Quarterly Reviews of Biophysics , (), pp. –. Printed in the United Kingdom

# Cambridge University Press

Moving one DNA double helix through

another by a type II DNA topoisomerase: the

story of a simple molecular machine

JAMES C. WANG

Department of Molecular and Cellular Biology, Harvard University, Cambridge,

Massachusetts �����, USA

.

. The double helix structure of DNA and its topological ramifications

. Entering the DNA topoisomerases

. Type II DNA topoisomerases

.

. Transporting one DNA double helix through another

. The clamp model: a type II DNA topoisomerase as an ATP-modulated

protein clamp

. The number of gates in the protein clamp

. Three-dimensional structures of type II DNA topoisomerase fragments

. How are the various subfragments connected?

. A molecular model of DNA transport by a type II DNA topoisomerase

.

. ATP utilization

. How tight is the coupling between DNA transport and ATP binding and

hydrolysis?

.. Dependence of the coupling efficiency on the degree of supercoiling of

the DNA substrate

.. Coupling efficiency and DNA binding

. DNA relaxation by a type II DNA topoisomerase: how is the high

efficiency of coupling achieved?

.. ATP hydrolysis and the closure of the enzyme clamp

.. Coupling of ATPase activity to DNA binding

.. Structural changes in the enzyme upon closure of the N-gate

.. ATP binding}hydrolysis and DNA cleavage

.. Entrapment of the T-segment and opening of the DNA gate

.. Closure of the DNA gate and exit of the T-segment

J. C. Wang

. Directionality of DNA transport: why is bacterial gyrase unique?

.. Structural basis of the ability of bacterial DNA gyrase to catalyse

the ATP-dependent negative supercoiling of DNA

.. The ATP-independent DNA relaxation activity of gyrase

.

.

.

.

.. The double helix structure of DNA and its topological ramifications

The discovery of the double helix structure of DNA led immediately to questions

on the mechanics of unravelling its intertwined strands during replication. If a

parental DNA is to be duplicated into two progeny molecules by separating its two

strands and copying each, then the strands must untwine rapidly during

replication (Watson & Crick, ).

That DNA indeed replicates in such a semiconservative fashion was soon

demonstrated by the Meselson–Stahl experiment (). At first, it appeared that

the unravelling of the intertwined strands should not pose an insurmountable

mechanical problem. The two strands at one end of a linear DNA, for example,

can be pulled apart with concomitant rotation of the double-stranded portion of

the molecule around its helical axis. If the strands of a DNA double helix are to

separate at an estimated replication rate of base pairs (bp) per minute, then

the speed of this rotation would be revolutions per minute from the bp

per turn helical geometry of the double helix. This speed, though impressive,

seemed reasonable: owing to the slender rod-like shape of the double helix, the

estimated viscous drag for this rotational motion is actually rather modest

(Meselson, ).

Two findings in the s, however, heightened interests in the DNA uncoiling

problem. First, evidence began to accumulate that the lengths of DNA molecules

from natural sources are much longer than previously realized. Once it was found

that mechanical shear during DNA preparation can cause extensive breakage of

DNA, new records of longer and longer DNA molecules were set. It soon became

clear that an entire chromosome containing tens of millions of base pairs is

consisted of a single DNA molecule, with a length of several millimetres along its

contour (Kavenoff et al. ). For such a long DNA, the simple idea of

unravelling its intertwined strands by rotating the entire thread-like molecule

around its helical axis is untenable. Second, the s also saw the discovery of

various ring-shaped or ‘circular’ DNA molecules. Of particular importance was

the finding that the small DNA of polyoma virus is a double-stranded ring with

intact strands (Dulbecco & Vogt, ; Weil & Vinograd, ). It was shown that

for such a ‘covalently closed’ DNA, the two complementary strands are

topologically linked, as predicted by the double-helix structure: a pair of

intertwined single-stranded DNA rings in a double-stranded DNA ring can not

A simple molecular machine

(b)(a)

A

B

C

Fig. . Schematic drawings illustrating two topological problems of DNA during

semiconservative replication. In (a), the general problem of separating the two parental

strands is illustrated for a double-stranded DNA ring. In this two-line representation of the

duplex DNA ring, each of the circular DNA strands is represented by a closed line. Because

of the double helix geometry of DNA, the two strands are topologically linked; the degree

of linkage between them must gradually reduce as DNA replication proceeds, and the two

strands must become completely unlinked at the end of a round of replication. For the two

closed lines drawn in the illustration, one way of unlinking them is to carry out multiple

cycles of a simple operation: in each cycle, one line is transiently broken, the other line is

then passed through the break once before the break is resealed. The linking number (Lk)

between the two closed strands is the minimal number of such cycles that is required to

unlink them. Inspection of the particular drawing shows that all crossovers are of the same

sign and each strand passage event removes two crossovers ; Lk is thus equal to one half of

the number of crossovers. For other ways of defining Lk, see Crick (), Cozzarelli et al.

(), Wang (). In (b), the conversion of twists between the strands of an unreplicated

DNA segment to intertwines between two fully replicated progeny DNA molecules is

illustrated. Two replication forks are shown to approach each other near the end of a round

of DNA replication (drawing A). If copying of the unreplicated portion of the parental

DNA segment is faster than unlinking the parental strands coiled around each other, two

intertwined progeny molecules are formed (C). Under such conditions, the replication

product of a circular DNA would be a pair of multiply linked rings. Illustration taken from

Varshavsky et al. ().

be separated without breaking at least one of the strands. A parameter defining the

degree of topological linkage between the two strands of a duplex DNA ring is

termed the linking number (Lk). As this quantity will pop up again in subsequent

discussions, an operational definition of Lk is illustrated in Fig. a.

Whereas the problem of unravelling the complementary strands of a long linear

DNA during replication might appear to be kinematic in nature, the same

problem for a covalently closed DNA ring is surely a topological one. Actually,

because a linear chromosome inside a cell is organized in a compact form with

J. C. Wang

multiple loops, the problem of separating its strands is not all that different from

that for a covalently closed DNA ring. Therefore, it would appear that during

evolution, as DNA became very long or circular, a solution must be found for the

disentanglement of its topologically intertwined strands during replication.

The separation of two parental DNA strands during replication is the best

known topological manifestation of the double helix structure. It is not the only

one, however, and a variation of the theme is illustrated in Fig. b. In the top

drawing (A), a DNA molecule near the end of its replication is shown, with a short

stretch of the molecule remaining unreplicated. If unravelling of the parental

strands in this unreplicated segment is incomplete before the strands are

completely unpaired and copied, then the residual intertwists between the

parental strands would be converted to intertwines between the newly replicated

progeny molecules (B and C) (Sundin & Varshavsky, , ; Varshavsky et

al. ). These interwoven DNA duplexes must be resolved if the progeny

molecules are to be segregated into two newly divided cells.

Historically, replication had inspired much of the earlier considerations on the

topological problems of DNA. It is now well-known that the necessity of

disentangling DNA strands or duplexes also arises in nearly all other cellular

transactions of DNA, including transcription, chromosome condensation and

decondensation, and recombination. These aspects have been reviewed elsewhere

(see Wang, , and references therein). As illustrated by the two examples

summarized above, many of the topological problems of DNA are deeply rooted

in its double-helix structure.

. Entering the DNA topoisomerases

Nature invented a family of enzymes to solve the topological problems of DNA.

These enzymes, termed the DNA topoisomerases (Wang & Liu, ), catalyse

the interpenetration of DNA strands or double helices. In their presence, DNA

strands and double helices can go through one another as if there were no physical

boundaries in between. Since the discovery of the first member of this family of

enzymes in the bacterium Escherichia coli nearly three decades ago (Wang, ),

all organisms have been found to possess several of these enzymes. The limited

scope of this review does not permit even a cursory coverage of the topoisomerase

literature; references on earlier studies can be found in a number of monographs

(Cozzarelli & Wang, ; Bates & Maxwell, ; Liu, a, b).

The DNA topoisomerases perform their magic through transient breakage of

DNA strands. They can be divided into two types: the type I enzymes break one

DNA strand at a time and the type II enzymes both strands of a double helix in

concert (Brown & Cozzarelli, ; Liu et al. ). The type I enzymes can be

further divided into two subfamilies: the IA subfamily represented by bacterial

DNA topoisomerases I and III and eukaryotic DNA topoisomerase III, and the

type IB subfamily by eukaryotic DNA topoisomerase I and pox virus DNA

topoisomerases. Members of the same subfamily are closely related in their amino

acid sequences and reaction characteristics, and members of different subfamilies

A simple molecular machine

share little sequence homology and are mechanistically distinct (for reviews, see

Wang, , and references therein). The type II DNA topoisomerases are

thought to form a single subfamily, but recent studies of an enzyme DNA

topoisomerase VI from archaeal hyperthermophiles suggest that the type II

enzymes may also be divided into two subfamilies IIA and IIB (Bergerat et al.

) : subfamily IIB is represented by archaeal DNA topoisomerase VI, and

subfamily IIA by all other type II topoisomerases, including bacterial gyrase

(DNA topoisomerase II), bacterial DNA topoisomerase IV, yeast and Drosophila

DNA topoisomerase II, mammalian DNA topoisomerases IIα and IIβ, and T-

even phage DNA topoisomerases. Because mechanistic analysis of archaeal DNA

topoisomerase VI is still in its infancy, this fascinating enzyme will not be

discussed in this review. It is plausible, however, that the archaeal enzyme may

share many of the mechanistic features to be described and discussed below.

All DNA topoisomerases catalyse transient DNA strand breakage by

transesterification, a mechanism first postulated in (Wang, ). For the

type I enzymes, the phenolic oxygen of an active-site tyrosyl residue undergoes

transesterification with a phosphoryl group in a DNA strand, breaking the DNA

phosphodiester bond and forming a phosphotyrosine linkage (Tse et al. ;

Champoux, ). Rejoining of the DNA strand occurs through an apparent

reversal of the DNA breakage reaction. Because the enzyme–DNA complex is

likely to undergo large conformation changes between the DNA breakage and

rejoining steps (Lima et al. ), the two steps are not necessarily the exact

microscopic reversal of each other (Stivers et al. ). For DNA topoisomerases

other than the type IB enzyme, the tyrosyl residue becomes linked to a DNA «-phosphoryl group in the formation of the covalent intermediate (Fig. ) ; for the

type IB enzymes, it becomes linked to a «-phosphoryl group. The type I enzymes

act as monomers in their breakage and rejoining of DNA strands one at a time; the

type II enzymes are dimeric, and a pair of active-site tyrosyl residues undergo

transesterification with a pair of phosphoryl groups in the two strands of a duplex

DNA (Morrison and Cozzarelli, ; Sander & Hsieh, ; Liu et al. ).

. Type II DNA topoisomerases

In this review, the focus is on the reaction mechanism of the type II DNA

topoisomerases. The type II enzymes are of special interest in a number of ways.

They unlink DNA catenanes and resolve intertwined chromosome pairs during

mitosis, and in their absence cells die (reviewed in Yanagida & Sternglanz, ).

These enzymes have been identified as the targets of a large number of natural

toxins, antimicrobial agents, and anti-tumour therapeutics (Liu, b ; Maxwell,

). They are also DNA-dependent ATPases, and how they couple their

manipulation of DNA to ATP binding and hydrolysis raises fascinating

mechanistic questions.

The first members of the type II subfamily of DNA topoisomerases, bacterial

DNA gyrase (DNA topoisomerase II), was discovered by Gellert et al. () as

J. C. Wang

5′ DNA

OO

O

OP DNA 3′CH2

CH2

OH 5′ DNA

3′ OH

OO

OP

O

CH2 DNA 3′

CH2

Fig. . Formation of a covalent intermediate between a DNA and a DNA topoisomerase.

Nucleophilic attack of an enzyme tyrosyl group on a DNA backbone phosphorus leads to

the breakage of the DNA strand and the simultaneous formation of a phosphotyrosine bond

between the enzyme and the DNA. In the illustration shown, the phenolic oxygen of the

tyrosyl group is attacking from the opposite side of a «-oxygen, and the enzyme tyrosyl

becomes linked to a DNA «-phosphoryl group in the covalent intermediate. Such a

covalent intermediate has been identified in reactions catalyzed by the type IA and type II

DNA topoisomerases; for the type IB DNA topoisomerases, the enzyme tyrosyl group

attacks from the opposite site of a «-oxygen and becomes linked to a DNA «-phosphoryl

group in the covalent intermediate.

an ATP-dependent DNA negative supercoiling activity. The term negative

supercoiling refers to reducing the linking number between the two strands of a

duplex DNA ring. If a DNA ring is in its most stable structure (termed a relaxed

DNA), its linking number Lk° at a temperature of – °C and in a dilute

aqueous buffer around neutral pH is readily estimated to be N}±, where N is

the size of the DNA ring in base pairs, and ± is the helical periodicity of a

typical DNA under these conditions (Wang, ; Rhodes & Klug, ). If Lk

of a DNA ring deviates significantly from Lk°, torsional and flexural strains are

introduced into the molecule, and the DNA ring becomes contorted in space

much like the deformation of a torsionally unbalanced rope. It is this contorted

A simple molecular machine

409 679 1201 1428

(C)(B)(A)

ATP Y* (dimer)

Yeast DNA topoisomerase II

1

8041 1 875

GyrAGyrB

39 60 52

Phage T4

Fig. . A schematic alignment of the amino acid sequences of three representative type II

DNA topoisomerases. Yeast DNA topoisomerase II is composed of a single subunit, E. coli

DNA gyrase of two (GyrA and GyrB), and phage T DNA topoisomerase of three (encoded

by genes , , and ). A, B, and C along the yeast polypeptide denote three protease-

sensitive sites. For the yeast enzyme, the region between sites A and B, corresponding to the

C-terminal half of the E. coli GyrB protein, is termed the B« subfragment; the region

between B and C, corresponding to the N-terminal two thirds of the E. coli GyrA protein,

is termed the A« subfragment. Thick lines represent regions with significant homology and

thin lines regions with marginal homology. The approximate locations of the ATPase site,

the active-site tyrosine, and residues that form the primary dimer interface in the yeast

enzyme are indicated by ATP, Y*, and (dimer) above the line representing the enzyme.

Illustration drawn after fig. in Berger & Wang ().

shape that inspired the term supercoiling, supertwisting, or superhelix formation.

By definition, a negatively supercoiled DNA is one with Lk!Lk°, and a

positively supercoiled DNA one with Lk"Lk°. DNA gyrase can apparently

utilize the chemical energy of ATP hydrolysis to drive a DNA ring to an

energetically unfavourable state with an Lk!Lk°.The discovery of bacterial DNA gyrase was soon followed by the discovery of

phage T DNA topoisomerase (Liu et al. ; Stetler et al. ) and eukaryotic

DNA topoisomerase II (Baldi et al. ; Hsieh & Brutlag, ; Miller et al.

). The phage and eukaryotic enzymes differ from bacterial DNA gyrase in

that they catalyse the relaxation of a positively or negatively supercoiled DNA in

the presence of ATP, but not DNA supercoiling. These enzymes also differ in

their quaternary structures. Bacterial gyrases are comprised of two subunits

(Gellert et al. ), the eukaryotic enzymes one (Miller et al. ; Sander &

Hsieh, ; Goto & Wang, ), and the phage T enzyme three (Liu et al.

; Stetler et al. ). The amino acid sequences of the enzymes showed

clearly, however, that the enzymes are closely related (see for example, the

compilation of Caron & Wang, ) ; a schematic alignment of the amino acid

sequences of three type II DNA topoisomerases of different quaternary structures

is illustrated in Fig. .

J. C. Wang

.

. Transporting one DNA double helix through another

A breakthrough in understanding the mechanism of the type II DNA

topoisomerases came around . In earlier years, it was assumed that in

reactions catalysed by all topoisomerases only one of the two strands of a DNA

double helix would be transiently broken so that the intact strand could hold the

broken ends close to each other for their subsequent rejoining. Several lines of

evidence showed, however, that bacterial DNA gyrase and phage T DNA

topoisomerase do not act in this fashion. First, it was observed in that the

addition of a protein denaturant to a plasmid-bound gyrase would linearize rather

than nick the plasmid in the presence of nalidixate, an antibiotic that targets

bacterial gyrase (Sugino et al. ; Gellert et al. ). Furthermore, a protein

moiety was found to be covalently attached to each of the two « ends of the linear

DNA produced in this reaction (Morrison & Cozzarelli, ), and this protein

moiety was shown to be the A-subunit of gyrase (Tse et al. ; Sugino et al.

). These findings demonstrated that the double-stranded breakage of DNA

by gyrase is a consequence of covalent adduct formation between the enzyme and

both strands of a duplex DNA. Second, it was found that the bacterial and phage

enzymes can interconvert several novel topological forms of duplex DNA rings,

for examples the unlinking of DNA catenanes or the knotting of unknotted rings

(Liu et al. ; Kreuzer & Cozzarelli, ; Mizuuchi et al. ). These

topological transformations strongly suggested that the bacterial and phage

enzymes can catalyse the passage of one DNA double helix through a transient

double-stranded break in another. Third, quantitative measurements of linking

number changes of DNA rings by DNA gyrase or phage T DNA topoisomerase

showed that these enzymes alter Lk in units of two (Brown & Cozzarelli, ; Liu

et al. ; Mizuuchi et al. ). Such an even-number change in Lk can best be

interpreted in terms of a mechanism in which an enzyme mediates the passage of

one DNA segment through another of the same DNA ring during each round of

reaction (Brown & Cozzarelli, ; Liu et al. ; Mizuuchi et al. ). It

turned out that the dyadic change in Lk by such an operation was foretold in a

mathematical analysis of the linking number between the two edges of a closed

ribbon (Fuller, ), and could also be inferred from earlier examples illustrating

the topological properties of a DNA ring (Crick, ). Studies of eukaryotic type

II DNA topoisomerases also led to the same conclusion that these enzymes act by

a double-stranded DNA breakage, passage, and rejoining mechanism (Baldi et al.

; Hsieh & Brutlag, ).

Once it was recognized that the type II enzymes transiently cleave double-

stranded DNA, it was straightforward to determine the relative positions of the

pair of cleavage sites such an enzyme would make in a DNA duplex. From these

measurements, DNA breakage by a type II topoisomerase was shown to involve

transesterification between the active-site tyrosyl residues of the enzyme and a pair

of staggered phosphoryl groups four base pairs apart (Morrison & Cozzarelli,

A simple molecular machine

; Sander & Hsieh, ; Liu et al. ). In the case of bacterial DNA gyrase,

the A-subunit is often referred to as the ‘DNA breakage-and-rejoining subunit ’

because of the presence of an active-site tyrosyl group in each A-subunit (Tse et

al. ; Sugino et al. ; Horowitz & Wang, ). However, essential

residues from both A- and B-subunits are most likely present in each catalytic

pocket for transesterification. Whereas purified DNA gyrase A-subunit has no

detectable DNA cleavage}rejoining activity (Higgins et al. ), the combination

of the A-subunit and a C-terminal fragment of the B-subunit does (Brown et al.

; Gellert et al. ). Similarly, yeast DNA topoisomerse II lacking the

‘GyrB’ portion of the polypeptide has no DNA cleavage activity, but a fragment

comprised of amino acid residues to about , corresponding to the C-

terminal half of the gyrase B-subunit plus the N-terminal two thirds of the gyrase

A-subunit, is capable of forming enzyme–DNA covalent adduct (Berger et al.

). Site-directed mutagenesis studies of yeast DNA topoisomerase II further

confirmed the requirement of residues in both A« and B« domains of the enzyme

(Q. Liu and J. Wang, to be published; see the legend to Fig. for the boundaries

of the A« and B« domains).

. The clamp model: a type II DNA topoisomerase as an ATP-modulated

protein clamp

From the above description, a type II DNA topoisomerase presumably binds to

a double-stranded DNA segment, to be termed the gate-segment or G-segment,

with the pair of active-site tyrosyl residues of the enzyme positioned near a pair

of DNA phosphorus atoms for creating an opening or gate in the DNA double

helix. The DNA-bound enzyme must then capture a second double-stranded

DNA segment, termed the transported-segment or T-segment, in order to move

it through the G-segment. How does this occur?

Studies carried out between the mid s and early s led to the proposal

of the clamp model, in which a type II DNA topoisomerase is assumed to act as

an ATP-modulated protein clamp (Roca & Wang, ). According to this model,

the enzyme possesses a pair of jaws, which come into contact to close a molecular

gate upon binding of ATP, and come apart to re-open the gate upon ATP

hydrolysis and product release. The closing and opening of the jaws can occur

either in a free enzyme, or in one bound to a G-segment. The ATP-modulated

cycling time is faster when the enzyme is DNA-bound, and the magnitude of this

increase in rate can be deduced from the DNA dependence of the rate of ATP

hydrolysis. In the case of yeast DNA topoisomerase II, the Michaelis constant KM

and the apparent turnover number kcat

were measured to be ± m and ± s−" per

dimeric enzyme in the absence of DNA (at °C in a medium containing m

Tris acetate, pH ±, m potassium acetate, m magnesium acetate, and

m -mercaptoethanol). In the presence of DNA, positive cooperativity

between the two ATPase sites was observed. The concentration of ATP at half-

maximal velocity drops to about ± m, and kcat

is increased by about -fold

(Lindsley & Wang, a). Thus at saturating ATP concentrations the motion of

J. C. Wang

G G

T+ATP

Fig. . The protein clamp model of a type II DNA topoisomerase. An enzyme molecule is

depicted as a dimeric protein clamp bound to a DNA segment termed the gate or G-

segment. According to this model, closing and opening of the pair of jaws of the protein

clamp are modulated by ATP binding and hydrolysis and release of the hydrolytic products.

A second DNA segment, termed the transported or T-segment, can enter the DNA-bound

protein clamp; ATP-mediated closure of the jaws captures the T-segment and drives it

through the transiently opened DNA gate in the G-segment. Drawing taken from Roca &

Wang ().

clamp opening and closing is probably accelerated by about -fold when the yeast

enzyme is bound to DNA. For E. coli DNA gyrase, binding to DNA was also

found to stimulate the ATPase activity of the enzyme (Sugino et al. ; Sugino

& Cozzarelli, ; Staudenbauer & Orr, ; Maxwell & Gellert, ).

The ATP-modulated opening and closing of the enzyme clamp is assumed to

be directly related to the type II topoisomerase-mediated transport of one DNA

double helix through another. When a DNA-bound protein clamp is in the open

state, a T-segment can enter through the open jaws into the enzyme interior. This

entrance is presumably facilitated by weak DNA–protein interactions and is

also influenced by the topology of the DNA substrate. Closing of the jaws by

ATP-binding then captures the T-segment and drives it through the G-segment

(Fig. ).

The clamp model was first hinted by images of DNA gyrase molecules in

electron micrographs (Kirschhausen et al. ). Experimental evidence

supporting the model came initially from studies of DNA binding by yeast

DNA topoisomerase II in the absence and presence of «-adenylyl-β,γ-

imidodiphosphate (ADPNP), a non-hydrolysable β,γ-imido analogue of ATP. In

the absence of the nucleotide, the yeast enzyme binds readily to linear or circular

DNA, with a slight preference for supercoiled DNA (by about a factor of ). Pre-

incubation of the enzyme with ADPNP, however, converts it to a form capable of

binding linear DNA but not any form of DNA rings (Roca & Wang, ). The

specific binding of linear DNA to the ADPNP–enzyme complex is not caused by

interactions between the DNA ends and the enzyme, as the ends of an enzyme-

bound linear DNA can be readily joined by DNA ligase. Thus it appeared that the

A simple molecular machine

binding of ADPNP to the enzyme converts it to an annular form with a hole

sufficiently large for a linear DNA to thread through (Roca & Wang, ).

Results of similar experiments with E. coli DNA gyrase were, however, not as

clear cut (A. Maxwell and M. Gellert, personal communications) ; it is plausible

that measurements with the bacterial enzyme are subject to complications arising

from the association}dissociation of the DNA gyrase subunits. In a later section,

mechanistic differences between DNA gyrase and other type II DNA

topoisomerases will be discussed.

The clamp model explains readily an earlier observation with Drosophila DNA

topoisomerase II that addition of ADPNP makes the Drosophila enzyme resistant

to dissociation by salt if it is bound to a DNA ring, but not if it is bound to a linear

DNA (Osheroff, ). For a type II topoisomerase bound to a DNA ring, closing

the protein clamp by ADPNP would introduce a salt-resistant topological link

between the protein and DNA (Pommier et al. ; Roca & Wang, ). The

same interpretation explains why a linear DNA bound to the enzyme–ADPNP

complex is readily dissociated from the complex upon exposure to high salt, but

ligation of the ends of the linear DNA before salt addition prevents its separation

from the protein (Roca & Wang, ).

According to the clamp model, the jaws of a type II DNA topoisomerase bound

to a DNA G-segment must close after the entrance of the T-segment in order to

effect the transport of the T-segment through the G-segment. This notion was

supported by an order-of-mixing experiment in which two different DNA rings

and ADPNP were added sequentially to yeast DNA topoisomerase II. Under

conditions such that an enzyme-bound DNA could not be displaced by a

subsequently added second DNA, it was shown that the G-segment always resides

in the first DNA ring added, and catenation between the two sequentially added

DNA rings is possible only if the second ring is added before the addition of

ADPNP (Roca & Wang, ).

. The number of gates in the protein clamp

A protein clamp has at least one molecular gate which closes when the jaws of the

clamp come into contact and opens when they come apart. The question was

raised, however, about whether there are separate gates for the entrance and exit

of the T-segment, or there is only a single gate through which the T-segment

would first enter and later exit (Roca & Wang, ). Fig. illustrates the two

situations. In each case, an ATP-modulated enzyme clamp bound to a G-segment

is shown, and a T-segment can enter the clamp in its open state as described in the

section above. In the two-gate model (a), closing the jaws denoted by N drives the

T-segment through the transiently opened G-segment and out of a second protein

gate on the opposite side of the entrance gate [right-side drawing in (a)]. In the

one-gate model (b), closing of the protein clamp drives the T-segment through the

G-segment into a holding dock; following ATP-hydrolysis, the clamp re-opens

for the exit of the T-segment through the same gate it had entered earlier. As

illustrated in the right-side drawing in (b), in the one-gate model at least a part of

J. C. Wang

(a)

(b)

G

T

T

N N N N

G

T

T

T

G

N N

Fig. . The two-gate (a) and one-gate (b) model of DNA transport by a type II DNA

topoisomerase. See the text for explanation. Illustration taken from Roca & Wang ().

the G-segment must dissociate from the enzyme prior to the exit of the T-

segment, because backtracking of the T-segment through the G-segment would

mean no net transport.

In the two-gate model, exist of the T-segment requires transient disruption of

protein–protein interaction between the pair of jaws constituting the exit gate; in

the one-gate model, transient disruption of DNA-protein interaction is involved

instead. The two-gate model was first proposed in for DNA gyrase

(Mizuuchi et al. ; Wang et al. ). Because gyrase interacts with about

bp of the DNA G-segment (Liu & Wang, a, b ; Klevan & Wang, ;

Kirkegaard & Wang, ; Morrison & Cozzarelli, ; Fisher et al. ; Rau

et al. ; Orphanides & Maxwell, ), in the absence of experimental data it

appeared that separating two interacting protein domains would be no more

difficult than dissociating a long stretch of DNA from the protein surface.

The first attempt to distinguish the two models was made more recently

(Roca & Wang, ). A series of experiments were carried out to examine the

unlinking of two singly linked DNA rings by yeast DNA topoisomerase II upon

A simple molecular machine

G

N N

1

2

N N

G

T

T

N N 3

ATP

T

G

N N

4

T

G

N N

5ADP + Pi

Fig. . Reaction steps in the two-gate model of ATP-dependent DNA transport by a type II

DNA topoisomerase. An enzyme molecule in its open-clamp conformation can bind a G-

segment, and the jaws labelled N in the drawings can be in the open or closed state

depending on the absence or presence of enzyme-bound ATP. A T-segment can enter the

enzyme when it is in the open-clamp conformation. The closure of the enzyme clamp upon

ATP-binding traps the T-segment and forces it to first go through the DNA gate and then

an exit gate on the opposite side of the entrance gate (the N-gate) formed by the pair of jaws

labelled N.

the addition of ADPNP. Because the non-hydrolysable nucleotide can trigger the

closure of the N-gate but does not allow its re-opening, the one-gate model would

predict that both of the unlinked DNA rings would be topologically linked with

the annular enzyme. The two-gate model, on the other hand, would predict that

only the DNA ring containing the G-segment would be bound in such fashion, as

the ring containing the T-segment would be expelled from the enzyme through

the exit gate. The results of the experiments are in agreement with the two-gate

model : decatenation of the singly linked rings was found to occur readily upon the

addition of ADPNP, and analysis of the products showed that the ring in which

the T-segment resided was free in solution and the ring in which the G-segment

resided remained topologically linked to the annular enzyme (Roca & Wang,

). Incorporation of this finding into the ATP-modulated protein clamp model

of type II DNA topoisomerases led to the model depicted in Fig. .

. Three-dimensional structures of type II DNA topoisomerase fragments

The structures of several type II DNA topoisomerase fragments have been solved

by X-ray crystallography in recent years (Wigley et al. ; Berger et al. ;

Lewis et al. ; Morais Cabral et al. ). Fig. depicts a ± AI structure of

J. C. Wang

a -kDa N-terminal fragment of the B-subunit of E. coli DNA gyrase (Wigley et

al. ). This fragment, comprised of amino acid residues –, contains the

ATPase domain of the enzyme. The crystal was obtained in the presence of

ADPNP, and there is one bond ADPNP molecule in each half of the dimeric

protein molecule; therefore the structure is likely to resemble the ATP-bound

form of the protein fragment. Two lysyl side-chains lining the nucleotide-binding

pocket, Lys- and Lys-, were previously identified to form covalent links

with an ATP affinity analogue pyridoxol «-diphospho-«-adenosine (Tamura &

Gellert, ).

Several studies showed that the -kDa E. coli GyrB fragment is monomeric in

solution but dimerizes upon binding of ADPNP, but not ADP (Ali et al. ,

). Measurements of ATP hydrolysis by E. coli DNA gyrase (Maxwell et al.

) and yeast DNA topoisomerase II (Lindsley & Wang, a), as well as

measurements of ADPNP-binding by DNA gyrase (Tamura et al. ), also

suggest a cooperative interaction between the two nucleotide binding sites. Taken

together, these results indicate that the binding of an ATP to one of the ATPase

catalytic pockets of a type II DNA topoisomerase triggers a conformation change

of the ATPase domain, which in turn leads to intramolecular dimerization of the

domains and facilitates the binding of a second ATP. An important conclusion

that can be drawn from the crystallographic and biochemical experiments is that

ATP hydrolysis can not occur without dimerization of the ATPase domains of the

enzyme (Wigley et al. ; Tamura et al. ; Ali et al. ). This point will

be elaborated upon in a later section on the coupling between ATP usage and

DNA transport by the type II DNA topoisomerases.

Because ADPNP binding is known to trigger the closure of the enzyme clamp

in the case of yeast DNA topoisomerase II, from the above discussion it can be

deduced that the pair of ATPase domains in a type II DNA topoisomerase

constitute the entrance gate; this gate has been termed the N-gate (Figs and )

because in a single-polypeptide eukaryotic enzyme the ATPase domain is close to

the N-terminus.

In the structure shown in Fig. (Wigley et al. ), an N-terminal arm (amino

acid residues –) of the DNA gyrase B-subunit extends from the ATPase

domain surface to form dimer contacts with the other monomer. The C-terminus

proximal helix (residues –) also extends from the core of the C-terminal

subdomain of the structure to contact the symmetry-related helix of the other

monomer. This C-terminal interface is likely to represent a crystallographic rather

than native protein contact, however.

Fig. depicts a ± AI crystal structure of a -kDa fragment of yeast DNA

topoisomerase II spanning residues – (Berger et al. ). This fragment

lacks the ATPase subfragment as well as amino acid residues at the C-

terminus of the intact enzyme, but it is capable of covalent adduct formation with

double-stranded DNA. From the alignment shown in Fig. , the –

fragment can be subdivided into two parts, termed the B« subfragment (residues

to about ), which corresponds to the C-terminal half of the gyrase B-

subunit, and the A« subfragment, which corresponds to the gyrase A-subunit

A simple molecular machine

lacking the -residue C-terminal fragment. In the cases of E. coli DNA gyrase

and yeast and Drosophila DNA topoisomerase II, this missing C-terminal

segment is known to be dispensable for catalysis of DNA transport (Reece &

Maxwell, a ; Shiozaki & Yanagida, ; Crenshaw & Hsieh, ; Caron et

al. ; Kampranis & Maxwell, ). The dispensability of the C-terminal

segment is also consistent with its absence in phage T and T DNA

topoisomerase (see Fig. ). Therefore, in combination, the structures of the gyrase

ATPase subfragment and the yeast DNA topoisomerase II B«A« fragment provide

a D sketch of a functional type II DNA topoisomerase.

In the heart-shaped (B«A«)#structure of the yeast enzyme fragment, the two B«

subfragments contact each other to form an arch, which caps the V-shaped A«–A«dimer to enclose a large hole AI wide at its base. The two A« polypeptides

contact at their C-terminal region to form the A«–A« dimer interface. This

interface represents the major contact between the two B«A« halves and it buries

over AI # of surface. Three of the four known yeast top� mutations that result

in cold-sensitivity of the mutant cells were mapped to this interface (Thomas et

al. ). The B«–B« contacts are less extensive but appear significant (Berger et

al. ).

A pair of semicircular grooves formed by residues from both the A« and B«subfragments are present near the top of the V-shaped A«–A« dimer. These

grooves, with an approximately AI opening, are decorated with positive surface

charges and have been implicated in the binding of the G-segment (Berger et al.

). A double-stranded DNA segment with a -nucleotide «-overhang can be

modelled into each groove in the protein structure, with the single-stranded

nucleotides extending through a narrow tunnel and joining to the active-site

tyrosyl residue Tyr- (Fig. ). Thus the -kDa structure is believed to

correspond closely to the conformation of the enzyme after it has cleaved the G-

segment and pulled it apart (Berger et al. ). In the proposed model, a part of

the DNA-binding surface is formed by a domain comprised of residues –.

This domain has the same fold as the DNA binding domain of the E. coli

catabolite activator protein (CAP) and histone H, and has been termed the CAP

domain (Berger et al. ). Recent lysine footprinting data are consistent with the

assignment of the G-segment binding surface described above (Li & Wang, ).

Fig. illustrates a ± AI structure of a -kDa fragment (residues –) of E.

coli DNA GyrA protein (Morais Cabral et al. ). As expected from amino acid

sequence comparisons (Fig. ), the overall architecture of this fragment is very

similar to that of the A« subfragment of yeast DNA topoisomerase II (Fig. ).

There are significant differences between the relative positions of the various

domains within the E. coli and the yeast enzyme fragment, however. Whereas the

pair of CAP domains in the yeast enzyme are well-separated, they are rotated from

their positions in the yeast enzyme and brought into contact to form the ‘head

dimer interface’ in the gyrase structure shown in Fig. (Morais Cabral et al.

). The three long helices (α, α, and α) connecting the N-proximal core

and the primary dimer interface also adopt different conformations in the yeast

and E. coli fragment structures, which will be discussed in a later section.

J. C. Wang

. How are the various subfragments connected?

No high-resolution structure of a type II DNA topoisomerase capable of moving

one DNA through another is presently available. Electron microscopic

examination of E. coli gyrase and its subunits (Kirschhausen et al. ), human

DNA topoisomerase II (Schultz et al. ), and intact and fragments of yeast

DNA topoisomerase II (Benedetti et al. ) has provided low-resolution

information on the organization of the subdomains. Visualization of negatively

stained human enzyme by scanning transmission electron microscopy, for

example, showed a V-shaped tripartite structure with a dense globular domain,

about AI in diameter, flanked by two smaller spheres about – AI in diameter

(Schultz et al. ). The pair of smaller spheres at the opening of the V-shaped

molecule appeared to be connected to the larger globular body by linkers – AIin length that were barely visible in the micrographs, and the angle extended by

the smaller spheres was typically in the range of –°. A similar tripartite

structure was seen by transmission microscopy with metal-shadowed yeast DNA

topoisomerase II (Benedetti et al. ). Deletion of the N-terminal residues

of the yeast enzyme resulted in a large decrease in the average size of the smaller

spheres in the tripartite structures (Benedetti et al. ), indicating that these

spheres represent the N-terminal or the GyrB-halves of the dimeric enzyme.

Interestingly, with either the human or the yeast enzyme, preincubation of the

enzyme with ADPNP caused a major change in the appearance of the images; in

the majority of molecules, the two smaller spheres appeared to be in contact with

each other following incubation with the ATP analogue (Schultz et al. ;

Benedetti et al. ). Micrographs of the human enzyme showed a preponderance

of bipartite structures in the presence of ADPNP, with a larger globular structure

of about AI in diameter and a smaller one of about AI in diameter (Schultz et

al. ).

In combination, the X-ray crystallography and electron microscopy results

suggest that each polypeptide of eukaryotic DNA topoisomerase II is organized

into two modules A and B, corresponding approximately to the A- and B-subunits

of gyrase. The pair of A modules in a dimeric enzyme are normally in contact with

each other in the presence or absence of bound ATP. The B modules, on the other

hand, form strong contacts only upon binding of ADPNP (and by implication

ATP); in the absence of bound ATP, the pair of B modules often stay apart.

Remarkably, there seem to be few protein–protein contacts between the A and B

modules in the absence of DNA, as suggested by the separation of the two

modules in electron micrographs of the molecules (Schultz et al. ; Benedetti

et al. ), and by the results of protein footprinting experiments designed to test

whether any lysyl residue in the GyrA or GyrB half of yeast DNA topoisomerase

II might be protected against citraconylation by the presence of the other half of

the polypeptide (Li & Wang, ).

The structural studies, especially the crystal structures of the -kDa yeast

DNA topoisomerase II fragment and the -kDa E. coli DNA gyrase A subunit

fragment, also reveal a high degree of flexibility in the detailed positioning of the

A simple molecular machine

various domains of the type II DNA toposiomerases. This structural variability

is presumably related to the ability of a type II DNA topoisomerase to assume

very different conformations during its catalytic steps. At the same time, this

structural flexibility increases the difficulty of reconstructing the various

conformational states of a functional enzyme from the structures of its parts.

. A molecular model of DNA transport by a type II DNA topoisomerase

The structural data summarized above provided additional support of the model

depicted in Fig. and added much molecular details to the model. Fig.

illustrates a refined model combining the known biochemical and structural data

(Berger et al. ). The particular enzyme in this illustration is that of yeast DNA

topoisomerase II lacking the C-terminal amino acid residues that are dispensable

for its catalytic actions. It is most likely, however, that the basic features of the

model are shared by all type II DNA topoisomerases. In a later section, the

plausible difference between bacterial DNA gyrase and type II enzymes

exemplified by yeast DNA topoisomerase II will be discussed.

The structure denoted by in Fig. is a representation of the free enzyme in

an open-clamp conformation. The purple (A«)#structure in is the same as that

seen in the -kDa crystal structure except that the pair of B« subfragments with

the ATPase subfragments tagged on, are not in contact. Such an open

conformation with detached B« subfragments is presumably one of the major

conformations of the free enzyme, and this open conformation is necessary for the

binding of the DNA G-segment (shown as a rod above the enzyme).

Upon binding of the DNA G-segment, there is a significant change in the

conformation of the enzyme (). The two enzyme halves move toward each other

by a large distance, estimated to be – AI (Berger et al. ), and the pair of

active-site tyrosyl residues are now positioned near the scissile phosphates for

nucleophilic attack. The structure of the purple (A«)#

dimer depicted in

resembles more the one shown in Fig. for the A« dimer of E. coli gyrase (Morais

Cabral et al. ) than the one shown in Fig. for the -kDa fragment of the

yeast enzyme. The precise positions of the B« subfragments in are uncertain. It

has been suggested (Berger et al., ) that these subfragments may be oriented

very differently from their positions in the -kDa fragment structure of Berger

et al. (). The pair of ATPase subfragments in can combine or separate

depending on the state of ATP binding, and a T-segment can enter the enzyme

clamp when these jaws are apart. Closure of the jaws upon the binding of ATP

would then trap the T-segment and initiate a conformational cascade in the

enzyme-DNA complex: the DNA G-segment is transiently broken and splayed

apart, and the T-segment is forced through the DNA gate into the large cavity

below (illustrations to in Fig. ).

The conformation of the (B«A«)#part of the enzyme shown in of Fig. is that

seen in the crystal structure of the -kDa fragment of the yeast enzyme. When

the two enzyme halves retract toward each other following the passage of the T-

segment through the DNA gate, the size of the cavity containing the T-segment

J. C. Wang

is reduced and the T-segment exits the enzyme through the A«–A« dimer interface

(illustration of Fig. ). The A«–A« interface has been termed the C-gate because

its constituents are close to the carboxyl termini of the A« polypeptides. The open

state of the C-gate is probably a transient one. Following the exit of the T-segment

and closure of the C-gate, the N-gate can re-open after ATP hydrolysis and the

release of the hydrolytic products ( in Fig. ), and the enzyme is posed for

another cycle of DNA transport.

The availability of detailed structural information had also made it possible to

test more directly a key postulate of the two-gate protein clamp model, namely the

role of the C-gate for the exit of the T-segment (Roca et al. ). Based on the

crystal structure of the -kDa fragment of the yeast enzyme, site-directed

mutagenesis was carried out to replace Lys- and Asn- by cysteines. In

the mutant enzyme, a pair of disulphide bonds can form across the A«–A« dimer

interface to lock the C-gate. When the single-step decatenation experiment of

Roca & Wang () was repeated with the mutant enzyme, with its C-gate locked

by the formation of disulphide bonds across the A«–A« interface, the unlinked

DNA rings following one round of DNA transport were found to remain

topologically linked to the enzyme–ADPNP complex (Roca et al. ).

Furthermore, exit of the DNA ring containing the T-segment was observed upon

subsequent addition of -mercaptoethanol, which reduces the disulphide bonds

and hence unlocks the C-gate, to the reaction products (Roca et al. ). These

and other data, while providing strong evidence in support of the two-gate

mechanism, do not preclude the possibility that under certain conditions (for

example when the G-segment can dissociate readily from the enzyme) a type II

DNA topoisomerase could promote DNA transport by a one-gate mechanism

(Lindsley, ). A special one-gate mechanism to account for the ATP-

independent relaxation of negative supercoiled DNAs by bacterial DNA gyrase

will be discussed in Section ...

.

A type II DNA topoisomerase can be viewed as a molecular machine that utilizes

ATP as the fuel in the transport of one DNA double helix through another. From

a thermodynamic point of view, only DNA supercoiling by bacterial DNA gyrase,

and not the removal of supercoils by the other type II DNA topoisomerases, is

expected to require a high-energy cofactor like ATP. It is clear, however, that

kinetically the transport of DNA by a type II DNA topoisomerase is strongly

dependent on ATP. Removal of DNA supercoils by yeast DNA topoisomerase II

is undetectable in the absence of ATP, and mutating a critical ATPase domain

residue Gly- of the enzyme to alanine also abolishes its DNA relaxation

activity in the presence of ATP (Linsley & Wang, a).

A low level of DNA transport activity in the absence of ATP is present,

however, in several type II DNA topoisomerases including E. coli DNA gyrase

(Gellert et al. , ; Sugino et al. ; Higgins et al. ; Kreuzer &

A simple molecular machine

Cozzarelli, ; Mizuuchi et al. ; Krasnow & Cozzarelli, ; Marians,

), phage T DNA topoisomerase (Liu et al. , ) and Drosophila DNA

topoisomerase II (Osheroff et al., ). A derivative of gyrase lacking the ATPase

domain was also found to relax supercoiled DNA in the presence or absence of

ATP (Brown et al. ; Gellert et al. ). In the sections below, the available

information on how a type II DNA topoisomerase couples ATP usage to DNA

transport is summarized. Attempts are made, at the risk of oversimplification, to

interpret this intricate coupling in terms of the known or predicted structural

features of the various enzyme–DNA complexes. The low level ATP-independent

DNA transport activity of DNA gyrase will be discussed in a separate section.

. ATP utilization

Because the binding of the non-hydrolysable ATP analogue ADPNP to a type II

topoisomerase can effect one round of DNA transport, it is generally assumed that

DNA transport by the enzyme is driven by ATP binding, and ATP hydrolysis and

product release are only necessary for enzyme turnover (Sugino et al. ;

Peebles et al. ). It is plausible, however, that ATP hydrolysis may further

accelerate the coupled process. An example is provided by the recent study of the

roles of nucleotide cofactors in the reaction steps catalysed by the elongation factor

G, a GTPase involved in the transloction of bacterial ribosomes along messenger

RNA (Rodnina et al. ). It is known that non-hydrolysable GTP analogues can

effect one round of translocation. With GTP itself, however, rapid hydrolysis of

the nucleotide was found to precede translocation. These observations suggest

that although the binding of a non-hydrolysable GTP analogue can effect

translocation, GTP hydrolysis may further accelerate the rearrangements of the

ribosome that drive translocation (Rodnina et al. ).

The intricacy of ATP utilization in the transport of DNA by the type II DNA

topoisomerases is illustrated by the recent pre-steady state kinetic analysis of ATP

hydrolysis by yeast DNA topoisomerase II (Harkins & Lindsley, a, b). As

expected, the DNA-bound enzyme consumes two ATP per enzyme turnover

event. These bound ATP molecules are hydrolysed sequentially, however. Rapid

hydrolysis of one ATP and the release of the hydrolytic products appear to occur

before the second ATP is hydrolysed. Deducing the structural changes

accompanying this stepwise hydrolysis of bound ATP should provide further

insight on the intricate coupling of ATP usage to DNA transport.

. How tight is the coupling between DNA transport and ATP binding and

hydrolysis?

.. Dependence of the coupling efficiency on the degree of supercoiling of the

DNA substrate

Measurements with yeast DNA topoisomerase II indicate that the coupling

efficiency can be very high under optimal conditions. Quantitative analysis of

linking number changes of DNA rings showed that for a moderately supercoiled

J. C. Wang

DNA, % of all bound enzyme molecules are able to promote one round of

DNA transport upon the addition of excess ADPNP (Roca & Wang, ).

Because a fraction of the bound enzyme molecules was probably not active, the

actual coupling efficiency could be even higher and close to the theoretical limit of

%. The same study showed that with a completely relaxed DNA ring the

efficiency is considerably lower: about one out of four bound enzyme molecules

would complete a DNA transport event upon ADPNP addition.

Two sets of measurements of the coupling efficiency in the presence of ATP

itself rather than its non-hydrolysable analogue were also carried out with the

yeast enzyme (Lindsley & Wang, a). In one, parallel measurements of the

rate of ATP hydrolysis and the rate of removal of DNA supercoils were carried

out at a saturating ATP concentration of m. The reaction temperature was set

below ± °C so that the ATPase and DNA transport rates were slowed down

sufficiently to permit manual quenching of the reactions. From these

measurements it was calculated that about ±³ ATP were hydrolysed per DNA

transport event. When the same measurements were carried out at °C and

lower ATP concentrations, however, the number of ATP hydrolysed per DNA

transport event was measured to be as low as ±, with an average of ±³±.

These results suggest that at a high ATP concentration when ATP binding is

rapid, closure of the enzyme clamp often occurs without trapping a T-segment;

at a low ATP concentration, at least in the case of a supercoiled DNA the rates of

ATP binding and the closure of the enzyme clamp are both slow relative to the

rate of entrance of the T-segment into an enzyme in its open-clamp conformation,

leading to a very high coupling efficiency.

In terms of the number of ATP required for one round of DNA transport by

a type II DNA topoisomerase, the theoretical limit is probably close to one ATP

per DNA transport event at very low ATP concentrations: the binding of an ATP

to one half of the dimeric enzyme could be sufficient in triggering dimerization of

the N-terminal domain and the conformational cascade that follows, leading to the

transport of a T-segment through the enzyme-bound G-segment. Studies with a

heterodimeric yeast enzyme with one wild-type polypeptide and one mutant

polypeptide with a mutation in the ATPase domain suggest that the binding of a

single ATP to a dimeric enzyme is sufficient to trigger the closure of the enzyme

clamp (Lindsley & Wang, b). At high ATP concentrations, however, the

binding of ATP to wild-type yeast DNA topoisomerase II is cooperative, and the

binding of an ATP to one protomer of the dimeric enzyme is likely to be followed

rapidly by the closure of the N-gate and the binding of a second ATP to the other

protomer (Lindsley & Wang, a). Therefore at high ATP concentrations the

theoretical limit of the coupling efficiency is expected to be two ATP for each

DNA transport event.

For E. coli DNA gyrase, the apparent coupling efficiency was also found to be

strongly dependent on DNA topology. With a positively supercoiled DNA, the

coupling efficiency approaches %, and a linking number reduction of two per

(BA)#

molecule was observed upon the addition of ADPNP (Bates et al. ).

The coupling efficiency decreases steadily with DNA substrates of increasing

A simple molecular machine

degrees of negative supercoiling, and the ADPNP-mediated reduction of Lk

becomes undetectable for a moderately negatively supercoiled DNA with a

specific linking difference of ®± [Bates et al. ; the specific linking

difference is defined as the fractional deviation of Lk from Lk° or (Lk®Lk°)}Lk°].Earlier measurements also showed that the addition of the non-hydrolysable

nucleotide to E. coli DNA gyrase bound to a relaxed DNA reduces Lk by about

± per A protomer or ± per dimeric enzyme molecule (Sugino et al. ),

corresponding to an efficiency of about %. With ATP as the cofactor, earlier

measurements of negative supercoiling of a relaxed DNA by gyrase suggested that

the apparent coupling efficiency is high during the early phase of the reaction, but

drops to zero when the specific linking difference of the DNA approaches ®±

(reviewed in Reece & Maxwell, b).

The results summarised above show that in terms of the relaxation of a

positively supercoiled DNA, the coupling efficiency for bacterial DNA gyrase is

not very different from that for yeast DNA topoisomerase II. Because the yeast

enzyme does not catalyse DNA negative supercoiling, it is not possible to compare

the efficiencies of the two enzymes in this reaction. There are two complications

in the interpretation of the coupling efficiencies of DNA negative supercoiling by

gyrase. First, in nearly all measurements with the bacterial enzyme, only the net

change in the average linking number of a DNA ring was examined. For a DNA

gyrase bound to a negatively supercoiled DNA, the formation of a positive-noded

configuration between the G-segment and the incoming T-segment is favoured by

the DNA-enzyme binding energy, but disfavoured by the free energy of DNA

supercoiling (see Section ..). Therefore the number of DNA transport events

can be equated to one half of the net decrease in Lk only if the DNA–enzyme

binding energy is the dominant term, otherwise ATP-triggered inversion of both

positive and negative nodes may occur in a population of negatively supercoiled

DNA molecules. Second, DNA gyrase is known to possess an ATP-independent

DNA relaxation activity that is specific for negatively supercoiled DNA (Sugino

et al. ; Gellert et al. , ; Higgins et al. ). This uncoupled

pathway may significantly reduce the apparent coupling efficiency, especially if a

highly negatively supercoiled DNA substrate is used. In experiments on the

coupling efficiency of ADPNP triggered DNA transport in a positively supercoiled

DNA, these complications are largely absent (the first complication can also be

eliminated entirely by the use of a DNA substrate consisting of a single linking

number isomer, so that the inversion of both positive and negative nodes can be

measured directly from an analysis of the distribution of the linking numbers of

the products; see Roca & Wang, ).

.. Coupling efficiency and DNA binding

In terms of the two-gate protein clamp model, the above data can be understood

by assuming that normally the binding of ADPNP or ATP to an enzyme bound

to a G-segment always leads to the closure of the N-gate and the sequential

transport of a T-segment, if one is present inside the enzyme, through the DNA

gate and the protein exit gate. This interpretation is supported by two

J. C. Wang

observations. First, as shown by the single-step decatenation experiments of Roca

& Wang (), the DNA ring containing the T-segment always exits the enzyme

clamp following its passage through the DNA gate. Second, through the use of a

mutant enzyme YF, which is incapable of opening the DNA gate owing to the

replacement of the pair of active site tyrosyl residues by phenylalanines, it was

found that the binding of ADPNP to a G-segment bound mutant enzyme can

trigger the closure of the N-gate, but a second DNA ring can not be captured

(J. Roca, W. Li, and J. Wang, unpublished). Together, these two observations

demonstrate that the probability of capturing a T-segment without its being

transported through the DNA gate and the exit gate of the enzyme is low.

If the binding of ADPNP to a G-segment-bound enzyme is always followed by

the transport of a T-segment that has entered the enzyme clamp, then the

dependence of the overall coupling efficiency on the topology of the DNA

substrate must reflect differences in the probability of finding a T-segment within

an open enzyme clamp with DNA substrates supercoiled to different extents. For

type II DNA topoisomerases resembling yeast DNA topoisomerase II, the

coupling efficiencies measured in experiments using ADPNP indicate that for a

supercoiled DNA substrate the probability of finding a T-segment within an open

enzyme clamp is very high, and the same is true for bacterial DNA gyrase bound

to a positively supercoiled DNA substrate. The situation with DNA gyrase bound

to a negatively supercoiled DNA is probably rather different, however, as

discussed above. In experiments using ATP as the cofactor, the coupling

efficiencies are likely to be determined by the relative rates of T-segment entrance

and clamp closure, rather than by the probability of finding a T-segment in an

equilibrium population of enzyme molecules.

For yeast DNA topoisomerase II bound to a relaxed DNA, the measured

coupling efficiency of DNA transport upon addition of ADPNP is around %

(Roca & Wang, ). This result suggests that the binding of a T-segment to the

enzyme in its open-clamp conformation is rather weak under the experimental

conditions (in m Tris HCl, pH , m EDTA, m KCl, m MgCl#

and m -mercaptoethanol, at °C). The T-segment presumably goes in and

out of the enzyme clamp in the absence of the nucleotide, and is not in contact with

the enzyme most of the time.

Stable binding of eukaryotic DNA topoisomerase II to DNA crossovers has

been implicated by results obtained by electron microscopy (Zechiedrich &

Osheroff, ), by the formation of DNA knots (Hsieh, ; Wassermann &

Cozzarelli, ; Roca et al. ), and by experiments measuring the residual

number of supercoils upon relaxation of a supercoiled DNA in the presence of

varying amounts of yeast DNA topoisomerase II (Roca et al. ; relaxation was

carried out with a type IB DNA topoisomerase in the absence of ATP so that the

DNA transport activity of the type II enzyme was not manifested). It is uncertain

in these measurements, however, whether a DNA crossover bound by an enzyme

molecule represents a pair of G- and T-segments. Under conditions that favour

crossover binding, the efficiency of DNA transport upon the addition of ADPNP

is low (Roca et al. ). Furthermore, in the ADPNP-triggered decatenation of

A simple molecular machine

a supercoiled DNA ring singly linked to a relaxed DNA ring, yeast DNA

topoisomerase II was found to preferentially bind to the supercoiled ring, which

is consistent with the binding of the enzyme to a DNA crossover; this one-step

DNA transport event was found to efficiently unlink the pair of rings, however,

indicating that the T-segment is from the relaxed DNA ring rather than a

crossover in the supercoiled ring (Roca & Wang, ). The plausible binding of

type II DNA topoisomerases to crossovers that do not include a T-segment was

also invoked to account for a remarkable property of these enzymes that was

revealed by recent experiments (Rybenkov et al. ). It was observed that in the

presence of ATP, a variety of the type II DNA topoisomerases can reduce the

steady state fraction of knotted or catenated DNA molecules to levels up to two

orders of magnitude below those at thermodynamic equilibrium.

. DNA relaxation by a type II DNA topoisomerase: how is the high efficiency

of coupling achieved?

In essence, the available data for the type II DNA topoisomerases suggest that in

the relaxation of a positively supercoiled DNA by either yeast DNA topoisomerase

II or E. coli DNA gyrase, the overall high efficiency of coupling of DNA transport

to ATP usage is achieved by the high efficiencies of two consecutive reactions.

First, the binding of two ATP molecules to a DNA-bound dimeric enzyme at a

high ATP concentration almost always leads to the closure of the N-gate. Second,

provided that a T-segment has already entered the enzyme clamp, the closure of

the N-gate almost always leads to the transport of the trapped T-segment, first

through the DNA gate and then the exit gate. The details of these and additional

factors involved in the coupling of ATP usage to DNA transport are discussed in

the sections below.

.. ATP hydrolysis and the closure of the enzyme clamp

A key step in the ATP-triggered conformational cascade of a type II DNA

topoisomerase is the dimerization of the ATPase domains of the enzyme following

ATP binding. Therefore a high efficiency of coupling requires that dimerization

must precede ATP hydrolysis. This is nicely accomplished by the participation of

amino acid residues of both protomers in each of the ATPase site of the dimeric

enzyme: ATP hydrolysis can not occur unless the N-gate is closed to form the

ATPase catalytic pocket (Wigley et al. ; Tamura et al. ; Ali et al. ;

O’Dea et al. ).

.. Coupling of ATPase activity to DNA binding

In the presence of a high concentration of DNA, the ATPase activity of yeast

DNA topoisomerase II is stimulated by about -fold (Lindsley & Wang, a).

For the residue N-terminal fragment of the yeast enzyme, the ATPase activity

is only marginally stimulated, by about %, by the presence of excess DNA (S.

Olland and J. Wang, unpublished). Similarly, for purified E. coli DNA GyrB

J. C. Wang

protein, the ATPase activity is not significantly stimulated by DNA in the absence

of the GyrA protein (Sugino et al. ; Sugino & Cozzarelli, ; Staudenbauer

& Orr, ; Maxwell & Gellert, ). Because the binding of a DNA G-

segment is likely to involve both the GyrA and GyrB regions of a type II DNA

topoisomerase (see Section .), these results are consistent with the notion that

stimulation of the ATPase activity of a type II DNA topoisomerase by DNA is

largely through the binding of the enzyme to the G-segment. The binding of the

two halves of a dimeric enzyme molecule to a contiguous DNA segment may

facilitate dimerization of the N-gate jaws to form the active form for ATP

hydrolysis ; allosteric changes within the ATPase domain, triggered by the

binding of a dimeric enzyme to DNA, may further enhance the ATPase activity

(see the section below).

Whether the DNA dependence of the ATPase activity involves interaction

between the enzyme and the T-segment as well is less clear. Studies with E. coli

DNA gyrase showed that the stimulation of its ATPase activity by DNA is

dependent on the length of the DNA; DNA shorter than bp can stimulate the

ATPase only at a high concentration, suggesting that DNA must bind to two or

more sites of the enzyme to have an effect on the ATPase activity (Maxwell &

Gellert, , ). These data can not distinguish, however, between a model

in which multiple sites are involved in the binding of the G-segment, and one in

which the multiple sites represent both G- and T-segment binding sites (Maxwell

& Gellert, , ).

Following the solution of the three-dimensional structure of the -kDa E. coli

GyrB fragment (Wigley et al. ), site-directed mutagenesis within this

fragment was carried out to replace Arg- by a glutamine (Tingey & Maxwell,

). Arg- is situated at a constriction of the channel formed by the closing

of the N-gate (Fig. ), and was postulated to be a part of a binding surface for the

T-segment (Wigley et al. ). Interestingly, the RQ mutant enzyme retains

the intrinsic ATPase activity of the wild-type enzyme but the activity is not

stimulated by DNA (Tingey & Maxwell, ). These observations led to the

suggestion that the binding of the T-segment within the enzyme clamp may be

required for stimulation of the ATPase activity. The effect of the RQ mutation

on the DNA dependence of the ATPase activity could also be attributed to,

however, an allosteric change within the ATPase clamp of the enzyme when a wild

type but not an RQ mutant enzyme binds to a G-segment (see Section ..).

It might seem ideal for a very high efficiency of coupling between ATP usage

and DNA transport if ATP hydrolysis requires the entrance of a T-segment into

a G-segment bound enzyme. This is not so. As described earlier, it appears that

at a high concentration of ATP a G-segment-bound enzyme clamp may often

close without capturing a T-segment. In such a situation, if the binding of a

T-segment is indeed a prerequisite for ATP hydrolysis, an enzyme bound to a

G-segment would get stuck in the closed clamp conformation whenever it fails to

capture a T-segment upon closure of the N-gate: it would have to rely on its

DNA-independent ATPase activity to free itself from the closed-clamp

conformation.

A simple molecular machine

.. Structural changes in the enzyme upon closure of the N-gate

Following ATP binding and the closure of the N-gate, a conformational cascade

ensues to effect the transport of the T-segment through the G-segment. Several

structural changes in the enzyme have been detected upon the binding of ADPNP,

the most conspicuous of which being the switch of an SV endoproteinase

cleavage site in yeast DNA topoisomerase II (Lindsley & Wang, ). In the

absence of ADPNP, this proteinase cleaves the yeast enzyme after Glu- (site

A in Fig. ) and around residue (site C in Fig. ) ; in the presence of ADPNP,

cleavage occurs after Glu- (site B in Fig. ) and site C. Limited proteolysis of

the -kDa N-terminal E. coli gyrase B-subunit also showed that in the presence

of ADPNP, cleavage by trypsin occurs after Lys-, but the -kDa N-terminal

product is resistant to further cleavage that would occur in the absence of the

nucleotide (Ali et al. ).

Probing the chemical reactivity of individual yeast DNA topoisomerase II lysyl

residues showed that ADPNP binding affects lysine citraconylation at at least six

sites. Three of these sites, around Lys-}, }, and }, showed a

reduced level of citraconylation in the presence of ADPNP (Li & Wang, ).

Based on the crystal structure of the ATPase domain of E. coli gyrase and amino

acid sequence alignment of type II DNA topoisomerases, it was suggested that the

Lys-} and } sites are probably contacted directly by a bound

ADPNP, and reactivity at the Lys-} site is probably affected by ADPNP

because of its location at the N-gate dimerization interface. Citraconylation at

Lys-, , and , on the other hand, was found to be enhanced by ADPNP

binding, suggesting that these lysyl residues become more exposed to the solvent

through allosteric conformational changes induced by the binding of the

nucleotide to the ATPase catalytic sites. ADPNP binding also affects reactivity of

one or more lysyl residues around positions and in a less direct way: these

positions are protected against citraconylation by DNA binding, and this DNA-

mediated protection appears to be reduced in the presence of the non-hydrolysable

ATP analogue (Li & Wang, ).

.. ATP binding}hydrolysis and DNA cleavage

There are conflicting hints on the relation between the binding and hydrolysis of

ATP by a type II DNA topoisomerase and the cleavage and rejoining of the G-

segment. Two lines of evidence suggest that ATP utilization affects transient

cleavage of DNA mediated by type II DNA topoisomerases and vice versa. First,

covalent adduct formation between DNA and type II DNA topoisomerases, as

revealed by the addition of a protein denaturant to the DNA–enzyme complexes,

is often enhanced by the presence of ATP or ADPNP (Sugino et al. ; Peebles

et al. ; Fisher et al. ; Sander & Hsieh, ; Kreuzer & Alberts, ;

Osheroff, ). Second, when the active site tyrosine Tyr- of yeast DNA

topoisomerase II is mutated to phenylalanine, and thus abolishing the DNA

cleavage activity of the enzyme, the DNA-dependent ATPase activity of the

enzyme was found to be largely abolished (W. Li, J. Roca, and J. Wang,

J. C. Wang

unpublished; J. Lindsley, personal communication; S. Gasser, personal

communication; A. Maxwell, personal communication). In contrast, there have

also been ample examples that the type II DNA topoisomerases and their

truncation derivatives can cleave DNA in the absence of ATP (see for examples,

Sander & Hsieh, ; Osheroff, ; Reece & Maxwell, a). Furthermore,

enzymes lacking the ATPase activity because of the presence of point mutations

or deletions were nevertheless found to have DNA cleavage activities similar to

that of the wild-type enzyme (Jackson & Maxwell, ; Lindsley & Wang,

a ; Berger et al. ; O’Dea et al. ; Tingey & Maxwell, ).

These seemingly conflicting observations are yet to be placed in the same

mechanistic framework. It is noteworthy, however, that the experimentally

measured efficiency of DNA cleavage by a type II DNA topoisomerase probably

represents a composite of contributions from different conformations of the

enzyme. The effect of ATP or ADPNP binding on DNA cleavage by the enzyme

is probably realized only if the enzyme is in the proper closed-clamp conformation.

In terms of the coupling between ATP usage and DNA transport, the effect of the

nucleotide cofactor on the cleavage of the G-segment by the enzyme in its closed-

clamp conformation is likely to be the important one, as the DNA transport event

presumably occurs only inside the enzyme in its closed-clamp conformation.

Judging from the large drop in the ATPase activity of yeast DNA topoisomerase

II upon mutating the active site tyrosyl residue Tyr- to phenylalanine, the

binding and}or hydrolysis of ATP is likely to have a significant effect on the

cleavage of the G-segment in the closed-clamp conformation of the enzyme.

.. Entrapment of the T-segment and opening of the DNA gate

Closely related to the question on the effects of ATP binding and hydrolysis on the

cleavage of the G-segment is whether the cleavage reaction involves the T-

segment. Studies with a bp DNA showed that the efficiency of Drosophila

DNA topoisomerase II-mediated cleavage of this short DNA has a sigmoidal

dependence on the concentration of the DNA (Corbett et al. ). Based on this

observation, the authors suggested that the enzyme might interact with more than

one -mer in its cleavage of one of them, and that the presence of a T-segment

might be required for the cleavage of a G-segment (Corbett et al. ).

Studies of DNA cleavage by the closed clamp form of yeast DNA topoisomerase

II show however, that the transient breakage of the G-segment does not

require the entrapment of a T-segment (Roca & Wang, ). In one experiment,

yeast DNA topoisomerase II was first converted to the closed clamp conformation

by the addition of ADPNP, and a linear DNA was allowed to thread through the

enzyme annulus to provide a G-segment. When etoposide and SDS were

sequentially added to the enzyme–DNA mixture, it was found that the G-segment

was efficiently cleaved by the enzyme. Because only one DNA double helix can be

threaded through an enzyme in its closed-clamp conformation, the finding of this

experiment demonstrates that the entrapment of a T-segment is not required for

the transient cleavage of the G-segment (Roca & Wang, ).

The opening of the DNA gate by a type II DNA topoisomerase, however,

A simple molecular machine

requires not only breakage of the DNA strands through transesterification

between the enzyme and DNA, but also the movement of the enzyme-linked

DNA ends away from each other to permit the passage of the T-segment. These

two steps are akin to unlocking a gate and opening it. Whereas the T-segment is

not required for unlocking the DNA gate, its entrapment may be strongly coupled

to opening it. From the crystal structure of the ATPase subfragment of bacterial

gyrase B-subunit, the dimension of the archway formed by the dimerized

fragment is not sufficiently large to accommodate a duplex DNA (Tingey &

Maxwell, ). Therefore steric repulsion is likely to accompany the trapping of

a T-segment, and this repulsion may force the widening of the DNA gate (Berger

et al. ).

In one experiment with the YF mutant of yeast DNA topoisomerase II, it

was found that closure of the N-gate of a G-segment-bound mutant enzyme

would occur normally upon the addition of ADPNP, but this closure could not

trap a second DNA ring present at a high concentration (J. Roca, W. Li, and J.

Wang, unpublished). This finding suggests that the widening of the DNA gate

may be coupled to the trapping of the T-segment. For the mutant enzyme,

unlocking and opening the DNA gate are not possible, and hence the G-segment-

bound enzyme can not assume the closed clamp conformation even if a T-segment

has entered the enzyme clamp. In a functional enzyme, on the other hand,

widening the DNA gate upon entrapment of a T-segment allows the T-segment

to move through the DNA gate into the large cavity on the other side of the G-

segment. The positive surface charges lining this cavity is likely to provide further

attraction for the T-segment to move into it.

.. Closure of the DNA gate and exit of the T-segment

Once the T-segment has passed the G-segment and entered the large hole lined

with positive surface charges (structure in Fig. ), it would seem to be in a

favourable micro-environment. The question is then what drives its subsequent

exit?

The answer to the above question is likely to lie in the energetics of the enzyme-

DNA complex in which the G-segment is cleaved and the DNA gate is wide open.

This state is most likely thermodynamically unstable; the enzyme–DNA complex

is probably driven into this state only when the T-segment is trapped in the closed

N-terminal jaws of the protein. Therefore once the T-segment has passed through

the open DNA gene into the large hole, the DNA–enzyme complex would resume

its more stable conformation by closing the DNA gate. The movement of the two

halves of the enzyme–DNA complex toward each other would reduce the size of

the large hole, which could in turn force the exit of the T-segment (Berger et al.

).

From the above discussion, it is clear that the stepwise movement of a T-

segment through the multiple ports and locks of the canal in a G-segment-bound

enzyme requires the existence of energetically well-balanced conformational states

of the various enzyme–DNA complexes. The existence of these conformational

states is in turn dependent on the presence of intricately balanced structural

J. C. Wang

elements or hinges in the enzyme–DNA complexes. For example, based on the

crystal structures of the -kDa yeast DNA topoisomerase II fragment and the

-kDa bacterial gyrase fragment, it was suggested that conformational changes at

a pair of hinge structures in a type II DNA topoisomerase may lead to the opening

of the C-gate for the exit of the T-segment, as illustrated in Fig. (Morais Cabral

et al. ).

. Directionality of DNA transport: why is bacterial gyrase unique?

Among the type II DNA topoisomerases, bacterial DNA gyrase is unique in two

respects. First, whereas all type II DNA topoisomerases are ATP dependent, only

DNA gyrase is capable of catalysing the negative supercoiling of DNA. Second,

DNA gyrase and a derivative of it lacking the N-terminal half of the B-subunit

possess an ATP-independent DNA relaxation activity, which differs from the

ATP-dependent DNA transport activities of the other type II enzymes in that it

is specific for negatively supercoiled DNA (Sugino et al. ; Gellert et al. ,

; Higgins et al. ). [An early report also ascribed to bacterial DNA gyrase

an ADPNP-dependent activity capable of removing positive supercoils

catalytically, but recent measurements showed that positive supercoil removal in

the presence of the non-hydrolysable analogue is stoichiometric and amounts to a

maximum of two supercoils per gyrase molecule (Bates et al. ).]

... Structural basis of the ability of bacterial DNA gyrase to catalyse the

ATP-dependent negative supercoiling of DNA

What might be the structural features that set gyrase apart from the other

members of the type II subfamily of DNA topoisomerases? It was suggested

many years ago that the ability of DNA gyrase to catalyse DNA supercoiling is a

consequence of the unique way DNA gyrase binds to DNA (Wang, ). Only

in the case of DNA gyrase the G-segment is wrapped around the enzyme with a

right-handed writhe, and about bp of the segment are in contact with the

enzyme (Liu & Wang, a, b ; Klevan & Wang, ; Fisher et al. ;

Kirkegaard & Wang, ; Morrison & Cozzarelli, ; Rau et al. ;

Orphanides & Maxwell, ) ; for the other type II DNA topoisomerases, a much

shorter region of the G-segment (about – bp) is contacting the enzyme

without a significant writhe of the segment (Spitzer & Muller, ; Lee et al.

; Thomsen et al. ).

How can the wrapping of a DNA segment around a type II DNA toposiomerase

affect the outcome of the enzyme-mediated topological transformation of a DNA

ring? When a T-segment enters the interior of an enzyme bound to a G-segment

on the same DNA ring, the two DNA segments can form a crossover with either

a plus sign or a minus sign, as illustrated in Fig. . The enzyme-mediated

transport of the T-segment through the G-segment always inverts this nodal sign.

Therefore, in order for a type II DNA topoisomerase to catalyse the negative

supercoiling of a DNA ring, the T- and G-segment must form a plus node before

the DNA transport event; the linking number of the DNA ring is then reduced

A simple molecular machine

(d)

(b)

(a)

(c)

G T

–

+

+

–

Fig. . The sign of the node between a pair of G and T-segments in a DNA ring before the

DNA transport event. The schematic in (a) illustrates a G-segment-bound enzyme (circle

with a cross inside) viewed down the molecular dyad from the N-gate to the C-gate. The T-

segment can enter the opened N-gate to form a positive node (b) or a negative node (c) with

the G-segment. If the enzyme does not impose a significant writhe in the G-segment, the

nodal sign between the incoming T-segment and the enzyme-bound G-segment must be

determined by the topology of the DNA ring: it would be positive in a positively

supercoiled DNA, negative in a negatively supercoiled one, and either positive or negative in

a relaxed DNA ring. In drawing (d ), the two crossing-segments are shown to be

perpendicular to each other; the plus and minus mode drawn are structurally equivalent. In

the case of bacterial DNA gyrase, the enzyme imposes a right-handed writhe in the G-

segment. The nodal sign is then influenced by two terms: enzyme–DNA interaction favours

a positive node, and DNA conformation favours a particular nodal sign depending on the

sense of supercoiling. Illustration taken from Roca & Wang ().

by two by the inversion of the plus node to a minus one. The wrapping of a DNA

segment around gyrase apparently imposes a strong bias in favour of the formation

of a plus node between the two segments. For the other type II DNA

topoisomerases, wrapping of the DNA around the enzyme is not involved, and the

nodal sign an incoming T-segment makes with the G-segment is specified by the

conformation of the DNA: the segments would preferentially assume a plus sign

J. C. Wang

in a positively supercoiled DNA and a minus sign in a negatively supercoiled

DNA (Roca & Wang, ). Thus for type II DNA topoisomerases other than

gyrase, enzyme-mediated transport of a T-segment through a G-segment always

reduces the net number of supercoils in the DNA.

Recent studies have provided strong evidence in support of the above

interpretation. A C-terminal -kDa fragment of E. coli DNA gyrase A-subunit

(residues –) was shown to be involved in the right-handed wrapping of

DNA around the enzyme: mutant gyrase lacking this -kDa domain does not

impose a right-handed writhe in the DNA segment bound to it (Kampranis &

Maxwell, ), but the purified -kDa protein does by itself (Reece & Maxwell,

a). In addition, deletion of this domain from DNA gyrase converts the

enzyme from a DNA-negative supercoiling activity to one that catalyses the ATP-

dependent removal of supercoils (Kampranis & Maxwell, ). Unlike DNA

gyrase but similar to the other type II DNA topoisomerases, this deletion mutant

of gyrase is also efficient in unlinking DNA catenanes in the presence of ATP

(Kampranis & Maxwell, ).

.. The ATP-independent DNA relaxation activity of gyrase

The ATP-independent DNA relaxation activity of gyrase is of particular

mechanistic interest in that it is specific for negative supercoils. In other words,

in the absence of ATP the enzyme retains a directionality in its transport of a

DNA segment through another segment of the same DNA ring, but this

directionality is the opposite of that in the presence of ATP.

One interpretation of the observed directionality of gyrase-mediated DNA

transport in the absence of ATP is that the ATP-independent reaction follows a

reverse pathway of the ATP-driven reaction: the T-segment enters through the

C-gate, passes through the G-segment, and exits the N-gate (Kampranis &

Maxwell, ; Cullis et al. ; Critchlow et al. ). This interpretation

implies that the C-gate of DNA gyrase might not be tightly closed and might often

pop open to allow a T-segment to slip through. There is no data to suggest,

however, that interaction across the A–A dimer interface in DNA gyrase is weaker

than that in yeast DNA topoisomerase II. The A subunit of E. coli as well as

Micrococcus luteus DNA gyrase exists as stable dimers in solution under a wide

range of conditions (see for examples, Klevan & Wang, ; Kirschhausen et al.

), and the crystal structure of the -kDa gyrase A fragment shows an A–A

dimer interface burying an area of over AI # (Morais Cabral et al. ).

A second problem with the reverse-path interpretation of the ATP-independent

DNA relaxation activity of DNA gyrase is that it provides no satisfactory

explanation for the specificity of the reaction, namely only negatively supercoils

are removed. This problem can be better seen by first considering a gyrase

molecule bound to a G-segment in a supercoiled DNA. As discussed before, the

wrapping of the G-segment around the enzyme positions a T-segment, T", which

makes a positive node with the G-segment, and is presumably located close to the

normal entrance gate of the enzyme. This means that the T-segment which would

follow the reverse path could not be this well-positioned T"

segment and must

A simple molecular machine

therefore be a different T-segment T#. The specificity for the removal of negative

supercoils would then require that the enzyme could somehow position T"and T

#

near the normal entrance and exit gate of the enzyme, respectively, and that the

nodal sign between the G-segment and T#

must be a negative one. There is

presently no structural hint of such an arrangement.

The above discussion demonstrates that alternatives to the reverse-path

interpretation should be considered. One model that is consistent with the

available data is that the ATP-independent removal of every two negative

supercoils involves the binding of a DNA gyrase molecule at a negative node in

a negatively supercoiled DNA, perhaps through the assembly of one GyrA dimer

and two GyrB monomers at that node, to form an enzyme–DNA complex similar

to the one which is formed in the normal ATP-driven pathway immediately after

the passage of the T-segment through the DNA gate. In this complex, one of the

arms of the negative DNA node forms the G-segment, and the other arm, which

would subsequently serve as the T-segment, is present inside the cavity bounded

by the A protomers and the G-segment. In the absence of the ATP dependent

closure of the entrance gate and the conformational cascade which would normally

follow, this pre-assembled T-segment could not pass through the normal exit gate

(the C-gate). It could escape through the DNA gate whenever the gate opens,

however, to form the more stable enzyme-DNA complex in which the T-segment

assumes a positive-noded orientation relative to the G-segment. This process

would remove two negative supercoils.

In the mechanism postulated above, thermodynamic considerations predict that

the formation of such a pre-assembled complex with a negative node between the

two DNA segments could occur only in a negatively supercoiled DNA. Assembly

of gyrase subunits at a positive node in a positively supercoiled DNA could occur,

of course, but the energetics would strongly favour the formation of the normal

enzyme-DNA complex before the transport of the T-segment, in which the DNA

wraps around the enzyme to position a T-segment in a positive-noded orientation

with respect to the G-segment. Such a thermodynamically stable complex can not

be relaxed further in the absence of ATP, which explains why the ATP-

independent DNA relaxation activity of bacterial gyrase does not remove positive

supercoils.

.

Biochemical, genetic, and crystallographic studies of the type II DNA

topoisomerases in the past two decades have led to a detailed model on how such

an enzyme accomplishes its remarkable feat of moving one DNA duplex through

another. A type II DNA topoisomerase is in every way an ATP-fuelled molecular

machine. It is now well recognized that in the biological world most if not all

processes are carried out by intricate macromolecular machines, usually composed

of a complex assembly of components (see for examples, Alberts, , ;

Echols, ). The type II DNA topoisomerases are unusual in that they act as

single molecules: a single dimeric enzyme does all the work of binding to one

J. C. Wang

DNA double helix, capturing a second, opening a gate in the first DNA segment,

and transporting the captured segment through the transiently opened DNA gate,

all in a split second! The beauty of this simple and yet efficient machine is

awesome. Because of their simplicity, the type II DNA topoisomerases also

provide opportunities for deeper understanding of coupled processes; their

indispensability in cells and their importance as targets of antimicrobial and

anticancer therapeutics add yet another bonus to their studies.

.

I am grateful to many who made the study of DNA topoisomerases a joyful

undertakening, to Qiyong Liu for his help in the preparation of several of the

figures, to Janet Lindsley and James Berger for sharing results prior to publication,

to Tao-shih Hsieh and Tony Maxwell for discussions, and to NIH for three

decades of support of the research of my laboratory on DNA topoisomerases. I

thank Richard Henderson and Kurt Wu$ thrich for their patience and prodding

during the writing of this review; without patience they would have given up on

this review, and without their prodding I would have abandoned my charge.

.

A, B. M. (). The DNA enzymology of protein machines. Cold Spring Harb.

Symp. Quant. Biol. , –.

A, B. (). The cell as a collection of protein machines: Preparing the next

generation of molecular biologists. Cell , –.

A, J. A., J, A. P., H, A. J. & M, A. (). The -kDa N-

terminal fragment of the DNA gyrase B protein hydrolyses ATP and binds coumarin

drugs. Biochemistry , –.

A, J. A., O, G. & M, A. (). Nucleotide binding to the -

kilodalton N-terminal fragment of the DNA gyrase B protein. Biochemistry ,

–.

B, M. I., B, P., M, E. & T-V, G. P. (). In

vitro catenation and decatenation of DNA and a novel eukaryotic ATP dependent

topoisomerase. Cell , –.

B, A. D., O’D, M. H. & G, M. (). Energy coupling in Escherichia coli

DNA gyrase: the relationship between nucleotide binding, strand passage, and DNA

supercoiling. Biochemistry , –.

B, A. D. & M, A. (). DNA Topology. Oxford: IRL Press.

B, P., S, A., F, P. & W, J. C. (). Study of yeast DNA

topoisomerase II and its truncation derivatives by transmission electron microscopy.

J. Biol. Chem. , –.

B, J. M., F, D., W, J. C. & H, S. C. (). Structural similarities

between topoisomerases that cleave one or both DNA strands. Proc. natn. Acad. Sci.

U.S.A. , –.

B, J. M., G, S. J., H, S. C. & W, J. C. (). Structure and

mechanism of DNA topoisomerase II. Nature , –.

B, J. M. & W, J. C. (). Recent developments in DNA topoisomerase II

structure and mechanism. Curr. Op. Struct. Biol. , –.

A simple molecular machine

B, A., M, B., G, D., V, P. C., N, A. & F,

P. (). An atypical topoisomerase II from Archaea with implications for meiotic

recombination. Nature , –.

B, P. O. & C, N. R. (). A sign inversion mechanism for enzymatic

supercoiling of DNA. Science , –.

B, P. O., P, C. L. & C, N. R. (). A topoisomerase from

Escherichia coli related to DNA gyrase. Proc. natn Acad. Sci. U.S.A. , –.

C, P. R. & W, J. C. (). Alignment of primary sequences of DNA

topoisomerases. In DNA Topoisomerase and Their Applications in Pharmacology (ed.

L. Liu), pp. –. San Diego: Academic Press.

C, P. R., W, P. W, J. C. (). The C-terminal domain of

Saccharomyces cerevisiae DNA toposiomerase II. J. Mol. Cell. Biol. , –.

C, M. (). Ribbons .. J. appl. Crystallogr. , –.

C, J. J. (). DNA is linked to the rat liver DNA nicking-closing enzyme by

a phosphodiester bond to tyrosine. J. biol. Chem. , –.

C, A. H., Z, E. L. & O, N. (). A role for the passage helix

in the DNA cleavage reaction of eukaryotic topoisomerase II. J. biol. Chem. ,

–.

C, N. R., B, T. C. & W, J. H. (). In DNA Topology and its

Biological Effects (ed. N. R. Cozzarelli and J. C. Wang), pp. –. Cold Spring

Harbor: Cold Spring Harbor Laboratory Press.

C, N. R. & W, J. C. (). DNA Topology and its Biological Effects. Cold

Spring Harbor: Cold Spring Harbor Laboratory Press.

C, D. G. & H, T. (). Function of the hydrophilic carboxyl terminus

of type II DNA topoisomerase from Drosophila melanogaster. I. In vitro studies. J.

biol. Chem. , –.

C, F. H. C. (). Linking numbers and nucleosomes. Proc. natn. Acad. Sci.

U.S.A. , –.

C, S. E., O’D, M. H., H, A. J., C, M., G, M. &

M, A. (). The interaction of the F plasmid killer protein, CcdB, with

DNA gyrase: induction of DNA cleavage and blocking of transcription. J. molec. Biol.

, –.

C, P. M., M, A. & W, D. P. (). Exploiting nucleotide

thiophosphates to probe mechanistic aspects of Escherichia coli DNA gyrase.

Biochemistry , –.

D, R. & V, M. (). Evidence for a ring structure of polyoma virus DNA.

Proc. natn. Acad. Sci. U.S.A. , –.

E, H. (). Nucleoprotein structures initiating DNA replication, transcription,

and site-specific recombination. J. biol. Chem. , –.

F, L. M., M, K., O’D, M. H., O, H. & G, M. (). Site-

specific interaction of DNA gyrase with DNA. Proc. natn. Acad. Sci. U.S.A. ,

–.

F, F. B. (). Decomposition of the linking number of a closed ribbon. Proc.

natn. Acad. Sci. U.S.A. , –.

G, M., F, L. M. & O’D, M. H. (). DNA gyrase: purification and

catalytic properties of a fragment of gyrase B protein. Proc. natn. Acad. Sci. U.S.A.

, –.

G, M., F, L. M. & O, H., O’D, M. H. & M, K. ().

J. C. Wang

DNA gyrase: site-specific interactions and transient double-stranded breakage of

DNA. Cold Spring Harbor Symp. Quant. Biol. , –.

G, M., M, K., O’D, M. H., I, T. & T, J.-I. ().

Nalidixic acid resistance: a second genetic character involved in DNA gyrase activity.

Proc. natn. Acad. Sci. U.S.A. , –.

G, M., M, K., O’D, M. H. & N, H. A. (). DNA gyrase: An

enzyme that introduces superhelical turns into DNA. Proc. natn. Acad. Sci. U.S.A.

, –.

G, T. & W, J. C. (). Yeast DNA topoisomerse II is encoded by a single-

copy, essential gene. Cell , –.

H, T. T. & L, J. E. (a). Pre-steady state analysis of ATP hydrolysis

by Saccharomyces cerevisiae DNA topoisomerase II. . A DNA-dependent burst in

ATP hydrolysis. Biochemistry , –.

H, T. T., L, T. J. & L, J. E. (b). Pre-steady state analysis of

ATP hydrolysis by Saccharomyces cerevisiae DNA topoisomerase II. . Kinetic

mechanism for the sequential hydrolysis of two ATP. Biochemistry , –.

H, N. P., P, C. L., S, A. & C, N. R. (). Purification of

subunits of Escherichia coli DNA gyrase and reconstitution of enzymatic activity. Proc.

natn. Acad. Sci. U.S.A. , –.

H, D. S. & W, J. C. (). Mapping the active site tyrosine of Escherichia

coli DNA gyrase. J. biol. Chem. , –.

H, T. (). Knotting of circular duplex DNA by type II DNA topoisomerase

from Drosophila melanogaster. J. biol. Chem. , –.

H, T. & B, D. (). ATP-dependent DNA topoisomerase from D.

melanogaster reversibly catenates duplex DNA rings. Cell , –.

J, A. P. & M, A. (). Identifying the catalytic residue of the ATPase

reaction of DNA gyrase. Proc. natn. Acad. Sci. U.S.A. , –.

K, S. C. & M, A. (). Conversion of DNA gyrase into a

conventional type II topoisomerase. Proc. natn. Acad. Sci. U.S.A. , –.

K, R., K, L. C. & Z, B. H. (). On the nature of chromosome-sized

DNA molecules. Cold Spring Harbor Symp. Quant. Biol. , –.

K, T., W, J. C. & H, S. C. (). DNA gyrase and its

complexes with DNA: direct observation by electron microscopy. Cell , –.

K, K. & W, J. C. (). Mapping the topography of DNA wrapped

around gyrase by nucleolytic and chemical probing of complexes of unique DNA

sequences. Cell , –.

K, L. & W, J. C. (). DNA gyrase-DNA complex containing base

pairs of DNA and an α#β#

protein core. Biochemistry , –.

K, M. A. & C, N. R. (). Catenation of DNA rings by

topoisomerases. Mechanism of control by spermidine. J. biol. Chem. , –.

K, K. N. & A, B. M. (). Site-specific recognition of bacteriophage T

DNA by T type II DNA topoisomerase and Escherichia coli DNA gyrase. J. biol.

Chem. , –.

K, K. N. & C, N. R. (). Formation and resolution of DNA

catenanes by DNA gyrase. Cell , –.

L, M. P., S, M. & H, T. (). Nuclease protection by Drosophila DNA

topoisomerase II: enzyme}DNA contacts at the strong topoisomerase II cleavage

sites. J. biol. Chem. , –.

L, R. J., S, O. M., S, C. V., S, T., M, A., W,

A simple molecular machine

A. J. & W, D. B. (). The nature of inhibition of DNA gyrase by the

coumarins and the cyclothiolidines revealed by X-ray crystallography. EMBO J. ,

–.

L, W. & W, J. C. (). Footprinting of yeast DNA topoisomerase II lysyl side

chains involved in substrate binding and interdomainal interactions. J. biol. Chem.

, –.

L, C., W, J. C. & M! , A. (). Three dimensional structure of the

K N-terminal fragment of E. coli DNA topoisomerase I. Nature , –.

L, J. E. (). Intradimerically tethered DNA topoisomerase II is catalytically

active in DNA transport. Proc. natn. Acad. Sci. U.S.A. , –.

L, J. E. & W, J. C. (). Proteolysis patterns of epitopically labeled yeast

DNA topoisomerase II suggest an allosteric transition in the enzyme induced by ATP

binding. Proc. natn. Acad. Sci. U.S.A. , –.

L, J. E. & W, J. C. (a). On the coupling between ATP usage and DNA

transport by yeast DNA topoisomerase II. J. biol. Chem. , –.

L, J. E. & W, J. C. (b). Study of allosteric communication between

protomers by immunotagging. Nature , –.

L, L. F. (a). DNA Topoisomerases: Biochemistry and Molecular Biology. Boca

Raton: Academic Press.

L, L. F. (b). DNA Topoisomerases and Their Applications in Pharmacology. Boca

Raton: Academic Press.

L, L. F., L, C. C. & A, B. M. (). T DNA topoisomerase: a new ATP-

dependent enzyme essential for initiation of T bacteriophage DNA replication.

Nature , –.

L, L. F., L, C. C. & A, B. M. (). Type II DNA topoisomerases:

Enzymes that can unknot a topologically knotted DNA molecule via a reversible

double-stranded break. Cell , –.

L, L. F., R, T. C., Y, L, T, K. M. & C, G. L. (). Cleavage of

DNA by mammalian DNA topoisomerase II. J. biol. Chem. , –.

L, L. F. & W, J. C. (a). DNA–DNA gyrase complex: the wrapping of the

DNA duplex outside the enzyme. Cell , –.

L, L. F. & W, J. C. (b). Micrococcus luteus DNA gyrase: active components

and a model for its supercoiling of DNA. Proc. natn. Acad. Sci. U.S.A. ,

–.

M, K. J. (). DNA gyrase-catalysed decatenation of multiply linked DNA

dimers. J. biol. Chem. , –.

M, A. (). DNA gyrase as a drug target. Trends Microbiol. , –.

M, A. & G, M. (). The DNA-dependence of the ATPase activity of

DNA gyrase. J. biol. Chem. , –.

M, A., R, D. C. & G, M. (). Mechanistic studies of DNA gyrase.

In Biomacromolecular Stereodynamics III: Proceedings of the Fourth Conversation in

the Discipline of Biomolecular Stereodynamics (ed. R. H. Sarma and M. H. Sarma),

pp. –. Albany, New York: Adenine Press.

M, M. (). Formation of hybrid DNA by rotary diffusion during genetic

recombination. J. molec. Biol. , –.

M, M. & S, F. W. (). The replication of DNA in E. coli. Proc. natn.

Acad. Sci. U.S.A. , –.

M, K. G., L, L. F. & E, P. T. (). A homogeneous type II DNA

topoisomerase from HeLa cell nuclei. J. biol. Chem. , –.

J. C. Wang

M, K., F, L. M., O’D, M. H. & G, M. (). DNA gyrase action

involves the introduction of transient double-stranded breaks in DNA. Proc. natn.

Acad. Sci. U.S.A. , –.

M C, J. H., J, A. P., S, C. V., S, N., M, A. &

L, R. C. (). Crystal structure of the breakage-reunion domain of DNA

gyrase. Nature , –.

M, A. & C, N. R. (). Site-specific cleavage of DNA by E. coli

DNA gyrase. Cell , –.

M, A. & C, N. R. (). Contacts between DNA gyrase and its

binding site on DNA: features of symmetry and asymmetry revealed by protection

from nucleases. Proc. natn. Acad. Sci. U.S.A. , –.

O’D, M. H., T, J. K. & G, M. (). Mutations in the B subunit of

Escherichia coli DNA gyrase that affect ATP-dependent reactions. J. biol. Chem. ,

–.

O, G. & M, A. (). Evidence for a conformational change in the

DNA gyrase–DNA complex from hydroxyl radical footprinting. Nucleic Acids Res.

, –.

O, N. (). Eucaryotic topoisomerase II. Characterization of enzyme

turnover. J. biol. Chem. , –.

O, N., S, E. R. & B, D. (). DNA topoisomerase II from

Drosophila melanogaster: Relaxation of supercoiled DNA. J. biol. Chem. ,

–.

P, C. L., H, N. P., K, K. N., M, A., B, P. O., S,

A. & C, N. R. (). Structure and activities of Escherichia coli DNA

gyrase. Cold Spring Harbor Symp. Quant. Biol. , –.

P, Y., K, D. & K, K. (). Topological complexes between DNA

and topoisomerase II and effects of polymines. Biochemistry , –.

R, D. C., G, M. & M, A. (). Structure of the DNA gyrase-DNA

complex as revealed by transient electric dichroism. J. molec. Biol. , –.

R, R. J. & M, A. (a). Probing the limits of the DNA breakage-reunion

domain of the Escherichia coli DNA gyrase A protein. J. biol. Chem. , –.

R, R. J. & M, A. (b). DNA gyrase: structure and function. CRC Critical

Rev. Biochem. Mol. Biol. , –.

R, D. & K, A. (). Helical periodicity of DNA determined by enzyme

digestion. Nature , –.

R, J., B, J. M., H, S. C. & W, J. C. (). DNA transport by a

type II topoisomerase: Direct evidence for a two-gate mechanism. Proc. natn. Acad.

Sci. U.S.A. , –.

R, J., B, J. M. & W, J. C. (). On the simultaneous binding of

eukaryotic DNA topoisomerase II to a pair of double-stranded DNA helices. J. biol.

Chem. , –.

R, J. & W, J. C. (). The capture of a DNA double-helix by an ATP-

dependent protein clamp: a key step in DNA transport by type II DNA

topoisomerases. Cell , –.

R, J. & W, J. C. (). DNA transport by a type II DNA topoisomerase:

evidence in favor of a two-gate mechanism. Cell , –.

R, J. & W, J. C. (). The probabilities of supercoil removal and decatenation

by yeast DNA topoisomerase II. Genes to Cells , –.

A simple molecular machine

R, M. V., S, A., K, V. I. & W, W. ().

Hydrolysis of GTP by elongation factor G drives tRNA movement on the ribosome.

Nature , –.

R, V. V., U, C., V, A. V. & C, N. R. ().

Simplification of DNA topology below equilibrium values by type II topoisomerases.

Science , –.

S, M. & H, T. (). Double strand DNA cleavage by type II DNA

topoisomerase from Drosophila melanogaster. J. biol. Chem. , –.

S, P., O, S., O, P. & H, R. (). Structure and

conformational changes of DNA topoisomerase II visualized by electron microscopy.

Proc. natn. Acad. Sci. U.S.A. , –.

S, K. & Y, M. (). Functional dissection of the phosphorylated

termini of fission yeast DNA topoisomerase II. Mol. Cell. Biol. , –.

S, J. R. & M, M. T. (). A consensus sequence for cleavage by

vertebrate topoisomerase II. Nucleic Acids Res. , –.

S, W. L. & O, E. (). DNA gyrase: affinity chromatography on

novobiocin-sepharose and catalytic properties. Nucleic Acids Res. , –.

S, G. L., K, G. J. & H, W. M. (). T DNA-delay proteins required

for specific DNA replication form a complex that has ATP-dependent DNA

topoisomerase activity. Proc. natn. Acad. Sci. U.S.A. , –.

S, J. T., S, S. & M, A. S. (). Vaccinia DNA topoisomerase, I:

Kinetic evidence for general acid-base catalysis and a conformational step.

Biochemistry , –.

S, A. & C, N. R. (). The intrinsic ATPase of DNA gyrase. J. biol.

Chem. , –.

S, A., H, N. P., B, P. O., P, C. L. & C, N. R. ().

Energy coupling in DNA gyrase and the mechanism of action of novobiocin. Proc.

natn. Acad. Sci. U.S.A. , –.

S, A., H, N. P. & C, N. R. (). DNA gyrase subunit

stoichiometry and the covalent attachment of subunit A to DNA during DNA

cleavage. Nucleic Acids Res. , –.

S, A., P, P. L., K, K. N. & C, N. R. (). Mechanism of

action of nalidixic acidL: purification of E. coli nalA gene product and its relationship

to DNA gyrase and a novel nicking-closing enzyme. Proc. natn. Acad. Sci. U.S.A. ,

–.

S, O. & V, A. (). Terminal stages of SV DNA replication

proceed via multiply intertwined catenated dimers. Cell , –.

S, O. & V, A. (). Arrest of segregation leads to accumulation of

highly intertwined catenated dimers: dissection of the final stages of SV DNA

replication. Cell , –.

T, J. K. & G, M. (). Characterization of the ATP binding site of

Escherichia coli DNA gyrase. Affinity labeling of Lys- and Lys- of the B-

subunit by pyridoxol «-diphospho-« adenosine. J. biol. Chem. , –.

T, J. K., B, A. D. & G, M. (). Slow interaction of «-adenylyl-

beta, gamma-immidodiphosphate with Escherichia coli DNA gyrase. Evidence for

cooperativity in nucleotide binding. J. biol. Chem. , –.

T, W., S, R. M., M, M. E. & H, C. (). Genetic analysis of the

gyrase A-like domain of DNA topoisomerase II of Saccharomyces cerevisiae. Genetics

, –.

J. C. Wang

T, A. P. & M, A. (). Probing the role of the ATP-operated clamp in

the strand-passage reaction of DNA gyrase. Nucleic Acids Res. , –.

T, B., B, C., L, K., A, A. H., S, B. S. &

W, O. (). Characterization of the interaction between topoisomerase

II and DNA by transcriptional footprinting. J. molec. Biol. , –.

T, Y.-C., K, K. & W, J. C. (). Covalent bonds between protein and

DNA. Formation of phosphotyrosine linkage between certain DNA topoisomerases

and DNA. J. biol. Chem. , –.

V, A., L, L., S, O., B, J., O, E., S, P.

& F, D. (). Cellular and SV chromatin: replication, segregation,

ubiquitination, nuclease-hypersensitivity sites, HMG-containing nucleosomes, and

heterochromatin-specific protein. Cold Spring Harbor Symp. Quant. Biol. , –.

W, J. C. (). Interaction between DNA and an Escherichia coli protein ω. J.

molec. Biol. , –.

W, J. C. (). Helical repeat of DNA in solution. Proc. natn. Acad. Sci. U.S.A.

, –.

W, J. C. (). DNA topoisomerases. In Nucleases (ed. R. Roberts and S. Linn),

pp. –. New York: Cold Spring Harbor Laboratory.

W, J. C. (). An introduction to DNA supercoiling and DNA topoisomerase-

catalyzed linking number changes of supercoiled DNA. In DNA Topoisomerase and

Their Applications in Pharmacology, (ed. L. Liu), pp. –. San Diego: Academic

Press.

W, J. C. (). DNA topoisomerases. Ann. Rev. Biochem. , –.

W, J. C., G, R. I., J, K. J., K, K., K, L.,

K, M. L. & T, Y-C. (). DNA topoisomerases. In Mechanistic Studies

of DNA Replication and Genetic Recombination (ed. B. Alberts), pp. –. New

York: Academic Press.

W, J. C. & L, L. F. (). DNA topoisomerases: enzymes that catalyze the

concerted breaking and rejoining of DNA backbone bonds. In Molecular Genetics,

Part III, (ed. J. H. Taylor), pp. –. New York: Academic Press.

W, S. A. & C, N. R. (). Supercoiled DNA-directed knotting by

T topoisomerase. J. biol. Chem. , –.

W, J. D. & C, F. H. C. (). Genetic implications of the structure of

deoxyribonucleic acid. Nature , –.

W, R. & V, J. (). The cyclic coil helix and cyclic forms of polyoma viral

DNA. Proc. natn. Acad. Sci. U.S.A. , –.

W, D. B., D, G. J., D, E. J., M, A. & D, G. ().

Crystal structure of the N-terminal domain of DNA gyrase B protein. Nature ,

–.

Y, M. & S, R. (). Genetics of DNA topoisomerases. In DNA

Topology and its Biological Effects (ed. N. R. Cozzarelli and J. C. Wang), pp. –.

Cold Spring Harbor: Cold Spring Harbor Laboratory Press.

Z, E. L. & O, N. (). Eukaryotic topoisomerases recognize

nucleic acid topology by preferentially interacting with DNA crossovers. EMBO J. ,

–.