Proc. Nati. Acad. Sci. USAVol. 78, No. 7, pp. 4165-4169, July 1981Biochemistry

Site-specific interaction of DNA gyrase with DNA(DNA sequence analysis/site-specific DNA breakage/oxolinic acid/DNase protection methods/Escherichia coli)

L. MARK FISHER, KIYoSHI MIZUUCHI, MARY H. O'DEA, HARUO OHMORI*, AND MARTIN GELLERT

Laboratory of Molecular Biology, National Institute of Arthritis, Metabolism and Digestive Diseases, NationalInstitutes of Health, Bethesda, Maryland 20205

Communicated by Gary Felsenfeld, March 31, 1981

ABSTRACT DNA gyrase, in the presence of the inhibitor ox-olinic acid, can induce double-strand DNA breakage at specificsites. The sequences at several sites have been determined. Inaddition, the structure ofcomplexes formed between DNA gyraseand restriction fragments containing an oxolinic acid-promotedcleavage site has been examined by DNase protection methods.DNA gyrase protects more than 120 base pairs of DNA againstpancreatic DNase in a region surrounding the cleavage site. Pro-tection is observed both in the presence and absence of oxolinicacid. Protected DNA flanking the cleavage site contains DNaseI-sensitive sites spaced on the average 10 or 11 base pairs apart.This result supports the view that, in the DNA gyrase-DNA com-plex, the DNA is largely wrapped on the outside of the enzyme.

Recent work has shown that the mode of action of DNA gyraseinvolves the passage of a duplex DNA segment through a tran-sient double-strand break in DNA. Evidence supporting thisconclusion has come from two kinds ofexperiments. First, DNAgyrase can unknot knotted duplex DNA and can form and re-solve DNA catenanes (1-3). Second, both the supercoiling andrelaxing activities of DNA gyrase alter the linking number ofthe substrate DNA in steps of two (2, 4). Several mechanisticmodels incorporating transient double-strand DNA breakagehave been proposed (2, 4, 5). In this paper, we describe ex-periments designed to examine the topography of complexesformed between DNA gyrase and DNA.

In the presence ofoxolinic acid, DNA gyrase forms a complexwith DNA that, on disruption with NaDodSO4, generates dou-ble-strand breaks in the DNA (6, 7). Because of the likely rel-evance ofthis reaction to transient DNA breakage by DNA gyr-ase, we have mapped and determined the sequences of severaloxolinic acid cleavage sites. Cleavage generates breaks on com-plementary strands staggered by 4 base pairs (bp) and resultsin covalent attachment of protein to each newly formed pro-truding 5'-phosphate end. These results complement previousstudies on DNA gyrase cleavage sites (8-11).

By using nuclease protection methods (12), we have inves-tigated the structure ofcomplexes formed between DNA gyraseand DNA restriction fragments containing potential oxolinicacid-cleavage sites. Earlier studies on the binding ofDNA gyr-ase to DNA established that the enzyme protects 140 bp ofDNA against micrococcal nuclease digestion (13, 14). More-over, in the DNA gyrase-DNA complex, the DNA appearedto be wrapped on the outside of the enzyme (15). We find that,both in the presence and in the absence of oxolinic acid, DNAgyrase protects >120 bp of DNA against pancreatic DNase. Aregion of =40 bp is most strongly protected. DNA flanking thisregion is less well protected and exhibits DNase I-sensitive sitesspaced 10 or 11 bp apart in a pattern reminiscent ofthat obtainedby DNase I digestion of nucleosomal DNA (16). These obser-

vations indicate that the DNA flanking the cleavage-site regionis wrapped on the outside of DNA gyrase.

Somewhat similar results have recently been obtained fromDNase protection experiments involving complexes formedwith M. luteus DNA gyrase (11).

MATERIALS AND METHODSMaterials. [y-32P]ATP (3000 Ci/mmol; 1 Ci = 3.7 X 1010

becquerels) was from New England Nuclear. ATP, [(3, y-imido]ATP, T4 polynucleotide kinase, and terminal nucleotidyltrans-ferase were from P-L Biochemicals. Calfintestine alkaline phos-phatase was from Boehringer Mannheim. Escherichia coli DNAgyrase A and B proteins were purified to >99% homogeneityas described (2). Pancreatic DNase (DNase I) from Worthingtonwas stored at 40C in 2.5 mM HC1 and diluted before use into20 mM Tris-HCl, pH 7.5/10 mM MgCJ2/0.5 mM CaCl2/0.1mM dithiothreitol. Restriction enzymes and E. coli exonucleaseIII (5.6 x 104 units per ml) were from New England BioLabs.The supercoiled DNA ofplasmids pBR322 and pVH51 was pre-pared by standard methods (2).DNA Sequence Analysis. DNA restriction fragments were

treated with alkaline phosphatase and then labeled at their 5'ends by using [y-32P]ATP and T4 polynucleotide kinase. Re-striction fragments specifically labeled at one 5' end were ob-tained by recutting with a second restriction enzyme. The de-sired fragments were isolated by acrylamide gel electrophoresis,and the sequence analyses were carried out by the method ofMaxam and Gilbert (17).The structures and sequences at sites of oxolinic acid-pro-

moted DNA cleavage by DNA gyrase were determined as fol-lows. Restriction fragments (100-250 bp long), known fromrough mapping to contain potential DNA gyrase cleavage sites,were labeled with 32P at both 5' ends and subjected to oxolinicacid-promoted gyrase cleavage followed by proteinase K treat-ment (for conditions, see ref. 6). The two resulting double-stranded fragments were separated and analyzed in each of twoways. (i) The lengths ofthe 32P-labeled strands were determinedby electrophoresis in a denaturing gel alongside the Maxam-Gilbert sequencing products obtained for each labeled strandof the original uncleaved restriction fragment. [Gyrase-cleavedfragments, having 3'-OH termini (see Results) run 0.5 nucleo-tide positions slower than the sequencing product of the samenucleotide length (18). This follows because the Maxam-Gilbertmethod generates DNA fragments that terminate in a 3' phos-phate. ] One must also note that the sequence analyses identifyDNA fragments from which the nucleotide of interest has beenremoved. [The same considerations apply in locating the se-quence position of 3'-OH termini generated by DNase I (12)or exonuclease III (see below).] (ii) The separated gyrase-

Abbreviation: bp, base pair(s).* Present address: Institute for Virus Research, Kyoto Univ., Kyoto,Japan.

The publication costs ofthis article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertise-ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact.

4165

4166 Biochemistry: Fisher et aL

cleaved fragments were themselves subjected. to the Maxam-Gilbert reactions and to denaturing gel electrophoresis-auto-radiography to establish which base was at the newly formed3'-OH end.DNase Protection. The method of Galas and Schmitz was

used (12). In this technique, aDNA restriction fragment labeledwith 32P at one 5' end is lightly digested with pancreatic DNasein the presence or absence of the DNA binding protein understudy.: The resulting set of labeled DNA fragments is separatedaccording to length. DNA sequences protected from nucleaseattack are revealed as missing bands in the ladder ofMDNA frag-ments resolved by the gel.

For our experiments, the assay;buffer contained (in 48 ,ul)46 mM Tris*HCI, pH' 7.5/6.9 mM MgCl2/75 mM KCV0.38mM CaCI2/6mM dithiothreitol/0. 19mM Na3EDTA/4. 1% (wt/vol) glycerol, bovine serum albumin at 38 Ag/ml, and -0.05,ug of [32P]DNA. Where included, an 8- to 10-fold molar excessof DNA gyrase over DNA was used (i.e., 500 units of gyrase Bprotein and an excess of gyrase A protein). Oxolinic acid whenadded was present at 73 Ag/ml. Samples were incubated at250C for 75 min. Pancreatic DNase (2 A.l at 5 Aug/ml) was added,and the solutions were incubated at 25°C for 1 min. Reactionwas terminated by the addition of 50 1.d of stop solution [1.2 Mammonium acetate/0. 1 M Na3EDTA containing sonicated calfthymus DNA at 80 ,ug/ml]. The samples were heated to 75°Cfor 10 min, and then the DNA was ethanol precipitated andsubjected to electrophoresis in a 10% polyacrylamide/urea gel.

Oxolinic acid-induced DNA cleavage by DNA gyrase underthe protection conditions was done by adding 5% NaDodSO4(4 yl) and proteinase K (8 ,ul at 0.2 mg/ml) instead of DNaseI. Samples were incubated at 37°C for 35 min, and then theDNA was isolated as described above.

Conditions for the protection ofDNA against exonuclease IIIwere the same as those used with DNase I (see above). Fifty-six units of exonuclease III was added to each sample, and in-cubation was at 25°C. Reaction was terminated at 1 min or 3 minby the addition of DNase I stop solution. DNA isolation anddenaturing polyacrylamide gel electrophoresis were as de-scribed above.

RESULTSSite-Specific DNA Cleavage by DNA Gyrase. As a first step

in determining the structures and sequences of sites of oxolinicacid-mediated cleavage ofDNA by DNA gyrase, we ascertainedthe nature of the DNA termini produced in the reaction. Wefound that labeling of ends was possible by using terminal nu-cleotidyltransferase but not by using polynucleotide kinase (19).This suggested the presence offree 3'-OH ends but blocked 5'-phosphate termini and parallels previous findings (9). Recently,it has been demonstrated that a protomer of the gyrase A sub-unit becomes covalently linked to each 5'-phosphate terminusgenerated-by DNA gyrase cleavage (20, 21).

The Maxam-Gilbert sequence analysis method (17) was usedto determine the structures and sequences of several preferredgyrase cleavage sites. We chose to examine sites in plasmidpBR322, whose entire sequence is known (22), and in the ColE 1derivative pVH51, for which a large extent of the sequence hasbeen determined'(ref. 18; unpublished results). The sites stud-ied include one in the tetracycline-resistance region ofpBR322,one near the origin of replication of pVH51, and three in theAlu I D fragment of pVH51. Two of the three sites in the AluI D fragment were spaced only three nucleotides from eachother but were readily distinguished by their different cleavageefficiencies. The sequences around the cleavage sites are shownin Fig. 1. For all the sites, complementary DNA strands werebroken with a 4-bp stagger, yielding protruding 5' ends.

pVH51-HadII C

pVH51Alul 0(1)

pVH51-Alul 0(2)

pBR322*HpaII 15

f5' -GTATCTTTATASTCCTrTOIGBTT

CATAGAAATATCASSACASGCCAA-5'

5' -TACGBGCATGATGTTACTGGGGATATGCCCGTACTACAATBACCCCTA-5'

5' -ATCTTBGAGGGAGGATACA&GCGAACCTAGAACCTCCCTCCTATGTCCGCTTGG-5'

5' -GAGGCTGGATGGCCTTCCCCATTACTCCGACCTACCGGAAGGGGTAAT-'5'

4

FIG. 1. Sites of DNA cleavage induced byDNA gyrase in the pres-ence of oxolinic acid. The sites are contained within the restrictionfragments shown. The actual restriction fragments used in the se-quence determinations and the site locations (i.e., the nucleotide po-sition halfway between the two staggered breaks) were as follows. AnHpa fl fragment of 206 bp is contained within the pVH51 Hae II Cregion. The cleavage site is centered at position (-77) relative to theorigin of replication (18). The sequences ofpVH51 and pNT1 are iden-tical in this segment (unpublished results). Two fragments from thepVH51-Alu ID region were examined: a 105-bpHpa H-Tac I fragment,designated Alu I D(1), which extends beyond the left end [in the re-striction map of Tomizawa et al. (18)] of Alu I D by 50 bp, and a 91-bpTac I fragment [Alu I D(2)], located immediately to the right. Thecleavage site in the first fragment is at +23 from the left end of theAlu I D fragment; the two sites in the second fragment are at +86 and+89, with stronger cleavage at +86. The sequence of this region hasbeen determined (unpublished results). In.pBR322, the 90bp Hpa IIfragment 15 contains a cleavage site at position 991 in the plasmidsequence (22).

From the sequence data (Fig. 1), it is seen that, for threesites, cleavage on one strand occurs within the dinucleotideGpG, but this feature is not found universally. Again, althoughseveral sites show adjacent clusters of guanosines, other sitesdo not (see also, ref. 9). Inspection of several sites in pBR322and pVH51 exhibiting less efficient cleavage than observed forsites shown in Fig. 1 (see, e.g., Fig. 3), also failed to reveal anobvious common sequence element. In summary, we find noreadily apparent sequence rule determining sites ofDNA cleav-age by DNA gyrase.

Binding ofDNA Gyrase atDNA Cleavage Sites. The bindingspecificity of DNA gyrase to restriction fragments containingpotential cleavage sites was investigated by DNase I protectionmethods (12). Two different restriction fragments were used:a 203-bp Ava II/Alu I fragment from plasmid pBR322 DNA[spanning nucleotide positions 886-1089 (22)], containing thesite in pBR322 centered 98 bp from the Alu I end (Fig. 1), anda 200-bp Hae II/Hae III fragment from pVH51, in which thecleavage-site sequence is centered 92 bp from the Hae III end.This Hae II/Hae III 8 fragment (18) contained the cleavage-sitesequence located near the origin of replication of pVH51 (seelegend to Fig. 1). The restriction fragments were labeled spe-cifically at one or the other 5' end and each was subjected toprotection analysis. In the presence of oxolinic acid, the com-plex of DNA gyrase competent to give DNA cleavage is formedslowly (6). To maximize the formation of this complex and tofacilitate useful comparison of DNA protection afforded byDNA gyrase in the presence and absence of oxolinic acid, theenzyme and DNA were subjected to an initial incubation priorto the addition of the nuclease. The protection patterns areshown in Fig. 2. For both cleavage sites, DNA-gyrase protectedboth DNA strands in an extensive region encompassing thecleavage site sequence. Protection of a region of DNA aroundeach site occurred whether oxolinic acid was present in the in-cubation mixture (d lanes) or absent (e lanes). No protection was

Proc. Nad- Acad. Sci. USA 78 (1981)

Proc. Nati Acad. Sci. USA 78 (1981) 4167

Aa bedef

* :;

* 2

Babcdef

......-

.. -* d;|s

ca bedef

_-

0~~~~~~~~~~4f

FIG. 2. DNase I protection patterns of DNA gyrDNA restriction fragments containing potential gyrase

(A and B) Protection of the 203-bp Ava H/Alu I fragm[spanning nucleotide positions 886-1089 (22), with that position 991], 32P labeled at the Ava I 5' end (A) or t(B). (C andD) Protection of the 200-bpHae 1/Hae 8pVH51 (18), 32p labeled at the Hae.II 5' end (C) or the(D). Lanes a and b inA and B are the Maxam-Gilbertactions for guanine and cytosine, and lanes a and b in (sequencing reactions for (G + A) and (C + T}. In all fouc-f show the results of an experiment on one end-labement; DNA samples were treated with pancreatic DI}sence (c lanes) or in the presence (d and e lanes) of DN(d and f lanes) or without (e lanes) oxolinic acid; f lanesincubated with DNA gyrase under the protection-metand then treated with NaDodSO4 and proteinase K tcleavage. Lanes c-e contained equal amounts of radicples were subjected to electrophoresis in a 10% polyacdenaturing gel (81 cm x 33 cm x 0.5mm). Arrows indiof the cleavage site, located halfwaybetween the staggerthe filled circles are at intervals of 10 bases from this

observed when either the gyrase A or B subunit a](results not shown).

Structure of Complexes Formed Between DIGyrase. The protection patterns shown in Fig. 2tural information about the enzyme-DNA comp]the details of DNA protection were somewhat diftwo DNA binding sites studied (see below), instretch of at least 120 bp of DNA was protected-40 bp surrounding the cleavage-site sequence isprotected. The DNA on either side of this region iprotected, and there are bands corresponding thanced sensitivity to DNase I. These features areent in Fig. 3, which summarizes the data in Fig. 2,the locations of DNase I-sensitive sites. These are11 bp apart, and sites on complementary strandserage staggered by 2 bp.The characteristics of DNA protection confer

gyrase are somewhat different for the two binding

D (Fig. 3). For the pBR322 site, protection is roughly symmetri-cal extending at least 60 bp on either side of the cleavage site.For the pVH51 site, however, protection extends at least 90 bp

____t___ on one side of the site while stopping 35 bp on the other side.Asymmetric binding of DNA gyrase at the pVH51 site is alsoobserved in experiments involving E. coli exonuclease III. This

- enzyme processively degrades double-strand DNA in the 3'* -P 5' direction starting at the 3' ends. The Hae II/Hae III-8

fragment ofpVH51 labeled at one or the other 5'-phosphate endwas treated with exonuclease III in the presence and absence

* ofDNA gyrase. In the presence ofbound DNA gyrase, digestion. by the exonuclease was arrested at locations 15 and 18 bp on

one side of the cleavage site and at 90 bp on the opposite side.The dissimilarity in protection by gyrase at the two sites is alsohighlighted by the patterns of DNase I-sensitive sites. It is ev-ident that the number and location of DNase I-sensitive sitesis rather different for the two DNA fragments.

Protection by DNA gyrase at a particular- site was differentin the presence and absence of oxolinic acid (Figs. 2 and 3).Inclusion of the drug resulted in protection of a more extensiveregion ofDNA (see, e.g., Fig. 3B). Moreover, DNA sequencesimmediately adjacent to the cleavage-site sequence became lessaccessible to the nuclease. The distribution of DNase I-sensitivesites is also different in the presence and absence of the drug(Fig. 3). The strong band at the cleavage site seen only whenoxolinic acid was included (Fig. 2, d and e lanes) deserves com-ment. Cleavage at the same position was observed when theDNase I digestion was omitted (data not shown). Therefore, this

ase binding to band presumably arises by adventitious DNA cleavage inducedcleavage sites. by DNA gyrase during the incubation and isolation ofthe DNA.tent of pBR322 Nucleotide Effects on DNA Cleavage and DNA Protectionte cleavage site by DNA Gyrase. The effects of ATP and the nonhydrolyzable;he Alu I5' end analogue [,f,y-imido]ATP on oxolinic acid cleavage were in-'frgetfoe Hae 115' end vestigated for the site in the Ava II/Alu I fragment of pBR322sequencing re- and that in the pVH51 8 fragment. The inclusion of ATP had'andD are the little effect on the efficiency or position of cleavage for their panels, lanes pBR322 site, although weaker double-strand cleavage was pro-Bled DNA frag- moted at a secondary site (Fig. 4, lanes i and j). Results with'ase in the ab- [(,3y-imido]ATP were similar (data not shown). However, for

3, samples were the pVH51 8 fragment, the addition of ATP or [,p, y-imido]ATPhod conditions shifted the major cleavage site by 1 bp, altering the intensitySo induce DNA of cleavage from >95% at site 1 to =60% at site 2 (Fig. 3B).Dactivity. Sam- Again, the addition of nucleotide induced weaker cleavage atrylamide/urea secondary sites. The diverse effects of ATP on cleavage haveicate the center previously been reported for several sites in ColEl, 4X174, andredbeas and ..

position. simian virus 40 DNAs (10).Nucleotide effects on the protection of DNA were studied

for the gyrase binding site in the Ava II/Alu I fragment fromlone was used pBR322 (Fig. 3A). DNA gyrase, in the presence or absence of

oxolinic acid, gave essentially the same DNase I protection pat-NA and DNA tern irrespective of whether ATP or [,8,y-imido]ATP was in-contain struc- cluded (Fig. 4).lex. Although DISCUSSIONfferent for thei each case a An important question concerns the nature of the recognition1. A region of element that directs DNA gyrase to bind to a particular DNAmost strongly sequence e.g., at DNA cleavage sites. From Fig. 1, it is ap-isless strongly parent that there is no unique nucleotide sequence at or im-to sites of en- mediately adjacent to the cleavage site that is common to all sitesmade appar- studied here. Thus, DNA cleavage by DNA gyrase differs fromin particular, that produced by type II DNA restriction enzymes (23), whosespaced 10 or specificity is generally determined by sequences at or very near

are on the av- the restriction site. In previous work, it has been suggested thatone of the two single-strand breaks generated by cleavage oc-

rred by DNA curs predominantly in the sequence TpG (9). This generaliza-g sites studied tion is not supported by our results for strong sites in the plas-

Biochemistry: Fisher et aL

-

4168 Biochemistry: Fisher et aL

A b

II I 11 lill1. 11.1

II

KST(6CGCCA(AAA(JTsTUGCAt(AGGCE~ttAT(aG(CGttATGG(GGC(GA(GCG(TGGGTAGTTTG(TGGTUGTTCG(6ACG6AGT6ttAtsTttT6ATG(T(TCGCt~TTCtt6Ct6GAlTCGGATGl(C(GETT6(AGGtCATGCTGTtt(CtTt( Att^CtsaGGtGIGGTGG TCUAAAG((l(TCT6TtCGTCC6AATAGtCGITAt( 6tt66tTt(K16At(t(GATGC (G(afAtt GCAttTGCG(lt(tGt(TACCttvAAtst6TAATA(TA6AG6AIGAGCGA' thC TIAG~tCTA(GGZAAC6lT(CetA(GA(A66tTCt6T(tAltiT(TAT6(TA6T(CCT6TC.B ~ ~ ~ I II I B lia Il ll l 1

6GGTG6A6t6(TT(Ct(Tt(6ITTGTGftCATA66tCiTT(Gt(GTC(tACCTsTrGet(T(Ttr rsstG GrT((C(TTf ~6CWA(CAUUMATrAr66A(AAG(( T66AAtr G(6ATAAAtIT(A6rAtKtGT((((1t6((T(66A7TACTTTTrTrl66taAT~u2 # 11 11 I111 AII I III III 11 r °a I 11 11 III 11 11

11b

'2 11 - II I

11 111FIG. 3. Summary of DNA gyrase protection against DNase I digestion of the Ava ll/Alu I DNA restriction fragment of pBR322 (A) and the

Hae Hl/Hae m a fragment of pVH51 (B). Filled triangles denote the major sites of oxolinic acid-mediated DNA cleavage by DNA gyrase. Opentriangles indicate sites of weaker cleavage (all <5% cleavage efficiency of the major site). The locations of DNase I-sensitive sites within the DNA-DNA gyrase complex observed in the absence (a) and presence (b) of oxolinic acid are shown by vertical lines. In each case, the line length signifieswhether sensitivity of the site to DNase I is enhanced (longer line) relative to the control DNA alone or the same (shorter line). Several prominentDNase I-sensitive sites in Fig. 2 are not indicated in Fig. 3 because they are cleaved to a lesser extent than in the control withoutDNA gyrase. Linelengths (both long and short) are shown larger in a than in b to focus attention on protection observed in the absence of oxolinic acid. Distinct limitsfor the protected DNA region could be readily discerned only in B and are shown for the absence and presence of oxolinic acid by filled and opensquares, respectively. Arrows in B-show positions at which exonuclease m digestion of the DNA was arrested.

mids pBR322 and pVH51, where only one out offive examplesshows this feature (Fig. 1). Scrutiny of DNA sequences ex-tending.on either side of the cleavage sites also failed to revealany obvious sequence rule. Further analysis of DNA cleavageby DNA gyrase will be necessary to establish the determinantgoverning the binding of the enzyme to DNA.The mechanism ofDNA gyrase involves the passage of a du-

plex DNA segment through a transient double-strand break inDNA (2, 4). The enzyme is thought to be part ofa "gate" throughwhich the translocated DNA strand is passed. The intimatedetails of this process are at present poorly understood. Valu-able mechanistic information can be provided by DNase I pro-tection experiments as these reveal details of-the time averagetopography of complexes formed between DNA gyrase andDNA. For the two binding sites examined here, DNA gyraseprotects more than 120 bp ofDNA against digestion by DNaseI (Figs. 2 and 3). The following observations suggest that thisprotection arises from the binding of only one DNA gyrasemolecule to each restriction fragment. First, it has been shownthat a single DNA gyrase A2B2 tetramer protects =w140 bp ofDNA against digestion by micrococcal nuclease, a length similarto that protected against DNase I (14). Second, when either ofthe DNA-DNA gyrase complexes studied here was subjectedto oxolinic acid-induced DNA cleavage, the DNA was cleavedspecifically and essentially completely at only one site on eachstrand (Fig. 2, f lanes). Protection by DNA gyrase results fromsequence-specific rather than nonspecific binding of the en-zyme to'DNA. Nonspecific binding with DNA gyrase protecting>120 bp of DNA against DNase I would predict symmetricalprotection of the DNA restriction fragment with the centralportion of the DNA most strongly protected. However, theprotection observed for the Hae II/Hae III 8 fragment (Figs.2 C and D and 3) is clearly asymmetric rather than symmetricwith respect to the ends of the DNA. Furthermore, we havestudied the protection of a 312-bp Hae II/Alu I restriction frag-ment of pBR322 containing the 203-bp Ava II/Alu I fragment

shown in Fig. 3A. In the 312-bp fragment, the cleavage site isasymmetrically located 98 bp from the Alu I end. Nevertheless,the DNA protectecbby DNA gyrase was localized to the regionimmediately surrounding the cleavage site (data not shown).Thus, these findings suggest that the protection patterns ob-served are those ofspecific 1:1 complexes formed between DNAgyrase and the DNA binding site.

Within the DNA segment protected by DNA gyrase, a regionof 40 bp is most strongly protected. This region contains thesequence at which oxolinic acid-mediated gyrase cleavage canbe induced. If the view is adopted that sites of oxolinic acid-promoted gyrase cleavage are sites of transient double-strandDNA breakage during catalysis, then the most strongly pro-tected region ofDNA presumably forms part ofthe gate throughwhich the other duplex DNA segment is passed. The relativeinaccessibility ofthis portion ofthe DNA.in the enzyme complexmay arise because the DNA is recessed into a site on the enzymeat which transient DNA strand scission can take place. Thisbinding locus quite likely involves the two gyrase A subunits;a protomer of the gyrase A subunit becomes covalently linkedto each 5'-phosphate end of the double-strand break generatedin the gyrase cleavage reaction (20, 21).The protected DNA flanking the cleavage sequence appears

to be wrapped on the outside of the enzyme in the DNA-DNAgyrase complex. This follows from the observation ofpancreaticDNase digestion at. sites separated by' 10 or 11 bp. A similarpattern of DNase I-sensitive sites has been observed for nu-cleosomal DNA (16) and for DNA adsorbed to calcium phos-phate precipitates (13). This feature is generally thought to bediagnostic of DNA adsorbed .to a surface. The interpretation isconsistent with the observation that DNA gyrase in the absenceof ATP does induce a positive superhelical wrapping of DNAon the enzyme (15). The coiling of DNA on the enzyme isthought to be the controlling factor that ensures that a nega-tively supercoiled product is formed in the supercoiling reaction(2, 5).

'Proc. Nad Acad. Sci. USA 78 (1981)

Ar

Proc. Natd Acad. Sci. USA 78 (1981) 4169

abcdefghi j...... wF- ...... . ''bl~~~~~~~~~~~~~~~~~~~~im!ago"*iir

a

*S,

.0 .

,.* ii*!11

!is*l!;i:

*~ 5.

*1D- - i~~

FIG. 4. Effect of ATP on DNase I protection and DNA cleavage byDNA gyrase. The 203-bp Ava ll/Alu I restriction fragment frompBR322, 32p labeled at theAlu I 5' end, was subjected to DNase I diges-tion (lanes a-c and f-h) or oxolinic acid-induced DNA cleavage (lanesi and j) in the absence (lanes a and h) and presence (lanes b, c, f, g, i,and j) of DNA gyrase. Lanes: b, g, and j, 0.5 mM ATP included; f-j,oxolinc acid included; d and e, C + T and G + A sequencing reactions,respectively. Thin layer chromatography on PEI-cellulose showed that<10% hydrolysis of ATP occurred during the incubation. Gel electro-phoresis/autoradiography was done as described in the legend to Fig.2 using a 40 cm X 33 cm x 0.5 mm gel. Filled circles and the arroware as defined in the legend to Fig. 2; other symbols are defined in thelegend to Fig. 3.

DNA strand passage catalyzed by DNA gyrase most likelyinvolves conformational transitions within the enzyme-DNAcomplex. Detection of these changes is essential to understand-ing the mechanism ofthe enzyme. Ifa conformational transitionalters the accessibility ofthe DNA to DNase I, then in principleit can be detected by a change in the DNase I protection pat-tern. It is interesting that the binding of oxolinic acid to DNAgyrase induces a marked change in both the degree and extentof protection conferred by the enzyme and results in a redis-tribution of DNase I-sensitive sites (Fig. 3 A and B). From thepoint of view of DNA supercoiling by DNA gyrase, the effectsofATP and the nonhydrolyzable analogue, [.3, y-imido]ATP, areespecially relevant. The binding of either of these nucleotidesto a DNA gyrase complex formed with the Ava II/Alu I DNAfragment of pBR322 does not result in any substantial change

in the DNase footprinting pattern (Fig. 4). Protection of bulkPM2 DNA by M. luteus DNA gyrase was also unaltered by theinclusion ofATP (13). These results support the interpretationthat the DNA is largely on the outside of the enzyme-DNAcomplex and not greatly affected by conformational changes in-duced by nucleotide binding.The experiments presented here reveal several structural

aspects of the DNA-DNA gyrase complex and how this is af-fected by the binding of various ligands. The results are com-patible with recently proposed models for DNA gyrase (2, 4,5). Further investigation of the interaction ofDNA gyrase withDNA during catalysis will be necessary to delineate the topolog-ical pathway of the gyrase reaction and to determine whetherthe oxolinic acid cleavage sites are catalytically productive DNAbinding sites for the enzyme.Note Added in Proof. Another study describing protection ofDNA byE. coli DNA gyrase has recently appeared (24).We thank J. Tomizawa for help with early experiments and for val-

uable criticism of the manuscript. We are also grateful to J. C. Wangfor communicating unpublished results. L. M.F. was supported by apostdoctoral fellowship from the Damon Runyon-Walter WinchellCancer Fund.

1. Liu, L. F., Liu, C.-C. & Alberts, B. M. (1980) Cell 19, 697-708.2. Mizuuchi, K., Fisher, L. M., O'Dea, M. H. & Gellert, M. (1980)

Proc. Nati Acad. Sci. USA 77, 1847-1851.3. Kreuzer, K. N. & Cozzarelli, N. R. (1980) Cell 20, 245-254.4. Brown, P. 0. & Cozzarelli, N. R. (1979) Science 206, 1081-1083.5. Wang, J. C., Gumport, R. I., Javaherian, K., Kirkegaard, K.,

Klevan, L., Kotewicz, M. L. & Tse, Y.-C. (1980) MechanisticStudies of DNA Replication and Genetic Recombination, eds.Alberts, B. M. & Fox, C. F. (Academic, New York), pp. 769-784.

6. Gellert, M., Mizuuchi, K., O'Dea, M. H., Itoh, T. & Tomizawa,J. (1977) Proc. Natl. Acad. Sci. USA 74, 4772-4776.

7. Sugino, A., Peebles, C. L., Kreuzer, K. N. & Cozzarelli, N. R.(1977) Proc. Natl Acad. Sci. USA 74, 4767-4771.

8. Gellert, M., Mizuuchi, K., O'Dea, M. H., Ohmori, H. & Tom-izawa, J. (1978) Cold Spring Harbor Symp. Quant. Biol. 43, 35-40.

9. Morrison, A. & Cozzarelli, N. R. (1979) Cell 17, 175-184.10. Morrison, A., Higgins, N. P. & Cozzarelli, N. R. (1980) J. Biol

Chem. 255, 2211-2219.11. Kirkegaard, K. & Wang, J. C. (1981) Cell 23, 721-729.12. Galas, D. J. & Schmitz, A. (1978) Nucleic Acids Res. 5, 3157-

3170.13. Liu, L. F. & Wang, J. C. (1978) Cell 15, 979-984.14. Klevan, L. & Wang, J. C. (1980) Biochemistry 19, 5229-5234.15. Liu, L. F. & Wang, J. C. (1978) Proc. Natl Acad. Sci. USA 75,

2098-2102.16. McGhee, J. D. & Felsenfeld, G. (1980) Annu. Rev. Biochern. 49,

1115-1156.17. Maxam, A. M. & Gilbert, W. (1977) Proc. Natl Acad. Sci. USA

74, 560-564.18. Tomizawa, J., Ohmori, H. & Bird, R. E. (1977) Proc. Natl. Acad.

Sci. USA 74, 1865-1869.19. Gellert, M., Fisher, L. M., Ohmori, H., O'Dea, M. H. & Mi-

zuuchi, K. (1980) Cold Spring Harbor Symp. Quant. BioL, inpress.

20. Tse, Y.-C., Kirkegaard, K. & Wang, J. C. (1980)J. Biol Chem.255, 5560-5565.

21. Sugino, A., Higgins, N. P. & Cozzarelli, N. R. (1980) NucleicAcids Res. 8, 3865-3874.

22. Sutcliffe, J. G. (1979) Cold Spring Harbor Symp. Quant. Biol 43,77-90.

23. Modrich, P. (1979) Q. Rev. Biophys. 12, 315-369.24. Morrison, A. & Cozzarelli, N. R. (1981) Proc. Natl Acad. Sci. USA

78, 1416-1420.

Biochemistry: Fisher et aL

4