Fas/CD95 death receptor and lipid rafts: New targets for apoptosis-directed cancer therapy

1.1. Apoptosis threshold and carcinogenesis 

The term apoptosis (from the Greek words απó -apo-, “away from, off, detached”, and πτώσιζ -ptosis-, “fall”; meaning “falling off”, as leaves from a tree), originally coined by the Australian pathologist John F.R. Kerr together with his Scottish colleagues Andrew H. Wyllie and Alastair R. Currie in 1972 (Kerr et al., 1972), refer to a type of a physiological cell death that was initially described by its morphological characteristics, including membrane blebbing, chromatin condensation, nuclear fragmentation, cell shrinkage, and breakdown of the cells into small membrane-surrounded fragments (apoptotic bodies), which were cleared by phagocytosis without prompting an inflammatory response. Apoptosis is essential for mammalian physiology and apoptosis malfunctioning is critical in the pathogenesis of many human diseases: cancer and autoimmune diseases where there is too little apoptosis; stroke damage and neurodegenerative diseases where there is too much.

Carcinogenesis is a multistage process that involves damage to the genome accumulating mutations in specific genes. This leads to alterations in either the activity or the amount of the encoded proteins, thereby perturbing normal cell functioning. These changes are perceived and appraised by cellular “sensors”, involved in restoring normal function. When this is not possible, sensors trigger signaling pathways that lead eventually to apoptosis. This implies the existence of a molecular threshold for the engagement of cell death, named “apoptosis threshold” (Gajate and Mollinedo, 2002), in response to damage that is set differently in different cell types (Fig. 1). Apoptosis thresholds depend on:

Tumor cells may either inhibit the molecular processes that lead to their own death by apoptosis and/or increase markedly the apoptosis threshold required to sensitize them to apoptotic signals (Gajate and Mollinedo, 2002). In this regard, the tumor suppressor gene p53, involved in cell-cycle control, apoptosis, and maintenance of genetic stability, is mutated or inactivated in over 50% of all cancers. There are significant intrinsic differences among different cell types in their capacity to invoke an apoptotic response after damage. Most (approximately 86%) of the human cancers derive from epithelial cells. One explanation might be that epithelial tissue has a high apoptosis threshold and can tolerate a high amount of damage before mounting an apoptotic response; hence, mutations might accumulate leading ultimately to cancer. The high apoptotic threshold of epithelium could be due to the fact that it is exposed to environmental insults and some damage risk might be an acceptable trade off for avoiding a high cell death rate. In contrast, the bone marrow, which is protected from external insults, would show a lower threshold for engaging apoptosis and this might explain why hematopoietic tumors are less frequent (about 8%). This low apoptotic threshold also explains the deleterious effects and the ease of chemotherapy- and radiation-induced cell death in the bone marrow.

Impaired apoptosis signaling is common in cancer cells and plays an important role in tumor initiation, progression and metastasis, as cells with genomic damage or deregulated cell cycle are normally eliminated by apoptosis. Resistance of cancer cells to apoptosis is especially deleterious, because it results in a higher survival capacity under adverse conditions, enhancing the malignant potential of the tumor, favoring accumulation of mutations, metastasis and rendering tumor cells resistant to therapy as well as to host defense mechanisms. The ability to metastasize makes cancers hard to eradicate and leads to the high death toll of cancer patients, though less than 0.1% of the cancer cells released into circulation by tumors survive and succeed in founding metastatic colonies.

The gene for caspase-8 (previously known as FADD-like interleukin-1β converting enzyme -ICE- (FLICE), or MORT1-associated CED-3 homolog (MACH)), a proapoptotic cysteine protease, is frequently silenced through DNA methylation and gene deletion in neuroblastoma (Teitz et al., 2000), the most common childhood solid tumor rising from the peripheral nervous system. Caspase-8-null neuroblastoma cells are resistant to death receptor- and doxorubicin-mediated apoptosis, deficits that are corrected by programmed expression of the enzyme (Teitz et al., 2000). Furthermore, loss of caspase-8 has been recently shown to potentiate neuroblastoma metastasis (Stupack et al., 2006). In addition, caspase-8 expression is lost in tumors other than neuroblastoma, including small-cell lung carcinoma, rhabdomyosarcoma, medulloblastoma, and retinoblastoma (Pingoud-Meier et al., 2003, Shivapurkar et al., 2002). These findings highlight the importance of apoptosis in tumor development, metastasis and treatment.

1.2. Death receptors and cancer chemotherapy 

Mammals have evolved a receptor/ligand mechanism that enables the organism actively to direct individual cells to self-destruct through the presence of cell surface death receptors, which transmit apoptosis signals initiated by specific death ligands. Apoptosis-targeted therapy through activation of death receptors can engage an apoptotic response that bypasses the action of sensors, such as p53, and therefore their frequent mutant state in cancer should be irrelevant to this therapeutic approach (Fig. 1). Death receptors are members of the tumor necrosis factor (TNF) receptor gene superfamily, which consists of more than 20 proteins with a broad range of biological function, including the regulation of cell death and survival, differentiation or immune regulation (Debatin and Krammer, 2004). Death receptors share regions of high homology including cysteine-rich extracellular domains and a cytoplasmic domain of about 80 amino acids called “death domain”, which plays a crucial role in transmitting the death signal from the cell's surface to intracellular signaling pathways. The best-characterized death receptors in their potential to induce apoptosis are Fas, TNF receptor 1 (TNFR1), TNF-related apoptosis-inducing ligand (TRAIL) receptor 1 (TRAIL-R1) (death receptor 4, DR4) and TRAIL receptor 2 (TRAIL-R2) (death receptor 5, DR5). Differential expression of TRAIL-R1/DR4 and TRAIL-R2/DR5 has been described for various tumor types, usually with TRAIL-R2/DR5 being the most prevalent. Recent data suggest that TRAIL-R2/DR5 may contribute more than TRAIL-R1/DR4 to TRAIL-induced apoptosis in cancer cells that express both death receptors (Kelley et al., 2005). Hence, the corresponding death ligands TNF, Fas ligand (FasL) and TRAIL are interesting candidates for antitumor therapy (Shankar and Srivastava, 2004, van Geelen et al., 2004). However, systemic TNF administration has shown low antitumor activity and higher doses induce a severe inflammatory response syndrome that resembles septic shock. Nevertheless, TNFα has been used in human cancer patients through local cytokine administration to avoid or reduce the induction of undesired systemic symptoms (Hohenberger and Tunn, 2003). Systemic FasL administration in humans has not been attempted because of severe liver toxicity in mice. Administration of FasL or agonistic antibody to Fas in tumor-bearing mice is lethal because of apoptosis induction in hepatocytes that abundantly express Fas (Ogasawara et al., 1993, Tanaka et al., 1997). Unlike TNF and FasL, TRAIL seems to be unique, since it has been reported to be nontoxic to normal cells at concentrations where it kills a broad range of tumor cells (Yagita et al., 2004); hence, this molecule is a promising anticancer therapeutic agent. Agonistic TRAIL-R1/DR4 and TRAIL-R2/DR5 antibodies are also being investigated as potential therapeutic options. Native TRAIL is expressed as a homotrimeric type II transmembrane protein that can be proteolytically cleaved into soluble homotrimeric TRAIL cells (Ashkenazi et al., 1999). Both membrane and soluble TRAIL can interact with TRAIL-R1/DR4 and TRAIL-R2/DR5, which initiate apoptosis via their intracellular death domains, and with two antagonist receptors TRAIL-R3 (decoy receptor 1, DcR1) and TRAIL-R4 (decoy receptor 2, DcR2), which are unable to engage apoptosis due to the absence or truncation of the cytoplasmic death domain (Ashkenazi and Dixit, 1998). The different TRAIL receptors are widely expressed in a variety of normal tissues and malignant cell types. Initially, TRAIL-R3/DcR1 and TRAIL-R4/DcR2 were thought to be predominantly expressed in normal cells, thus sparing normal cells from apoptosis. However, no correlation between TRAIL sensitivity and expression of TRAIL-R3/DcR1 or TRAIL-R4/DcR2 has been found (Lincz et al., 2001). Consequently, the mechanism for the tumor-selective activity of TRAIL remains elusive. Nevertheless, several reports have described apoptotic activity of TRAIL toward various normal human cells, including primary human hepatocytes (Jo et al., 2000), keratinocytes (Leverkus et al., 2000), prostate epithelial cells (Nesterov et al., 2002), and brain tissue (Nitsch et al., 2000).

Similar to trastuzumab (Herceptin®), the humanized antibody to the erbB-2/HER2 receptor tyrosine kinase overexpressed in 20–25% of invasive breast cancers, previous assessment of the status of molecular target in the tumor to predict response (Nahta and Esteva, 2006), may also be appropriate for death-receptor-directed therapy. Moreover, because ligands of the TNF family and their cognate receptors play a key role in liver pathogenesis (Faubion and Gores, 1999), hepatotoxicity is a major challenge for the clinical application of death receptor-targeted therapy. If liver toxicity could be circumvented, Fas would be a worthy anticancer target due to its potent proapoptotic activity and widespread expression in tumor cells. It is expected that small molecules targeting these death receptors will be designed to lower toxicity and increase antitumor activity.

1.3. Fas death receptor induces apoptosis through its oligomerization and a cascade of protein–protein interactions 

Yonehara et al. (1989) reported an IgM monoclonal antibody that could kill several human cell lines, and termed Fas (FS7-associated cell surface antigen) the cell surface protein recognized by the antibody. In July of the same year, Peter H. Krammer and his associates reported a mouse monoclonal antibody, named anti-APO-1 antibody, which promoted apoptosis in human leukemic cells and activated lymphocytes (Trauth et al., 1989). Two years later, Nagata et al. succeeded in cloning the membrane protein recognized by the killing antibody, the Fas antigen (Itoh et al., 1991) that turned out to be identical to the APO-1 protein identified later in Krammer's group (Oehm et al., 1992). Then, Nagata's group cloned the corresponding physiological ligand of the Fas death receptor, FasL (Suda et al., 1993) that belongs to the TNF family and can be found as a 40-kDa membrane-bound or a 26-kDa soluble protein (Nagata, 1997). Subsequent findings identified the Fas/FasL system as the major regulator of apoptosis at the cell membrane in mammalian cells through a receptor/ligand interaction.

Mature Fas (Fig. 2) is a 48-kDa type I transmembrane receptor of 319 amino acids with a single transmembrane domain of 17 amino acids (from Leu-158 to Val-174), an N-terminal cysteine-rich extracellular domain (18 cysteine residues in 157 amino acids) and a C-terminal cytoplasmic domain of 145 amino acids that is relatively abundant in charged amino acids (28 basic and 20 acidic amino acids). The cytoplasmic portion of Fas contains a domain of about 85 amino acids termed “death domain”, which plays a crucial role in transmitting the death signal from the cell's surface to intracellular pathways (Nagata, 1997). Unlike the intracellular regions of other transmembrane receptors involved in signal transduction, the death domain does not possess enzymatic activity, but mediates signaling through protein–protein interactions. The death domain has the propensity to self-associate and form large aggregates in solution (Huang et al., 1996) (Gajate and Mollinedo, unpublished results). The tertiary structure of the Fas death domain, revealed by NMR spectroscopy (Huang et al., 1996), consists of six antiparallel, amphipathic α helices (Huang et al., 1996). Helices α1 and α2 are centrally located and flanked on each side by α3/α4 and α5/α6. This leads to an unusual topology in which the loops connecting α1/α2 and α4/α5 cross over each other (Huang et al., 1996). The presence of a high number of charged amino acids in the surface of the death domain is probably responsible for mediating the interactions between death domains. Stimulation of Fas by FasL results in receptor aggregation (Chan et al., 2000), previously assembled in trimers (Papoff et al., 1999, Siegel et al., 2000), and recruitment of the adaptor molecule Fas-associated death domain-containing protein (FADD) (Chinnaiyan et al., 1995) through interaction between its own death domain and the clustered receptor death domains. FADD also contains a “death effector domain” (DED) that binds to an analogous domain repeated in tandem within the zymogen form of caspase-8 (Boldin et al., 1996). Upon recruitment by FADD, procaspase-8 oligomerization drives its activation through self-cleavage, activating downstream effector caspases and leading to apoptosis (Ashkenazi and Dixit, 1998). Thus, activation of Fas results in receptor aggregation and formation of the so-called “death-inducing signaling complex” (DISC) (Kischkel et al., 1995), containing trimerized Fas, FADD and procaspase-8 (Fig. 3).

View Large Image Download to PowerPoint

Fig. 2. Schematic diagram of the human Fas death receptor. Mature human Fas protein consists of 319 amino acids (aas) with an N-terminal extracellular domain of 157 aas, a short transmembrane region (17 aas) and a C-terminal cytoplasmic domain of 145 aas. Relevant domains for Fas oligomerization and apoptotic activity are shown. An N-terminal extracellular oligomerization domain (NOD) of 49 aas (Arg-1 to Pro-49) responsible for the FasL-independent oligomerization of the receptor. Three cysteine-rich domains (CRD1-Gln31 to Val-67, CRD2-Pro-68 to Cys-111-, and CRD3-Arg-112 to Lys150-) containing four, six and eight Cys residues in each domain, respectively. A cytoplasmic death domain (DD) of 85 aas (Ser-214 to Ile-298) is crucial for apoptotic signaling. The last 15 amino acids (Asp-305 to Val-319) of the Fas amino acid sequence represent a C-terminal inhibitory domain (CID). Domains and membrane are not to scale.

View Large Image Download to PowerPoint

Fig. 3. Schematic representation of Fas activation through its aggregation in membrane rafts. Fas molecules are brought together and concentrated in membrane rafts facilitating the formation of DISCs, following protein-protein interactions between Fas-FADD through their respective death domains (DD), and FADD-procaspase-8 through their respective death effector domains (DED). DISC formation leads to activation of unprocessed procaspase-8 by driving its dimerization and autoproteolysis, resulting in the release of mature, active caspase-8 (composed of a p20/p10 heteromer) into the cytoplasm. The asteriks represent the active-site cysteine residues of caspase-8, the dash lines indicate proteolytic processing in trans, and the arrowheads point to the sites of proteolytic cleavage. Actin cytoskeleton through ezrin is involved in the clustering of Fas in lipid rafts.

Mice carrying the lymphoproliferation (lpr) point mutation which converts Ile-225 to Asn-225 in the cytoplasmic region of the mouse Fas antigen, are characterized by a deficient Fas antigen that leads to a lymphoproliferation syndrome showing lymphadenopathy and a systemic lupus erythematosus-like autoimmune disease (Watanabe-Fukunaga et al., 1992). The corresponding mutation in human Fas (V238N) leads to inhibition of apoptosis, together with a dramatic inhibition in Fas death domain self-association and binding to FADD (Huang et al., 1996), suggesting that this point mutation alters the protein structure of the death domain. These data suggest that the intracellular portion of the Fas molecule is critical for death receptor oligomerization required for apoptotic activity. Current evidence indicates that the molecular ordering of the initial events in physiological Fas-mediated signaling after binding of Fas to its cognate ligand include four successive steps (Algeciras-Schimnich et al., 2002):

2.1. Fas oligomerization without interaction with its ligand 

Several of the early paradigms in Fas activation have been recently challenged by recent evidence. An early view of the molecular events leading to Fas activation considered that, upon binding to homotrimers of FasL, the Fas receptor homotrimerized through the intracellular death domains. However, Fas is now assumed to constitutively trimerize prior to FasL binding (Papoff et al., 1999, Siegel et al., 2000). Pre-associated Fas complexes were found in living cells by means of fluorescence resonance energy transfer between variants of green fluorescent protein (Siegel et al., 2000). A FasL- and death domain-independent oligomerization domain in the extracellular region of the Fas receptor, mapping to the N-terminal 49 amino acids, mediates homo- and hetero-oligomerization of the death receptor (Papoff et al., 1999) (Fig. 2). In addition, the notion that Fas requires interaction with its ligand to trigger an apoptotic response has also been challenged. Apoptosis can be triggered in the absence of FasL by overexpression of the Fas cytoplasmic domain or a Fas receptor lacking the N-terminal 42 amino acids (Papoff et al., 1999), suggesting that the extracellular oligomerization domain of Fas is not required to initiate signaling, and that self-association of the death domain is necessary and sufficient to induce cell death and occurs in the absence of an intact extracellular oligomerization domain. Thus, two major oligomerization domains are present in the Fas receptor (Fig. 2), one mapping to the extracellular region of the receptor, likely related to the regulation of the non-signaling state, and another one, involved in apoptotic signaling, mapping to the intracytoplasmic region, the death domain (Papoff et al., 1999). The intracellular death domains of death receptors show a high tendency to self-associate, and when overexpressed by gene transfer in eukaryotic cells trigger apoptotic signaling (Boldin et al., 1995). These findings indicate that the Fas receptor plays an active role in its own clustering and suggest the existence of cellular mechanisms that restrict its self-association, thus preventing constitutive signaling. Taken together, these data show that Fas oligomerization can be achieved in the absence of FasL.

2.2. FasL is dispensable in Fas-mediated apoptosis 

Doxorubicin-induced apoptosis in human T-leukemic cells has been proposed to be mediated by FasL expression with subsequent autocrine and/or paracrine induction of cell death through binding of FasL to the membrane Fas receptor (Friesen et al., 1996). This led Friesen et al. to postulate that Fas/FasL interactions could account for chemotherapy-associated apoptosis (Friesen et al., 1996). Additional anticancer drugs, such as methotrexate or bleomycin, were also reported to promote induction of FasL expression and up-regulation of membrane FasL, leading to autocrine or paracrine Fas/FasL-dependent apoptosis (Friesen et al., 1996, Fulda et al., 1997, Muller et al., 1997). Cell lines resistant to Fas were reported to be insensitive to anticancer drug-induced apoptosis, and drug-induced cell death was prevented by Fas-neutralizing antibodies (Friesen et al., 1996). However, the involvement of Fas/FasL interactions in chemotherapy-induced apoptosis rapidly became a controversial issue in the late 1990s as several research groups were unable to reproduce the original findings, showing that blockade of Fas/FasL interactions did not prevent apoptosis induced by doxorubicin and other cytotoxic drugs, and that anticancer drug-induced apoptosis did not require de novo synthesis of FasL (Gajate et al., 2000a, Gamen et al., 1997, Tolomeo et al., 1998, Villunger et al., 1997). However, although many cytotoxic drugs have been shown to act independently of the Fas system, we and others detected FasL-independent activation of Fas in the mechanism of action of a number of antitumor drugs, including edelfosine, cisplatin, etoposide, and vinblastine (Gajate et al., 2000a, Micheau et al., 1999). Cells deficient in Fas were resistant to the proapoptotic action of edelfosine, but became sensitive to the antitumor ether lipid when transfected with Fas (Gajate et al., 2000a). The presence or absence of Fas expression on the cell surface of cancer target cells correlated with their sensitivity to the proapoptotic activity of edelfosine (Gajate et al., 2000a). Down-regulation of FADD by transient transfection with an antisense FADD construct inhibited tumor cell sensitivity to cisplatin, etoposide or vinblastine, whereas overexpression of FADD sensitized tumor cells to drug-induced cell death (Micheau et al., 1999). Transfection of cells with FADD dominant negative decreased apoptosis induced by cisplatin or antitumor ether lipids (Matzke et al., 2001, Micheau et al., 1999), and transient transfection with either MC159 or E8, two viral proteins that inhibit apoptosis at the level of FADD and caspase-8, respectively (Bertin et al., 1997), protected cells from cisplatin-induced cytotoxicity (Micheau et al., 1999). Nevertheless, incubation with blocking anti-Fas antibodies (such as ZB4 and SM1/23 antibodies), or with the soluble Fas–IgG fusion protein (to prevent the interaction of Fas with FasL) failed to inhibit drug-induced apoptosis and drug-mediated induction of FasL expression was not always detected in distinct tumor cells (Gajate et al., 2000a, Micheau et al., 1999). These data suggest that, at least some, anticancer drugs induce cell death through a Fas/FADD pathway in a FasL-independent manner.

2.3. Intracellular triggering of Fas activation in a FasL-independent way, a new selective approach to kill tumor cells 

Combining transfection and microinjection experiments, we successfully demonstrated the intracellular activation of Fas independently of FasL through the elucidation of the unique mechanism of action of the antitumor ether lipid edelfosine (Gajate et al., 2004, Gajate and Mollinedo, 2002, Mollinedo et al., 1997, Mollinedo et al., 2004) (Fig. 4). This antitumor ether lipid mediates selective apoptosis in cancer cells through Fas activation, independently of FasL, once the drug is inside the cell (Gajate et al., 2004). Tumor cells take up edelfosine and are sensitive to the drug, whereas normal cells are spared because they are unable to incorporate significant amounts of the drug (Gajate et al., 2000a, Mollinedo et al., 1997). Fas-expressing cells that do not take up edelfosine from the culture medium, like normal cells (Gajate et al., 2000a, Mollinedo et al., 1997), are unaffected by the exogenous addition of the ether lipid, but they undergo apoptosis following microinjection of edelfosine (Gajate et al., 2000a). Fas-deficient cells were spared from the drug action even following microinjection, but became sensitive after bestowing Fas expression by gene transfer (Gajate et al., 2000a, Gajate et al., 2004). Deletion of the Fas 57 C-terminal amino acids that included part of the Fas cytoplasmic death domain prevented apoptosis (Gajate et al., 2004). In addition, edelfosine-sensitive Fas-expressing Jurkat cells turned drug-resistant when these cells became Fas-deficient (Gajate et al., 2004). Taking together, these findings indicate that Fas is required for edelfosine-mediated apoptosis and that edelfosine must be inside the cell to trigger cell death through intracellular activation of Fas (Gajate et al., 2004, Mollinedo et al., 2004). Because edelfosine is selectively taken up by the cancer cell, and once inside the cell, edelfosine triggers Fas activation leading to apoptosis, this mechanism represents the first selective activation of Fas in tumor cells (Gajate et al., 2004) (Fig. 4). Whether edelfosine triggers Fas activation by direct interaction with the cytoplasmic part of the death receptor or through an indirect process remains to be elucidated. However, computational docking studies have allowed us to propose a molecular model for the putative interaction of edelfosine with the intracellular Fas death domain (Mollinedo et al., 2004). Thus, edelfosine takes advantage of an apparently selective drug uptake in tumor cells (Gajate et al., 2000a, Gajate et al., 2004, Mollinedo et al., 1997) and of a general Fas-mediated apoptotic signaling. This intracellular activation of Fas is an attractive way to target cancer cells from within the cell, thus avoiding the deleterious systemic activation of Fas death receptor in normal cells, especially in liver (Gajate et al., 2004, Mollinedo et al., 2004). It also sets a conceptual framework for designing novel and more selective proapoptotic antitumor drugs.

View Large Image Download to PowerPoint

Fig. 4. Selective antitumor action of edelfosine. Normal cells do not take up the drug, and therefore they are spared, whereas cancer cells incorporate edelfosine and undergo apoptosis mediated by intracellular triggering of Fas.

2.4. FasL-independent translocation and clustering of Fas into membrane rafts, a novel approach in cancer chemotherapy 

As discussed above, we found that the antitumor drug edelfosine induced the translocation and concentration of Fas into lipid rafts (Gajate et al., 2004, Gajate et al., 2000a, Gajate and Mollinedo, 2001), implicating for the first time membrane rafts in Fas-mediated apoptosis and cancer chemotherapy. Resveratrol, a polyphenol found mainly in grape skin with antitumor chemopreventive properties (Pervaiz, 2004), as well as the antitumor drugs cisplatin and aplidin have also been found to redistribute Fas in rafts independently of FasL (Delmas et al., 2003, Gajate and Mollinedo, 2005, Lacour et al., 2004). Hence Fas/FasL interaction, although it can enhance cell death, is not essential for drug-induced apoptosis; a growing number of agents and experimental conditions can induce Fas activation without the participation of FasL (Table 1). These data suggest a common mechanism whereby divergent stimuli can activate membrane-associated events that target the Fas apoptotic pathway in a manner that precludes its natural ligand FasL. This FasL-independent activation of Fas is blocked by dominant negative-FADD, antisense FADD, caspase-8 inhibitors, or by MC159 and E8, involving FADD and caspase-8 in this process (Beltinger et al., 1999, Bush et al., 2001, Chen and Lai, 2001, Delmas et al., 2003, Luo et al., 2003, Micheau et al., 1999). These findings suggest that Fas clustering promotes FADD recruitment and DISC formation, independently of FasL (Fig. 3).

Table 1. FasL-independent activation of Fas by different agents

	Inducer	Experimental evidence (Refs.)
		FasL-independent activation of Fas	Fas clustering/capping	Co-capping of Fas and rafts
	Aplidin	Gajate et al. (2003); Gajate and Mollinedo (2005)	Gajate and Mollinedo (2005)	Gajate and Mollinedo (2005)
	Camptothecin	Shao et al. (2001)
	Cisplatin	Huang et al. (2003); Micheau et al. (1999)	Huang et al. (2003); Lacour et al. (2004); Micheau et al. (1999)	Lacour et al. (2004)
	Curcumin	Bush et al. (2001)	Bush et al. (2001)
	Deoxycholic acid	Gupta et al. (2004); Qiao et al. (2001)
	Edelfosine (ET-18-OCH3)	Gajate et al., 2004, Gajate et al., 2000a; Gajate and Mollinedo (2001)	Gajate et al., 2004, Gajate et al., 2000a; Gajate and Mollinedo (2001)	Gajate et al. (2004); Gajate and Mollinedo (2001)
	γ-irradiation	Huang et al. (2003)	Huang et al. (2003)
	Glutamine deprivation-mediated cell shrinkage	Fumarola et al. (2001)	Fumarola et al. (2001)
	HCV core protein	Moorman et al. (2003)	Moorman et al. (2003)
	JNK activation (via MKK7)	Chen and Lai (2001)	Chen and Lai (2001)
	Mithramycin A	Leroy et al. (2006)	Leroy et al. (2006)
	Resveratrol	Delmas et al. (2003)	Delmas et al. (2003)	Delmas et al. (2003)
	Reactive oxygen species (ROS)	Huang et al. (2003)	Huang et al. (2003)
	TGF-β1	Kim et al. (2003)
	TK/GCV	Beltinger et al. (1999)	Beltinger et al. (1999)
	Ultraviolet light	Aragane et al. (1998); Rehemtulla et al. (1997); Zhuang and Kochevar (2003)	Aragane et al. (1998); Rehemtulla et al. (1997); Zhuang and Kochevar (2003)
	Vanadate	Luo et al. (2003)	Luo et al. (2003)
	Vinblastine	Micheau et al. (1999)	Micheau et al. (1999)
	Etoposide	Micheau et al. (1999)	Micheau et al. (1999)

Fas clustering was visualized by immunofluorescence confocal microscopy or assessed by immunoprecipitating Fas using limiting antibody concentrations (Rehemtulla et al., 1997). Co-capping of Fas and rafts was visualized by immunofluorescence confocal microscopy and further assessed by identifying Fas in isolated lipid rafts (Gajate and Mollinedo, 2001). HCV, hepatitis C virus. TK/GCV, herpes simplex thymidine kinase/ganciclovir. TGF-β1, transforming growth factor-β1.

2.5. Aggregation of apoptotic molecules leads to apoptosis 

Formation of lipid raft platforms, where a large amount of signaling molecules are brought together, increases DISC formation and therefore potentiates Fas signaling. Because activation of caspase-8 is induced by proximity (Muzio et al., 1998), its concentration in lipid rafts will favor caspase-8 activation, triggering downstream apoptotic signaling. It can be envisaged that the intrinsic enzymatic activity of caspase-8, upon interaction with additional procaspase-8 molecules mediated by the adapter FADD molecules, attains a sufficient concentration to activate the apoptosis pathway. Using chimeras of caspase-8 with either CD8 or Tac, Martin et al. (1998) found that oligomerization at the cell membrane powerfully induces caspase-8 autoactivation and apoptosis. Based on these findings, it can be envisaged that these oligomerization processes would be facilitated enormously in the large Fas aggregates formed during stimulation, leading to activation of caspase-8 and generation of downstream apoptotic signals.

Thus, Fas clustering could be an efficient way to elicit apoptosis through recruitment of the DED-containing proteins FADD and procaspase-8 into Fas clusters (Fig. 3). In addition, it has been demonstrated that FADD and caspase-8 coalesce into what appear to be perinuclear “death effector filaments” (DEFs), inducing receptor-independent apoptotic signals and apoptosis (Siegel et al., 1998). Overexpression of either FADD or caspase-8 induces apoptosis through the formation of unique filament structures that contain the death effector domains of these proteins (Siegel et al., 1998), accordingly being named death effector filaments. Thus, formation of death effector filaments leads to intracellular assemblies of apoptosis-signaling complexes that can initiate or amplify apoptotic stimuli independently of receptors at the plasma membrane. Cycloheximide has been shown to induce cell death in human leukemic Jurkat and CEM C7 T-cell lines in a FADD-dependent and receptor-independent manner through DEF formation (Tang et al., 1999). Also, a number of antitumor drugs, including microtubule-disrupting agents, may induce apoptosis via caspase-8 activation independently of the Fas/FasL system (Goncalves et al., 2000).

As stated above, the initial events in Fas signaling are largely dependent on the local concentration of the three major components of the DISC – Fas, FADD and caspase-8 – oligomerization of each one being sufficient to mount an apoptotic response. Thus, formation of Fas caps leads to the recruitment of these molecules in a limited space, increasing the probability of interactions among them, and thereby promoting a strong apoptotic response.

2.6. Recruitment of death receptors and downstream signaling apoptotic molecules in lipid rafts 

As suggested from the translocation of Fas into membrane rafts following antitumor chemotherapy, the concentration of death receptors in a rather small area of the cell would potentiate death receptor ligands to achieve cell death. In this regard, resveratrol (Delmas et al., 2004) and aplidin (Gajate and Mollinedo, 2005) have been reported to redistribute Fas, TNFR1, and TRAIL receptor into lipid rafts, and this redistribution sensitizes the cells to death receptor stimulation by their cognate ligands or agonistic cytotoxic antibodies (Delmas et al., 2004) (Gajate and Mollinedo, unpublished results).

Our recent studies have shown that not only Fas together with FADD and procaspase-8 are recruited into lipid rafts, forming the DISC, but additional downstream apoptotic signaling molecules, including procaspase-10, c-Jun N-terminal kinase (JNK), and BH3-interacting domain death agonist (Bid) are also translocated into membrane rafts following cancer chemotherapy (Gajate et al., 2004, Gajate and Mollinedo, 2005) (Fig. 5). Persistent JNK activation is associated with apoptosis (Chen et al., 1996), and Bid has been shown to act as a bridge between Fas signaling and the mitochondrial-dependent pathway of apoptosis (Li et al., 1998). The recruitment in membrane rafts of JNK and Bid following treatment of human leukemic cells with edelfosine (Gajate et al., 2004) and aplidin (Gajate and Mollinedo, 2005) may explain the dependence of edelfosine- and aplidin-mediated apoptosis on both JNK and mitochondrial signaling (Gajate et al., 1998, Gajate et al., 2003). This redistribution of death receptors and downstream signaling molecules into lipid rafts does not require protein synthesis, and therefore it is achieved from the pre-existing protein pool (Gajate et al., 2004). Because Fas clustering can occur without the participation of FasL, and ceramide enhances Fas clustering, it could be suggested that either the different treatments exerting Fas clustering promote sphingomyelinase-dependent ceramide generation or cause physical changes in the plasma membrane similar to those elicited by ceramide, inducing coalescence of rafts leading to large raft platforms and subsequent capping. However, the clustering of Fas and downstream signaling molecules in lipid rafts, leading to Fas-mediated apoptosis, upon treatment of human leukemic cells with edelfosine was independent of sphingomyelinase activation (Gajate et al., 2004). Edelfosine did not activate either neutral or acidic sphingomyelinase, and did not induce any significant increase in endogenous ceramide levels, suggesting that sphingomyelinase activation was not required in edelfosine-induced apoptosis (Gajate et al., 2004). A number of antagonists of ceramide-induced apoptosis, including cAMP, the free radical scavenger C60, the metal chelator pyrrolidinedithiocarbamate (PDTC), and the SAPK (stress-activated protein kinase)/ERK (extracellular signal-regulated kinase) kinase (SEK) dominant negative mutant, could not suppress Fas-mediated cell death, suggesting that the apoptotic signal of Fas is not mediated by ceramide (Hsu et al., 1998).

View Large Image Download to PowerPoint

Fig. 5. Clustering of Fas death receptor and downstream signaling molecules in lipid rafts in cancer chemotherapy. Edelfosine and Aplidin induce apoptosis in cancer cells through aggregation of Fas, downstream signaling molecules, including FADD, procaspases 8 and 10, JNK and Bid, and actin-linking proteins (ezrin, moesin) in clusters of lipid rafts. Aplidin also induces recruitment of FasL into lipid rafts, facilitating Fas/FasL killing between neighboring cells.

We hypothesize that accumulation of Fas into aggregates of stabilized membrane lipid domains from a highly dispersed distribution may represent a general mode of regulating Fas activation. Thus, membrane rafts could serve, in addition to generating high local concentration of Fas, as platforms for coupling adaptor and effector proteins required for Fas signaling (Fig. 5). This is of particular importance in Fas-mediated signal transduction as the initial signaling events depend on protein–protein interactions. Furthermore, this could facilitate and amplify signaling processes by local assembly of various cross-interacting signaling molecules.

4.1. How are Fas clusters in lipid rafts formed by intracellular signals? 

As shown in Table 1, FasL-independent activation of Fas is mediated by Fas clustering, and recent evidence shows co-capping of Fas in membrane rafts (Delmas et al., 2003, Gajate et al., 2004, Gajate and Mollinedo, 2001, Gajate and Mollinedo, 2005, Lacour et al., 2004). How are these Fas clusters generated? Because FasL is not strictly required, signals from inside the cell must be able to regulate this process. The formation of Fas clusters and the recruitment of Fas into membrane rafts in a FasL-independent manner could involve intracellular processes, changes in the physicochemical properties of cell membranes or both.

Vanadate-elicited Fas aggregation and Fas–FADD association as well as caspase-8 activation, were dependent on JNK activation (Luo et al., 2003). These results highlight a major role for JNK in the signaling mechanisms leading to FasL-independent Fas activation. In fact, selective JNK activation by overexpressing the mitogen-activated protein kinase kinase 7 (MKK7) induced cell death mediated by FADD and Fas activation, independently of FasL (Chen and Lai, 2001). Persistent JNK activation is required for apoptosis (Chen et al., 1996, Gajate et al., 2002) and leads to clustering of Fas (Chen and Lai, 2001). A JNK-associated protein named JAMP (JNK1-associated membrane protein) has been recently identified as a membrane-anchored regulator of the duration of JNK1 activity in response to diverse stress stimuli (Kadoya et al., 2005). The extent of JNK activation can also be determined by several mitogen-activated protein kinase (MAPK) phosphatases, including MKP5 (Theodosiou et al., 1999), as well as by scaffold proteins, including JIP-1 (Whitmarsh et al., 1998) and POSH (Xu et al., 2003, Xu et al., 2006). Irrespective of the molecular mechanism involved in JNK sustained activity, it is interesting to note that several inducers of FasL-independent Fas capping lead to a rapid and persistent activation of JNK, such as edelfosine (Gajate et al., 1998) and vanadate (Luo et al., 2003). These data suggest that persistent JNK activation could be at least one of the signaling events leading to Fas clustering (Fig. 7). In this regard, ceramide, which also favors Fas aggregation, induces apoptosis through sustained JNK activation (Verheij et al., 1996). However, the molecular events between JNK activation and Fas clustering remain to be elucidated.

View Large Image Download to PowerPoint

Fig. 7. Putative processes involved in the clustering of Fas in lipid rafts. Persistent activation of JNK, increases in ROS or ceramide levels as well as reorganization of the actin cytoskeleton have been suggested to lead to clustering of Fas. Ezrin links Fas with the actin cytoskeleton.

Another putative mechanism implicated in Fas clustering involves the cytoskeleton (Fig. 7), a dynamic intracellular structure that due to its continuous assembly/disassembly could be perfectly equip to translocate proteins and transmit signals (Mollinedo and Gajate, 2003). The interactions between plasma membrane and cytoskeleton play an essential role in various cellular functions, and a link between raft-mediated signaling and the interaction of actin cytoskeleton with raft membrane domains has been suggested (Harder and Simons, 1999). Ezrin, a major protein of the ERM (ezrin, radixin, moesin) proteins linking the actin cytoskeleton to the plasma membrane (Mangeat et al., 1999), interacts with Fas and mediates Fas cell membrane polarization during Fas-induced apoptosis in human T lymphocytes (Fais et al., 2005, Parlato et al., 2000) (Fig. 3). Furthermore, interference with actin cytoskeleton prevented Fas clustering and apoptosis triggered by the antitumor agent aplidin (Gajate and Mollinedo, 2005).

Treatment with tert-butyl hydrogen peroxide induced a rapid Fas aggregation on the surface of Jurkat cells (Huang et al., 2003), involving reactive oxygen species (ROS) in this process (Fig. 7). In addition, free radical scavengers abrogated apoptosis and Fas aggregation induced by γ-irradiation or cisplatin (Huang et al., 2003). These data indicate that ROS participate in γ-irradiation- and cisplatin-induced Fas clustering. However, free radical scavengers did not affect apoptosis triggered by the agonistic Fas antibody CH11, indicating that the ROS-sensitive stage is upstream of a Fas-mediated apoptotic pathway. ROS has been implicated in FasL expression (Bauer et al., 1998), but FasL was dispensable for cell death induced by either cisplatin (Huang et al., 2003) or additional stimuli that trigger both ROS and Fas clustering (Gajate et al., 2000a, Gajate et al., 2000b, Gajate and Mollinedo, 2001). The mechanism by which ROS influences Fas clustering is unknown at present. Presumably, ROS activate signals that eventually result in cytoskeleton reorganization and membrane protein clustering (van Wetering et al., 2002, Wang et al., 2001). In this context, ROS generation has been shown to regulate actin polymerization by reversible glutathionylation of 42-kDa actin (Wang et al., 2001).

5.1. Daxx 

Fas death domain-associated protein (Daxx) was originally identified as a protein that specifically binds to the death domain of Fas and potentiates Fas-induced apoptosis (Yang et al., 1997). Daxx contains a region (amino acids 434–493) of 60 amino acids with a high content (71.7%) of glutamic acid and aspartic acid and comprises two small proline-rich regions. Overexpression of Daxx enhances Fas-mediated apoptosis and activates the JNK pathway. A C-terminal portion of Daxx interacts with the Fas death domain, while a different region activates both JNK and apoptosis. The Fas-binding domain of Daxx behaves as a dominant-negative inhibitor of both Fas-induced apoptosis and JNK activation, while the FADD death domain partially inhibits death, but not JNK activation, and hence Daxx and FADD apparently define two distinct apoptotic pathways downstream of Fas (Yang et al., 1997). However, the importance of the Fas–Daxx–JNK pathway has been questioned when transfection of a Fas death domain mutant that selectively binds Daxx, activated JNK but failed to induce apoptosis (Chang et al., 1999). In addition, Jnk1−/−Jnk2−/− primary murine embryonic fibroblasts showed no inhibition of Fas-mediated apoptosis (Tournier et al., 2000), challenging the putative role of Daxx in enhancing the Fas pathway of apoptosis through JNK activation. These findings suggest that although Daxx can activate JNK upon Fas ligation, JNK activation is not per se sufficient to trigger cell death, at least in certain cell types.

Intriguingly, although Daxx was first reported to mediate the apoptotic signal from Fas to JNK in the cytoplasm, a large proportion of Daxx is mainly located in the nucleus acting as a transcriptional regulator. Daxx associates with the promyelocytic leukemia (PML) nuclear body (PML-NB), which has been proposed to participate in a nuclear pathway for apoptosis (Zhong et al., 2000). The cellular target responsible for the nuclear export of Daxx has been identified as chromosomal region maintenance 1 (CRM1), which is a carrier protein for nuclear export and a receptor for the nuclear export signal of Daxx. Phosphorylation of Ser-667 is required for Daxx binding to CRM1 and for its subsequent nuclear export (Song and Lee, 2004). Phosphorylation of Daxx is mediated through activation of the apoptosis signal-regulating kinase 1 (ASK1)-SEK1-JNK1-homeodomain-interacting protein kinase 1 (HIPK1) signal transduction pathway, and the activated HIPK1 is probably involved in the relocalization of Daxx from the nucleus to the cytoplasm (Song and Lee, 2003). Phosphorylated Daxx is translocated to the cytoplasm, bind to ASK1, and subsequently lead to ASK1 oligomerization (Song and Lee, 2003). Daxx has been found to interact with and activate the upstream JNK kinase kinase ASK1 upon Fas stimulation (Chang et al., 1998). Overexpression of a kinase-deficient ASK1 mutant inhibited Fas- and Daxx-induced apoptosis as well as JNK activation. Cellular localization of Daxx is determined by the relative concentration of ASK1, which controls the dual function of Daxx as a transcriptional repressor in the nucleus and as a proapoptotic signal mediator in the cytoplasm (Ko et al., 2001). ASK1 sequesters Daxx in the cytoplasm, and Daxx binds to the activated Fas only in the presence of ASK1, accelerating Fas-mediated apoptosis (Ko et al., 2001). Thus, Daxx requires ASK1 for its cytoplasmic localization and Fas-mediated signaling. Ectopic expression of Daxx in malignant Jurkat T-cells substantially increases the rate of apoptosis upon incubation with death receptor agonists as well as after incubation with the cytotoxic drug doxorubicin (Boehrer et al., 2005a). Overexpression of Daxx in Jurkat cells slightly sensitized neoplastic cells to the apoptosis-inducing effects of specific chemotherapeutic agents, including bendamustine, cladribine, cytosine-arabinoside and mitoxantrone (Boehrer et al., 2005b). However, the major role of Daxx in promoting apoptosis has been challenged by demonstrating that targeted deletion of Daxx in mice results in early embryonic lethality (E9.5), with extensive apoptosis, thus supporting an anti-apoptotic role of Daxx during embryonic development (Michaelson et al., 1999). In this regard, specific RNA interference for Daxx in cell lines was reported to increase apoptosis (Michaelson and Leder, 2003) and to sensitize cells to the apoptosis induced by Fas, UV or TNFα (Chen and Chen, 2003), further suggesting an anti-apoptotic role for Daxx. Expression of Daxx also inhibits cell death induced by CD43 (leukosialin, sialophorin) in the hematopoietic progenitor cell line TF-1 (Cermak et al., 2002), further supporting an antiapoptotic role for Daxx. Despite a large number of studies attempting to determine Daxx function in cell death, its precise role is only partially understood and remains largely controversial (Salomoni and Khelifi, 2006). Daxx acts as a promiscuously interacting protein, having been found to bind to more than 20 proteins, many of them involved in cell death regulation and transcriptional regulation (Salomoni and Khelifi, 2006).

5.2. FADD 

As shown above FADD is a key adaptor molecule transmitting the death signal mediated by death receptors, and together with procaspase-8 form the DISC. FADD- or caspase-8-deficient mouse embryo fibroblasts and lymphocytes are resistant to Fas-induced apoptosis (Juo et al., 1998, Zhang et al., 1998). These data highlight the importance of the FADD-caspase-8 route for Fas-induced apoptosis. Human FADD gene has a quite simple organization consisting of two exons (286 and 341bp) separated by a unique intron of approximately 2kb. The 208-amino acid sequence of human FADD contains two domains that are particularly well-conserved between species and play a crucial role in transducing the apoptotic signal mediated by death receptors: the death domain at the C-terminus of the protein and the DED at the N-terminus of the protein (Tourneur et al., 2005). Lack of FADD protein expression in cancer cells has been found to be a relevant phenomenon in human malignancies, predicting resistance to chemotherapy and poor outcome (Tourneur et al., 2004). Post-translational modification of FADD by PKCζ, which is up-regulated in a number of tumors, regulates Fas receptor-mediated apoptosis in cells by inhibiting DISC formation following Fas receptor activation, this inhibition being reversed by overexpressing the PKC inhibiting protein prostate apoptosis responsive 4 (PAR-4) (de Thonel et al., 2001).

5.3. FAF1 

Fas-associated factor-1 (FAF1) was identified by yeast two-hybrid assay using the cytoplasmic domain of Fas as bait (Chu et al., 1995). FAF1 is a Fas-associating molecule, which enhances Fas-mediated apoptosis (Chu et al., 1995). The N-terminus of FAF1 binds to the death domain of Fas even though it does not contain the typical death domain (Ryu et al., 1999). Ryu et al. have recently shown that FAF1 is a component of the DISC, formed by interaction of the DED-like region (181–381 amino acid region) of FAF1 and the DEDs of procaspase-8 and FADD, providing a molecular explanation for the proapoptotic role of FAF1 in Fas-mediated signaling (Ryu et al., 2003). Overexpression of human FAF1 in Jurkat cells caused significant apoptotic death, but the FAF1 deletion mutant lacking the N-terminus where Fas, FADD, and procaspase-8 interact protected Jurkat cells from Fas-induced apoptosis leading to a dominant-negative phenotype (Ryu et al., 2003). Park et al. (2005) has found that FAF1 mediates chemotherapeutic-induced apoptosis via participation in the formation of the cytoskeleton-like structure DEFs, found in death receptor-independent apoptosis. Overexpression of FAF1 enhanced DEF assembly and cell death induced by chemotherapeutic agents such as staurosporine, cisplatin and etoposide, whereas antisense FAF1 construct inhibited DEF assembly and chemotherapeutic-induced apoptosis (Park et al., 2005). Confocal microscopy revealed that FAF1 was present in DEFs together with FADD and caspase-8 (Park et al., 2005). In addition, FAF1 has been shown to interact physically with nuclear factor-κB (NF-κB) p65 preventing translocation of RelA (NF-κB p65) into the nucleus and hence inhibiting its DNA-binding activity (Park et al., 2004). The NF-κB suppressor activity of FAF1 was also mapped to the DED-interacting domain (amino acids 181–381) (Park et al., 2004). Thus, FAF1 is involved in dual signaling mechanisms and could potentiate apoptosis not only by strengthening apoptotic signaling via DISC, but also by suppressing the cell's survival potential by down-regulation of NF-κB. FAF1 has been found to be significantly reduced in gastric carcinomas (Bjorling-Poulsen et al., 2003), suggesting a role for this protein in cancer progression.

5.4. FAIM 

Fas apoptosis inhibitory molecule (FAIM) was identified as a Fas antagonist from a differential display strategy to detect cDNAs present in B cells rendered Fas resistant, but absent in those rendered Fas sensitive (Schneider et al., 1999). FAIM behaves as an inducible effector molecule that mediates Fas resistance produced by surface Ig in B cells (Schneider et al., 1999). However, FAIM is broadly expressed in various tissues and its sequence is highly conserved in evolution from Caenorhabditis elegans to humans (Rothstein et al., 2000, Schneider et al., 1999). Recently, an alternatively spliced isoform of FAIM, FAIM-L, has been identified, which is predominantly expressed in the brain (Zhong et al., 2001). An additional function of FAIM in the nervous system lies in promoting neurite outgrowth by a mechanism involving activation of the Ras-extracellular signal-regulated kinase (ERK) pathway and NF-κB, but has no effect on neuronal survival (Sole et al., 2004), which contrasts with its role in modulating Fas signaling and cell survival in the immune system.

5.5. FAP-1 

Fas-associated phosphatase-1 (FAP-1) is a tyrosine phosphatase that was identified as a protein that associates with the negative regulatory domain (C-terminal 15 amino acids) of Fas using the yeast two-hybrid system (Sato et al., 1995). FAP-1 is one of the largest known non-receptor protein tyrosine phosphatases (about 270kDa) and is the only protein known to associate with the C-terminal negative regulatory domain of Fas. FAP-1 contains an N-terminal leucine zipper motif, an ezrin-like cytoskeleton binding domain, and six PSD95/Dlg/Zo-1 homology (PDZ) domains (Sato et al., 1995). The C-terminal amino acids (SLV) of human Fas are necessary and sufficient for its interaction with the third PDZ domain (PDZ3) of FAP-1 (Yanagisawa et al., 1997). Stable introduction of an FAP-1 cDNA in Fas-sensitive Jurkat cells (Li et al., 2000, Sato et al., 1995) as well as in other cancer cells (Li et al., 2000, Ungefroren et al., 2001) has been shown to protect these cells from Fas-mediated cytotoxicity. In addition, FAP-1 is strongly expressed in human tumor cells which are largely refractory to Fas-induced apoptosis, such as pancreatic adenocarcinoma, colon carcinoma cells, and astrocytomas (Foehr et al., 2005, Ungefroren et al., 2001, Yao et al., 2004), but pretreatment of cells with synthetic SLV tripeptide abolishes Fas resistance (Ungefroren et al., 2001, Yao et al., 2004). Association of FAP-1 with Fas inhibits Fas trafficking to the cell surface, reducing cell surface Fas levels (Ivanov et al., 2003) and enhancing colocalization of Fas/FAP-1 in the Golgi complex (Ungefroren et al., 2001). Down-regulation of FAP-1 expression by specific RNA interference restores Fas export (Ivanov et al., 2006), and leads to increased sensitivity to Fas-induced cell death (Foehr et al., 2005). It has been recently reported that FasL induces tyrosine phosphorylation of Fas in astrocytoma cells, and FAP-1 dephosphorylates tyrosine-275 in the C-terminus of Fas (Foehr et al., 2005). This finding indicates that Fas can be regulated by reversible phosphorylation. This notion is supported by previous observations that showed a physical interaction between Fas and the tyrosine kinase p59fyn (Atkinson et al., 1996), and that Fas ligation induced early tyrosine phosphorylation of multiple proteins and inhibitors of protein tyrosine kinases block Fas-induced DNA fragmentation and prolong cell survival (Eischen et al., 1994). These results suggest that protein tyrosine kinase activation is an early and obligatory signal in Fas-induced apoptosis (Eischen et al., 1994).

5.6. FIST/HIPK3 

Using the yeast two-hybrid system to screen for proteins that bind to the cytososlic domain of murine Fas, another Fas-interacting protein was identified and named Fas-interacting serine/threonine-protein kinase (FIST), which was identical to homeodomain-interacting protein kinase 3 (HIPK3). FIST/HIPK3 is a 130-kDa serine/threonine kinase that causes FADD phosphorylation. FIST/HIPK3 overexpression inhibited FasL-induced JNK activation, but did not affect apoptosis, suggesting that Fas-associated FIST/HIPK3 modulates one of the two major signaling pathways of Fas (Rochat-Steiner et al., 2000).

5.7. Lifeguard 

Lifeguard (LFG) was isolated from a cDNA library generated in a retroviral vector from a human lung fibroblast cell line, MRC5, which was not sensitive to FasL in the absence of the protein synthesis inhibitor cycloheximide. HeLa cells were transduced with the retroviral vectors containing the cDNA library, maintained in the presence of the mouse anti-human Fas agonistic antibody CH11, and then LFG was isolated from the genomic DNA of the surviving pool of cells and was shown to inhibit death mediated by Fas (Somia et al., 1999). LFG binds directly to the Fas receptor, but not to the Fas adaptor protein FADD, and does not inhibit binding of FADD to Fas (Somia et al., 1999). LFG is expressed in most tissues, except spleen and placenta, being highly abundant in the brain (Somia et al., 1999). LFG expression seems to be mediated, at least in part, by the phosphatidylinositol 3-kinase (PI 3-kinase)-Akt/protein kinase B (PKB) pathway (Beier et al., 2005).

5.8. Sentrin/SUMO-1 

Sentrin was originally isolated by the yeast two-hybrid system with the death domain of Fas as a bait (Okura et al., 1996). Sentrin (a.k.a. small ubiquitin-related modifier-1, SUMO-1) is a protein of 101 amino acids, that when overexpressed provides protection against Fas-mediated cell death (Okura et al., 1996). Sentrin/SUMO-1, is a ubiquitin-like protein that can covalently modify a large number of cellular proteins, including signaling and nuclear proteins with important roles in regulating transcription, chromatin structure, and DNA repair (Gill, 2004), by a process named sentrinization (or sumoylation) in a manner analogous to ubiquitination. Despite sentrin has a ubiquitin-like domain (amino acids 22–97), it contains four additional amino acids (HSTV) at the C-terminus, which are cleaved so as to allow the conjugation of sentrin to other proteins via the glycine residue at the C-terminus (Kamitani et al., 1997). Modification by sentrin requires activation of the Gly-97 by the ubiquitin-conjugating enzyme Ubc9 and the activating enzyme complex UBA2/AOS1 (Gong et al., 1999). Unlike ubiquitin modification, sentrinization/sumoylation is not a signal for degradation, but it regulates activity and/or localization of proteins, being suggested to be involved in nuclear translocation of target proteins and obstruction of cell death signaling by prevention of FADD binding to Fas (Gill, 2004, Okura et al., 1996). Interestingly, sentrin has been shown to interact with Daxx (Ryu et al., 2000).

5.9. Ubc9 

The ubiquitin-conjugating enzyme Ubc9, which is essential for sentrinization/sumoylation, has been found to bind to Fas at the interface between the death domain and the membrane-proximal region of Fas (Becker et al., 1997). Ubc9 also binds Daxx at the same region as Fas and sentrin/SUMO-1 (Ryu et al., 2000). MCF7 human breast tumor cells expressing a Ubc9 dominant-negative mutant were found to accumulate more cytoplasmic Daxx than the control cells and were more sensitive to anticancer agents (Mo et al., 2004). Ubc9 expression levels are elevated in ovarian tumors compared to the matched normal ovarian specimens, suggesting that Ubc9 may play a role in tumorigenesis. Inoculation of MCF-7 cells overexpressing a dominant-negative mutant of Ubc9 and wild-type Ubc9 as xenografts in mice revealed that tumors expressing the wild-type version of Ubc9 grew better than the vector control, while tumors expressing the Ubc9 mutant reduced growth, further supporting a role for Ubc9 in tumorigenesis (Mo et al., 2005).

6.1. FLASH 

FLICE-associated huge protein (FLASH) is a protein with binding activity to the DEDs of procaspase-8 and FADD through its DED-like domain, being a component of the DISC (Imai et al., 1999). FLASH was identified by using the yeast two-hybrid technique and two tandem-repeated DED domains of procaspase-8 as a probe (Imai et al., 1999). FLASH contains a motif structurally related to C. elegans protein CED-4/apoptotic protease activating factor-1 (Apaf-1) and two tandem-repeated DED homologous domains (Imai et al., 1999). FLASH enhances the activation of caspase-8 in Fas-mediated apoptosis, indicating that DED-containing proteins seem to modulate the apoptotic process.

6.2. FLIP 

The cellular FLICE-inhibitory protein (c-FLIP) is another DED-containing protein that acts as a major regulator of apoptosis and can switch life/death signals in tumor cell (Tucker et al., 2004). Although more than 10 isoforms of FLIP mRNA have been described, only two of them have been significantly studied at the protein level, c-FLIP-long (c-FLIPL) (480 amino acids, 55kDa) and c-FLIP-short (c-FLIPS) (221 amino acids, 28kDa). Both proteins have two N-terminal DED motifs that are very similar to the DED's on procaspase-8. c-FLIPL in addition has a C-terminal domain that is homologous to the catalytic domain of caspase-8, but devoid of enzymatic activity. Both proteins can be recruited to the DISC, binding to the Fas–FADD complex and inhibiting the recruitment of caspase-8 to DISC, and preventing induction of apoptosis mediated by death receptors (Kataoka, 2005). Interestingly, c-FLIP expression is increased in some human tumors, and overexpression of cFLIP in tumor cells results in their escape from T-cell immunity (Dutton et al., 2006).

6.3. TOSO 

A novel human gene dubbed Toso was identified as a molecule that blocks Fas-mediated apoptosis by Hitoshi et al. using retroviral cDNA library-based functional cloning (Hitoshi et al., 1998). The investigators named the molecule Toso after the Japanese liquor “that is drunk on New Year's Day to celebrate long life and eternal youth” (Hitoshi et al., 1998). Toso was found to be expressed mainly by lymphocytes and was suggested to regulate apoptosis by inhibiting caspase-8 processing, potentially through up-regulation of cFLIP (Hitoshi et al., 1998). Toso-overexpressing primary T lymphocytes from mouse Toso transgenic mice are resistant to Fas/FasL-induced apoptosis, but sensitive to glucocorticoid-induced apoptosis (Song and Jacob, 2005). Toso knock-out mice are embryonically lethal. Toso is a type I membrane protein where its C-terminal is involved in FADD binding, and hence it could interfere with the recruitment of caspase-8 to FADD molecule and its activation (Song and Jacob, 2005).