Citation
Bruin, E. C. de. (2008, February 27). Apoptosis in cancer : regulation and prognostic value. Retrieved from https://hdl.handle.net/1887/12616
Version: Not Applicable (or Unknown)
License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/12616
Note: To cite this publication please use the final published version (if applicable).
Apoptosis in Cancer
regulation and prognostic value
ISBN 978-90-9022820-4
Layout: Chris Bor, Medical Photography and illustration, AMC, Amsterdam Printed by: Buijten & Schipperheijn, Amsterdam
Cover: the left pictures show apoptosis in vitro; melanoma cells untreated, or treated with etoposide to induce apoptosis, whereas the right pictures show apoptosis in vivo; two different rectal carcinoma samples with low or high numbers of apoptotic (brown) cells.
The work presented in this thesis was financially supported by Stichting Vanderes and the Dutch Cancer Society (RUL 2002-2733).
Publication of this thesis is financially supported by Dutch Cancer Society, Stichting Klinische Oncologie Leiden, BD Biosciences, Corning, Pfizer and Tebu-bio.
Financial support regarding the thesis defence by Abbott Diagnostics, Adriaan van Erk Werkmaatschappijen B.V., Bouwbedrijf De Vries en Verburg and Stichting Nationaal Fonds tegen Kanker is greatly acknowledged.
Apoptosis in Cancer
regulation and prognostic value
Proefschrift
ter verkrijging van
de graad van Doctor aan de Universiteit Leiden, op gezag van Rector Magnificus prof. mr. P.F. Van der Heijden,
volgens besluit van het College voor promoties te verdedigen op woensdag 27 februari 2008
klokke 15.00 uur
door
Elizabeth Cornelia de Bruin
geboren te Bergambacht in 1977
Prof. dr. J.W.R. Nortier Prof. dr. C.J.H. van de Velde
Prof. dr. J.P. Medema (AMC, Amsterdam)
Co-promotor:
Dr. C.A.M. Marijnen
Referent:
Prof. dr. J. Borst (NKI, Amsterdam)
Overige leden:
Prof. dr. J.H.J.M. van Krieken (Radboud UMCN, Nijmegen) Prof. dr. J. Morreau
Dr. L.T.C. Peltenburg
Prof. dr. T.N.M. Schumacher (NKI, Amsterdam)
C.S. Lewis
Chapter 1
General introduction and Outline of this thesis 9
Part I: Apoptosis regulation in vitro 43
Chapter 2
c-Myc is able to sensitize human melanoma cells to diverse apoptotic triggers.
45 Melanoma Research, 2004; 14: 3-12
Chapter 3
Expression and function of the apoptosis effector Apaf-1 in melanoma.
65 Cell Death and Differentiation, 2005; 12: 678-679
Chapter 4
A serine protease is involved in the initiation of DNA damage- induced apoptosis.
71 Cell Death and Differentiation, 2003; 10: 1204-1212
Part II: Apoptosis regulation and prognostic value in vivo 91
Chapter 5
Prognostic value of apoptosis in rectal cancer patients of the Dutch total mesorectal excision trial: radiotherapy is redundant in intrinsically high-apoptotic tumors.
93
Clinical Cancer Research, 2006; 12(21): 6432-6436
Chapter 6
Caspase-3 activity predicts local recurrence in rectal cancer. 105 Clinical Cancer Research, 2007; 13(19): 5810-5815
Chapter 7
Macrodissection versus microdissection of rectal carcinoma: minor influence of stroma cells to tumor cell gene expression profiles.
119 BMC Genomics, 2005; 6: 142
Apoptosis, 2007;12:1671-1680
Chapter 9
Epithelial HLA-DR expression predicts reduced recurrence rates and prolonged survival in rectal carcinoma patients
171 Clinical Cancer Research (in press)
Chapter 10
Summary and General discussion 187
Nederlandse samenvatting 197
List of publications 203
Curriculum vitae 205
Dankwoord 207
1
C h a p t e r
Introduction
Introduction
The process by which a normal cell develops into a malignant cell with the capacity to form a tumor is thought to be a multistep process, which requires several cellular alterations.
Evasion of apoptotic cell death is one of the proposed alterations that enable malignant cells to expand and progress (1). Importantly, evasion of apoptosis is also recognized as a factor that makes tumors resistant to anti-cancer therapies (2). Much research on anti- cancer therapies has therefore focused on finding ways to overcome this resistance in order to improve the induction of apoptosis. However, the contribution of apoptosis resistance to treatment failure remains a matter of debate, especially in solid tumors (3). In addition, though apoptosis has taken a central position in cell death research for a long time, increasing attention is being directed towards other types of cell death, such as mitotic catastrophe, autophagy and necrosis. These alternative types of cell death may compensate for the resistance to apoptosis. It follows that knowledge regarding the regulation of apoptosis and non-apoptotic death pathways will help us to better understand their impact on tumor development and treatment response in vivo. We therefore summarize and discuss our current knowledge on the role of the different types of cell death in tumor development and treatment, focusing primarily of colorectal cancer.
Cell Death Pathways
Apoptosis
Apoptosis is a tightly regulated type of cell death mediated by caspases, which are cys-dependent Asp-specific proteases (4,5). The term ‘apoptosis’ was based on the morphological characteristics of the dying cells, which include cellular shrinkage, nuclear condensation, membrane blebbing and eventually fragmentation into membrane bound apoptotic bodies (6), as illustrated in Figure 1. During apoptosis, the cell membranes lose their asymmetry, resulting in exposure of phosphatidylserine (PS) on the cell surface. This PS functions as ‘eat me’ signal for macrophages, which results in effective clearance of the dying cell (7). This type of cell death is therefore suggested not to trigger inflammation.
Apoptosis can be initiated by two different types of signals: extracellular ligands or intracellular stress signals (illustrated in Figure 2). When extracellular ligands such as Fas ligand, TNFα or TRAIL (TNF-related apoptosis-inducing ligand) bind to their receptors, the intracellular death domains of these receptors recruit adaptor proteins (such as FADD and TRADD) and initiator caspase-8 and -10. Together these comprise the death-inducing signaling complex (DISC) (8,9). Caspase-8 and -10 are activated at the DISC, due to induced proximity of the caspases (10-12). This caspase activation is controlled by c-FLIP (cellular FLICE inhibitory proteins). c-FLIP is detected in two forms: short and long, which are called c-FLIPS and c-FLIPL. Whereas c-FLIPS has been shown to prevent caspase-8 activation (13,14), the effect of c-FLIPL is more ambiguous. The proposed model is that its effect depends on the ratio of c-FLIPL to caspase-8. c-FLIPL is suggested to compete with caspases
for binding to the DISC, and prevents caspase binding when highly expressed, but facilitates caspase binding and activation when lowly expressed (13,15-20). Once caspase-8 is active, it propagates apoptosis via cleavage of executioner caspases.
Intracellular stress signals, such as growth factor withdrawal, DNA damage or oxidative stress, induce apoptosis via permeabilization of the mitochondrial outer membrane resulting in the release of cytochrome c into the cytosol. The mitochondrial outer membrane permeabilization (MOMP) is tightly regulated by the Bcl-2 family of proteins. This family consists of pro- and anti-apoptotic proteins. The multidomain pro-apoptotic Bax and Bak are essential for permeabilization, since mitochondria deficient for Bax and Bak fail to release cytochrome c (21). Bax and Bak probably induce MOMP by forming pores upon oligomerization (22). The pro-apoptotic BH3-only family members (such as Bid, Bim, Bad, Noxa and Puma) promote cytochrome c release. Several peptide studies show that they activate Bax and/or Bak either through direct binding or indirectly through binding of the anti-apoptotic Bcl-2 proteins (such as Bcl-2 and Bcl-xl). This latter binding relieves the inhibitory function of the anti-apoptotic proteins, and can result in activation of Bax and/
mitotic catastrophe apoptosis
PS PS PS
PS
PS PS
necrosis autophagy
LC3 LC3
mitotic catastrophe apoptosis
PS PS PS
PS
PS PS
PS PS PS
PS
PS PS
necrosis necrosis autophagy
LC3 LC3
LC3 LC3
Figure 1. morphological characteristics of apoptosis and non-apoptotic cell death.
Apoptosis is characterized by membrane blebbing, cytoplasmic shrinkage, chromatin condensation, exposure of PS on the cell surface, and finally the formation of apoptotic bodies. In experimental assays, apoptotic cell death is often determined by binding of Annexin V to the exposed PS or by caspase-cleaved proteins or fragmented DNA. Death by autophagy is characterized by the double-membrane vesicles containing cytosolic organelles. The autophagic-vesicle-associated form of Atg8/LC3 is used as marker of autophagy, since its cleavage, lipidation and recruitment to the autophagosomes results an increased mobility with Western blot assays, and a punctate staining of the protein which can be visualized by fluorescence microscopy using green-fluorescent protein (GFP)-fused Atg8/LC3. Cells dying from mitotic catastrophe are usually large and contain uncondensed chromosomes. The main characteristic of mitotic catastrophe is the formation of multiple micronuclei, and also aberrant mitotic spindle formation can be involved. During necrosis, cells swell and loose their membrane integrity.
or Bak (23-26). Importantly, the p53 tumor suppressor protein can activate transcription of BH3-only proteins in the context of DNA damage, and thereby promotes MOMP (27).
When cytochrome c is in the cytosol, it forms a complex - the apoptosome - with Apaf-1 and initiator caspase-9. Similarly to caspase-8, caspase-9 is activated by the induced proximity in the apoptosome (11,28,29). Active caspase-9 propagates the apoptotic pathway by cleavage and thereby activation of executioner caspases.
The extracellular and intracellular apoptotic pathways cross at the level of the mitochondria since caspase-8 can cleave the protein Bid into its active form tBid. Being a pro-apoptotic member of the Bcl-2 family, tBid translocates to mitochondria where it induces MOMP (30).
Both apoptotic pathways lead to activation of the executioner caspases, caspase-3, -6 and -7, which are the main proteases that degrade the cell. Their activity is, at least to some extent, kept under control by IAPs (Inhibitor of Apoptosis Proteins) (31,32). IAPs themselves are inhibited by the proteins SMAC/DIABLO (33,34) and the serine protease HtrA2/Omi (35). These proteins are also released from the mitochondria, possibly simultaneously with cytochrome c to alleviate the inhibitory signal and to enhance the apoptotic signal.
Non-classical induction of apoptosis
Although the activity of caspases is a characteristic feature of apoptosis, other proteases than caspases can be involved in the induction and/or propagation of apoptosis as well.
Several of these proteases – cathepsins, calpains and serine proteases - have now been identified. A well-known serine protease that initiates apoptosis is Granzyme B, a protease released from cytotoxic lymphocytes. Granzyme B can induce apoptosis via cleavage of Bid and activation of the mitochondrial pathway, since both Bcl-2 overexpression and Bid- deficiency block its pro-apoptotic activity (36-38). Alternatively, cell-free studies show that Granzyme B can cleave and activate caspases-8 and -3 directly (39,40). Direct or indirect caspase-3 activation may thus both occur, and in part depend on the recently described differences in Granzyme B-induced pathways between human and mouse (40,41). Next to its role in caspase-dependent apoptosis, Granzyme B might also play a role in caspase- independent apoptosis. This likely involves cleavage of the Rho-regulated kinase Rock II (42), and cleavage of several other substrates that recently have been identified using proteomic screens (43).
Another serine protease involved in apoptosis is the mitochondrial protein HtrA2/Omi.
When released into the cytosol, HtrA2/Omi can enhance apoptosis by two mechanisms.
One mechanism is through binding and inactivation of IAPs. Studies with mutated proteins show that HtrA2/Omi does not require its protease activity for inactivating IAPs. However, the protease domain has a yet unidentified role in inducing apoptosis, independent of the IAP- binding region (44-46). This latter HtrA2/Omi-induced cell death can occur independent of caspases, as has been shown for TNF-induced apoptosis of neutrophils (47).
Next to Granzyme B and HtrA2/Omi, an unknown serine protease is reportedly involved in the induction of apoptosis upon several apoptotic stimuli. This protease is sensitive to the protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride (AEBSF). Inhibition
of this serine protease in combination with inhibition of caspases renders cells resistant to apoptosis induced by DNA damage or endoplasmatic reticulum (ER) stress, whereas neither one of these inhibiters could rescue the cells on its own (48,49). How this serine protease induces apoptosis is currently unclear, but a recent study shows a role for an AEBSF-sensitive protease in cytochrome c release, independent of Bax and Bak activation (50).
Next to serine proteases, also cysteine proteases have been reported to induce apoptosis, such as lysosomal cathepsins. When released into the cytosol, these cathepsins likely induce apoptosis via cleavage of Bid (51). In agreement with this hypothesis, studies using cell- free systems show that lysosomal extracts can cleave cytosolic Bid, whereafter it induces cytochrome c release from isolated mitochondria. A role for Bid is further supported by the observation that cytosolic extracts from Bid-deficient cells show minor cytochrome c release in response to lysosomal extracts (52). Alternatively, Cathepsin B can also cause caspase-independent cell death with morphological apoptosis characteristics. For instance, cathepsin B has been shown to function as the dominantexecution protease in death receptor-triggered apoptosis in WEHI-S, which occurs independently of effector caspase
Bcl-2 Bcl-xL Bcl-w Mcl-1
Bad
Noxa Bim
Puma
Bak Bax Bid
cytochrome c
Casp-9
Apaf-1
Casp-3 Casp-8
IAPs
SMAC/DIABLO HtrA2/Omi death receptor
ligand
DISC
Bcl-2 Bcl-xL Bcl-w Mcl-1
Bad
Noxa Bim
Puma
Bak Bax Bid
cytochrome c
Casp-9
Apaf-1
Casp-3 Casp-8
IAPs
SMAC/DIABLO HtrA2/Omi death receptor
ligand
DISC
Figure 2. Regulation of apoptosis.
Triggering of death receptors results in assembly of the DISC which result in activation of executioner caspases (e.g. caspase 8) that in turn directly activate effector caspases (e.g. caspase-3). Release of cytochrome c is mainly regulated by the Bcl-2 family of proteins, consisting of pro- and anti-apoptotic family members (illustration adopted from ref 26). Cytochrome c together with Apaf-1 activates the executioner caspase-9 that in turn activates effector caspases. The released mitochondrial proteins Smac/
DIABLO and HtrA2/Omi antagonize that inhibitors of apoptosis (IAPs). There is cross-talk between the two pathways as caspase-8 can activate Bid, which facilitates cytochrome c release.
activation (53). Besides a role in apoptosis, as reviewed by Stoka et al. (54), these lysosomal proteases play also important roles in non-apoptotic types of cell death (55).
Another class of cysteine proteases is the calpains. Since these cytosolic proteases are activated by Ca2+, they are linked to ER stress. A role for calpains in apoptosis is shown in a study using calpain-deficient MEFs. These cells display a defective caspase-12, -9 and -3 activation upon ER stress, and are indeed resistant to ER stress-induced apoptosis (56). Biochemical studies show that calpain is involved in the cleavage and activation of caspase-12 (57). Once active, caspase-12 can directly activate caspase-9 (58). Caspase-12 knockout mouse cells are indeed resistant to ER stress-induced apoptosis (59). However, contrary to those observations in rodents, human caspase-12 have no significant effect on apoptotic sensitivity to diverse stimuli, including agents that provoke apoptosis through ER stress (60). As reviewed by Vandenabeele et al., calpains can also cleave pro- and anti- apoptotic proteins of the Bcl-2 family, and have been described to induce the release of the apoptosis inducing factor (AIF) from the mitochondria in order to promote apoptosis (61).
Combined, all these experimental systems show that different proteases can regulate the demise of a cell via the induction of apoptosis, which allows a rapid and tidy disappearance of unwanted cells. Nevertheless, several other cell death programs have been described that may also regulate cellular degeneration.
Autophagy
Autophagy is defined as a process in which proteins and organelles are degraded by lysosomal proteases. Morphologically, this process is characterized by the formation of autophagosomes - double-membrane vesicles containing cytosolic organelles - that fuse with lysosomes (illustrated in Figure 1). As reviewed by Klionsky and Emr, autophagy is not only important for protein degradation and organelle turnover, it also provides an alternative source of nutrients (62). In yeast, autophagy is induced under nutrient- limiting conditions as a mechanism to survive. However, in Drosophila melanogaster, the same autophagic structures are formed during morphogenesis, suggesting a role for autophagy in cell death (63). It has therefore been considered that autophagy might start as an adaptive response in order to enhance survival, but can result in cell death beyond a certain threshold.
Genetic studies in yeast have identified more than twenty genes involved in autophagosome formation and induction of autophagy. Based on homology, several human autophagy- associated Atg proteins have been identified, involved in two conjugation systems: Atg12 and Atg8/LC3 (62), as illustrated in Figure 3. Both in yeast and mammalian cells, Atg12 and Atg8/LC3 are activated by Atg7 and conjugate to Atg5 or phosphatidylethanolamine (PE) respectively (64,65). The two systems are related, since the Atg12-Atg5 complex is required for Atg8/LC3 targeting to the vesicle membranes (66,67). Studies in yeast show that the kinase TOR lies upstream of all autophagy-associated genes. It thus follows that TOR plays an important role in the initiation of autophagy (68). Such an important role has also been shown for mammalian TOR (mTOR); inhibition of mTOR enhances autophagy, whilst autophagy is suppressed by Akt, an activator of mTOR (69,70). Another initial step is
activation of a class III phosphatidylinositol 3-kinase, shown to depend on the formation of a complex in which the Atg6/Beclin1 protein is involved (71).
Mitotic catastrophe
Mitotic catastrophe is defined as a type of cell death that is caused by aberrant mitosis. It is originally described in yeast, where cells die as a result of aberrant chromosome segregation (72). In mammalian cells, mitotic catastrophe is often associated with deficiencies in cell cycle checkpoints. These checkpoints regulate transition from one phase of the cell cycle to the next and only allow cells to proceed when all processes are completed. Particularly in tumor cells, mitotic catastrophe is associated with the inability of G2/M checkpoints to arrest cell cycle progression upon (treatment-induced) DNA damage (73,74). To recognize the occurrence of mitotic catastrophe, both morphological characteristics (such as enlarged and multinucleated cells) as well as the presence of mitotic defects (such as incomplete
Atg6 PI3KC3
Atg12
Atg7
Atg12 Atg7
Atg12 Atg10
Atg12 Atg5
Atg12 Atg5
Atg16
Atg8
Atg4 Atg8
Atg7
Atg8 Atg7
Atg8
Atg3 Atg8 PE
mTOR
lysosome lysosome
Akt GF
Figure 3. Regulation of autophagy.
Starvation triggers autophagy by modifying mTOR signaling, which represses autophagy under growth factor (GF) conditions. Atg proteins are involved in autophagosome formation, which is positively regulated by Atg6 and class III PI3K. This autophagosome formation involved two conjugation pathways. Atg12,-5 and -16 are physically associated with the membrane, whilst Atg8/LC3 is conjugated to the PE that is inserted in the membrane. The autophagosome fuses with a lysosome in order to degrade its content.
nuclear condensation, chromosome alignment defects, unequal DNA separation or mitosis in the presence of DNA damage) are used (73,75-78) (illustrated in Figure 1).
Since the G2/M checkpoint is responsible for blocking mitosis in case of damaged DNA, altered expression of proteins involved in this checkpoint are likely associated with mitotic catastrophe. Several studies indeed show a role for G2/M regulatory proteins; mitotic catastrophe is enhanced by high expression levels of proteins that promote entry of mitosis (such as Cdk1 and cyclinB) as well as by inhibition or knockout of proteins that prevent premature mitosis (such as ATR, ATM, Chk1, Chk2, Plk and 14-3-3σ) (79-83). Since p53 induces a G2-arrest upon DNA damage via transcriptional activation of the Cdk-inhibitor p21, both p53 and p21 might play a role in preventing mitotic catastrophe as well (84).
Next to defects in the G2/M checkpoint, defective mitotic spindle checkpoints have been linked to mitotic catastrophe as well. Such defects can lead to missegregation of chromosomes. Adequate functioning of the mitotic spindle depends on proteins involved in the spindle formation, such as Mad and Bub, and on chromosomal passenger proteins, such as Survivin and Aurora kinases (85-88). Conditional deletion of the survivin gene indeed leads to disorganized mitotic spindles in early passage cells, and these cells finally die with morphological characteristics of mitotic catastrophe (89). Also drugs that affect the mitotic spindle as well as specific downregulation of spindle checkpoint proteins or inhibition of aurora B kinase activity can result in aberrant mitosis (90,91).
All these experiments show that the cell cycle checkpoints play an important role in preventing aberrant mitosis. However, the exact molecular pathways are largely unknown and molecular markers have not been defined. It is therefore still a matter of debate whether mitotic catastrophe is a specific death process or just functions as a trigger for apoptosis (73,74).
Necrosis
In contrast to the regulated nature of apoptosis, necrosis has been considered as an uncontrolled form of cell death. Morphologically, necrosis is characterized by vacuolization of the cytoplasm, loss of membrane integrity and cellular swelling followed by the release of intracellular components which can provoke an inflammatory response (illustrated in Figure 1). Usually, necrosis is a consequence of pathological traumas, such as infection or ischemia. Treatment of cells with TNFα or Fas ligand can also induce necrosis via their respective death receptors, especially when apoptosis is inhibited by zVAD (92,93). This latter observation points to the fact that necrosis may not be such an uncontrolled form of cell death as initially suggested. Indeed, growing evidence supports the idea that necrosis can also be regulated. Death receptor-induced necrosis might depend on the kinase RIP1 (receptor-interacting protein 1); downregulation of RIP1 as well as RIP1-deficient Jurkat cells show partial resistance to Fas-induced cell death (94). RIP1 likely targets the mitochondria in order to induce excess formation of reactive oxygen species (ROS), as reviewed by Festjens et al. (95). ROS are considered to play a central role in necrosis, since the ROS scavengers efficiently prevent necrosis induced by several treatments (96-98).
Besides death-receptor-triggered necrosis, DNA damage (e.g. MNNG-induced) can also result in necrosis. This necrotic death is mediated by PARP-1, a protein involved in DNA damage repair. Overactivation of this enzyme results in a drop of cellular NAD+ and ATP, suggesting a link to mitochondrial functioning (99,100). In agreement with this hypothesis, PARP-1 activation has been shown to induce a translocation of AIF from the mitochondria to the nucleus. This translocation mediates a caspase-independent death in cells treated with DNA damaging agents (101). In addition, it has been shown that PARP-1-mediated necrosis depends on the proteins RIP1, TRAF2 and JNK1 (102,103).
Combined, these observations indicate that necrosis should no longer be exclusively viewed as an unregulated process. A regulated form of necrosis - also called necrosis-like programmed cell death – might also be considered as a different type of cell death, besides accidental necrosis (104).
Crosstalk apoptosis and other death pathways
When apoptosis is blocked in experimental settings, cells can become more sensitive for other types of death. It thus follows that the death pathways are somehow interconnected (illustrated in Figure 4). As reviewed by Baehrecke and by Maiuri et al., various studies indicate a connection between apoptotic and autophagic cell death pathways (105,106).
In support of the notion that inhibition of apoptosis can sensitize for autophagy, TNF- triggered L929 cells die through autophagy when caspase-8 is inhibited (107). Also apoptosis-resistant Bax/Bak double knockout cells die through autophagy when treated with apoptosis-inducing agents (108). However, the opposite has been described as well, since nutrient-starvation induces apoptosis in Hela cells when autophagy is genetically or pharmacologically inhibited (109). Glioma cells, that normally die through autophagy upon treatment with temozolomide, die through apoptosis when autophagy is inhibited by bafilomycin A1 (110). It has been proposed that autophagy can delay apoptotic cell death via the removal of damaged mitochondria (111,112). An incompatible link between autophagy and apoptosis is however not very likely, since both processes can occur simultaneously in dying cells. In Drosophila melanogaster for example, autophagosomes appear after caspase activation in dying salivary glands (113). Also in mammalian MCF10-A cells, TRAIL can induce apoptosis and autophagy in parallel (114).
Several studies describe a specific protein which might function at the interplay of apoptosis and autophagy. The autophagic protein Atg6/Beclin1, for example, can bind the anti-apoptotic Bcl-2 (115). The observation that this binding prevents autophagy upon starvation indicates that Bcl-2 might block both apoptosis and autophagy (116). In addition, the autophagy-associated Atg5 protein can induce cytochrome c release from isolated mitochondria after calpain-mediated cleavage, indicating a role for Atg5 in the induction of apoptosis (117). Furthermore, p53 can activate the transcription of the protein DRAM (damage-regulated autophagy modulator), which is suggested to induce both apoptosis and autophagy to promote a full cell death response (118).
In the case of mitotic catastrophe, it is still a matter of debate whether this death is conducted by apoptosis. Since mitotic catastrophe is often accompanied by apoptosis
characteristics, such as mitochondrial membrane permeabilization and caspase activation, it could be merely a stress signal that leads to apoptosis (77,85,91,119). However, inhibition of apoptosis does not necessarily inhibit mitotic catastrophe-related death. One study using the drug combretastain-A4 to induce death with typical morphology of mitotic catastrophe (giant, multinucleated cells) shows that this death could not be blocked by the pan-caspase inhibitor zVAD nor by a specific caspase-9 inhibitor (120). Also when apoptosis is suppressed by Bcl-2, cells can die through mitotic catastrophe alone (73). Hence, it has been proposed that mitotic catastrophe does not depend on apoptosis per se, but might be followed by apoptosis in apoptosis-competent cells (121,122). Besides, the extent of damage may also influence the mode of death, since low doses of doxorubicin induce cell death that is accompanied by characteristics of mitotic catastrophe, whereas high doses induce death through apoptosis (75).
At the protein level, Castedo et al. reviewed several possible links between apoptosis and mitotic catastrophe; transcription of pro-apoptotic p53 target genes, like Bax and PUMA, during p53-dependent mitotic catastrophe, as well as activation of caspase-2 may induce apoptotic cell death via the mitochondrial pathway (74). The role of Survivin during mitosis and the suggestion that it can act as apoptosis inhibitor (123) provide another link (88,124).
Apoptosis and necrosis are likely incompatible types of cell death, since several experimental studies show that necrosis and apoptosis cannot occur simultaneously.
Treatment of L929sAFas cells with interferon and dsRNA only results in apoptosis when necrosis is suppressed by ROS scavengers (125). Reciprocally, TNFα and Fas ligand induce necrosis provided that caspases are inhibited (92,93). Also apoptosis-resistant melanoma
Mitotic catastrophe
Apoptosis
Necrosis Autophagy
trigger
Mitotic catastrophe
Apoptosis
Necrosis Autophagy
trigger
Figure 4. Relation apoptosis and non-apoptotic cell deaths
The relation between apoptosis and autophagy is complex: these pathways exhibit some degree of mutual inhibition, but a mixed phenotype can be detected as well. In case of mitotic catastrophe and apoptosis, it is highly debated whether these are distinct types of death, or whether mitotic catastrophe functions as a stress signal that triggers apoptosis. Apoptosis and necrosis negatively regulate each other: caspase activity can be prevented by the necrosis-associated low levels of ATP, whilst caspases can cleave proteins involved in necrosis.
cells and Jurkat cells deficient in FADD or caspase-8 are more susceptible to necrosis (94,125-127). It thus seems that caspases – activated during apoptosis – can negatively regulate the necrotic pathway. The RIP1 kinase has been proposed to play a dominant role in a cell’s decision to die through apoptosis or necrosis upon death receptor triggering (95). It has been shown that caspase-8 can cleave RIP1 and overexpression of the cleaved fragment increases TNF-induced apoptosis (128). Conversely, a non-cleavable form of RIP1 restores the induction of necrosis in RIP1-deficient Jurkat cells upon TNF treatment (129), suggesting that RIP1 can function as a switch. Also DNA damage-triggered necrosis is inhibited by apoptosis, through caspase-mediated cleavage of PARP1. On the other hand, low amounts of intracellular ATP have been shown to switch cell death from apoptosis to necrosis, which is likely due to preventing executioner caspase activation for which ATP is required (130).
Together, these observations suggest that necrosis predominantly occurs when apoptosis is inhibited, which may indicate a preference for apoptotic cell death. Whether this also holds for the more regulated pathways to necrosis remains to be determined.
Taken together, the above described data indicates that inhibition of apoptosis does not necessarily prevent overall cell death. In experimental settings, it can actually sensitize cells for other modes of death. Since resistance towards apoptosis is considered as one of the fundamental hallmarks of tumors, it follows that non-apoptotic types of death might play a role in (treatment-induced) tumor cell death in vivo.
The Role of Apoptosis and Non-Apoptotic Death Pathways in Tumorigenesis
The role of apoptosis in tumorigenesis
Since tumor cells have to survive under stressful conditions, such as limited supplies of growth factors and oxygen and the host immune attacks, cells are expected to acquire several modifications that provide resistance to death stimuli. The frequent mutations leading to a loss of function of the tumor suppressor protein p53 can directly be linked to a failure to induce apoptosis after cellular stress. In agreement, mice deficient for p53 are highly prone to develop tumors (131-133). Another anti-apoptotic modification observed in human tumors is high expression of Bcl-2. In a subtype of B-cell lymphomas, Bcl-2 is overexpressed as a consequence of a bcl-2 gene translocation next to an immunoglobulin gene (134). This translocation indeed increases the incidence of spontaneous B-cell tumors in mice (135). In human tumors, Bcl-2 overexpression is often found together with high expression of the oncogene c-myc, shown to enhance tumorigenesis in mice (136). Loss of functional pro-apoptotic Bcl-2 family members, Bax and Bak, can also be found in several human tumors (137,138). Furthermore, modifications in the death receptor pathways may play a role in the apoptosis resistance of human cancers. For example, the Fas receptor expression is high in normal colon mucosa but is reduced or even lost in colon carcinomas (139). Absence of Fas allows tumor cells to evade the immune destruction mediated by cytotoxic lymphocytes via this pathway. In addition, c-FLIP is specially overexpressed
in colon cancers, which has been shown to protect tumor cells against cytotoxic T cell- induced apoptosis in vivo (140,141). To conclude, apoptosis-resistance is the common outcome of all the different anti-apoptotic modifications found in human tumors (2). It follows that apoptosis is an important barrier that must be circumvented by tumor cells in order to survive and proliferate, and that resistance to apoptosis plays an important role in tumorigenesis.
The prognostic value of apoptosis in colorectal cancer
Since resistance to cell death is an inherent property of a tumor cell, it follows that there is a correlation with tumor cell survival. In order to evaluate whether apoptosis has indeed clinical relevance, the sensitivity towards apoptosis must be screened in human tumor tissues. Several studies on colorectal cancer have scored the expression of a single apoptosis-associated protein, such as p53, Bcl-2 and Bax, and attempt to correlate this with prognosis. As reviewed by Brown and Wilson, these studies present conflicting data; some show significant correlations with good or poor prognosis, whereas others describe no significant associations (142). A limitation in scoring a single protein is that the expression of this single protein may not reflect the level of apoptosis as apoptosis is regulated by a complex network of proteins. Indeed, the expression of Bcl-2 or p53 do not often correlate with the number of apoptotic cells (143-145). Therefore, sensitivity towards apoptosis might be more adequately determined by evaluating the exact number of apoptotic cells in the tumor tissue. Apoptotic cells can be identified by a TUNEL assay (measures DNA fragmentation) or by staining with the M30 antibody (recognizes caspase-cleaved cytokeratin-18). Indeed, a review of studies on prognostic markers in rectal carcinoma by Smith et al. shows that high amounts of TUNEL-positive cells in pretreated biopsies corresponds with more tumor regression upon pre-operative radiochemotherapy (146).
A correlation between apoptosis and tumor cell survival implies that resistance to apoptosis will be correlated with a poor prognosis for colorectal cancer patients. The results of studies that evaluated this correlation are summarized in Table 1. For rectal cancer patients, although no effect of apoptosis on survival is found, a positive effect on local control for patients treated with surgery only is repeatedly observed (145,147,148).
One study shows that the level of apoptosis in pre-treated biopsies is also associated with less overall recurrence development after receiving radiochemotherapy (149), but this observation is not supported by other publications (144,150). Thus, the prevalent finding in rectal cancer patients is a positive correlation with higher spontaneous apoptosis favoring less local recurrence development. The development of distant recurrences and survival are however not affected by the level of apoptosis in the primary tumor. For colon cancer patients, contradictory data have been reported. Whereas some studies found a correlation between high levels of apoptosis and worse survival, others show no correlation or a correlation with worse survival (145,151-157). This lack of consistent correlations with survival for both rectal and colon cancer patients might be due to the fact that survival depends on distant metastases rather than local control.
Two studies evaluated whether resistance to apoptosis influences the metastatic capacity of the tumor cells. However, both lower and higher levels of apoptosis in metastatic compared with primary tumors are reported (158,159). These conflicting data might be the result of differences in specificity of TUNEL assays and M30 stainings (160). Nevertheless, most colorectal tumor cells have acquired anti-apoptotic modification during the development of the primary tumor, such as loss of functional p53. The fact that no clear correlations with tumor stage and metastasizing capacity have been reported, suggests that the initially acquired apoptosis resistance may be sufficient for the cells to survive these later stages of tumor development.
The role of autophagy in tumorigenesis
The double-membrane bound vesicles that are typical for autophagy can be observed in several types of human tumors, and indicate that autophagy occurs in tumors in vivo (161).
Several findings support the hypothesis that autophagy plays a tumor suppressive role in the early stages of tumorigenesis. Mice with mutations in genes that result in decreased autophagy, such as Beclin1/- and Atg4C-/-, are more prone to develop tumors (162,163).
Overexpression of Beclin1 in MCF7 cells promotes autophagy, and lowers the incidence of tumor formation when injected in nude mice (115). In human tumors, the notion of a tumor suppressive role is supported by the observation that pro-autophagic proteins, as Beclin1 and DRAM, are found to be lowly expressed in several tumor types (115,118).
Table 1. Prognostic value of apoptosis for (colo)rectal cancer patients
1st author Patients treatment Outcome
Tannapfel 32 rectal neoadjuvant CRT no correlation with recurrences = Schwandner 160 rectal adjuvant CRT for
TNM II - III
no correlation with recurrences and survival
=
Adell 162 rectal randomized for
neoadjuvant RT
high apoptosis less local recurrences, not survival
+
Rodel 44 rectal neoadjuvant CRT high apoptosis less recurrences +
Hilska 124 rectal no correlation with survival =
de Bruin 1198 rectal randomized for
neoadjuvant RT
high apoptosis less local recurrences, not survival
+
Langlois 74 colorectal high apoptosis better survival +
Sinicrope 64 proximal colon no correlation with survival =
82 distal colon high apoptosis better survival +
Michael-Robinson 100 colorectal no correlation with survival =
Elkablawy 53 colorectal no correlation with survival =
Rupa 278 colorectal high apoptosis worse survival -
Noguchi 80 colorectal no correlation with survival =
Bendardaf 49 colorectal high apoptosis worse survival -
Hilska 239 colon high apoptosis worse survival -
in case of neoadjuvant chemoradiotherapy (CRT), apoptosis was determined in pre-treated biopsies
Furthermore, components of the Ras and PI3K signaling pathway are often mutated. Since the PI3K-mTOR pathway plays a role in the autophagic process, the observed mutations in this pathway may also point to an inhibition of autophagy in tumors (164,165). These anti-autophagic mutations found in human tumors indicate that inhibition of autophagy possibly may provide a selective growth advantage for tumor cells.
Several mechanisms are proposed by which autophagy can prevent tumorigenesis.
Autophagy may reduce metabolic stress by degrading damaged organelles, or degrade specific proteins that induce or enhance tumor formation (166,167). In this way, autophagy can prevent a normal cell to become a malignant cell. Furthermore, since monoallelic loss of beclin1 is found to be associated with chromosome gains and losses, autophagy may limit chromosome instability and thereby limit tumor progression (168). Alternatively, autophagy may kill developing premalignant cells and thereby prevent tumorigenesis.
Although autophagy is potentially suppressive during the early stages of tumorigenesis, it seems to play a tumor-promoting role during later stages of tumor growth, since it provides a protective mechanism against stressful conditions. In this respect, growth factor-dependent cells from Bax/Bak-deficient mice activate autophagy upon growth factor withdrawal, which enables them to survive for several weeks. These cells die when autophagy is inhibited, indicating that autophagy functions as a survival mechanisms (169). In agreement, activation of p53 in myc-induced lymphoma reveals that most cells die through apoptosis, but that autophagy was occurring in the surviving cells. Inhibition of autophagy in this setting enhances tumor cell apoptosis and tumor regression, confirming a cytoprotective role of autophagy in established tumors (170).
Although the above mentioned anti-autophagic mutations observed in vivo indicate a tumor suppressive role for autophagy, its definitive role remains unclear. The observation that the beclin1 gene is just monoallelically lost suggests that a certain level of Beclin1 expression is required for tumor cell survival. In addition, a recent mutational screen of the beclin1 gene in more than 500 primary tumors of different origin found point mutations in only 2%
of the tumors. Such a rare event seems unlikely to play a major role in tumorigenesis (171).
Despite some connections between autophagy and tumorigenesis, lack of good markers to easily detect autophagic cells in vivo limit investigations in human tumor tissues at present.
The role of mitotic catastrophe in tumorigenesis
Mitotic catastrophe is a form of cell death that results from abnormal mitosis. Cells that survive abnormal mitosis can potentially divide asymmetrically, leading to aneuploid cells which are generally more tumorigenic. Also cells that go into mitosis with damaged DNA are more likely to acquire tumorigenic capacity as these cells are genetically unstable. This is illustrated in patients with hereditary defects in DNA mismatch repair genes, who are predisposed to develop colon cancer.
Several studies provide evidence that mitotic catastrophe can play a tumor suppressive role. Experiments using Hela cells in which apoptosis is blocked show that radiation- induced mitotic catastrophe indeed leads to a reduced clonogenic outgrowth of the cells (73). Induction of mitotic catastrophe also reduces the clonogenicity of p53-deficient MEFs
that were genetically modified in order to prevent a DNA damage-induced cell cycle arrest (77). In these experiments, mitotic catastrophe is however induced by DNA damaging treatments. It is therefore difficult to conclude that mitotic catastrophe might prevent tumorigenesis. Nevertheless, studies on the Survivin protein provide some evidence for such a role. When the expression of Survivin is blocked in MEFs, these cells show disorganized mitotic spindles, are no longer able to proliferate and finally die upon mitotic catastrophe (89). In human myeloma cells, knockdown of the survivin gene decreases the overall growth of the cells. Interestingly, Survivin expression strongly correlates with the stage of disease for patients with multiple myeloma; the highest expression is found in patients with a tumor relapse (172). Furthermore, while Survivin is rarely detected in normal tissues, it is expressed in almost all human tumors (173). Failure of cells to undergo mitotic catastrophe might thus play a role in tumor development.
The prognostic value of mitotic catastrophe in colorectal cancer
Some findings indicate that failure in mitotic catastrophe is associated with a poor prognosis for colorectal cancer patients. High expression of the polo-like kinase 1 (Plk1) protein is found to be associated with worse prognosis (174). In vitro studies show that a constitutively active mutant of this protein can override growth arrest induced by DNA damage, resulting in aberrant mitosis (175). High expression of another protein involved in entering mitosis, the CDC25B phosphatase, is also associated with poor prognosis for colorectal cancer patients (176). The association between aberrant mitosis and a poor prognosis might be due to the inability of the tumor cells to induce mitotic catastrophe- associated death. However, such aberrant mitosis does not necessarily point to a prognostic value of mitotic catastrophe. Aberrant mitosis can also be related to genetic instability and/or a higher proliferation rate of the tumor cells, which both can negatively influence patients’ prognosis.
High Survivin expression correlates with shorter survival for colorectal cancer patients after curative resection (142,177,178). Survivin indeed regulates mitotic progression and might suppress mitotic catastrophe, since deletion of the gene results in mitotic catastrophe (89). However, this protein has been originally identified as a member of the Inhibitor of Apoptosis Proteins (IAPs) (179). Even though structural studies have shown that a direct inhibitory effect on caspases is unlikely (124), a role for Survivin as apoptosis inhibitor cannot be excluded. Moreover, the expression of Survivin inversely correlates with the level of apoptosis in colorectal tumors (177,180-182). Besides its effect on apoptosis, Survivin expression also correlates with tumor cell proliferation in colorectal tumor tissues (180). It is therefore difficult to dissociate the effects of Survivin on apoptosis and proliferation from its effect on mitosis and mitotic catastrophe. Due to lack of markers to distinguish mitotic catastrophe form other forms of cell death, research has been limited so far. Hence, the role of mitotic catastrophe in tumor development and patients’ prognosis is still unclear.
The role of necrosis in tumorigenesis
Since necrosis can induce an inflammatory response, it may decrease tumorigenesis by killing potentially malignant cells. As reviewed by Aggarwal et al., TNFα is originally isolated as an anticancer cytokine, able to kill tumor cells and to induce tumor regression in mice (183). However, when TNFα persists in the system for a long time, its effect becomes tumor promoting. In agreement, mice with impaired TNFα signaling, such as TNFα-/- and TNFR1-/- mice, are less prone to develop liver tumors in models of hepatocarcinogenesis, and are also more resistant to chemical induction of skin cancer (184,185). Other experimental evidence for a link between inflammation and cancer has recently been reviewed by Karin, who proposes an important role for the NF-κB group of transcription factors. These transcription factors induce transcription of several anti- apoptotic proteins, and also of other proteins that promote tumorigenesis (186). Hence, it is not clear whether necrosis plays a major role in tumorigenesis. On one hand, tumor cell necrosis can provoke an inflammatory response, and stimulate an immune response towards the tumor cells. On the other hand, chronic inflammation is thought to promote tumor development.
For colorectal cancer patients, chronic inflammation is indeed involved in tumor development. Patients with chronic inflammatory bowel diseases (IBD) have an increased risk of cancer development (187), and patients with the familial adenomatous polyposis (FAP) syndrome show a significant reduction in the number and size of colorectal adenomas upon treatment with the nonsteroidal anti-inflammatory drugs celecoxib (188,189). Since necrosis can lead to inflammation and a sustained inflammatory response can stimulate tumor development, these data provide some indirect evidence for a role of necrosis in tumor development. A more direct link has been described as well; colorectal cancer patients with more than 10% necrosis in their primary tumor had worse survival than patients with little necrosis (190). However, it remains unclear whether there is a causal relationship between necrosis and prognosis, and thus whether necrosis plays a major role in tumor suppression.
The Role of Apoptosis and Non-Apoptotic Death Pathways in Treatment Response
The role of apoptosis in treatment response
Anticancer treatments such as chemo- and radiotherapy aim to induce cell death. As reviewed by Fulda and Debatin, most anticancer strategies currently used in clinical oncology has been linked to activation of apoptosis signal transduction pathways in cancer cells (191). Thus, failure to undergo apoptosis may result in treatment resistance.
Indeed, in vitro experiments often show that anti-apoptotic modifications in tumor cells, such as Bcl-2 overexpression, suppress cell death induced by radiation or chemotoxic drugs. However, these anti-apoptotic modifications seem to predominantly affect cell death measured in short-term assays, but not always those measured in long-term survival
assays. This has been nicely illustrated in HCT116 colorectal cancer cells, in which Bcl-2 overexpression reduces the number of apoptotic cells upon radiation, while it has no affect on the clonogenic outgrowth of the cells (3). Since such a clonogenic survival assay integrates all forms of cell death for a longer time period, it follows that other types of cell death may be induced when apoptosis is blocked. Indeed, a compensatory increase in other forms of cell death has been observed in apoptosis-resistant cells (192,193).
These experimental data indicate that resistance towards apoptosis might not dramatically influence the overall treatment effect. However, effects observed by clonogenic survival of single cells do not always reflect the in vivo tumor response (2);
loss of functional p53 or overexpression of Bcl-2 in Eµ-myc lymphomas in mice result in resistance towards apoptosis, and this resistance correlates with a failure to respond to several chemotoxic drugs in vivo, although it does not affect the clonogenic survival (133,194). However, not all in vivo tumor models show this correlation between resistance to apoptosis and chemotherapeutic drugs. Established HCT116 tumors, for example, show hardly any difference in treatment response between cells overexpressing Bcl-2 or not, while these cells clearly differ in apoptosis-resistance (3). Hence, the effect of apoptosis resistance to the overall treatment effect is likely to depend on the specific tumor type. It has been suggested that resistance towards apoptosis might be more important for therapy-resistance of hematopoietic malignancies rather than solid tumors of epithelial origin (3).
The role of apoptosis in treatment response of colorectal cancer patients
As recently reviewed by Smith et al., the level of spontaneous apoptosis correlates with treatment response in six out of seven clinical studies on (colo)rectal cancer patients;
tumors with high levels of spontaneous apoptosis in pretreatment biopsies show more tumor regression upon pre-operative chemoradiotherapy (146). This indicates that high levels of spontaneous apoptosis may be related to sensitivity towards anti-cancer therapies.
Several anticancer treatments have been shown to trigger apoptosis in tumor tissues.
Treatment with 5-fluorouracil as well as radiotherapy increases the amount of apoptotic cells in colorectal tumors (147,195). However, the clinical outcome of the patients appears not to depend largely on the level of this induced apoptosis. For rectal cancer patients, is has been shown that radiotherapy reduces local recurrence rates irrespective of the level of apoptosis (147,148). In agreement with this finding in rectal cancer patients, two reviews describe a modest role for apoptosis in response to radiotherapy in several other solid tumor types (3,196). Apparently the success of treatment does not solely depend on the induction of apoptosis. Exploring non-apoptotic types of cell death might therefore provide new opportunities for a more effective anti-cancer approach.
The role of autophagy in treatment response
Autophagy can be induced by radiation and chemotherapy in tumor cells (73,197-200).
However, seemingly conflicting data have been published as to whether autophagy exerts a positive or negative effect on treatment response. As mentioned before, autophagy
might function as survival mechanism, protecting cells from treatment-induced death.
In agreement with this hypothesis, inhibition of autophagy by specific drugs enhances the radiosensitivity of malignant glioma cells in an experimental setting (200). Also in myc-induced lymphoma in mice, it has been shown that inhibition of autophagy enhances p53-induced apoptosis, increases tumor regression and results in delayed tumor outgrowth (170). Conversely, induction of autophagy by rapamycin protects various tumor cell lines against the induction of apoptosis (111,112). Autophagy can thus protect cells from death upon treatment-induced damage. In this case, inhibition of autophagy as adjuvant treatment would then enhance the effect of apoptosis-inducing anti-cancer therapies for cancer patients.
Seemingly conflicting observations are made with cells that have impaired apoptosis, by using caspase inhibitors or Bax/Bak double knockout cells. These cells show more autophagy and are more sensitive to radiation than wild-type cells (201-203). This radiosensitivity was determined using clonogenic survival assays, measuring clonogenic outgrowth of cells. Such assays are, however, not only affected by cell death but also by a delay or stop in proliferation;
effects suggested to occur upon induction of autophagy (161). Autophagy can thus still be a survival mechanism, which reduces but preserves the cells’ clonogenic potential for a period of time. Another possible explanation for these contradictory findings could be that autophagy beyond a certain threshold will result in cell death. If this is the case, induction of autophagy as adjuvant treatment could enhance the effect of anti-cancer therapies. In line with this idea is the observation that clinical application of mTOR inhibitors result in a prolonged survival of patients with metastatic renal cancer or breast cancer (204-206).
Obviously, these data are not fully conclusive regarding the role of autophagy in treatment response, since the anti-tumor effects of inhibiting mTor could reflect its role in cell cycle regulation or translation as well. It is thus still unclear whether treatment-induced autophagy functions as a cell death mechanism or as a mechanism by which tumor cells try to survive.
The role of mitotic catastrophe in treatment response
The observation that tumor cells are frequently deficient in cell-cycle checkpoints implies that particularly tumor cells can be susceptible to the induction of mitotic catastrophe (207,208). Especially DNA-damaging agents may induce mitotic catastrophe, since a lack of proper checkpoint controls means that the cells will enter mitosis without an arrest that allows for DNA repair. Mitotic catastrophe has indeed been pointed out as an important form of cell death induced by radiation in solid tumors (3,73,196,209). This notion is supported by several studies. Changing the sensitivity towards apoptosis does not always affect the sensitivity of tumor cells or tissues to radiation or other anti-cancer drugs.
Instead, overall cell death can be compensated by an increase in mitotic catastrophe, as shown in experimental settings (73,192,193). In addition, twelve out of fourteen solid- tumor cell lines display mitotic catastrophe when treated with the DNA damaging drug doxorubicin, whereas only two lines die through apoptosis (73). A recent study shows that the induction of mitotic catastrophe can play a role in tumor regression in vivo. When
