Effects of DNA methylation on topoisomerase I and II cleavage activities.

DNA methylation is deregulated during oncogenesis. Since several major anti-cancer drugs act on topoisomerases, we investigated the effects of cytosine methylation on topoisomerase cleavage activities. Both topoisomerase I and II cleavage patterns were modified by CpG methylation in c-myc gene DNA fragments. Topoisomerase II changes, mainly cleavage reduction, occurred for methylation sites within 7 base pairs from the topoisomerase II breaks and were different for VM-26 and azatoxin. For topoisomerase I, cleavage enhancement as well as suppression were observed. Using synthetic methylated oligonucleotides, we show that hemimethylation is sufficient to alter topoisomerase I activity. Cytosine methylation on the scissile strand within the topoisomerase I consensus sequence had strong effects. Cleavage was stimulated by methylation at position -4 and was strongly inhibited by methylation at position -3 (with position -1 being the enzyme-linked nucleotide). This inhibitory effect was attributed to the presence of a methyl group in the major groove, since the transition uracil to thymine also inhibited cleavage. Altogether these results suggest an interaction of topoisomerase I with the DNA major grove at positions -3 and -4. In addition, DNA methylation may have profound effects on the activity of topoisomerases and may alter the distribution of cleavage sites produced by anticancer drugs in chromatin.

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank-IEMBL Data Bank with accession number(s) X00364, K01908, V00501. $  DNA imprinting, inactivation of one X chromosome for the female and in the inactivation of some tissue-specific genes (7). DNA topoisomerases are essential for RNA transcription and DNA replication by relieving DNA torsional tension (swiveling) and resolving intertwined DNA molecules (unknotting and decatenation) (8)(9)(10)(11). Topoisomerase reactions involve the transesterification of a DNA phosphodiester linkage to a tyrosine residue of the enzyme. The protein-associated strand breaks, called cleavable complexes, form transient gaps in the DNA backbone which allow DNA topoisomerization reactions. Topoisomerase inhibitors which are among the most active anticancer agents trap the enzyme cleavable complexes (12). The resulting lesions are probably responsible for the cytotoxicity of these drugs (13)(14)(15). Camptothecin derivatives selectively inhibit topoisomerase I (topl)' (15,16), whereas amsacrine, mitoxantrone, anthracycline, and epipodophyllotoxin derivatives inhibit topoisomerase I1 (top21 (16,17).
The aim of the present study was to investigate the effects of DNA methylation on the activity of topl and top2 using specific inhibitors, since chromatin modifications induced by methylation might change both the enzyme activities and the DNA sequence selectivity of topoisomerase inhibitors. We have examined the in vitro effects of cytosine methylation on the cleavage activity of topl and top2 in the first exon and first intron of the human c-myc gene, and investigated more precisely the interactions of topl with DNA using synthetic methylated oligonucleotides.
EXPERIMENTAL PROCEDURES Materials, Chemicals, and Enzymes-Etoposide (VP-16) and teniposide (W-26) were obtained from Bristol-Myers Co., Wallingford, CT. 20-S-Camptothecin and amsacrine were obtained from the Drug Synthesis and Chemistry Branch (National Cancer Institute, Bethesda, MD). Azatoxin was provided by Dr. T. Macdonald, Department of Chemistry of Virginia, Charlottesville, VA. Drug stock solutions were made in dimethyl sulfoxide at 10 m, and further dilutions were made in distilled water immediately before use. Human c-myc DNA inserted in pBR322, restriction enzymes, T4 polynucleotide kinase, Taq DNA polymerase, calf thymus DNA topoisomerase I, and polyacrylamide/ bisacrylamide were purchased from Life Technologies, Inc., from the American Type Culture Collection (Rockville, MD), from Perkin-Elmer, or from New England Biolabs (Beverly, M A ) . [Y-~~PIATP and d 2 Pcordycepin were purchased from DuPont NEN. Top2 was purified from mouse leukemia L1210 cell nuclei as described previously (18). Oligonucleotides were prepared using a 392 DNA synthesizer from Applied Biosystem (ABI, Foster city, CA) and purified using oligonucleotide purification cartridges (ABI). Phosphoramidite of modified deoxynucleotides were purchased from Glen Research (Sterling, VA).
Preparation of End-labeled DNA Fragments by PCR-Three sets of labeled DNA fragments were prepared from the human c-myc gene by PCR. A 254-base pair DNAfragment from the first intron was prepared topoisomerase 11; CPT, camptothecin; MeC, methylcytosine; MIF, c-myc The abbreviations used are: topl, DNA topoisomerase I; top2, DNA intron factor; PCR, polymerase chain reaction; W-26, teniposide; VP- 16, etoposide.

DNA Methylation Affects
Topoisomerase Activities DNA methylation by SssI methylase affects topl and top2 cleavage sites in the c-myc first intron. DNA fragments were prepared by PCR using one primer labeled with 32P at the 5' terminus. Cleavage reactions were a t 37 "C for 30 min with purified enzymes and stopped by adding SDS and proteinase K. DNA electrophoresis was in 7 8 denaturing acrylamide gels (7 u urea) in TBE buffer. Treatments are indicated above each lane, including DNA methylation a t CpG sequences using SssI methylase (+ and -); CPT: topl + 10 PM CPT, VM-26: top2 + covalently linked to top2 and letters on the left to topl cleavage sites. Stars correspond to sites for the SssI methylase. A, c-myc sense strand labeled 10 pu VM-26; FA: purine ladder obtained after formic acid reaction. Numbers on the right correspond to genomic position of the nucleotide a t position 3035. CpG methylation induced a topl site, site C, in the presence of CF'T. B, c-myc antisense strand labeled a t position 3288. between positions 3035 and 3288, with numbers referring to GenBank genomic positions, as described previously (19). A 403-base pair DNA fragment from the junction between the first intron and first exon was TGCCGCATCCACGAAACTMGC-3' as sense primer and 5"GAACT-prepared between positions 2671 and 3073 using oligonucleotides: 5'-GTTCAGTGTMACCCCG-3' as antisense primer. A 481-base pair DNA fragment from the first exon was prepared between positions 2265 and 2745 using oligonucleotides: 5'-GATCCTCTCTCGCTAATCTCCGC-CC-3' as sense primer and 5'-TCC'R'GCTCGGGTGTTGTAAG'l"CC-3' as antisense primer. Single end labeling of these DNA fragments was obtained by 5'-end labeling of the adequate primer oligonucleotide (19). Approximately 0.1 pg of the c-myc DNA that had been restricted by SmaI and PuuII (fragment 2265-2745 which contained the exon I),XhoI and XbaI (fragment 2671-3073 corresponding to the junction exon Iintron I and fragment 3035-3288 corresponding to the first part of intron I) was used as template for the PCR (19).
Enzymatic Methylation of c-myc DNA Fragments-PCR products prepared for the mapping of the topoisomerase cleavage sites were methylated for 2 h using 10 units of SssI methylase (New England Biolabs, Beverly, MA). The unmethylated DNAcontrol was incubated in parallel without methylase. DNAsamples were purified by ethanol precipitation after two cycles of protein extraction using Strataclean ResinT" (Stratagene, La Jolla, CA).
Oligonucleotides-Modified oligonucleotides containing methylcytosine or uracil bases were prepared using the methylated 2"deoxycytidine or the 2"deoxyuridine phosphoramidite reagents from Glen Research (Sterling, VA), respectively. Duplex oligonucleotides derived  Upper profile corresponds to DNA pretreated with SssI methylase (Me). The boxed sequence corresponds to a palindrome containing a prominent top2 cleavage site at the beginning of the first intron. It is also part of the MIF-2 binding site (24). Numbers correspond to genomic position of the nucleotide covalently linked to the topl, and stars correspond to sites for SssI methylase. from the c-myc first intron and starting at the genomic position 3221 (GenBank numbering) were modified at the underlined positions. The sense strand (S) is indicated S , when methylated at position 3237. The antisense strand (A) is indicated A,,, when methylated a t position 3238. Preparation of 3"labeled oligonucleotides on the scissile sense strand (upper strand above) was performed as described previously (20).
Topoisomerase-induced DNA Cleavage Reactions-DNA fragments ,5' were equilibrated with or without drug in 1% dimethyl sulfoxide, 10 m M Tris-HCI, pH 7.5, 50 m~ KCI, 5 mM MgC12, 2 nm dithiothreitol, 0.1 r n~ Naz EDTA, 1 mM ATP, and 15 pg/ml bovine serum albumin for 5 min before addition of purified top2 (40-70 ng) in 2 0 4 final reaction volume. For reactions with topl, ATP was omitted, and 7-30 units of enzyme were used. Unless otherwise specified, reactions were performed a t 37 "C for 30 min. Reactions were stopped by adding 1% sodium dodecyl sulfate (SDS) and 0.4 mg/ml proteinase K (final concentrations) followed by an additional incubation at 50 "C for 30 min. Samples were ethanol-precipitated before separation of the topoisomerase-cleaved fragments on denaturing polyacrylamide gels.
Electrophoresis and Data Analysis-Sequencing gels were made of 7 or 16% polyacrylamide in 1 x TBE buffer (90 mM Tris borate, 2 m M EDTA, pH 8.3). Electrophoresis was a t 2500 V (60 watts) for 2-5 h. Gels were quantified using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) or a Betascope 603 Blot Analyzer (Betagen Inc., Waltham, MA). For oligonucleotides, the percentage of cleavage a t a given site was calculated as the radioactive signal at this site minus the radioactivity signal in the oligonucleotide control at the corresponding position (background counts) divided by the total amount of radioactivity signal in the lane.  The oligonucleotides were 3"labeled and reactions were run a t 37 "C for 30 min as described in the legend to Fig. 1

. Lanes I , DNA
alone; lanes 2, + topl; lanes 3, + topl + 10 1.1~ CPT. Numbers on the right of the autoradiograph correspond to the genomic position of the nucleotide covalently linked to the topl and stars to methylation sites. topoisomerase cleavage sites in vivo (21). It is also rich in CpG sequences that can be methylated in vitro using SssI methylase. Fig. 1 shows the topl and top2 cleavage sites obtained in a 5'-region of the first intron on the sense and the antisense strands. Top2 sites were mapped and designated by the genomic position of the covalently linked nucleotide (at the 5' termini of the top2 breaks). Nonintercalators (W-16, V"26, and azatoxin (22)) were used to stimulate top2 cleavages in order to avoid a direct effect of DNA methylation on drug binding. Also, these drugs induce many cleavage sites and are among the less sequence-specific top2 inhibitors (22). Therefore alterations of drug-induced top2 cleavage correspond to the enzyme response to DNA methylation.
Top2 cleavage sites were affected by CpG methylation. Reduction of cleavage intensity can be observed near the methylation sites (Figs. 1 and 2, sites 3121,3149,3167,3175,3152, and 3170). The inhibitory effect can be observed with CpG methylation as far as 7 base pairs from the top2 sites (site 3149, Fig. 2) and did not extend past 11 bases (site 3145 was not affected by MeC 3128, Fig. 2). Fig. 2 also shows the 4 base pair stagger for top2 double strand breaks and the coordinated inhibition of cleavage on both strand by CpG methylation (site 3149). Additionally, the strong intensity of the double strand cleavage site 314513148 has never been reported before. Interestingly, this site is within an AT palindrome (231, which is a regulatory element binding site (24). Top2 sites can also be stimulated by DNAmethylation and stimulation can depend on the drug used (sites 2744, 2748, and 2757 were stimulated by azatoxin but not by VP-16) (Fig. 3). Assuming complete methylation, CpG methylation on the bases immediately flanking the cleavage site or within the top2 base pair stagger can affect cleavage intensity (Fig. 3, sites 2746,2748,2757, and 2761). In contrast, site 3129, in Fig. 2, was not affected by MeC 3156 that Topoisomerase Activities induced a reduction of site 3121. Additional experiments showed that the strong top2 site induced by amsacrine a t the P2 promoter (21) was strongly decreased by DNA methylation (results not shown). In conclusion, CpG methylation within 7 bases from top2 cleavage sites affects top2 cleavage, in agreement with the minimal size requirements for top2 binding (25). This effect was dependent on the drugs used.
Effect of CpG Methylation on Drug-induced Topl Cleavage Sites-Top1 cleavage sites induced in the presence of the CFT were affected by methylation sites nearby (Fig. 1). However, the 5'4abeling used in the experiments shown in Fig. 1 did not allow precise mapping of topl cleavage sites. Indeed, despite proteinase K digestion, topl polypeptides remain covalently linked to the 5'-DNA termini. As a result, the electrophoretic migration of the DNA fragment is retarded. Therefore, cleavage sites were labeled with letters. Reductions of the cleavage intensities were seen (Fig. LA, site E , and several sites in the exon 1; results not shown). Interestingly, CpG methylation could also induce topl cleavage sites (Fig. LA, site C). This induction was unexpected regarding the association of topl activity with the DNA transcription and the inhibitory effect of DNA methylation on transcription. Further investigations were performed using methylated oligonucleotides modified during the synthesis by incorporation of methylcytosine. This approach allows, at will, MeC positioning around the cleavage site.
Mapping of sites B and C (this latter induced by CpG methylation) was performed using an oligonucleotide prepared from c-myc genomic positions 3221-3278. MeC were included in the CpG sequence at position 3237 on the sense strand and 3238 on the antisense strand, and the oligonucleotide was 3"end-labeled with 32P-cordycepin on the sense strand. Genomic positions of topl cleavage site were determined by comparison with a purine lane bearing a similar 5' terminus. Site B was mapped at position 3256 and site C at position 3240 (Fig. 4). Utilization of synthetic oligonucleotides also enabled us to study the effect of hemimethylation. We compared CPT-induced cleavage in unmethylated, hemimethylated, and fully methylated oligonucleotides (Fig. 5). Site 3240 was barely visible in the unmethylated DNA (SA), and cleavage was most enhanced by full methylation on the sense and antisense strand ( S A , 1. Hemimethylation of the scissile strand ( S A ) had a greater enhancing effect than hemimethylation of the uncleaved strand (SA, 1.

Kinetics of Formation and Resealing of the Topl Sites 3240
and 3256-Previous studies have suggested an inverse relationship between the stability of a given cleavage site and its intensity (26). Accordingly, the induction of topl site 3240 after cytosine methylation may result from its greater stability. This hypothesis was tested using unmethylated, fully methylated, and hemimethylated oligonucleotides (Fig. 6). Cleavage sites were allowed to form at room temperature for 5 min before their resealing was initiated by adding sodium chloride at 0.35 M final concentration. The rapid kinetics of resealing for site 3256 is characteristic of most topl cleavage sites (26L2 Unexpectedly, the kinetics of resealing for site 3240 was extremely slow regardless of the methylation status (Fig. 6B 1. Kinetics of formation of topl cleavage was then investigated in the unmethylated (SA) and fully methylated (SA,) oligonucleotides (Fig. 7, A and B). Reactions were performed a t 10 "C to slow down reaction kinetics and to allow more accurate quantifications. Site 3256 was induced very rapidly in both oligonucleotides, whereas site 3240 was only induced in the S,k, oligonucleotide. Its formation kinetics was slower than that of site 3256 (Fig. 7B 1. Therefore, the effect of methylation on site 3240 appeared to be essentially an inductive effect rather than an inhibition of religation.

Relationship between the Position of the Methylcytosine Group around the Topl Cleavage Site and Topl Cleavage
Zntensity-We took advantage of the presence of 4 consecutive cytosines 5' to the 3256 topl site to test the effect of a MeC at positions -3 to -6 (Fig. 8). These hemimethylated substrates are not physiological, but they allow further study of the DNA-enzyme interactions. MeC at position -3 strongly reduced cleavage (by 95%), whereas MeC a t position -4 increased cleavage (by 50%), consistent with the effect of hemimethylation a t position -4 from the cleavage site 3240. Methylation at positions -5 or -6 had a modest inhibitory effect (around 30%). Mutations at positions +1 on the scissile strand or +1 and +2 on the uncleaved strand of an oligonucleotide containing the strong topl site isolated in the tetrahymena rDNA (27) did not reveal any cleavage alteration when C was replaced by MeC at these positions (data not shown). Since T-1 is strongly preferred for topl cleavage (20, 261, MeC at positions -1 and -2 were not investigated. Taken Time (min) 7897 together, our results indicate that cytosine methylation within the topl consensus sequence (the 4 bases upstream the cleavage site) has profound effects on topl cleavage activity.

The Presence of a Methyl Group in the Major Grove at Position -3 Appears to Suppress Topl
Cleavage-% confirm that the addition of a methyl group at position -3 was responsible for the inhibitory effect on topl cleavage, we investigated the effect of different methylated pyrimidines. The base at position 3254 was changed from cytosine (C) to MeC and from uracil (U) to thymine (T). Because the introduction of a T or a U required the modification of the complementary strand, the U versus T and C versus MeC oligonucleotides should be compared (Fig. 9). Interestingly, the addition of a methyl group on the cytosine and the uracil was similarly responsible for inhibition of cleavage a t site 3256. Therefore, the presence of a methyl group in the major grove at position -3 was responsible for the reduction of the 3256 topl cleavage site. Additionally, the introduction of

DISCUSSION
Topoisomerases interact preferentially with specific DNA sequences. Mapping of topoisomerase sites was first performed in vitro using purified DNA and enzymes in many different systems such as the SV40 DNA (23,26,29,30), the c-myc gene (21, 31), and the rDNA(32). More recently, topoisomerase sites were also mapped in vivo using SV40 (281, the c-myc gene (21, 33, 341, the fos gene (351, the rDNA(32), the histone genes (36), and the heat shock genes (36,37). Comparison of in vitro and in vivo cleavage sites mapping shows differences. Mainly in vivo sites are less numerous and correspond to a subset of the in vitro  sites (38,39). Remarkably, when mapping data are available a t the nucleotide level, in vivo sites are visible in vitro, and, a t least for topl, the same consensus sequences have been obtained in cellular and purified systems (20, 26, 28). The quantitative reduction of topoisomerase cleavage sites observed in vivo may have a variety of origins related to chromatin structure. Among the possible factors are the gene transcription activity (37, 4 W 3 ) , the presence of nucleosomes, which protects the bound DNA and reinforces the cleavage activity in the linker region (38), and the DNA modifications, such as DNA glycosylation in phage T4 (44). In the present work, we demonstrate that DNA methylation a t CpG sequences can alter the topoisomerase-DNA interactions and may contribute to the cleavage differences observed in vitro and in vivo.
We find that cytosine methylation alters CPT-induced topl cleavage sites in the c-myc DNA proto-oncogene (GenBank identification for the human c-myc oncogene is HSMYCC) as well as in synthetic oligonucleotides. The most important effects resulted from the methylation on the 5'-flank of the cleavage site (Fig. 81, in agreement with the localization of essential bases for topl binding (20, 26, 27). Cleavage was enhanced by MeC a t position -4 on the scissile strand and to a lesser extent by MeC a t position -3 on the uncleaved strand. In fully methylate? CpG, the two methyl groups lay side by side at about 7.6 A from each other symmetrically from the longitudinal major groove axis (45). Therefore, they may interact with a common hydrophobic pocket of topl. Enhancement of DNA-binding protein activity in the presence of MeC has also been reported for the human immunodeficiency virus integrase (46) and for the MeCpG binding proteins (5). By contrast, N7 guanine methylation at position -3 on the uncleaved strand inhibits topl cleavage activity (27). Also, topl cleavage was inhibited by cytosine methylation at positions -3, -5, or -6 from the cleavage site on the scissile strand. The most prominent effect was a t position -3 (Fig. 81, within the consensus sequence for topl DNA binding (20, 26, 27). It is unlikely that CpG methylation in the oligonucleotides could alter DNA structure (45) and facilitate a B to Z transition, since CpG methylation was neither in alternating CpG repeats nor in proximity of poly(A) (47). Rather, the effect of CpG methylation is a direct consequence of the addition of a methyl group in the major groove, because cleavage inhibition was also observed after the replacement of uracil by thymine at position -3 on the scissile strand (Fig. 9). The importance of DNA major groove interactions for topl binding has recently been reported for the vaccinia enzyme (48). Hence, topl interacts closely and precisely with the DNA major groove a t positions -3 and -4 upstream from the cleavage site. Methylation at position -3 of the scissile strand suppresses cleavage, whereas methylation at -4 increases cleavage.
The present results also show that the presence of T at position -3 on the scissile strand strongly suppressed topl cleavage and shifted the cleavage site by one nucleotide in the upstream direction (Fig. 9). These data are consistent with those obtained by base sequence analysis in vitro (20,26) or in vivo (28). Additionally, MeC and T share the same positive and negative effects on topl cleavages. Their methyl groups are identically positioned in the major groove of B-DNA. Consequently, methyl group hindrance is responsible for the strong inhibitory effect of T or MeC at position -3. In contrast, the same methyl group at position -4 on the scissile strand (Figs. 5  and 8) or at position -3 on the uncleaved strand (Fig. 5) in- Cytosine 3254 was replaced by MeC, uracil (U), or thymine (T); for the last two substitutions, base complementary was preserved by performing simultaneously a transition G to A on the uncleaved strand. Reactions were performed as described in the legend to Fig. 8. Lanes 1, DNA alone; lanes 2, + topl alone; lunes 3, + topl + 10 p~ CPT.
creases the topl-induced DNA cleavage. These results suggest that DNA-top1 interactions in the consensus sequence must take place in the major groove at least a t -3 and -4 positions on both strands.
Current mechanisms for transcription inhibition by CpG methylation involve the fixation of specific proteins with (6) or without (5,49) DNA sequence selectivity which interferes with the binding of transcription factors. Alternatively, DNA methylation may directly interfere with the DNA binding of tran- repeat (HIV-LTR) (51), the retinoblastoma binding factor 1 (52), the CAMP response element-binding protein (53), AP-2 (54), and E2F (55). Hemimethylation can be sufficient to inhibit the binding of transcription factors (56) and cellular transcription (57). CpG hemimethylation was also sufficient to affect topl cleavage activity (Fig. 5). Top2 cleavage activity was also altered by CpG methylation, including at palindromes that are likely to be involved in transcription regulation (site 3145 in the first intron within the MIF-2 binding sequence (24), Fig. 2; and P2 promoter (21)). Hence, alteration of the topoisomerase functions may contribute to the effects of CpG methylation on transcription. Topoisomerase I1 inhibitors differ in their DNA sequence selectivity (11) which has been attributed to the preferential stacking of drugs with the bases that flank the cleavage sites (58). Among topoisomerase I1 inhibitors, activity differences observed in the clinics might be related to differences of top2induced DNA genomic lesions. Consequently, modifications of topoisomerase cleavage patterns by DNA methylation might have profound effects on drug activity because methylation status changes with cell differentiation, aging (3, 59) and oncogenesis (60,61). Since alterations of gene methylation might be reversed after efficient therapy (62,63), and gene methylation might be involved in the evolution to chemotherapy resistance (64-66), we postulate that methylation might modulate the tissue selectivity and efficiency of topoisomerase inhibitor cytotoxicity.