Sie sind auf Seite 1von 17

Available online at www.sciencedirect.

com

ScienceDirect
Physics of Life Reviews 18 (2016) 118134
www.elsevier.com/locate/plrev

Review

Disentangling DNA molecules


Alexander Vologodskii
New York University, New York, NY 10003, United States
Received 6 February 2016; received in revised form 29 April 2016; accepted 2 May 2016
Available online 4 May 2016
Communicated by E. Shakhnovich

Abstract
The widespread circular form of DNA molecules inside cells creates very serious topological problems during replication. Due
to the helical structure of the double helix the parental strands of circular DNA form a link of very high order, and yet they have to
be unlinked before the cell division. DNA topoisomerases, the enzymes that catalyze passing of one DNA segment through another,
solve this problem in principle. However, it is very difficult to remove all entanglements between the replicated DNA molecules
due to huge length of DNA comparing to the cell size. One strategy that nature uses to overcome this problem is to create the
topoisomerases that can dramatically reduce the fraction of linked circular DNA molecules relative to the corresponding fraction
at thermodynamic equilibrium. This striking property of the enzymes means that the enzymes that interact with DNA only locally
can access their topology, a global property of circular DNA molecules. This review considers the experimental studies of the
phenomenon and analyzes the theoretical models that have been suggested in attempts to explain it. We describe here how various
models of enzyme action can be investigated computationally. There is no doubt at the moment that we understand basic principles
governing enzyme action. Still, there are essential quantitative discrepancies between the experimental data and the theoretical
predictions. We consider how these discrepancies can be overcome.
2016 Elsevier B.V. All rights reserved.

Keywords: DNA; DNA topology; DNA topoisomerases; Topology simplification

1. Introduction

DNA molecules are usually very long and often form circles inside cells. This creates very serious topological
problem during DNA replication. Due to the helical structure of the double helix the parental strands of circular DNA
form a link of very high order. When each strand is duplicated, the link between the complementary parental strands
is converted into the link of the daughter double helices. On the other hand, the daughter DNA molecules have to
be unlinked before cell division. In the case of linear DNA there is no topological restriction on separation of the
daughter molecules since the free ends allow unwinding of the interwound double helices. However, for very large
linear molecules such unwinding would take too much time due to the hydrodynamic friction. So, the problem of
unlinking of newly replicated DNA molecules exists both for circular and linear DNA molecules in living cells.

E-mail address: alex.vologodskii@nyu.edu.

http://dx.doi.org/10.1016/j.plrev.2016.05.001
1571-0645/ 2016 Elsevier B.V. All rights reserved.
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 119

Fig. 1. A typical conformation of supercoiled DNA. A key feature of supercoiled DNA is interwinding of the double-stranded segments. Tight
interwound conformations of supercoiled DNA dramatically reduce the equilibrium probability of their linking one with another [5]. The simulated
conformation shown corresponds to 10 kb DNA and was obtained by the author as described in [6].

DNA topoisomerases, the enzymes that catalyze passing of one DNA segment through another, solve this prob-
lem, in principle. However, even inside the cell conformations of DNA molecules have some degree of randomness.
Therefore, the unlinking is a stochastic process, and the enzymes can not only unlink but also introduce links between
DNA molecules.
The process of unlinking of the parental DNA strands looks rather straightforward till the very end of DNA repli-
cation. The replication complex moving along parental DNA cannot rotate with sufficient speed around this DNA due
to the hydrodynamic friction. As a result, the positive torsional stress is accumulating in the parental DNA ahead of
the complex. The only way to eliminate this stress is to form a swivel on the parental DNA ahead of the replication
complex, and DNA topoisomerases create such a swivel [1,2]. Thus, the torsional stress governs the direction of the
topological changes during most of the replication process. Still, at the very end of the process there is not enough
room on the parental DNA for this swivel, and some number of interlinks have to be removed after completion of
replication [3]. These final steps of unlinking are more random, and the individual strand-passages can either reduce
or increase the linking of daughter DNA molecules.
It is very difficult to remove all the entanglements between the replicated DNA molecules because DNA length
exceeds the size of the cells by a few orders of magnitude. Special mechanisms are needed to provide complete
unlinking, and a few such mechanisms have been developed by nature.
First, compression of DNA into thicker and shorter chromosomes can greatly simplify the problem. This is similar
to spooling, that is used to keep very long and thin threads in order. The extreme case of compactness are chromosomes
in the mitotic phase of the cell cycle, where they look more like extended particles, rather than polymer chains. Clearly,
the individual chromosomes cannot be entangled with one another in this phase. DNA replication, however, takes place
in interphase, when chromosomes are much less compact. We do not know much about local shape of chromosomes
around the moving replication complex, but it seems clear that nearby DNA regions should be sufficiently loose. Thus,
although compaction is a very efficient way to avoid topological entangling, is has limited application during DNA
replication.
Two other mechanisms that contribute to complete unlinking are related to the thermodynamic concept of topo-
logical equilibrium, formally defined in the next section. Let us imagine that we have a solution of circular DNA
molecules with phantom backbones, so that molecular segments can easily pass through one another. As a result
of strand-passing, thermal motion will permanently change the topology of the molecules, until the distribution of
topological forms reaches a steady state. This steady state corresponds to topological equilibrium. At a very high con-
centration of DNA in cells (around 20 mg/ml in bacterial nucleoids) the equilibrium fraction of linked circular DNAs
would be high as well. DNA topoisomerases make DNA segments, to some extent, phantom. If the enzymes drive
circular DNA molecules to topological equilibrium, they would hardly solve the unlinking problem. One mechanism
of facilitating the problem is based on shifting the topological equilibrium to a smaller fraction of linked molecules.
This is achieved via DNA supercoiling. Prokaryotic circular DNA molecules are always negatively supercoiled inside
the cells. This means that the linking number of the complementary strands, Lk, is smaller, by 57%, than the value
which corresponds to the torsionally unstressed molecules, Lk0 . This deficit in Lk, or the linking number difference,
Lk = Lk Lk0 , causes torsional deformation of the double helix and also results in the interwinding of DNA seg-
ments (see [4], for example). Although the overall sizes of supercoiled and torsionally relaxed DNA molecules are
not so different, interwound conformations of supercoiled DNA (Fig. 1) are much tighter than the conformations of
relaxed molecules (see Fig. 5). It is much less probable that a random line will pass inside an interwound superhelix
120 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

Fig. 2. The simplest knots and links. Knots and links are classified by the minimum number of crossing on their projection, nc . All types of knots
(a) and links (b) with nc less than 7 are shown here.

than inside a loose random coil formed by a relaxed circular DNA of the same size. Thus, supercoiling greatly reduces
the equilibrium probability that two circular DNA will be linked.
The second of these mechanisms, which is the subject of the current review, is the ability of enzymes, type II DNA
topoisomerases, to deliberately reduce the level of entanglements in DNA molecules. Type II DNA topoisomerases
catalyze the passing of double-stranded DNA segments one through another (see [7,8] for reviews). It is natural to
assume that they have to drive the distribution of DNA topological states in solution to its equilibrium form. However,
a striking discovery, made nearly two decades ago, shows that the enzymes can dramatically reduce fractions of
linked and knotted circular DNA molecules relative to the corresponding fractions at thermodynamic equilibrium [9].
Clearly, this property is very helpful for unlinking DNA molecules inside the cell.
The discovery posed a puzzle that has attracted a great deal of attention in the biophysical community. The problem
is that topology is a global property of circular molecules and cannot be determined by enzymes that interact with
DNA only locally. Thus, there is no way for the enzymes to unambiguously determine the topological consequences
of a particular strand-passing event that can either unlink or link two circular DNA molecules. We needed to figure
out how the enzymes accomplish the simplification of DNA topology.
This review considers the experimental studies of the phenomenon and analyzes the theoretical models that have
been suggested in attempts to explain it. How various models of the enzyme action can be investigated computationally
is outlined. There is no doubt at the moment that we have a model that provides good qualitative understanding of
the phenomenon. Experimental studies have convincingly proved that the enzyme-catalyzed reaction has all features
required by the model. Still, there are essential quantitative discrepancies between the experimental data and the model
predictions. At the end of the review we will discuss a way to overcome these discrepancies.

2. Simplification of DNA topology by type II DNA topoisomerases

Before we describe the phenomenon of topology simplification we need to clarify the concept of topological
equilibrium and describe, in some details, the enzymes of interest, type II DNA topoisomerases.

2.1. Equilibrium distribution of topological states

Circular DNA molecules, as well as other circular polymers, can exist in different topological states. They can form
knots of various types and can be linked with one another by various topological links [10,11]. There is an infinite
number of topologically distinguished knots and links. Knots and links are classified according to the minimal number
of intersections in their projection on a plane [12]. Such projections are called the standard representations of knot
and links. A few of the simplest knots and links in their standard representations are shown in Fig. 2.
Since circular polymer molecules cannot change their topological states without breaking their backbones, the
distributions of the topological states of circular DNA molecules in a particular sample do not change over time, so
the distributions are not ergodic. In this respect the distributions of topological states are different from distributions
of other conformational properties of polymers that always relax to their equilibrium forms in dilute solutions. Still,
we can talk about the equilibrium distributions of topological states. The definition of the equilibrium distributions is
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 121

not different from the definition of the equilibrium distributions of other conformational properties. The equilibrium
probability of topology i, P (i), is specified as

states with exp(Ek /kB T )
topology i
P (i) =  , (1)
all states exp(Ek /kB T )
where Ek is the energy of state (conformation) k, and kB T is the Boltzmann temperature factor. This definition is used
to find the equilibrium distributions of topological states by computer simulations (see [13] for review). All that one
needs to do for this goal is to analyze the topology of each conformation of a simulated equilibrium conformational set
(obtained for the corresponding phantom chains). This can be done by calculating a topological invariant for each of
the simulated conformation, as was suggested by Frank-Kamenetskii and co-workers in the early 1970s [11,14,15]. Of
course, the invariant has to be powerful enough to have different values for chains with different topologies appearing
in the simulation. Over the years researchers have used this approach to investigate various aspects of knotting in
polymer chains (reviewed in [16,17]).
The equilibrium distributions of topological states can be also obtained experimentally, by slow cyclization of DNA
molecules in solution [5,1820]. Such slow cyclization can be achieved by joining complementary single-stranded
ends of the molecules, sticky ends. Some natural phage DNAs have such sticky ends; they can also be created in labo-
ratory by using special enzymes that introduce sequence-specific single-stranded nicks in double-stranded DNA (see
[21], for example). It is well established that the equilibrium distributions obtained experimentally and by computer
simulations are in very good quantitative agreement [5,19].
Throughout this review, when we talk about knots in circular DNA molecules and links between them, we assume
that the molecules are double-stranded, with at least one single-stranded nick. In this case the molecules cannot
have torsional stress, due to unrestricted rotation of one strand around the other. So, the molecules behave like simple
polymer chains. If both complementary strands of double-stranded DNA are closed (and the topology of the DNA axis
corresponds to its unknotted circular contour), the topological link between the complementary strands is specified
by their linking number, Lk, which may be determined in the following way. One of the DNA strands defines the
edge of an imaginary surface (any surface gives the same result). Lk is the algebraic (i.e., sign-dependent) number of
intersections of the surface by the other strand. Lk depends only on the topological state of the strands and hence is
maintained through all conformational changes that occur in the absence of strand breakage [22,23]. It follows from
the definition that Lk is an integer. By convention, the Lk of a closed circular DNA formed by a right-handed double
helix is positive. It is easy to obtain the equilibrium distribution over Lk experimentally by ligating single-stranded
nicks in nicked circular DNA molecules [2427]. Usually, the number of molecules in a sample of closed circular
DNA, A, has a Gaussian distribution over the values of Lk, A(Lk):
 
(Lk Lk)2
A(Lk) = A0 exp (2)
2(Lk)2 
where A0 is a coefficient, Lk and (Lk)2  are the distribution average and its variance, correspondingly. The
equilibrium distribution over Lk is centered around the average value of Lk in the nicked molecules, Lk0 , so in this
case Lk = Lk0 . For DNA molecules a few thousand base pairs in length, the distribution over Lk typically includes
a few topoisomers (molecules with different values of Lk) in a comparable amount. The topoisomers can be separated
by gel electrophoresis [28], and this allows reliable determination of (Lk)2 . The equilibrium distribution over Lk
can be also obtained by computer simulation [2933].

2.2. Type II DNA topoisomerases

In general, DNA topoisomerases catalyze chains of chemical transformations that result in passing one DNA seg-
ment through another [34]. They catalyze the strand-passing reaction by cutting a single strand or both strands of
the double helix, and depending on this the topoisomerases belong to type I or II, respectively. Both types of DNA
topoisomerases are necessary for DNA replication and transcription in all living cells. Type II topoisomerases in-
cludes type IIA and IIB. The latter type is not so widespread, and we will mean type IIA enzymes when we talk about
type II topoisomerases (topo II). The enzymes consist of two identical halves that can be monomers, dimers, or even
trimers (see [8] for review). In all cases the structures of topo II have twofold symmetry. The reaction cycle catalyzed
122 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

Fig. 3. Diagram of type IIA topoisomerase action. In state 1 the enzyme is bound with the first DNA segment (G segment, shown by yellow). The
segment is bent in the complex. In its open-clamp conformation the enzyme waits for another DNA segment (T segment, shown by light brown).
When a potential T segment is inside the clamp (state 2), the clamp can trap the segment there (state 3), or the segment can diffuse out. Closure
of the clamp (state 3) triggers breaking the G segment and opening the DNA gate (circled in state 1), allowing the T segment to reach the central
cavity of the enzyme (state 4). Then the DNA gate closes and the G segment is religated. Closure of the DNA gate triggers opening the exit gate
(C gate) and release of the T segment (state 5). (For interpretation of the references to color in this figure legend, the reader is referred to the web
version of this article.)

by topo II consists of a few successive steps (Fig. 3) [7,8,34]. First, the enzyme binds a DNA segment that will be
broken later (G segment). Then the topoisomerase captures, using its clamps, a second segment (T segment) of the
same or another DNA molecule. This capture serves as a signal to introduce a transient double-stranded break in the
G segment. The broken bonds in the DNA strands are replaced by covalent bonds between two amino acids of the
enzyme and the newly formed DNA ends. So, one strand at each end of the double helix becomes covalently bound
to the enzyme. The reaction triggers a conformational change in the enzyme which creates a gap between the ends of
the broken G segment. The T segment diffuses through the gap and, eventually, leaves the enzyme through another
gate. The break in the G segment is resealed after the passage of the T segment, by the reverse chemical reaction.
So, the enzyme works like a sophisticated molecular machine. It is important that the chemical transformations of
the reaction cycle do not require energy, and type I DNA topoisomerases catalyze the same transformations without
energy consumption [35]. In case of topo II energy is needed for the unidirectional transfer of a T segment through
the enzyme, and it comes from ATP hydrolysis incorporated in the reaction cycle. This unidirectional transfer was
established in 1996 by Wang and co-workers in their very elegant experiments [36]. It was not clear at that time,
however, why the enzymes need a particular directionality of strand transfer.

2.3. Simplification of DNA topology

In 1997 Rybenkov et al. unexpectedly found that topo II enzymes can greatly reduce the fractions of knotted
and linked circular DNA molecules compared with the corresponding equilibrium values, up to hundreds of times
(Fig. 4) [9]. The experiments were started by adding topo II to solutions of nicked circular DNA with various initial
concentrations of knots/links. The enzymes maintain their activity in such experiments over some hours. After that
time the reaction products were separated by gel electrophoresis followed by quantitative determination of their frac-
tions. The quantitation was based on the radioactive labeling of the DNA molecules and therefore allowed accurate
measurements of DNA bands in the gels over a very wide dynamic range. It was found that when the molar ratio of
topo II to the molecules of DNA was between 0.1 and 1, the resulting fractions of knots and links (also called cate-
nanes) were independent of the enzyme concentration and the initial distribution of topological states (see Fig. 4b).
We will only consider results obtained for this range of enzyme/DNA ratios. At lower enzyme concentrations the yield
of fractions was affected by the exhaustion of enzyme activity. At higher concentrations of topo II a few molecules
of the enzyme bound to a DNA molecule disturbed its conformational distribution and the fractions of knotted/linked
molecules increased.
Thus, topoisomerases are capable of creating new steady-state distributions of topological states. In these distribu-
tions the fractions of knots and links are greatly reduced compared to the equilibrium fractions at the same conditions.
In 7 kb DNA the steady-state fraction of trefoil knots (only the simplest knots, trefoils, were detected in the exper-
iments) was reduced up to a factor of 90 relative to the equilibrium value. The maximum reduction in the catenane
fraction (only the simplest links 21 were detected) was a factor of 15. In this review, these ratios of equilibrium to
steady state fractions will be called R-factors. The values of R-factors obtained for topo II from various organisms
were different, but always larger than 1.
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 123

Rybenkov et al. also investigated the distribution of closed circular DNA over Lk of the complementary strands
created by topo II. They found that topo II reduces the distribution variance, (Lk)2 , comparing to its equilibrium
value [9]. This effect was also measured in three other studies [3739]. All results on the distribution narrowing are in
reasonable agreement. However, the maximum reduction of the steady state variance relative to the equilibrium value
does not exceed a factor of 2. So, the effect is relatively small, and its biological meaning remains elusive. Theoretical
analysis of (Lk)2  reduction is more difficult since more than two topological states of the molecules have to be
taken into account. The corresponding analysis has been performed for only one model of the phenomenon [40]. We
will not discuss the effect in the review.
Of course, shifting the distributions of topological states from their equilibrium forms requires energy, and the en-
zymes receive it by coupling the strand-passing reaction with ATP hydrolysis. Thus, the finding is in agreement with
thermodynamic principles. This property of topo II makes clear biological sense, since unlinking DNA molecules
during their replication and segregation into daughter cells is its major function. The phenomenon of topology sim-
plification by topo II attracted a lot of attention because it was very difficult to understand how the enzymes could
achieve this goal. The problem is that topology is a global property of circular molecules. The average size of ran-
dom coils formed in solution by the DNA molecules used in the experiments is many times larger than the size of
proteins (Fig. 5). Thus, topology of the DNA molecules cannot be deduced by enzymes that only interact with DNA
locally. Correspondingly, there is no way for the enzymes to determine the topological consequences of a particular
strand-passing event that can either unknot or knot DNA molecule, link two molecules one with another or unlink
them. It has been realized, eventually, that the only way for the enzymes to achieve topology simplification is to use
the fact that the probability of some specific local conformations of DNA segments depends on DNA topology. If the
enzymes only catalyze strand-passing for local conformations that appear more often in knotted/linked molecules,
they can reduce the steady state fractions of knots and links. Implicitly or explicitly this idea has been the basis for all
models suggested in attempts to explain the phenomenon [4143]. We consider these models in detail below.

3. Theoretical analysis

3.1. General consideration

We start the analysis from knots formed by nicked circular DNA. Knotting/unknotting is a monomolecular process
and for this reason it is simpler for theoretical analysis than linking/unlinking. Both experimental and theoretical data
show that only trefoil knots appear with notable probability in DNA molecules a few kb in length at thermodynamic
equilibrium [19,20], so we can consider only two topological states, unknotted, or trivial knots, and trefoil knots.
Thus, the steady-state concentrations of knotted and unknotted molecules created by the topoisomerases, Ck and Cu
respectively, depend on the rates of exchanges between the two topological states:
dCk /dt = kuk Cu kku Ck = 0, (3)
where kuk is the rate constant for conversion of unknotted molecules into knotted ones and kku is the reverse rate
constant. So, the analysis is reduced to the estimation of kuk and kku . It will be assumed below that the assembly of
the reaction complex is not a diffusion-limited process, so that the reactions rates are proportional to the probability of
the complex appearance in the DNA conformational space. Available data, obtained for other DNA-protein systems,
support this assumption [44,45], although it has not been tested for the strand-passing reaction catalyzed by topo II.
For now we can only suggest that the analysis in the limit of a diffusion-limited process would bring approximately the
same results. It would be definitely interesting to simulate the process in this limit, but it is a very hard computational
problem. Thus, all computational analysis of the phenomenon has been performed under the assumption that the
reaction rate is not limited by the rate of diffusion of DNA segments [42,4649]. Under this assumption we can
present the rate constants as
kuk = Aruk P (j |u) and kku = Arku P (j |k), (4)
where ruk and rku are fractions of strand passages that change topology from unknotted to knotted state and from
knotted to unknotted state, correspondingly, P (j |u) and P (j |k) are the probabilities of specific enzyme-required
mutual conformations of two juxtaposed DNA segments, j , for unknotted and knotted chains, and A is a coefficient
that depends on the enzyme concentration and activity. It is important to emphasize that the values of ruk , rku , P (j |u),
124 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

and P (j |k) depend on j . The terms ruk and rku are very important because many strand passages do not change
topology of DNA molecules. Combining Eqs. (3) and (4) we obtain the expression for the steady state fraction of
knotted molecules, fkss :
Ck Ck ruk P (j |u)
fkss = = = , (5)
Cu + Ck Cu rku P (j |k)
since Cu is always much larger than Ck .
It should be noted that ruk , rku , P (j |u), and P (j |k) depend not only on the DNA length but also on the size of two
loops created by the juxtaposition of the segments. Thus, the products ruk P (j |u) and rku P (j |k) have to be properly
averaged over the loop sizes. The simulation procedures used in the calculations provide such averaging.
Eq. (5) can be applied, in particular, to imaginary phantom chains whose segments can freely pass one through
another regardless of their mutual conformation. Of course, for such chains the steady state distribution of topological
states is the equilibrium distribution. Also, the probabilities of the complex formation, P (j |u) and P (j |k), are simply
the probabilities of a segment collision in each state, P (c|u) and P (c|k). Thus, we obtain that
eq ruk P (c|u)
fk = , (6)
rku P (c|k)
eq
where fk is the equilibrium fraction of knots.
Although Eq. (5) could be used to investigate all the models of the phenomenon considered below, the correspond-
ing computational procedure is not very efficient, and cannot be adapted for the linkingunlinking reaction of two
circular DNA molecules. Liu et al. developed an efficient way to estimate R-factors bypassing the direct modeling
of the complex formation [46]. Their algorithm is based on conformational averaging with the preformed juxtaposi-
tion of two segments, with mutual conformation j . The needed equation can be derived from Eq. (5), if we consider
probabilities P (j |u) and P (j |k) as the corresponding conditional probabilities. So, P (j |u) and P (j |k) are the prob-
abilities of juxtaposition j under condition that the chain is unknotted or knotted, correspondingly. By the definition
of conditional probability, P (j |u) and P (j |k) can be expressed as
P (j, u) P (j, k)
P (j |u) = and P (j |k) = , (7)
P (u) P (k)
where P (j, u) and P (j, k) are the joint probabilities that the chain has a particular topology and two segments are
properly juxtaposed, P (u) and P (k) are the total probabilities of the corresponding topologies. The values of P (j |u)
and P (j |k) do not depend on the distribution of topological states, so Eqs. (7) are valid, in particular, for the equilib-
rium distribution. Thus, from Eqs. (5) and (7) we can obtain
ruk P (j, u)eq P (k)eq
fkss = . (8)
rku P (j, k)eq P (u)eq
The probabilities P (j, u)eq and P (j, k)eq can be expressed through other conditional probabilities,
P (j, u)eq = P (u|j )eq P (j )eq , P (j, k)eq = P (k|j )eq P (j )eq , (9)
where P (u|j )eq and P (k|j )eq are the equilibrium probabilities of unknotted and knotted states under condition that
two of the chain segments form the required juxtaposition, and P (j )eq is the total probability of the juxtaposition in
the equilibrium ensemble. Combining Eqs. (8) and (9) and using the definition of R-factor we obtain that
rku P (k|j )eq
Rk = . (10)
ruk P (u|j )eq
Eq. (10) allows one to calculate Rk by sampling an equilibrium set of chain conformations with the preset juxtaposi-
tion j . During this sampling the chain segments can freely pass one through another, changing the chain topology. In
the prepared set we have to calculate, for each conformation, the initial topology of the chain and its topology after
strand-passing at the juxtaposition, to determine the variables in Eq. (10).
The same approach, although in less general form, was suggested for calculation of the R-factor for the link-
ing/unlinking process [48]. The experiments on linking/unlinking by topo II were performed at sufficiently low DNA
concentration so that links between three or more chains are not detected [9]. It was also found that for a circular DNA
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 125

a few kb in length only two topological states appear with significant probability, one where the molecule is not linked
with another one (unlinked state, u) and the state where it forms the simplest link 21 (see Fig. 2) with another cir-
cular DNA (linked state, c). Therefore, a theoretical analysis of the linking/unlinking process needs also to account
only for these two states. Under these conditions the R-factor for the steady state fraction of links (catenanes), Rc , can
be calculated by an equation which is similar to Eq. (10),
P (c|j )eq
Rc = , (11)
P (u|j )eq
where P (c|j )eq and P (u|j )eq are the equilibrium probabilities of linked (forming the catenane) and unlinked states
under condition that segments of two circular molecules form juxtaposition j . Comparing Eqs. (10) and (11) one can
note that there are no coefficients specifying the probability of topology change upon the strand-passing in Eq. (11).
It is easy to prove that for the linking/unlinking reaction, where only these two topological states are possible, the
coefficients equal 1.

3.2. Computer simulation of DNA conformational properties

Computer simulation is crucially important for the analysis of various issues related to the phenomenon. Indeed,
we need to study how various conformational properties of DNA molecules depend on their topology, and so far
we have no way to approach this kind of problems analytically. On the other hand, computer simulation of DNA
conformational properties has been well developed and allows accurate quantitative modeling of both equilibrium and
dynamic large-scale statistical properties of the molecules (see [50] for review).
The equilibrium distributions of DNA conformations, needed for the analysis, can be accurately simulated by a
chain of rigid, impenetrable cylinders. An example of such a chain is shown in Fig. 5. In the case of nicked DNA
this model has only three parameters, the bending rigidity between adjacent cylinders of the chain, g, the cylinder
diameter, d, and its length, l. The value of g is specified by the DNA persistence length, a, and the chosen value of l:
g = kB T a/ l. (12)
Eq. (12) is valid under the condition that l  a. For double-stranded DNA the value of a is close to 50 nm [5153].
The value of d accounts both for the geometrical diameter of the double helix and for electrostatic repulsion between
the negatively charged DNA segments. Therefore, the value of d is larger than the geometric diameter. The concept
of effective diameter is the simplest but sufficiently accurate way to account for the electrostatic interaction between
the DNA segments [54,55]. The value of d strongly depends on ionic conditions in solution; for the conditions of
the experiments with topo II the value of d is close to 5 nm [27]. The choice of the third, computational parameter
l depends on the conformational properties which one wants to simulate [50]. In general, the value of l should be
chosen so that a simulated property of interest is not changing on further reduction of l. In the case under consider-
ation, it should not exceed the size of the enzyme, so l of 10 nm was used in the majority of the simulations. The
sampling of the conformational space is performed by the Metropolis procedure [50]. The analysis requires topology
determination of the simulated conformations, as well as the topological outcome of the strand-passing reaction. This
is done by calculating a topological invariant that has different values for closed contours with different topologies.
The Alexander polynomial is the most convenient invariant that allows to distinguish among the simplest topologies
[14,15,56], although stronger but more time-consuming invariants were also used in the computations [47,49].
In the case of knots the method based on the preformed juxtaposition (Eq. (10)) requires proper averaging over
the sizes of two loops created by the juxtaposition. This can be done by incorporating in the Metropolis procedure
a special move that changes the loop size [49]. We, performing the simulation, used the reptation move to transfer a
chain segment from one loop to another during the simulation [6].
We consider below three models which were well defined for a quantitative analysis. We will not discuss models
which have evident flaws, like the one suggested in the original study [9], or the models which were not specified in
detail sufficient for a quantitative analysis [37,39].

3.3. The kinetic proofreading model

The first proposed model of the phenomenon uses the fact that segment collisions should occur more often in DNA
molecules with more complex topology due to their higher compactness [41]. The key assumption of the model is that
126 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

Fig. 4. Reduction of the fractions of knots and catenanes in solution by topo II from E. coli [9]. The fraction of knots was measured for nicked DNA
molecules 10 kb in length (a); the fraction of catenanes (links between two molecules) was determined for links between nicked circular molecules
of 7 and 10 kb in length (b). The data for two initial concentrations of catenanes are shown by two different shades of grey. The equilibrium
fractions were measured in separate experiments in which 10 kb DNA was slowly cyclized via long cohesive ends, and the values were confirmed
by Monte Carlo simulations [5,19]. The reduction of the fraction of knots/links reached the steady state level when the molar ratio of topo II to the
molecules of DNA was between 0.1 and 1.

the G segment-bound enzyme needs two successive binding events of another segment to complete the strand-passing
reaction. The first binding event only activates the enzyme but does not proceed to the strand-passing. The activating
segment dissociates from the complex and only binding of another segment with the activated G segment-bound topo
II results in the strand-passing. If the segment collisions occur more often in knotted molecules than in unknotted,
the requirement of the second collision should enhance the effect of this difference. Analyzing the model its authors
assumed, implicitly, that each strand-passing reaction changes the topology of DNA molecules, so the coefficients ruk
and rku were omitted from the analysis. This omission leads to the conclusion that P (c|u)/P (c|k) should be equal
eq eq
to the equilibrium fraction of knots, fk , rather than fk rku /ruk (see Eq. (6)) [41]. Under the requirement of two
successive collision with a potential T segment for each strand-passing, the authors found that
 
P (c|u) 2  eq 2
fk =
ss
= fk . (13)
P (c|k)
eq
Correspondingly the R-factor for the reduction of knot fraction, Rk , should be equal to 1/fk . For DNA molecules
eq
and ionic conditions used in [9] fk is close to 0.02 (see Fig. 4a), so the analysis predicts the magnitude of the effect
close to the experimental result.
The computer simulation for the model chain in which segments can freely pass through one another shows,
however, that the coefficients ruk and rku cannot be ignored. It was found that for DNA molecules 10 kb in length the
great majority of strand passages in unknotted states do not change the chain topology, while nearly all strand passages
from the knotted states result in unknotting [42]. If we take into account this result and Eq. (6), we have to conclude
that segment collisions have comparable frequency in the knotted and unknotted states of 10 kb circular DNA. The
eq
value of fk is small for such DNA only because the majority of collisions in the unknotted state, if extended to the
strand-passages, would not change the chain topology. Thus, the correct analysis has to account for the missing terms
in the equation for fkss . Using Eq. (6) we obtain that
 
ruk P (c|u) 2 rku  eq 2
fkss = = f . (14)
rku P (c|k) ruk k
The ratio rku /ruk is close to 15 for these molecules [42], so the corrected analysis of the model predicts only small
eq
reduction of fkss relative to fk , by a factor of 3. Clearly, this is not enough.
The kinetic proofreading model can be combined with the model of bent G segment (see the next section). This
means that only juxtapositions with the specific mutual conformations of the colliding segments contribute to the
strand-passing. Although such a combined model predicts sufficiently large R-factors, it would not explain all the ex-
perimental observations (see below). It should be also noted that there is no experimental indication that two collisions
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 127

Fig. 5. Typical simulated conformation of circular DNA molecule 10 kb in length. The size of topo II in this scale corresponds to the black circle in
the figure. The diameter of the chain segments in the figure corresponds to the geometrical diameter of the double helix. The conformation shown
corresponds to a trefoil knot.

with a potential T segment are needed for the strand-passing reaction. The experimental study by Seol et al. found just
the opposite, that a single collision with a potential T segment is needed for the strand-passing reaction catalyzed by
topo II [57]. Still, the kinetic proof-reading model was a first rational theoretical model that suggested a mechanism
of topology simplification by the locally acting enzyme.
The above consideration of the kinetic proof-reading model shows that topology-related properties of random coils
can hardly be analyzed without detailed computer simulation. This point is even more evident for the second suggested
model of the phenomenon.

3.4. The model of bent G segment

The model of bent G segment had two key assumptions [42]. First, it assumed that the enzyme creates a sharp bend
in the G segment upon its binding. Second, the model assumed that the enzyme catalyzes unidirectional transport of
T segment from inside the bent G segment to outside. This directionality of the strand passage is only local, since the
bend can have any orientation relative to the DNA coil. Thus, it is not evident that the model should provide any effect
on the steady state distribution of topological states. The computer simulation shows, however, that this mechanism
can result in a large decrease of the steady state fractions of knots and links relative to the corresponding equilibrium
fractions.
At the time when the model was suggested there were no indications that the G segment is bent in the complex. On
the other hand, it was known that the enzyme catalyzes unidirectional passage of T segments relative to its own body.
So, the second assumption of the model only additionally assumed that in the complex of topo II with the G segment
the entrance for the T segment is located inside the bent G segment (as it shown in Fig. 3).
Evaluation of the model ability to simplify DNA topology was based on Eq. (5) [42]. All four variables of the
equation were estimated directly by the computer simulation of equilibrium sets of DNA conformations. The protein-
bound G segment was modeled by 34 straight segments of the chain, and its bent conformation was kept rigid during
the simulation, while the joints between other segments had their normal flexibility. Typical simulated conformation
with a potential T segment located inside the protein-bent G segment is shown in Fig. 6. Such conformations that had
another segment inside the bent G segment were selected from very large equilibrium sets of simulated conformations
with a particular topology. This allowed direct estimation of P (j |u) and P (j |k). To estimate the values of ruk and rku
the researches calculated topological outcome of the strand-passing reaction for each of the selected conformations.
For DNA molecules 7 kb in length the maximum value of R-factor for knotting/unknotting reaction, Rk , for 180
bend angle in the G segment, was equal to 14.
128 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

Fig. 6. Mechanism of topology simplification in the model of bent G segment. In this simulated conformation of knotted DNA, a potential T segment
is located inside the sharp bend made by the protein-bound G segment (shown by black). Among all of the crossings on this projection only three of
them (marked by the numbers) contribute to the knot. Two of these crossings involve the bent G segment, so the segment facilitates knot localization.
Since conformations with localized knot have an entropic advantage, the bent G segment increases the probability of the juxtaposition of the T and
G segments in knotted molecules, accelerating unknotting by the enzyme.

The mechanism of topology simplification in this model originates from the entropic effect of a knot/link localiza-
tion. It was shown that localization of a knot in a small part of a large closed chain increases the chain entropy because
the rest of the chain receives more conformational freedom [16,58,59]. The sharply bent G segment simplifies such
localization and therefore increases the probability of juxtaposition of T and G segments in knotted molecules. The
effect of G segment bending is well illustrated by the chain conformation shown in Fig. 6. One can see in the figure
that the T segment located inside the protein-bent G segment makes two of three crossings that define the trefoil. This
is nearly always true, if the G segment is sharply bent and has another segment inside the bend. As a result, the trefoil
is nearly localized in a small area of the chain. Similar arguments apply to the linking/unlinking process in this model.
Six years after this model was suggested, Dong and Berger obtained the X-ray structure of the complex between
yeast topo II and a G segment [60]. They found that the G segment in the complex was bent at 150 . They also found
that the enzyme entrance for the T segment is located inside the bend, in full agreement with the second assumption
of the model. Thus, the bent G segment model definitely captures the main mechanism of the phenomenon. Indeed,
the phenomenon of topology simplification directly follows from the fact that the enzyme catalyzes transport of a T
segment from inside the bent G-segment to its outside. Still, the model falls short in explaining all the experimental
findings, suggesting that there is something else there what we do not know yet.
The model predicts that Rk for 7 kb DNA cannot be larger than 14 [42], while Rybenkov et al. found that a few
topo IIs show substantially higher Rk , up to 90 [9]. Similar to this, the steady-state fraction of catenanes created by
the same enzymes were lower than the values predicted by the model. So, the model is not able to explain the whole
magnitude of the effect.
Also, it is necessary to clarify how topo IIs from different organisms are able to simplify DNA topology to dif-
ferent extent [9]. The model of the bent G segment suggests a simple explanation for this observation: enzymes from
different organisms could introduce bends of different magnitude in the protein-bound G segment. It turns out, how-
ever, that this explanation contradicts the experimental data. Two groups performed direct measurements, by atomic
force microscopy, of the bend angles created in DNA molecules by various DNA-bound topo IIs [61,62]. Both studies
found very similar bending of the G segment in complexes with different enzymes.

3.5. The model of hooked juxtapositions

The two discrepancies between the experimental data and the prediction of the bent G segment model, listed above,
are resolved, at least formally, by the model of hooked juxtapositions suggested by Buck and Zechiedrich [43]. These
authors assumed that the enzymes simultaneously bind two segments, G and T, when the segments form a so-called
hooked juxtaposition (Fig. 7). If the hooked juxtapositions appear more often in knotted/linked molecules, the model
should promote unknotting and unlinking. According to the simulation results the model of hooked juxtapositions
predicts large values of Rk , close to the experimental values [49]. We will return to these results in the next section.
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 129

Fig. 7. The hooked juxtaposition of DNA segments.

The model has an important drawback, however. The experimental data show that the enzymes do not bind two
DNA segments simultaneously as the model assumes. Instead, they bind a G segment alone and only then bind a T
segment (see Refs. [7,8], for example). In general, all enzymes studied that interact with two remote DNA sites bind
the sites successively [63]. It is a natural way of assembling complexes between two DNA segments and an enzyme,
since successive binding is much more efficient kinetically. In addition to this, strongly bent segments appear quite rare
in an equilibrium ensemble of DNA conformations, and much more rarely two bent segments appear simultaneously
in the hooked juxtaposition. As a result, the enzymes would wait nearly forever trying to catch a strongly hooked
juxtaposition.

3.6. Merging two models

We now consider a new model which represents an extension of the bent G segment model, or a modification of the
model of hooked juxtapositions. Although the latter model makes the unrealistic assumption that topo II can only bind
the hooked G and T segments simultaneously, there is no need for this assumption. The same topology simplification
is obtained in a more realistic model when a bent potential T segment binds the complex of topo II with the bent G
segment. Indeed, under the assumption that the reaction rate is not limited by the rate of segment diffusion, the rate
is proportional to the equilibrium probability of properly juxtaposed G and T segments. The topology simplification
is determined by the dependence of the required juxtaposition on DNA topology. This dependence on DNA topology
is not affected by the fact that the protein-bound G segment was bent before binding a T segment. Eventually, the
new model requires that two segments form the hooked conformation, but it happens in two steps. In this two-step
pathway, the probabilities of hooked juxtapositions in both knotted and unknotted states will be increased by the same
very large factor. It is important to emphasize here, however, that while the G segment is bent by the enzyme during its
binding with the protein, a T segment has to be caught by the enzyme in the bent and properly oriented conformation.
Only under this condition can the conformation of the T segment enhance the topology simplification. Bending of the
T segment after its binding with the enzyme cannot affect the topological outcome of the strand-passing reaction and,
correspondingly, the steady state distribution of topological states.
Clearly, assembling the complex by binding the protein-bent G segment with the bent T segment has to be many
times faster than assembling in the model of hooked juxtapositions. Still, if the T segment can only be trapped by the
enzyme in the strongly bent and properly oriented conformation, complex formation would be too slow. One way of
making this model more practical is to introduce the two step binding mode of the T segment (Fig. 8). We assume
that there are two binding sites for a potential T segment, on the two sides of the enzyme. The sites are oriented by
such a way that only strongly bent DNA segment can interact with both sites simultaneously. We can further assume
that binding with both sites is a necessary condition for the strand-passing reaction. All binding steps are diagrammed
in Fig. 8. According to this model, a potential T segment first binds with a single binding site, on the left or right
side of the enzyme in the figure. It can then dissociate from the enzyme or, after strong transient bending, bind the
second site of the enzyme. The bending of the T segment will affect the topology simplification in this model only if
the rate constant of dissociation from a single binding site is much larger than the rate constant for binding the second
site of the enzyme (k21  k23 ). In this case the majority of the potential T segments bound to a single site dissociate
from the complex rather than proceed to the strand-passing reaction. Under this assumption the segment binding to a
single DNA site increases the probability of the juxtaposition with a properly bent T segment, but does not affect the
dependence of this probability on DNA topology. If, in addition, the transiently bent conformation shown in position
3 of Fig. 8 decays much faster than binding to the second site occurs (k32  k34 ), the rate of the reaction remains
130 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

Fig. 8. Model of topo II action that includes bending of the T segment. The G segment (yellow) bound with the enzyme (green) is strongly bent.
The T segment (orange) can be bound with two binding sites on the enzyme surface (red) which are arranged so that the T segment can interact
with both of them only in the strongly bent conformation. Therefore, a potential T segment first binds with a single binding site of the enzyme and
waits for a bending fluctuation which allows binding with the second binding site. During this time the segment can dissociate from the complex.
Under the condition that k21  k23 this two-step binding mode for the T segment can increase the ability of the enzyme to simplify DNA topology.
The areas of interaction between the T segment and the enzyme are marked by blue. If, in addition, k32  k34 , the reaction rate is proportional to
the equilibrium probability of the hooked juxtaposition (state 3 in the figure). (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)

Fig. 9. Predicted topology simplification by topo II for DNA molecules of different length. The calculated R-factors for knots (a) and links (b)
correspond to the model that requires hooked juxtaposition of bent G and T segments. The bend angles in each of the segments equal 150 . The
calculation for the knotting/unknotting and linking/unlinking reactions was based on Eqs. (10) and (11), correspondingly. The angle between the
planes of the G and T segments was equal to 90 . The data shown were obtained by the author.

proportional to the equilibrium probability of appearance of a bent segment inside the protein bound G segment.
So, this two-step binding mode of a T segment should not affect the predicted ability of the model to modulate the
topology simplification. If all the above assumptions hold, only the final geometry of the juxtaposed sites matters.
So, the computational analysis used for the hooked juxtaposition model can be directly applied to this new model.
Effectively, in this model the probabilities of the hooked juxtapositions in each topological state will be multiplied by
the same very large factor.
The values of R-factors for unknotting and unlinking of circular DNA, calculated for this model, are shown in
Fig. 9. We see that the effect of topology simplification is very large in general, although it strongly depends on the
size of DNA molecules. The value of Rk can be as large as 1000 for DNA circles of 1.5 kb in length, and exceeds 20
even for 10 kb DNA molecules. The values of Rc are always smaller than Rk for the molecules of the same length. As
one can expect, the effect of topology simplification gradually disappears with increase of DNA length.
Comparison of the model prediction with the experimental data of Rybenkov et al. is shown in Fig. 10. The
bend angle in the G segment is known to be close to 150 [6062], and this value of the angle was maintained in
the simulation shown in the figure. On the other hand, nothing is known about the conformation of the T segment
needed for binding, so we performed the simulation for the various values of this angle, from 0 till 150 . The angle
variation provided the changes of the R-factors in the simulation. One can see from the figure that the simulated
values of R-factors are in reasonable agreement with the experimental data. In addition to this, the values of Rk can
be increased by factor 23 by reducing the angle between the planes of bent G and T segments in the complex [49],
although we found that this angle does not affect the values of Rc .
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 131

Fig. 10. Comparison of the experimental data on the simplification of DNA topology with the theoretical prediction based on the reaction model
that requires hooked juxtaposition of bent G and T segments. The R-factors for the knotting/unknotting reaction catalyzed by various topo II
enzymes, Rk , are plotted versus R-factors for the linking/unlinking reaction, Rc . These values, obtained by [9], vary for various topo II (!). The
computations were performed for 150 bend angle in the G segment and 0 , 60 , 90 , and 150 bend angle in the T segment (F). The angle
between the planes of the G and T segments was set to 90 . The R-factors shown correspond to 7 kb DNA for the knotting/unknotting reaction,
and the linking/unlinking reaction between 7 and 10 kb DNA molecules. The computations were performed by the author.

It is worth noting that the model suggests an additional mode of R factor variation. The simulation results in
Figs. 9 and 10 were obtained under the assumption that k21  k23 (see Fig. 8). If the opposite assumption were
valid, that is k21  k23 , the bending of the T segment and its binding with the second site of the enzyme would
follow the first binding of a potential T segment with the first binding site. Therefore, bending of the T segment
becomes irrelevant for the topology simplification since it cannot affect the topological outcome of the strand-passing
reaction. In simulations, such a situation corresponds to zero bend angle in the T segment. Of course, all intermediate
relationships between k21 and k23 are also possible.
Overall, comparison of the simulation results with experimental data shows that the model can explain, with rea-
sonable accuracy, all major observations. However, there is no experimental data to support the assumptions of the
model related to the binding of the T segment. Clearly, additional experimental studies of the phenomenon are needed.

4. Discussion

Simplification of DNA topology by type II DNA topoisomerases is an important biological phenomenon that helps
unlink newly replicated DNA molecules. It is an amazing physical phenomenon because in this process the enzymes
deliberately change the distribution of topological states of very large DNA molecules that form large random coils
in solution. Topology is a global property and it was not easy to understand how it can be accessed by enzymes that
can only interact with DNA locally. The fact that topoisomerases consume the energy of ATP hydrolysis incorporated
in the reaction cycle does not offer much help here. However, over the years we have achieved remarkable progress
in the investigation of this phenomenon. It is important to emphasize that this progress became possible only due
to complementation between computational and experimental studies. Indeed, the possibility of topology simplifica-
tion is based on very delicate statistical properties of polymer chains which are hardly accessible without detailed
computer simulation. On the other hand, any model requires experimental testing, even if it successfully explains the
phenomenon.
We now understand the major principle used by the enzymes to simplify DNA topology. There is no doubt that
the enzymes achieve this goal by catalyzing strand-passing for local DNA conformations that appear more often in
knotted/linked molecules. Still, details of the topoisomerase mechanism require further investigation.
We believe that the new model which originates from the model of bent G segment and the model of hooked
juxtapositions and described in detail in the previous section, is feasible and can account for the experimental data. It is
important that much part of the model, constituting the bent G segment, has been convincingly proved experimentally.
Of course, only experimental studies can prove or disprove the assumptions of the additional part of the new model,
related to the binding of the second DNA segment with the reaction complex.
132 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

Fig. 11. The diagram of strand-passages in the conformational space of circular DNA. The diagram assumes that only two topological states
of the chain are possible. The gray lines correspond to the chain conformations with at least one pair of juxtaposed segments suitable for an
enzyme. Possible strand-passages are shown by black arrows. Crossings of the internal lines correspond to strand-passages that do not change
the chain topology. (a) Diagram corresponds to an imaginary enzyme I that can catalyze the reaction for any juxtaposition of two segments. The
enzyme-catalyzed reactions are completely reversible, so the arrows cross all lines in both directions. (b) The strand-passing reactions catalyzed
by real topo II. The enzyme-required juxtapositions are asymmetrical, so the possible strand-passages are not reversible. Correspondingly, there
is no line in the conformational space that can be crossed in both directions at the same point. In the panel all lines corresponding to the required
juxtapositions have two different sides, shown by dark and light gray, and the arrows cross them in one direction only.

It is interesting to analyze the strand-passing reactions in the considered models in the conformational space of
circular DNA. The corresponding diagrams are presented in Fig. 11 for an imaginary enzyme (enzyme I) that can
catalyze strand-passages for all possible conformations of the juxtaposed segments, and for real topo II. The DNA
conformational space in both panels is divided into two parts, unknotted chains and trefoil knots. Each strand-passage
that can be catalyzed by enzyme I can be catalyzed in the reverse direction as well (Fig. 11a). A key property of the
real topo II, the unidirectional transferring of a T segment relative to the bent enzyme-bound G segment, means that
reverse strand-passage is impossible. In accordance with this, the reactions that change the chain topology, u k and
k u, originate at the separated areas of the conformational space (Fig. 11b).
If topo II catalyzes strand-passing in the opposite direction, from outside to inside of the bent G segment, it would
increase rather than decrease the steady state fractions of linked DNA molecules relative to the corresponding equi-
librium fractions. From a biological standpoint increasing the fraction of linked circular molecules might be useful in
handling replication of the giant network of mitochondrial DNA of kinetoplastids [64]. These DNA networks contain
thousands of topologically linked circular DNAs 24 kb in length (kDNA). These DNA molecules form a chain mail-
like structure. Each circular DNA is presented by a few dozens of copies, and two daughter networks are not exact
copies of the parental network [6567]. During replication of the kinetoplast DNA (kDNA) minicircles are released
from the network, undergo replication, and then their progeny are reattached to the network. Network reassembly after
minicircle replication has a lot of randomness. Therefore, a topo II enzyme that promotes linking versus unlinking
may prove essential with the progeny reattachment. Several observations on kDNA replication support the suggestion
that the known kinetoplast topo II enzyme has this unusual property. First, the enzymes participate in reattachment
of the newly replicated minicircles to the network but not in their release [68,69]. It was also found that a significant
fraction of kDNA minicircles is knotted [70,71] while the equilibrium fraction of knots for these very small minicir-
cles should be undetectable. The knotting activity of the kinetoplast topo II may be responsible for the appearance of
these knots. It would be very interesting to investigate this issue experimentally.

Acknowledgement

The author thanks A. Grosberg and N. Kallenbach for helpful discussions of the review.

References

[1] Wang JC, Liu LF. DNA replication: topological aspects and the roles of DNA topoisomerases. In: Cozzarelli NR, Wang JC, editors. DNA
topology and its biological effects. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press; 1990. p. 32140.
[2] Alexandrov AI, Cozzarelli NR, Holmes VF, Khodursky AB, Peter BJ, Postow L, et al. Mechanisms of separation of the complementary strands
of DNA during replication. Genetica 1999;106:13140.
[3] Sundin O, Varshavsky A. Terminal stages of SV40 DNA replication proceed via multiply intertwined catenated dimers. Cell 1980;21:10314.
[4] Vologodskii A. Biophysics of DNA. Cambridge, UK: Cambridge University Press; 2015.
A. Vologodskii / Physics of Life Reviews 18 (2016) 118134 133

[5] Rybenkov VV, Vologodskii AV, Cozzarelli NR. The effect of ionic conditions on the conformations of supercoiled DNA. II. Equilibrium
catenation. J Mol Biol 1997;267:31223.
[6] Vologodskii AV, Levene SD, Klenin KV, Frank-Kamenetskii M, Cozzarelli NR. Conformational and thermodynamic properties of supercoiled
DNA. J Mol Biol 1992;227:122443.
[7] Wang JC. Moving one DNA double helix through another by a type II DNA topoisomerase: the story of a simple molecular machine. Q Rev
Biophys 1998;31:10744.
[8] Schoeffler AJ, Berger JM. DNA topoisomerases: harnessing and constraining energy to govern chromosome topology. Q Rev Biophys
2008;41:41101.
[9] Rybenkov VV, Ullsperger C, Vologodskii AV, Cozzarelli NR. Simplification of DNA topology below equilibrium values by type II topoiso-
merases. Science 1997;277:6903.
[10] Frisch HL, Wasserman E. Chemical topology. J Am Chem Soc 1961;83:378995.
[11] Frank-Kamenetskii MD, Lukashin AV, Vologodskii MD. Statistical mechanics and topology of polymer chains. Nature 1975;258:398402.
[12] Rolfsen D. Knots and Links. Berkeley, CA: Publish or Perish, Inc.; 1976.
[13] Vologodskii A. Monte Carlo simulation of DNA topological properties. In: Monastryrsky M, editor. Topology in molecular biology. Berlin,
Heidelberg, New York: Springer; 2007. p. 2341.
[14] Vologodskii AV, Lukashin AV, Frank-Kamenetskii MD, Anshelevich VV. Problem of knots in statistical mechanics of polymer chains. Sov
Phys JETP 1974;39:105963.
[15] Vologodskii AV, Lukashin AV, Frank-Kamenetskii MD. Topological interaction between polymer chains. Sov Phys JETP 1975;40:9326.
[16] Grosberg AY. A few notes about polymer knots. Polym Sci, Ser A 2009;51:709.
[17] Micheletti C, Marenduzzo D, Orlandini E. Polymers with spatial or topological constraints: theoretical and computational results. Phys Rep
2011;504:173.
[18] Wang JC, Schwartz H. Noncomplementarity in base sequences between the cohesive ends of coliphages 186 and lambda and the formation of
interlocked rings between the two DNAs. Biopolymers 1967;5:95366.
[19] Rybenkov VV, Cozzarelli NR, Vologodskii AV. Probability of DNA knotting and the effective diameter of the DNA double helix. Proc Natl
Acad Sci USA 1993;90:530711.
[20] Shaw SY, Wang JC. Knotting of a DNA chain during ring closure. Science 1993;260:5336.
[21] Vologodskii A. Bridged DNA circles: a new model system to study DNA topology. Macromolecules 2012;45:43336.
[22] Vinograd J, Lebowitz J, Radloff R, Watson R, Laipis P. The twisted circular form of polyoma viral DNA. Proc Natl Acad Sci USA
1965;53:110411.
[23] Fuller FB. The writhing number of a space curve. Proc Natl Acad Sci USA 1971;68:8159.
[24] Depew RE, Wang JC. Conformational fluctuations of DNA helix. Proc Natl Acad Sci USA 1975;72:42759.
[25] Pulleyblank DE, Shure M, Tang D, Vinograd J, Vosberg HP. Action of nicking-closing enzyme on supercoiled and nonsupercoiled closed
circular DNA: formation of a Boltzmann distribution of topological isomers. Proc Natl Acad Sci USA 1975;72:42804.
[26] Horowitz DS, Wang JC. Torsional rigidity of DNA and length dependence of the free energy of DNA supercoiling. J Mol Biol 1984;173:7591.
[27] Rybenkov VV, Vologodskii AV, Cozzarelli NR. The effect of ionic conditions on DNA helical repeat, effective diameter, and free energy of
supercoiling. Nucleic Acids Res 1997;25:14128.
[28] Keller W. Determination of the number of superhelical turns in simian virus 40 DNA by gel electrophoresis. Proc Natl Acad Sci USA
1975;72:487680.
[29] Vologodskii AV, Anshelevich VV, Lukashin AV, Frank-Kamenetskii MD. Statistical mechanics of supercoils and the torsional stiffness of the
DNA. Nature 1979;280:2948.
[30] Frank-Kamenetskii MD, Lukashin AV, Anshelevich VV, Vologodskii AV. Torsional and bending rigidity of the double helix from data on
small DNA rings. J Biomol Struct Dyn 1985;2:100512.
[31] Shimada J, Yamakawa H. Moments for DNA topoisomers: the helical wormlike chain. Biopolymers 1988;27:65773.
[32] Klenin KV, Vologodskii AV, Anshelevich VV, Klisko VY, Dykhne AM, Frank-Kamenetskii MD. Variance of writhe for wormlike DNA rings
with excluded volume. J Biomol Struct Dyn 1989;6:70714.
[33] Geggier S, Kotlyar A, Vologodskii A. Temperature dependence of DNA persistence length. Nucleic Acids Res 2011;39:141926.
[34] Wang JC. Untangling the double helix: DNA entanglement and the action of the DNA topoisomerases. Cold Spring Harbor: CSHL Press;
2009.
[35] Wang JC. Interaction between DNA and an Escherichia coli protein . J Mol Biol 1971;55:52333.
[36] Roca J, Berger JM, Harrison SC, Wang JC. DNA transport by a type II topoisomerase direct evidence for a two-gate mechanism. Proc Natl
Acad Sci USA 1996;93:405762.
[37] Trigueros S, Salceda J, Bermudez I, Fernandez X, Roca J. Asymmetric removal of supercoils suggests how topoisomerase II simplifies DNA
topology. J Mol Biol 2004;335:72331.
[38] Stuchinskaya T, Mitchenall LA, Schoeffler AJ, Corbett KD, Berger JM, Bates AD, et al. How do type II topoisomerases use ATP hydrolysis
to simplify DNA topology beyond equilibrium? Investigating the relaxation reaction of nonsupercoiling type II topoisomerases. J Mol Biol
2009;385:1397408.
[39] Martinez-Garcia B, Fernandez X, Diaz-Ingelmo O, Rodriguez-Campos A, Manichanh C, Roca J. Topoisomerase II minimizes DNA entangle-
ments by proofreading DNA topology after DNA strand passage. Nucleic Acids Res 2014;42:182130.
[40] Yan J, Magnasco MO, Marko JF. Kinetic proofreading can explain the suppression of supercoiling of circular DNA molecules by type-II
topoisomerases. Phys Rev E 2001;63(11):031909.
[41] Yan J, Magnasco MO, Marko JF. A kinetic proofreading mechanism for disentanglement of DNA by topoisomerases. Nature 1999;401:9325.
[42] Vologodskii AV, Zhang W, Rybenkov VV, Podtelezhnikov AA, Subramanian D, Griffith JD, et al. Mechanism of topology simplification by
type II DNA topoisomerases. Proc Natl Acad Sci USA 2001;98:30459.
134 A. Vologodskii / Physics of Life Reviews 18 (2016) 118134

[43] Buck GR, Zechiedrich EL. DNA disentangling by type-2 topoisomerases. J Mol Biol 2004;340:9339.
[44] Polikanov YS, Bondarenko VA, Tchernaenko V, Jiang YI, Lutter LC, Vologodskii A, et al. Probability of the site juxtaposition determines the
rate of protein-mediated DNA looping. Biophys J 2007;93:272631.
[45] Catto LE, Bellamy SRW, Retter SE, Halford SE. Dynamics and consequences of DNA looping by the FokI restriction endonuclease. Nucleic
Acids Res 2008;36:207381.
[46] Liu Z, Mann JK, Zechiedrich EL, Chan HS. Topological information embodied in local juxtaposition geometry provides a statistical mechan-
ical basis for unknotting by type-2 DNA topoisomerases. J Mol Biol 2006;361:26885.
[47] Burnier Y, Weber C, Flammini A, Stasiak A. Local selection rules that can determine specific pathways of DNA unknotting by type II DNA
topoisomerases. Nucleic Acids Res 2007;35:522331.
[48] Vologodskii A, Rybenkov VV. Simulation of DNA catenanes. Phys Chem Chem Phys 2009;11:1054352.
[49] Liu Z, Zechiedrich L, Chan HS. Local site preference rationalizes disentangling by DNA topoisomerases. Phys Rev E 2010;81:031902.
[50] Vologodskii A. Simulation of equilibrium and dynamic properties of large DNA molecules. In: Lankas F, Sponer J, editors. Computational
studies of DNA and RNA. Dordrecht: Springer; 2006. p. 579604.
[51] Taylor WH, Hagerman PJ. Application of the method of phage T4 DNA ligase-catalyzed ring-closure to the study of DNA structure. II.
NaCl-dependence of DNA flexibility and helical repeat. J Mol Biol 1990;212:36376.
[52] Vologodskaia M, Vologodskii A. Contribution of the intrinsic curvature to measured DNA persistence length. J Mol Biol 2002;317:20513.
[53] Geggier S, Vologodskii A. Sequence dependence of DNA bending rigidity. Proc Natl Acad Sci USA 2010;107:154216.
[54] Stigter D. Interactions of highly charged colloidal cylinders with applications to double-stranded DNA. Biopolymers 1977;16:143548.
[55] Vologodskii AV, Cozzarelli NR. Modeling of long-range electrostatic interactions in DNA. Biopolymers 1995;35:28996.
[56] Alexander JW. Topological invariants of knots and links. Trans Am Math Soc 1928;30:275306.
[57] Seol Y, Hardin AH, Strub MP, Charvin G, Neuman KC. Comparison of DNA decatenation by Escherichia coli topoisomerase IV and topoiso-
merase III: implications for non-equilibrium topology simplification. Nucleic Acids Res 2013;41:46409.
[58] Katritch V, Olson WK, Vologodskii A, Dubochet J, Stasiak A. Tightness of random knotting. Phys Rev E 2000;61:55459.
[59] Marcone B, Orlandini E, Stella AL, Zonta F. What is the length of a knot in a polymer?. J Phys A 2005;38:L1521.
[60] Dong KC, Berger JM. Structural basis for gate-DNA recognition and bending by type IIA topoisomerases. Nature 2007;450:12015.
[61] Hardin AH, Sarkar SK, Yeonee Seol Y, Liou GF, Osheroff N, Neuman KC. Direct measurement of DNA bending by type IIA topoisomerases:
implications for non-equilibrium topology simplification. Nucleic Acids Res 2011;39:572943.
[62] Thomson NH, Santos S, Mitchenall LA, Stuchinskaya T, Taylor JA, Maxwell A. DNA G-segment bending is not the sole determinant of
topology simplification by type II DNA topoisomerases. Sci Rep 2014;4:6158.
[63] Halford SE, Welsh AJ, Szczelkun MD. Enzyme-mediated DNA looping. Annu Rev Biophys Biomol Struct 2004;33:124.
[64] Shapiro TA, Englund PT. The structure and replication of kinetoplast DNA. Annu Rev Microbiol 1995;49:11743.
[65] Maslov DA, Simpson L. The polarity of editing within a multiple gRNA-mediated domain is due to formation of anchors for upstream gRNAs
by downstream editing. Cell 1992;70:45967.
[66] Thiemann OH, Maslov DA, Simpson L. Disruption of RNA editing in Leishmania tarentolae by the loss of minicircle-encoded guide RNA
genes. EMBO J 1994;13:5689700.
[67] Liu Y, Englund PT. The rotational dynamics of kinetoplast DNA replication. Mol Microbiol 2007;64:67690.
[68] Melendy T, Sheline C, Ray DS. Localization of a type II DNA topoisomerase to two sites at the periphery of the kinetoplast DNA of Crithidia
fasciculata. Cell 1988;55:10838.
[69] Wang Z, Englund PT. RNA interference of a trypanosome topoisomerase II causes progressive loss of mitochondrial DNA. EMBO J
2001;20:467483.
[70] Ryan KA, Shapiro TA, Rauch CA, Griffith JD, Englund PT. A knotted free minicircle in kinetoplast DNA. Proc Natl Acad Sci USA
1988;85:58448.
[71] Sheline C, Melendy T, Ray DS. Replication of DNA minicircles in kinetoplasts isolated from Crithidia Jasciculata: structure of nascent
minicircles. Mol Cell Biol 1989;9:16976.

Das könnte Ihnen auch gefallen