H. Stopper Department of Toxicology, University of Würzburg, 97078 Würzburg, Germany

Address for correspondence:

Dr. Helga Stopper Department of Toxicology University of Würzburg Versbacher Str. 9 97078 Würzburg Germany Fax: (0)931-201-3446 E-mail:stopper@toxi.uni-wuerzburg.de

Keywords: Micronuclei, methylation, azacytidine, mutation, mitotic recombination

Download file as a Word document: Stopper.doc

Introduction

There are many compounds that induce genotoxic effects without directly damaging DNA. Over the last years we have developed a working hypothesis to explain some of the mechanisms involved (Fig. 1). An interaction of compounds with DNA in a not directly damaging way can cause a change in DNA conformation. We hypothesize that the interaction of proteins with such a DNA may be impaired as a consequence. If proteins of the mitotic machinery are disturbed, micronuclei can be formed. Depending on the genes encoded on the DNA enclosed in a micronucleus, its loss may result in the formation of mutants. On the other side, most mutants are formed without micronucleus induction, for example by loss of heterozygozity (LOH). Among the mechanisms leading to LOH, mitotic recombination may be more important than previously assumed.

Genotoxicity of azacytidine-analogs

The cytidine-analog 5-azacytidine induces tumors in rodents (Carr et al. 1984; Schmahl et al. 1985) and the mechanism of tumor induction is not clear. The compound does not directly damage DNA.

We found that it induced micronuclei (Stopper et al. 1995b; Stopper et al. 1993b; Stopper et al. 1992). Micronuclei in general can either contain a whole chromosome or a chromosomal fragment. The first kind occurs if a chromosome is not at the right place in mitosis or if chromosome distribution is not correct; these micronuclei usually appear soon after substance treatment. The second kind requires chromosome breakage and usually appears much later after treatment. Since 5-azacytidine-induced micronuclei appeared early after treatment we investigated whether 5- azacytidine interferes with the arrangement of chromosomes at mitosis (e.g. by spindle disturbance) and analyzed mitotic ring arrangements (Stopper et al. 1993a). While the spindle disturbing compound diethylstilbestrol (DES) increased the number of mitotic ring arrangements with displaced chromosomes (example shown in Fig. 2) from 0 to 4.6%, 5-azacytidine did not induce such misalignments.

Using a modified BrdU incorporation method, we then identified sensitive cell cycle phases for micronucleus induction with 5- azacytidine (Stopper et al. 1993a). Syrian hamster embryo fibroblasts that had been treated with 5-azacytidine and BrdU were analyzed for micronucleus formation 12 hours after the end of the treatment. Only those cells that had replicated DNA at the time of 5-azacytidine incubation could now have antibody-detectable BrdU in their DNA. By analyzing the micronucleus containing cell population we found that 5-azacytidine induced micronuclei only in cells that had replicated DNA. Thus, it is very likely that 5- azacytidine can only induce micronuclei after its incorporation into DNA. DES, which was included as a positive control for spindle disturbance, was only active in those cells that had not replicated DNA during substance exposure.

Further elucidation of the mechanism of 5-azacytidine-induced genotoxicity was attempted by applying supravital UV-microscopy (Schiffmann and De Boni 1991; Stopper et al. 1993a) (Fig. 3).

In this method, a cell culture is being seeded into gas-tight chambers, the DNA is stained with low concentrations of bisbenzimid 33258, and the chamber is placed under a fluorescent microscope. Neutral density filters are used to reduce the UV light to a level that does not damage the cells within the observation time. The use of a CCD-intensifier-camera and an image enhancement system (Hamamatsu Co., Herrsching/Germany) permits the obervation of the live cells during the course of mitosis on a video monitor (Fig. 4). 5-Azacytidine treated cells entered mitosis normally and formed normal metaphase arrangements. This was in agreement with the analysis of the metaphase arrangements in fixed cells. However, chromatid separation at anaphase was disturbed and chromatin bridges (thin threads of chromatin) were formed. When the daughter cells progressed into interphase and moved apart from each other, these chromatin threads sometimes became very thin and elongated. In some cases, the chromatin threads ruptured and micronuclei were formed.

In our search for possible reasons for this chromosomal instability we questioned whether

5-azacytidine-induced changes in the endogenous cytosine-methylation could be involved. Normally, the pattern of cytosine-methylation in differentiated mammalian cells is stable and heritable at cell division and cytosine- methylation is presumed to be involved in many regulatory events like differentiation or neoplastic transformation of the cell (Laird and Jaenisch 1996). 5-azacytidine cannot be methylated like cytidine and thus the pattern and extent of cytosine-methylation is changed by 5-azacytidine incorporation into DNA. We investigated 4 different analogs of cytosine that inhibit methylation to a different degree (5-fluoro-2`-deoxyazacytidine; 5- azacytidine; 5,6-dihydroazacytidine; 6-azacytidine) (Stopper et al. 1995a). These analogs had been found to induce cell differentiation (Jones and Taylor 1980) and to be mutagenic (McGregor et al. 1989) in the same order of intensity as they inhibited methylation. We now found them to be cytotoxic and to induce micronuclei in the same order of effectiveness, although 6- azacytidine, which does not influence methylation, induced some micronuclei at high doses.

These findings lead us to the following hypothesis for the mechanisms of 5-azacytidine-mediated genotoxicity: A change in the pattern of methlyation can lead to a change in DNA conformation. Changes in chromosome condensation have been described (Hori 1983; Schmid et al. 1984). The observed thin chromatin bridges in supravital UV-microscopy can be interpreted as such a change in conformation. Furthermore, on a molecular level we found a 10% change in curvature dependent gel mobility of a centromeric DNA fragment with and without cytosine-methylation (S. Diekmann, unpublished results). Certain proteins may be impaired in their interaction with DNA by this change in conformation. In the case of the kinetochore complex this can lead to spindle-unattached chromosomes in metaphases or to weakly attached chromosomes which are prone to malsegregation in mitosis. In fact, we observed a lower percentage of kinetochore positive micronuclei (antibody staining; 19%) than centromere positive (minor satellite in situ hybridization; 35%) in 5-azacytidine-induced micronuclei. In the case of topoisomerase II this impaired interaction with DNA can result in difficulties in chromatid separation since the chromosomes have not been dekatenated sufficiently when the cell attempts their separation at anaphase (Downes et al. 1991) (Kirchner et al. 1995). We observed such anaphase problems with supravital UV-microscopy experiments and we detected that the consequence can be chromatid bridges and micronucleus formation.

To use another (possibly more physiological) way of hypomethylation, we investigated F9 mouse teratocarcinoma cells. These cells can be induced to differentiate in vitro over a time period of 2 weeks (Alonso et al. 1991) and it has been shown that the pattern of cytosine-methylation changes dramatically over that time (Razin et al. 1986). For example, the minor satellite region which contains the kinetochore binding sequence (Masumoto et al. 1989) and most likely many topoisomerase II binding regions (Sumner 1991), becomes hypomethylated (Teubner and Schulz 1994). When we analyzed the frequency of spontaneously occuring micronuclei during the time period of differentiation (Stopper et al. 1997c), we found an increase. Kinetochore staining and in situ hybridization with a DNA probe that binds to the minor satellite region revealed that the percentage of signal positive micronuclei was increased in fully differentiated cells (time point 216 hours; 35.3/31.0%) as compared to the time points before (up to 168 hours, less than 25.0/17.5%). A possible interpretation would be that the kinetochore protein complex still bound to the DNA but that its function was impaired, more chromosomes were not attached correctly to the spindle and included in micronuclei as a consequence. Another explanation might be increased tetraploidization as a result of impaired mitosis, and then increased genomic instability of the tetroploid cells resulting in more micronuclei with whole chromosomes included. Both explanations would be in aggreement with our working hypothesis.

To investigate the effects of methylation changes on topoisomerase II, we used a cell free assay, in which the decatenation of mitochondrial DNA rings from Crithidia fasciculata is measured (Marini et al. 1980). When we compared unmethylated catenated DNA and CpG-methylated catenated DNA we found a marked difference in topoisomerase II activity (results will be published separately).

In F9 cells, we measured topoisomerase II-mediated strand breaks in the comet assay in dependence of the differentiation time. Although ethylmethane sulfonate induced comets at all times, the topoisomerase II inhibitor etoposide only induced comets in undifferentiated cells (Stopper et al. 1997c). Western blots (antibodies No. 680-topoisomerase II -N, No. 779-topoisomerase II- á-C; (Boege et al. 1995)) showed that this was not due to a decrease in topoisomerase or á expression. Thus, the idea that methylation changes influence the quality of DNA as a substrate for topoisomerase II was further supported.

When we treated mouse lymphoma L5178Y cells with a sequential combination of 5-azacytidine and different topoisomerase II inhibitors, the increase in micronucleus frequency was synergistic (over-addititve) (Stopper et al. 1997c). This would be in accordance with the idea that both treatments, 5-azacytidine as well as the topoisomerase II inhibitors, influence the same cellular target, namely topoisomerase II. In other words, hypomethylation of the DNA inhibits topoisomerase II.

Thus, several lines of experimental evidence are all in accordance with our hypothesis that changes in the pattern of endogenous cytosine-methylation can influence the conformation of DNA and as a consequence impair the binding of DNA-interacting proteins such as topoisomerase II and the kinetochore complex. This can lead to difficulties in mitosis which can manifest as micronucleus formation. Other non-covalent interactions of compounds with DNA, such as intercalation or binding to the minor groove may exert comparable effects.

Fate of micronuclei

In the next step of our working hypothesis (Fig. 1)

we question the fate of micronucleus containing cells (Fig. 5) (Stopper et al. 1994). The inclusion of a tumorsupressorgene in the micronucleus and the subsequent inactivation or loss of that micronucleus could lead to a transformed cell and therefore be a step in carcinogenesis. The death of a micronucleus containing cell would eliminate such a potentially dangerous cell. Reintegration of micronuclear material into the main nucleus might render the cells phenotypically normal. The situation for the growth of mutant colonies in the L5178Y tk mutation assay depends on a comparable set of events. The survival of trifluorthymidine-resistant mutant colonies in this cell line depends on the loss of the thymidine- kinase (tk) gene. The tk gene is located on chromosome 11. Allele 11+ contains the active form and allele 11- contains an inactive form of the gene. Thus, loss of chromosome 11 by micronucleus formation and inactivation/loss might be a mechanism for the formation of mutant colonies in this cell system.

However, when we investigated induction of mutation by the four known aneugenic compounds colcemid, diethylstilbestrol (DES), griseofulvin and vinblastine, we did not find an increase in mutant frequency (Stopper et al. 1994). In a recent publication by Sofuni et al. (Sofuni et al. 1996) several labs found DES and griseofulvin positive in the same mutation assay system. However, DES was only positive after exogenous metabolic activation, which may have lead to the loss of its aneugenic properties, and griseofulvin was mutagenic under conditions of substance precipitation. Although we used a comparable dose-range, microscopic inspection did not reveal any precipitation below 150 æg/ml. Since no detailed description of experimental procedures is given by Sofuni et al., the reason for the different precipitation behaviour is not known. However, the precipitated compound may have exerted non-aneugenic effects. Under our treatment conditions we found more than 87% of the induced micronuclei to be kinetochore positive, indicating aneugenicity (Stopper et al. 1994).

The failure to induce mutation under our conditions allows at least 3 different explanations. First, chromosome 11 was not included in micronuclei at sufficient frequency. Second, the DNA in the micronuclei was reintegrated into the nuclear genome and the cells resumed a non-mutant phenotype. Third, cells that have chromosome 11 included in a micronucleus cannot survive to form mutant colonies. Whole chromosome in situ hybridization ("chromosome painting") in micronuclei (Caspary et al. 1997) showed that chromosome 11 was included at sufficient frequency that should have allowed the detection of an elevated mutant frequency (Stopper et al. 1997b). Reintegration of micronuclear material is theoretically possible, but would require the micronucleus to be in cycle with the main nucleus to be successful; otherwise the result may be premature chromosome condensation in the micronucleus and subsequent destruction of that DNA (Obe et al. 1975). Furthermore, a large percentage of all micronuclei would have to be reintegrated to explain the described findings. This explanation is thus considered unlikely. Our own analyis of mutants and other published data never demonstrated a viable L5178Y cell line that had lost chromosome 11+ completely without having reduplicated chromosome 11-. Thus, we conclude that the most likely explanation is that chemically induced loss of a whole chromosome in mouse lympoma L5178Y cells through micronucleus formation leads to eventual cell death. The probability that primary human cells lose a whole chromosome containing a tumorsuppressorgene through micronucleus formation and survive to develop into a neoplastic cell is now also estimated to be very low. The situation may be different at later stages of neoplastic development (e.g., in tumor progession), where more unstable karyoptypes exist.

In the case of clastogenic compounds the loss of the material in micronuclei may not be quite as detrimental to the cell and mutants might be formed through the hypothesized pathway. However, comparison of micronucleus and mutant frequencies shows that other, additional mechanisms for the induction of mutants must exist for not-directly genotoxic compounds.

Mitotic recombination

Another mechanism of mutation that is mediated by the interaction between proteins and DNA is mitotic recombination (Fig.6). These proteins might also be influenced by the DNA secondary structure (conformation). By changing the DNA conformation, not directly DNA damaging genotoxic compounds may alter the protein-DNA interaction and the frequency of mitotic recombination might increase. To be able to investigate that hypothesis in the L5178Y mouse lymphoma mutation assay system, a new combination of a method for the detection of loss of heterozygozity with chromosome specific in situ hybridization was required (Caspary et al. 1997; Liechty et al. 1995; Liechty et al. 1997)). Short repetitive stretches of DNA composed of simple repeate sequences are interspersed throughout the mammalian genomes and are called microsatellites.

The length of these simple repeat sequences is often heterozygous between homologous chromosomes (Fig. 7). After PCR-ampification, two DNA fragments with different lengths occur. Loss of heterozygozity can be detected as the loss of one of the two fragments. Mechanisms for LOH can be deletion, mitotic recombination, the loss of a whole chromosome (e.g. by nondisjunction) with or without duplicaton of the other allele (Caspary et al. 1997).

This type of analysis can determine the extent of LOH along the chromosome more exact than other, previously available methods. A limitation of this method is that it is not possible to identify the mechanisms leading to LOH. Specifically, translocations and changes in chromosome number cannot be detected and mitotic recombination cannot be clearly distinguished from deletion. To overcome these shortcomings, the results from LOH-analysis can be combined with those from whole chromosome in situ hybridization.

Analysis of mutants

The advantages of the combination of these two techniques are best explained by the following example (Fig. 8; Original examples see (Caspary et al. 1997)). Let us assume that LOH-analysis of a mutant colony showed only one signal at all tested microsatellites along chromosome 11. The first interpretation might have been that this mutant had developed as a result of chromosome 11+ loss with or without duplication of the other allele (11-). Another possibility would be mitotic recombination. Chromosome painting then showed the presence of two chromosome 11 alleles.

Thus, chromosome loss without duplicaton could be ruled out. This cell system has an additional feature that permits futher analysis. The centromeres of the two chromosome 11 alleles have different sizes and are not covered by in situ hybridization. Let us assume that the size was different in our case. Then only one interpretation was possible. There must have been mitotic recombination in which the former 11+ allele exchanged its material between a breakpoint adjacent to the centromere and the distal end (including the tk gene) with material from the 11- allele. Chromosome painting would also have revealed translocations or chromosome number changes. When spontaneous mutants were investigated (Liechty et al. 1997), it was found that about 30% showed a gene mutation and all others (about 70%) showed LOH to some extend. About one third of these showed LOH almost along the whole chromosome. Chromosome painting of these mutants is still in progress. However, a vast majority of the analyzed mutants with larger LOH were due to mitotic recombinations and not deletions or chromosome loss (W.J. Caspary, M.C. Liechty, and J.C. Hozier; unpublished results).

We used the topoisomerase II inhibitor etoposide as a first model compound for the analysis of mutants in this cell system with this combination of LOH-detection and chromosome in situ hybridization ((Stopper et al. 1997a) ; original data will be published separately). Etoposide induced a 10-fold increase in mutant frequency. From the mutants that we have analyzed so far, less than 10% were due to point mutations and/or small deletions (intragenic) mutation and more than 80% showed LOH to some extent. From those showing LOH, about 20% were due to mitotic recombination. There was no chromosome loss without reduplication of the other allele. Overall, the achievable percentage of mutations that are due to large chromosomal changes is higher in this system which is heterozygous for the selectable (tk) gene than in hemizygous (hprt) systems (McGregor et al. 1996). As a conclusion, we consider mitotic recombination an important mechanism for mutant induction. Mutant assay systems that are able to detect mitotic recombination should thus be preferred in genotoxicity testing.

Acknowledgements

I would like to thank the many collaboration partners that contributed to various areas of the results described here. Specifically, studies with the UV-microscopy technique and several other micronucleus-experiments were a collaboration with Dr. D. Schiffmann (Rostock, Germany); many of the methylation-related aspects were investigated together with Dr. W.J. Caspary (NIEHS/NIH, RTP, USA); F9 mouse teratocarcinoma cell experiments were a collaboration with Dr. W. Schulz (Düsseldorf, Germany). DNA curvature measurements were performed by Prof. Dr. S. Diekman (Jena, Germany); the methods for mutant analysis were developed by Liechty et al. (Liechty et al. 1997) and introduced to us by Dr. W.J. Caspary. Members of my working group, that delivered the original data for the results presented here, are Ms. I. Eckert, Dr. S. Kirchner, Mr. G. Boos, Mr. C. Korber and Ms. N. Herrmann.

References

Alonso A, Breuer B, Steuer B, Fischer J (1991) The F9-EC cell line as a model for the analysis of differentiation. Int. J. Dev. Biol. 35: 389-397
Boege F, Andersen A, Jensen S, Zeitler R, Kreipe H (1995) Proliferation-associated nuclear antigen Ki-S1 is identical with topoisomerase IIa. Delination of a carboxy-terminal epitope with peptide antibodies. Am. J. Pathology 146: 1302- 1308
Carr BI, Garett Reilly J, Smith SS, Winberg C, Riggs A (1984) The tumorigenicity of 5-azacytidine in the male Fisher rat. Carcinogenesis 5: 1583-1590
Caspary WJ, Stopper H, Hozier JC, Liechty MC, Davis LM (1997) Mechanisms leading to the expression of recessive alleles: the use of polymorphic microsatellites and whole-chronsomsome painting probes to analyze mouse tumors, mutants, and micronuclei. Rec. Res. Canc. Res. 143: 161-182
Downes CS, Mullinger AM, Johnson RT (1991) Inhibitors of topoisomerase II prevent chromatid separation in mammalian cells but do not prevent exit from mitosis. Proceedings of the National Academy of Science USA 88: 8895-8899
Hori TA (1983) Induction of chromosome decondensation, sister- chromatid-exchange and endoreduplications by 5-azacytidine, an inhibitor of DNA-methylation. Mutation Research 121: 47-52
Jones PA, Taylor SM (1980) Cellular differentiation, cytidine analogs and DNA methylation. Cell 20: 85-93
Kirchner S, Schiffmann D, Stopper H (1995) The influence of DNA methylation on topoisomerase II activity and its possible link with genomic instability in different cell lines of the Syrian hamster. Toxicology in Vitro 9(4): 519-525
Laird PW, Jaenisch R (1996) DNA methylation and cancer. Human Molecular Genetics 3: 1487-1495
Liechty MC, Hall BK, Scalzi JM, Davis LM, Caspary WJ, Hozier JC (1995) Mouse chromosome-specific painting probes generated from microdissected chromosomes. Mammalian Genome 6: 592-594
Liechty MC, Hozier J, Davis LM, Crosby Jr. H, Spencer DL, Caspary WJ (1997) Identification by loss of heterozygozity of a region distal to Tk1 influencing growth rates in mutants from L5178Y mouse lymphoma cells.
Marini JC, Miller KG, Englund PT (1980) Decatenation of kinetoplast DNA by topoisomerase. J. Biol. Chem. 255: 4976- 4979
Masumoto A, Masukata H, Muro Y, Nozaki N, Okazaki T (1989) A human centromere antigen (CENP-B) interacts with a short specific sequence in alphoid DNA, a human centromeric satellite. J. Cell Biol. 109: 1963-1973
McGregor DB, Brown AG, Cattanach P, Shepherd P, Riach C, Daston DS, Caspary WJ (1989) TFT and 6TG resistance of mouse lymphoma cells to analogs of azacytidine. Carcinogenesis 10: 2003-2008
McGregor DB, Riach C, Cattanach P, Edwards I, Shepherd W, Caspary WJ (1996) A comparison of mutagenic response of L5178Y mouse cells at the tk and hprt loci. Environm. Molec. Mutagen. in press
Obe G, Beek B, Vaidya F (1975) The human leukocyte test system III. Premature chromosome condensation from chemically and X- ray induced micronuclei. Mutation Research 27: 89-101
Razin A, Szyf M, Kafri T, Roll M, Giloh H, Scarpa S, Carotti D, Cantoni GL (1986) Replacement of 5-methylcytosine by cytosine: a possible mechanism for transient DNA-demethylation during differentiation. Proc. Natl. Acad. Sci. USA 83: 2827-2831
Schiffmann D, De Boni U (1991) Dislocation of chromatin elements in prophase induced by diethylstilbestrol: a novel mechanism by which micronuclei can arise. Mutat Res 246: 113-22 Issn: 0027-5107
Schmahl W, Geber E, Lehmacher W (1985) Diaplacental carcinogenic effect of 5-azacytidine in NMRI-mice. Cancer Letters 27: 81-90
Schmid M, Haaf T, Grunert D (1984) 5-Azacytidine-induced undercondensation in human chromosomes. Hum. Genet. 67: 257- 263
Sofuni T, Honma M, Hayashi M, Shimida H, Tanaka N, Wakuri S, Awogi T, Yamamoto KI, Nishi Y, Nakadate M (1996) Detection of in vitro clastogens and spindle poisons by the mouse lymphoma assay using the microwell method: interim report of an international collaborative study. Mutagenesis 11: 349-355
Stopper H, Eckert I, Boos G, Caspary WJ (1997a) Analysis of loss of heterozygozity and chromosome painting of etoposide-induced mutants. Naunyn-Schmiedeberg`s Archives of Pharmacology submitted
Stopper H, Eckert I, Liechty M, Hozier J, Davis L, Caspary WJ (1997b) Aneugen induced micronuclei containing the selectable gene do not lead to mutagenesis. Mutagenesis : in preparation
Stopper H, Eckert I, Schiffmann D, Spencer DL, Caspary WJ (1994) Is micronucleus induction by aneugens an early event leading to mutagenesis? Mutagenesis 9(5): 411-416
Stopper H, Eckert I, Wagener P, Schulz W (1997c) Formation of micronuclei and inhibition of topoisomerase II in the comet- assay in mammalian cells with altered DNA-methylation. Recent Results in Cancer Research 143: 183-193
Stopper H, K"rber C, Gibis P, Spencer DL, Caspary WJ (1995) Micronuclei induced by modulators of methylation: analogs of 5- azacytidine. Carcinogenesis 16(7): 1647-1650
Stopper H, Korber C, Schiffmann D, Caspary WJ (1993) Cell-cycle dependent micronucleus formation and mitotic disturbances induced by 5-azacytidine in mammalian cells. Mutat Res 300: 165-177
Stopper H, Pechan R, Schiffmann D (1992) 5-azacytidine induces micronuclei in and morphological transformation of Syrian hamster embryo fibroblasts in the absence of unscheduled DNA synthesis. Mutat Res 283: 21-28
Sumner AT (1991) Scanning electron microscopy of mammalian chromosomes from prophase to telophase. Chromosoma 100: 410- 418
Teubner B, Schulz W (1994) Exemption of satellite DNA from demethylation in immortalized derivatives of F9 mouse embryonal carcinoma cells. Exp. Cell Res. 210: 192-200