Analysis of covalent ellipticine-and doxorubicin-derived adducts in DNA of neuroblastoma cells by the 32 P-postlabeling technique

Background. Ellipticine and doxorubicin are antineoplastic agents, whose action is based mainly on DNA damage such as intercalation, inhibition of topoisomerase II and formation of covalent DNA adducts. The key target to resolve which of these mechanisms are responsible for ellipticine and doxorubicin anticancer effects is the development of suitable methods for identifying their individual DNA-damaging effects. Here, the 32P-postlabeling method was tested to detect covalent DNA adducts formed by ellipticine and doxorubicin. Methods. The standard procedure of 32P-postlabeling assay, this procedure under ATP-deficient conditions, the version using extraction of adducts with n-butanol and the nuclease P1 enrichment version were used to analyze ellipticineand/or doxorubicin-derived DNA adducts. Results. Two covalent ellipticine-derived DNA adducts, which are associated with cytotoxicity of ellipticine to human UKF-NB-3 and UKF-NB-4 neuroblastoma cell lines, were detected by the 32P-postlabeling method. These adducts are identical to those formed by the ellipticine metabolites, 13-hydroxyand 12-hydroxyellipticine. In contrast, no covalent adducts formed by doxorubicin in DNA of these neuroblastoma cells and in DNA incubated with this drug and formaldehyde in vitro were detectable by the 32P-postlabeling assay. Conclusions. The results presented in this paper are the first to demonstrate that in contrast to covalent DNA adducts formed by ellipticine, the adducts generated by formaldehyde-mediated covalent binding of doxorubicin to DNA are not detectable by the 32P-postlabeling assay. No DNA adducts were, detectable either in vitro, in incubations of DNA with doxorubicin or in DNA of neuroblastoma cells treated with this drug. The results also suggest that covalent binding of ellipticine to DNA of UKF-NB-3 and UKF-NB-4 neuroblastoma cell lines is the predominant mechanism responsible for the cytotoxicity of this drug. To understand the mechanisms of doxorubicin anticancer effects on neuroblastoma cells, development of novel methods for identifying covalent doxorubicin-derived DNA adducts is the major challenge for further research.


INTRODUCTION
Neuroblastoma, a tumor of the peripheral sympathetic nervous system, is the most frequent solid extracranial neurological tumor in children and is a major cause of death from neoplasia in infancy 1,2 .These tumors are biologically heterogeneous, with cell populations differing in their genetic programs, maturation stage and malignant potential 3 .Neuroblastoma consists of three principal neoplastic cells 4,5 : i) neuroblastic or N-type: undifferentiated, round and small cells with scant cytoplasm and neuritic processes; ii) stromal or S-type: large, flattened and adherent differentiated cells; and iii) intermediate or I-type with morphological features of both above mentioned types i.e. cells with short neurite-like processes such as adherent growth.As neuroblastoma cells seem to have the capacity to differentiate spontaneously in vivo and in vitro 6 , their heterogeneity could affect treatment outcome, in particular the response to apoptosis induced by chemotherapy.Neuroblastoma may regress spontaneously in infants, mature to benign ganglioneuromas, or grow relentlessly and be rapidly fatal 3 .The prognosis for patients with high risk tumors is poor, in spite of intensive therapy including megatherapy with subsequent hematopoietic progenitor cell transplantation, biotherapy and immunotherapy because drug resistance arises in the majority of these patients who initially responded to chemotherapy 3 .
Chemotherapy agents used in combination have been found to be effective against neuroblastoma.Agents commonly used are platinum compounds (carboplatin), alkylating agents (cyclophosphamide, ifosfamide, melphalan), topoisomerase II inhibitors (etoposide), anthracycline antibiotics (doxorubicin) and vinca alkaloids (vincristine).Some novel regimens include also topoisomerase I inhibitors (topotecan and irinotecan), which have been found to be effective against recurrent disease 3 .Nevertheless, little improvement in therapeutic options has been made in the last decade, requiring a need for the development of new therapies.
Of anthracycline antibiotics, doxorubicin (Fig. 2A) is frequently used for neuroblastoma treatment 3 .The primary mechanism of action of doxorubicin (and other anthracyclines) appears to be poisoning of the enzyme topoisomerase II which results in double-strand DNA breaks, and the failure to repair these breaks leads to apoptosis 15,16 .More recently however, it has been demonstrated that doxorubicin also forms covalent adducts with DNA and these lesions are more cytotoxic than those induced by topoisomerase II impairment 17 .The adducts are formed predominantly at 50-GC-30 sites in DNA (ref. 18), where the doxorubicin sugar group (daunosamine) is covalently linked to the N-2 amino group of guanine via an aminal (N-C-N) bond [19][20][21][22] .The central carbon atom in the aminal bond is derived from formaldehyde, hence formaldehyde is an absolute requirement for adduct formation 20,21 .This compound is present in cancer cells, at levels which are often even higher in tumor cells (1.5-4.0 mM) than normal cells 23,24 .Moreover, formaldehyde is also produced in cancer cells treated with doxorubicin, from the oxidation of doxorubicin itself 20 .The resulting drug-DNA monoadduct is further stabilized through in- tercalation and hydrogen bonding with the second strand of DNA (ref. 21) (Fig. 2B).
Even though ellipticine and doxorubicin are cytotoxic to neuroblastomas 7 , the question arises whether the covalent DNA adduct formation is the predominant mechanism responsible for this cytotoxic effect.To answer this question, development of methods suitable for detecting and quantifying covalent DNA adducts formed by ellipticine and doxorubicin is the first and key target of the research.Using 14 C-labeled doxorubicin, doxorubicin-DNA adduct formation has been detected in several cancer cells 17,20,21,25 .Moreover, accelerator mass spectrometry has recently also been shown suitable for detecting the covalent 14 C-labeled doxorubicin-DNA adducts in cancer cells exposed to this drug [26][27][28] .Utilizing the 14 C-labeled doxorubicin has, however, limitations for use in human treatment.
0][31][32] ).However, whereas it is useful for detecting and quantifying covalent ellipticinederived DNA adducts 12,14,15,33,34 , its ability to determine covalent doxorubicin-DNA adducts remains to be resolved.The present study was therefore undertaken to investigate whether covalent doxorubicin-DNA adducts are detectable by the 32 P-postlabeling technique.DNA adducts formed by doxorubicin in DNA of neuroblastoma cells and in DNA incubated with this drug and formaldehyde were analyzed.Since neuroblastoma is heterogenous and this feature could affect its treatment, two types of neuroblastoma cell lines were tested for their response to treatment by ellipticine and doxorubicin, UKF-NB-3 cells (the invasive N-type), and UKF-NB-4 cells (the less-aggressive S-type).DNA from UKF-NB-3 and UKF-NB-4 cell lines treated with ellipticine and doxorubicin in concentrations that are toxic to these cells was isolated and formation of covalent DNA adducts by ellipticine and doxorubicin analyzed.

Chemical
Ellipticine was obtained from Sigma (St. Louis, MO, USA).Doxorubicin was obtained from EBEWE Pharma Ges.m.b.H. (Unterach, Austria), dimethyl sulfoxide (DMSO) from Amresco Inc. (Solon, OH, USA), phenolchloroform from Roth (Karlsruhe, Germany) and isopropanol from PLIVA-Lachema (Brno, Czech Republic).12-Hydroxy-and 13-hydroxyellipticine were isolated from multiple high-performance liquid chromatography (HPLC) runs of ethyl acetate extracts of incubations containing ellipticine and human and/or rat hepatic microsomes as described 12,33,34 .All these and other chemicals used in the experiments were of analytical purity or better.Enzymes and chemicals for the 32 P-postlabeling assay were obtained from sources described 12 .All other chemicals used in the experiments were of analytical purity or better.

Synthesis of covalent formaldehyde-mediated doxorubicin-DNA adducts
Multiple 0.5-ml reactions were run in parallel according the procedure described by Zeman et al. 21.Briefly, each reaction mixture contained 125 mM calf thymus DNA, 125 mM doxorubicin, and 0.37% formaldehyde in 20 mM sodium phosphate buffer (pH 7.0) containing 150 mM NaCl and 0.5 mM EDTA.Reactions were run at 10 °C for 4 h, the time determined for nearly 100% conversion of all DNA to covalent species 21 .DNA was isolated from the mixture by its precipitation with 2.5 times the volume of ethanol (-20 o C), washed twice with 5 ml of 70% ethanol, 5 ml of absolute ethanol and 5 ml of diethyl ether.DNA was dried under a stream of nitrogen and dissolved in distilled water.

Treatment of neuroblastoma cells with ellipticine and doxorubicin for DNA adduct analyses
Neuroblastoma cell lines were seeded 24 h prior to treatment at a density of 5 x 10 5 cells/ml in two 75 cm 2 culture flasks in a total volume of 20 ml of IMDM.Ellipticine and doxorubicin were dissolved in 5 μl of DMSO and distilled water, respectively.The final concentration was 1 or 10 μM.After 48 h the cells were harvested after trypsinizing by centrifugation at 2000 x g for 3 min and two washing steps with 5 ml of PBS yielded a cell pellet, which was stored at -80 ºC until DNA isolation.DNA was isolated and labeled as described in the next section.

DNA isolation
DNA from cells treated with ellipticine and doxorubicin was isolated by the phenol-chloroform extraction as described 7 .

P-postlabeling of ellipticine-derived DNA adducts
The 32 P-postlabeling of nucleotides using nuclease P1 enrichment procedure 31 , found previously to be appropriate to detect and quantify ellipticine-derived DNA adducts formed in vitro and in vivo 12,14,33,34 was used previously to detect and quantify the adducts formed in neuroblastoma cells 7 .DNA isolated from neuroblastoma cells as described in that study 7 and DNA from experiments performed earlier, namely, calf thymus DNA incubated with 13-hydroxy-and 12-hydroxyellipticine 33,34 were labeled with 32 P to show and compare adduct spot patterns.

P-postlabeling of doxorubicin-derived DNA adducts
The standard procedure 29 , this procedure under the ATP-deficient conditions 30 , the version using extraction of adducts with n-butanol 35 and the nuclease P1 enrichment version 31 of the 32 P-postlabeling assay were used.Labeled DNA digests were separated by two chromatographic methods on polyethylenimine (PEI)-cellulose plates.(i) Essentially as described 29 , except that D3 solvent was 3.5 M lithium formate, 8.5 M urea (pH 3.5); D4 solvent was 0.8 M lithium chloride, 0.5 M Tris, 8.5 M urea (pH 8.0), followed by a final wash with 1.7 M sodium phosphate (pH 6.0).D2 was omitted (method A). (ii) 32 P-labeled adducts were also resolved by a modification described by Reddy et al 32 .This procedure has been shown to be suitable for resolution of DNA adducts formed by oanisidine 36 or o-nitroanisole 37 .The solvents used in this case were: D1, 2.3 M sodium phosphate (pH 5.77); D2 was omitted; D3, 2.7 M lithium formate, 5.1 M urea (pH 3.5); D4, 0.36 M sodium phosphate, 0.23 M Tris-HCl, 3.8 M urea (pH 8.0).After D4 development and brief water wash, the sheets were developed (along D4) in 1.7 M sodium phosphate (pH 6.0) (D5), to the top of the plate, followed by an additional 30-40 min development with the TLC tank partially opened, to allow the radioactive impurities to concentrate in a band close to the top edge (method B)(ref. 32,36,37).Adduct levels were calculated in units of relative adduct labeling (RAL), which is the ratio of c.p.m. of adducted nucleotides to c.p.m. of total nucleotides in the assay.

DNA adduct formation by ellipticine and doxorubicin analyzed with 32 P-postlabeling
As shown in our previous study 7 , ellipticine and doxorubicin are cytotoxic to human UKF-NB-3 and UKN-NB-4 neuroblastoma cells.The toxicity of ellipticine and doxorubicin to both neuroblastoma cell lines was similar; IC 50 values ranged from 0.42 to 0.70 μM (ref. 7).It should be noted, however, that even though the IC 50 values for Fig. 3. Autoradiographs of PEI-cellulose TLC maps of 32 P-labeled digests of DNA isolated from neuroblastoma UKF-NB-3 exposed to 10 μM ellipticine for 48 h (A) (ref. 7), from calf thymus DNA reacted with 13-hydroxyellipticine (B) and 12-hydroxyellipticine (C) (ref. 33,34).Analyses were performed by the nuclease P1 version of the 32  ellipticine for both neuroblastoma cell lines did not differ significantly, the UKF-NB-4 cell line (the less-aggressive S-type line) was less sensitive to this drug than the UKF-NB-3 cell line.No viability of a UKF-NB-3 neuroblastoma cell line was found at 0.75 μM ellipticine in cultivation medium, while viability of UKF-NB-4 cells was found up to ellipticine concentrations of 6.3 μM.
DNA adducts were analyzed in DNA of neuroblastoma cells treated with 1 and 10 μM ellipticine and doxorubicin, the concentrations that are toxic to these cells.Formation of ellipticine-derived DNA adducts in UKF-NB-3 and UKN-NB-4 neuroblastoma cells has already been found in our previous work 7 .Therefore, their analysis in human UKF-NB-3 cells was used in the present study as positive control.As shown in Fig. 3A, two major ellipticine-derived DNA adducts formed in these neuroblastoma cells are generated from ellipticine-13-ylium and ellipticine-12-ylium (Fig. 1), the reactive species formed by dissociation of ellipticine metabolites, 13-hydroxy-and 12-hydroxyellipticine 33,34 (Fig. 3B,C).Since ellipticine-derived DNA adducts have already been detected and quantitated using the nuclease P1 version of 32 P-postlabeling assay 12,33,34 , this version of the method was first used also for analysis of DNA isolated from neuroblastoma cells treated with doxorubicin.
In contrast to these results, no adducts were found in DNA of cells treated with doxorubicin analyzed with the same version of the 32 P-postlabeling method (Fig. 3F).Because the nuclease P1 version of the 32 P-postlabeling method might have limitations for detecting some of DNA adducts 35 , we also used other versions of the 32 P-postlabeling method such as the standard procedure 28 , this procedure under the ATP-deficient conditions 30 and the version utilizing of extraction of adducts into nbutanol 35 to analyze doxorubicin-derived DNA adducts (Fig. 3D-G).Likewise, modifications of all versions that are appropriate for resolution of more polar adducts on thin layer of PEI cellulose such as the adducts generated in DNA by o-anisidine 36 or o-nitroanisole 37 (see method B in Material and Methods), were tested to determine doxorubicin-derived DNA adducts.Using all these methods, no adducts were again detected (data not shown).These findings might indicate at least two phenomena.First, low levels of covalent adducts (if any) that are undetectable by the 32 P-postlabeling methods might be formed by doxorubicin in DNA of neuroblastoma cells.Their formation in neuroblastoma cells cannot, however, be excluded.Recently, changes in structure of DNA isolated from neuroblastoma cells induced by their treating with doxorubicin have been detected by square-wave voltammetry 38 .Nevertheless, whether these changes are produced by formation of actual covalent doxorubicinderived DNA adducts remain to be investigated.Second, the 32 P-postlabeling method is not suitable to determine covalent adducts formed in DNA by doxorubicin.
In order to resolve whether the 32 P-postlabeling method is suitable for detection of covalent formaldehydemediated doxorubicin-DNA adduct, we prepared the DNA adduct synthetically, by incubation of DNA with doxorubicin in the presence of formaldehyde, the compound necessary to covalent doxorubicin-DNA adduct formation 21 .DNA incubated with doxorubicin and formaldehyde was isolated from the incubation mixture (see Material and Methods) and analyzed with all versions of the 32 P-postlabeling method and two modifications used for resolution of adducts on PEI-cellulose TLC (see Materials and Methods).No DNA adducts were again detected (data not shown).These results indicate that none of the versions of the 32 P-poslabeling assay is suitable for detecting formaldehyde-mediated covalent doxorubicin-DNA adducts.We can only speculate why formaldehyde-mediated covalent doxorubicin-DNA adduct is not detectable by 32 P-postlabeling.One of the reasons for this finding, could be inefficient labeling reaction for formaldehyde-mediated doxorubicin-derived adduct during the 32 P-postlabeling method, which is the conversion of the adducted nucleoside 3'-phosphate to its corresponding 3',5'-bisphosphate by T4 polynucleotide kinase.Further reasons could be incomplete digestion of modified DNA, loss of material during the experimental manipulations used in 32 P-postlabeling or retaining compounds at the origin of the PEI-cellulose TLC plates.However, investigation of these features was beyond the scope of the present study.

Cytotoxicity of and DNA adduct formation by ellipticine and doxorubicin in neuroblastoma cells cultivated under hypoxic conditions
Hypoxia frequently occurs in tumors because of their fast growth and inadequate vascularisation.This strongly correlates with advanced disease and poor outcome caused by chemoresistance.As reported 7 , growth inhibition is mediated by ellipticine and doxorubicin in neuroblastoma cells even under hypoxic conditions of cultivation.Nevertheless, whereas a low effect of hypoxia was found on toxicity of both tested compounds to a UKF-NB-3 cell line, the UKF-NB-4 neuroblastoma cells were less sensitive to both studied cytostatics under the hypoxic conditions of their cultivation 7 .
Under hypoxic conditions, the ellipticine-DNA adduct levels in neuroblastoma cells were lower 7 , whereas formation of DNA adducts generated by doxorubicin in these cells was again not detectable.In both neuroblastoma cell lines treated with 10 μM ellipticine, almost 2-fold decrease in levels of ellipticine-DNA adducts by hypoxia was found and resulted from a decreased formation of adducts 1 and 2 in both types of cells 7 .This finding shows that CYP enzymes, whose activities are dependent on oxygen 13,14,33,34 , are predominantly responsible for formation of adducts 1 and 2 in these neuroblastoma cells 7 .
A decrease in the levels of adducts 1 and 2 in neuroblastoma cells under hypoxic conditions corresponded to a decrease in toxicity of ellipticine under these conditions 7 .We therefore concluded that formation of ellipticine-DNA adducts was the predominant DNA-damaging mechanism of ellipticine action, resulting in its strong cytotoxicity to neuroblastoma cells.

CONCLUSIONS
The results presented here are the first to demonstrate that in contrast to covalent DNA adducts formed by ellipticine, the adducts generated by formaldehyde-mediated covalent binding of doxorubicin to DNA are not detectable by the 32 P-postlabeling assay.No DNA adducts were detectable in either in vitro, in incubations of DNA with doxorubicin or in DNA of neuroblastoma cells treated with this drug.The results of this and our previous study 7 also suggest that covalent binding of ellipticine in DNA of UKF-NB-3 and UKF-NB-4 neuroblastoma cell lines is the predominant mechanism responsible for cytotoxicity of this drug to these cells.The mechanism of toxicity of doxorubicin to neuroblastoma cells has, however, not been resolved by this study and the development of novel methods identifying covalent doxorubicin-derived DNA adducts is the major challenge for further research.ABBREVIATIONS CYP, cytochrome P450; DMSO, dimethyl sulfoxide; IMDM, Iscove's modified Dulbecco's medium; HPLC, high-performance liquid chromatography; PBS, phosphate buffered saline; PEI-cellulose, polyethylenimine-cellulose; RAL, relative adduct labeling; SDS, sodium dodecyl sulphate; TLC, thin layer chromatography.

Fig. 1 .
Fig. 1.Scheme of the metabolism of ellipticine by human CYPs and peroxidases showing the characterized detoxication metabolites and those proposed to form DNA adducts.The compounds shown in brackets are the hypothetical electrophilic metabolites postulated as ultimate arylating species or the postulated N²-deoxyguanosine adducts.
Fig.3.Autoradiographs of PEI-cellulose TLC maps of32 P-labeled digests of DNA isolated from neuroblastoma UKF-NB-3 exposed to 10 μM ellipticine for 48 h (A) (ref.7 ), from calf thymus DNA reacted with 13-hydroxyellipticine (B) and 12-hydroxyellipticine (C) (ref.33,34 ).Analyses were performed by the nuclease P1 version of the 32 P-postlabeling assay.(D-G) Autoradiographs of PEI-cellulose TLC maps of of 32 P-labeled digests of DNA isolated from neuroblastoma UKF-NB-3 exposed to 10 μM doxorubicin for 48 h.The standard procedure of the 32 P-postlabeling assay was used for the TLC map in panel (D), the standard procedure under ATP-deficient conditions for that in panel (E), the nuclease P1 version for that in panel (F) and the version utilizing extraction of adducts into n-butanol for that in panel (G).The method A (see the Material and methods section) was utilized for resolution of adducts in all panels.(A) Scans of the plates from the imager for 6.5 min; (B,C) autoradiographs of films exposed for 1 h at -80 o C; (D-G) scans of the plates from the imager for 1 h.Origins are located at the bottom left corners (D3 from bottom to top and D4 from left to right).