Current Biology Vol 16 No 19R838

Dispatches

DNA Topology: Dynamic DNA Looping

The DNA in repressive loops is often tightly bent. DNA flexibility imposessignificant constraints on their topology suggesting that they may existas perturbations in plectonemic DNA.

Andrew Travers

The formation of DNA loopsbetween proteins bound to widelyseparated sites can facilitatetranscriptional regulation and DNArecombination. Natural examplesof this regulation include theconstraint of tight repressive DNAloops by the GalR, AraC and LacRproteins [1–5]. These loops areoften short, around 80–250 basepairs, and because DNA istorsionally rigid this means that thetwo binding sites must be on thesame face of the double helix [2,5].In the lac operon two alternativeloops can be formed: onebetween the principal operatorsequence, O1, which overlaps thetranscriptional start site, anda second, weaker operator, O3, 92base pairs upstream; and the otherbetween O1 and another weakeroperator, O2, 401 base pairsdownstream (Figure 1A) [6]. In bothcases, the tetrameric Lac repressorcloses the loop with each of its tworeading heads binding to oneoperator. These reading heads arejoined by two flexible regions,which potentially allow theconformation of the repressor tovary between a V-shape anda more extended structure [7]. Thispotential variability raises twoquestions: does loop configurationaffect angle between heads? Andhow might repressor flexibilityoperate in the context of thebacterial chromatin?

To address these questionsSwigon et al. [8] have recentlycalculated the energy required toform short, bent loops when theDNA binding sites contact thereading heads in an anti-parallel orparallel orientation relative to thepath of the DNA (Figure 1B). Theyfound, using a new model ofsequence-dependent DNAelasticity, that, for a 74 base-pair

loop of relaxed DNA with a helicalrepeat of w10.5 base pairs, certainloop structures, notably those thatform anti-parallel loops with aV-shaped repressor and those witha more open loop and an extendedrepressor, are energeticallypreferred to a structure with a tightloop bound in a parallelconfiguration. The importantconclusion is that DNA flexibilityper se can, in isolation, determineboth loop and repressorconformation.

How then do these calculationsrelate to the natural environment ofa lac repressor loop, or, indeed,other repression loops? Normally,the Escherichia coli chromosome isnegatively supercoiled in vivo.In vitro, loop stability is enhancedby negative superhelicity [4,9],

which concomitantly increasesthe preferred helical repeat froma relaxed value of 10.5 base pairsto w10.8–10.9 base pairs [10].Similarly, in vivo the helical repeatof both the lac loop [11,12] andthose of loops formed by theAraC protein [13] and the Hininvertasome [14] are increasedto w11.1–11.3 base pairs. Thisvalue indicates strongly that thesetight loops have the configurationof a right-handed toroidcorresponding to the interwindingsof the plectonemic form ofsupercoiled DNA, rather than thatof the left-handed toroidal form (forwhich the helical repeat wouldbe less than the relaxed valueof w10.5 base pairs) [15].

In vivo, plectonemes are likelydynamic structures, oscillatingbetween an extended tighthigh-pitch configuration anda more compact open low-pitchform. They are thus essentially

Flexible hinges

Reading heads

B C

Antiparallel orientation

V-shaped repressor

Parallel orientation

V-shaped repressor

Parallel orientation

Extended repressor

AO2O0O3

92 bp 401 bp

Current Biology

Figure 1. DNA loop formation by interactions between bound proteins.

(A) Organisation of operators in the lac operon. (B) Structure of the tetrameric Lacrepressor showing reading heads and flexible regions. (C) Alternative configurationsof constrained loops and Lac repressor (reproduced with permission from [8]).

DispatchR839

both deformable and mobile andcan thus adapt their local structureto constraints imposed by thebinding of different proteins. In thiscontext, a simple solution would befor the LacR tetramer bound to O1

and O3 to constrain a right-handedsuperhelical loop as a continuationof the helical path of one of thetwo DNA duplexes making up theinterwindings. The resultingconstrained loop would remain anintegral part of the plectonemicstructure (Figure 2). For thealternative, longer loop formedbetween O1 and O2, theorganisation would be essentiallysimilar, but the large loop mightitself form an interwound structure.In this situation, LacR wouldstabilise a branch point in the largerplectoneme. Alternatively, LacRmight bridge between the twooperator sites now situated on bothof the interwinding duplexes andseparated by an apex.

The stabilisation of one turn ofan interwinding duplex is largelyconsistent with the describedcharacteristics of DNA binding byLacR. The bending of DNA awayfrom the two reading heads wouldallow the tetrameric repressor tostabilise the coherent bend formedby one complete interwoundturn [5]. This would correspondto the parallel configuration ofreading heads. Energetically, ascalculated by Swigon et al. [8], thisarrangement is disfavoured if theDNA has a helical repeat of w10.5base pairs. In contrast, atomicforce microscopy of the LacR(and GalR) repressor bound toa supercoiled DNA minicircle, butwith a separation of 197 base pairsbetween operators (correspondingto a helical repeat of 10.9 basepairs), suggested that the readingheads adopted an anti-parallelbinding configuration [16]. In vivo,however, the formation of tightloops, or even in this case possiblycomplete superhelical turns, isfacilitated by the HU proteins,which increase the bendingflexibility of DNA and thussubstantially reduce the requiredbending energy [17,18].

These considerations also reveala rationale for the flexibility of theLacR tetramer. In vivo, thesuperhelicity of DNA is quitevariable, changing in response to

Apex

PlectonemeToroid

Interwindings

Current Biology

Figure 2. Model for the O1 and O3 lac repression loop within a negatively supercoiledplectonemic DNA molecule with low pitch.

The variant shown introduces an additional turn into the path of one of the interwindingduplexes. The second duplex could also in principle pass within the constrained loop.The inset shows the organisation of plectonemic and toroidal forms of negativelysupercoiled DNA.

both growth phase and stress.Such changes could alter both thehelical repeat of DNA and the pitchof the interwindings in theplectoneme. In this situation, therepressor could respond to at leastsmall changes in superhelicaldensity by altering the hinge anglesand so maintaining the alignment ofthe reading heads with the operatorsites and, in turn, maintainingrepression. Possibly of moresignificance is the breaking of therepression loop on induction andits subsequent reestablishment onexhaustion of lactose.

Using a model system for O1 andO3, in which the strength of theoperator sites was inverted fromthe natural situation so that thestronger of the two operatorsoccupied a position equivalentto O3 relative to the promoter,Becker et al. [18] showed that theefficiency of induction was stilldependent on operator spacing,but unexpectedly the helical repeatdependence now had a lowervalue. This indicates that theaverage topology of the loopchanges in the presence ofinducer. In the natural promoter O3

both has the lower affinity for LacRand also overlaps an upstreamcontact for RNA polymerase [19].As the toroidal topology of DNA

constrained by polymerase and itsactivators can differ substantiallyfrom that of the repressionloop [20], the initial binding ofpolymerase could facilitate thebreaking of the loop while theflexibility of the LacR hinges couldallow the loop to reform easilyif the polymerase binding wereunproductive. Put another way, theflexible hinges in the repressorbuffer conformational fluctuationsin the repression loop to maintainoptimal regulation.

References1. Irani, M.H., Orsorz, H., and Adhya, S.

(1983). A control element withina structural gene: the gal operonof Escherichia coli. Cell 32,783–788.

2. Dunn, T.M., Hahn, S., Ogden, S., andSchleif, R.F. (1984). An operator at - 280base-pairs that is required for therepression of the araBAD operator:addition of DNA helical turns between theoperator and promoter cyclically hindersrepression. Proc. Natl. Acad. Sci. USA 81,5017–5020.

3. Mossing, M.C., and Record, M.T., Jr.(1986). Upstream operators enhancerepression of the lac promoter. Science233, 889–892.

4. Borowiec, J., Zhang, L., Sasse-Dwight, L.,and Gralla, J. (1987). DNA supercoilingpromotes formation of a bent repressionloop in lac DNA. J. Mol. Biol. 196,101–111.

5. Kramer, H., Niemoller, M., Amouyal, M.,Revet, B., von Wilcken-Bergmann, B., andMuller-Hill, B. (1987). lac repressor formsloops with linear DNA carrying twosuitably spaced lac operators. EMBO J. 6,1481–1491.

Current Biology Vol 16 No 19R840

6. Reznikoff, W.S., Winter, R.B., andHurley, C.K. (1974). The location of therepressor binding sites in the lacoperon. Proc. Natl. Acad. Sci. USA 71,2314–2318.

7. Lewis, M., Chang, G., Horton, N.C.,Kercher, M.A., Pace, H.C.,Schumacher, M.A., Brennan, R.G., andLu, P. (1996). Crystal structure of thelactose operon repressor and itscomplexes with DNA and inducer.Science 271, 1247–1254.

8. Swigon, D., Coleman, B.D., andOlson, W.K. (2006). Modeling the Lacrepressor-operator assembly: theinfluence of DNA looping on Lacrepressor conformation. Proc. Natl. Acad.Sci. USA 103, 9879–9884.

9. Whitson, P.A., Hsieh, W.T., Wells, R.D.,and Matthews, K.S. (1987). Supercoilingfacilitates lac operator-repressor-pseudooperator interactions. J. Biol.Chem. 262, 4943–4946.

10. Kramer, H., Amouyal, M., Nordheim, A.,and Muller-Hill, B. (1988). DNAsupercoiling changes the spacingrequirement of two lac operators for DNAloop formation with lac repressor. EMBOJ. 7, 547–556.

11. Law, S.M., Bellomy, G.R., Schlax, P.J.,and Record, M.T., Jr. (1993). In vivo

Growth Control: pAngel of CompenProliferation

Compensatory growth, or regenerattissue during animal development. Rfor Drosophila p53 in the compensatneeded to repair damaged tissues, afunction of the caspase protease Dr

Eric H. Baehrecke

Organ and tissue size is regulatedby cell proliferation, cell growthand cell death during animaldevelopment [1]. Remarkably,tissues with severe damage arecapable of regeneration duringdevelopment through a processof compensatory growth, resultingin tissues and structures of normalsize and pattern [2]. Such damageis triggered by cellular insults,including DNA damage, thatactivate programmed cell death(apoptosis). p53 is a key regulatorof the response to genotoxicstress, and its importance insuppressing tumor formation isunderscored by its inactivation inmany human cancers [3]. p53responds to cellular insults eitherby arresting the cell cycle, so thatDNA can be repaired, or by

thermodynamic analysis of repressionwith and without looping in lac constructs.Estimates of free and local lac repressorconcentrations and of physical propertiesof a region of supercoiled plasmid DNAin vivo. J. Mol. Biol. 230, 161–173.

12. Muller, J.D., Oehler, S., and Muller-Hill, B.(1996). Repression of lac promoter asa function of distance, phase and qualityof an auxiliary lac operator. J. Mol. Biol.257, 21–29.

13. Lee, D.H., and Schleif, R.F. (1989). In vivoDNA loops in araCBAD: size limits andhelical repeat. Proc. Natl. Acad. Sci. USA86, 476–480.

14. Haykinson, M.J., and Johnson, R.C.(1993). DNA looping and the helical repeatin vitro and in vivo: effect of HU proteinand enhancer location on the Hininvertasome. EMBO J. 12,2503–2512.

15. Cozzarelli, N.R., Boles, T.C., andWhite, J.H. (1990). DNA topology and itsbiological effects, N.R. Cozzarelli andJ.C. Wang, eds. (New York: Cold SpringHarbor Laboratory Press), pp. 139–184.

16. Virnik, K., Lyubchenko, Y.L.,Karymov, M.A., Dahlgren, P.,Tolstorukov, M.Y., Semsey, S.,Zhurkin, V.B., and Adhya, S. (2003).‘‘Antiparallel’’ DNA loop in Gal

53, the Guardiansatory

ion, is used to replace damagedecent work has revealed a new roleory proliferation of cells that arerole that requires the non-apoptoticonc.

triggering apoptosis [4]. A recentstudy published in Current Biologyby Wells et al. [5] implicatescell-death regulators, includingp53, in compensatory proliferationand regeneration of damagedtissues during development of thefruit fly Drosophila melanogaster.The discovery of cell-deathregulators that promote cellproliferation and regenerationprovides a new twist in ourunderstanding of the mechanismscontrolling life and death decisionsduring animal development.

Compensatory cell proliferationis used to repair developing adultstructures (imaginal discs) inDrosophila following the inductionof cell death [6–8]. ‘Undead cells’can be created in developing flytissues by activating cell deathwhile protecting against thedemise of these cells with

repressosome visualized by atomicforce microscopy. J. Mol. Biol. 334,53–63.

17. Van Noort, J., Verbrugge, S., Goosen, N.,Dekker, C., and Dame, R.T. (2004). Dualarchitectural roles of HU: formation offlexible hinges and rigid filaments. Proc.Natl. Acad. Sci. USA 101, 6969–6974.

18. Becker, N.A., Kahn, J.D., andMaher, L.J., III. (2005). Bacterialrepression loops require enhanced DNAflexibility. J. Mol. Biol. 349, 716–730.

19. Buckle, M., Buc, H., and Travers, A.A.(1992). DNA deformation in nucleoproteincomplexes between RNA polymerase,cAMP receptor protein and the lac UV5promoter probed by singlet oxygen.EMBO J. 11, 2619–2625.

20. Maurer, S., Fritz, J., Muskhelishvili, G.,and Travers, A. (2006). RNA polymeraseand an activator form discretesubcomplexes in a transcription initiationcomplex. EMBO J., in press.

MRC Laboratory of Molecular Biology,Hills Road, Cambridge CB2 2QH, UK.E-mail: aat@mrc-lmb.cam.ac.uk

DOI: 10.1016/j.cub.2006.08.070

expression of p35, an inhibitorof caspase proteases. This resultsin overgrowth of tissues becauseof compensatory proliferationof cells that is associated withectopic expression of thepatterning regulators Wingless(Wg) and Decapentaplegic (Dpp).Additionally, Jun N-terminal kinaseand the initiator caspase Dronchave been implicated in theregulation of overgrowth, butseveral mysteries about thistissue repair mechanism remainunsolved.

Wells et al. [5] recognized thatthe creation of large regions(compartments) of undead cellsin fly wing imaginal discs causesa 3–4 day developmental delayduring the third larval instar stage.In addition, wing imaginal discsthat contain undead cells undergogrowth arrest specifically duringthe third larval instar such thatthese tissues are smaller thanthose in control animals. Growtharrest in imaginal discs containingundead cells is transient, however,as these tissues can becomenearly 50% larger than controlsprior to the onset of pupariation.This prompted the authors toinvestigate the influence of undeadcompartments of cells on celldivision in wing imaginal discs.They discovered that undead