The Anticancer Drug Ellipticine Activated with Cytochrome P450 Mediates DNA Damage Determining Its Pharmacological Efficiencies: Studies with Rats, Hepatic Cytochrome P450 Reductase Null (HRN™) Mice and Pure Enzymes

Ellipticine is a DNA-damaging agent acting as a prodrug whose pharmacological efficiencies and genotoxic side effects are dictated by activation with cytochrome P450 (CYP). Over the last decade we have gained extensive experience in using pure enzymes and various animal models that helped to identify CYPs metabolizing ellipticine. In this review we focus on comparison between the in vitro and in vivo studies and show a necessity of both approaches to obtain valid information on CYP enzymes contributing to ellipticine metabolism. Discrepancies were found between the CYP enzymes activating ellipticine to 13-hydroxy- and 12-hydroxyellipticine generating covalent DNA adducts and those detoxifying this drug to 9-hydroxy- and 7-hydroellipticine in vitro and in vivo. In vivo, formation of ellipticine-DNA adducts is dependent not only on expression levels of CYP3A, catalyzing ellipticine activation in vitro, but also on those of CYP1A that oxidize ellipticine in vitro mainly to the detoxification products. The finding showing that cytochrome b5 alters the ratio of ellipticine metabolites generated by CYP1A1/2 and 3A4 explained this paradox. Whereas the detoxification of ellipticine by CYP1A and 3A is either decreased or not changed by cytochrome b5, activation leading to ellipticine-DNA adducts increased considerably. We show that (I) the pharmacological effects of ellipticine mediated by covalent ellipticine-derived DNA adducts are dictated by expression levels of CYP1A, 3A and cytochrome b5, and its own potency to induce these enzymes in tumor tissues, (II) animal models, where levels of CYPs are either knocked out or induced are appropriate to identify CYPs metabolizing ellipticine in vivo, and (III) extrapolation from in vitro data to the situation in vivo is not always possible, confirming the need for these animal models.


Introduction
A plant alkaloid ellipticine (5,11-dimethyl-6H-pyrido [4,3-b]carbazole, Figure 1) found in several Apocynaceae plants and its derivatives are efficient anticancer compounds that function through multiple mechanisms participating in cell cycle arrest and the initiation of apoptosis (for a summary see: [1][2][3][4][5][6]). Ellipticine was found (I) to arrest cell cycle progression due to modulation of levels of cyclinB1 and Cdc2, and phosphorylation of Cdc2 in human mammary adenocarcinoma MCF-7 cells [7]; (II) to initiate apoptosis by several mechanisms such as formation of reactive oxygen species (ROS) inducing DNA damage, the activation of mitogen-activated protein kinases (MAPKs), release of cytochrome c and apoptosis-inducing factor (AIF) from the mitochondrial membrane, caspase activation as well as a caspase-independent pathway [8,9], triggering of Fas/Fas ligand pathway and modulation of proteins of the Bcl-2 family in several tumor cell lines [10]; (III) to disrupt mitochondrial function and (IV) to cause the apoptotic signaling that is amplified by cross-talk between the Fas death receptor and the mitochondrial apoptotic pathway (for a summary see: [3,4,[7][8][9][10]). Several studies also demonstrated that the p53 tumor suppressor protein is involved in ellipticine-mediated induction of cell cycle arrest and apoptosis [9][10][11][12][13][14][15][16][17][18][19][20]. Ellipticine inhibits p53 protein phosphorylation by a selective inhibition of CDK2 kinase in Lewis lung carcinoma and the human colon cancer cell line SW480 [11], and this effect on p53 correlated with cytotoxic activity of ellipticine [11]. Treatment of Saos 2 cells transfected with mutant p53 with ellipticine restored the transactivation function of p53, resulting in the induction of p53-responsive p21 Waf1 and MDM2 genes at protein levels and activation of a p53-responsive luciferase reporter [15]. The results found in the study of Sugikawa and coworkers [15] indicate that ellipticine induces a shift of mutant p53 conformation towards wild-type and this activity is not caused by its function as an inhibitor of topoisomerase II, which is one of the DNA-damaging effects of ellipticine (for a summary see [1][2][3][4][5][6]). More importantly, ellipticine can even activate mutant p53 and induces p21 Waf1 and MDM2 gene expression in vivo, in nude mouse tumor xenografts [16]. Moreover, Wang et al. [19] demonstrate that in mutant-p53 lymphoma cells, ellipticine-mediated reactivation of mutant p53 sensitizes these cells to treatment with further DNA-damaging drugs (i.e., doxorubicin).
Ellipticine elevated the nuclear localization of endogenous p53 and exogenous mutant p53 in HCT116 colon cancer cells leading to transactivation of the p21 promoter. Nuclear localization of p53 is frequently the consequence of a genotoxic stress by compounds inducing DNA damage (i.e., inhibitors of topoisomerase II) [21]. The ellipticine-mediated abundance of nuclear p53 was not associated with an increase in DNA double strand breaks. Therefore, this effect of ellipticine seems not to be dependent on the mechanism mediated by topoisomerase II inhibition, but on another genotoxic stress [22]. Further, ellipticine induced nuclear translocalization of p53 and of the serine/threonine kinase Akt (an enzyme providing a survival signal protecting cells from stress-induced apoptosis) and recruitment of autophagosomes in human non-small cell lung cancer (NSCLC) epithelial cells A549 [18]. Akt-related cell death also occurred in p53-deficient cells with stable expression of exogenous p53. Hence, as a DNA-damaging agent, ellipticine is a regulator in autophagy-related cell death by cooperation of p53 and Akt [18]. Ellipticine also activates the p53 pathway in glioblastoma cells; its impact on these cancer cells depends on the p53 status [14]. In a U87MG glioblastoma cell line expressing wild-type p53, ellipticine provoked an early G0/G1 cell cycle arrest, whereas in a U373 cell line expressing a p53 mutant it induced arrest in S and G2/M phases of the cell cycle [14].
All studies investigating the mechanism of ellipticine antitumor action indicate complex pathways leading to cancer cell death by this drug. Chemotherapy-induced cell cycle arrest and induction of apoptosis were shown to frequently result from DNA damage caused by exposure to a variety of chemotherapeutics including ellipticine. In addition, genotoxic stress as a result of multiple DNA-damage increases levels of nuclear p53 [21,22], the tumor suppressor protein shown to be involved in ellipticine-mediated induction of cell cycle arrest and apoptosis [10][11][12][13][14][15][16]. These findings suggest that DNA damage by ellipticine is crucial for its cytotoxic effects.

DNA-Damaging Mechanisms of Ellipticine Cytotoxicity to Cancer Cells
The most important DNA-damaging mechanisms of ellipticine were considered to be intercalation into DNA [5,[23][24][25][26] and inhibition of DNA topoisomerase II activity [5,27,28]. Recently, Andrews and co-workers [29] demonstrated, however, that ellipticine and some of its derivatives are potent and specific inhibitors of RNA polymerase I (Pol-I) transcription and that this Pol-I inhibition occurs by a p53-and topoisomerase II-independent mechanism. They found that the drug influences the assembly and stability of preinitiation complexes by targeting the interaction between promoter recognition essential transcription factor SL1 and the rRNA promoter. In addition, Ghosh et al. [30] showed that along with DNA intercalation and/or topoisomerase II inhibition, interaction with the telomeric DNA region and the resultant inhibition of telomerase activity might be an additional mode of action of ellipticine. Moreover, we showed that this antitumor agent also causes damage to the structural integrity of DNA through covalent binding, by forming covalent DNA adducts after its enzymatic activation with cytochrome P450 (CYP) or peroxidases [1][2][3][4][31][32][33][34][35][36][37][38][39][40]. Cytotoxicity of ellipticine in cells of several cancer lines sensitive to this drug such as HL-60 promyelocytic leukemia, T-cell leukemia CCRF-CEM, glioblastoma U87MG, neuroblastoma UKF-NB-3 and UKF-NB-4, thyroid cancer BHT-101, B-CPAP and 8505-C and breast adenocarcinoma MCF-7 corresponded to levels of ellipticine-derived DNA adducts generated after its enzymatic activation in most of these cell lines [41]. This indicates that covalent binding to DNA of reactive species generated by enzymatic bioactivation of ellipticine is one of the most important mechanisms responsible for ellipticine cytotoxicity in these cancer cells. The formation of ellipticine-DNA adducts ultimately forces cancer cells to initiate cell death signaling [9]. Based on these results, we suggest that ellipticine acts as a prodrug, which is metabolically activated to reactive species forming covalent DNA adducts causing genotoxic stress. Therefore, information on which enzymes are involved in the metabolism of ellipticine is critical to identify the pharmacological effects of ellipticine. Several in vitro and in vivo approaches have been developed to study the role of specific CYP and peroxidase enzymes in ellipticine metabolism.
In this review we focus on comparison between the data found in the in vitro and in vivo studies investigating ellipticine metabolism and show a necessity of both approaches to obtain valid information on CYP enzymes participating in this process.
Using a variety of human recombinant CYPs, inhibitors of these enzymes in human hepatic microsomes and correlation analyses, the roles of individual CYPs in the formation of ellipticine metabolites was identified [31,34,35,40]. Human recombinant CYP1A1 and 1A2, followed by CYP1B1, are most effective in formation of 7-hydroxy-and 9-hydroxyellipticine detoxifying ellipticine ( Figure 3) [31]. The active metabolite, 13-hydroxyellipticine, forming the ellipticine-DNA adduct 1 ( Figure 2C), is generated predominantly by CYP3A4. Oxidation of ellipticine to another activation metabolite, 12-hydroxyellipticine, generating DNA adduct 2 ( Figure 2D) is also catalyzed by CYP3A4, but more efficiently by CYP2C19. The N 2 -oxide of ellipticine is generated mainly by CYP2D6 beside CYP3A4 [31,34,35,39]. These results demonstrate that CYP3A4 is the most effective enzyme leading to ellipticine-DNA adducts 1 and 2, while adduct 2 is also generated by CYP2C19 and 2D6. Moreover, orthologous CYP enzymes of rats and mice catalyze formation of these metabolites and DNA adducts [1][2][3][4]36,38]. This indicates that these animals might be suitable models mimicking the fate of ellipticine in human.
Recently, we could show that levels of the DNA adduct formed by 13-hydroxyellipticine increased if this ellipticine metabolite was conjugated with sulfate or acetate by human sulfotransferases 1A1, 1A2, 1A3 and 2A1, or N,O-acetyltransferases 1 and 2 ( Figure 3) [34,58].

Oxidation of Ellipticine by CYP Enzymes in Vivo
In order to extrapolate from the in vitro data to the in vivo situation, additional factors have to be considered such as route of administration, absorption, renal clearance, and tissue-specific expression of enzymes metabolizing ellipticine. To identify CYP enzymes responsible for activation of ellipticine in vivo, several animal models were used: (I) wild-type (WT) and Hepatic P450 Reductase Null (HRN) mice; (II) the same mouse models, in which expression of enzymes of the mixed-function oxidase system was induced by benzo[a]pyrene (BaP); and (III) Wistar rats. The data obtained with these animal models revealed a paradox: namely, that CYP1A enzymes appear to be more important for activation of ellipticine in vivo, despite being involved in its metabolic detoxification in vitro.

Utilization of Wild-Type (WT) and Hepatic P450 Reductase Null (HRN) Mice to Identify Enzymes Metabolizing Ellipticine in Vivo
In HRN mice, POR, the most important electron donor to mouse CYPs, is deleted specifically in hepatocytes. This model was developed by Henderson et al. [62] to evaluate the role of both hepatic POR and CYPs in xenobiotic metabolism. Deletion of this enzyme results not only in the loss of essentially all hepatic CYP function, but also in the lack of direct reduction of xenobiotics by POR, an additional property of this enzyme. This mouse model has been used successfully to investigate the role of hepatic versus extra-hepatic CYP-catalyzed metabolism and the disposition of several carcinogens and drugs including ellipticine [48,[62][63][64][65][66][67][68][69].
Using WT and HRN mouse lines, hepatic CYPs were demonstrated to be important in ellipticine-derived DNA adduct formation also in vivo, because up to seven ellipticine-specific DNA adducts were observed in liver, lung, kidney, spleen, colon and bladder (see Figure 2E,F for liver of WT and HRN mice). Deoxyguanosine adduct spots 1 and 2 derived from 13-hydroxy-and 12-hydroxyellipticine, respectively (Figures 2 and 3), were the predominant adducts in all mouse tissues examined.
The finding that ellipticine-DNA adducts are formed in all organs tested in these animals suggest that ellipticine or its metabolites are distributed via the blood stream to different organs and that these tissues may have the metabolic capacity to oxidatively activate ellipticine. As found by Chadwick and co-workers [52,53] ellipticine is very rapidly distributed from the blood, and its excretion is essentially complete by 24 h in several species including mice, rats, dogs, and monkeys. The rate of ellipticine elimination from blood was found to reflect the rate of metabolism of this drug [52]. The main organ responsible for its biotransformation was found to be the liver, forming predominantly 9-hydroxyellipticine, which is excreted mainly in bile as its glucuronide or sulfate conjugate [52,53]. Other in vivo pathways involving hydroxylation at as yet unknown positions in the molecule have also been found [52,53]. As mentioned above, in in vitro experiments, ellipticine is oxidized by CYPs in hepatic microsomes from a variety of species, including humans, rats, rabbits and mice [31,39,48,49,60,61] to several hydroxylated derivatives, with 9-hydroxy-, 12-hydroxy-and 13-hydroxyellipticine as the major metabolites in most species. However, because 13-hydroxy-and 12-hydroxyellipticine are reactive and have been found to form the two major ellipticine-DNA adducts [31,32,39,47], they will not be easily detectable in vivo. In addition, in these early studies radioactively labelled ellipticine was found to be deposited in a number of organs with the highest levels in the liver, followed by kidney, lung, intestine and spleen, and was located primarily in the nuclear fraction [52]. Our more recent data would suggest that covalent binding of ellipticine to DNA can explain this localization [32,33,48].
The levels of ellipticine-DNA adducts in the livers of HRN mice were lower (by up to 65%) than those in WT mice, demonstrating that CYP enzyme activity is important for the oxidative activation of ellipticine to metabolites generating these adducts. Whereas hepatic CYP-mediated ellipticine DNA binding was reduced in HRN mice, adduct levels in extrahepatic organs were up to 4.7-fold higher ( Figure 5).
These tissues therefore have the metabolic capacity to oxidize ellipticine and, more importantly, the same reactive species forming DNA adducts are produced as in the liver, probably by both CYP catalysis and maybe peroxidases. Experiments utilizing ex vivo incubations of ellipticine with hepatic microsomes of WT and HRN mice and those of these mice exposed to BaP, as well as employing inhibitors of the most important CYP enzymes catalyzing detoxification and activation of ellipticine in vitro, CYP1A and 3A, respectively, helped to resolve which of these CYPs play a role in the mouse models. As expected, treatment of WT and HRN mice with BaP significantly induced expression of CYP1A, predominantly of CYP1A1 in liver (up to 175-fold), both at the transcriptional and translation levels [61,65]. This carcinogen also increased the expression of POR protein and its enzymatic activity in livers of these mouse models, but to a much lower extent, up to 2.9-fold. More interestingly, exposure of WT and HRN mice to BaP was also found to result in an increased expression of cytochrome b5, a protein of the microsomal mixed-function-oxidase system, in livers of these mice [66].
In the presence of NADPH, a cofactor of POR-and CYP-dependent enzyme systems, the ex vivo incubations with ellipticine, DNA and hepatic microsomes of untreated (control) WT and HRN mice and mice treated with BaP led to activation of this drug to ellipticine-derived DNA adducts (Figures 2A,B  and 6), confirming the role of CYPs in ellipticine activation. Arachidonic acid, a cofactor for COX-dependent oxidation [48,64,[70][71][72], also mediated formation of ellipticine-DNA adducts 1 and 2 in hepatic microsomes of all mice used. This suggests that COX also activates ellipticine in mouse liver [48,61], but arachidonic acid as a cofactor was much less effective than NADPH ( Figure 6). Surprisingly, levels of ellipticine-derived DNA adducts formed in the ex vivo incubations of HRN mice liver microsomes with NADPH were only 1.4-fold lower than amounts formed by hepatic microsomes from WT mice ( Figure 6), even though POR expression in livers of HRN mice was two orders of magnitude lower. This finding indicates that ellipticine activation should be at least partially catalyzed also by enzymes with POR-independent activity [48]. Beside peroxidases that were found to activate ellipticine [32], the CYP2S1 enzyme, which is abundantly expressed in several tissues [73][74][75][76], might be such an enzyme, because it catalyzes the oxidation of compounds having polycyclic aromatic structures similar to ellipticine without participation of POR [75,76]. Whereas a role of a COX peroxidase in hepatic microsomes of WT and HRN mice was proven (see Figure 6) [48,61], the participation of CYP2S1 in ellipticine activation awaits further examination. Therefore, the human recombinant CYP2S1 enzyme heterologously expressed in Escherichia coli was prepared in our laboratory [77] and will be utilized to investigate efficiency of this CYP in ellipticine oxidation in an additional study.
At least two adducts (spots 1 and 2 in Figure 2A,B), which were identical to those generated in vivo in mice treated with ellipticine ( Figure 2E,F) were formed by mouse hepatic microsomes. Furthermore, ellipticine-derived DNA adduct, spot A, was found as a minor adduct (Figure 2A), predominantly in microsomes isolated from HRN mice [48]. In incubations containing hepatic microsomes of WT and HRN mice treated with BaP, an additional adduct spot, corresponding to the 10-(deoxyguanosin-N 2 -yl)-7,8,9-trihydroxy-7,8,9,10-tetrahydrobenzo[a]pyrene (dG-N 2 -BPDE) adduct of BaP-7,8-dihydrodiol-9,10-epoxide with DNA in vitro and in vivo [65,78] was also detected ( Figure 2B). This finding indicates that residual BaP is present in microsomes isolated from livers of WT and HRN mice treated with BaP, and is activated by CYP1A1 in combination with microsomal epoxide hydrolase to form this adduct.
Ketoconazole, a selective inhibitor of CYP3A [79,80], inhibited formation of ellipticine-DNA adducts in hepatic microsomes of untreated (control) WT and HRN mice, by ~60% ( Figure 6A), confirming a role of CYP3A in ellipticine activation in mouse liver. However, the effect of this inhibitor was much lower in hepatic microsomes of BaP-treated WT and HRN mice, only by ~10% ( Figure 6B). In mice exposed to BaP the contribution of CYP3A is much lower, because of the massive CYP1A induction by BaP. Surprisingly, this increased level of CYP1A, the enzymes that mainly detoxify ellipticine in vitro, led to higher amounts of ellipticine-DNA adducts formed (Figure 6), predominantly of adduct 1 [61]. Moreover, a selective inhibitor of CYP1A activities, α-naphthoflavone (α-NF) [79], inhibited formation of ellipticine-DNA adducts in all mice except HRN mice exposed to BaP (Figure 6). Induction of CYP1A in HRN mice by BaP also resulted in increased levels of ellipticine-DNA adducts, but α-NF caused an increase rather than a decrease in formation of ellipticine-DNA adducts ( Figure 6). Therefore, here the BaP induced CYP1A enzymes seem to increase ellipticine detoxification.
Exposure of mice to BaP induced CYP1A and resulted, as expected, in an increase in formation of 9-hydroxy-and 7-hydroxyellipticine (Figure 7). This result is consistent with previous studies where CYP1A1 and 1A2 were the major enzymes forming these metabolites [31,35,40]. Treatment of WT mice with BaP, however, also resulted in up to 2.5-fold higher levels of 13-hydroxy-and 12-hydroxyellipticine (Figure 7), the metabolites that were found to be formed in vitro mainly by CYP3A, and much less efficiently by CYP1A1 [31,39]. Hence not only CYP3A, but also CYP1A expressed in mouse liver are important for activation of ellipticine to ellipticine-DNA adducts in WT mice only, while in HRN mice the detoxification only is induced by BaP explaining the DNA adduct levels. These findings show that in contrast to the pure in vitro CYP systems (CYPs reconstituted with POR), CYP1A enzymes are responsible for ellipticine activation to form DNA adducts in mouse liver [48,61].

Ellipticine Metabolism in Wistar Rats
In order to further identify CYP enzymes responsible for activation and detoxification of ellipticine in vivo, Wistar rats were used. Also in this animal model, ellipticine treatment resulted in ellipticine-derived DNA adduct generation in several healthy organs (liver, kidney, lung, spleen, breast, heart and brain) (see Figure 2G for rat liver) [3,33,37] and in DNA of mammary adenocarcinoma [3]. The levels of ellipticine-derived DNA adducts generated in these adenocarcinomas were almost 2-fold higher than in normal healthy mammary tissue. This finding indicates that other CYPs such as CYP1B1, able to activate ellipticine may be expressed at higher levels in this adenocarcinoma [41] than in peritumoral tissues. Indeed, we [41], and others [81][82][83] previously showed CYP1B1 to be a typical CYP expressed in breast cancer.
Genotoxic side effects of ellipticine in healthy organs of experimental animals in vivo may of course also be determined by expression levels of CYP enzymes activating ellipticine in these tissues [33,37,48]. Indeed, several studies have found a positive correlation between DNA adduct levels of carcinogens or genotoxic agents, their persistence and their mutagenicity and/or tumorigenicity [84][85][86][87][88]. To better understand the role of ellipticine-DNA adducts in genotoxic side effects in healthy tissues, we have analyzed the persistence of ellipticine-DNA adducts in liver, lung, kidney, spleen, heart, and brain of rats to model the bioactivation of ellipticine in humans treated with ellipticine [33]. Only very low levels of adducts persisted only in some tissues. In addition, not all ellipticine-DNA adducts persist in the tissues analyzed in the study (only adducts 1, 2, 4, and 5) [33]. This finding demonstrates that healthy tissues of rats treated with ellipticine possess effective DNA repair systems to remove certain lesions and suggests a relatively low impact of the genotoxic side effects of ellipticine during cancer treatment in humans.
Also in rats, formation of ellipticine-DNA adduct 1 is dependent not only on levels of CYP3A, but also on those of CYP1A1. The levels of ellipticine DNA adduct 1 in analyzed organs correlated not only with expression levels of CYP3A, but also with those of CYP1A1 in the same organs, which again does not correspond to the situation in vitro. As outlined above, in the in vitro systems, CYP3A is mainly responsible for formation of ellipticine-DNA adducts and CYP1A predominantly oxidizes ellipticine to its detoxification metabolites [3,4,36]. In-vitro studies investigating the effect of cytochrome b5 on the metabolism of ellipticine explained the discrepancies between CYP oxidation of ellipticine in vitro and in vivo, because this protein has a crucial role in directing individual CYP enzymes to ellipticine activation or detoxification. Moreover, the induced expression of cytochrome b5 protein in liver of rats treated with ellipticine [35,51] suggests that cytochrome b5 may modulate the CYP-mediated bioactivation and detoxification of ellipticine in this animal model in vivo as well.

Studies with Human Microsomes and Cancer Cells
Because in the studies described above with isolated enzymes the ratios of the various partners of CYP are generated experimentally, a more physiological model to identify human enzymes responsible for ellipticine activation was used, namely, human hepatic microsomes [31,39]. These microsomal fractions contain a mixture of human CYPs, POR, cytochrome b5 and its reductase (NADH:cytochrome b5 reductase). Thus, they comprise the essential components of the enzymatic system metabolizing drugs, mimicking well a situation in human liver, where a majority of drug metabolism occurs [78,95]. Human hepatic microsomes oxidize ellipticine mainly to 13-hydroxy-and 12-hydroxyellipticine, whereas 7-hydroxy-, 9-hydroxyellipticine and ellipticine N 2 -oxide are generated at more than 10-fold lower amounts (see table 2 in [39]). Similar results were found in hepatic microsomes of rats [60]. As a consequence, high levels of both major ellipticine-DNA adducts are formed when DNA is added to the microsomal incubations. The amounts formed correlated with the activity of the major CYPs found previously to form the metabolites generating these DNA adducts (see above) [31,39].
All these results explained why the CYP1A enzymes are more important in ellipticine activation in vivo; their activity is modulated by cytochrome b5. These results also demonstrated that not only the expression levels of CYP1A and 3A in several species including human, but also the amounts of expressed cytochrome b5 dictate the oxidative activation and detoxification of ellipticine in vivo. Moreover, because ellipticine itself is capable of inducing expression of CYP1A (predominantly CYP1A1), POR, and cytochrome b5 [51,96], it increases its own metabolism leading predominantly to activation of this drug to reactive species forming DNA adducts [34,35], thereby modulating its own pharmacological potential.
The ellipticine-derived DNA adducts formed by enzymatic activation of ellipticine in vitro and in vivo were also found in human tumor cells, in which CYP enzymes are expressed. The adducts were found in human adenocarcinoma MCF-7 [41], neuroblastoma IMR-32, UKF-NB-3, and UKF-NB-4 [43,44], glioblastoma U87MG [46], and BHT-101, B-CPAP and 8505-C thyroid cancer cells [45] exposed to ellipticine. Cytotoxicity of ellipticine corresponded to the amounts of ellipticine-DNA adducts formed in the specific cancer cells and depended on expression levels of CYP enzymes metabolizing ellipticine (CYP1A1, 1B1, and 3A4) and/or cytochrome b5 in these cells [36,41,[43][44][45][46]. High expression levels of cytochrome b5 together with those of CYP1A1 and 3A4 lead to more ellipticine-DNA adducts and higher cytotoxicity of ellipticine predominantly in neuroblastoma UKF-NB-4, glioblastoma U87MG and thyroid cancer cells [14,36,41,[43][44][45]. These findings again demonstrate the importance of expression of CYPs, but also of cytochrome b5, in the tumor and normal tissue, because the ratios of these enzymes determine the pharmacological effects of ellipticine.

Conclusions
The data summarized in this review demonstrate that the DNA-damaging anticancer alkaloid ellipticine might be considered a prodrug, whose major mechanism of action is mediated by an enzymatic activation leading to formation of covalent DNA adducts in target tissues. These ellipticine-DNA adducts are formed in both healthy and tumor tissues and cells, but they do not persist in healthy tissues. The data show that cytotoxic effects of ellipticine in tumor tissues are dictated by (I) levels of CYP expression (and/or peroxidase expression); (II) levels of cytochrome b5 expression; and (III) its own potency to induce CYP1A1, CYP3A and cytochrome b5 in tumor tissues and cells. The results also demonstrate that animal models, where levels of biotransformation enzymes either knocked out or induced, are appropriate tools to identify enzymes responsible for the metabolic activation and detoxification of ellipticine. They also demonstrate that extrapolation from in vitro data to the situation in vivo is not always possible, confirming the need for these animal models.
Even though the role of cytochrome b5 in modulation of ellipticine metabolism in vitro was clearly shown, its effect in vivo is still quite enigmatic. Two mouse lines, one with a conditional hepatic deletion of cytochrome b5 (HBN, Hepatic cytochrome b5 Null) [97] and a double conditional mutant, HBRN (Hepatic cytochrome b5/P450 Reductase Null), in which both enzymes are deleted specifically in the liver [98], which were recently developed in the laboratory of Wolf and coworkers [97,98], may help to resolve the role of cytochrome b5 in CYP-mediated metabolism of ellipticine in vivo.
The results summarized in this review form the basis to further predict the susceptibility of human cancers to ellipticine and suggest this alkaloid for treatment in combination with CYP gene transfer (CYP-gene-directed enzyme-prodrug therapy) [99,100], which has the potential to provide efficient activation of ellipticine in target tumor tissue, thereby increasing the anticancer potential of this prodrug. Furthermore, two of the ellipticine metabolites formed by oxidation with CYPs in combination with cytochrome b5, 13-hydroxy-and 12-hydroxyellipticine, are reactive enough to decompose spontaneously to the carbenium ions forming DNA adducts that are predominantly responsible for killing cancer cells. Both these ellipticine metabolites are, therefore, excellent candidates for tumor-specific targeting by appropriate derivatives, including their encapsulated forms into nanocarriers. Research into such targeted carrier systems for active ellipticine metabolites is a major research aim on the path to clinical application of ellipticine in tumor therapy.