mirkin dna topology

DNA Topology:FundamentalsSergei M Mirkin, University of Illinois at Chicago, Illinois, USA

Topological characteristics of DNA and specifically DNA supercoiling influence all major

DNA transactions in living cells. DNA supercoiling induces the formation of unusual

secondary structure by specific DNA repeats which can also affect DNA functioning.

Introduction

A typical DNA molecule consists of two complementarypolynucleotide chains that are multiply interwound,forming a double helix. In the prevailing conformation,called B-DNA, this is a right-handed helix with a period ofapproximately 10.5 base pairs (bp) per turn at physiolo-gical conditions. Though locally (i.e. for a given sequence)DNA may be very different from the B conformation, thelatter accurately describes the overall structure of a DNAmolecule.

Topological aspects of DNA structure arise primarilyfrom the fact that the two DNA strands are repeatedlyintertwined.Untangling these two strands, which occurs inall major genetic processes may prove rather difficult. Inthe simplest case of a linearDNA in solution, untangling ispossible due to the free rotation of the ends of the DNA.However, for all natural DNAs, free end rotation is eitherrestricted or forbidden altogether. Consequently, untan-gling the two DNA strands becomes topologicallyimpossible. Figure 1 illustrates this for the imaginary case

of a circular DNA molecule where the two strands aretangled only once.A DNA segment constrained so that the free rotation of

its ends is impossible is called a topological domain(Figure 2). A canonical example of a topological domainis circular DNA, which is typical of bacteria, mitochon-dria, chloroplasts, many viruses, etc. In this case, there areobviously noDNA ends at all, since bothDNA strands arecovalently closed. Although eukaryotic chromosomes arelinear overall, they consist of large DNA loops firmlyattached to the nuclear matrix. These loops represent

Article Contents

Introductory article

. Introduction

. Linking Number, Twist and Writhe

. DNA Supercoiling

. Knots and Catenanes

. Supercoil-dependent Structural Transitions in DNA

. Methods of Detection and Analysis

. Role of DNA Topology in Genome Functioning

. Biological Role of Alternative DNA Structures

Figure 1 A hypothetical circular DNA molecule in which two DNA strandsare linked only once. Untangling of the two strands is impossible, unlessone of them is broken.

(a) (b)

(c)

(d)

Figure 2 Examples of topological domains. (a) Circular DNA, (b)chromosomal DNA loops, (c) linear DNA attached to the membrane, (d)linear DNA attached to protein aggregates.

1ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

topological domains, i.e. they are equivalent to circularDNA topologically. The ends of linear DNA can also beaffixed to the membrane, as has been shown for someviruses, making this DNA topologically closed. Finally, astretch of DNA situated between the two massive proteinbodies can also be considered a topological domain. Forsimplicity, the features of topological domains will beconsidered below for circularDNAs, but the principles canbe applied for all other cases.

If strand separation within a topological domain isimpossible, how can intracellular DNA function at all? Toaddress this problem a special group of enzymes calledDNA topoisomerases has evolved. These enzymes intro-duce transient single- or double-strandedbreaks intoDNAto release torsional tension accumulating during strandseparation in a topological domain. Topoisomerases areessential for the resolution of numerous topologicalproblems in DNA. Having them in their arsenal, cells takeadvantage of the topologically constrained nature of theirDNA, as discussed at the end of this article.

Linking Number, Twist and Writhe

The fundamental topological parameter of a covalentlyclosed circular DNA is called the linking number (Lk).Assume that one DNA strand is the edge of an imaginarysurface and count the number of times that the otherDNAstrand crosses this surface (Figure 3). The algebraic sum of

all intersections (which accounts for a sign of everyintersection) is the Lk. Two important features of the Lkare evident from Figure 3. First, Lk is always an integer.Second, Lk cannot be changed by any deformation of theDNA strands, i.e. it is topologically invariant. The onlyway to change Lk is to introduce a break in one or bothDNA strands, rotate the twoDNA strands relative to eachother and seal the break. This is precisely the role of DNAtopoisomerases.Another characteristic of a circular DNA is called twist,

or Tw. Tw is the total number of helical turns in circularDNA under given conditions. Since DNA is a right-handed helix with � 10.5 base pairs (bp) per turn, Tw is alarge positive number for anynaturalDNA.Take a planar,circular DNA and try to locally separate the two DNAstrands, i.e. to decrease the Tw. Since Lk cannot change, adecrease in Tw will be compensated by several positivewrithes of the double helix (Figure 4). Writhing (Wr) is thethird important characteristic of circular DNA, describingthe spatial pass of the double helix axis, i.e. the shape of theDNA molecule as a whole. Wr can be of any sign, andusually its absolute value is much smaller than that of Tw.The above consideration can be formalized by thefollowing equation:

Lk5Tw+Wr [1]

Note, thatwhile Lk is an integer, neither TwnorWr shouldbe such.Also, neitherTwnorWrare topological invariantsand their values easily change with changes in ambientconditions, temperature and during DNA functioning.

DNA Supercoiling

The number of base pairs per DNA turn is designated as g.This parameter can vary depending on the ionic condi-

Figure 3 Linking number represents an algebraic sum of all theintersections made by one DNA strand across the imaginary surface carvedby another DNA strand. The Lk here is +8.

Figure 4 Local unwinding of relaxed circular DNA leads to positive DNAsupercoiling.

DNA Topology: Fundamentals

2 ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

tions, temperature, etc. If the Lk of theN-bp-long circularDNA molecule corresponds exactly to

Lk05Tw05N/g [2]

this DNA molecule is called relaxed. In fact, it will closelyresemble the planar circular DNA shown in Figure 4.

In real DNAs, however, the above equation is almostnever accurate.ADNAmoleculewhoseLkdiffers from theLk0 is called supercoiled. A quantitative measure of DNAsupercoiling is called linking difference (t):

t5Lk2Lk05Lk2N/g [3]

It is clear from the above equation that linking differencecan have either a positive or negative value.When the valueof t is negative, the corresponding DNA is negativelysupercoiled. In this case, the Lk is less than N/g, i.e.negatively supercoiled DNA is somewhat unwoundcompared with the relaxed DNA. When the t value ispositive, DNA is positively supercoiled, and is somewhatoverwound compared with the relaxed DNA.

The term ‘supercoiling’ reflects the shape of the DNA: itlooks like a normal DNA helix coiled into a helix of ahigher order. The two principal configurations of super-coiled DNA, called solenoidal and plectonemic, are shownin Figure 5. The plectonemic (or interwound) supercoil ischaracteristic of DNA in prokaryotes. Solenoidal super-coiling is typical for eukaryotes, where DNA is wrappedaround nucleosomal particles. The sign of a piece ofsupercoiled DNA cannot be determined based on thehandedness of the superhelix. Figure 5 illustrates this forsolenoidal and plectonemic configurations. Both DNAs inthe figure are negatively supercoiled but the plectonemicsuperhelix is right-handed while the solenoidal superhelixis left-handed. This is because to determine the sign of anynode, one should consider thewhole path of theDNA.Thearrows in Figure 5 take the DNA path into account. It isclear that the relative orientation of the DNA segments at

an intersection is the same for both superhelical config-urations.In addition to linking difference, a useful characteristic

of supercoiledDNA is the superhelical density (s), definedas:

s5 t/Lk05 gt/N [4]

It is more convenient to use s, rather than t, to comparesupercoiling between different DNAs, since s is normal-ized forDNA length. At a first approximation,s estimatesthe number of supercoils per helical turn of DNA. Forcircular DNAs isolated from living cells the absolute valueof s may vary between 0.02 to 0.09, i.e. there are 2–9supercoils per 100 helical turns of DNA.The relationships between the linking difference and

twist andwrithe canbedeterminingby combining eqns [1]–[3]:

t5Lk2Lk05 (Tw2Tw0)1Wr5DTw1Wr [5]

The above formula shows that topological stress caused bylinking difference in a circularDNAboth changes the twistfrom its optimal value and introduces writhe. Studyingnegatively supercoiled DNA using an array of differenttechniques it has been demonstrated that, at physiologicalconditions, Wr takes up approximately three-quarters ofthe linking difference, while the remaining quarter goes toDTw. The distribution of torsional stress in positivelysupercoiled DNA remains to be determined.Since supercoiling induces serious torsional and bending

deformations into DNA, it is energetically unfavourable.Extensive experimental and theoretical analysis haveshown that the free energy of negatively supercoiledDNA at physiological conditions is:

DG5 10RTNs2 [6]

where R is the gas constant, T is absolute temperature andN is the length of DNA in base pairs. Since DG isproportional to the square of s, relatively small changes inthe supercoiling density may result in substantial changesin the free energy. One should expect similar dependencefor a positively supercoiled DNA.Because supercoiling is energetically unfavourable, local

DNA changes leading to supercoil relaxation becomefavourable. Consider a 1050-bp-long negatively super-coiled molecule with a linking difference of t5 2 4. Tocompletely relax this stress, it is sufficient to unwind a 42-bp-long DNA segment (four turns of the double helix)within this DNA molecule. This example illustrates twoimportant features of negatively supercoiledDNA.First, achange in a DNA segment corresponding to only a smallpercentage of the total DNA length is sufficient to keep therest of molecule relaxed. Second, under the influence ofnegative supercoiling, DNA would tend to unwind.Positively supercoiled DNA, in contrast, would tend toovertwist.

Figure 5 Plectonemic (upper) and solenoidal (lower) DNA supercoils.Arrows illustrate that both molecules here are negatively supercoiled,despite different handedness.



Knots and Catenanes

Linking number is not the only topologically invariantcharacteristic for circular DNAs. In the process ofcyclization, long DNA molecules can form knots ofdifferent types and complexity. Importantly, after acovalent closure of a DNA molecule, the characteristicsof a knot cannot be changed by any conformationalchanges in DNA short of strand breakage. Thus, the knottype is another topological invariant. Figure 6a shows twoforms of the simplest knot, called a trefoil, of differentsigns. Knots are occasionally detected in living cells. Theyare believed to be the side products of various geneticprocesses, such as recombination.

It is also likely that two or more DNA molecules caninterlink in the process of cyclization.Again, after covalentclosure of those molecules, the linkage type becomesinvariant. Circular DNAs that are linked together arecalled catenanes. Clearly, there are numerous possibletypes of catenanes. Figure6bshows a simple catenane of twopossible signs. Catenanes are routinely detected insideliving cells. They are probably formed at the late stages ofDNA replication and can be subsequently resolved bytopoisomerases.

Supercoil-dependent StructuralTransitions in DNA

As discussed above, local changes in DNA secondarystructure that result in DNA unwinding (decrease the Tw)are energetically favourable in a negatively supercoiledDNA. Extensive studies of these transitions during the lasttwo decades have revealed several distinct DNA confor-mations that are fundamentally different from the cano-

nical B-DNA. These DNA structures are usually calledalternative DNA structures. A common feature of thesestructures is that they are formed by specific DNAsequences, usually of repeated nature, rather than byrandom DNA sequences. The best-studied alternativeDNA structures are considered below.

Cruciforms

Sequence elements called inverted repeats are remarkablywidespread in both pro- and eukaryotic genomes. Theseare DNA segments in which DNA bases that areequidistant from the symmetry centre in a DNA strandare Watson–Crick complements to each other (Figure 7).Under the influence of negative DNA supercoiling thesesequences can form DNA structures called cruciforms(Figure 7). These form when two complementary DNAstrands unpair and then eachDNA strand self-pairs. Thus,topologically, cruciform formation is equivalent to a totalunwinding of the inverted repeat.Since unpairing of a DNA duplex is energy consuming,

this stage represents an energetic barrier for cruciformformation. In addition, cruciforms contain single-strandedbases at their central loops and energetically costlyjunctions with the adjacent duplex DNA (the so-calledfour-way junctions). Altogether, this leads to a high energyfor cruciform formation, approaching 20 kcalmol2 1,which means that cruciform formation is not feasible inlinear DNAs. Because cruciforms are topologicallyequivalent to unwound DNA, however, their formationreleases torsional tension in negatively supercoiled DNA,providing necessary energy. Obviously, the longer aninverted repeat, the more supercoils are relaxed uponcruciform formation. This makes cruciform formation bylong inverted repeats much more thermodynamicallyfavourable than that by short ones (which are only feasibleat very high supercoiling densities). In fact, energeticscalculations show that the probability of cruciformextrusion increases exponentially with the length of aninverted repeat.

Z-DNA

Specific direct repeats that consist of regularly alternatingpurines (denoted R) and pyrimidines (denoted Y),

+ –

Figure 6 Elementary knots (upper panel) and catenanes (lower panel) ofdifferent signs.

Figure 7 B-DNA to cruciform transition. The red and blue strips arecomplementary halves of an inverted repeat. Black strips are adjacent DNA.Fractalized black strips are unwound regions.



d(R-Y)n, can adopt a DNA conformation called Z-DNA(Figure 8). These repeats are much more common ineukaryotic than in bacterial DNA. For example, a repeat(C-A)n.(T-G)n is one of the most common in microsatel-lites in eukaryotic DNA, present in roughly 50 000 copiesper human genome. Z-DNA is the best characterizedalternative DNA conformation since its crystal structurehas been solved at atomic resolution. Although it is adouble helix, it is fundamentally different from B-DNA.

First, Z-DNA is a left-handed double helix with a periodof 12 base pairs per turn. This means that topologically,during the B-to-Z transition, not only do the complemen-tary DNA strands completely unpair, but they also windup in the opposite direction. Thus, transition of n helicalturns of the B-DNA into the Z-conformation shouldrelease 1.8n negative supercoils. This makes formation ofZ-DNA in negatively supercoiled DNA exceptionallyfavourable, even for relatively short repeats. In practice,this structure is favoured only under rather exoticconditions such as very high ionic strength, or methylationof all cytosines in the Z-forming repeat in linear DNA.

Second, the structure is called Z-DNA because of thezig-zag configuration of its sugar–phosphate backbone.Due to the repetitive nature of the Z-forming DNAsequences, there are two defined steps along a DNA chain,YpRandRpY.The relative rotations of the adjacentDNAbases in those two steps are very different: 98 for the YpRstep and 518 for the RpY step. Thus, a sugar–phosphatebackbone at aYpRstep is nearly straight but is followedbya sharp turn at the RpY step. Consequently, thesymmetrical unit in Z-DNA is a dinucleotide, comparedwith a mononucleotide in the B-DNA. This also leads tothe fact that the double helix in the Z-DNA has only onedeep groove, corresponding to the minor groove in the B-DNA.

Two other distinguishing chemical features of Z-DNAare the conformations of the deoxyribose and DNA bases.Deoxyribose adopts the so-called C3’ endo conformationin Z-DNA, compared with the C2’ endo conformation ofdeoxyribose in B-DNA. Purines in Z-DNA are in a syn-conformation, while pyrimidines are in an anti-conforma-tion relative to deoxyribose. Thus, there exists a regularalternation of syn- and anti- base conformations along theDNA chain in the Z-structure. In fact, the requirement for

the Z-forming sequence to be a regular alteration ofpurines and pyrimidines largely depends on the fact thatthe syn-conformation is unfavourable for pyrimidines.

H-DNA

Mirror repeats are DNA segments in which DNA basesthat are equidistant from the symmetry centre in a DNAstrand are identical to each other. A subgroup of theserepeats that are homopurine–homopyrimidine, i.e. con-tain only purines in one DNA strand and only pyrimidinesin the another strand, are called H-palindromes. H-palindromes are enormously overrepresented in eukaryo-ticDNAbut occur at a chance frequency inbacterialDNA.These repeats can adopt an unusual conformation calledH-DNA in a negatively supercoiled state (Figure 9).Themajor element ofH-DNA is an intramolecular triple

helix. To build this structure, a DNA strand from one halfof the repeat folds back, forming a triplex with the duplexhalf of the repeat, while its complement remains single-stranded. As can be seen from Figure 9, the twocomplementary DNA strands are not linked in thisstructure, i.e. topologically, formation of H-DNA isequivalent to the unwinding of the entire homopurine–homopyrimidine stretch. Thus it is favoured in negativelysupercoiled DNA. Since the structure contains an amplesingle-strandedDNA segment, as well as triplex-to-duplexand duplex-to-single strand junctions, its nucleationenergy is rather high, � 18 kcalmol2 1. Consequently, H-DNA is unlikely to form in linear DNA.Depending on the chemical nature of the strand donated

to the triplex, either pyrimidine or purine, there are twosubclasses of H-DNA called H-y or H-r, respectively. TheH-y form is built from TA*T and CG*C1 triads(Figure 10a), where pyrimidines from the third strand aresituated in the major groove and form Hoogsteenhydrogen bonds with the purines of the duplex. The

Figure 8 Transition from the right-handed B-DNA into left-handed Z-DNA by an alternating purine/pyrimidine sequence. The green/red stripsshow the region of alternating purines/pyrimidines. Black strips areadjacent DNA. Fractalized black strips are unwound regions.

Figure 9 H-DNA is formed by homopurine-homopyrimidine mirrorrepeats. A DNA strand from one half of the repeat folds back forming atriplex with the repeat’s duplex half, while its complement remains single-stranded. The black ribbon represents the homopurine strand, the redribbon is the homopyrimidine strand, and green ribbons are adjacent DNA.



extingency for cytosine protonation makes this structurepreferred under mildly acidic pH. The H-r form can bebuilt of CG*G, TA*A and sometimes TA*T triads. In H-rtriads, DNA bases of the third strand form reverseHoogsteen hydrogen bonds with the purines of the duplex(Figure 10b). These triads are stable at physiological pHandare additionally stabilized in the presence of divalentcations.

Other alternative DNA conformations

There are other DNA conformations that could beexpected to occur in negatively supercoiled DNA. Someof them are well defined structurally but yet to be detectedexperimentally in superhelical DNA. Others have beendetected in superhelical DNA, but their fine structure isunclear.

H

G

C+

C

G

G

C

A

T

T

A

T

A

A

T

T

A. Hoogsteen Triads B. Reverse Hoogsteen Traids

Figure 10 H-DNA triads.



G-quartet

This structure (Figure 11a) can be formed by direct repeatscontaining tandemly arranged runs of guanines. Thebuilding elements are stacked G4 runs that are stabilizedby certainmonovalent cations (Figure11b). This structure isdefinitely formed by single-stranded G-rich direct tandemrepeats (such as telomeric repeats in eukaryotes) and isextensively characterized at atomic resolution. However,there are only fragmentary indications that it exists insuperhelical DNA.

S-DNA

Direct tandem repeats of random base composition canadopt a structure called slipped-stranded DNA (S-DNA).This structure (Figure 12) utilizes the multiply-repeatednature of the sequence: upon denaturing and renaturing,the complementary repeats can mispair, resulting in apeculiar combination of double-helical stretches inter-spersed with single-stranded loops. This conformation isthermodynamically unfavourable in linear DNA but canbe trapped kinetically. In superhelical DNA, it mightbecome favourable given the release of substantial

torsional stress. The loops can be stabilized by hydrogenbonds for some of the repeated units, making S-DNA evenmore feasible. However, unambiguous proof of theexistence of S-DNA in superhelical DNA is still lacking.

DNA-unwinding elements

These are extremely AT-rich sequences with a certain biasin the distribution of adenines and thymines between thetwo DNA strands. Other than that, there are no obvioussimiliarities between the sequences of different DNA-unwinding elements. Under the influence of negativeDNAsupercoiling, these elements undergo a transition into astably unwound conformation. This transition is non-cooperative, i.e. the length of the unwound area graduallyincreases with the increase in supercoiling density. Themechanism of this transition is poorly understood. DNA-unwinding elements are often found at the replicationorigins, matrix attachment sites and other importantelements of pro- and eukaryotic genomes.

Methods of Detection and Analysis

DNA supercoiling has been analysed by a variety ofapproaches and only the most common are describedbelow. One of the first methods used historically was thetitration of supercoiling density by sedimentation inethidium bromide sucrose density gradients. This methodis grounded in two facts: (1) because supercoiledmoleculesare more compact than relaxed ones, they sediment fasterthrough a sucrose density gradient; (2) ethidium bromideintercalates between the stackedDNA base pairs, unwind-ing the double helix by 268 per intercalated molecule. Asdiscussed above, DNA unwinding leads to a release ofnegative supercoils. When negatively supercoiled DNA issubjected to centrifugation through the sucrose density

(a) (b)

G

C2 N1

C6N3

C4

C5

O6

HN2H

H

N7 C5

O6

C6

C4

N3

N1

C2

H

N2

H

H

O6

C6

N7

C8

N9

H

R

H N2

H

C2 C4

N3

C6HC5N1

O6

R

R

N9

R C8

H

H

H

N7

N2

H

H

N1 C2

N3

H

C8

C5

C4

N9

N9

G

G

G

K+

N7

C8

G

G

G

G

G

G

G

G

G

G

G

G

G

G

G

G

Figure 11 G-quartet. (a) General overview. The black line is the DNA strand and the purple rectangles are the stacked G-quartets. (b) Structure of a G-quartet.

Figure 12 Formation of slipped-stranded DNA by direct tandem repeats.Upon strand separation, complementary repeats can mispair, resulting in acombination of double-helical stretches interspersed with single-strandedloops. The red and blue strips are complementary strands of a directtandem repeat; the black strips are adjacent DNA.



gradient with increasing concentrations of ethidiumbromide, its sedimentation coefficient first decreases untilall negative supercoils are removed, but then increasesfollowing the accumulation of positive supercoils. Bymeasuring a critical dye concentration required for thecomplete relaxation of a negatively supercoiled DNAsample its supercoiling density can be calculated. This is aprecise but laborious approach and it was subsequentlyreplaced by less complex electrophoretic methods.

Electrophoreticmethods are also based on the differencein shape between supercoiled and relaxedDNAmolecules.Circular DNA molecules become more compact with anincrease in supercoiling density andmigrate faster throughan agarose gel than their relaxed counterparts. Conse-quently, upon separating a mixture of DNA topoisomersin an agarose gel, a ladder of DNA bands can be observed,where the neighbouring bands are chemically identical butdiffer in t by 1. However, the resolution of a standardagarose gel is not sufficient for topoisomers with highdensity of supercoiling, and they co-migrate as a singleband. (This is why a plasmid DNA sample that wouldnormally have s� 2 0.05 migrates as a single band in agel.) To separate these highly supercoiled DNA topoi-somers, agarose gel electrophoresis is performed in thepresence of an intercalator, usually chloroquine. Byunwinding the double helix, this intercalator convertshighly negatively supercoiled topoisomers into less super-coiled ones, allowing their resolution in a gel.

The disadvantage of one-dimensional electrophoresis isthat it does not allow the analysis of complex mixtures ofDNA topoisomers that might simultaneously include bothpositively and negatively supercoiled topoisomers ofvarying densities. This goal can be achieved by usingtwo-dimensional agarose gel electrophoresis. Amixture ofDNA topoisomers is first separated in a standard agarosegel. Here the mobility of both positively and negativelysupercoiled DNA topoisomers increases with an increasein the absolute value of their t until it reaches saturation.Note that upon separation in the first dimension,topoisomers with the same number of supercoils ofopposite sign practically co-migrate. To resolve these co-migrating topoisomers, electrophoresis in a second direc-tion, perpendicular to the first one, is perfomed in thepresence of chloroquine. Since chloroquine unwindsDNA, negatively supercoiled topoisomers become lesssupercoiled and migrate more slowly, while positivelysupercoiled ones gain extra supercoils and migrate morerapidly. Consequently, mobilities of previously co-mi-grated topoisomers of opposite signs become different inaccordwith their actual t. In the arch-like picture shown in(Figure 13a), the right arm represents positively supercoiledtopoisomers while the left arm corresponds to negativelysupercoiled topoisomers.

While the above methods can be used to determine thesupercoiling density of circular DNA, they give littleinformation about the shape of supercoiled DNA mole-

cules. This question can be adequately addressed byelectron microscopic techniques. Electron microscopyallows the actual shape of a supercoiled molecule to beseen and the number of crossings per such amolecule to becounted. The disadvantage of conventional electronmicroscopy is that the DNA conformation may changeduring sample preparation. Also, it is difficult to determinethe sign of a crossing. These problems can be addressed bycryoelectron microscopy. Here, a DNA sample in a watersolution is rapidly cooled to 2 1508C. As a result, DNAmolecules are captured in a thin layer of vitrified water andtheir three-dimensional images can be obtained. Usingthesemethods, it has been shown that the ratioWr/t� 0.75in supercoiled DNA and that a significant fraction ofsupercoiled molecules contain branched structures.The most reliable approach for measuring DNA super-

coiling in vivo is based on the rate of psoralen photobindingto DNA. This compound intercalates into DNA and canform a crosslink with the adjacent pyrimidines in theopposite DNA strands when it absorbs 360 nm light. Sincebinding of an intercalator unwinds DNA, it preferentiallybinds to supercoiled, instead of relaxed DNA. Thus,treatment of cells with psoralen followed bymeasuring theefficiency of psoralen crosslinking provides an estimate ofsupercoiling density in vivo. An essential control here is themeasurement of psoralen photobinding to a completelyrelaxed intracellular DNA, which is usually achieved bysaturating X-ray irradiation of cells.Another valuable approach is based on the efficiency of

the formation of alternative DNA structures in vivo. Forexample by comparing the rate of cruciform formation invitro at physiological conditions with that of a cruciformformation in intracellular DNA (determined as describedbelow) a fair estimate of supercoiling density in vivo can beobtained.

2nd dimension

1st

dim

ensi

on

0+1

+5

+10–10

–5

–1

–15 –20(a)

2nd dimension

1st

dim

ensi

on

0+1

+5

+10–10

–5

–1

–15

–20

(b)

Figure 13 Two-dimensional electrophoretic analysis of a mixture of DNAtopoisomers (a) without local structural transitions and (b) with a localtransition into an alternative DNA conformation. Filled brown circlesrepresent relaxed topoisomers, filled red circles are positively supercoiledtopoisomers and filled blue circles are negatively supercoiled topoisomers.Empty blue circles in (b) show the expected mobility of topoisomers 215to 220 if no transition occurred.



There are also numerous approaches for studyingalternative DNA structures in vitro. A topological way ofdetecting such structures utilizes two-dimensional gelelectrophoresis. If a transition into a non-B conformationoccurs, accompanied by the release of some superhelicalstress, the mobility of the corresponding topoisomerswould decrease. Thus, the topoisomer under transitionwould co-migrate with a less supercoiled topoisomer in thefirst dimension of the electrophoresis. Due to the presenceof an intercalator in the second dimension of electrophor-esis, the superhelical tension is released and alternativeDNA structures convert into the B-form. Subsequently,mobilities of previously co-migrated topoisomers becomedifferent, according to their actual t. A gradual increase oftopoisomer mobility can be seen in the final picture, untilthere is a sharp drop, reflecting the transition (Figure 13b).From its two-dimensional pattern, two important char-acteristics of a structural transition can be obtained: (1) thenumber of supercoils released during transition and (2) thesupercoiling density required for the transition. If thelength of the sequence adopting a new conformation isknown, the number of supercoils will allow the topologicalstatus of this conformation to be deduced.Measurement ofthe supercoiling density will allow the free energy requiredto be calculated.

The formation of alternative DNA structures is accom-panied by the appearance of single-stranded stretcheseither within those structures (as in cruciform or H-DNA)or at their borders with adjacent B-DNA (as in Z-DNA).Thus, these structures can be detected by enzymes andchemicals that specifically recognize single-strandedDNA.These agents are widely used since they allow single-stranded regions to bemapped at a sequence resolution, i.e.they provide fine structural information. Single-strandedDNA-specific nucleases S1 and P1 are most commonlyused among enzymes. Superhelical DNA is treated withunsaturated amounts of a nuclease followed by restrictiondigestion, end-labelling and sequencing gel electrophor-esis. As a result, a pattern of cleavage at a base level can beseen.

There are also convenient chemicals that preferentiallymodify single-stranded DNA bases. Diethyl pyrocarbo-nate carboxyethylates purines at their N7 positions insingle-stranded DNA or the Z conformation. Osmiumtetroxide forms osmate esters with theC5–C6 double bondof single-stranded thymines. Chloroacetaldehyde formsethenoderivatives with the base-pairing positions ofadenines, cytosines and, less prominently, guanines.Potassium permanganate oxidizes single-stranded thy-mines and, to a lesser extent, cytosines. All of thesemodified residues are detectable at a nucleotide resolutionafter piperidine cleavage, followed by sequencing gelelectrophoresis.

The most reliable methods for studying alternativeDNAs in vivo are based on chemical probing. Some of thechemicals described above, including osmium tetroxide,

chloroacetaldehyde, potassium permanganate and psor-alen, penetrate both pro- and eukaryotic cells. IntracellularDNA is isolated after chemical modification and modifiedDNA bases are detected. This can be done either byconventional chemical DNA sequencing or, for chromo-somal DNA, by more sophisticated methods of genomicsequencing.

Role of DNA Topology in GenomeFunctioning

As was mentioned in the beginning of this article,operational units of practically all DNA genomes aretopological domains. Thesemay simply be circular DNAs,typical for essentially all bacteria, mitochondria, chlor-oplasts, etc., or they are large DNA loops attached to thenuclear matrix, as is true for eukaryotic chromosomes.Finally, the ends of linear DNA of some viruses can beaffixed to the cell membrane. This indicates that certainfeatures of topologically constrained DNA made themadvantageous in the course of natural selection.One important feature comes from eqn [1]. It shows that

for a topologically closed DNA domain, any change of asecondaryDNA structure (Tw) is immediately reflected bya change in its overall shape (Wr). Thus, a local change in atopologically closed DNA, say unwinding caused by aprotein binding, will be immediately reflected by a changein supercoiling of the whole molecule and vice versa. Thisgives two important biological benefits. First, if cellularmachinery can sense a link between local and globalchanges in DNA, it can be assured that both DNA chainsare integral, i.e. DNA is not damaged. Another assurancethat DNA is not broken comes from its supercoiling. Asdiscussed above, supercoiled DNA is torsionally stressed,thus the appearance of a sole single-stranded break withinit leads to immediate relaxation. As long as DNA issupercoiled, it does not contain DNA breaks. Clearly it isuseful to know whether DNA is damaged before decisionson its replication aremade.Not surprisingly, replicons are,by-and-large, topologically constrained. Note that thisissue has been most extensively studied in the prokaryoticsystems, where it has been demonstrated unambiguouslythat DNA supercoiling is a prerequisite for replicationinitiation. While some studies hint that the same is true foreukaryotes, more data are needed to support this conclu-sion.The second benefit is that changing DNA structure in

one DNA segment can be instantly sensed at a remoteDNA segment, i.e. topologically constrained DNAs areideally suited for communication at a distance. This isimportant for many genetic processes regulated viainteraction of two or more separated DNA segments.Transcriptional initiation in both pro- and eukaryotes, forexample, is often co-regulated by proteins interacting with



DNA segments separated by hundreds or even thousandsof base pairs. This is certainly out of reach of a directprotein–protein interaction in a linearDNAmolecule. It iswidely assumed, therefore, that some form of topologicalcommunication between those remote DNA segments isinvolved. Similarly, in many instances, genetic recombina-tion occurs between very distant DNA segments.

Besides being topologically closed, DNA in mostinstances is supercoiled. Intracellular DNA in bacteria isactually torsionally stressed and has all the properties ofsupercoiled DNA described above. The situation is morecomplicated in eukaryotes. Supercoils are accumulated inthe chromatin, where DNA is wrapped around nucleo-somes. Thus, unless nucleosomes are removed, thisDNA isnot torsionally constrained. During genome functioning,however, nucleosomes should be either removed orrelocated. It is believed, therefore, that eukaryotic DNAis at least transiently supercoiled. It is revealing to considerthe biological advantages of DNA supercoiling.

DNA is negatively supercoiled inmost living organisms,pro- and eukaryotic. The reason for that is probably thefact that DNA has to be unwound, at least temporarily, toparticipate in all major genetics processes. DNA replica-tion starts from the initial unwinding of the replicationorigin and subsequently expands into the rest of themolecule. During transcription, RNA polymerase un-winds an � 15-bp-long DNA segment at a promoterfollowed by its translocation along the DNA template.Two DNA molecules exchange their individual strands inthe course of homologous recombination, which requiresstrand separation in the partners. Since, as discussedabove, negative supercoiling makes DNA unwindingenergetically favourable, all the above processes shouldbe facilitated by negative supercoiling. Indeed, there isample evidence that DNA replication, transcription andrecombination are stimulated by negative supercoiling.Moreover in some cases, negative supercoiling is abso-lutely required for those processes. Note, however, thatmost reliable data on the role of DNA supercoiling wereobtained for prokaryotic system, where DNA is indeedtorsionally stressed. While there are fragmentary data onthe effects of supercoiling in eukaryotes, the situation thereis more complicated (see above) and warrants furtherstudies.

Interestingly, while basic genetic processes depend onDNA supercoiling, they can, in turn, change the latter. Thebest-studied case of such relationships is the process oftranscription. RNA polymerase unwinds a DNA segmentat a transcription start site and translocates it along thetranscribed gene. This translocation forces DNA to rotatearound the elongating RNA polymerase so that negativeand positive waves of supercoiling are generated upstreamand downstream of it, respectively (Figure 14). This so-called transcriptional supercoiling iswell documentedbothin vitro and in vivo. Since supercoiling waves dissipatequickly, transcriptional DNA supercoiling is principally

dynamic and differs from the steady-state supercoilingdescribed above. Yet it is known to affect structure andfunctioning of genes that are situated at significantdistances from a transcribed DNA segment.Some archaea live at extremely high temperatures, i.e.

under conditions where undesired separation of DNAstrands can happen. Negative supercoiling would onlymake this problem worse in this case. An elegant solutionfor this problem found by those thermophiles is to keeptheir DNA positively supercoiled. Since unwinding ofpositively supercoiled DNA is energetically unfavourable,this provides a topological barrier for the DNA strandseparation.Supercoiling also substantially improves the probability

of juxtaposition of remote DNA segments. It is clear thatmatching of the two distant DNA segments occurs as aresult of a randomDNAwalk. This walk proceeds in threedimensions in a nonconstrained (relaxed circular or linear)DNA molecule, making the collision of distant segmentsrelatively unlikely. Yet this is essentially a one-dimensionalwalk along the superhelix axis in rod-like supercoiledDNAmolecules. Consequently in superhelical DNA, the prob-ability of segment collision increases by roughly two ordersof magnitude. This feature of supercoiled DNA isimportant for genetic processes requiring collision ofdistant DNA segments, and is particularly well documen-ted for site-specific recombination in bacteria.

Biological Role of Alternative DNAStructures

Using the experimental approaches discussed above,alternative DNA structures, including cruciforms, Z-DNA and H-DNA, have been found to exist in vivo.Despite substantial efforts, however, their biologicalfunctions remain elusive. DNA cruciforms have been

Figure 14 A model of transcriptional DNA supercoiling. RNA polymerasetranslocates an unwound DNA segment along the transcribed gene. Thistranslocation forces DNA to rotate around DNA polymerase, so thatnegative and positive DNA supercoiling is generated upstream anddownstream of the enzyme, respectively. The grey circle represents RNApolymerase, the brown ribbons are unwound DNA, the red ribbon is apositive supercoiling wave and the blue ribbon is a negative supercoilingwave. Arrow shows the direction of polymerase movement.



implicated in the initiation of DNA replication, transcrip-tion and recombination in both pro- and eukaryotes. Z-DNA could be involved in the coupling of transcriptionalelongationwithRNAprocessing in eukaryotes. It has beenspeculated that H-DNA may positively regulate theinitiation of transcription and replication, inhibit tran-scription and replication elongation, and serve as a hotspotfor homologous recombination. DNA-unwinding ele-ments are believed to be involved in replication initiation.G-quartets might participate in maintaining the integrityof the chromosomal DNA ends in eukaryotes. Finally,several proteins that specifically recognize these unusualDNAconformations have been isolated and characterized.The best-studied protein of this type is a eukaryoticenzyme, dsRNA adenosine deaminase, that binds Z-DNAwith the highest degree of specificity. Future studies are

needed to distinguish between these numerous and some-times conflicting ideas.

Further Reading

Calladine CR and Drew HR (1997) Understanding DNA: The Molecule

and How it Works. 2nd edn. Sandiego, USA: Academic Press.

Frank-Kamenetskii MD (1997) Unraveling DNA: The Most Important

Molecules of Life. Reading, USA: Addison Wesley.

Sinden RR (1994) DNA Structure and Function. San Diego, USA:

Academic Press.

Travers A (1993) DNA–Protein Interactions. London, UK: Chapman

and Hall.

Vologodskii A (1992) Topology and Physics of Circular DNA. Boca

Raton, USA: CRC Press.



mirkin dna topology

Documents