Early Events in the Anoikis Program Occur in the Absence of Caspase Activation*

Adhesion of many cell types to the extracellular matrix is essential to maintain their survival. In the absence of integrin-mediated signals, normal epithelial cells undergo a form of apoptosis termed anoikis. It has been proposed that the activation of initiator caspases is an early event in anoikis, resulting in Bid cleavage and cytochrome c release from mitochondria. We have previously demonstrated that the loss of integrin signaling in mammary epithelial cells results in apoptosis and that this is dependent upon translocation of Bax from the cytosol to the mitochondria. In this paper, we ask whether caspases are required for Bax activation and the associated changes within mitochondria. We show that Bax activation occurs extremely rapidly, within 15 min after loss of integrin-mediated adhesion to extracellular matrix. The conformational changes associated with Bax activation are independent of caspases includ-ing the initiator caspase-8. We also examined downstream events in the apoptosis program and found that cytochrome c release occurs after a delay of at least 1 h, with subsequent activation of the effector caspase-3. This delay is not due to a requirement for new protein synthesis, since cycloheximide has no effect on the kinetics of Bax activation, cytochrome c release, caspase-3 cleavage, or apoptosis. Together, our data indicate that the cellular decision for anoikis in mammary epithelial cells occurs in the absence of caspase activation. More-over, although the conformational

Apoptosis is an important cellular mechanism for packaging and removing unwanted, damaged, or infected cells (1). Normal cells are kept alive through the action of signal transduction pathways triggered by extracellular ligands but undergo apoptosis if survival factors are withdrawn or if a cell detects inappropriate stress or damage. Apoptosis is mediated by dynamic changes within Bcl-2 family proteins that lead to alterations in mitochondrial homeostasis and activation of the apoptosome. Certain cells may alternatively enter an extrinsic apoptosis program if they are targeted by cells of the immune system, through ligand binding of death domain-containing surface receptors such as Fas (2). This results in formation of the death-inducing signaling complex and then cleavage and activation of initiator caspases. The subsequent activation of effector caspases can occur independently of mitochondria. Thus, the cellular decisions for these two modes of apoptosis regulation are distinct. On the one hand, kinase-mediated signaling pathways control the post-translational modification and/or subcellular redistribution of Bcl-2 proteins to mitochondria, with the indirect and subsequent activation of caspases, whereas on the other hand multiprotein complexes are assembled at the plasma membrane to orchestrate caspase activation directly. Subtle variations on this theme also exist because cross-talk can occur between the mitochondrial and death receptor pathways.
Many cell types are naturally adherent in vivo and form integrin-mediated interactions with the extracellular matrix (ECM). 1 Integrins are transmembrane receptors that link components of the ECM with intracellular structural elements of the cytoskeleton through adhesion complexes. Associated with such complexes are signaling proteins whose activation is critically required for migration, differentiation, cell cycle, and survival (3). Activation of signaling proteins such as pp125 FAK within the adhesion complex is achieved only when integrins are ligated to the ECM and does not occur in cells that have been detached from the matrix (4). Adherent cells normally require integrin-mediated adhesion in order to stay alive, and the downstream signals resulting from integrin activation cooperate with those from soluble factors to prevent apoptosis (5)(6)(7)(8)(9). Apoptosis mediated by the loss of integrin signaling is sometimes referred to as anoikis (10). Two broadly different mechanisms have been proposed for adhesion control of cell survival. One of these involves signaling pathways that link integrins with apoptosis commitment events in mitochondria, whereas the other invokes the activation of death effector domain proteins.
In primary fibroblasts, a pp125 FAK -p130 Cas complex drives integrin-mediated survival through a Ras/Rac1/Pak1/MKK4 pathway, which regulates the activity of c-Jun N-terminal kinase and p53 (9,11). In epithelial MDCK cells, integrin survival is via a Ras/phosphatidylinositol 3-kinase/protein kinase B signaling cassette that appears to be mediated in part by pp125 FAK and in part by Rac1 (7,12,13). Both the c-Jun N-terminal kinase and p53 pathways have been shown in some cell models to regulate the proapoptotic Bcl-2 family protein, Bax, thereby controlling apoptosis through the mitochondrial pathway (14,15). Moreover, adhesion signals triggered through pp125 FAK in mammary epithelial cells cause the se-questration of Bax within the cytosol, thereby preventing its activation and translocation to mitochondria (16). Thus, in some cells, adhesion appears to regulate the cellular decision to enter apoptosis by controlling events at the mitochondrion.
By contrast, several publications have suggested that death receptors may be involved in anoikis, thereby implicating mitochondria-independent caspase activation. In certain epithelial cell lines, loss of integrin-mediated adhesion to the ECM results in rapid activation of caspase-8, an initiator caspase that is normally associated with the ligation of death receptors (17,18). This leads to direct activation of effector caspases as well as cleavage of the BH3-only protein Bid. The role of caspases in detachment-induced cytochrome c mobilization, at least in MDCK and MCF-10A cells, is supported by the observation that both a dominant-negative death domain that blocks caspase-8 recruitment to the death-inducing signaling complex and a generic caspase inhibitor, z-VAD-fmk, prevent cytochrome c release (19). Furthermore, in endothelial cells, the FasL/Fas/caspase-8 axis is sensitized by detachment of cells from the ECM, although this occurs over a substantial (12-h) time frame (20).
We previously demonstrated an essential role for Bax activation in detachment-induced apoptosis of mammary epithelial cells (16). In view of the possibility that caspase-8 activation and Bid cleavage may be involved in anoikis of some epithelial cell lines, we have now addressed the question of whether or not after removing integrin-mediated adhesion signals, initiator caspases are required for Bax translocation to mitochondria, cytochrome c release, and effector caspase activation.

EXPERIMENTAL PROCEDURES
Cell Culture-FSK-7 is a mouse mammary epithelial cell line isolated from luminal epithelial cells (21). FSK-7 cells were cultured in Dulbecco's modified Eagle's medium/F-12 medium (BioWhittaker) supplemented with 5 ng/ml epidermal growth factor, 880 nM insulin, and 2% fetal calf serum at 37°C in a humidified atmosphere of 5% CO 2 . MDCK dog kidney epithelial cells were cultured similarly; we used one strain of MDCK cells already existing within our laboratory as well as early passage MDCK-II cells purchased from the European Collection of Cell Cultures and used within three further passages. In all experiments, adherent cell monolayers were harvested after growth on tissue culture plastic or glass coverslips for 48 h, or cells were detached from the substratum with 5 mg/ml trypsin (Sigma), immediately washed in serum-containing medium and plated in whole medium onto dishes coated with polyhydroxyethylmethacrylate (poly-HEMA; Sigma) for various lengths of time. In the case of inhibitor experiments, cells were preincubated 1 h prior to trypsinization and throughout culture on poly-HEMA with inhibitors at the following concentrations: 25 g/ml cycloheximide (Calbiochem catalog no. 239764); 100 M z-VAD-fmk (Calbiochem catalog no. 627610); 50 M IETD-CHO (Calbiochem catalog no. 218773); 10 M z-IETD-fmk (Calbiochem catalog no. 218759). These concentrations of z-VAD-fmk and z-IETD-fmk are effective for inhibiting caspase-8 in other cell systems (17). Pervanadate was used at 1 mM from sodium orthovanadate and hydrogen peroxide and treated with catalase to quench unreacted hydrogen peroxide. To confirm that cycloheximide inhibited protein synthesis, mammary cells were cultured with 100 Ci of [ 35 S]methionine in the presence or absence of 25 g/ml cycloheximide for 24 h, and cell lysates were separated by SDS-PAGE, stained with Coomassie Blue, and autoradiographed. Although the lysates contained similar amounts of total cell protein, cycloheximide completely blocked the incorporation of isotope into newly synthesized proteins.
Immunofluorescence-Cells were fixed in 3.7% paraformaldehyde in PBS for 10 min, cytospun onto polysine slides (BDH), and permeabilized in 0.5% Triton X-100 in PBS for 15 min. Cells were stained as previously described with an antibody specific for the Bax 62M epitope (6,22), an antibody to amino acids 163-175 of the caspase-3 P20 subunit, which detects the active form of caspase-3 but not the precursor (R&D Systems catalog no. AF835), or an antibody to cytochrome c (Pharmingen catalog no. 65971A), followed by either Cy-2-or Cy-3-conjugated secondary antibodies (Jackson Laboratories catalog no. 33697 and 31053, respectively). Cells were counterstained with 4 g/ml Hoechst 33258 (Sigma) to examine nuclear morphology and quantify apoptosis.
Cells were viewed on a Zeiss Axiophot photomicroscope equipped with epifluorescence, and images were captured with a digital camera using Metamorph software. For comparison of each staining, all exposures and subsequent image manipulations were identical. For visualization of YFP-Bax costained with mitochondrial HSP70, fixed and stained cells were embedded in ProLong mounting medium (Molecular Probes, Inc., Eugene, OR), and then images were collected on an Olympus IX70 inverted microscope equipped with a DeltaVision imaging system, using a ϫ 100 UPLAN-APO 1.35 NA objective. Images were captured with a Photometrics CH350L cooled CCD camera and were processed by constrained iterative deconvolution using SoftWoRx version 2.5 software (Applied Precision).
Flow Cytometry-Cells cultured on poly-HEMA were harvested and fixed in 3.7% paraformaldehyde for 10 min at room temperature, washed three times in PBS, and permeabilized in 0.5% Triton X-100 for 15 min. Cells were incubated for 1 h with 62M anti-Bax antibody at 4 g/ml in PBS containing 0.5% Triton X-100 and 0.1% horse serum. After three washes in PBS, the cells were incubated for 30 min with 1:500 dilution fluorescein isothiocyanate-conjugated donkey anti-rabbit IgG (Jackson catalog no. 41624). Analysis was performed on a FACS-Vantage flow cytometer. Fluorescein isothiocyanate fluorescence was detected at 530 Ϯ 30 nm, and fluorescence was acquired using logarithmic amplifiers. 10,000 cells were analyzed per sample at a flow rate of 300 cells/s.
Fluorimetric Assay for Caspase-3 Enzyme Activation-FSK-7 cells were cultured on poly-HEMA for different times, washed once with PBS, lysed in 100 l of NET buffer (150 mM NaCl, 1 mM EDTA, 50 mM Tris-Cl, 0.1% Nonidet P-40, pH 7.6) and kept on ice for 10 min prior to centrifugation at 20,000 ϫ g for 10 min. 100 l of 50 M caspase-3 substrate Ac-DEVD-AMC (Calbiochem catalog no. 235425) was added to the lysate from 10 6 cells, and the suspension was incubated at 37°C for 15 min. The reaction was terminated by adding 200 l of 5 mM NaOAc. 100 l of each sample was transferred to a 96-well plate, and the fluorescence of cleaved AMC was measured using a spectrofluorometer (Berthold Technologies) with excitation and emission wavelengths of 380 and 460 nm, respectively. (16). Upon cell detachment from ECM, Bax translocates to mitochondria. Concomitant with this, Bax undergoes a conformational change, which can be visualized using antibodies to cryptic epitopes within Bax that only become exposed following its activation. To determine whether Bax activation, as measured by a change in conformation, is an early initiator of the apoptotic program in response to cell detachment, we examined the kinetics of events following the removal of mammary epithelial cells from ECM. After plating onto the nonadhesive substratum, poly-HEMA, cells were cytospun at various time intervals. Bax activation was examined by immunostaining with a polyclonal anti-peptide antibody that has previously been shown to recognize only the activated form of Bax on a specific epitope overlapping the BH3 domain, 62M (22). As in adherent cells, there was little or no exposure of the 62M epitope within 5 min following detachment from ECM (Fig. 1a). However, Bax became immunoreactive 10 -15 min after loss of adhesion and was evident in the majority of the cells. Exposure of the Bax 62M epitope was maintained throughout 24-h culture on poly-HEMA and during the morphological changes associated with apoptosis.

FIG. 1. Kinetics of Bax activation following loss of integrin-mediated adhesion. a, adherent FSK-7 cells and FSK-7 cells maintained on
poly-HEMA for the indicated times prior to fixation and centrifugation onto polysine slides were immunostained with an antibody that recognizes an activation state epitope within Bax, 62M, and nuclei were counterstained with Hoechst 33258. Note the absence of Bax immunoreactivity in attached cells and those detached for 5 min and that staining becomes apparent 10 -15 min after detachment and is maintained thereafter. All of the Bax immunoreactivity could be abolished by preincubation with a Bax peptide corresponding to the immunogen. b, cells detached for the indicated times were immunostained with the anti-Bax 62M antibody and analyzed by flow cytometry. Note the rapid increase of fluorescence over the first 10 -15 min after loss of integrin-mediated adhesion, which reaches a maximum after 15 min in suspension. c, adherent FSK-7 cells and FSK-7 cells maintained on poly-HEMA for the times indicated prior to hypotonic extraction and Dounce homogenization were subfractionated by centrifugation at 100,000 ϫ g into a soluble (S) fraction and a pellet (P) fraction. Equal fractions of protein were resolved by SDS-PAGE, and immunoblots were probed with anti-Bax 62M and anti-actin antibodies. d, FSK-7 cells were transfected with a plasmid expressing Bax conjugated to YFP (YFP-Bax) and either left attached to the culture dish or detached from the dish and incubated on poly-HEMA for 15 min. The cells were fixed, stained with an antibody to mitochondrial HSP70 (mtHSP70) to identify mitochondria, and visualized by immunofluorescence microscopy. Note the even staining of YFP-Bax through the cytosol, virtually all of which costains with mitochondrial HSP70 after cell detachment from extracellular matrix.
To quantify Bax activation, detached cells were fixed at the same time points after incubation on poly-HEMA, and stained with the 62M antibody for FACS analysis (Fig. 1b). There was little Bax staining immediately following detachment, but this very rapidly increased over 10 -15 min. Exposure of the 62M epitope was maximal after 15 min in suspension and remained at this level over several hours. At later time points, Bax staining became slightly reduced, although this is probably due to cells undergoing the terminal stages of apoptosis. We did find, however, that cells containing apoptotic bodies still stain for active Bax.
Two experiments were performed to confirm that Bax translocated to mitochondria. First, mammary cells were extracted under detergent-free, hypotonic conditions, and the soluble and membrane fractions were separated. Under these conditions, Bax was largely present in the cytosol of adherent cells (Fig.  1c). Following detachment from the ECM, the majority of Bax was associated with the membrane fraction. This occurred within 15 min of detachment and correlated exactly with the kinetics of exposure of the 62M epitope of Bax. Second, cells were transfected with YFP-Bax, and its location was examined in adherent cells and in those detached from the substratum. YFP-Bax was present in a diffuse, cytosolic distribution in adherent cells (Fig. 1d). Following cell detachment from the ECM for 15 min, YFP-Bax became punctate and co-stained with the mitochondrial marker mitochondrial HSP70, indicating its redistribution to mitochondria.
These data suggest that following cell detachment, Bax ac- Detachment-induced cytochrome c release, caspase activation and apoptosis occur later than Bax activation. a, equal numbers of cells were incubated on poly-HEMA for the indicated times and then split into two parts, one of which was solubilized in radioimmune precipitation buffer for the whole cell lysate and the other of which was used to prepare the cytosolic fraction before analysis by SDS-PAGE and immunoblotting. Cells harvested immediately after detachment are referred to as 0-min samples. Note that cytochrome c becomes apparent in the cytosol 1 h after detachment, increasing to a maximum at 2 h. Activated caspase-3 was analyzed by immunoblotting the whole cell lysates. It was just detectable 2 h after detachment and significantly more pronounced after 8 h. Erk1/2 was used as a loading control. b, FSK-7 cells were cytospun onto slides as in Fig. 1a, and  tivation and its translocation to mitochondria is a very rapid event and is complete within 15 min of removal of integrin signaling. This appears to occur synchronously within the cell population.
A Delay Occurs between Bax Activation and the Release of Cytochrome c from Mitochondria-The rapidity of Bax activation is in contrast to mammary epithelial cell death, which occurs over several hours following loss of adhesion. To compare Bax activation with events previously indicated to characterize commitment to apoptosis, we examined the kinetics of cytochrome c release from mitochondria and the activation of caspase-3.
Cells maintained on poly-HEMA were extracted with digitonin to leave mitochondria intact, and the soluble extracts were immunoblotted for cytochrome c. Although the Bax 62M epitope became exposed within 15 min of cell detachment from ECM, cytochrome c was not released at this time. Rather, the appearance of cytochrome c in the cytosol was first apparent 1 h following detachment and subsequently increased to a maximal level at 2 h (Fig. 2a).
Cytochrome c release from mitochondria has been described to occur rapidly and, once initiated, to be complete within 5 min subsequent to the treatment of cells with a variety of apoptotic stimuli (24). This observation is not consistent with the gradual appearance of cytochrome c in the cytosol over a period of hours in ECM-detached mammary cells. To determine whether cytochrome c is released from mitochondria in a rapid and kinetically invariant manner following loss of integrin survival signals, we examined its subcellular distribution within individual cells in situ. After 1 h, when cytochrome c is first detectable in the cytosol by immunoblotting, most cells still showed it localizing to the mitochondria by immunofluorescence (Fig. 2b). At later time points, the percentage of cells without mitochondrial cytochrome c staining increased, and many of the cells displayed diffuse cytochrome c staining instead (Fig. 2, b (arrows) and c). These results suggest that the slow release of cytochrome c observed in immunoblotting is due to its release in an asynchronous manner. Importantly, the rapid and synchronous conformational changes that occur within Bax following loss of adhesion do not lead immediately to the release of cytochrome c from mitochondria.
Cytochrome c release into the cytosol is required for the formation of the apoptosome, within which the proenzymatic form of caspases are cleaved to generate the active enzyme (25). It has previously been suggested that the release of cytochrome c rapidly commits cells to caspase activation and apoptosis (26). We therefore examined the kinetics of caspase-3 activation in response to cell detachment and compared this with the temporal release of cytochrome c. Activation of the effector caspase, caspase-3, was monitored in cells cultured on poly-HEMA by examining the appearance of its cleaved active subunit. Active caspase-3 was just detectable 2 h after detaching mammary cells from ECM and was significantly more pronounced at 8 h, as measured both by immunoblotting (Fig. 2a) and immunofluorescence (Fig. 2d). This time course was confirmed using a fluorimetric assay to measure caspase-3 enzyme activity (data not shown). We also noticed that some of the cells with cytochrome c released from mitochondria contained nuclei with normal morphology (Fig. 2b, arrowheads at the 1-and 8-h time points). Rather than becoming apoptotic just after cytochrome c release, the number of cells with normal nuclei gradually disappeared over time as the proportion of cells without mitochondrial cytochrome c but containing apoptotic nuclei increased (Fig. 2c). Thus, in anoikis, there is a lag between the release of cytochrome c and the activation of caspases, leading to the subsequent morphological signs of apoptosis.
Taken together, these results define a sequential ordering of apoptotic events following cell detachment from the ECM. In mammary cells that adhere to the ECM via integrins, Bax is localized within the cytosol and excluded from mitochondria. 15 min after cell detachment, Bax is redistributed to mitochondria and undergoes conformational changes to reveal the 62M epitope. These changes occur synchronously within the population of detached cells. Over the next 1-8 h, cytochrome c is released asynchronously into the cytosol, possibly in a stochastic manner. This is followed by the activation of caspase-3 and subsequently by cell death. Importantly, our results firmly place the activation of Bax and its translocation to mitochondria as one of the earliest detectable events leading to apoptosis after loss of integrin signaling.
Anoikis in Mammary Cells Does Not Require de Novo Protein Synthesis-Although the ECM is a survival factor for mammary cells, other types of extracellular ligand can also regulate survival through Bax translocation in some cell systems. For example, in sympathetic neurons, the removal of neurotrophins results in Bax translocation to mitochondria and apoptosis in a In each case, one set of samples was harvested immediately after detachment (0 min). a, cells were cytospun onto polysine slides and immunostained for the Bax 62M epitope (left panels) or were analyzed by flow cytometry to detect Bax activation (right panels) as in Fig. 1a. b, cell lysates of cycloheximide-treated cells were prepared for immunoblotting analysis to determine cytochrome c release and caspase-3 activation, as in Fig. 2a. c, the percentage of apoptotic cells was determined by counting the number of cells with apoptotic nuclei. In each case, the results typical of three independent experiments are shown. Note that cycloheximide has no effect on the kinetics of Bax activation, cytochrome c release (compare b with Fig. 2a), or apoptosis. mechanism that involves protein synthesis (27). We have previously demonstrated that the redistribution of Bax does not require new protein synthesis following loss of integrin-mediated adhesion in mammary epithelia (16). However, our new data indicate that there is a considerable delay between Bax translocation and the subsequent release of cytochrome c, suggestive of a protein synthesis requirement.
We examined whether this was the case by detaching cells in the presence of cycloheximide and measuring the kinetics of Bax activation, cytochrome c release from mitochondria, and apoptosis. Cycloheximide did not delay the kinetics of Bax activation as measured by exposure of the Bax 62M epitope in both immunofluorescence and FACS analysis (Fig. 3a). Similarly, cytochrome c release still occurred when protein synthesis was blocked. Immunoblots of isolated cytosolic fractions show that cytochrome c was released after 1 h, with caspase-3 activation after 2-4 h (Fig. 3b). This time course compares with that in the absence of cycloheximide (Fig. 2a). Finally, there was no difference in the rate of apoptosis over a 24-h time course in detached cells treated with or without cycloheximide, indicating that apoptosis occurs equally in the presence or absence of protein synthesis (Fig. 3c).
These data indicate that despite the delay between Bax translocation and cytochrome c release and caspase-3 activation, the anoikis program in mammary epithelial cells is independent of protein synthesis. Moreover, the mechanism for Bax redistribution and the subsequent events in the progression of apoptosis vary in response to the removal of different types of survival factor (i.e. trophic factors from sympathetic neurons and ECM from mammary epithelial cells).
Premitochondrial Caspase Activation Is Not Involved in Anoikis-It has been reported that anoikis may be initiated by death receptor-mediated activation of caspase-8 (17,18,20). Our results have indicated that Bax is redistributed rapidly to mitochondria in adhesion-regulated apoptosis (Fig. 1). Since caspase-8 can cleave the BH3-only protein Bid and it has been suggested that cleaved tBid may have a role in the activation of Bax, there is a possibility that Bax may be activated indirectly following caspase-8 activation (28 -32). We therefore examined the possibility that caspases, in particular the death receptoractivated caspase-8 (which is expressed in mammary epithelial cells) (33), might play a role in mammary cell anoikis.
We first compared the effects of a broad spectrum inhibitor of caspase activation, z-VAD-fmk, with a specific caspase-8 inhibitor, IETD-CHO, on Bax activation. Both of these inhibitors prevent the activation of caspase-8 in cells by staurosporine (Fig. 4d). Neither z-VAD-fmk nor IETD-CHO blocked exposure of the Bax 62M epitope, when examined by immunostaining or FACS analysis (Fig. 4, a and b). Furthermore, the kinetics of Bax conformation changes were not altered following caspase inhibition, with maximal activation occurring 15 min after cell detachment from ECM.
To determine whether caspase-8 is required for cytochrome c release, we examined its appearance in the cytosol in the presence or absence of z-VAD-fmk or IETD-CHO. Neither of these caspase inhibitors delayed the release of cytochrome c from mitochondria (Fig. 4c). Together, the results imply that caspase-8 is not required for detachment-induced Bax activation and cytochrome c release.
Death receptors are present on mammary cells (34,35). Fas ligand, tumor necrosis factor-␣, and TRAIL all induced apoptosis in mammary epithelial cell cultures, indicating that the death receptor axis can be activated, provided the appropriate signals are present. 2 To confirm the lack of involvement of death receptor-activated caspases in mammary epithelial anoikis, we examined the cleavage of one of its substrates, Bid, following detachment from ECM. Although treatment of mammary cell lysates with purified caspase-8 resulted in the appearance of cleaved tBid, which was blocked by z-VAD-fmk and IETD-CHO (Fig. 5, a and b), and staurosporine induced the disappearance of full-length Bid in MDCK epithelial cells (data 2 K. Green and C. H. Streuli, unpublished data.

FIG. 4. Caspase activation is not required for detachment-induced
Bax activation and cytochrome c release. FSK-7 cells were maintained on poly-HEMA for the indicated times in the presence or absence of z-VAD-fmk or IETD-CHO before harvesting. a, cells were immunostained with anti-Bax 62M antibody and Hoechst 33258. b, flow cytometry analysis was performed to detect Bax activation. c, cytosolic fractions were isolated to determine cytochrome c release by immunoblotting analysis. Note that neither inhibition of caspase-8 with IETD-CHO nor a general inhibition of caspases with z-VAD-fmk has any effect on the kinetics of Bax activation or cytochrome c release. In similar experiments, we found that z-IETD-fmk also did not inhibit Bax activation (not shown). d, in a control experiment, caspase-8 is activated after staurosporine (STS) treatment, but adding IETD-CHO or z-VAD-fmk to the culture medium blocks this, confirming the ability of these inhibitors to enter cells and inhibit endogenous caspase-8 activation.
not shown), we could not find any evidence for Bid cleavage in mammary cells plated in suspension on poly-HEMA (Fig. 5c).
Since caspases are not required for Bax activation or cytochrome c release (Fig. 4), we investigated whether the caspase inhibitors had any effect on later apoptotic events. Although z-VAD-fmk greatly delayed caspase-3 activation, IETD-CHO had no appreciable effect (Fig. 6a). In agreement with this result, only z-VAD-fmk, but not IETD-CHO, delayed cell death in suspension (Fig. 6b). As a control for these experiments, to demonstrate that Bax activation can be inhibited by pharma-cological agents following the loss of adhesion to ECM, detached cells were also treated with the protein-tyrosine phosphatase inhibitor, vanadate (Fig. 7). This efficiently blocks both Bax activation and apoptosis.
Together, our results demonstrate that premitochondrial caspases are not required to mediate the initial events that lead to cytochrome c release and apoptosis. Instead, a rapid death receptor-independent signal mediated by loss of integrin ligation results in Bax activation. DISCUSSION All mammalian cells require the receipt of extracellular signals to survive. In many cell types, interactions with adhesive ligands, in addition to soluble factors, are essential for the prevention of apoptosis. An implication of this is that cell survival in vivo is mediated not only by the milieu of locally acting factors but also by spatial cues within the cell and tissue microenvironment. Adhesive interactions with the ECM can occur through integrins, and there is now abundant evidence that these receptors provide essential survival signals in vivo (36 -38). However, the intracellular signaling mechanisms by which integrins suppress apoptosis are varied and have not yet been explored fully.
We have previously demonstrated that apoptosis resulting from loss of integrin-mediated adhesion in mammary epithelial cells occurs through translocation of Bax from the cytosol to mitochondria (16). Bax activation, as revealed by conformational changes that lead to exposure of the 62M antibody epitope, occurs after its translocation to mitochondria. 3 Here we have confirmed that Bax structure alters after mammary 3 A. J. Valentijn and A. P. Gilmore, unpublished data. epithelial cells are detached from the ECM but demonstrate that it occurs far more rapidly than previously thought, within 15 min of loss of integrin signaling; at this time, the majority of Bax has translocated from the cytosol to membrane as determined by subcellular fractionation studies. Moreover, the changes are synchronous within the cell population, since all cells show conformationally altered Bax by immunofluorescence and FACS analysis.
Although the initial integrin survival signal appears to be mediated through focal adhesion kinase, it is not yet known what the downstream signals are that lead to changes in the conformation of Bax. One possible mechanism might involve activation of the extrinsic pathway for apoptosis, which has been implicated in several cell systems where apoptosis is triggered following altered integrin signaling. This pathway is normally activated in response to ligation of death receptors by their ligands, an event that involves formation of the deathinducing signaling complex and activation of caspase-8 and thereby both the direct activation of effector caspases and their indirect activation via Bid and the mitochondrial route (1). Bid is a substrate for caspase-8, whose product, tBid, translocates to mitochondria and ultimately leads to cytochrome c release (39). In some situations, Bid is essential for the extrinsic apoptosis pathway, since an activating Fas antibody does not kill hepatocytes from Bid-null mice (40).
In some strains of MDCK cells, detachment from ECM induces caspase-dependent cytochrome c release from mitochon-dria, and specific inhibitors of caspase-8 block both cytochrome c release and apoptosis in MDCK and endothelial cells (19,20). In another cell culture model, artificial overexpression of integrins in adherent carcinoma cells leads to recruitment and activation of caspase-8, thereby triggering apoptosis (41). These mechanisms for apoptosis induction are distinct, since the former appears to involve death receptor adaptors, whereas the latter entails the direct recruitment of caspase-8 to integrin tails.
Other types of cells, such as mammary epithelial cells, are also dependent on integrins for their survival, and we have therefore asked whether anoikis in this system involves activation of caspase-8. However, our data indicate that this is not the case, since the early events of the anoikis program in mammary cells are independent of caspase activation. The evidence for this is that conformational changes within Bax occur very rapidly after the inhibition of integrin signals, with kinetics that are not altered by either a caspase-8 or a generic caspase inhibitor. Moreover, the subsequent release of cytochrome c is also not dependent on caspase activation in mammary cells. This is supported by our study, where Bid cleavage is not apparent following loss of adhesion of mammary cells.
Thus, our data demonstrate that anoikis does not always require the activation of initiator caspases but can in certain cell types be dependent on the mitochondrial apoptosis program. Interestingly, we have found that in some strains of MDCK cells, anoikis is also not dependent on caspase-8 activation (data not shown). Caspase-independent changes in integrin signaling lead to rapid changes in Bax, and we are currently identifying the pathway of enzymes involved in mammary epithelial cells. Possibilities include the phosphatidylinositol 3-kinase-protein kinase B pathway, which has a role in staurosporine-induced Bax translocation in HeLa cells, or c-Jun N-terminal kinase, since a constitutively active version mediates apoptosis via Bax in embryonic fibroblasts and Chinese hamster ovary cells (15,42). One model for Bax activation invokes the involvement of upstream BH3-only proteins, and Bim is a possible candidate because certain splice variants of this protein influence the conformation and activation of Bax (43). Whichever pathway is involved, it is likely to be direct rather than via a transcriptional intermediate. In neuronal cells, Bax activation and apoptosis following nerve growth factor withdrawal are dependent on RNA and protein synthesis (27,44). This is not the case in mammary cells, since cycloheximide has no effect on either the kinetics of Bax activation or apoptosis following loss of integrin signaling.
Once Bax has become conformationally altered after the removal of adhesion survival signals, the ultimate consequence is a change in mitochondrial structure that precedes export of cytochrome c (and other proapoptotic proteins) across the outer mitochondrial membrane and thus activation of executioner caspases. In some cell systems, activated BH3-only proteins trigger this process efficiently and rapidly. tBid remodels the structure of isolated liver mitochondria and induces cytochrome c release in permeabilized HepG2 cells within 10 s (39,45). Although cytochrome c export can happen rapidly once it has been triggered and occurs synchronously from all mitochondria within individual cells (as judged by kinetic analysis of green fluorescent protein-cytochrome c movements), not all cells release cytochrome c at the same time (24). UV-treated HeLa cells release cytochrome c stochastically over a period of 5 h, beginning 3 h after the death stimulus. Similarly, we find that cytochrome c is not released from all mammary epithelial cells simultaneously after removing integrin survival signals. Rather, it occurs stochastically, beginning ϳ2 h after detachment from ECM. FIG. 7. Bax activation during detachment-induced apoptosis is controlled by a kinase signaling pathway. Cells were incubated on poly-HEMA for the indicated times in the presence or absence of pervanadate and then harvested to assess activation of Bax, as described in the legend to Fig. 1a (a) and the percentage of apoptotic cells as described in Fig. 2 (b). Note that pervanadate blocks Bax activation and detachment-induced apoptosis.
The changes within mitochondrial structure and the precise mechanisms for release of cytochrome c are not known and currently are a subject of considerable debate (46). However, the significant lag of more than 2 h between Bax conformational change in mammary cells and observable cytochrome c release indicates that the changes are profound and complex and suggest that apoptosis commitment events may require other cellular decisions. Our culture model may provide an excellent opportunity for experimental dissection of the apoptotic decision events that occur following the initial activation of Bax. Moreover, it may help to resolve the question of why the initial, synchronous changes in Bax conformation lead only to a nonsynchronous, possibly stochastic release of cytochrome c from mitochondria.
In summary, this paper reports two significant and novel findings about the mechanism of integrin-regulated survival. The first is that loss of adhesion of mammary epithelial cells to ECM results in the activation of an intrinsic apoptosis pathway, which does not involve caspase-8. The second is that altered integrin signals induce rapid and synchronous changes in the conformation of Bax, but the apoptosis commitment event leading to cytochrome c release and caspase activation occurs asynchronously within the cell population and after a time delay.