Proc. Natl. Acad. Sci. USAVol. 89, pp. 10603-10607, November 1992Biochemistry

Dynamics of DNA supercoiling by transcription in Escherichia coli(DNA gyrase/topoisomerase I/topology/gene expression)

DAVID N. COOK*, DZWOKAI MA, NING G. PON, AND JOHN E. HEARSTMelvin Calvin Laboratory, Department of Chemistry, University of California at Berkeley, and Division of Structural Biology, Lawrence Berkeley Laboratory,Berkeley, CA 94720

Communicated by I. Tinoco, Jr., July 20, 1992

ABSTRACT The relative rotation between RNA polymer-ase and DNA during transcription elongation can lead to super-coiling of the DNA template. However, the variables thatinfluence the efficiency of supercoiling by RNA polymerase invivo are poorly understood, despite the importance of supercoil-ing for DNA metabolism. We describe a model system tomeasure the rate ofsupercoiling by transcription and to estimatethe rates of topoisomerase turnover in Eseherichia coil. Tran-scription in a strain lacking topoisomerase I can lead to optimalsupercoiling, wherein nearly one positive and one negativesuperturn are produced for each 10.4 base pairs transcribed.This rapid efficient supercoiling is observed during transcriptionof membrane-associated gene products, encoded by tet (the genefor tetracycline resistance) and phoA (the gene for E. coilalkaline phosphatase), when the genes are oppositely oriented.Replacement of tetby cat, the gene from Tn9 encoding resistanceto chloramphenicol, whose gene product is soluble in the cytosol,reduces the efficiency of supercoiling by RNA polymerase. In awild-type topoisomerase background, both gyrase and topo-isomerase I are kinetically competent to relieve superturnsproduced by transcription. These results suggest that the level ofDNA supercoiling in vivo is probably determined by topoisom-erase activity, not by transcription.

Transcription can lead to localized supercoiling of DNAbecause the topology of the elongation complex requires arelative rotation between RNA polymerase and DNA (1, 2).Experimental evidence for this model was originally based ontwo observations concerning the topology ofplasmid pBR322in Escherichia coli: (i) inhibition ofDNA gyrase results in theformation of positively supercoiled plasmids (3); and (ii)transcription of tet in AtopA strains leads to the accumulationof a heterogeneous population ofpBR322 topoisomers, someof which are hypernegatively supercoiled (4). These seminalobservations led Liu and Wang (2) to propose that therequired relative rotation ofRNA polymerase about DNA (1)generates positive supercoils downstream from and negativesupercoils upstream from the elongating polymerase. Manystudies have verified the essential features of this model bothin vitro and in bacteria and yeast (5-10). An importantconclusion from these studies is that topoisomerases functionas swivels to relieve torsional stress during transcription (11).

Since transcription is a ubiquitous process in cells, afundamental biological question is the extent to which tran-scription determines the level ofDNA supercoiling in vivo (5,12, 13). The mechanistic factors which influence the effi-ciency and extent of localized supercoiling during transcrip-tion, however, are not well understood. First, the forceswhich may anchor RNA polymerase in vivo and thus neces-sitate the rotation of DNA through the transcription ternarycomplex have not been clearly identified. Frictional drag onpolymerase with its nascent RNA and associated ribonucle-

oprotein complexes in the viscous environment of the cellmay be sufficient to cause supercoiling of the template (1, 2).Alternatively, polymerase may rotate freely about the DNAunless it is directly attached to some larger cellular structure.Second, those superturns which are produced during tran-scription may migrate rapidly by torsional diffusion manykilobases (kb) away from the site of transcription. Sincetranscription naturally generates equal numbers of supercoilsof opposite sign, torsional diffusion may lead to the rapidequilibration of the superhelix density along a DNA mole-cule. Finally, topoisomerases may relieve localized super-helical stress so quickly that effects of transcription on DNAtopology are normally not apparent. This last possibilityimplies that the superhelix density of DNA is effectivelydetermined by the activity of topoisomerase enzymes, whichhas been the traditional view of the control of DNA super-coiling, particularly in eubacteria (14, 15).The data presented here provide a quantitative basis for

evaluating the efficiency of supercoiling by RNA polymeraseand the activity of topoisomerases in E. coli. In a topoisom-erase I-deficient strain of E. coli, we show that RNA poly-merase can produce superturns at a rate which approachesthe maximal efficiency of one positive and one negativesuperturn per 10.4 base pairs transcribed. This efficientsupercoiling is observed during transcription of membrane-associated gene products. Transcription of soluble geneproducts reduces the efficiency of supercoiling by RNApolymerase. In a wild-type strain of E. coli, we demonstratethat both topoisomerase I and gyrase are kinetically compe-tent to relieve supercoils introduced by RNA polymerase.The mechanism of anchoring RNA polymerase during tran-scription and the consequences of topoisomerase activity forchromosomal supercoiling are discussed.

MATERIALS AND METHODSBacterial Strains and Plasmids. The strains used in this

study were RL88, a topoisomerase I-deletion mutant [A&tonB-cysB (A206) his-68 gyrA223 tyrA2 galK2 malAl xyl-7 man-2strr (rpsLJ25)] (from B. Bachmann, Yale University), andD1210, a lacIQ derivative of HB101 (from N. Linderoth, UCBerkeley). In RL88, lacIQ was provided in trans by pRG1(16).To make plasmids which differed only by the orientation of

an inserted gene, the double-stranded DNA oligomer

5'-TA TCT AGA GGG CCC CTC GAG-3'3'-AGA TCT CCC GGG GAG CTC AT-5'

containing Xho I and Xba I sites was inserted in bothorientations into the Nde I site of pBR322 to yield twomolecules, pBRXXopp and pBRXXpar. Fusions ofphoA (E.coli alkaline phosphatase) bounded byXho I at the 5' end andXba I at the 3' end (see below) were then ligated into

Abbreviation: IPTG, isopropyl P-D-thiogalactoside.*To whom reprint requests should be addressed.

10603

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

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

Proc. Natl. Acad. Sci. USA 89 (1992)

pBRXXopp and pBRXXpar to generate a pair of molecules,designated pXXopp and pXXpar. To make a strongly ex-pressed tet-phoA pair, a tet (tetracycline resistance) genefusion to the T7A1 promoter and RBSII translation initiationsequence was also constructed (see below) and used toreplace the 5' portion of tet in pXXopp and pXXpar. Thesemolecules, designated pNPopp and pNPpar, are shown inFig. 1. To reverse the orientation of tet in the pNP plasmids,the EcoRI-Sty I fragment of pBRXXopp and pBRXXpar wascut, its ends filled in with the Klenow fragment of DNApolymerase, and religated to yield plasmids in which theconstitutive tet was in reverse orientation from that onpBR322. A phoA fusion was inserted at the Xho I and Xba Isites and the 5' portion of tet was then replaced as above toyield pNPOtr and pNPPtr, with strongly expressed phoA andtet in opposed and parallel orientation, respectively. pMCoppand pMCpar were similarly derived by replacing the entire tetcoding sequence of pXXopp and pXXpar to the Msc I site ofpBR322 with a cat (chloramphenicol acetyltransferase) genefusion analogous to the earlier tet fusions.

All gene fusions were constructed in the transcription-translation vector pUHE21-2 (provided by N. Linderoth;made in the laboratory of H. Bujard, University of Heidel-berg) by PCR amplification and modification of phoA, cat,and tet. Care was taken to maintain a direct fusion with thephage T5-derived RBSII ribosome binding sequence, sinceoptimizing translation as well as transcription was an impor-tant goal. These fusions resulted in minor changes in theN-terminal sequence of each gene. Each of these fusionproteins retained its biological activity as evidenced by theability of our Tet and Cat fusion polypeptides to conferantibiotic resistance and the ability of the PhoA fusionproduct to hydrolyze p-nitrophenylphosphate.Topoisomer Analysis. Distributions of topoisomers were

trapped by following the method of Okazaki for quenchingenzymatic processes in cells (17). Cultures were grown tomid-logarithmic phase (OD600 = 0.5) and isopropyl j3-D-thiogalactoside (IPTG) was added to a final concentration of

A

I-:

.Xi

phoAA

...llp t

lp~~~~~~~~~~~~~~~~~~~~~~~i~~~~~~~~~~~~~~~~~~~

.; .:~~~~~~~~~~~~~1phoA

B

-35 -10

FIG. 1. Plasmids for the production and trapping of supercoilsgenerated by transcription. (A) Plasmids pNPopp and pNPpar are

6.4-kb derivatives of pBR322 with a 2.03-kb insert containing phoA.The phoA gene was inserted in both orientations at the Nde I site ofpBR322. tet is a modified version of the gene found on pBR322. Thedistance between promoters is approximately 2.1 kb for both pN-Popp and pNPpar. (B) Both tet and phoA are expressed from a

modified phage T7A1 promoter (black boxes) which is controlled bylac repressor (binding sites denoted by arrows). In addition, each

0.5 mM to derepress transcription. Transcription and othermetabolic processes were arrested by the addition of 1 vol of75% EtOH/2% phenol buffered with 21 mM NaOAc, pH 5.3,and 2 mM EDTA. Cells were spun down from this solution,and DNA was worked up by using standard miniprep proto-cols (ref. 18, pp. 1.38-1.41).Topoisomer gels (15 x 20 cm) consisting of 1.2% agarose

in 0.5 x Tris/phosphate/EDTA buffer (ref. 18, p. 8.23) wererun at 2 V/cm for 24 hr at 40C with constant recirculation ofbuffer. Chloroquine diphosphate was used at the concentra-tion indicated in the figure legends. Two-dimensional gelswere run in the first dimension as above, soaked with freshbuffer containing chloroquine at the appropriate final con-centration for 5 hr, and run in the second dimension for anadditional 10-12 hr.

RESULTSModel Plasmids for Transcription-Dependent Supercoiling.

Several plasmids were constructed with the goal of optimiz-ing the rate at which RNA polymerase supercoils the tem-plate (Fig. 1A). Repressible, strongly expressed genes were

juxtaposed in either parallel or opposed orientation. Strongexpression was achieved by cloning genes downstream froma modified T7 Al promoter flanked by operator sequences forthe lac repressor (19) (Fig. 1B). Translation of each genedepends on the ribosome binding sequence RBSII from phageT5, which confers a high level of translation (20). Genes wereselected whose protein products are targeted to the innermembrane (tet) or periplasmic space (phoA) of E. coli.Transcription (4, 5) and translation (21) of tet has beenpreviously shown to be essential for the production ofhypernegatively supercoiled species of pBR322 in topA mu-tants. Lodge et al. (21) have postulated that membraneinsertion of the nascent tet gene product during transcriptionis essential for hypemegative supercoiling of pBR322. How-ever, their experiments did not rule out the possibility thattranslation is required because ribosomes contribute to thehydrodynamic drag on the transcription apparatus.

Strong expression of genes on pNPpar and pNPopp shouldlead to continuous transcription by RNA polymerase and toformation of topological domains. If polymerase is efficientlyanchored, transcription of genes in opposed orientation on

pNPopp will trap positive supercoils formed downstream andnegative supercoils formed upstream from transcription (Fig.1A). This buildup ofDNA supercoils can be relieved only bytopoisomerases. Transcription ofgenes in parallel orientationon pNPpar introduces positive and negative superturns intoeach topological domain such that rotation of the DNA helixabout its axis can mitigate the buildup of supercoils.

Kinetics of Supercoiling During Transcription of Mem-brane-Associated Gene Products. To measure the kinetics ofsupercoiling of plasmids pNPopp and pNPpar, the change inlinking number, ALk, was measured by gel electrophoresisafter the induction of transcription in an E. coli strain, RL88,that lacks the gene for topoisomerase I. In bacteria, thedivision of labor between topoisomerases is such that topo-isomerase I removes negative superturns formed upstreamfrom transcription and gyrase removes positive superturnsformed downstream from transcription (5). The absence oftopoisomerase I should lead to the accumulation of negativesupercoils after induction of transcription.There is a lag of approximately 15-30 sec in the production

of supercoils on pNPopp (Fig. 2A). Since induction oftranscription of genes under control of the lac repressor isalmost instantaneous (22), this lag presumably represents thetime required to produce an efficient anchor plus the time ittakes for gyrase to act upon the overwound domain of the1|__ J B., AC _A

.st + 2 k X D ;

Xhol .4 .4

m

10604 Biochemistry: Cook et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

Proc. Natl. Acad. Sci. USA 89 (1992) 10605

A Time aifter IPTG addition, sec0 1 5. 30 4A5 60 90

B Time after IPTG addition. min0 (0.5 1 2 5 10

rS

_- Second dimension

FIG. 2. Topoisomer gels showing the time course for transcrip-tional supercoiling of plasmids pNPopp and pNPpar in E. coli strainRL88. (A) Bacterial cells harboring pNPopp were induced fortranscription by IPTG and quenched by EtOH/phenol at times up to90 sec after induction. DNA was electrophoresed in the presence ofchloroquine diphosphate at 60 ,ug/ml to resolve topoisomers. Thefast-migrating band found at times longer than 30 sec results from ahighly underwound DNA species (see text and Fig. 3). Formation ofthis species is complete within 60 sec after the induction of tran-scription. DNA at time 0 is positively supercoiled in this gel systemdue to the high concentration of chloroquine used. (B) Bacterial cellscontaining pNPpar were induced for transcription and quenched attimes up to 10 min and electrophoresed, using the same protocol asin A. A heterogeneous smear of topoisomers is produced aftertranscription is induced but no highly underwound plasmid DNA isseen. Two-dimensional gel electrophoresis shows that the 1-minsample is positively supercoiled under the conditions used here.

wound band is detected. Supercoiling by transcription in thisexperiment occurs more than 10-fold faster than previouslyreported (3, 10). This experiment demonstrates that RNApolymerase can, under certain circumstances, form an effec-tive barrier to the diffusion of supercoils.

In contrast, supercoiling of the parallel plasmid is muchless efficient (Fig. 2B). At times as long as 10 min there is aheterogeneous distribution of underwound topoisomers. Theslow supercoiling of the parallel plasmid is not due toinhibition oftranscription, however, since mRNA productionfrom phoA and tet is 2-3 times higher on pNPpar comparedwith pNPopp as measured by Northern blot analysis (unpub-lished data), and the copy number of pNPpar and pNPopp isthe same (unpublished data). Taken together, these resultsindicate that torsional diffusion of supercoils on pNPpar mustbe fast relative to gyrase turnover.To demonstrate that supercoiling by transcription depends

only on the orientation ofphoA and tet, plasmids have beenconstructed in which the orientation of tet is reversed com-pared with the plasmids shown in Fig. 1A. The same rapidsupercoiling is observed only when tet is arranged in opposedorientation to phoA, not when the two genes are parallel(unpublished data). Thus, it is the relative orientation ofphoAand tet, not their orientation with respect to some other locuson the plasmid, such as amp, which determines the efficiencyof supercoiling by transcription.The rate of supercoiling of pNPopp has been estimated by

using a two-dimensional topoisomer gel to separate the prod-ucts at the 45-sec time point (Fig. 3). The least negativelysupercoiled bands in this sample have a superhelix densityidentical to the distribution before addition of IPTG (unpub-lished data), while the most negatively supercoiled bands havea mobility characteristic of the fast-moving, underwoundspecies produced after 1 min of transcription (Fig. 2A). Be-tween the 45- and 60-sec time points the most positivelysupercoiled species in the distribution after 45 sec is com-pletely converted to the highly underwound form seen at 60sec. By counting topoisomer bands, we can estimate the

FIG. 3. Two-dimensional gel of the topoisomer distribution pro-duced 45 sec after the induction of transcription on pNPopp. Analiquot of the DNA from the 45-sec sample from Fig. 2A was rununder standard conditions with chloroquine diphosphate at 60 ,Ag/mlin the first dimension and at 250 jtg/ml in the second dimension. Theright arm of the arc comprises the most positively supercoiled DNA,and the most negatively supercoiled DNA extends to the end of thelower left arm of the arc. Every fifth band in the pattern is marked,beginning at the most positively supercoiled topoisomer and extend-ing to the point where the gel is no longer able to resolve discreetnegative topoisomers. The smear marked D is a dimer of the plasmidpRG1, which provides the lac repressor in trans. The band markedN is the nicked monomer of pNPopp. Dimeric pNPopp, not shown,also formed an arc of topoisomers in this gel.

maximum change in linking number, ALkmax, for a plasmidduring this interval and thus arrive at a rate for the introductionof superturns by RNA polymerase. Every fifth topoisomer isdenoted with a hash mark in Fig. 3 up to ALk = -65. The gelloses the ability to resolve the most underwound topoisomers,giving rise to a smear at the tail end of the distribution. Weestimate that there are 20 or more topoisomers in this smear.With an estimate of ALk,[,R, = -85 in -15 sec, a rate ofsupercoiling by RNA polymerase of .5.7 negative superturnsper sec can be calculated. Since this rate is the result oftranscription oftwo genes, the rate ofsupercoiling for one genecan exceed 2.8 superturns per sec. Taking an average elon-gation rate of 40 nucleotides per sec, RNA polymerase isintroducing supercoils at a rate which is greater than 70%o ofthe maximal theoretical rate. Furthermore, assuming thatneither topoisomerase III nor topoisomerase IV is involved inrelaxing positive supercoils produced by transcription, thisrapid change in linking number indicates that DNA gyrase canwork with a high efficiency in vivo.

Supercoiling During Transcription of a Soluble Gene Prod-uct. Plasmids pNPopp and pNPpar were designed with theidea that concerted transcription, translation, and membraneinsertion should be sufficient to effectively anchor RNApolymerase (21). To test whether membrane insertion isnecessary, we have constructed similar plasmids in which tethas been replaced by the coding sequence for cat, the genefrom Tn9 encoding resistance to chloramphenicol. Since thecat gene product is a cytosolic protein, these constructs testwhether two membrane-associated gene products are re-quired for efficient supercoiling by RNA polymerase.

Biochemistry: Cook et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

Proc. Natl. Acad. Sci. USA 89 (1992)

A two-dimensional gel of the topoisomer distributions ofpMCpar and pMCopp 1 min after the induction of transcrip-tion is shown in Fig. 4. pMCopp produces an arc of topo-isomers ranging from the superhelix density of uninducedplasmid to a highly underwound species reminiscent of theproduct band seen for pNPopp in Fig. 2A. In contrast,pMCpar is underwound to a much smaller extent. Productionof highly underwound pMCopp 1 min after induction oftranscription demonstrates that anchoring of polymerasedoes not depend absolutely on transcription of a membrane-associated gene product. However, supercoiling ofpMCoppis far less efficient than for pNPopp, as measured by the yieldofunderwound DNA produced after 1 min. Interestingly, thetopoisomer distribution of pMCopp remains largely un-changed between 1 and 30 min after induction oftranscription(unpublished data). This occurs in spite of the fact thatexpression ofCat protein is linear with time even 30 min afterinduction, and Cat protein is expressed at a high level,accumulating to 4% of total soluble protein during thisinterval (unpublished data). The possible significance of thisobservation is discussed below.Eiciency of Topoisomerase I and Gyrase in a Wild-Type

Strain. To test the efficiency of topoisomerase I and gyrase,the time course of supercoiling was studied in a wild-typetopoisomerase strain containing plasmids pNPpar and pN-Popp (Fig. 5). All of the topoisomers migrate as negativesupercoils due to the low concentration of chloroquine usedin these gels; an increase in mobility after transcription hasbeen induced indicates an increase in negative supercoiling.Two things are apparent upon inspection of these gels: (i)Each plasmid topoisomer distribution undergoes a change inLk of about -10 after induction of transcription; this newsteady-state Lk is established within 1 min after inductionand does not change over the next 10 min. (ii) There is nodifferential change in topology between opposed and parallelplasmids. We interpret this ALk to be a result of polymerasebinding and opening a transcription bubble which introduces

FIG. 4. Two-dimensional gel of the topoisomer distributions ofpMCpar and pMCopp 1 min after induction of transcription. The gelwas handled as described in the legend to Fig. 3. The right-most armof each arc is positively supercoiled with a superhelix densityequivalent to that before induction of transcription. The left-mostarm of the arc is highly negatively supercoiled DNA as in Fig. 3. Theband marked N is the nicked monomer of the plasmids, and the bandmarked laciQ is plasmid pRG1 after cutting with the restrictionenzyme Apa I, which does not cut pMCpar or pMCopp.

TrnMc. pNN-)pITIllill: U) 1 10)

'K I i :1A 1 1i

FIG. 5. Kinetics of supercoiling by transcription in a wild-typetopoisomerase strain. The distribution of topoisomers at 0, 1, and 10min after induction oftranscription is shown for pNPpar and pNPoppin strain D1210. This gel was run as described for Fig. 2 except thatthe chloroquine diphosphate concentration was reduced to 6 jg/ml.Under these conditions all of the topoisomers are negatively super-coiled.

about 1.7 positive turns per polymerase elsewhere in theDNA (1). These positive turns are rapidly removed bygyrase, and the net result is a decrease in the linking numberirrespective of gene orientation. A ALk of -10 is equivalentto the binding of about six polymerases per plasmid. Thisinterpretation suggests that both genes on each plasmid arecontinuously transcribed. In addition, we infer that bothgyrase and topoisomerase I efficiently remove transcriptionalsupercoils, since there is no differential change in linkingnumber between the parallel and opposed plasmids.

DISCUSSIONThe observation that transcription can lead to changes inDNA supercoiling has led to speculation that transcriptionmay determine the local superhelix density in vivo (5) and thatgenes may be regulated by transcription-induced changes insupercoiling (12, 13). But the in vivo parameters that influ-ence the ability of polymerase to supercoil the template havebeen largely unknown. These parameters include: (i) thetendency of RNA polymerase to rotate about the template,(ii) the rate at which supercoils are dissipated by torsionaldiffusion, and (iii) the ability of topoisomerases to removesuperturns introduced by transcription. The data presentedabove provide quantitative insight into these variables.

Kinetics of Supercoing by RNA Polynerase. We haveobserved efficient supercoiling by transcription on a plasmidwith strongly expressed, membrane-associated gene prod-ucts arranged in opposed orientation. We define "efficientsupercoiling" to mean that the rate of supercoiling for eachgene approaches 1 superturn per 10.4 base pairs transcribedand that the plasmid population undergoes a concertedchange in supercoiling during a brief (approximately 30-sec)time interval. In a AtopA genetic background, superturnsaccumulate in plasmid pNPopp at a rate of .5.7 per sec (Fig.3); the entire population of plasmids is extensively under-wound in the interval between 30 and 60 sec after inductionof transcription (Fig. 2A).Our data also provide insight into the rate of diffusion of

supercoils produced by transcription. With a parallel arrange-ment of genes, torsional diffusion can lead to the annihilationof supercoils, since positive and negative supercoils are pro-duced in a single topological domain. If torsional diffusionwere slow in vivo due to interactions ofDNA with proteins orother cellular factors, then rapid supercoiling should occurindependent ofgene orientation. But superturns accumulate ata much slower rate on the parallel plasmids than on theopposed ones (Figs. 2 and 4), and so torsional diffusion must

10606 Biochemistry: Cook et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0

Biochemistry: Cook et al.

be fast relative to the rate at which gyrase removes superturnsin vivo.Plasmid pMCopp, containing cat in place of tet, is less

efficiently supercoiled than pNPopp by the criteria outlinedabove. Although some plasmid molecules are extensivelyunderwound 1 min after inducing transcription, the bulk ofthe plasmid population is underwound only to an intermedi-ate superhelix density on this time scale (Fig. 4). Further-more, the observed topoisomer distribution does not changesignificantly between 1 and 30 min after induction of tran-scription, as assayed by two-dimensional gel electrophoresis(unpublished data). This slow supercoiling of pMCopp is notdue to the inhibition of gene expression at long times,however, since Cat protein accumulates linearly during thistime interval and the alkaline phosphatase production issimilar to that observed for pNPopp (unpublished data).The forces that immobilize RNA polymerase during tran-

scription in vivo are still obscure. It is tempting to suggest thatmembrane-associated gene products can anchor RNA poly-merase by concerted transcription, translation, and membraneinsertion of nascent polypeptides (21), but the fact that afraction of the pMCopp plasmid population is rapidly super-coiled implies that membrane attachment by both gene prod-ucts is not required to immobilize RNA polymerase. How-ever, the inefficiency of supercoiling at times longer than 1 minon pMCopp presents difficulties for any model which postu-lates that efficient driven rotation of DNA is an intrinsicproperty of elongating RNA polymerase (23). With such amodel, one would expect to observe rapid, complete super-coiling for any two genes expressed in opposed orientation. Itis likely that anchoring of RNA polymerase is a complexfunction of the local cellular environment of a given DNAmolecule.

Activity of Topoisomerase I and Gyrase in Vivo. The resultspresented here provide quantitative evidence for the kineticefficiency of topoisomerases in coping with supercoiling bytranscription. In experiments with strain RL88, ALk..JI isdue to DNA gyrase, since topA is deleted. The rapid changein linking number of pNPopp, -5.7 superturns per sec (Fig.3), indicates that gyrase can work with a high turnover ratein vivo. This result is somewhat surprising in light of thegyrA223 compensatory mutation in RL88. Compensatorymutations normally decrease the intrinsic activity of gyrase(24, 25). This rate is significantly faster than in vitro estimatesof about 0.5-1 turnover per sec (equivalent to 1-2 superturnsper sec) (26). It is possible that there is more than one gyraseacting on each plasmid in vivo, and our measured rate istherefore not a turnover number per enzyme. It is alsoconceivable that in a region where positive superturns arerapidly introduced by transcription or replication, slippage bygyrase is reduced, increasing enzyme efficiency (27).

Experiments in a wild-type topoisomerase strain supportthe notion that both gyrase and topoisomerase I can removesuperturns at a rate equal to their introduction by RNApolymerase. In the wild-type strain, if either enzyme laggedbehind the rate of supercoiling by RNA polymerase, then thesteady-state Lk would differ for the parallel and opposedplasmids. Torsional diffusion can contribute to the relaxationof superturns on the parallel but not the opposed plasmid.Since a differential change in Lk is not observed, we concludethat both topoisomerase I and gyrase effectively relieve therapid introduction of supercoils by transcription. This, inturn, implies that there is excess topoisomerase activity in E.coli such that a rapid burst of transcriptional activity afterinduction by IPITG can be accommodated. This result hasimplications for chromosomal supercoiling, since presum-ably the topoisomerases respond with equal rapidity toinduction of transcription on the chromosome.The kinetic efficiency of the topoisomerases supports the

idea from many other studies that these enzymes make up a

Proc. Natl. Acad. Sci. USA 89 (1992) 10607

homeostatic system for the maintenance of superhelix den-sity around an optimal set point in E. coli (24, 25, 28, 29). Itfurther suggests that it is the topoisomerases-and not DNA-tracking processes such as transcription-that are the pri-mary determinants of the level of DNA supercoiling in E.coli. Supercoiling by polymerase could be transiently evidentduring the time between induction of a gene and the arrivalof topoisomerase I orgyrase at a supercoiling hot spot. Thus,fluctuations in superhelix density, as opposed to stablechanges due to transcription, may be important in a varietyof biological situations. Consistent with this idea, B- toZ-DNA transitions have been reported upstream of the tetgene on pBR322 in the presence of wild-type topoisomerases(7). Presumably these structural transitions occur on a subsetof plasmid molecules for which gyrase, but not topoisomer-ase I, has removed superturns generated by transcription. Ifsupercoiling by transcription is a transient phenomenon, thenone might expect its effects to be most relevant to biologicalprocesses that occur infrequently. Lodge and Berg (30), forinstance, have presented evidence that transposition of TnSis favored at sites upstream of the tet gene. Since transpo-sition is necessarily a rare event, it might be preferentiallystimulated by localized fluctuations in supercoiling. Generalrecombination may also be sensitive to fluctuations in super-coiling due to transcription.We thank Paul Selvin, Nathan Hunt, Rolf Sternglanz, James

Wang, and Nick Cozzarelli for helpful conversations, Nora Lin-deroth for supplying technical advice and materials, and MarieAlberti for DNA sequencing. This work was supported by NationalInstitutes of Health Grant FD 8R1 GM 41911A-03-NF-03/92.1. Gamper, H. B. & Hearst, J. E. (1982) Cell 29, 81-90.2. Liu, L. F. & Wang, J. C. (1987) Proc. Natl. Acad. Sci. USA 84,

7024-7027.3. Lockshon, D. & Morris, D. R. (1983) Nucleic Acids Res. 11,

2999-3017.4. Pruss, G. J. & Drlica, K. (1986) Proc. Natl. Acad. Sci. USA 83,

8952-8956.5. Wu, H.-Y., Shyy, S., Wang, J. C. & Liu, L. F. (1988) Cell 53,

433 440.6. Tsao, Y.-P., Wu, H.-Y. & Liu, L. F. (1989) Cell 56, 111-118.7. Rahmouni, A. R. & Wells, R. D. (1989) Science 246, 358-363.8. Ostrander, E. A., Benedetti, P. & Wang, J. C. (1990) Science 249,

1261-1265.9. Giaever, G. N. & Wang, J. C. (1988) Cell 55, 849-856.

10. Brill, S. J. & Sternglanz, R. (1988) Cell 54, 403-411.11. Wang, J. C. (1991) J. Biol. Chem. 266, 6659-6662.12. Pruss, G. J. & Drlica, K. (1989) Cell 56, 521-523.13. Lilley, D. M. J. & Higgins, C. F. (1991) Mol. Microbiol. 5,779-783.14. Gellert, M. (1981) Annu. Rev. Biochem. 50, 879-910.15. Wang, J. C. (1985) Annu. Rev. Biochem. 54, 665-697.16. Griffin, T. G. & Kolodner, R. D. (1990) J. Bacteriol. 172, 6291-6299.17. Okazaki, R. (1974) in Methods in Molecular Biology, ed. Wickner,

R. B. (Dekker, New York).18. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1988) Molecular

Cloning: A Laboratory Manual (Cold Spring Harbor Lab., ColdSpring Harbor, NY).

19. Deuschle, U., Kammerer, W., Gentz, R. & Bujard, H. (1986) EMBOJ. 5, 2987-2994.

20. Bujard, H., Gentz, R., Lanzer, M., Stuber, D., Muller, M., Ibra-himi, I., Hauptle, M. T. & Dobberstein, B. (1987) Methods En-zymol. 155, 416-433.

21. Lodge, J., Kazic, T. & Berg, D. E. (1989) J. Bacteriol. 171,2181-2187.

22. Straney, S. B. & Crothers, D. M. (1987) Cell 51, 699-707.23. Heggeler-Bordier, B., Wahli, W., Adrian, M., Stasiak, A. & Du-

bochet, J. (1992) EMBO J. 11, 667-672.24. DiNardo, S., Voelkel, K. A., Sternglanz, R., Reynolds, A. E. &

Wright, A. (1982)- Cell 31, 43-51.25. Pruss, G. J., Manes, S. H. & Drlica, K. (1982) Cell 31, 35-42.26. Higgins, N. P., Peebles, C. L., Sugino, A. & Cozzarelli, N. R.

(1978) Proc. Natl. Acad. Sci. USA 75, 1773-1777.27. Westerhoff, H. V., O'Dea, M. H., Maxwell, A. & Gellert, M. (1988)

Cell Biophys. 12, 157-181.28. Tse-Dinh, Y.-C. (1985) Nucleic Acids Res. 13, 4751-4763.29. Menzel, R. & Gellert, M. (1983) Cell 34, 105-113.30. Lodge, J. K. & Berg, D. E. (1990) J. Bacteriol. 172, 5956-5960.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 22

, 202

0