins, in the FasL-triggered death signaling pathway, and our conclusions were as follows. FasL stimulation upregulates the ICE protease cascade as recently clarified, which facilitated the dephosphorylation of ERM proteins. This dephosphorylation event is required for ERM proteins to translocate from the plasma membrane to the cytoplasm, which induces gross dissociation of actin filaments from the plasma membrane.

The first question is how activated proteases induce the dephosphorylation of ERM proteins and their subsequent cytoplasmic translocation. In most types of cells, approximately half of the ERM protein is distributed in the cytoplasm (Sato et al., 1992). As described in the introduction, these cytoplasmic soluble ERM proteins take an inactive form, in which the amino- and carboxy-terminal halves mutually suppress their functions through head-to-tail association (Henry et al., 1995; Magendantz et al., 1995; Martin et al., 1995). Some kinds of activation signals may release this mutual suppression by disrupting the intramolecular or intermolecular head-to-tail interaction, allowing the activated ERM proteins to bind to the plasma membranes and actin filaments simultaneously (Arpin et al., 1994; Tsukita et al., 1997a,b). The serine/threonine phosphorylation of ERM proteins has been reported to cause their translocation from the cytoplasm to the plasma membrane (Urushidani et al., 1989; Chen 1995; Nakamura et al., 1995). These observations are consistent with our present finding that the dephosphorylation of ERM proteins is coupled with their cytoplasmic translocation during apoptosis, although dephosphorylation was detected not only at serine/threonine but also at tyrosine residues. Ezrin has been reported to be a good in vivo substrate for tyrosine kinases, but its physiological relevance remains unclear (Bretscher, 1989; Hunter and Cooper, 1981). As during apoptosis, phosphothreonine of ERM proteins appeared to be preferentially dephosphorylated as compared to phosphoserine and phosphotyrosine, and as calyculin A, a potent serine/threonine phosphatase inhibitor, suppressed the apoptosis-associated dephosphorylation of ERM proteins, it is likely that the FasL-activated ICE protease cascade directly or indirectly downregulates some serine/threonine kinase or activates some serine/threonine phosphatase, resulting in the dephosphorylation of ERM proteins. On the other hand, Rho, a small GTP-binding protein, was reported to regulate the activation of ERM proteins through its downstream signaling molecules (Hirao et al., 1996). It is thus also possible that the FasL-activated ICE protease cascade affects the activity of ERM proteins through the Rho-dependent signaling pathway. In this context, it is interesting that D4-GDI, a hematopoietic cell-specific homologue of Rho-GDI, is cleaved by ICE proteases in FasL- and staurosporine-induced apoptosis of T-cell lines (Songqing et al., 1996).

The second question was the subsequent event of the cytoplasmic translocation of ERM proteins, i.e., the breakdown of microvilli at the early phase of apoptosis. ERM proteins are general cross-linkers between actin filaments and the plasma membranes. In this sense, microvilli can be regarded as membrane domains that are specialized for the ERM protein–based interaction of actin filaments with the plasma membrane. Thus the depletion of ERM proteins from the plasma membrane, i.e., the breakdown of microvilli, is expected to result in the gross dissociation of actin filaments from the plasma membrane. The DiO-stained single-layered membrane preparations developed in this study allowed us to evaluate this expectation and led us to conclude that at an early phase of FasL-induced apoptosis, the actin-based cytoskeleton and plasma membrane are completely dissociated, leaving actin filament–free plasma membranes. Many actin filament–plasma membrane cross-linkers other than ERM proteins have been identified, but such gross dissociation of actin filaments and plasma membranes has not been observed by the downregulation of their expression, suggesting the predominant role of ERM proteins in the regulation of actin filament/plasma membrane interaction. Furthermore, this gross dissociation is consistent with our previous results obtained with antisense oligonucleotides that the suppression of ERM protein expression affected several independent cellular events, such as cell–cell adhesion, cell–matrix adhesion, and microvillar formation, where actin filament/plasma membrane interactions may play important roles (Takeuchi et al., 1994b).

During apoptosis, breakdown of microvilli has been reported to be followed by marked changes in cytoskeletal organization. Several cytoskeletal proteins and related signaling molecules, such as actin, fodrin, Gas2, and protein kinase Cδ, have been identified as potential targets for ICE proteases (Brancolini et al., 1995; Chinnaiyan and Dixit, 1996; Fraser and Evans, 1996). These targets may play synchronized roles together with the dephosphorylation of ERM proteins, resulting in marked changes in cytoskeletal organization.

Considering that microvillar breakdown is a common early event in apoptosis, it is likely that the gross dissociation between actin filaments and plasma membranes occurs commonly at an early phase of various types of apoptosis. Actually, the apoptosis-associated cytoplasmic translocation of ERM proteins was observed not only in LHF cells but also in cultured epithelial cells (MTD-1A) and promyelocytic leukemic cells (HL-60). Furthermore, this gross dissociation is located downstream of activation of the ICE protease cascade, which is commonly activated in apoptosis (Enari et al., 1996). This raises questions regarding the physiological relevance of this gross dissociation in apoptosis in general and whether this apoptosis-associated cytoplasmic event is independent from the nuclear events such as chromatin condensation and DNA fragmentation. It is noteworthy that calyculin A significantly suppressed the FasL-induced apoptosis of LHF cells under the conditions used in this study (data not shown). Calyculin A has also been reported to delay the apoptotic process in lymphoma cells induced by gamma-irradiation, tetrandrine, bistratene A, cisplatin, and etoposide (Song and Lavin, 1993; Morana et al., 1996). Considering that calyculin A retained ERM proteins on plasma membranes, it is tempting to speculate that the ERM-based tight association of the actin-based cytoskeleton with the plasma membranes makes cells resistant to death signals, i.e., that destruction of ERM-based actin filament/plasma membrane interactions is required for the apoptotic changes in the nucleus. This speculation is similar to the recent observation that mutated lamin, a nuclear skeletal protein that is resistant to ICE protease–dependent proteolysis, delayed apoptosis (Rao et al., 1996). Of course, it is possible that calyculin A does not suppress the nuclear apoptosis through ERM proteins, but through other proteins directly involved in the nuclear apoptosis. Further analysis of the physiological relevance of the gross dissociation of actin filaments from plasma membranes, i.e., the disappearance of microvilli, in apoptosis will lead to a better understanding of the total picture of the molecular mechanism of apoptosis.

We thank all the members of our laboratory (Department of Cell Biology, Faculty of Medicine, Kyoto University) for their helpful discussions. T. Kondo thanks Prof. H. Saito (Nagoya University) for providing him with the opportunity to work in the Department of Cell Biology, Faculty of Medicine, Kyoto University.

This work was supported in part by a Grant-in-Aid for Scientific Research (B) (to Sa. Tsukita), a Grant-in-Aid for Cancer Research and a Grant-in-Aid for Scientific Research (A) from the Ministry of Education, Science, and Culture of Japan (to Sh. Tsukita).

DAPI

4′,6′-diamidino-2-phenylindole

ERM

ezrin/radixin/moesin

FasL

Fas ligand

ICE

interleukin-1β-converting enzyme