Site-specific Recognition of Bacteriophage T4 DNA by T4 Type I1 DNA Topoisomerase and Escherichia coli DNA Gyrase*

The site specificity of bacteriophage T4-induced type I1 DNA topoisomerase action on double-stranded DNA has been explored by studying the sites where DNA cleavages are induced by the enzyme. Oxolinic acid addition increases the frequency at which 4x174 duplex DNA is cut by the enzyme by about 100-fold, to the point where nearly every topoisomerase molecule causes a double-stranded DNA cleavage event. The effect of oxolinic acid on the enzyme is very similar to its effect on another type I1 DNA topoisomerase, the Escherichia coli DNA gyrase. A filter-binding method was developed that allows efficient purification of to-poisomerase-cleaved DNA fragments by selecting for the covalent attachment of this DNA to the enzyme. Using this method, T4 topoisomerase recognition of mutant cytosine-containing T4 DNA was found to be relatively nonspecific, whereas quite specific recognition sites were observed on native T4 DNA, which contains glucosylated hydroxymethylcytosine residues. The increased specificity of native T4 DNA rec- ognition seems to be due entirely to the glucose modification. In contrast, E. coli DNA gyrase shows a high level of specificity for both the mutant cytosine-con-taining DNA and native T4 DNA, recognizing about five strong cleavage sites on both substrates. An un-expected feature of DNA recognition by the T4 topo- isomerase is that the addition of the cofactor ATP

initiation of T4 replication in uiuo, since topoisomerase-deficient mutants display a "DNA-delay'' phenotype, characterized by a delayed production of DNA (Mufti and Bernstein, 1974). McCarthy et al. (1976) determined that the rate of replication fork movement in cells infected with these mutants is the same as in cells infected with wild type bacteriophage, and therefore concluded that the process of replication fork initiation is defective. The residual T4 DNA replication seen in topoisomerase-deficient mutants has been reported to be totally dependent on host DNA gyrase, indicating that this host type I1 DNA topoisomerase can partially substitute for the T4 enzyme (McCarthy, 1979).
As a first step toward reconstructing the replication fork initiation process in uitro, we have characterized the interaction of the T4 topoisomerase with T4 DNA, searching for possible origin-specific DNA recognition by the topoisomerase. In the reaction catalyzed by type I1 topoisomerases, the enzyme creates a transient break in a double-stranded DNA molecule through which a second segment of unbroken double-stranded DNA is passed (for reviews, see Cozzarelli, 1980;Gellert, 1981). The presumptive reaction intermediate consists of broken DNA with a subunit of the topoisomerase covalently attached to each of the two 5' ends that are newly formed at the break. This intermediate can be detected a t a very low frequency after detergent treatment of T4 topoisomerase reactions (L. F. . Studies of a similar intermediate in the DNA gyrase reaction were greatly facilitated by the finding that nalidixic acid and oxolinic acid, two closely related antibacterial agents, trap this intermediate and enable nearly quantitative recovery of the enzyme in the covalent complex (Sugino et al., 1977;Gellert et al., 1977). Oxolinic acid also enhances double-strand DNA breakage by the T4 DNA topoisomerase. Using this drug treatment followed by a new method for isolation of covalent topoisomerase-DNA complexes, we show that both the T4 DNA topoisomerase and Escherichia coli DNA gyrase recognize native T4 DNA in a highly specific fashion.

EXPERIMENTAL PROCEDURES
modification of the method of L. F. , as described by Materials-Homogeneous T4 topoisomerase was prepared by a Kreuzer and Jongeneel (1983). Purified subunits A and B of E. coli DNA gyrase were kindly provided by Dr. N. Cozzarelli (University of California, Berkeley). Restriction enzymes were purchased from New England Biolabs, and T4 DNA polymerase was purified as described (Morris et al., 1979). T4 DNAs were prepared from purified phage particles by gently extracting with neutralized, water-saturated phenol and then ether, followed by extensive dialysis into 10 mM Tris-CI (pH 7.8), 1 mM Na3EDTA. HMC glu T4 DNA' was purified ' The abbreviations used are: HMC glu T4 DNA, native T4 DNA with the normal hydroxymethylated and glucosylated cytosine residues; C T4 DNA, T4 DNA (prepared from a multiple mutant) containing unmodified cytosine residues; HMC T4 DNA, T4 DNA containing nonglucosylated hydroxymethylcytosine residues; kb, kilobase pairs; SDS, sodium dodecyl sulfate; ATPrS, adenosine 5 ' -0 -(thiotriphosphate). from wild type phage particles; HMC T 4 DNA was purified from T4 agt-Pgt-phage grown in E. coli K803 (both kindly provided by Dr. J. Childs, Chalk River Nuclear Laboratory, Ontario, Canada); and C T4 DNA was purified from T4 56-42-denB alc phage particles grown in E. coli ED8689. The authenticity of these DNAs was verified by restriction enzyme analysis: all three were cleaved by TaqI, only C  T4 DNA and HMC T4 DNA were cleaved by XbaI, and only C T4 DNA was cleaved by XhoI.
Oxolinic acid was kindly provided by Martin Block of Warner-Lambert Co., [a-32P]dTTP was purchased from New England Nuclear and Amersham, agarose (medium electroendosmosis) was from SeaKem, glass fiber filters (GF/C and GF/A) were from Whatman, and human serum albumin was from Worthington Biochemicals.
Topoisomerase Cleavage Reactions-Except where indicated, all T 4 topoisomerase and E. coli DNA gyrase cleavage reactions (20 p1) contained 40 mM Tris-C1 (pH 7.8), 60 mM KCl, 10 mM MgCI,, 0.5 mM dithiothreitol, 0.5 mM Na3EDTA, 0.5 mM ATP, human serum albumin a t 30 pg/ml and oxolinic acid a t 500 pg/ml. The reactions were incubated for 15 min a t 30 "C, and then SDS was added to 0.2%. When necessary, the covalent attachment of the topoisomerase to DNA was eliminated by incubation for 60 min a t 37 "C with proteinase K (EM Biochemicals) a t 100 pg/ml. Gel Electrophoresis-DNA fragments were analyzed on horizontal agarose gels in TBE buffer (90 mM Tris base, 90 mM boric acid, 2.5 mM NaREDTA). Samples were loaded in final concentrations of 4% Ficoll and 1% SDS, with 0.1% bromphenol blue and xylene cyanole added as tracking dyes. After electrophoresis, agarose gels were dried onto filter paper, and autoradiography was carried out at -70 "C with Cronex Lightning Plus intensifying screens.
Spin Dialysis-A rapid method of gel filtration, called spin dialysis, is based on the original method of Neal and Florini (1973) which was modified by McGhee and von Hippel (1977). Spin dialysis columns were prepared by delivering 0.8 ml of a preequilibrated slurry (60% settled volume) of Sepharose CL-GB to a 1.5-ml Eppendorf tube with a hole (made with a 27-gauge needle) punched in the bottom. The gel matrix (but not the liquid) was retained in this tube by covering the hole with a small amount (30 pl) of siliconized glass beads (average diameter = 0.1 mm) before adding the slurry. The tube was placed into a second 1.5-mI Eppendorf tube (with a hole punched in the side to relieve pressure which develops during centrifugation), and the nested pair of tubes was spun for 2 min a t setting 80 in a Clay Adams table-top centrifuge. This spin removed the gel buffer, and the top tube, containing the packed gel beads, was then nested into an empty 1.5-ml Eppendorf tube. The DNA sample (60 pl) was delivered to the center of the packed column, and the nested tubes were centrifuged as above. The DNA was recovered from the bottom tube, free of low molecular weight materials tie. SDS, unincorporated triphosphates, phenol, salts, etc.). The DNA yield in this procedure was generally 50 to 75%.
Preparation of End-labeled T4 DNA Restriction Digests-The substrates used for Figs. 2, 3, and 6 were end-labeled by replacement synthesis (O'Farrell, 1981). After the appropriate restriction enzyme treatment, T4 DNA polymerase was added and exonucleolytic digestion allowed for 5 min a t 37 "C; resynthesis in the presence of [ C U -~~P ] dTTP and nonradioactive dCTP, dGTP, and dATP results in a labeled region of about 100 bases a t each 3' end. Labeling reactions were terminated by adding SDS to 0.2%, and the DNA samples were then purified by spin dialysis.
Filter-binding Method for Covalent Complexes-The filter-binding procedure for purifying covalent topoisomerase-DNA complexes is a modification of t,he glass fiber (GF/C) filter method of Coombs and Pearson (1978) and Thomas et al. (1979). Topoisomerase reactions were terminated (with induction of the covalent complex) by adding SDS to 0.276, as described above. The SDS was then removed by spin dialysis, and one-fifth final volume of 5X binding buffer was added to the eluate (1X binding buffer = 50 mM Tris-C1, pH 7.8 200 mM KC1; 10 mM MgCI2; 0.5 mM Na,EDTA). Covalent protein-DNA complexes were then collected on GF/C filters as follows. Two 7-mm diameter GF/C filters (one on top of the other) were placed onto a larger GF/A filter, and 150 p1 of binding buffer was delivered slowly to the top filter (the buffer soaks through the filter pair by capillary action). After this prewash, the filter pair was moved to a dry area of the GF/A filter, and the spin dialysis eluate was applied, followed immediately by 150 pl of binding buffer. The filter pair was then washed four more times with 150 pl of binding buffer, moving the filter pair to a dry area of the GF/A filter before each wash. The top GF/C filter disc was then eluted twice with 20 p1 of 10 mM Tris-C1, pH 7.8, and 0.1% SDS. The filter eluate was collected and treated with proteinase K a t 100 pg/ml for 30 min a t 37 "C, it was then subjected to spin dialysis before gel electrophoresis.
Filter-binding Method for Covalent and Noncoualent Complexes-The filter-binding procedure was also used for purifying noncovalent complexes of topoisomerase and DNA by delivering the topoisomerase reaction mix directly to the prewashed filter pair, without the SDS addition used in the detection of covalent complexes. The filter pair was washed five times with 150 p1 of binding buffer, and the eluate from the top filter was then analyzed as described above. This procedure detects both covalent and noncovalent topoisomerase-DNA complexes.

RESULTS
Oxolinic Acid Induces DNA Cleavage by the T4 Topoisorneruse-As discussed above, an intermediate in the reaction of type I1 DNA topoisomerases consists of broken DNA with covalently attached protein. In order to facilitate further analysis of the cleavage reaction catalyzed by the T4 topoisomerase, conditions were sought that increase the efficiency of cleavage (and presumably the lifetime of the reaction intermediate). It had previously been shown that nalidixic acid and its more potent analogue oxolinic acid are specific inhibitors of E. coli DNA gyrase and that their addition causes nearly quantitative cleavage of the DNA double helix by the enzyme (Sugino et al., 1977;Gellert et al., 1977). While we detect no effect of oxolinic acid on the T4 topoisomerase at a drug concentration that inhibits E. coli DNA gyrase (10 pg/ ml), the drug blocks the T4 topoisomerase at higher concentrations, with half-maximal inhibition a t roughly 250 pg/ml (L. F.  data not shown). In the presence of 500 pg/ml oxolinic acid, the T4 topoisomerase converts circular duplex 6x174 DNA into a series of linear fragments when SDS is added, as judged by gel electrophoresis. Quantitation of this cleavage by scanning densitometry shows that the amount of DNA cleavage obtained is about 100-fold greater than that caused by SDS addition alone (Fig. 1). Assuming that the native molecular weight of the topoisomerase is 260,000 (see Kreuzer and Jongeneel, 1983), an equimolar ratio of DNA and protein was reached when about 20 ng of enzyme was added in this experiment, and thus the data in duplex DNA, the indicated amount of topoisomerase, and either no oxolinic acid (0) or 500 pg/ml oxolinic acid ( 0 ) . After 10 min a t 30 "C, the reactions were terminated by adding SDS, and then treated with proteinase K as described under "Experimental Procedures." The reaction products were separated on a 0.8% agarose gel, and the disappearance of the linear duplex DNA was quantitated by scanning a negative of the ethidium bromide-stained gel, using a Zeineh soft laser densitometer. Both the efficient oxolinic acid-induced cleavage of duplex 6x174 DNA and the low level of cleavage seen in the absence of the drug require treatment with a detergent such as SDS to trap the DNA in its broken form, and a prior treatment with excess EDTA blocks DNA cleavage. In addition, in either the presence or absence of oxolinic acid, protein becomes covalently attached to the broken DNA as judged both by filter-binding experiments and comparison of the rate of electrophoretic migration of the DNA with and without proteinase K treatment (data not shown).

Filter-binding Method for Identification of Strong Topoisomerase Cleavage Sites on the T4
Genome-The analysis of T4 topoisomerase-binding sites on T 4 DNA is complicated by the fact that native T4 DNA contains glucosylated hydroxymethylcytosine residues, which makes it refractory to nearly all restriction enzymes. Moreover, the T4 genome is large (165 kb), and the chromosome sequence is circularly permuted. A third complication is that the T4 topoisomerase binds to a large number of sites, many with a low efficiency (see below).
We began our analysis by using radioactively labeled restriction enzyme digests of C T4 DNA, prepared from a special multiple mutant bacteriophage (O'Farrell et al., 1980). The restriction enzyme digest was treated with topoisomerase in the presence of oxolinic acid, and the reaction was terminated with SDS to create covalent complexes of topoisomerase and cleaved DNA. A strong cleavage site on the DNA, being located a unique distance from each of the two ends of a DNA restriction fragment, should produce two discretely sized fragments of DNA over a background of DNA fragments created at weaker cleavage sites. T o detect such strong sites, the topoisomerase-cleaved DNA was purified away from the substrate DNA restriction fragments by a modified glass fiber filter-binding technique that selects only DNA molecules that have covalently bound protein (see "Experimental Procedures"). with the DNA fragments obtained by the above filter-binding method either without or with T4 topoisomerase treatment (lanes 2 and 3, respectively). As shown by a comparison of lanes 2 and 3, only DNA fragments with bound topoisomerase stick to the filters. About 30 discrete radioactive bands were created by topoisomerase cleavage, in addition to a background smear of nonspecific cleavage products. Thus, T4 topoisomerase recognizes on the order of 15 high affinity cleavage sites on the C T4 DNA (30 divided by 2, since each cleavage generates two fragments), in addition to a much larger number of lower affinity sites. The addition of ATP to the enzyme reaction had essentially no effect on the pattern of fragments produced, although it did slightly increase the total frequency of cleavage (Fig. 2, lane  4 ) . Oxolinic acid could be added either at the start of the reaction (lane 4 ) or 10 s before terminating the reaction with SDS (lane 5 ) , without changing the cleavage products obtained. This provides some evidence that the oxolinic acid does not change the distribution of sites to which the enzyme is bound.
Topoisomerase Recognition of Native T4 DNA-Since the mutant T4 cytosine-containing phages are seriously disturbed with respect to various aspects of nucleic acid metabolism in vivo, it seemed imperative to compare the topoisomerase recognition of T4 cytosine-containing DNA with that of the native, glucosylated hydroxymethylcytosine-containing DNA. As mentioned above, native T4 DNA is, in general, refractory to restriction enzyme cleavage. However, after screening a large number of restriction enzymes, we discovered that TaqI is unusual in cleaving native T4 DNA efficiently.* The TaqI restriction digests of C and HMC glu T4 DNA are compared in lane 1 of Fig. 3, A and R, respectively. Many of the faster migrating bands on the gel contain several co-migrating restriction fragments, and we estimate that more than 100 Tag1 sites are present in T4 DNA (data not shown).
There are differences in the pattern of DNA fragments obtained by TaqI digestion of C and HMC glu T4 DNAs. These differences arise from at least three separate causes: 1) the C T 4 DNA contains a deletion removing the 162.3-165.5kb region of the T4 genome; 2) HMC glu T4 DNA has a few Tag1 sites that are protected from restriction enzyme cutting: and 3) the glucose moieties change the electrophoretic migration of the HMC glu T4 DNA fragments. Because of these differences in the pattern of fragments, it is not possible to compared to C T4 DNA. Because DNA fragments generated identify the same DNA fragment in the two digests with from C T4 DNA and HMC glu T4 DNA substrates are not certainty simply on the basis of electrophoretic migration. directly comparable with respect to their electrophoretic mi-Nevertheless, the topoisomerase cleavage sites on these two gration, it is not possible at this time to assess whether some DNAs can be compared both qualitatively and quantitatively. of the same strong topoisomerase cleavage sites are recognized (together with numerous weaker ones) were generated by the The addition of ATP increased the frequency of topoiso-T4 topoisomerase using the TaqI-digested C T 4 DNA submerase-induced cleavage of HMC glu T4 DNA roughly 5-fold strate (Fig. 3A, lanes 3 and 4 ) . At the highest enzyme level, a (Fig. 3B, lanes 6-8). A t the highest enzyme level, multiple smear of low molecular weight fragments was generated (lane cutting of the largest DNA cleavage fragments was observed, In the absence of ATP, several strong cleavage fragments on the two DNAs.

5),
indicating that many of the Ta9I restriction fragments were cleaved more than once by the topoisomerase. The addition of ATP had little or no effect on the particular fragments recovered, but it did increase the efficiency of cleavage by about 2-fold (lanes 6-8). These results are consistent with those obtained above using XhoI-digested C T4 DNA as substrate, and demonstrate that the T 4 topoisomerase recognizes a large number of cleavage sites on C T 4 DNA.
The results of the cleavage site assay with TaqI-digested HMC glu T4 DNA were quite different (Fig. 3 B ) . In the absence of ATP, even at the highest enzyme level (Fig. 3B,  lane 5), only about eight strong and eight weaker cleavage fragments were generated, and the pattern was not obscured by multiple cutting of the larger fragments. Thus, in the absence of ATP, the T4 topoisomerase shows a marked increase in specificity when recognizing HMC glu T4 DNA as but about 12 to 14 specific cleavage fragments were still evident (lune 8). Note that the strong cleavage fragments generated in the absence of ATP are also seen at the lower enzyme levels in the presence of ATP, and therefore, the same strong recognition sites were generally recognized in both cases. The stimulatory effect of ATP could be either a simple quantitative effect on the efficiency of cleavage, or alternatively, indicate that numerous weaker recognition sites will react only in the presence of the cofactor.
Since the activity of the topoisomerase in the strand-passing reaction is coupled to ATP hydrolysis (L. F. Liu, et ul., 1979), the ATP-induced increase in the efficiency of topoisomerase cleavage of HMC glu T4 DNA could involve some ATP hydrolysis-dependent reaction of the topoisomerase, such as localized supercoiling or movement of the enzyme along the DNA double helix. This hypothesis was tested by determining whether the nonhydrolyzable ATP analogue ATPyS would mimic the effect of ATP; this analogue is a potent inhibitor that blocks turnover of the enzyme (L. F. . As shown in Fig. 3B (lane 9 ) , ATPyS closely mimics ATP in its effect on topoisomerase-induced DNA cleavage. Thus, it seems likely that the ATP effect shown in Fig. 3B simply reflects a conformational change of the enzyme caused by ATP binding (see below).
The increased specificity with which the topoisomerase recognizes native T4 DNA could require either the hydroxymethylation of the cytosine residues or the glucosylation of these hydroxymethylcytosines. These two possibilities were distinguished by preparing "2P-labeled Tag1 restriction fragments of hydroxymethylated T4 DNA free of glucose residues, which was isolated from an ngt, @gt double mutant T 4 bacteriophage (Revel and Luria, 1970). The topoisomerase-induced cleavage patterns generated from the glucose-free DNA (Fig.  3C) were nearly identical with those seen with C T4 DNA (Fig. 3A). Moreover, the dramatic stimulation of cleavage seen with native T4 DNA in the presence of ATP (Fig. 3 B ) was not obtained with the glucose-free DNA. We therefore conclude that glucosylation causes the increase in the specificity of topoisomerase recognition of native T 4 DNA.
Since E. coli DNA gyrase has been reported to substitute partially for the T4 topoisomerase during DNA replication in uiuo (McCarthy, 1979), we compared its cleavage specificity with that of the T4 topoisomerase on T4 DNA substrates. With HMC glu T4 DNA, only about 10 major cleavage fragments were observed, indicating that gyrase recognizes approximately five strong cleavage sites on this substrate, and the background of nonspecific cleavage was lower than for the T 4 topoisomerase at comparable levels of DNA cleavage (compare Fig. 4, lane 3 with lane 1 ) . The pattern of cleavage fragments generated after gyrase treatment of C T 4 DNA also showed about 10 strong cleavage fragments and a low background of nonspecific cleavage fragments (data not shown). The addition of ATP to the gyrase reactions had essentially no effect on the pattern of cleavage fragments observed (Fig.  4, lane 4 ) . Some of the DNA gyrase-induced cleavage fragments co-migrated with the T4 topoisomerase-induced fragments, but our preliminary mapping analysis indicates that these cleavage fragments originate from different regions of the T4 genome.
One possible limitation of the filter-binding assay used here is that the topoisomerases might not bind to some possible recognition sites, due to their location close to the end of a DNA fragment (i.e. near a TaqI restriction site). This was tested by repeating the filter-binding experiment in such a way that the cleavage fragments were generated by topoisomerase on intact T 4 DNA, with the Tag1 digestion only performed after the topoisomerase reaction (Fig. 4, lanes 5-8). Except for relatively minor differences in intensities, the recognition sites for T 4 topoisomerase in the absence (lanes 1 and 5) or presence of ATP (lanes 2 and 6), and for DNA gyrase in the absence (lanes 3 and 7) or presence of ATP (lanes 4 and 8), were the same on intact and on TaqI-cleaved T4 DNA. Thus, the spectrum of cleavage fragments detected is not significantly affected by DNA substrate size.
The high level of specificity demonstrated for the T4 topoisomerase and the E. coli DNA gyrase on native T4 DNA could, in principle, be explained by either of the following situations. 1) The initial binding of the enzyme occurs a t strongly preferential sites on the DNA, and all of these sites undergo the strand-scission reaction that results in a covalent complex. 2) Only selected DNA binding sites are revealed by the covalent complex. For example, topoisomerase action, showed no detectable labeled products upon autoradiography of the gel (data not shown). In order to generate end-labeled and TagIcleaved reaction products, the filter-binding assay for covalent complexes was modified as follows. Immediately after the first spin dialysis to remove SDS, the reaction products were digested with TaqI at 65 "C for 30 min. The TqI digestion was terminated with SDS and the products were subjected to a second spin dialysis step to remove SDS. The reaction products were then fractionated by filter-binding and eluted from the filters with SDS as described under "Experimental Procedures." After another spin dialysis to remove SDS, the reaction products were end-labeled by the replacement synthesis method described in the text. These labeled fragments were treated with proteinase K in the presence of SDS to remove the covalently attached topoisomerase, and a last spin dialysis step was performed to remove unincorporated [a-"PldTTP. The reaction products were then analyzed by electrophoresis through a 1.2% agarose gel.
involving cleavage of the phosphodiester backbone with concomitant formation of the covalent protein-DNA complex, might occur at only a subset of the enzyme-binding sites.
The above two possibilities were distinguished by comparing the DNA fragments retained in two different filter-binding assays. For detecting both covalent and noncovalent complexes, 32P-labeled TaqI restriction fragments of native T 4 DNA were incubated with various levels of T4 topoisomerase in the presence of oxolinic acid, and half of each reaction was applied without SDS treatment to a GF/C filter; the bound DNA was then eluted with SDS and analyzed (Fig. 5A,  The assay mix was as described under "Experimental Procedures," except that ATP was omitted and the reaction volume was increased to 40 pl. The reactions were incubated for 20 min at 30 "C and then split into two equal aliquots. For each reaction, one aliquot was subjected to the filter-binding assay that does not require covalent DNA-protein binding ( A ) , while the other was subjected to the filterbinding assay for covalent complexes ( B ) . In both A and B, lane 1 shows the '*P-labeled DNA substrate, lane 2 is the filter eluate from a control reaction without enzyme, lanes 3 through 5 are the filter eluates of reactions with decreasing amounts of the T4 topoisomerase (1600, 160, and 16 ng, respectively), and lane 6 is the filter eluate of a reaction with 70 units of E. coli DNA gyrase. Because the amount of radioactivity in DNA in each of these reactions was 10-to 20-fold less than that used in the other cleavage experiments reported in this paper, only a faint pattern of cleavage fragments is observed in the covalent assays.
to at least one site on virtually every Tag1 restriction fragment, and no strong site specificity was seen even after dilution of the enzyme. Because most of the DNA fragments recovered co-migrated with the original TaqI restriction fragments, most of the enzyme-bound DNA was apparently not cleaved by the topoisomerase. In contrast, the usual pattern of specific topoisomerase-cleaved DNA fragments was observed in the assay that selects for only covalent complexes. We conclude that only a small subset of the initial binding sites can produce a covalent complex. Presumably, this reflects the fact that most binding sites on HMC glu T 4 DNA are unable to undergo the topoisomerase-induced, reversible strand breakage event under these conditions (see "Discussion" below).
A markedly different result is obtained when the same type of comparison is made for DNA gyrase binding to native T4 DNA. Here, essentially the same specific DNA fragments are recovered in the two DNA-binding assays (Fig. 5, A and B,  lane 6). Thus, the site specificity of DNA gyrase on native T 4 DNA results directly from the initial selection of binding sites, and there is no evidence for a class of binding sites that are unable to undergo the strand cleavage reaction.
As shown in Fig. 1, the addition of oxolinic acid to a reaction containing a DNA substrate with unmodified cytosine residues causes nearly every T 4 DNA topoisomerase molecule to cleave a duplex DNA molecule once SDS is added. In contrast, a FIG. 6. T4 DNA topoisomerase-induced cleavage fragments of TaqI-digested native T4 DNA in the absence of oxolinic acid. The substrate for these T4 topoisomerase cleavage reactions was 3zP end-labeled, TaqI-digested HMC glu T4 DNA, and each reaction also contained 0.2 pg of unlabeled, TaqI-digested HMC glu T4 DNA as carrier. The reaction products were separated by electrophoresis through a 1.2% agarose gel. Lane 1 shows the restriction digest substrate without fractionation by the filter-binding procedure, and lanes 2 through 4 show the GF/C filter eluates of the isolated DNA cleavage products from topoisomerase cleavage reactions performed in the absence of oxolinic acid, but otherwise as described under "Experimental Procedures." Lane 2 is the eluate of a control reaction without enzyme, lane 3, the eluate from a reaction containing 1600 ng of T4 topoisomerase without ATP, and lane 4, the eluate from a reaction containing the same amount of enzyme in the presence of 0.5 mM ATP. only a minority of the enzyme molecules cleave modified HMC glu T 4 DNA under the same conditions (Fig. 5). We find that oxolinic acid stimulates cleavage of HMC glu T 4 DNA by roughly 20-fold in the absence of ATP, and nearly 100-fold in the presence of the cofactor (data not shown). The patterns of specific DNA cleavage products obtained when SDS is added to reactions without (Fig. 6) or with oxolinic acid (Fig.  3 B ) are significantly different. (Some of the fragments recovered in the two experiments have the same electrophoretic mobility, but further analysis is necessary to determine if these fragments are identical.) Another difference is that there is no apparent effect of ATP on the cleavage observed in the absence of oxolinic acid (Fig. 6, lanes 3 and 4 ) . The cleavage sites observed with and without oxolinic acid apparently represent different classes of binding sites on native T 4 DNA (see "Discussion" below).

DISCUSSION
The T4 DNA topoisomerase and E. coli DNA gyrase share several properties, including the double strand passage mechanism of topoisomerization, a multisubunit composition, and an ATP hydrolysis requirement (L. F. L. F. Liu et al., 1980). The results presented here extend the similarities between the two enzymes. Oxolinic acid, a potent inhibitor of DNA gyrase, also blocks T 4 topoisomerase, albeit at higher drug concentrations. For both enzymes, the drug appears to interfere with resealing of the transient reaction intermediate consisting of a DNA helix broken in both strands with the enzyme covalently attached. In addition, both enzymes cleave a broad spectrum of sites with widely varying efficiencies in 4x174 duplex DNA, and if very weak cleavage sites are counted, both recognize roughly one cleavage site every 50 base pairs.4 A detailed analysis comparing sites in limited regions of the genome suggests that about one-fourth of the cleavage sites in 4x174 duplex DNA are recognized in common by the two enzymes, but the relative efficiencies of cleavage at commonly recognized sites are very different for the two enzyme^.^ The rules of DNA sequence recognition by type I1 topoisomerases are elusive (Morrison and Cozzarelli, 1979), but a detailed comparison of the sequences cleaved by these two related enzymes could help to define these rules.
The filter-binding assay introduced here uses SDS treatment to destroy all noncovalent protein interactions with DNA, followed by a rapid method of gel filtration called "spin dialysis" to remove both the SDS and the noncovalently bound protein. It provides an effective method for selectively purifying covalent protein-DNA complexes, and is so sensitive that covalent complexes of topoisomerase bound a t specific locations on exogenously added, native T4 DNA can be detected even if the source of topoisomerase is a crude extract of T4-infected cells.4 The pattern of cleavage fragments and the effect of ATP in these extracts are the same as described for the purified topoisomerase, suggesting that no elements that alter topoisomerase specificity are lost during topoisomerase purification. This assay should prove useful in studying the binding of any protein that interacts covalently with DNA.
The efficiency with which duplex DNA is cleaved by the T4 DNA topoisomerase is not only strongly enhanced by the inhibitor oxolinic acid, but also influenced by the presence of ATP and the modification of cytosine residues in the substrate DNA. With an unmodified substrate, such as duplex 4x174 DNA, oxolinic acid increases the efficiency of cleavage by about 100-fold (Fig. l ) , and there is no dramatic effect of ATP on DNA cleavage. In this case, oxolinic acid appears to very effectively trap the reaction intermediate consisting of broken duplex DNA covalently attached to the topoisomerase, since virtually every enzyme molecule cleaves a duplex DNA substrate molecule in the presence of the inhibitor. The cleavage of unmodified (cytosine-containing) T4 DNA by the T4 topoisomerase seems quite similar. However, the efficiency with which the T4 topoisomerase cleaves native HMC glu T4 DNA in the absence of ATP is greatly reduced. As judged by a comparison of the DNA fragments recovered in the noncovalent and the covalent binding assays (Fig. 5), only a subset of' the initial binding sites on native T4 DNA is cleaved by the topoisomerase in the presence of oxolinic acid. Interestingly, the addition of ATP dramatically increases the frequency of cleavage of this modified DNA substrate (Fig, 3 B ) . The fact that this stimulation is also caused by the nonhydrolyzable analogue ATPyS suggests that it is not caused either by ATP-induced movement of the enzyme to new cleavage sites or by an induction of local DNA supercoiling. Instead, it seems likely that the ATP stimulation reflects a conformational change of the enzyme induced by the binding of ATP or ATPyS. Analogously, an ATP-induced shift in DNA gyrase cleavage sites on ColEI DNA has been shown to depend on a conformational change induced by ATP binding (Morrison et al., 1980). K. N. Kreuzer, unpublished data. _ _ _ _ _~. _ _ _ _ For both DNA gyrase and the T4 topoisomerase, binding of ATP (or a nonhydrolyzable ATP analogue) apparently induces a conformational change that allows a single breakage-reunion reaction cycle, and the actual hydrolysis of ATP is required for reuse of the enzyme in a second reaction cycle (Sugino et al., 1978;L. F. Liu et al., 1979). One might therefore expect that both enzymes would require the binding of ATP to induce double strand cleavage. In this view, the considerable ATP dependence shown for the cleavage of native T4 DNA by the T4 topoisomerase is what one would naively expect in all cases. Why is there a tighter coupling to ATP when the T4 topoisomerase acts on its native substrate than when it acts on T4 DNA lacking glucosyl modifications (Fig.  3)? Could the topoisomerization reactions catalyzed by the T4 topoisomerase likewise be different in the presence of the modified cytosine residues? Previous studies revealed that the enzyme is unable to actively supercoil circular duplex DNA substrates containing unmodified cytosine residues Stetler et al., 1979), but it seems important to retest for possible DNA supercoiling using covalently closed duplex DNA circles constructed from native (HMC glu) T4 DNA.
The oxolinic acid-enhanced cleavage of native T4 DNA by the T4 DNA topoisomerase occurs at strongly preferred sites (Fig. 3B). In general, the same preferred sites are cleaved in the presence or absence of ATP, even though the efficiency of cleavage is increased by the presence of the cofactor. We cannot be certain that these strong cleavage sites represent the strongest binding sites of the enzyme on T4 DNA, since only a minority of the enzyme molecules generate a covalent DNA-protein complex (Fig. 5 ) . However, we are unable to detect any strong sites that do not result in cleavage, because no specific TuqI restriction fragments are recovered preferentially upon dilution of the enzyme in an experiment that detects noncovalent DNA binding (Fig. 5A).
The findings that ATP has no effect on the cleavage reaction and that the preferred sites of DNA cleavage are different in the absence of oxolinic acid suggest that the cleavage that occurs in the absence of oxolinic acid may occur at a special subset of topoisomerization sites where the reformation of the intact DNA double helix from the covalent reaction intermediate is unusually slow. In this case, the strong cleavage sites observed in the presence of oxolinic acid would be more indicative of the strong sites of T4 topoisomerase action. Further experiments are in progress to test this supposition directly.
The mechanism of initiation of T4 DNA replication in vivo has been the object of intensive study. Several groups have shown that T4 DNA replication normally begins at preferred regions of the T4 chromosome; these "replication origins" have been approximately localized by hybridizing newly synthesized DNA to defined fragments of cytosine-containing T4 DNA (Halpern et al., 1979;Mosig et al., 1981;King and Huang, 1982). Differences in the locations of the origins detected in different laboratories have recently been explained as being primarily due to physiological perturbations induced by the particular experimental protocols used by the various groups, and it now seems that T4 can utilize as many as five different replication origins depending on the conditions (Mosig, 1983).s Recently, it has also been shown that T4 replication fork initiation occurs in two separate modes (Luder and Mosig, 1982). The initial mode, which produces replication intermediates consisting of standard "replication bubbles," requires a direct involvement of the host RNA polymerase, while the later mode is independent of RNA polymerase but Site Specificity of T 4 Topoisomerase on T4 DNA dependent on phage-encoded genetic recombination enzymes. The RNA polymerase-dependent mode functions at an origin located roughly in the 15-20-kb region of the T4 chromosome, but it may also operate a t other origins. In addition, at least three of the origin regions correlate with locations of recombinational hot spots, and these hot spots may result in preferred initiation in these regions by the second mode described above.
Mosig's group has also obtained evidence for a role of the T4 DNA topoisomerase in both modes of replication fork initiation . In the experiments presented in this paper, we find that the T4 topoisomerase acts preferentially at a small number of sites on native T4 DNA. The preliminary results of genome-mapping experiments have been published . One strong oxolinic acid-induced topoisomerase cleavage site is located near the strongest early T4 promoter (Niggemann et al., 1981), in the vicinity of a replication origin that functions in the RNA polymerase-dependent mode. Another strong cleavage site is located near a major recombinational hot spot on the T4 chromosome, near or within gene 35. Recombination promoted at this hot spot is dependent on glucosylation of the participating DNAs (Levy and Goldberg, 1980), which is reminiscent of our finding that topoisomerase recognition of T 4 DNA is grossly altered by glucosylation of the DNA. Further experiments will be necessary to determine whether the topoisomerase site specificity studied in this report is involved in the selection and utilization of T4 replication origins by one or both of the two different modes of replication fork initiation.