Loss of Caspase-8 in Tumor Cells: Mechanism to Overcome Integrin-Mediated Death?
Successful metastasis is determined by the ability of tumor cells to invade surrounding tissue, survive in the circulation, extravasate, arrest, and proliferate within the secondary organ. During this process, mutations in oncogenes and tumor suppressor genes lead to genetic instability, which bestow tumor cells with a selective growth advantage and aid the acquisition of the malignant phenotype (1). At the core of the metastatic process are the reciprocal interactions between the tumor cells and their microenvironment, as mediated by direct cell-cell contact or cytokine and growth factor signaling. The complexity of the crosstalk among tumor cells, stromal cells, and immune cells determines the response of the host microenvironment, which in turn modifies the invasive behavior of tumor cells. During metastasis, cells leave their “familiar neighborhoods” and in “foreign territory” they rely on interactions with extracellular matrix (ECM) to sense their surroundings (2). The ECM is a dynamic environment, varying from tissue to tissue, as well as at different times and states within a single tissue. The interaction of heterodimeric cell-surface receptors called integrins with the ECM transduces signals between the cells and the ECM (Figure 1⇓). These matrix-initiated signals profoundly affect many cellular functions, including cell survival, proliferation, polarity, motility, differentiation, and even cell proliferation vs survival (3). Cancer cells have the ability to switch the types of integrins they express in order to overcome death signals and to activate pro-survival pathways (4); such resistance to apoptotic signals represents an important aspect of tumor progression, metastasis, and resistance to chemotherapy.
Altered expression and activity of various components of the apoptotic pathway, including receptors, ligands, adaptors, and caspases, can contribute to malfunction of the apoptotic machinery and, ultimately, to a more malignant phenotype. Such altered expression is exemplified by the elimination of caspase-8 in high-risk neuroblastomas, the most common group of early childhood tumors (5). Caspase-8 is a key apoptotic factor involved in both the extrinsic and the intrinsic apoptotic pathways (6, 7) (Box 1). In the most aggressive, disseminated stage of neuroblastoma, the caspase-8 gene, CASP8, is deleted or silenced by methylation, a change strongly associated with amplification of the MYCN gene, a known molecular marker of poor prognosis (8, 9). Highly aggressive neuroblastomas are generally unresponsive to treatment and account for most of the pediatric cancer deaths (5). The overall loss of caspase-8 expression in neuroblastomas is estimated to be 25–35% and up to 70% in highly aggressive tumors (5, 9).
The Two Major Apoptotic Pathways: Extrinsic (Death Receptor-Induced) and Intrinsic (Mitochondrial, Stress-Induced)
The extrinsic pathway is initiated by ligation of transmembrane death receptors belonging to the tumor necrosis factor (TNF)-receptor family (e.g., Fas/Apo-1/CD95, TNF-R1 or TRAMP), and the assembly of the death-inducing signaling complex (DISC). The DISC is the site of activation for caspase-8. Apoptosis through the extrinsic pathway is responsible for eliminating cells during development and immunosurveillance (6). The intrinsic pathway becomes activated upon response to ionizing radiation, chemotherapeutic agents and mitochondrial damage, leading to cytochrome c release and recruitment of caspase-9 into an “apoptosome.” Once activated, caspase-9 can cleave and activate other caspases, including caspase-8 (7).
A recent study by Stupack et al. investigated the effects of caspase-8 deletion on the metastatic potential of neuroblastomas (10). The authors utilized a chorioallantoic membrane (CAM) model, a widely utilized in vivo experimental system for studying tumor cell metastasis in which tumor cells are grafted onto the highly vascularized chorioallantoic membrane of the developing chick embryo through an opening in an egg shell (11). This model has some limitations owing to the instability of embryonic tissue (12) and the requirement that the CAM must be wounded in order for the tumor cells to migrate (13). At the same time, this a convenient system which allows one to assess tumor growth, invasion, and spontaneous metastasis within the same animal—events not as readily achieved in mouse models of neuroblastoma (14). The CAM system has been used by several groups for semiquantitative assessment of metastasis (15–18) and, in combination with real time polymerase chain-reaction (PCR), allows for a highly sensitive and quantitative detection of the metastatic dissemination of tumor cells (11). Utilizing this quantitative technique, Stupack et al. report a much higher incidence of apoptosis in caspase-8+, locally invasive extratumoral cells as compared to caspase-8– cells. Conversely, knockdown of caspase-8 expression by RNA interference promotes metastasis but has no effect on primary tumor growth. Their hypothesis that loss of caspase-8 promotes metastatic disease is further validated in mice, where decreased levels of caspase-8 mRNA are observed in metastases.
Caspase-8 expression is a key determinant of sensitivity to apoptosis induced by death-inducing ligands or cytotoxic drugs (19)—apoptosis that is mediated through the extrinsic pathway. Decreases in or an absence of caspase-8 mRNA and protein render neuroblastoma cell lines resistant to apoptosis induced by tumor necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL), CD95-specific antibody, or TNF-α (8, 19). This raises an intriguing question: do caspase-8–expressing neuroblastoma cells die in a death receptor-dependent manner? The study by Stupack et al. provides evidence for increased apoptosis in the absence of pro-apoptotic cycloheximide in caspase-8–expressing cells in vivo (10). Cycloheximide is known to augment the effect of death ligands and pre-sensitize neuroblastoma cells to death-receptor-mediated killing (8, 9). In the study by Stupack et al., an in vitro treatment with cycloheximide led to an increased apoptosis in caspase-8+ cells as compared to caspase-8− cells; however, no difference in the rates of apoptosis among these cell lines was detected in the absence of cycloheximide (10). Therefore, an increase in apoptosis in caspase-8+ cells in the absence of cycloheximide in vivo, in combination with previous results showing that etoposide- and staurosporine-mediated death does not depend on caspase-8 expression (10, 20), might suggest that caspase-8–expressing neuroblastomas undergo apoptosis in a caspase-8–dependent manner that is independent of death receptor activation.
One such form of stress response- and death receptor-independent apoptosis is integrin-mediated death (IMD), a process that occurs in adherent cells and, thus, is distinct from anoikis, a form of programmed cell death resulting from a loss of integrin-mediated cell-matrix contact (21). IMD results from altered clustering of integrins and depends on the ability of the integrin β-subunit to recruit and activate caspase-8 at the cell surface (22). IMD may also depend on activity of protein kinase A (PKA), which can activate caspase-8 when integrin ligation (but not cell adherence) is perturbed (23) as a result of integrins failing to bind their ECM ligands (Figure 1⇓). For instance, αvβ3-expressing endothelial cells undergo rapid apoptosis when seeded in a three-dimensional collagen I matrix (which does not ligate the αvβ3 integrin) or upon treatment with an αvβ3 integrin antagonist (22, 24). Conversely, inhibitors of caspase-8, but not inhibitors of caspase-9, restore neovascularization in tissues treated with an αvβ3 integrin antagonist, suggesting that αvβ3-mediated apoptosis is caspase-8–dependent (22). A similar association between unligated αvβ3 integrins and abundant caspase-8 expression has been linked to the regulation of osteoclast apoptosis (25).
In addition to being affected by αvβ3 integrins, apoptotic signaling pathways are triggered by non-ligation or antagonism of β1 integrins (22). For example, β1 integrin signaling has been implicated in mediating resistance to chemotherapy in cancers of lung, colon, and breast (26–28); a recent study by Park et al. showed that β1 integrin antagonism is a promising therapeutic approach in breast cancer therapy (29). In neuroblastoma, tumors with poor prognosis lack integrins with β1 subunits, consistent with increased apoptosis in the presence of β1. It is interesting to note that there is an inverse relationship between the loss of β1 integrin and overexpression of N-myc, whose expression is associated with silencing of CASP8 (30, 31). Coincidentally, the report by Stupack et al. demonstrated that caspase-8 co-precipitates with unligated β1 integrins from neuroblastoma cells, and neuroblastoma cells expressing caspase-8 undergo apoptosis on collagen I, a matrix which is not a ligand for the integrins (α3β1, α5β1, or αvβ5) expressed on these cells (10). Further investigation of α3β1 integrin, the most abundant receptor on the cell lines analyzed in the Stupak et al. study, revealed that α3β1 antagonism selectively enhances apoptosis of caspase-8-expressing neuroblastoma cells and prevents their metastasis in vivo.
The above findings indicate that neuroblastoma cells can escape IMD either by loss of caspase-8 or by antagonism or unligation of specific integrins. Historically, integrins have been shown to promote cell survival by mediating cellular adhesion to the ECM (32). The recent findings by Stupack et al. (10) suggest a new role for integrins in tumor invasion and metastasis. Integrins seem to have a capacity to selectively suppress or promote apoptosis depending on the environmental cues they receive from ECM. Their life- or death-promoting activities depend on the presence of suitable ligands, which places them in a newly appreciated category of apoptotic triggers: the dependence receptors (22, 33). These phenomena are most likely responsible for the discrepancies in the literature regarding the roles of α5β1 integrin, a fibronectin receptor (Figure 1⇓). α5β1 has been shown in some studies to induce metastasis and correlate with poor prognosis, yet in other studies to suppress tumorigenicity and angiogenesis (22), perhaps depending on interactions with different ECMs.
Caspase-8 seems to function downstream of unligated integrins (22), and suppression of its activity might represent the mechanism by which neuroblastoma cells overcome IMD, survive, and metastasize (10). Because integrins govern so many different aspects of normal cell functions, it is only natural to speculate that in addition to direct activation of caspase-8 they might also have an indirect effect on caspase-8-mediated apoptosis through interaction with other signaling molecules. Some of the possible direct and indirect interactions between integrins and caspase-8 are illustrated in Figure 2⇓. For example, integrin signaling increases the expression of the caspase-8 inhibitor c-FLIP (Fas-associated death domain–like interleukin 1β–converting enzyme (FLICE)-inhibitory protein) (2), and overexpression of anti-apoptotic c-FLIP has been demonstrated in high-stage neuroblastoma (34). In addition, human keratinocytes, which undergo apoptosis as a result of β1 integrin antagonism, fail to die when they overexpress c-FLIP (35). Bcl-2, a known inhibitor of the intrinsic apoptotic pathway can bind and sequester caspase-8 (36). Bcl-2 is frequently expressed in neuroblastoma cells and its presence inversely correlates with apoptosis (34). Interestingly, the expression of α5β1 integrins or αvβ3 integrins by neuroblastoma cells can induce expression of bcl-2, a process promoting survival (37). Integrin antagonism may activate the tumor suppressor p53, which forms a complex with caspase-8, thus suggesting involvement of p53 in IMD (22).
Many cytotoxic drugs induce caspase-8-dependent apoptosis, thus the loss of caspase-8 in neuroblastoma might represent an important determinant of resistance to chemotherapy. Sensitivity to chemotherapy depends not only on genetic alterations of the tumor cell but also on the tumor microenvironment. Tumor cells are more dependent on suppression of apoptosis than other cells—a strong rationale for employing pro-apoptotic strategies to restore sensitivity to chemotherapeutics. Many of the novel therapeutics currently in development do, indeed, target components of the caspase-8-dependent IMD pathway and they include among others direct modulators of caspase-8 expression (e.g., interferon gamma) as well as c-FLIP, members of the bcl-2 family, and the caspase-8 inhibitors termed X-linked inhibitor of apoptosis protein (XIAP) and survivin (38). Several integrin inhibitors are also under investigation for their possible use as therapeutic agents (24). As recent advances improve our understanding of molecular mechanisms regulating metastatic processes, we also have to keep in mind they raise more questions than they answer. Complex and sometimes redundant pathways and interactions among tumor cells, host cells and the surrounding ECM require that the therapeutic targets are chosen carefully, as in addition to exerting response in the tumor cells they will likely affect other important physiological functions.
The integrin receptors and their major ECM ligands. Each heterodimeric transmembrane integrin receptor consists of one α and one β subunit. α subunits: 
α1, 
α2, 
α3, 
α4, 
α5, 
α6, 
α7, 
α8, 
α9, 
α10, 
α11, 
αv; β subunits: 
β1, 
β3, 
β4, 
β5, •β6, 
β7, 
β8. Composition of each subunit determines ligand specificity. Figure depicts seventeen distinct integrins and their binding
preferences for collagen, laminin, tenascin, fibronectin, and vitronectin.
Simplified schematic of apoptosis pathways focused on integrin-mediated death (IMD). Integrins ligated by the appropriate ligands serve to mediate cell survival pathways (purple). In their unligated or antagonized state, integrins promote the recruitment and activation of caspase-8 at the cell surface. This results in initiation of the IMD pathway (red), which occurs independently of the death receptor-mediated extrinsic pathway (blue) or stress-induced intrinsic pathway (green). Integrin antagonism can potentiate caspase-8 activity by activating protein kinase A (PKA), which in turn activates caspase-8, or by suppressing the expression of inhibitors, c-FLIP and bcl-2. c-FLIP, Fas-associated death domain–like interleukin 1β–converting enzyme (FLICE)-inhibitory protein.
- © American Society for Pharmacology and Experimental Theraputics 2006
References:
Izabela Podgorski, PhD, is a post-doctoral research associate in Bonnie F. Sloane’s laboratory in the School of Medicine at Wayne State University. She received a postdoctoral fellowship from the Department of Defense to support her research on molecular mechanisms underlying the metastasis of prostate cancer to bone. E-mail: ipodgors{at}med.wayne.edu
Bonnie F. Sloane, PhD, is Distinguished Professor and Chair of the Department of Pharmacology at Wayne State University in Detroit. She received her BS and MA from Duke University and her PhD in Physiology from Rutgers. She is internationally recognized for her research examining proteolytic enzymes and their impact on the progression of breast cancer and is a leader in applying imaging to the analysis of proteolytic enzymes in cancers. She co-founded the International Proteolysis Society as well as the Protease Consortium. Her research has been honored nationally through her selection as one of the four initial Avon Foundation/AACR International Scholar Hosts for Breast Cancer Research and she is director of a Department of Defense Breast Cancer Center of Excellence. She also serves as special advisor to the NCI Cancer Imaging Program. E-mail: bsloane{at}med.wayne.edu; fax (313) 577-6739.
