Cleavage of DNA by Mammalian DNA Topoisomerase 11*

Using the P4 unknotting assay, DNA topoisomerase I1 has been purified from several mammalian cells. Similar to prokaryotic DNA gyrase, mammalian DNA topoisomerase I1 can cleave double-stranded DNA and be trapped as a covalent protein-DNA complex. This cleavage reaction requires protein denaturant treatment of the topoisomerse 11-DNA complex and is reversible with respect to salt and temperature. The product after reversal of the cleavage reaction remains supertwisted, suggesting that the two ends of the pu- tatively broken DNA are held tightly by the topoisomerase. Alternatively, the enzyme-DNA interaction is noncovalent, and the covalent linking of topoisomerase to DNA is induced by the protein denaturant. Detailed characterization of the cleavage products has revealed that topoisomerase I1 cuts DNA with a four-base stag- ger and is covalently linked to the protruding 5"phos-phoryl ends of each broken DNA strand. Calf thymus DNA topoisomerase I1 cuts SV40 DNA at multiple and specific sites. However, no sequence homology has been found among the cleavage sites as determined by direct nucleotide-sequencing studies.


I1
from several mammalian cells. We report here our studies of the interaction between mammalian DNA topoisomerase I1 and DNA.

MATERIALS AND METHODS
Enzymes and Nucleic Acids-HeLa DNA topoisomerase I1 was purified according to the published procedure (13). Calf thymus DNA topoisomerase I1 was purified to homogeneity by a similar procedure, except that the P4 unknotting assay was used to monitor the enzyme activity(l4).' HeLa DNA topoisomerase I, and calf thymus DNA *This research was supported by National Institutes of Health Grant GM-27731 and Chicago Community Trust/Searle Scholars Program. 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.
B. D. Halligan and L. F. Liu, unpublished results. topoisomerse I were purified as the 100 kDa form (15).' Most experiments were performed with topoisomerases isolated from both mammalian sources. Plasmid DNA pBR322 dimer was purified by the standard procedure (16). SV40 DNA was purified as described previously (17). All restriction endonucleases and the large fragment of Escherichia coli DNA polymerase I were purchased from Bethesda Research Laboratories. T4 polynucleotide kinase was a gift from Dr. L. Klevan (Bethesda Research Laboratories).
Conditions for Topoisomerase-induced Cleavage of DNA-Cleavage of DNA by mammalian DNA topoisomerase I was done in a reaction mixture (50 pl) containing 10 mM Tris, pH 9, 0.5 mM EDTA, and 10 pg/ml of bovine serum albumin. Cleavage by mammalian DNA topoisomerase I1 was done in a reaction mixture (50 ~1 ) containing 10 mM Tris, pH 7.0,1 mM MgCl,, 0.5 mM EDTA, and 10 pg/ml of bovine serum albumin. Appropriate amounts of DNA and topoisomerases were added to initiate the reactions. After 5 min a t 37 "C, the reactions were terminated by the treatment with SDS' (1% final).
End Labeling of DNA Restriction Fragments-For labeling of the SV40 restriction fragments, 30 pg of SV40 DNA, digested with HpaII restriction endonuclease, were dephosphorylated with bacterial alkaline phosphatase (18). For 5'-end labeling, half of the sample was treated with T4 polynucleotide kinase and [Y-~'P]ATP (18). For 3'end labeling, the other half of the sample was treated with a large fragment of E. coli DNA polymerase I, [a-"P]dGTP, and unlabeled dCTP at 10 "C for 1 h (18). Unincorporated triphosphates were removed by gel filtration through a G-50 column equilibrated in 0.5 M NaC1, 50 mM Tris, pH 7.7, and 0.5 mM EDTA. Labeled DNA samples were concentrated by ethanol precipitation and then digested with KpnI restriction endonuclease. End labeling of pBR322 restriction fragments was done similarly.
S D S Precipitation of Topoisomerase-DNA Complexes-(a) SDS precipitation of double-stranded DNA-topoisomerase complexes. Reactions (50 pl each) were terminated by adding 100 pl of a stop solution containing 2% SDS, 2 mM EDTA, and 0.5 mg/ml of salmon sperm DNA, and heated to 65 "C for 10 min. Precipitation of topoisomerase-DNA complex was achieved by the addition of 50 p1 of 0.25 M KC1 and incubation on ice for 10 min. The precipitate was collected by centrifugation in an Eppendorf centrifuge for 15 min in the cold room. The supernatant was aspirated and the pellet was washed once with 200 pl of a solution containing 10 mM Tris, pH 8, 100 mM KCI, 1 mM EDTA, and 100 pg/ml of carrier salmon sperm DNA a t 65 "C for 10 min. Following cooling on ice and recentrifugation, the pellet was resuspended in 200 pl of H 2 0 with heating to 65 "C. The suspension was then quickly transferred to a vial containing 4 ml of scintillation fluid (Liquiscint, National Diagnostics) and counted. (b) SDS precipitation of single-stranded DNA-topoisomerase complexes. In order to denature the DNA, 0.2 N NaOH was included in the stop solution described above. Heating was done a t 37 rather than 65 "C. Precipitation of the topoisomerase-DNA complexes was achieved by the addition of the KC1 solution (0.25 M), which was modified to include 0.4 M Tris-HC1 in order to neutralize the NaOH. The rest of the procedure was identical with (a).
DNA Nucleotide Sequencing of Topoisomerease II Cleavage Sites-Plasmid DNA pCJl was constructed by replacing the EcoRI-Hind111 (31 bp) fragment of pBR322 DNA with the EcoRI-Hind111 (74 bp, nucleotide 1692-1769) fragment of SV40 DNA. E. coli "294 cells were used for transformation. 5'-end labeling of the EcoRI-Hind111 restriction fragment of SV40 DNA was done as follows. For labeling at the EcoRI site, pCJl DNA was cut with EcoRI and labeled with T4 polynucleotide kinase and [-,-3'P]ATP. Following Hind111 digest, The abbreviations used are: SDS, sodium dodecyl sulfate; bp, base pairs. the labeled I.:coRI-HindIII fragment (74 bp, labeled at the BC~JRI site) was gel isolated. For labeling at the Hind111 site, pCJl DNA was cut with Hind111 and 5'-end labeled as before. After labeling, the product was digested with EcoRI, and the EcoRI-Hind111 fragment (74 bp, laheled at the Hind111 site) was gel isolated. DNA was extracted from the low melting point agarose by the published procedure (18). Sequencing of the topoisomerase I1 cleavage sites on the isolated KcoRI-IfindIII fragments was done on a 20% polyacrylamide sequencinggel.

RESULTS
Rmersibilitv of the Topoisomerase II Cleavage Reaction-At high concentrations of calf thymus topoisomerase 11, cleavage of DNA substrates into the nicked form and the linear form was observed when the reaction mixture was terminated by the addition of SDS to 1% (Fig. 1, lanes A and R ) . This apparent nuclease activity of calf thymus topoisomerase 11, however, is quite different from the classical DNA endonucleases. The cleavage reaction was complete in less than 1 min after incubation at 37 "C. Prolonged incubation did not increase the cleavage product. Furthermore, addition of high concentrations of NaCl (0.25 and 0.5 M NaCl for samples in lanes H and I, respectively) to the reaction mixture after the incubation rapidly reversed the cleavage reaction (compare lanes A, B, H , and I). We have also observed that shifting the temperature from 37 to 0 "C also gradually reversed the cleavage reaction. Both form I1 and form I11 plasmid DNA were converted back to form I. This simple salt-induced reversal reaction suggested that cleavage of DNA by topoisomerase I1 was the result of SDS treatment which trapped a putative intermediate of the topoisomerase reaction. The fact that both form I1 and form I11 plasmid DNA were converted back to the supertwisted form but not to the relaxed form further suggested that the two broken DNA ends of the putative enzyme-DNA complex were held tightly by the enzyme. Alternatively, the enzyme-DNA interaction is noncovalent and the covalent linking of enzyme to DNA was induced when the protein denaturant was applied. At this time, we are unable to distinguish between these two possibilities. The require- mM 'his, pH 7.0. 1 mM MgCI,, 0.5 mM EDTA, 10 pg/ml of bovine serum albumin, 120 pg/ml of pRR322 DNA dimer, and 100 pg/ml of calf thymus topoisomerase 11 were incubated at 37 "C for 10 min. After incubation, a 4-pl aliquot (lune H ) was withdrawn and stopped immediately. The rest were distributed (4 p1 of each) into tubes each containing 1 pl of NaCl at various concentrations (lanes C to I). Incubation was continued for another 1 min and then stopped. All reactions were stopped with 30 pl of 1 5 SDS and then treated with 0.1 mg/ml of proteinase K at 37 "C for 30 min. I h e A is the control (no enzyme). Final NaCl concentrations of samples in lunes C to I were 8. 16 ments for this reversible cleavage reaction have been characterized. ATP (or dATP) which is required for the catalytic activity of topoisomerase 11, is not required for the cleavage reaction. Contrary to mammalian DNA topoisomerase I which does not show a pH optimum for cleavage, topoisomerase I1 cleavage reaction shows a sharp pH optimum at about 5. Mg(I1) ion stimulates cleavage but is not absolutely required. The cleavage reaction is strongly inhibited by salt above 100 mM NaCI. Single-stranded DNA is also cleaved by topoisomerase I1 a t about the same efficiency. The catalytic reactions of topoisomerase I1 (e.g. unknotting reaction) are very sensitive to sulfhydryl reagents such as N-ethylmaleimide, while the cleavage reaction is quite resistant to sulfhydryl inactivation. It is possible to inactivate the catalytic activity of mammalian DNA topoisomerase I1 almost completely, without affecting the efficiency of the cleavage activity, by using sulfhydryl reagents (data not shown).

A B C D E F G H I
Covalent Association of Topoisomerase II to Each .5'-end of the Broken DNA Strands-When the topoisomerase I1 cleavage product was electrophoresed in an agarose gel containing 0.196 SDS (16), the mobilities of both the linear form and the nicked form of plasmid DNA were retarded suggesting that topoisomerase I1 is still tightly linked to the cleavage products (data not shown). In order to study the cleavage reaction of topoisomerase I1 more quantitatively, we have developed a procedure which selectively precipitates the protein bound nucleic acids (see "Materials and Methods"). The cleavage product can thus be quantitated by the precipitated counts if the DNA is labeled with radioactivity. Fig. 2 shows such an  (Fig. 2, A and R, solid circles), or topoisomerase I1 (Fig. 2, C and D, solid circles), the radioactivity in the precipitates also increased, suggesting that more covalent topoisomerase-DNA complexes were formed. T o confirm that the SDS precipitation procedure only precipitated the covalent topoisomerase-DNA complexes, topoisomerase cleavage products were parallelly quantitated by alkaline agarose gel electrophoresis (discussed in the next section). The electrophoresis results confirmed that the SDS precipitation procedure indeed selectively precipitated DNA molecules which had covalently linked proteins (data not shown). This simple SDS precipitation procedure was also exploited to determine the polarity of the topoisomerase linkage. In this experiment, singlestranded DNA-topoisomerase complexes were precipitated (see "Materials and Methods)" (Fig. 2, triangles). Depending on whether the topoisomerase was linked to the 3'or 5"ends of the broken DNA strands and whether 3'or 5'-end labeled DNA was used, protein-DNA complex may or may not be associated with the labeled ends. A similar analysis has been used to determine the polarity of the E. coli DNA topoisomerase I linkage (19). In the case of mammalian DNA topoisomerase I, labeled protein-DNA complexes were recovered by precipitation only when 5'-end labeled DNA was used (Fig.  2B, triangles). No labeled DNA was precipitated when 3'-end labeled DNA was used ( Fig. 2A, triangles), even though protein-DNA complexes did form as revealed by precipitation under neutral conditions ( Fig. 2A solid circles). This result strongly suggests that calf thymus DNA topoisomerase I is covalently linked to the 3'-ends of the broken DNA strands, consistent with the previous report (20). In the case of calf thymus DNA topoisomerase 11, the opposite was observed (Fig. 2, C and D), suggesting that calf thymus DNA topoisomerase I1 is covalently linked to the 5'-ends of the broken DNA strands.
Electrophoretic Analyses of t h Topoisomerase Cleavage Sites on SV40 DNA-In order to map the cleavage sites, linearized sV40 DNA was uniquely labeled at one end by a procedure described in Fig. 3. SV40 DNA was linearized with HpaII restriction endonuclease and end labeled at either the 3'-or 5'-ends with "'P (see "Materials and Methods"). KpnI restriction endonuclease was then used to clip a 50-bp fragment from one end of the labeled DNA. Since the 50-bp fragment (containing one labeled end) is small enough, no attempt was made to separate it from the 5193-bp fragment which contained the other labeled end. Cleavage of labeled SV40 DNA by topoisomerase I and I1 was analyzed both by native and alkaline agarose gel electrophoresis (18). Fig. 4 shows the autoradiograms of the topoisomerase cleavage products analyzed by native agarose gel electrophoresis. Very little double-stranded DNA cleavage was observed with calf thymus topoisomerase I using either 3'-end labeled SV40 DNA (lanes R to E) or 5'-end labeled SV40 DNA (lanes F to I). At the highest topoisomerase I concentration, a low level of double- varied from preparation to preparation. The basis for this variation is not known.
T o investigate whether all the cleavage products have topoisomerases covalently linked to the ends, cleavage products Cleavage of DNA by Mammalian DNA Topoisomerase II which were not treated with proteinase K were also analyzed by alkaline agarose gel (Fig. 5, lanes D, G, K , and N). When 3'-end labeled DNA was used for topoisomerse I cleavage, omission of proteinase K treatment did not change the mobility of the cleavage products in the gel (Fig. 5, compare lanes   C and D). However, all the cleavage products were retarded in gel mobility when 5'-end labeled DNA was used (Fig. 5,  compare lanes F and C). This result strongly suggested that all the topoisomerase I cleavage products had proteins cova-  K , M, and N ) . Samples in lanes D, C, K, and N were the same as those in lanes C, F, J , and M, respectively, except that proteinase K treatment was omitted. Lanes B and I were the controls (no enzyme) for the 3'-end labeled SV40 DNA. Lanes E and L were the controls (no enzyme) for the 5'-end labeled DNA. Lanes A and H were the molecular weight marker. The mobility shift (compare l a t . . s F with C, and J and K ) in alkaline agarose gel is characteristic for covalently linked protein-DNA complexes. The polarity of the topoisomerase linkage can also be determined from this analysis.
A Eco R I lently linked to the 3'-ends of the broken DNA strands. The same experiment was repeated using calf thymus topoisomerase I1 (Fig. 5, lanes J, K , M, and N). In this case, the result was totally the opposite. The mobility shift was only observed when the 3'-end labeled DNA was used. We thus concluded that topoisomerase I1 was covalently linked to each 5'-end at the cleavage sites.
The 3'-ends of the Cleavage Product of Calf Thymus Topoisomerase II Possess Hydroxyl Croups-The 3'-ends of the cleavage products produced by calf thymus DNA topoisomerase I1 were analyzed by the enzymatic modification using both terminal transferase and E. coli DNA polymerse I. We had shown that the cleavage products of topoisomerase I1 could be labeled by both the terminal transferase and polymerase I, suggesting that the 3'-ends of the cleavage products possessed hydroxyl groups and were recessed (data not shown). To test whether all the cleavage products had the same 3'-OH ends, the following experiment was performed (Figs. 6 and 7). as substrate) were labeled with [a-:"P]dATP using the large fragment of E. coli DNA polymerase I, cut with EcoRI (Fig.   6A), and then analyzed by native agarose gel electrophoresis ( Fig. 7, land B). The cleavage products, which were labeled, appeared as specific bands in t h gel (Fig. 7, lane B). To prove that these labeled bands were labeled at the cleavage sites by the polymerase reaction, a different experiment was performed ( Fig. 6, panel B). In this experiment, the same plasmid DNA was linearized first with EcoRI, end labeled with polymerase, and then cleaved by topoisomerase I1 (Fig. 6B,  were determined by DNA sequencing (Fig. 8). The cleavage sites on both complementary strands were schematically shown in Fig. 9. Several interesting features of the cleavage sites were noted. 1) There were more than 12 cleavage sites on this 74-bp fragment. The cleavage efficiency varies greatly from site to site. 2) For each cleavage site, the two cuts on the complementary strands were staggered by four base pairs. Site X was an exception, as we could not locate the cut on the complementary strand. 3) No consensus sequence can be deduced for all the cleavage sites.

DISCUSSION
Eukaryotic DNA topoisomerase I1 has enzymatic activities analogous to bacteriophage T4-induced DNA topoisomerase (10,21). Our present characterization of the cleavage reaction of topoisomerase I1 further demonstrated the similarity between prokaryotic and eukaryotic DNA topoisomerase 11.
Similar to DNA gyrase, mammalian DNA topoisomerase I1 cuts DNA with a four-base stagger and is covalently linked to the 5'-phosphoryl end of each broken DNA strand (5). Different from DNA gyrase, however, mammalian DNA topoisomerase I1 can efficiently cleavage DNA in the absence of any added drugs. Similar to other DNA topoisomerases, cleavage of DNA by mammalian DNA topoisomerase I1 is the result of protein denaturant treatment of the topoisomerase 11-DNA complex (5, 22). It is particularly significant that the salt-or temperature-induced reversal of the cleavage reaction gave products that were superhelical rather than relaxed. This result strongly suggests that the topoisomerase 11-DNA complex must be organized in such a way that either the two ends of the broken DNA are tightly bound by the enzyme a t all times or the covalent linking of enzyme to DNA occurs only when the protein denaturant was added. It is interesting that ATP is not required for the cleavage reaction. It is possible that the ATPase function is associated with the transport of another DNA segment through this breakage and rejoining site.
We do not know the in uiuo significance of this partial reaction of topoisomerase 11. The cleavage reaction of eukaryotic topoisomerase I1 is rather efficient under the physiological conditions, compared with that of the prokaryotic DNA gyrase which is highly dependent on oxolinic acid (3-5). Whether this cleavage reaction of topoisomerase I1 is related to certain illegitimate recombination processes in eukaryotic cells remains to be tested. The cleavage reaction of topoisomerase I1 (or I) may also provide a useful way for the mapping of the topoisomerase I1 (or I) binding sites on chromatin. However, we have not yet correlated the cleavage sites with the binding sites.
The biological functions of eukaryotic DNA topoisomerases are still unknown. The similarity between eukaryotic DNA topoisomerase I1 and prokaryotic DNA topoisomerase I1 (especially T4 DNA topoisomerase) suggests possible similar functions. it is likely that, similar to prokaryotic DNA topoisomerases, the multiple eukaryotic DNA topoisomerases regulate the topological structure of chromatin and thus control a variety of genetic processes. We have shown that eukaryotic DNA topoisomerase I1 is capable of catenating SV40 chromatin in vitro.:' Whether eukaryotic DNA topoisomerase I1 is capable of supertwisting chromatin in vivo is of fundamental importance to our understanding of the structure and function of chromatin. The effect of DNA supercoiling on transcription in eukaryotic cells has been noted ( 2 3 ) . It has been suggested that committed genes in eukaryotic cells may be under negative superhelical tension (24). Whether or not this tensioned state of DNA is controlled by DNA topoisomerases is still not known. Many attempts have been made to demonstrate the supertwisting reaction of eukaryotic topoisomerase I1 in vitro, and all have failed. It is possible that the supercoiling reaction is intimately related to the wrapping of DNA around the enzyme, as demonstrated in the case of DNA gyrase (25, 26).
Perhaps specific DNA sequences and/or a missing component(s) might stabilize such a wrapping around eukaryotic topoisomerase I1 in uiuo.