Mutation Research 414 Ž1998. 125–129

Factors affecting the genotoxic potency ranking of natural anthraquinones in mammalian cell culture systems Stefan O. Mueller, Werner K. Lutz, Helga Stopper

)

Department of Toxicology, UniÕersity of Wurzburg, 97078 Wurzburg, Germany ¨ ¨ Received 9 December 1997; revised 12 March 1998; accepted 13 March 1998

Abstract We had reported that the plant-derived 1,8-dihydroxyanthraquinone derivatives, emodin and danthron, were clearly genotoxic in mouse lymphoma L5178Y cells, whereas chrysophanol was only weakly genotoxic and physcion not at all. Danthron was more potent than emodin. Furthermore, we had found that these compounds bound non-covalently to DNA and inhibited topoisomerase II activity. Interestingly, in these systems emodin was more potent than danthron. This inversion of the ranking prompted us to investigate the underlying mechanism. Since emodin shows a high serum-protein binding affinity, horse serum used as a media-supplement in the mouse lymphoma genotoxicity assays was analyzed for a potential selective scavenging of emodin. Non-covalent DNA-binding in mouse lymphoma L5178Y cells was investigated in the absence or presence of serum. In the presence of 10% serum, the DNA-binding potency of emodin was markedly reduced and was lower than that of danthron. We also applied mutation assays with mouse lymphoma cells and AS52 cells and varied the serum concentration used. In the absence of serum emodin showed slightly higher mutagenicity in AS52 cells than danthron. At reduced serum concentration Ž0.5%. emodin was strongly cytotoxic to the mouse lymphoma cells. For chrysophanol and physcion, a considerable reduction of the non-covalent DNA-binding potency in intact cells was found when compared to danthron, in concordance with their lower genotoxic potency. Overall, these data support the understanding that the genotoxicity of anthraquinones is, at least in part, mediated by non-covalent DNA-binding. q 1998 Elsevier Science B.V. All rights reserved. Keywords: Chrysophanol; Danthron; Emodin; Physcion; Genotoxicity; Non-covalent DNA-binding; Mechanism; Modulating factor

1. Introduction In three recent studies we investigated the genotoxicity of the naturally occurring anthraquinones emodin, danthron, chrysophanol and physcion ŽFig.

) Corresponding author. Department of Toxicology, University of Wurzburg, Versbacher Str. 9, 97078 Wurzburg, Germany. Tel.: ¨ ¨ q 49-931-201-3427; fax: q 49-931-201-3446; E-m ail: [email protected]

1. using the micronucleus test, the comet assay and mutation assays in mammalian cells. In mouse lymphoma cells, danthron was a more potent genotoxin than emodin w1x. Non-covalent DNA-binding and topoisomerase II inhibition was found as one relevant, activation-independent mechanism of genotoxicity w1x. For these endpoints, emodin was more effective than danthron, not reflecting the compounds’ genotoxic potencies. This inversion of the ranking prompted us now to investigate the underlying mechanism.

1383-5718r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 1 3 8 3 - 5 7 1 8 Ž 9 8 . 0 0 0 4 7 - 3

126

S.O. Mueller et al.r Mutation Research 414 (1998) 125–129

Fig. 1. Chemical structures of the investigated compounds.

For emodin, a high serum-protein binding affinity has been shown w2x. Since horse serum is used as a media-supplement in the mouse lymphoma genotoxicity assays, a potential selective scavenging of emodin was investigated. For this purpose, we subjected emodin and danthron to the mouse lymphoma mutation assay varying the serum concentration in the incubation media and we compared the mutagenic responses with that from the AS52 mutation assay w1x were no serum supplement is used. In addition, we investigated the non-covalent DNAbinding affinities of emodin and danthron in intact cells in the presence of serum proteins. In another study from this laboratory wMueller et al., submittedx, we reported only weak genotoxic effects of chrysophanol and no genotoxicity of physcion in mammalian cells Žsee Fig. 1 for structures.. Therefore, we also tested their ability to bind non-covalently to DNA in solution and in intact cells in order to verify whether their genotoxic potency reflects this mode of action. 2. Materials and methods 2.1. Chemicals Chrysophanol ŽCAS No. 481-74-3., danthron Ž117-10-2., emodin Ž518-82-1. and physcion Ž521-

61-9. were acquired from Roth ŽKarlsruhe, Germany.. m-Amsacrine Ž54301-15-4., ethylmethanesulfonic acid ethyl ester Ž62-50-0., etoposide Ž3341942-0., DMSO Ž67-68-5. and bisbenzimide Hoechst 33342 Ž23491-52-3. were purchased from Sigma ŽDeisenhofen, Germany.. Test compounds were dissolved in DMSO. The final concentration of DMSO in the cell culture medium did not exceed 1%. 2.2. Cell culture Mouse L5178Y cells were cultured in suspension in RPMI-1640 as described w1x. Cell cultures were grown in a humidified atmosphere with 5% CO 2 in air at 378C. 2.3. DNA-binding assay with DNA in solution and in liÕing cells This method is based on the elevated fluorescence of Hoechst 33342 when it binds non-covalently to DNA w3,4x. The addition of other DNA-binding compounds causes a reduction of this fluorescence, presumably because of competition of binding sites of the DNA. Fluorescence was monitored by a Kontron spectrofluorometer in 1 = 1 cm cuvettes under continous stirring at 378C. 500 ng calf thymus DNA or a

S.O. Mueller et al.r Mutation Research 414 (1998) 125–129

suspension of 10 6 L5178Y cellsrml in 2 ml PBSrG were mixed with 2 m M Hoechst 33342 and incubated until fluorescence equilibrium was reached Ž30 min for calf thymus DNA; 4 h in living cells.. Horse serum was added thereafter Ž10% vrv. when necessary. Experiments were initiated after equilibrium of fluorescence by addition of the respective compound and the decrease of fluorescence was monitored over time. Steady state of the decreased fluorescence after substance addition was reached within few minutes Žup to 5 min. for all test compounds. The excitation wavelength was set at 347 nm and the emission wavelength at 454 nm. Total fluorescence of cells with Hoechst 33342 in equilibrium was about 100 fluorescence units ŽFU.. Non-specific fluorescence of the compounds was always less than 1 FU. Quenching of Hoechst 33342 by anthraquinones and m-AMSA was always less than 1 FU Žabsorbance.. IC 50 values were determined in at least three independent experiments using concentrations of the test compounds including the respective range of IC 50 values. The viability of the cells was determined with the fluoresceindiacetatrethidiumbromide method as described w5x and was between 70 and 85%. 2.4. Mutation assay Mutation assays were performed with mouse lymphoma L5178Y tk " cells at the tk locus ŽMLA. and with AS52 cells at the gpt locus as previously described w1x. The MLA was modified by reducing the concentration of the horse serum supplemented to the incubation medium, from 5% to 0.5% Žvrv.. 2.5. Statistics Statistical analyses were performed applying Student’s t-test. Statistical significance is indicated by asteriks Ž) p F 0.05..

3. Results and discussion 3.1. Non-coÕalent DNA-binding to DNA in solution and in intact cells We tested the non covalent DNA-binding to calf thymus DNA and to DNA in intact mouse lym-

127

phoma cells. For this, we modified a published method w6x using a fluorescence dye readily capable to reach DNA in intact cells w4x, as described wMueller et al., submittedx. We analyzed the competition of our test compounds with bisbenzimide Hoechst 33342 for DNA-binding sites under non-cytotoxic conditions. The DNA intercalator and topoisomerase II poison m-amsacrine Ž m-AMSA. was used as a positive control. In Table 1, the IC 50 values for the binding of Hoechst 33342 for each compound are given. This indirect measure is assumed to correspond to the ability of the respective compound to bind non-covalently to the DNA. The known topoisomerase II poison etoposide as a non DNA intercalator showed no detectable effects with DNA in solution or in intact cells, and was used as a negative control. With DNA in solution, the IC 50 values are assumed to correspond to the non-covalent DNA binding affinity of the tested compounds. They were in the same range for all tested anthraquinones and the positive control m-AMSA ŽTable 1.. In intact cells, the IC 50 values are assumed to correspond to the ability of the respective compound to bind non-covalently to DNA. This ability depends on the respective intracellular concentration and may not necessarily reflect the intrinsic DNA-binding affinity. Chrysophanol and physcion showed a six- to nine-fold increase in IC 50 values. The lack of genotoxicity of chrysophanol and the weak effects of physcion in mouse lymphoma cells could be explained by their lower ability to reach DNA in intact cells. For emodin and danthron, and also for m-AMSA, only a two-fold decrease was seen for their binding to DNA in solution vs. in intact cells. Surprisingly, emodin showed higher DNA-binding affinities than danthron, although statistically not significant Ž p s 0.1., a ranking opposite to their genotoxic potencies w1x. Since emodin exhibits a high serum protein binding w2x we suspected the horse serum, which is supplemented to the cell-culture medium in the genotoxicity test systems Ž5–10% vrv., to act as scavenger and analyzed the non-covalent DNA-binding potencies in the presence of 10% horse serum ŽTable 1, last column.. Indeed, in the presence of supplemented serum, emodin showed a statistically significant weaker Ž p F 0.05. DNA-binding potency than danthron, now reflecting the observed genotoxicity

S.O. Mueller et al.r Mutation Research 414 (1998) 125–129

128

Table 1 IC 50 values of different compounds for the non-covalent DNA-binding of Hoechst H33342 IC 50 Ž m M. with DNA in solution Emodin Danthron Chrysophanol Physcion m-AMSA Etoposide

a

IC 50 Ž m M. in intact cells a

12 " 4 14 " 4 a 15 " 3 13 " 2 10 " 1a `ab

IC 50 Ž m M. in intact cells q10% serum 32 " 9 c ) 22 " 3 d ) y y 16 " 4 `b

23 " 2 28 " 5a 110 " 40 73 " 25 22 " 3 a `ab

a

Data taken from Mueller et al., submitted. ` means no inhibition measurable at any tested concentration ŽIC 50 4 500 m M Žhighest concentration tested... c Statistical significance of differences between IC 50 values in the presence or absence of 10% serum ŽStudents’s t-test. are indicated by an asterisk Ž ) p F 0.05.. d Statistical significance of differences between IC 50 values of emodin and danthron ŽStudents’s t-test. are indicated by an asterisk Ž ) p F 0.05.. IC 50 values represent the concentrations at which the DNA bound fluorescence of Hoechst 33342 was reduced to 50%. Data are means" S.D. Ž N s 3.. b

ranking w1x. To summarize, the presence of serum in the reaction buffer selectively decreased the activity of emodin Žstatistically significant, p F 0.05., while danthron and m-AMSA were not affected Žor even showed a tendency towards an increased potency.. 3.2. Mutagenicity of emodin is dependent on the presence of serum We analyzed whether the selective scavenging effect of serum on the non-covalent DNA-binding potency of emodin results also in a selectively re-

duced genotoxicity in mammalian cells. For this purpose we tested emodin and danthron in a modified mouse lymphoma mutation assay ŽMLA. and in the AS52 mutation assay ŽAS52., a test system in which no serum is supplemented to the culture media during substance incubation. In the standard MLA, cell culture media are supplemented with 5% horse serum Žvrv. during substance incubation. We now applied the same protocol with reduced serum content during the incubation Ž0.5% vrv.. Compounds were tested in a concentration range from 20 m M Žlowest effective dose in the standard test. up to 100 m M Žlimit of solubility.. In Table 2, the highest

Table 2 Dependence of the mutagenicity of emodin and danthron in mammalian cells on the serum concentration Supplemented serum Žvrv.

Emodin Ž20–100 m M. Danthron Ž20–100 m M. DMSO Ž1%, solvent control. EMS Žpositive control. 2 mM 4 mM a

Mutation assay in AS52 Cells

Mouse lymphoma cells

0%

0.5%

5%a

6.4 b Ž1.0. 5.6 b Ž0.9. 1Ž0.84.

cytotoxic 6.2Ž0.43. 1Ž0.59.

2.8 b Ž0.44. 4.3 b Ž0.40. 1Ž0.52.

y 8.7Ž0.72.

7.1Ž0.41. y

3.5Ž0.27. y

Data taken from Ref. w1x. Differences between emodin and danthron were statistically not significant ŽStudents’s t-test; p ) 0.1.. Assays were repeated at least twice with consistent results. Compounds were tested in a concentration range from 20 up to 100 m M. Given are the highest achieved relative mutant fractions and in parenthesis the respective titers ŽAS52. or cloning efficiencies Žmouse lymphoma cells.. b

S.O. Mueller et al.r Mutation Research 414 (1998) 125–129

relative mutant fractions achieved in this concentration range are shown. The highest mutant fractions were observed in the absence of serum, with emodin slightly more potent Žstatistically not significant, p s 0.4. than danthron ŽAS52 cells, column 1 in Table 2.. In the presence of 5% serum, emodin showed a weaker mutagenicity Žstatistically not significant, p s 0.3. than danthron ŽMLA, column 3 in Table 2.. In the MLA performed with 0.5% serum, emodin was strongly cytotoxic at all concentrations used so that no mutagenic properties could be determined. Overall the data show, that in the absence of serum emodin showed stronger genotoxic and cytotoxic effects than danthron.

4. Conclusions The weaker genotoxic potency in mammalian cells for emodin as compared to danthron is most likely caused by selective scavenging effects by the supplemented serum in the cell culture media. Furthermore, the weak genotoxic potencies of chrysophanol and physcion correlate with their high IC 50 value. The data support the understanding that the anthraquinone-induced genotoxicity is, at least in part, mediated by non-covalent DNA-binding. More general, we conclude from this study that the use of non-defined supplements such as serum in cell culture genotoxicity tests cannot only diminish effects, but even selectively reduce effects of a particular chemical within a class of structurally related compounds.

129

Acknowledgements This work was supported by the Swiss Federal Office of Public Health ŽBAG grant No. FE 316.95.0500.. We thank Dr. J. Schlatter, monitoring scientist of the BAG, for valuable advice in all parts of this work and Ms. M. Gerhard for her expert technical assistance.

References w1x S.O. Muller, I. Eckert, W.K. Lutz, H. Stopper, Genotoxicity of ¨ the laxative drug components emodin, aloe-emodin and danthron in mammalian cells: topoisomerase II mediated?, Mutat. Res. 371 Ž1996. 165–173. w2x J.W. Liang, S.L. Hsiu, P.P. Wu, P.D.L. Chao, Emodin pharmacokinetics in rabbits, Planta Med. 61 Ž1995. 406–408. w3x T. Bogush, G. Smirnova, I. Shubina, A. Syrkin, J. Robert, Direct evaluation of intracellular accumulation of free and polymer-bound anthracyclines, Cancer Chemother. Pharmacol. 35 Ž1995. 501–505. w4x F. Belloc, F. Lacombe, P. Dumain, F. Lopez, P. Bernard, M.R. Boisseau, J. Reifers, Intercalation of anthracyclines into living cell DNA analyzed by flow cytometry, Cytometry 13 Ž1992. 880–885. w5x G.H.S. Strauss, Non-random cell killing in cryopreservation: Implication for performance of the battery of leukocyte tests ŽBLT.: I. Toxic and immunotoxic effects, Mutat. Res. 252 Ž1991. 1–15. w6x T. Bogush, J. Robert, Comparative evaluation of the intracellular accumulation and DNA binding of Idarubicin and daunorubicin in sensitive and multidrug-resistant human leukaemia K562 cells, Anticancer Res. 16 Ž1996. 365–368.