Tumor Cell-protective Catalase as a Novel Target for Rational Therapeutic Approaches Based on Specific Intercellular ROS Signaling

Multistep Oncogenesis

Multistep oncogenesis is characterized by multiple distinct steps, including abrogation of senescence control, oncogene activation, tumor suppressor gene inactivation (1-3), independence of exogenous proliferation signals through autocrine mechanisms (4-6), independence of control by neighboring cells (7-11), escape from immune surveillance (2), release of prostaglandin E2 (12-16), resistance to hypoxia-induced p53-mediated cell death (17, 18), tumor angiogenesis (19) and others (reviewed in 20-24). These basic features of oncogenesis are interconnected with three additional central mechanisms related to the action of reactive oxygen (ROS) and nitrogen species (RNS): i) oncogene-controlled extracellular superoxide anion production through NADPH oxidase (NOX1) (25-42; for review see 43-49), ii) acquisition of the ‘H₂O₂-catabolizing phenotype’ during tumor progression (12-16), and iii) resistance of tumor cells to intercellular ROS signaling through membrane-associated catalase expression (50-54).

ROS and Oncogenesis

Generation of extracellular superoxide anions through a membrane-associated NADPH oxidase (NOX1) is associated with oncogene activation and represents one of the hallmarks of the transformed state (25-42, 55, for review see 43-49). The rat sarcoma oncogene (RAS) and and the RAS-related small GTPase RAC1 play central roles for the activation of NOX1 (25, 56). Activated NADPH oxidase seems to be required for the control of proliferation and the maintenance of the transformed state (25-27, 29, 37, 57), changes in the cytoskeleton of transformed cells (58) and induction of cell motility (59). A causal connection between oncogene activation, superoxide anion production and transformation, as shown convincingly in vitro, has also been demonstrated to be relevant for the situation in vivo (37, 39, 40, 42, 60). Specific overexpression of NOX1 has been found in human tumors (39, 60), and is dependent on activation of RAS (39). Inhibition of NOX1 activity causes subsequent inhibition of tumor growth, as shown by the work of Mitsushita et al. (37). Overexpression of RAC1 in oral squamous cell carcinomas (61) and RAC1 gene mutations in human brain tumors (62) also point to the significance of the RAS-RAC1-NOX1 network in tumor development. In addition, ROS have been shown to be relevant for tumor angiogenesis (63, 64) and metastasis (65, 66). Together with induction of genomic instability (67), the effects of ROS on angiogenesis and metastasis thus contribute to tumor progression in a highly dynamic way, beyond the establishment and the control of the transformed state. On the flip side of the coin, extracellular superoxide anions generated by transformed cells, drive both the efficiency and selectivity of intercellular induction of apoptosis specifically in malignant cells, a hitherto unrecognized potential control step during multistage oncogenesis (30, 32, 33, 35, 68-79, for review see 21-24). During intercellular induction of apoptosis, transformed cells are selectively induced to die by apoptosis after a concerted action of transformed cell-derived ROS and signaling components released by surrounding non-transformed cells (classical intercellular induction of apoptotis), or the transformed cells themselves (autocrine, ROS-mediated apoptotic self-destruction) (21-24, 30, 32, 52-54). For simplicity, the comprehensive term ‘intercellular ROS signaling’ is used for both aspects of ROS-mediated apoptosis induction, as they utilize the same basic signaling chemistry. Besides its potential role for the understanding of ROS-related processes in oncogenesis, intercellular ROS signaling represents a rather unique experimental system for the study of site-specific signaling effects of radical and nonradical ROS and RNS (80). The perception of ROS and RNS is often restricted to shotgun-like nonspecific damaging agents, although their specific signaling potential (in addition to their destructive potential) has been proposed many years ago (81). Specific apoptotic ROS signaling, as discussed here, depends on a fine interplay between different radical and nonradical species, with short or long free diffusion path length, whereby the membrane of malignant cells is both the origin and the target of some of the relevant molecules. Four pathways involved in intercellular ROS-mediated signaling have been elucidated so far: i) the HOCl signaling pathway (30); ii) the NO/peroxynitrite signaling pathway (30, 35, 82); iii) the nitryl chloride signaling pathway (83); and iv) the metal ion catalyzed Haber-Weiss reaction (84). The selectivity of the ‘target cell function’ of transformed cells is based on their specific extracellular superoxide anion production by NOX1. ‘Effector function’ in this system of intercellular ROS-mediated induction of apoptosis is based on the release of peroxidase and/or nitric oxide (NO). This effector function can be exerted by non-transformed as well as by transformed cells, leading either to classical intercellular induction between non-transformed and transformed cells, or autocrine ROS-mediated apoptosis induction within the population of transformed cells. Efficient apoptosis induction in this system requires a sufficient density of the target cells (to ensure optimal dismutation of superoxide anions to hydrogen peroxide) as well as a sufficient effector cell number in order to reach an optimal overall concentration of effector molecules. The combination of these two requirements represents the prerequisite for successful intercellular ROS-mediated induction of apoptosis in vitro and may also be relevant in vivo. Whereas cells transformed in vitro are regularly and specifically sensitive to intercellular induction of apoptosis and autocrine self-destruction, independently of the origin of tissue and the transforming principle, bona fide tumor cells established from tumors are resistant to intercellular induction of apoptosis (50-54). Therefore, ROS-mediated apoptotic signaling in transformed cells has been discussed as a potential early control step during oncogenesis (21). Mathematical modeling supports this assumption (85), demonstrating that elimination of malignant cells through ROS-mediated induction of apoptosis should be able to reduce the pool of transformed cells despite their ROS-driven proliferation. Malignant cells may, however, escape this control and form tumors as soon as they acquire resistance through expression of membrane-associated catalase. Tumor cells, like cells transformed in vitro, exhibit marked NOX1-dependent superoxide anion generation (28, 47, 52-55). The resistance of tumor cells to intercellular ROS signaling depends on the expression of membrane-associated catalase on the outer surface of the tumor cells (52-54). Catalase-mediated protection of tumor cells from intercellular ROS signaling was found in all human and rodent tumor cell lines studies so far (more than 70 human tumor cell lines have been tested; Bauer, unpublished finding). The specific location of tumor cell-protective catalase on the cell surface was proven through i) staining of live tumor cells (but not of non-transformed cells) with antibody towards catalase and competition of free catalase to this staining; ii) re-establishment of intercellular ROS signaling after binding of neutralizing antibodies directed towards catalase; iii) abrogation of resistance after intensive trypsinization (54). After trypsinization, the antibody against catalase exhibited no additional effect, indicating that the same target had been attacked by trypsinization and the antibody to catalase (54). Finally, tumor cells were protected against extracellular peroxynitrite and this protection was counteracted by catalase inhibition, indicating that catalase must be active on the outside of the cell, allowing peroxynitrite to be neutralized before it reaches the cell membrane (54). In contrast, non-transformed cells were not protected against apoptosis induction by exogenous peroxynitrite. The degree of their sensitivity for peroxynitrite was not enhanced when their intracellular catalase was inhibited by a cell-permeable inhibitor, as intracellular catalase cannot prevent the reaction of peroxynitrite with the cell membrane when peroxynitrite approaches the membrane from outside the cell. Catalase-mediated protection of tumor cells from apoptotic ROS signaling is perfectly matching with the H₂O₂-catabolizing phenoptype, as defined by the classical work of Galina Deichman's group (12-16). Their experimental tumor progression studies demonstrated a remarkable increase in tumorigenicity when in vitro transformed hamster cell populations, that had been inoculated into syngeneic animals for the establishment of tumor formation, were compared to bona fide tumor cells obtained from the arizing tumors. This increase in tumorigenicity was of several magnitudes and was strictly associated with the expression of resistance to oxidative stress (H₂O₂-catabolizing phenotype). Resistance to oxidative stress was discussed as efficient protection from the attack by neutrophils and macrophages that utilize specific ROS signaling for their attack towards malignant cells (12). A direct comparison of five in vitro transformed cell lines (transformed by different transformation principles) with corresponding tumor cell lines that had been isolated from tumors established through inoculation of the transformed cells, confirmed that all transformed and tumor cells exhibited marked extracellular superoxide anion production, but showed that the tumor cells were protected from ROS-mediated apoptosis induction through expression of membrane-associated catalase (Deichman and Bauer, in preparation). The characterization of the H₂O₂-catabolizing phenotype of tumor cells by Deichman's group represents the central intellectual concept and experimental key for the understanding of tumor cell resistance to ROS signaling. The findings on superoxide anion production by malignant cells, the characterization of ROS-dependent intercellular induction of apoptosis specifically in malignant cells, the acquisition of the H₂O₂-catabolizing phenotype during tumor progression and the correlation between this phenotype and the occurrence of protective membrane-associated catalase, allow a rather coherent and novel picture of the role of ROS during multistep oncogenesis to be drawn (Figure 1). Activation of oncogenes, inactivation of tumor suppressor genes and acquisition of independence of senescence control seem to be the primary prerequisites for the establishment of the transformed state. Transformed cells have the potential to form tumors, but they are different from bona fide tumor cells (i.e. cells finally found in tumors), with respect to the effects of ROS signaling. Transformed cells, as well as tumor cells generate extracellular superoxide anions that establish a ROS-dependent proliferation stimulus. This proliferation stimulus has been suggested to be mediated by the superoxide anions directly (86, 87), or through their dismutation product hydrogen peroxide (55, 57, 88). The generation of superoxide anions by transformed and tumor cells occurs in a rather sustained way, whereas a limited and tightly regulated superoxide anion production also plays a role for the control of proliferation of non-transformed cells (89-91). The low level of superoxide anion production of non-transformed cells is not sufficient to establish HOCl- or NO/peroxynitrite-dependent induction of apoptosis (31, 35), in contrast to transformed cells. Whereas transformed cells are subject to ROS-dependent intercellular induction of apoptosis (either through interaction with normal neighbouring cells, or in an autocrine mode), tumor cells possess acquired resistance against intercellular ROS-mediated apoptosis induction through expression of membrane-associated catalase (52-54). The expression of membrane-associated catalase in tumor cells is under constant positive control by H₂O₂. Removal of H₂O₂, thus, has a strong modulatory effect on catalase-mediated protection of tumor cells (Bauer, unpublished findings). The difference between transformed and tumor cells with respect to the expression of membrane-associated catalase is not as clear-cut as that demonstrated in Figure 1 for reasons of simplicity, rather it represents a gradual though profound difference. Transformed cells (but not non-transformed cells) are also found to express detectable concentrations of catalase on their surface, but the concentration of the enzyme is too low to prevent ROS signaling. In contrast, tumor cells adjust their membrane-associated catalase to a level that completely blocks ROS signaling. The involvement of ROS in tumor initiation and progression, as well as further pro-carcinogenic ROS-related effects, such as the induction of genomic instability (67), effects on angiogenesis (63, 64) and modulation of the metastatic potential (65, 66), are not shown in Figure 1. Multiple pro-carcinogenic effects of ROS, thus, seem to be counterbalanced by induction of ROS-dependent apoptosis of malignant cells. The exact knowledge of the biochemical basis for intercellular ROS signaling and resistance to this feature of malignant cells allows novel therapeutic approaches to be under-taken in the future, that will be based on the inhibition or inactivation of catalase, or the prevention of its expression. These aspects are the focus of this article.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Central reactive oxygen species (ROS)-related changes during transformation and tumor progression. Among many transformation-relevant cellular changes the continuous production of extracellular superoxide anions by membrane-associated NADPH oxidase (NOX1), controlled by the RAS-related small GTPase RAC1, represents one central hallmark of the transformed state (25-49, 55). Superoxide anions and their dismutation product H₂O₂ control the proliferation and the transformed state of the cells. However, they are also the basis for selective and efficient ROS-dependent intercellular induction of apoptosis that causes specific elimination of the transformed cells and that therefore seems to represent a basic tumorpreventive control system (21-24, 30, 32, 33, 35, 68-79). Tumor progression requires acquisition of resistance against intercellular ROS signaling through expression of membrane-associated catalase (CAT) (52-54), thus establishing the H₂O₂-resistant phenotype as defined by Deichman (12-16). Tumor cells maintain generation of extracellular superoxide anion through NADPH oxidase (NOX1) (28, 34, 39, 47, 52-55). Inhibition or inactivation of catalase is proposed as a novel therapeutic approach, based on the re-establishment of specific intercellular apoptotic ROS signaling of malignant cells. The figure does not show further ROS-related effects in tumor development and progression, such as induction of genomic instability, involvement in tumor inititation and tumor promotion, induction of cell motility, effects on angiogenesis and enhancement of metastasis.

ROS-dependent Apoptotic Signaling in Malignant Cells

The basic features of intercellular ROS-dependent induction of apoptosis are presented in Figure 2: Superoxide anions that are generated specifically by transformed cells through membrane-associated NOX1 are the basis for selective induction of apoptosis through the HOCl- and the NO/peroxynitrite pathways. The HOCl signaling pathway (steps 1-7 in Figure 2) depends on the dismutation of superoxide anions to hydrogen peroxide (30). Hydrogen peroxide is then utilized by a novel peroxidase for the synthesis of HOCl. This peroxidase is released by non-transformed, as well as transformed, cells. HOCl in the micromolar concentration range does not affect the cells directly (31, 33). However, the interaction of HOCl with superoxide anions (92-94) leads to the generation of hydroxyl radicals with the ability to trigger the onset of apoptosis through lipid peroxidation. As superoxide anions and hydroxyl radicals have relatively short free diffusion path lengths (95, 96), apoptosis induction is selectively directed towards the transformed target cells. As non-transformed cells generate much fewer extracellular superoxide anions than their transformed counterparts (30-32, 35), they are not affected by HOCl in the micromolar concentration range. In the case of a vast excess of hydrogen peroxide compared to peroxidase, consumption of HOCl through the reaction with hydrogen peroxide (H₂O₂ + HOCl → H₂O₂ + O₂ + H⁺ + Cl⁻) may lead to a significant blunting of HOCl signaling. The

NO/peroxynitrite signaling pathway depends on the release of NO by non-transformed or transformed cells. NO is generated within the cells through the action of NO synthase (NOS), either inducible NOS (iNOS), neuronal NOS (nNOS) or endothelial NOS (eNOS), which utilize arginine as their substrate (97, 98). The available concentration of NO in the intercellular space is controlled by a complex consumption reaction between NO and hydrogen peroxide. This reaction depends on the oxidation of NO, subsequent generation of N₂O₃ and the reaction of N₂O₃ with hydroperoxide anions (Bauer, unpublished findings). Free NO and superoxide anions derived specifically from transformed cells interact and form peroxynitrite in a diffusion-controlled reaction (99-102). After protonation, peroxynitrite rapidly decomposes into NO₂ and apoptosis-inducing hydroxyl radicals (103, 104). The nitryl chloride signaling pathway (83) and the metal ion-catalyzed Haber-Weiss reaction (84) are of minor importance and are not shown in Figure 2. Both pathways depend on the availability of superoxide anions and hydrogen peroxide. The intercellular ROS-dependent signaling chemistry outlined in Figure 2 is further substantiated in Figure 3. This figure focuses on the signaling chemistry during autocrine ROS signaling for the sake of simplicity of the scheme. This reduction is justified, as the interaction between non-transformed and transformed cells follows the same signaling chemistry as the autocrine process, as can be deduced from Figure 2. This deduction has been experimentally verified. The combination of siRNA-based analysis, inhibitor studies and reconstitution experiments allowed the essential players and their reactions during ROS-mediated signaling to be defined. Figure 3 points to the role of activation of the rat sarcoma oncogene (RAS) and and the RAS-related small GTPase RAC for the activity of the NADPH oxidase NOX1 (steps 1 and 2). The HOCl-synthesizing peroxidase is coded by the peroxidase domain of dual oxidase (DUOX) (105-107). The peroxidase domain is freed through the action of matrix metalloproteases that are inhibited by galardin (step 3) (108). Release of the peroxidase is not a prerequisite for its activity. The cells release inactive transforming growth factor type-beta1 (TGF-β1), a complex between TGF-β1 and the large latency-associated protein, which is subsequently activated through a change of the conformation of the inhibitory latency-associated protein (step 4). The interaction of active TGF-β1 with its receptor leads to an induction of NOX1 as well as to DUOX expression. In vitro, intercellular ROS signaling is significantly stimulated by the addition of exogenous TGF-β1 and severely inhibited by neutralizing antibodies directed towards TGF-β1, or through siRNA-based knockdown of either TGF-β1 or its receptor. The level of free NO is controlled at several central points that co-operate in the modulation of the efficiency of the NO/peroxynitrite signaling pathway. First of all, the level of arginine is controlled by the activity of cellular arginase (step 13). The concentration of active NO synthase then defines the rate of NO synthesis (step 14). A substantial amount of NO is converted to nitritate through the action of NO dioxygenase (NOD) (step 15) (109-112), which is interconnected to the activity of cytochrome P450-dependent oxido-reductase POR (111). NO passes the cell membrane readily and is then channelled either into the complex consumption reaction with hydrogen peroxide (step 18) or to the interaction with superoxide anions, resulting in the formation of peroxynitrite (step 16). The complex scenario outlined in Figure 3 also shows the positive effect of ROS signaling for the cells, simplified by the proliferation-stimulating activity of hydrogen peroxide (step 8) (55, 57). Other groups suggest a direct role of superoxide anions for proliferation stimulus (86, 87). These positive ROS-dependent effects on the cells are counterbalanced by severe negative effects directed against transformed cell survival. These effects are mediated by hydroxyl radicals derived either from the interaction between superoxide anions and HOCl (92-94), or the decomposition of protonated peroxynitrite (103, 104). Hydroxyl radicals cause lipid peroxidation that transmits the apoptosis-inducing signal through sphingomyelinase activation, ceramide formation, activation of the mitochondrial pathway of apoptosis, caspase-9 and caspase-3 activity (data not shown). During tumor progression, tumor cells acquire resistance to intercellular ROS signaling through expression of membrane-associated catalase (52-54). The tumor cell phenotype characterized by the combination of activated NOX1 and membrane-associated catalase correlates with the hydrogen peroxide-catabolizing phenotype of tumor cells as characterized by Deichman et al. (12-16). Figure 4 demonstrates that membrane-associated catalase of tumor cells interferes with HOCl signaling through removal of hydrogen peroxide (and thus prevention of HOCl production) and with NO/peroxynitrite signaling, through decomposition of peroxynitrite. Catalase also efficiently blocks the nitryl chloride signaling pathway, as the inhibition of HOCl production also prevents formation of nitryl chloride through HOCl/nitrite interaction. Finally, catalase prevents the metal ion-dependent Haber-Weiss reaction, as this signaling pathway depends on decomposition of hydrogen peroxide through Fenton chemistry. Thus, the expression of only one phenotypic marker, i.e. membrane-associated catalase, prevents induction of apoptosis by all four ROS-dependent intercellular signaling pathways in tumor cells. This feature defines membrane-associated catalase of tumor cells as an attractive target for selective antitumor therapy based on ROS-mediated induction of apoptosis. The protective function of catalase depends on its multiple reaction pathways, as summarized in Figure 5: HOCl signaling (step 1) is prevented through the classical catalase reaction (steps 2 and 3), in which the native catalase (CATFe^III) interacts with hydrogen peroxide in a first step, resulting in the formation of compound I (CAT FeIV=O^+.) and water. Compound I then interacts with a second molecule of hydrogen peroxide, yielding native catalase, water and molecular oxygen. The inhibitory effect of catalase on the NO/peroxynitrite pathway (step 4) depends on the potential of catalase to use peroxynitrite as a substrate for a two-step reaction (steps 5 and 6), that involves compound I formation in an analogous way as during the interaction of the enzyme with hydrogen peroxide. The differentiation between true peroxynitrite decomposition by catalase from a theoretically conceivable action of catalase on an hydrogen peroxide-mediated step induced by peroxynitrite was recently shown (54). The differentiation utilized the ability of catalase compound I to subtract two hydrogen atoms from methanol, resulting in formation of formaldehyde that can be easily detected by purpald staining. It was shown that in the presence of peroxynitrite (and the absence of hydrogen peroxide), catalase indeed caused formaldehyde formation from methanol. This confirms the direct interaction of peroxynitrite with catalase, leading to enzymatic peroxynitrite decomposition. This conclusion is in perfect agreement with the findings of Gebicka and Didil (113) and Kono et al. (114). The work of other groups has shown that NO can inhibit catalase reversibly (step 7) (115). This inhibition is characterized by an enzyme inhibitor constant (Ki) of 0.18 μM NO. Finally, compound I has the potential to oxidize NO and thus to remove NO from the system (step 8) (116). This potential adds to the protection of tumor cells as it counterbalances the inhibitory effect of NO on catalase and also interferes with NO/peroxynitrite signaling.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Intercellular and autocrine reactive oxygen species (ROS)-dependent apoptotic signaling. Transformed target cells generate extracellular superoxide anions (step 1), whereas neighboring non-transformed cells, as well as the transformed cells themselves contribute two effector molecules: peroxidase (POD, step 3) and NO (step 8). The interplay between superoxide anions and the effector molecules establish two major signaling pathways. The HOCl pathway depends on dismutation of superoxide anions to H₂O₂ (step 2), synthesis of HOCl through peroxidase (step 4), formation of hydroxyl radicals through HOCl/superoxide anion interaction (step 5). Hydroxyl radicals induce apoptosis through lipid peroxidation, followed by sphingomyelinase activation, ceramide formation and induction of the mitochondrial pathway of apoptosis. Excess H₂O₂ blunts HOCl signaling through a consumption reaction (step 7): H₂O₂ + HOCl → H₂O + O₂ + H⁺ + Cl⁻. NO is generated by NO synthase (NOS) (step 8) and is efficiently removed in a complex consumption reaction (step 9). Free NO readily interacts with superoxide anions and thus generates peroxynitrite (step 10). The protonated form of peroxynitrate decomposes into NO₂ and hydroxyl radicals (step 11). This figure also shows that selective apoptosis induction in transformed cells by either interaction with non-transformed cells or within the transformed cell population (autocrine self-destruction) is based on the same signaling chemistry.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Detailed picture of central players of intercellular reactive oxygen species (ROS)-mediated apoptotic signaling. siRNA-based analysis, reconstitution experiments and inhibitor studies have allowed to define the essential players in intercellular ROS signaling. NOX1 represents the NADPH oxidase that is activated by rat sarcoma oncogene (RAS) through the RAS-related small GTPase RAC1 (25). The peroxidase domain of dual oxidase (DUOX) is removed from DUOX through matrix metalloproteases (MMP). Inactive transforming growth factor type-β (TGF-β) is released by the cells and, upon activation through conformational change activates its receptor. As a result, synthesis of NOX1 and DUOX is increased, resulting in enhanced superoxide anion production and peroxidase release. Arginase controls the levels of arginine, the substrate for NO synthase (NOS). A substantial part of NO generated by NOS is dioxygenated by NO dioxygenase (NOD) that is controlled by cytochrome P450 oxido-reductase. Please find details in the text.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Catalase protects tumor cells against intercellular reactive oxygen species (ROS) signaling. Efficient tumor progression resulting in tumor formation requires the establishment of resistance to ROS signaling (52-54). This is achieved uniformly in all tumor cell systems studied so far, through expression of membrane-associated catalase on the outer surface of the tumor cells. Catalase interferes with HOCl signaling (steps 1-5) through decomposition of H₂O₂ (step 11) and with the NO/peroxynitrite pathway (steps 9 and 10) through decomposition of peroxynitrite (step 12) and oxidation of NO through compound I (not shown in Figure 4, for details see Figure 5). Decomposition of H₂O₂ also inhibits the nitryl chloride pathway and the metal ion-catalyzed Haber-Weiss reaction (data not included in this figure).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Multiple protective functions of catalase against reactive oxygen species (ROS) signaling. HOCl signaling (step 1) is prevented by catalase through removal of H₂O₂ in a two-step reaction (steps 2 and 3), in which native catalase (CAT Fe^III) reacts with the first molecule of H₂O₂, resulting in formation of water and compound I (CAT Fe^IV=O^+.), which, in a reaction with H₂O₂, generates water and molecular oxygen and is thereby converted back to CAT Fe^III. Peroxynitrite is generated through the reaction between superoxide anions and NO (step 4). Peroxynitrite is decomposed by catalase in a two-step reaction (steps 5 and 6) in analogy to H₂O₂, involving intermediate formation of compound I. Free NO can inhibit catalase in a reversible mode (step 7, K_i=0.18 μM) or is oxidized by compound I (step 8) and thus removed from signaling.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Restoration of intercellular reactive oxygen species (ROS)-mediated apoptotic signaling of tumor cells through inhibition of catalase activity or its expression. Inhibitors of catalase like 3-aminotriazole (3-AT), monoclonal antibodies towards catalase, and siRNA-mediated knockdown of catalase have the potential to restore intercellular ROS signaling and subsequent selective apoptosis induction in tumor cells.

Sensitization of Tumor Cells to ROS-mediated Intercellular Apoptotic Signaling through Catalase Inhibition or siRNA-mediated Knockdown of its Expression

Figure 6 demonstrates that tumor cells can be sensitized to their own specific intercellular ROS signaling when membrane-associated catalase is inhibited by small inhibitors such as 3-aminotriazole (3-AT), or monoclonal antibodies directed towards catalase, or when the expression of catalase is prevented through siRNA-based knockdown (52-54). These three approaches have been utilized for the analysis of catalase-mediated protection of tumor cells (52-54). The specificity of these approaches was assured as i) intercellular ROS signaling-controlled induction of apoptosis was activated after catalase inhibition or knockdown of its expression, as ii) the apoptosis-inducing ROS-signaling dependent effects of specific antibody towards catalase or siRNA-based knockdown were specifically competed with by the addition of exogenous catalase, and iii) the apoptosis-inducing effect of the catalase inhibitor 3-AT was abrogated through addition of the catalase mimetic chloro[[2,2’-[1,2-ethanediylbis [(nitrilo-κN)methylidyne]]bis[6-methoxyphenolato-κO]]]-manganese (EUK-134). The salen-manganese complex EUK-134 exhibits an activity analogous to that of catalase but is not inhibited by 3-AT (117). Whereas 3-AT and siRNA-based knockdown affect both membrane-bound catalase (specific for tumor cells) and intracellular peroxisomal catalase (which is not specific for tumor cells), the monoclonal antibody acts exclusively on the membrane-associated catalase due to its lack of cell permeability. The inactivation of intracellular catalase may contribute to the efficiency of ROS-mediated intercellular induction of apoptosis, as this process is linked to intracellular ROS-dependent effects. Normal cells which lack the potential for intercellular signaling and have no extracellular catalase, were not affected by the inhibition of their intracellular catalase. This is explained by the lack of extracellular ROS-derived lipid peroxidation and the redundancy of intracellular antioxidant defence, where catalase, glutathione, glutathione peroxidase and other enzymes work together to cope with intracellular oxidative stress. Nevertheless, the selective inhibition of tumor cell membrane-associated catalase by monoclonal antibodies seems to represent a superior approach for future therapeutic approaches based on catalase inhibition and induction of ROS-mediated induction of apoptosis due to its specific action on tumor cell membrane-associated catalase. Experimental data demonstrating the sensitization of tumor cells to ROS-mediated induction of apoptosis after catalase inhibition are shown in Figure 7. The data summarize our findings for the action of the catalase inhibitor 3-aminotriazole (3-AT) on the gastric carcinoma cell line MKN-45. Analogous results have been obtained for the application of antibodies towards catalase and for knockdown of catalase by increasing concentrations of specific siRNA (54). Figure 7 demonstrates that untreated MKN-45 tumor cells only have a minor background apoptotic activity in the absence of the catalase inhibitor. At low concentrations of 3-AT (6 mM), apoptosis was induced and depended exclusively on the NO/peroxynitrite pathway, as it was blocked by inhibition of NADPH oxidase through 4-(2-aminoethyl)benzene-sulfonyl fluoride (AEBSF), scavenging of superoxide anions through superoxide dismutase (SOD), inhibition of NO synthesis through the NOS inhibitor N-ω-nitro-L-arginine methylester hydrochloride (L-NAME), decomposition of peroxynitrite through the peroxynitrite decomposition catalyst 5-,10-,15-,20-tetrakis(4-sulfonatophenyl)porphyrinato iron(III) chloride (FeTPPS) and scavenging of hydroxyl radicals through mannitol. The effect of 6 mM 3-AT was slightly increased when either an exogenous NO donor was added, or endogenous NO generation was enhanced by the addition of arginine. Inhibitors specific for the HOCl pathway [such as the HOCl scavenger taurine and the peroxidase inhibitor 4-aminobenzoyl hydrochloride (ABH)] did not inhibit apoptosis induced by 6 mM 3-AT. Prevention of peroxidase release by galardin-mediated inhibition of matrix metalloproteinase activity caused abrogation of NO/peroxynitrite signaling, as the peroxidase localized at the cell membrane decomposes peroxynitrite in a specific enzyme reaction, as seen by the interference with the peroxidase inhibitor ABH. This site-specific effect is not seen when the peroxidase is released from the cells and thus its local concentration at the membrane becomes much lower. When 75 mM 3-AT were added, overall induction of apoptosis increased and were mainly dependent on HOCl signaling, as seen from the inhibition profile. Decomposition of hydrogen peroxide by 2 μM of the catalase mimetic EUK-134, peroxidase inhibition by ABH as well as scavenging HOCl by taurine, caused nearly complete inhibition of apoptosis induction. The involvement of superoxide anions and hydroxyl radicals in intercellular signaling under these conditions was assured by the inhibitory effects of AEBSF, SOD and mannitol. The addition of FeCl₂ caused a nearly complete block of the reaction, as this leads to Fenton chemistry-dependent decomposition of hydrogen peroxide and HOCl, distant from the membrane. The resulting hydroxyl radicals are too far away to reach and attack the cell membrane, due to their very low free diffusion path length. At the same time, essential components of HOCl signaling, i.e. H₂O₂ and HOCl, then no longer contribute to specific apoptosis induction (53). The inhibitory effect of the NO donor sodium nitroprusside (SNP) on HOCl signaling is explained by the consumption reaction between NO and hydrogen peroxide. Apoptosis induction at both concentrations of 3-AT was dependent on the activity of caspase-3 and caspase-9, whereas caspase-8 inhibitor had no signficant inhibitory effect. This finding is in good agreement with apoptosis induction through the mitochondrial pathway of apoptosis. The singlet oxygen scavenger histidine had no effect under the conditions of this experiment. These data demonstrate that a low degree of catalase inhibition can be sufficient to reactivate NO/peroxynitrite signaling, whereas a higher degree of inhibition is necessary to establish HOCl signaling. This sequence of events is reasonable, as protection against peroxynitrite, which is directly decomposed at the membrane, requires a high local concentration of protective catalase and therefore reacts sensitively to even low degrees of catalase inhibition. In contrast, the HOCl signaling pathway depends on sufficient concentrations of hydrogen peroxide and therefore requires a higher degree of catalase inhibition than does the NO/peroxynitrite pathway. The reverse sequence of events was seen when intercellular ROS-dependent apoptosis induction in sarcoma oncogene (SRC)-transformed cells was gradually inhibited by addition of increasing concentrations of exogenously added catalase (54): Low concentrations of catalase efficiently blocked intercellular induction in a concentration-dependent mode. Signaling was initially dependent on the sole action of the HOCl signaling pathway, but switched to NO/peroxynitrite signaling with increasing concentrations of exogenous catalase. At defined catalase concentrations, HOCl signaling was completely blocked and NO/peroxynitrite signaling was enhanced due to interference of catalase with the consumption between hydrogen peroxide and NO. Further increase in catalase concentration finally, also abrogated NO/peroxynitrite signaling, but rather high concentrations of catalase were necessary to achieve complete inhibition. This is a reasonable finding, as it requires high concentrations of soluble catalase to mimic the high local concentration at the cell membrane that is a prerequisite for protection from peroxynitrite. These data point to the significance of site-specific interactions in this complex intercellular signaling system and its control. The exact signaling profile of malignant cells depends on the relative ratios between superoxide anions, peroxidase and NO generated by the tumor cells. In certain tumor cell systems, such as neuroblastoma cells and Ewing sarcoma cells, a relative abundance of NO may be responsible for exclusive signaling through the NO/peroxynitrite pathway, at all 3-AT concentrations tested (54). In these cells, apoptosis induction by 3-AT up to 100 mM was exclusively dependent on superoxide anions (inhibition by AEBSF) and NO (inhibition by the nNOS-specific inhibitor 3-bromo-7-nitroindazole), but was not affected by the HOCl-scavenger taurine, nor by the singlet oxygen-scavenger histidine (54). A similar signaling profile has been found for mammary and ovarian carcinoma cells (Bauer, unpublished results). Work in progress shows that the signaling profiles of human tumor cells in the presence of a catalase inhibitor can be modulated in a predictable way, through modulation of the central players of intercellular ROS signaling. The increase in superoxide anion production shifts signaling towards HOCl signaling, whereas reduction of superoxide anion production and an increase in NO production causes dominant NO/peroxynitrite signaling.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

NO/peroxynitrite and HOCl signaling after differential inhibition of catalase in MKN-45 gastric carcinoma cells. A total of 12,500 MKN-45 human gastric carcinoma cells in 100 μl RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), were incubated in the absence or the presence of 6 mM or 75 mM or the catalase inhibitor 3-aminotriazole (3-AT). Assays contained no further additions or the indicated inhibitors: 100 μM of the NADPH oxidase inhibitor 4-(2-aminoethyl)benzene-sulfonyl fluoride (AEBSF), 60 U/ml Mn superoxide dismutase (MnSOD), 0125 μM or 2 μM of the catalase mimetic chloro[[2,2’-[1,2-ethanediylbis[(nitrilo-κN)methylidyne]]bis[6-methoxyphenolato-κO]]]-manganese (EUK-134), 150 μM of the peroxidase inhibitor 4-amino benzoyl hydrazide (ABH), 50 mM of the HOCl-scavenger taurine, 10 mM of the hydroxyl radical-scavenger mannitol, 20 μM FeCl₂, 2.4 mM of the NOS inhibitor N-ω-nitro-L-arginine methylester hydrochloride (L-NAME), 20 μM of the peroxynitrite decomposition catalyst 5-,10-,15-,20-tetrakis(4-sulfonatophenyl)porphyrinato iron(III) chloride (FeTPPS), 0.75 mM arginine, 0.05 mM of the NO donor sodium nitroprusside (SNP), 10 μM of the matrix metalloprotease inhibitor galardin, 2 mM of the singlet oxygen-scavenger histidine, 25 μM of the caspase-8 inhibitor, 25 μM of the caspase-9 inhibitor or 50 μM of the caspase-3 inhibitor. After 4 h at 37°C, the percentage of apoptotic cells was determined by the classical morphological criteria for apoptosis (nuclear condensation, nuclear fragmentation and membrane blebbing). Please find technical details for cell culture, reagents and quantitation of apoptotic cells in reference 54 and the discussion of the signaling chemistry in the text.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Establishment of intercellular reactive oxygen species (ROS) signaling through addition of exogenous signaling components. Addition of exogenous HOCl causes apoptosis induction in tumor cells through HOCl/superoxide anion interaction and subsequent hydroxl radical generation. Whereas tumor cell catalase efficiently prevents HOCl synthesis of tumor cells, it ensures optimal action of exogenously added HOCl through removal of H₂O₂. In the presence of the ferric ion chelator deferoxamine (DFO), consumption of HOCl and H₂O₂ through Fenton chemistry is inhibited. Fenton chemistry distant from the cell membrane does not contribute to apoptosis induction in the tumor cells (as HOCl and H₂O₂ are relatively far-ranging molecular species and hydroxyl radicals generated through the action of ferrous ions cannot reach the cell membrane) but rather inhibits the signaling process. Abrogation of HOCl and H₂O₂ consumption through removal of ferric ions (which are centrally involved in Fenton chemistry after being reduced to ferrous ions by superoxide anions) causes apoptosis induction due to enhancement of the specific signaling effects at the cell membrane that are no longer controlled by catalase. Addition of H₂O₂ –generating glucose oxidase (GOX) causes apoptosis induction through re-establishment of the HOCl signaling pathway. This effect is synergized by addition of exogenous peroxidase such as myeloperoxidase (MPO). Re-establishment of HOCl signaling has been initially interpreted as mere overrunning of the protective potential of catalase (51, 52). Recent experiments show that a more complicated process is involved in this process, as further discussed in Figure 9.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Excess H₂O₂ causes singlet oxygen generation through displacement of peroxynitrite from catalase. Excess H₂O₂ generated by exogenously added glucose oxidase (GOX) competes with peroxynitrite for catalase and leads to the displacement of peroxynitrite. As a result, H₂O₂ and peroxynitrite interact and form singlet oxygen. Singlet oxygen causes a APO/FAS receptor and caspase-8-mediated amplification step through induction of NADPH oxidase (NOX1) and NO synthase (NOS). As a result, increased concentrations of H₂O₂ and peroxynitrite generate more singlet oxygen that inactivates catalase and allows subsequent intercellular ROS signaling. Please find details of the reactions in the text.

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

Excess peroxynitrite causes singlet oxygen generation through displacement of H₂O₂ from catalase. Addition of exogenous peroxynitrite to tumor cells causes displacement of H₂O₂ from catalase and subsequent singlet oxygen generation through H₂O₂/peroxynitrite interaction. The subsequent steps, amplification of singlet oxygen generation, catalase inactivation and re-establishment of intercellular ROS-mediated apoptotic signaling, are identical to those described in Figure 9.

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

The exogenous singlet oxygen generator photofrin sensitizes tumor cells to intercellular reactive oxygen species (ROS)-mediated apototic signaling through inactivation of protective catalase. Addition of high concentrations (10 μM) photofrin and application of visible light causes sufficient singlet oxyxgen generation for direct inactivation of catalase (step 2a) and subsequent ROS-mediated apoptotic signaling. Addition of lower photofrin concentrations requires amplification of singlet oxygen generation through APO/FAS receptor activation (step 2b), NADPH oxidase (NOX1) and NO synthase (NOS) stimulation (steps 3a and 3b), reaction between peroxynitrite (step 4a) and hydrogen peroxide (step 4b) which results in singlet oxygen formation (step 5), before catalase is inactivated and ROS-mediated signaling is re-established.

Reestablishment of Intercellular ROS Signaling through Addition of Signaling Components or Deferoxamine (DFO)-dependent Prevention of Side Reactions

ROS-dependent induction of apoptosis in tumor cells can also be achieved by the addition of exogenous HOCl (Figure 8), mimicking a situation in which tumor cells are confronted with activated neutrophils that generate HOCl through activated NADPH oxidase and myeloperoxidase (118-120). Apoptosis induction in tumor cells by exogenous HOCl requires free superoxide anions for the interaction with HOCl, leading to the formation of apoptosis-inducing hydroxyl radicals (31). Apoptosis induction by HOCl is more efficient in the presence of active catalase which removes hydrogen peroxide that otherwise might consume HOCl. Addition of DFO, a chelator of ferric ions, causes apoptosis induction in tumor cells. This is explained by the interference of DFO with nondirected Fenton chemistry of hydrogen peroxide and HOCl distant from the cell membrane. As hydrogen peroxide and HOCl are rather far-ranging molecular species, hydroxyl radicals derived from their interaction with ferrous ion are too far away to reach the membrane of the target cells. Therefore, these two central signaling components are used up in a cycle that is driven by superoxide anion-dependent reduction of ferric ions back to reactive ferrous ions. Removal of ferric ions by DFO, thus, causes a dramatic net concentration of signaling components in the milieu of the tumor cells and subsequent apoptosis induction. This system is challenging for analytical goals but does not seem to be too promising for therapeutic approaches, as DFO might interfere with the synthesis of other metal-containing essential enzymes in normal tissue. DFO has been used in tumor therapy (121) and the effects have been discussed in relation to the role of iron for the growth of neuroblastoma cells (122). It is possible that the mechanism described here also contributed to the outcome of the study by Blatt (121). As protection of endogenous hydrogen peroxide and HOCl proved to be sufficient for apoptosis induction in tumor cells, addition of excess hydrogen peroxide (e.g. through steady generation by glucose oxidase), alone or in combination with additional peroxidase (e.g. myeloperoxidase), should allow restoration of intercellular ROS signaling despite the presence of protective catalase in the cell membrane. This prediction was indeed confirmed in model experiments (52, 53), and allowed to draw the conclusion that an excess of the substrate H₂O₂ might have overrun catalase and thus allowed subsequent ROS-mediated apoptosis. Control experiments ensured that the effect of glucose oxidase alone, and in combination with peroxidase, was indeed due to HOCl/superoxide anion interaction leading to the generation of apoptosis-inducing hydroxyl radicals.

• Download figure
• Open in new tab
• Download powerpoint

Figure 12.

NO-mediated catalase inhibition causes subsequent singlet oxygen-dependent catalase inactivation. The increase of free NO in tumor cells causes a transient and reversible inhition of catalase through formation of CATFe^IIINO. As a consequence, the decomposition of H₂O₂ and peroxynitrite by catalase is reduced and these two molecules can interact and form singlet oxygen that triggers its own amplification and subsequent catalase inactivation in analogy to the mechanisms described in Figures 9, 10 and 11.

• Download figure
• Open in new tab
• Download powerpoint

Figure 13.

Increase in endogenous NO after arginine addition causes singlet oxygen-dependent catalase inactivation and ROS-mediated intercellular signaling. A total of 25,000 human lymphoma Gumbus cells in 100 μl medium containing 10% fetal bovine serum (FBS) were treated with the indicated concentrations of arginine. Assays received either no inhibitors or 2 mM of the singlet oxygen scavenger histidine, 2.4 mM of the NOS inhibitor N-ω-nitro-L-arginine methylester hydrochloride (L-NAME), 25 μM of the peroxynitrite decomposition catalyst 5-,10-,15-,20-tetrakis(4-sulfonatophenyl)porphyrinato iron(III) chloride (FeTPPS) or 50 mM of the HOCl-scavenger taurine. Inhibitors were added either immediately before (A) or 1 h after arginine application. The percentages of apoptotic cells were determined 3 h after (B) arginine addition. Please find details for cell culture, inhibitors and quantitation of apoptosis in (52) and (54). Gumbus cells were a generous gift of Dr. G. Dölken, University of Greifswald, Germany.

• Download figure
• Open in new tab
• Download powerpoint

Figure 14.

Inhibition of NOD enhances peroxynitrite-dependent effects. A total of 12,500 transformed 208Fsrc3 cells in 100 μl medium, supplemented with 5% fetal bovine serum (FBS) were treated with increasing concentrations of the NO donor diethylamine NONOate (DEA NONOate), in the presence and absence of the NO synthase (NOS) inhibitor N-ω-nitro-L-arginine methylester hydrochloride (L-NAME), which had been added 10 min before the NO donor. Parallel assays received 20 μg/ml itraconazole plus L-NAME 10 min before addition of DEA NONOate. The percentages of apoptotic cells were determined 3.5 h after addition of the NO donor. Induction of apoptosis was solely dependent on the concentration of exogenously added NO. Endogenously produced NO did not contribute to the reaction measured under these conditions. In the presence of itraconazole (and the NOS inhibitor) apoptosis induction was efficiently enhanced at lower concentrations of the NO donor. When 1 mM DEA NONOate was applied, the enhancement by itraconazole was only marginal. This finding is in agreement with stabilization of the free NO concentration through inhibition of NOD that converts NO to nitrate. Control experiments (data not shown) confirmed that apoptosis induction measured in this experiment was dependent on superoxide anions and had been caused by peroxynitrite formation.

Singlet Oxygen-dependent Modulation of Intercellular ROS-dependent Apoptotic Signaling

Although experimental addition of glucose oxidase to tumor cells re-establishes intercellular ROS signaling (Figure 8; references 52 and 53), the initial steps after generation of excess H₂O₂ by glucose oxidase are far more complex than originally anticipated (Bauer, manuscript in preparation). The use of additional inhibitors revealed that singlet oxygen-dependent reactions were also involved in this scenario. Thereby, the formation of singlet oxygen through peroxynitrite/H₂O₂ interaction (123) and the potential of singlet oxygen to inactivate catalase (124, 125), possibly also through interaction with the essential histidine at position 74 (126), are of crucial importance. The ability of singlet oxygen to induce the death receptor APO/FAS in a ligand-independent mode (127) and the ability of the APO/FAS receptor to stimulate NOX1 activity (128, 129) and NOS expression (130) proved to be additional components in this autoamplificatory signaling cascade, whose details are summarized in Figure 9. The figure visualizes that by competition, the presence of an excess of H₂O₂ excludes peroxynitrite from destruction by catalase. Peroxynitrite now has a greater chance of interacting with one of the abundant H₂O₂ molecules not yet decomposed by catalase and thus to generate singlet oxygen. The concentration of singlet oxygen generated this way may not be sufficiently high to inactivate the number of catalase molecules necessary to release signaling from negative control. However, if singlet oxygen hits an APO/FAS receptor (127), the subsequent caspase-8-dependent induction of NOX1 (128, 129) and NOS activity (130) results in overall increased concentrations of hydrogen peroxide, as well as of peroxynitrite. This allows for subsequent increased formation of singlet oxygen (123), at a concentration to cause optimal catalase inactivation and subsequent apoptotic ROS signaling. As soon as lipid peroxidation has been induced in the cell membrane by hydroxl radicals derived from ROS signaling, a further loop of singlet oxygen generation is conceivable, as peroxynitrite also forms singlet oxygen in the reaction with biological hydroperoxides (131), based on the reaction with the model compound tert-butyl hydroperoxide (132). The reaction steps with amplificatory potential have been elucidated by kinetic addition of inhibitors and by the use of specific siRNA-mediated knockdown of certain players; this study is to be published in detail elsewhere. The kinetic study of inhibitor action revealed that APO/FAS-triggered caspase-8 was only required during the first minutes for activation of NOX1 and NOS, but not for the subsequent induction of apoptotic cell death which was controlled by caspases 9 and 3. If the reaction scheme proposed in Figure 9 was correct, the opposite approach, i.e. outcompeting of H₂O₂ by a vast excess of peroxynitrite, should lead to the same effect as generation of relatively high concentrations of H₂O₂. Recently completed experiments (Bauer, manuscript in preparation) show this assumption to be correct. As summarized in Figure 10, an excess of exogenous peroxynitrite seems to outcompete H₂O₂, leading to singlet oxygen generation, amplification of singlet oxygen generation through singlet oxygen-dependent APO/FAS activation, increased NOX1 activity and NOS enzyme concentration. This allows for subsequent catalase inactivation by singlet oxygen formed through the interaction between excess hydrogen peroxide and peroxynitrite. For both scenarios (Figures 9 and 10), early (but not late) requirement of APO/FAS and caspase-8 activity, as well as catalase inactivation (enabling late intercellular ROS signaling) have been experimentally verified (data not shown). When synergistic conditions with parallel enhancement of peroxynitrite and H₂O₂ generation were experimentally established, the amplification step by the APO/Fas receptor and caspase-8 was no longer required. The central role of singlet oxygen during a complex set of biochemical steps leading to ROS-driven catalase inactivation, prompted testing of whether the generation of extracellular singlet through application of a singlet oxygen generator such as photofrin might lead to the inactivation of tumor cell-protective catalase and subsequent specific ROS signaling. When normal cells and tumor cells were incubated with photofrin in the dark, and visible light was applied after the photosensitizer had reached the cytoplasm, both cell types died readily through apoptotic and necrotic cell death. However, when visible light was applied immediately after the addition of photofrin to the cells, before the photosenitizer had a chance to enter the cells, the non-transformed cells were not affected, but the tumor cells died through specific ROS signaling, as shown by the effects of inhibitor application (Riethmüller and Bauer, in preparation). Detailed studies revealed that singlet oxygen by itself was not sufficient to induce apoptosis directly under these conditions, but singlet oxygen-dependent catalase inactivation was necessary to allow intercellular apoptotic ROS signaling based on tumor cell-specific extracellular superoxide formation and re-establishment of NO/peroxynitrite and/or HOCl signaling (data not shown). When high concentrations of photofrin were applied, the generated concentration of singlet oxygen was sufficient to directly inactivate catalase (step 2a) and to allow for immediate intercellular ROS signaling. Potential effects of singlet oxygen on the cell membrane were not suffient for apoptosis induction and the requirement for catalase inactivation (through singlet oxygen/histidine interaction) was directly proven by kinetic inhibitor experiments. When low concentrations of photofrin were applied (Figure 11), an amplification step through activation of the APO/Fas receptor, and potentially also of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) receptor, was necessary (step 2 b), leading to the same biochemical amplification machinery as outlined in Figures 9 and 10. The generation of sufficient singlet oxygen after these amplification steps, caused sufficient catalase inactivation that allowed for ROS-dependent signaling by the known signaling pathways. Under suboptimal inactivation conditions, the NO/peroxynitrite pathway dominated, whereas optimal inactivation of catalase enabled re-establishement of HOCl signaling. These data demonstrate the central importance of singlet oxygen for the inactivation of catalase. A modification of photodynamic therapy (which is based on the generation of singlet oxygen) in a way that cell-impermeable photosensitizers are used for selective apoptosis induction in tumor cells is thus proposed. As yet, photodynamic therapy depends on an increased uptake of photofrin and related compounds into the tumor, but not on a selective action at the cellular level. Photosensitizers acting strictly in the extracellular space, thereby inactivating tumor cell catalase and establishing intercellular ROS signaling, might improve selectivity and efficiency of photodynamic compounds. In parallel, this approach might reduce unwanted side-effects on nonmalignant tissue.

• Download figure
• Open in new tab
• Download powerpoint

Figure 15.

ROS-mediated apoptosis induction in the human lymphoma cell line Gumbus, by application of the NOD inhibitor miconazole. A total of 25,000 human lymphoma Gumbus cells in 100 μl RPMI-1640 medium containing 10% fetal bovine serum (FBS) were treated with the indicated concentrations of miconazole. Assays received either no inhibitor or 2 mM of the singlet oxygen-scavenger histidine, 2.4 mM of the NO synthase (NOS) inhibitor N-ω-nitro-L-arginine methylester hydrochloride (L-NAME), 25 μM of the peroxynitrite decomposition catalyst 5-,10-,15-,20-tetrakis(4-sulfonatophenyl)porphyrinato iron(III) chloride (FeTPPS) or 50 mM of the HOCl-scavenger taurine. Inhibitors were added either immediately before (A) or 1 h after (B) miconazole application. The percentage of apoptotic cells was determined 10 hs after miconazole addition.

• Download figure
• Open in new tab
• Download powerpoint

Figure 16.

Distinct ways to generate cell-derived catalase-inactivating singlet oxygen by modulation of free NO. Experimental evidence obtained in a comprehensive survey showed a remarkable variety of strategies using modulation of NO availability for apoptosis induction in tumor cells, based on preceeding singlet oxygen generation and catalase inactivation. Inhibition of arginase or addition of arginine cause singlet-oxygen-mediated catalase inactivation in many different tumor cell systems. Likewise, induction of NOS expression by interferon γ, APO/FAS ligand/receptor interaction and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) ligand/TRAIL receptor interaction caused an analogous effect. Death receptor effects were most prominent in those cell systems that did not show apoptosis by activation of the receptor alone. In addition to the previously reported inhibitors of NO dioxygenase (NOD) (111), a large number of secondary plant compounds were found to inhibit either NOD directly or cytochrome P450 oxido-reductase (POR), that is interconnected with NOD. Intriguingly, certain established chemotherapeutics were also found to act at this step, unexpectedly (data not shown). Finally, a finely tuned addition of exogenous catalase abrogated the consumption reaction between H₂O₂ and NO, but allowed for peroxynitrite/H₂O₂-dependent singlet oxygen generation, catalase inactivation and ROS-mediated apoptotic signaling. The knowledge of these biochemical interactions should allow for establishment of novel therapeutic approaches that are based on the re-establishment of tumor cell apoptotic ROS signaling after catalase inactivation. A large body of knowledge on potentially interesting synergistic effects within the signaling network presented here has been worked out and will be presented elsewhere. The establishment of synergistic effects should allow the concentrations of the interactive compounds to be reduced to a small percentage of the concentration required for application of any single compound alone. Application of synergistic effects might optimize the specific antitumor effects and thereby mimimize unwanted side-effects on normal tissue.

NO-Mediated Catalase Inactivation Based on Singlet Oxygen Generation

The knowledge on the multiple functions of catalase and the multitude of chemical interactions related to singlet oxygen, as outlined in the previous figures, allowed speculations on NO-mediated induction of tumor cell sensitivity for intercellular signaling, despite the fact that the NO/peroxynitrite pathway is efficiently controlled by catalase. As summarized in Figure 12, an increase in the free NO concentration (Ki=0.18 μM) causes a transient inhibition of catalase (115). This situation should lead to less decomposition of H₂O₂ and peroxynitrite and therefore these two molecules now may interact and form singlet oxygen. Experimental biochemical evidence (as outlined in Figures 9, 10 and 11) allows the prediction that this should lead to APO/FAS receptor and caspase-8 activation, induction of NOX1 and NOS, and the subsequent generation of more singlet oxygen, based on the relative abundance of H₂O₂ and peroxynitrite. As a result, catalase should be sufficiently inactivated to allow ongoing intercellular ROS–mediated apoptotic signaling. The following figures demonstrate essential parts from the experimental verification of this concept. The addition of arginine represents a simple way to increase the synthesis of NO in tumor cells. As shown in Figure 13, addition of arginine to the human lymphoma cell line Gumbus causes apoptosis in a concentration-dependent mode. Addition of the peroxynitrite decomposition catalyst FeTPPS, the singlet oxygen-scavenger histidine, the NOS inhibitor L-NAME, or the HOCl-scavenger taurine immediately before arginine addition, caused nearly complete inhibition of arginine-mediated induction of apoptosis (Figure 13A). When histidine was added one hour after arginine, there was no inhibition at all, indicating that the singlet oxygen-dependent step was restricted to the initial phase of this process (Figure 13B). Addition of L-NAME or FeTPPS one hour after arginine resulted in strong inhibition of apoptosis in the low concentration range of applied arginine and decreasing inhibition at higher arginine concentrations. This finding is consistent with the interpretation that there is an early NO- and peroxynitrite-dependent step at all arginine concentrations and a later NO/peroxynitrite-dependent step only in the assays containing lower concentrations of arginine initially. Finally, the inhibitory effect of taurine was independent of the time of addition of the inhibitor, indicating that HOCl-dependent signaling represented a late effect at all applied arginine concentrations. Together with further control experiments that directly illuminated the inactivation of catalase in this scenario, the following model can be suggested: The increase of NO after arginine addition causes transient inhibition of catalase and subsequent peroxynitrite-dependent singlet oxygen generation through peroxynitrite/H₂O₂ interaction. This rather fast reaction causes inactivation of protective membrane-bound catalase and subsequent intercellular ROS-mediated apoptotic signaling. When lower concentrations of arginine were applied, the NO/peroxynitrite and the HOCl pathway acted co-operatively in the induction of apoptosis. Therefore, inhibition of one of the two pathways alone, even at one hour after the addition of arginine, caused nearly complete inhibition of apoptosis. At higher concentrations of initial arginine, late intercellular ROS signaling leading to cell death was nearly completely dependent on HOCl signaling. As indicated before (Figure 3), the level of available NO is also controlled by NOD, an enzyme that convertes NO into nitrate (109-112). Initially found as a microbial enzyme with profound effects on the resistance of microbes to NO-mediated attack by monocytes and neutrophils, NOD has been also described to be present in mammalian cells, including tumor cells (109-111). Several inhibitors have been described for NOD, including certain azoles (some of them with antifungal activity), quercetin and garlic extract (111). Figure 14 demonstrates that addition of itraconazole to transformed cells in the presence of an NO donor (DEA NONOate) and an inhibitor of NOS (L-NAME), caused a dramatic increase in the efficiency of NO-mediated, peroxynitrite-dependent apoptosis induction. This effect is well explained by the stabilizing effect on NO through inhibition of NOD. It is definitely not due to a modulation of intracellular NO synthesis, as it depends on exogenously added NO and not on active NOS. An increase in the activity of NOX1 rather than an increase in free NO, through NOD inhibition, would also cause an increase in the efficiency of peroxynitrite-mediated apoptosis induction. However, the curves obtained through NOD inhibition, compared to superoxide anion induction are quite different: at low concentrations of the NO donor, NOD inhibition results in marked enhancement of apoptosis induction, whereas superoxide anion induction may result in a decrease of apoptosis induction due to overwhelming consumption of NO by hydrogen peroxide. At high concentrations of the NO donor, NOD inhibition only caused marginal effects as seen in Figure 14, whereas induction of superoxide anion synthesis resulted in a substantial increase in apoptosis induction. NOD inhibition by azoles seems to be sufficient to induce apoptosis in human tumor cells, as shown for the human lymphoma cell line Gumbus, treated with the antifungal azole miconazole (Figure 15). Apoptosis induction was dependent on the concentration of miconazole. Again, the inhibitor profile and the window of action of certain inhibitors, indicates that miconazole causes an early, NO–, peroxynitrite- and singlet oxygen-dependent step (directly defined in parallel control experiments as catalase inactivation) and a subsequent intercellular signaling reaction that is characterized by the co-operation between the NO/peroxynitrite and the HOCl signaling pathways. The knowledge on ROS-mediated interactions during singlet-oxygen formation and singlet oxygen-dependent catalase inactivation, in combination with the multiple pathways induced by the molecular players whose action potentials are no longer suppressed after catalase inactivation, allowed several alternative pathways to be experimentally defined. They are based on the increase in free NO concentration. These pathways cause ROS-related amplification steps that finally lead to singlet oxygen-dependent catalase inactivation. This scenario is summarized in Figure 16: The concentration of NO can be alternatively increased by the addition of arginine, inhibition of arginase, induction of NOS (e.g. by interferon γ), inhibition of NOD or cytochrome P450-dependent oxido-reductase and even through abrogation of the consumption reaction between hydrogen peroxide and NO. The latter mechanism requires a fine-tuned catalase concentration that is sufficiently high to abrogate NO consumption, but still too low to successfully decompose peroxynitrite site-specifically, at the cell membrane. All of these experimental procedures that cause an increase of NO also cause initial singlet oxygen formation, autoamplification of singlet oxygen formation (through APO/FAS receptor, caspase-8, NOX1 and NOS), as demonstrated in the previous figures, throughout this article. They finally lead to catalase inactivation and subsequent apoptotic cell death due to release of intercellular ROS signaling from negative control by catalase.

The Significance of ROS-driven Catalase Inactivation and Subsequent Apoptotic ROS Signaling: Potential for a Highly Specific Approach in Tumor Therapy

Targeting tumor cell-specific membrane-associated catalase for selective apoptosis induction should allow establishment of very selective and efficient therapeutic strategies that are dependent on two specific phenotypic features of tumor cells, namely extracellular superoxide anion production, through NOX1 and the presence of membrane-associated catalase. The therapeutic use of antibodies directed towards catalase allows for targeting specifically membrane-associated catalase, the characteristic feature of tumor cells, and should have no effect on normal tissue as the antibodies are not cell-permeable. This first selective step in the use of an antibody against catalase is linked to a second selective step, as only malignant cells with sustained NOX1 activity have the potential to respond with apoptotic ROS signaling, driven by their own superoxide anions and subsequent HOCl and/or NO/peroxynitrite signaling that causes their own cell death. The suggested therapeutic use of compounds that increase free NO concentration (Figures 12, 13, 14, 15 and 16), has three levels of control with respect to selectivity for tumor cells: i) the initial step of extracellular singlet oxygen generation requires sustained superoxide anion production and is therefore restricted to malignant cells with activated oncogenes; ii) singlet oxygen generation takes place on the outside of the cells and therefore preferentially affects membrane-associated catalase, the major controlling factor of tumor cells in apoptotic ROS signaling. Intracellular catalase (essential in malignant and nonmalignant cells) with its protective function against hydrogen peroxide generated by mitochondria or certain metabolic pathways is not affected; iii) singlet oxygen-mediated inactivation of catalase only has an apoptosis-mediating effect on cells that show sustained NOX1 activity, which is characteristic of malignant cells with activated oncogenes. Based on the established knowledge about the interconnected levels of ROS signaling between malignant cells, several modes of synergistic interactions have been recently defined. These may be instrumental for further optimization and reduction of the necessary concentrations of individual compounds, thus mimimizing the risk of unwanted side effects on normal tissue. These aspects will be presented elsewhere.

Dynamic Aspects of ROS-driven Catalase Inactivation and Subsequent Apoptotic ROS Signaling

Targeting tumor cell-specific catalase for selective apoptosis induction not only represents a mechanism-based, rational approach for selective destruction of malignant cells, it also has a valuable dynamic aspect. Ongoing experiments indicate that the inactivation of catalase in a small group of cells within a population of tumor cells causes a bystander effect-like spread of catalase inactivation and subsequent apoptotic ROS signaling. This process is driven by peroxynitrite/H₂O₂ interaction in the vicinity of the catalase-negative cells, leading to singlet oxygen generation and subsequent action of singlet oxygen on neighboring catalase-positive tumor cells (Bauer, in preparation). In this way, catalase inactivation and ROS signaling, finally occur in the whole population of cells. The knowledge of the exact biochemical parameters of this enhancing effect may be useful to optimize future therapeutic approaches based on ROS-dependent signaling in tumor cells.

ROS Signaling and Tumor Cell Dormancy

Tumor cell metastasis followed by dormancy, enharboring the potential for later reactivation of tumor cell proliferation represents one of the challenging problems of tumor therapy. The established knowledge of ROS-mediated signaling in malignant cells, as presented here, allows a rather concise working hypothesis: Suppression of NOX1 activity or scavenging of NOX1-derived superoxide anions and H₂O₂ by a dominant antioxidant milieu might prevent tumor cell proliferation and expression of the malignant phenotype. As catalase expression in the membrane of tumor cells seems to be regulated by constant H₂O₂ signals (Bauer, unpublished findings), these dormant tumor cells most probably would not exhibit strong catalase expression in situ and thus would neither be easily detectable by specific anti-catalase staining nor be subject to therapy based on targeting catalase. Elimination of dormant tumor cells of this phenotype would require enhancement of NADPH oxidase activity and reduction of antioxidant control. As a result, dismutation of superoxide anions might generate sufficient H₂O₂ for positive feedback on catalase expression. This would then allow targeting of tumor cells with an antibody directed towards catalase, or with ROS-driven singlet oxygen generation, followed by apoptotic ROS signaling. The resulting positive therapeutic effect of this complex scenario would however be counterbalanced by the H₂O₂-dependent proliferation stimulus of the tumor cells. This counterbalance needed to be outweighed by optimal ROS signaling for an overall removal of malignant cells through apoptosis induction.

ROS Signaling and Tumor Prevention

The focus of this article is on the utilization of specific ROS signaling of malignant cells for novel therapeutic approaches. Nevertheless, the tumor-preventive potential of intercellular ROS-mediated apoptotic signaling should not be completely neglected in this context. As outlined in Figure 1, ROS signaling of transformed cells (without sufficient catalase protection) has been suggested as representing an early control step in oncogenesis (21, 30). This ROS-driven control step thus might counterbalance the multiple pro-carcinogenic effects of ROS in tumor initiation and promotion. Mathematical modeling is in favour of this conclusion (85). The results obtained in vivo by the group of Deichman (12-16) are also in favor of this assumption, as they are best explained by selection of malignant cells with protective catalase and parallel elimination of catalase-negative transformed cells. The regular finding of membrane-associated catalase in human tumor cell lines further supports this concept. Based on the signaling chemistry presented in Figures 12, 13, 14, 15 and 16, it may be concluded that ROS-dependent elimination of malignant cells in vivo should not be necessarily restricted to early stages of transformed cells without sufficient catalase protection, but might still become effective on microtumors with established catalase protection. Natural compounds that modulate the concentration of free NO might induce singlet oxygen generation, catalase inactivation and ROS-mediated apoptosis, in analogy to the mechanisms described in Figures 12, 13, 14, 15 and 16. An ongoing survey in our laboratory has shown that many secondary plant compounds indeed exhibit the potential to modulate free NO concentration and do show the predicted effects on tumor cells in vitro. Further characterization of these effects may be useful for our understanding of tumor prevention by nutrition and also may aid in drug development for the establishment of novel therapeutic approaches.