Involvement of Raft Aggregates Enriched in Fas/CD95 Death-Inducing Signaling Complex in the Antileukemic Action of Edelfosine in Jurkat Cells

Background Recent evidence suggests that co-clustering of Fas/CD95 death receptor and lipid rafts plays a major role in death receptor-mediated apoptosis. Methodology/Principal Findings By a combination of genetic, biochemical, and ultrastructural approaches, we provide here compelling evidence for the involvement of lipid raft aggregates containing recruited Fas/CD95 death receptor, Fas-associated death domain-containing protein (FADD), and procaspase-8 in the induction of apoptosis in human T-cell leukemia Jurkat cells by the antitumor drug edelfosine, the prototype compound of a promising family of synthetic antitumor lipids named as synthetic alkyl-lysophospholipid analogues. Co-immunoprecipitation assays revealed that edelfosine induced the generation of the so-called death-inducing signaling complex (DISC), made up of Fas/CD95, FADD, and procaspase-8, in lipid rafts. Electron microscopy analyses allowed to visualize the formation of raft clusters and their co-localization with DISC components Fas/CD95, FADD, and procaspase-8 following edelfosine treatment of Jurkat cells. Silencing of Fas/CD95 by RNA interference, transfection with a FADD dominant-negative mutant that blocks Fas/CD95 signaling, and specific inhibition of caspase-8 prevented the apoptotic response triggered by edelfosine, hence demonstrating the functional role of DISC in drug-induced apoptosis. By using radioactive labeled edelfosine and a fluorescent analogue, we found that edelfosine accumulated in lipid rafts, forming edelfosine-rich membrane raft clusters in Jurkat leukemic T-cells. Disruption of these membrane raft domains abrogated drug uptake and drug-induced DISC assembly and apoptosis. Thus, edelfosine uptake into lipid rafts was critical for the onset of both co-aggregation of DISC in membrane rafts and subsequent apoptotic cell death. Conclusions/Significance This work shows the involvement of DISC clusters in lipid raft aggregates as a supramolecular and physical entity responsible for the induction of apoptosis in leukemic cells by the antitumor drug edelfosine. Our data set a novel framework and paradigm in leukemia therapy, as well as in death receptor-mediated apoptosis.


Introduction
In the last few years, a growing amount of evidence suggests that apoptosis induced by Fas/CD95 death receptor is mediated by the formation of Fas/CD95 aggregates in lipid rafts [1][2][3][4][5][6][7]. Clustering of death receptor Fas/CD95 can be achieved not only by interaction with its natural ligand FasL/CD95L, but through non-physiological agents independently of its ligand [1,4,8], providing a new framework for novel therapeutic interventions [6]. This ligand-independent activation of Fas/CD95 has a great potential therapeutic utility as it avoids the toxic side effects derived from the use of FasL/CD95L and agonistic anti-Fas/ CD95 antibodies in vivo, that lead to a fatal hepatic damage with symptoms similar to fulminant hepatitis [9,10].
Edelfosine (ET-18-OCH 3 , 1-O-octadecyl-2-O-methyl-racglycero-3-phosphocholine) is the prototypic compound of a promising family of synthetic antitumor lipids, collectively named as synthetic alkyl-lysophospholipid analogues [11], which induce selective apoptosis in tumor cells while sparing normal cells [12]. Edelfosine acts through activation of the apoptotic machinery in cancer cells, involving death receptors, caspase activation, JNK/c-Jun signaling and mitochondria [1,4,[13][14][15][16][17][18], and therefore this apoptosis-targeted drug could be appropriate for diseases where apoptosis is altered. Edelfosine was the first antitumor drug reported to promote an apoptotic response through FasL/CD95Lindependent activation of Fas/CD95 by its recruitment in lipid rafts [1], linking for the first time membrane rafts and Fas/CD95mediated apoptosis in cancer chemotherapy. The proapoptotic capacity of edelfosine was higher against cancer cells derived from blood malignancies than from solid tumors [19].
Stimulation of Fas/CD95 results in receptor aggregation and recruitment of the adapter molecule Fas-associated death domaincontaining protein (FADD), through interaction between its own death domain and the clustered receptor death domains. FADD, in turn, contains a death effector domain that binds to an analogous domain repeated in tandem within the zymogen form of procaspase-8, forming the so-called death-inducing signaling complex (DISC), made up of Fas/CD95, FADD and procaspase-8 [20], which drives cells to apoptosis.
Despite previous findings that suggest a role of Fas/CD95 and lipid rafts in cancer chemotherapy [1,[4][5][6][7]21], compelling evidence and visualization of the involvement of DISC-enriched raft clusters in cancer treatment is still lacking. Here, by using genetic, biochemical and ultrastructural approaches, we demonstrate the formation and role of DISC co-clustering in membrane rafts during edelfosine-induced apoptosis, setting up a new framework in leukemia therapy.

Biochemical and ultrastructural evidence for the formation of DISC-enriched raft clusters in Jurkat cells treated with edelfosine
We have previously reported that edelfosine induced coaggregates of rafts and Fas/CD95 in leukemia Jurkat T-cells and multiple myeloma cells [1,4,7]. The proapoptotic complex DISC was generated upon multiple myeloma cell treatment with edelfosine [7]. Here, we extend these results by finding that Fas/ CD95 as well as downstream signaling molecules FADD and procaspase-8 were translocated into membrane rafts ( Figure 1A), forming the proapoptotic complex DISC in lipid rafts ( Figure 1B), upon edelfosine treatment of Jurkat leukemic T-cells. In contrast, co-immunoprecipitation assays conducted in untreated Jurkat cells rendered no DISC formation (data not shown). Lipid rafts were identified using cholera toxin (CTx) B subunit conjugated to horseradish peroxidase that binds to the oligosaccharide portion of ganglioside GM1 [22,23], mainly found in lipid rafts [24].
To visualize and further define the co-clustering of DISC and membrane rafts following edelfosine incubation, we used electron microscopy. This technique preserved the ultrastructural integrity of the membrane and allowed immunolabeling for DISC constituents. In untreated Jurkat cells, we only detected very disperse and weak staining of rafts and Fas/CD95 at the plasma membrane without co-localization between both labels (data not shown). However, edelfosine treatment of Jurkat cells induced clusters of lipid rafts, identified through CTx B subunit binding, at the cell surface ( Figure 2A). Fas/CD95 was located embedded in plasma membrane rafts upon edelfosine treatment ( Figure 2B), thus indicating the translocation and recruitment of Fas/CD95 death receptor into membrane rafts at the cell surface. Following drug treatment and using immunogold electron microscopy, FADD was immunolocalized together with Fas/CD95 and rafts, facing the cytoplasmic side of the membrane ( Figure 2C). Furthermore, procaspase-8 was also localized in the same region in edelfosine-treated cells, forming DISC aggregates in discrete zones of the plasma membrane enriched in lipid rafts after drugtreated Jurkat cells were labeled with specific antibodies to each DISC component and with CTx B subunit to identify membrane rafts ( Figure 2D). The presence of multiple gold particles, labeling rafts (Figure 2A-2D) and DISC components ( Figure 2B-2D), suggests a high local molecule concentration of DISC constituents in raft platforms, in keeping with the formation of DISC-rich raft clusters. Our data visualize for the first time the formation of coclusters of DISC in lipid rafts in cancer chemotherapy, and further demonstrate the role of membrane rafts as scaffolds to concentrate Fas/CD95 and downstream signaling molecules in small and specialized areas of the cell surface following edelfosine treatment.
Functional role of Fas/CD95, FADD, and caspase-8 in drug-induced apoptosis in Jurkat cells To investigate the role of DISC in edelfosine-induced apoptosis, we inhibited the endogenous expression of Fas/CD95 in Jurkat cells by RNA interference using short hairpin RNA (shRNA). This Fas/CD95 silencing resulted in a significant loss of Fas/CD95 protein expression ( Figure 3A) and inhibition of edelfosine-induced apoptosis ( Figure 3B). Transfection of Jurkat cells with one of the four target sequences used for Fas/CD95 silencing (target sequence 1 in the Materials and Methods section) led to clones with about 65% Fas/CD95 silencing ( Figure 3A). This downregulation of Fas/CD95 protein expression was further assessed by Western blotting and by mean fluorescence intensity measurements in flow cytometry analysis (data not shown). This partial Fas/CD95 silencing led to about 70% inhibition in edelfosineinduced apoptosis after 48-h drug incubation ( Figure 3B). Apoptosis rates were determined by measuring the percentage of cells at the sub-G 1 region in cell cycle analysis ( Figure 3B). Analysis of the distinct cell cycle phases showed that Fas/CD95 silencing promoted a slight G 2 /M arrest following edelfosine treatment ( Figure 3B). Cell transfection with a control vector containing a scrambled sequence did not affect either Fas/CD95 expression ( Figure 3A) or drug-induced apoptosis ( Figure 3B), and no changes were detected in the distinct cell cycle phases before the triggering of apoptosis ( Figure 3B). Likewise, additional Fas/CD95 shRNA vectors that did not silence Fas/CD95 expression were without effect on the apoptotic rate rendered by edelfosine (data not shown). Cells transfected with control vector behaved similarly to intact nontransfected Jurkat cells, regarding either Fas/CD95 protein expression or drug-induced apoptosis (data not shown).
Using stable transfection in Jurkat cells with a dominantnegative form of the FADD adapter protein (FADD-DN), which lacks the death effector domain and prevents death receptor signaling [25], we found that blockade of Fas/CD95 downstream signaling abrogated edelfosine-induced apoptosis (about 80% inhibition after 48-h drug incubation) ( Figure 4A and 4B). Conversely, transfection with a control pcDNA3 empty vector did not affect edelfosine-induced apoptosis ( Figure 4A). Cells transfected with pcDNA3 control vector behaved similarly to intact nontransfected Jurkat cells regarding drug-induced apoptosis (data not shown).
The above results show that silencing or inhibition of the three major components of DISC prevents the apoptotic response induced by edelfosine in Jurkat cells as assessed by cell cycle analysis. In order to further confirm these conclusions, we analyzed apoptosis by the terminal deoxynucleotidyl transferasemediated dUTP nick-end labeling (TUNEL) technique [26] as an in situ method for detecting the 39-OH ends of DNA exposed during the internucleosomal cleavage that occurs during apoptosis ( Figure 6). Labeling the 39-OH ends, generated by DNA fragmentation, through incorporation of fluoresecin-12-dUTP allowed visualization of apoptotic cells. In addition, cells were permeabilized and stained with propidium iodide to visualized all nuclei from both non-apoptotic and apoptotic cells in red, while TUNEL-positive cells were stained in green. Silencing of Fas/ CD95 by RNA interference ( Figure 6A and 6B), constitutive expression of FADD-DN ( Figure 6A and 6C), and inhibition of caspase-8 with z-IETD-fmk ( Figure 6A and 6D) strongly inhibited edelfosine-induced apoptosis, as assessed by TUNEL analysis. The apoptotic rate, measured by this TUNEL technique, of untreated cells or Jurkat cells treated only with the caspase-8 inhibitor z-IETD-fmk, run in parallel, was less than 3% in all cases (data not shown). Similar apoptosis rates were obtained using cell cycle (hypodiploidy) and TUNEL analyses (Figures 3-6).
Taken together, we found that targeting each of the three components of DISC precludes the induction of apoptosis by the alkyl-lysophospholipid analogue edelfosine. These results strongly indicate that DISC regulates edelfosine-induced apoptosis in leukemic cells.
Accumulation of edelfosine in lipid rafts and raft requirement for drug uptake and apoptosis   the parental drug [4,19], co-localized with rafts forming edelfosinerich lipid raft clusters in leukemic T-cells ( Figure 7B). A higher magnification of the images showed a good co-localization between the green staining of the raft marker fluorescein isothiocyanatelabeled CTx B subunit and PTE-edelfosine ( Figure 7C). Figures 7B and 7C also show the clustering and capping of lipid rafts induced by edelfosine treatment. Rafts as well as PTE-edelfosine were mainly concentrated in dense patches in one or two poles of the Jurkat cell ( Figure 7B and 7C).
Preincubation with parental edelfosine prevented labeling of cells with the fluorescent analogue (data not shown). Disruption of lipid rafts with methyl-b-cyclodextrin (MCD), which interferes with protein association to rafts by cholesterol depletion [1,27], inhibited drug uptake ( Figure 7D), DISC formation (data not shown) and apoptosis ( Figure 7E). No patchy localization of the edelfosine fluorescent analogue was observed at the cell surface of MCD-treated cells (data not shown). These results square with our previous observation that lipid raft disruption abrogated edelfosine-induced formation of Fas/CD95 clusters [1,4]. Drug uptake was monitored after exhaustive washing of cells with bovine serum albumin (BSA), for which edelfosine shows a high binding capacity [28,29]. Thus, incorporated edelfosine seems to be accumulated in the inner leaflet of the plasma membrane in order to be inaccessible to extracellularly added albumin.

Discussion
Our data demonstrate that the synthetic alkyl-lysophospholipid analogue edelfosine targets lipid rafts in Jurkat leukemic T-cells, and that intact rafts are crucial for both drug uptake and druginduced apoptosis, serving as scaffolds for DISC recruitment. The present results extend and complement our earlier findings on the role of CD95 signaling and lipid rafts in the triggering of apoptosis, and lead to the scheme depicted in Figure 8, as a novel framework in cancer therapy. Here we have shown, by using different experimental approaches, the formation and role of co-clusters of DISC and membrane rafts in the triggering of apoptosis during edelfosine antileukemic action. Crucial to this process is the accumulation of edelfosine in lipid rafts, which is ensued by the reorganization of membrane raft protein and lipid composition [4,7,30] that leads to the recruitment of DISC in rafts. The aggregation of DISC in membrane rafts would favor caspase-8 activation, and hence apoptosis, as the basal activity of procaspase-8 is put in function by proximity [31]. Thus, the concentration of DISC in a rather small region of the plasma membrane facilitates caspase-8 activation that eventually leads to apoptosis. Our present findings indicate that membrane rafts serve, in addition to generating high local concentrations of Fas/CD95, as platforms for coupling adapter (FADD) and effector (caspase-8) proteins required for Fas/CD95 signaling. This is of particular importance in Fas/CD95-mediated signal transduction as the initial signaling events depend on protein-protein interactions [1,4,8]. We have previously shown that edelfosine is taken up preferentially by tumor cells [12,16], and trigger a Fas/CD95-mediated apoptotic response from within the cell independently of the ligand FasL/ CD95L [4]. Taken together, these data suggest that formation of DISC-raft clusters can be pharmacologically modulated with promising therapeutic prospects in cancer therapy.
Recent evidence shows that the novel antitumor drugs aplidin and perifosine also induce translocation of Fas/CD95 and downstream signaling molecules into lipid rafts in leukemia cells [5,7,32]. In addition, cis-platin and resveratrol have been reported to elicit co-clustering of Fas/CD95 and rafts in solid tumors [21,33]. Thus, there is an increasing evidence suggesting that Fas/ CD95 translocation into lipid rafts represents a mechanism of action for anticancer drugs. The data reported here demonstrate the formation of DISC-enriched raft aggregates as a linchpin from which apoptosis is triggered. On these grounds, we postulate that co-aggregation of lipid rafts and Fas/CD95-DISC is a new framework and paradigm in anticancer therapy. Interestingly, edelfosine and aplidin, two anticancer drugs that promote recruitment of Fas/CD95 in lipid rafts, accumulate in death receptor-rich rafts [1,4,5]. Likewise, perifosine also binds to lipid rafts [34] and promotes Fas/CD95 translocation in rafts [7]. Taken together, these results establish membrane rafts as an attractive target in cancer therapy, and edelfosine antitumor action as a paradigm of this novel raft/death receptor-mediated mechanism of action. This translocation of Fas/CD95-DISC into raft clusters, leading to Fas/CD95-mediated apoptosis independently of FasL/CD95L, provides a new molecular insight into leukemia chemotherapy and in the triggering of death receptormediated apoptosis.

Cell culture
The human acute T-cell leukemia Jurkat cell line was grown in RPMI-1640 culture medium supplemented with 10% heatinactivated fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/ml penicillin, and 100 mg/ml streptomycin at 37uC in humidified 95% air and 5% CO 2 . The Jurkat-FADD-DN cell line, provided by M.L. Schmitz (Justus-Liebig-University, Giessen, Germany), is a Jurkat-derived clone stably transfected with a pcDNA3 expression vector encoding a dominant-negative form of the FADD protein [25]. This Jurkat-FADD-DN cell line constitutively expresses FADD-DN, thus blocking Fas/CD95 downstream signaling [25], and was maintained in complete

Apoptosis assay by flow cytometry
Quantitation of apoptotic cells, following treatment with edelfosine (INKEYSA, Barcelona, Spain; APOINTECH, Salamanca, Spain), was calculated by flow cytometry as the percentage of cells in the sub-G 1 region (hypodiploidy) in cell cycle analysis as previously described [17].

TUNEL assay
Apoptosis was also analyzed in situ by the TUNEL technique using the Fluorescein Apoptosis Detection System (Promega, Madison, WI), according to the manufacturer's instructions. Cells were fixed on microscope slides, permeabilized with 0.2% Triton X-100, stained for fragmented DNA using the above kit, and then propidium iodide was added for 15 min to stain both apoptotic and non-apoptotic cells as previously described [16,35]. Thus, propidium iodide stains apoptotic and non-apoptotic cells in red, whereas fluoresecin-12-dUTP is incorporated at the 39-OH ends of fragmented DNA [26], resulting in localized green fluorescence within the nucleus of apoptotic cells. Samples were analyzed with a Zeiss LSM 510 laser scan confocal microscope.

Edelfosine localization by fluorescence microscopy
Jurkat cells were treated for 9 h with 10 mM fluorescent analog PTE-edelfosine, provided by A.U. Acuñ a and F. Amat-Guerri (Consejo Superior de Investigaciones Científicas, Madrid, Spain), which behaves similarly to the parental compound [4,19], and then incubated with the raft marker fluorescein isothiocyanate-CTx B subunit to label lipid rafts as described previously [1,4]. Colocalization of the distinct fluorochromes was analyzed, by excitation of the corresponding fluorochromes in the same section of samples, using a fluorescence microscope (Axioplan 2; Carl Zeiss MicroImaging, Inc.) and a digital camera (ORCA-ER-1394; Hamamatsu).  anti-mouse IgG coupled to 15 nm gold particles and anti-rabbit IgG linked to 10 nm gold particles, both diluted 1:50 in blocking solution. In order to reduce background, NaCl concentration was elevated from 2.5 mM to 2.5 M in 4 changes of 6 min each and then switched to ultrapure water. Grids were counterstained with 2% uranyl acetate for 20 min in dark and examined in a Zeiss EM 900. Negative controls were prepared by replacing the primary antibody with a nonrelevant antibody, showing no staining of the samples. After a 20min incubation at room temperature, mixture was added to the cells in a final volume of 0.5 ml/well. Transfected cells were cultured at 37uC in humidified 95% air and 5% CO 2 for 48 h, and then neomycin-resistant clones were selected using G418 (1 mg/ ml). Silencing of Fas/CD95 expression was confirmed by flow cytometry and Western blotting. Only one (target sequence 1) of Figure 8. Schematic model for the involvement of DISC and lipid rafts in edelfosine-induced apoptosis in Jurkat cells. This diagram portrays a currently plausible mechanism for the role of DISC recruitment in membrane rafts in drug-induced apoptosis based on the results reported in this work. Initially, Fas/CD95 death receptor is not located at the membrane raft regions of plasma membrane. Incubation of Jurkat cells with edelfosine (EDLF) leads to its accumulation in membrane rafts, inducing raft clustering and recruitment of Fas/CD95 into lipid rafts. This translocation and concentration of Fas/CD95 in rafts brings together FADD and procaspase-8, forming the DISC, through protein-protein homotypic interactions between their respective death domains (DD) and death effector domains (DED). Thus, lipid raft clusters act as scaffolds where DISC is concentrated, hence achieving caspase-8 activation and eventually apoptosis. These DISC-raft co-clusters would behave as a supramolecular and physical entity crucial for the death receptor-mediated regulation of apoptosis. doi:10.1371/journal.pone.0005044.g008 the four Fas/CD95 shRNA sequences successfully inhibited Fas/ CD95 expression (about 65% inhibition).

Immunofluorescence flow cytometry
Cell surface expression of Fas/CD95 death receptor was analyzed by flow cytometry in 4610 5 cells as described previously [16] in a Becton Dickinson FACSCalibur TM flow cytometer using an anti-Fas/CD95 SM1/1 monoclonal antibody (Bender Med-Systems, Vienna, Austria). P3X63 myeloma culture supernatant, provided by F. Sánchez-Madrid (Hospital de La Princesa, Madrid, Spain), was used as a negative control.

Lipid raft isolation
Lipid rafts were isolated from 8610 7 cells by nonionic detergent lysis and centrifugation on discontinuous sucrose gradients as described [1,4]. Twelve fractions (1-ml) were collected from the top of the gradient and 20 ml of each fraction were subjected to SDS-PAGE, immunoblotting and enhanced chemiluminescence detection. Location of GM1-containing lipid rafts was determined using CTx B subunit conjugated to horseradish peroxidase (Sigma). Pools of fractions 3-5 (enriched in lipid rafts) and fractions 10-12 (largely deficient in lipid rafts) from these sucrose gradients led to the raft and non-raft fractions, respectively, which were subsequently subjected to Western blotting. Proteins were identified using specific antibodies: anti-48-kD Fas/CD95 (C-20) rabbit polyclonal antibody (Santa Cruz Biotechnology), anti-29-kD FADD (clone-1) (BD Transduction Laboratories) monoclonal antibody, and anti-55-kDa procaspase-8 (Ab-3) monoclonal antibody (Oncogene Research Products, Cambridge, MA).

Cholesterol depletion
For cholesterol depletion, 2.5610 5 cells/ml were pretreated with 2.5 mg/ml MCD for 30 min at 37uC in serum-free medium. Cells were then washed three times with PBS and resuspended in complete culture medium before edelfosine addition.