to become dorsal epidermispredominantly receive signals inthe form of Dpp and Scwhomodimers, which have a broaderdistribution due to their loweraffinity for Sog and Tsg. However,homodimers activate Smads lessefficiently than heterodimers, andthus a biphasic signalling profile isgenerated [3] (Figure 1C).

In addition to the transportsystem, a positive feedbackmechanism is proposed to promoteBMP–receptor interactions at thedorsal midline via an unidentifiedDpp target gene (Figure 1B). Thiscreates a biphasic profile ofBMP–receptor interactions, withheterodimers at the dorsal midlinehaving increased capacity forreceptor binding, whereashomodimer–receptor interactions indorsolateral regions are reduced. Inthis way, the peak of Dpp/Scwsignalling is refined to a tight stripein embryos at the onset ofgastrulation [4] (Figure 1D).

This model leaves at least twooutstanding questions. First, whatis the molecular mechanism forthe increased levels of active Madby Tkv–Put–Sax versus Tkv–Putor Sax–Put receptor complexes?Perhaps Smads are recruited tothe Tkv–Put–Sax receptorcomplex with greater efficiency oran increased stoichiometry.Alternatively, an inhibitor mayexist which is more readilydisplaced from Tkv–Put–Saxcomplexes [9], or Tkv–Put–Saxcomplexes may be preferentiallysorted into distinct endocyticvesicles which favour signalling[10]. Second, what is the Dpptarget gene which promotesBMP-receptor interactions? Highlevels of BMP signalling mayactivate a co-receptor whichincreases the affinity ofBMP–receptor interactions, or aninhibitor of post-transcriptionalreceptor downregulation therebyrestricting downregulation toregions of low signalling [4]. It ispossible that the target genereinforces Dpp/Scw synergy, forexample by stabilising the Saxreceptor.

Mathematical modellingsuggests that Dpp–Scwheterodimers are more robust tochanges in gene dosage thanhomodimers [3]. As heterodimers

are also more potent signallingmolecules, it is likely that otherBMPs function as heterodimersas well. There is somesupporting evidence for this inDrosophila [11,12], as well as invertebrates [13–16].

Formation of the Dpp–Scwgradient in the embryo is anunusual case in that dpp and scwtranscripts are uniform in thedorsal ectoderm. In contrast, in thewing imaginal disk a gradient ofDpp forms from a localised sourcethrough diffusion which isrestricted by heparan sulphateproteoglycans [17]. These distinctmechanisms for creating BMPgradients emphasise theresourcefulness of evolution insolving a biological problem indifferent developmental contexts.

References1. Hogan, B.L. (1996). Bone morphogenetic

proteins in development. Curr. Opin.Genet. Dev. 6, 432–438.

2. Raftery, L.A., and Sutherland, D.J. (2003).Gradients and thresholds: BMPresponse gradients unveiled inDrosophila embryos. Trends Genet. 19,701–708.

3. Shimmi, O., and O’Connor, M.B. (2005).Facilitated transport of a Dpp/Scwheterodimer by Sog/Tsg leads to robustpatterning of the Drosophila blastodermembryo. Cell 120, 873–886.

4. Wang, Y.C., and Ferguson, E.L. (2005).Spatial bistability of Dpp-receptorinteractions during Drosophila dorsal-ventral patterning. Nature 434, 229–234.

5. Podos, S.D., and Ferguson, E.L. (1999).Morphogen gradients: new insights fromDPP. Trends Genet. 15, 396–402.

6. Ashe, H.L. (2002). BMP signalling:visualisation of the Sog protein gradient.Curr. Biol. 12, R273–R275.

7. Eldar, A., Dorfman, R., Weiss, D., Ashe,H., Shilo, B.Z., and Barkai, N. (2002).Robustness of the BMP morphogengradient in Drosophila embryonicpatterning. Nature 419, 304–308.

8. Srinivasan, S., Rashka, K.E., and Bier, E.

(2002). Creation of a Sog morphogengradient in the Drosophila embryo. Dev.Cell 2, 91–101.

9. Dorfman, R., and Shilo, B.Z. (2001).Biphasic activation of the BMP pathwaypatterns the Drosophila embryonicdorsal region. Development 128,965–972.

10. Miaczynska, M., Pelkmans, L., andZerial, M. (2004). Not just a sink:endosomes in control of signaltransduction. Curr. Opin. Cell Biol. 16,400–406.

11. Ray, R.P., and Wharton, K.A. (2001).Context-dependent relationshipsbetween the BMPs gbb and dpp duringdevelopment of the Drosophila wingimaginal disk. Development 128,3913–3925.

12. Kawase, E., Wong, M.D., Ding, B.C., andXie, T. (2004). Gbb/Bmp signaling isessential for maintaining germline stemcells and for repressing bamtranscription in the Drosophila testis.Development 131, 1365–1375.

13. Aono, A., Hazama, M., Notoya, K.,Taketomi, S., Yamasaki, H., Tsukuda, R.,Sasaki, S., and Fujisawa, Y. (1995).Potent ectopic bone-inducing activity ofbone morphogenetic protein-4/7heterodimer. Biochem. Biophys. Res.Commun. 210, 670–677.

14. Nishimatsu, S., and Thomsen, G.H.(1998). Ventral mesoderm induction andpatterning by bone morphogeneticprotein heterodimers in Xenopusembryos. Mech. Dev. 74, 75–88.

15. Schmid, B., Furthauer, M., Connors, S.A.,Trout, J., Thisse, B., Thisse, C., andMullins, M.C. (2000). Equivalent geneticroles for bmp7/snailhouse andbmp2b/swirl in dorsoventral patternformation. Development 127, 957–967.

16. Butler, S.J., and Dodd, J. (2003). A rolefor BMP heterodimers in roof plate-mediated repulsion of commissuralaxons. Neuron 38, 389–401.

17. Belenkaya, T.Y., Han, C., Yan, D., Opoka,R.J., Khodoun, M., Liu, H., and Lin, X.(2004). Drosophila Dpp morphogenmovement is independent of dynamin-mediated endocytosis but regulated bythe glypican members of heparan sulfateproteoglycans. Cell 119, 231–244.

Faculty of Life Sciences, The Universityof Manchester, Oxford Road,Manchester M13 9PT, UK. E-mail: hilary.ashe@manchester.ac.uk

DOI: 10.1016/j.cub.2005.05.003

Dispatch R377

Andrew Travers

One of the most stringentexperimental tests of DNA flexibilityis the formation of small DNAcircles. But how does DNA form asmall circle in solution? A commonview is that the stacking betweenadjacent base pairs can vary within

rather small limits, thus enablingthe narrowing or widening of theDNA grooves that is a concomitantof DNA bending and circularisation.These conformational fluctuationswould also accommodate smallcorrelated changes in DNA twist.As a circle must necessarilycontain an integral number of

DNA Dynamics: Bubble ‘n’ Flip forDNA Cyclisation?

A recent demonstration of the facile in vitro formation of DNAmicrocircles of fewer than 100 base pairs throws new light on the basisof DNA flexibility.

double helical turns, circleformation from a linear moleculerequires that on ligation the twoends be in helical register. And this,in turn, implies that DNA fragmentswith an integral number of turns, n,will form circles much more easilythan those with n±½ turns.

Indeed, such a property waselegantly demonstrated by theclassic experiments of Shore andBaldwin [1], where for fragments of240–250 bp the difference incyclization probability betweenfragments containing n and thosecontaining n±½ turns was ~25 fold.These experiments showed thatboth undertwisting andovertwisting could compensatesymmetrically for twist deficits, andenabled the calculation of a valuefor the torsional flexibility of DNA.

With smaller fragments, both therequired bend angle and twistcompensation for cyclizationincrease so that considering onlysmooth bending and associatedtwisting the formation of circles ofonly ~100 bp should, on currentmodels, be energeticallyimplausible in solution in theabsence of a facilitating protein.Nevertheless, despite the apparentimprobability of such events,Cloutier and Widom [2,3] haverecently shown that circles can beformed relatively easily in vitro, notonly from fragments of 94 and105 bp (9 and 10 double helicalturns, respectively) but, even more

surprisingly, from fragments of 89and 99 bp (8½ and 9½ turns), againwith only an ~25 fold lowerprobability.

These observations stronglysuggest that the behaviour of DNAin solution is more complex thanpreviously modelled. The classictheory of DNA flexibility isessentially an exercise inengineering, considering the longDNA polymer as a moderatelyflexible rod that can both twist andbend [4]. However, although thismodel very successfully predictsthe probability of formation ofmoderately sized circles [5], itignores the highly dynamicproperties of the component base-pairs. In solution, both base-pairseparation and base-flipout occuron a millisecond time scale [6,7].The rates of both these (related)processes depend on the stackingand melting energies of individualbase-steps and their sequencecontext. What, then, is the sourceof enhanced flexibility?

The finger points to the usualsuspect, the base-step TpA, as amajor player [8]. This is the leaststable of all the ten possible base-steps, with a melting energy insolution of only ~0.12 kcal/mole, avalue which is six-fold smaller thanthat of TpG/CpA, the next leaststable step, and some 22-foldlower than that of GpC, the moststable step [9]. Notably, the energyrequired for melting the TpA step is

less than the thermal energy atroom temperature —~0.6 kcal/mole — implying thatconformational fluctuations of thisstep are facile and could dominatethe dynamics. Indeed whenCloutier and Widom [3] substitutedphased TpA steps with TpG theprobability of circularisation for89–105 bp fragments decreased by~12-fold to a value comparable tothat for random sequences.

What structural perturbationscould be responsible for thisphenomenon? A kink in the DNAduplex increasing the local bendangle is one possibility. Indeedlocal kinks or near-kinks restrictedto a single base-step occur atpositions of high DNA curvature inthe nucleosome core particle [10].As predicted by Crick and Klug[11], these kinks are found wherethe minor groove of DNA points intowards the histone octamer[10,12]. Tellingly, in these positionsthere is a preference for the threebase-steps, TpA, TpG/CpA andCpG, with the lowest stackingenergy [13–15]. However, thesekinks are also likely stabilised bythe charge neutralisation of thephosphates on the inside of thekink, thus minimising repulsionbetween closely approachingphosphates. This implies that, infree solution, kinks are at besttransient entities, but they couldnevertheless facilitate tight bendingalthough not necessarily twisting.

For such a situation two othermodels, the kinked worm-like chainmodel (KWLC) and the flexible-hinge bubble model, have recentlybeen proposed (Figure 1) [16,17].Both postulate local lesions whichstrongly enhance both bending andbidirectional torsional flexibility. Isthere any evidence that suchlesions can exist? One indication isthe occurrence in both supercoiledand relaxed DNA of short stretchesof ~3 bp in extent that areaccessible to an endonucleasespecific for single-stranded DNA[8]. These lesions are invariablycentred on a TpA step and notablywhen the sequence contextjuxtaposes a TpA and a relativelyunstable TpG step the accessibleregion increases to ~10 bp. Thisobservation strongly supports theconcept of a flexible-hinge bubble,but it remains to be established

Current Biology Vol 15 No 10R378

Figure 1. Mechanisms for the formation of microcircles.

(A) Smooth bending. (B) Kinking at sites of low stacking energy. (C) Hinge formation atbubble sites.

OR OR

A B CCurrent Biology

whether such bubbles actuallyfacilitate the formation ofmicrocircles.

Although transient bubbles orrelated structures might increasethe probability of forming smallcircles, longer DNA molecules areprobably similarly dynamic.Increased flexibility, however, isnot the universal panacea for eitherDNA wrapping or circularization. Itcomes with a cost. Static bendswhich allow a DNA molecule toeasily adopt an appropriateconfiguration can considerablyenhance both circularisation andnucleosome formation [18].Nevertheless when the flexibility ofa DNA sequence is increased atthe expense of loss of staticbending, DNA wrapping is impairedrather than enhanced [14].

This is presumably because themore flexible a sequence is, themore configurations it can easilyadopt, so that the entropic penaltyon wrapping is consequentlygreater. This implies that theoccurrence of transient,hyperflexible lesions might bedetrimental to the formation oflarger circles, where the averagebend angle per double helical turnis within the normal fluctuations formaintaining the stacking consistentwith smooth bending. As thecalculation of the torsionalflexibility from experimentallydetermined cyclization rates couldinclude a component, eitherpositive or negative, from thehyperflexible lesions the relativecontributions of discontinuous andsmooth fluctuations clearly need tobe refined.

To what extent are theseconsiderations applicable to DNAin chromatin — whether eukaryoticor bacterial? Much of the genomicDNA is packaged, and soconstrained, by non-specific DNAbinding proteins. Yet even whennot complexed with proteins, otherstabilising factors — divalentcations and/or polyamines — couldlimit the dynamic flexibility of DNA.Moreover eukaryotes and bacteriahave both evolved functionallyanalogous but structurally distinctproteins — HMGB proteins ineukaryotes and HU in bacteria —that facilitate circularisation in partby introducing one or more kinksinto the DNA. However, HMGB-

induced circularisation of DNA,even for very small circles, appearsto a first approximation to beindependent of helical phase,implying that these proteins notonly bend the DNA but alsoincrease torsional flexibility [19,20].One possible implication is thatcells adopt an authoritarian attitudetowards their most precious assetand act to suppress dynamicrandom melting events that couldpromote deleterious events such asadventitious transcription initiation.

References1. Shore, D., and Baldwin, R.L. (1983).

Energetics of DNA twisting. I. Relationbetween twist and cyclizationprobability. J. Mol. Biol. 170, 957–981.

2. Cloutier, T.E., and Widom, J. (2004).Spontaneous sharp bending of double-stranded DNA. Mol. Cell 14, 355–362.

3. Cloutier, T.E. and Widom, J. (2005). DNAtwisting flexibility and the formation ofsharply looped protein-DNA complexes.Proc. Natl. Acad. Sci. USA 102, 3645-3650.

4. Doi, M., and Edwards, S.F. (1985).Theory of Polymer Dynamics (Oxford:Oxford University Press).

5. Zhang, Y., and Crothers, D.M. (2003).Statistical mechanics of sequence-dependent circular DNA and itsapplication for DNA cyclization. Biophys.J. 84, 136–153.

6. Pardi, A., Morden, K.M., Patel, D.J., andTinoco, I., Jr. (1982). Kinetics forexchange of imino protons in the d(C-G-C-G-A-A-T-T-C-G-C-G) double helix andin two similar helices that contain a G. Tbase pair, d(C-G-T-G-A-A-T-T-C-G-C-G),and an extra adenine, d(C-G-C-A-G-A-A-T-T-C-G-C-G). Biochemistry 21,6567–6574.

7. Spies, M.A., and Schowen, R.L. (2002).The trapping of a spontaneously‘flipped-out’ base from double helicalnucleic acids by host-guestcomplexation with b-cyclodextrin: theintrinsic base-flipping rate constant forDNA and RNA. J. Am. Chem. Soc. 124,14049–14053.

8. Drew, H.R., Weeks, J.R., and Travers,A.A. (1985). Negative supercoilinginduces spontaneous unwinding of a

bacterial promoter. EMBO J. 4,1025–1032.

9. Protozanova, E., Yakovchuk, P., andFrank-Kamenetskii, M.D. (2004).Stacked-unstacked equilibrium at thenick site of DNA. J. Mol. Biol. 342,775–785.

10. Richmond, T.J., and Davey, C.A. (2003).The structure of DNA in the nucleosomecore. Nature 423, 145–150.

11. Crick, F.H.C., and Klug, A. (1975). Kinkyhelix. Nature 255, 530–533.

12. Hogan, M.E., Rooney, T.F and Austin,R.H. (1987). Evidence for kinks in DNAfolding in the nucleosome. Nature 328,554–557.

13. Thastrom, A., Bingham, L.M., andWidom, J. (2004). Nucleosomal locationsof dominant DNA sequence motifs forhistone-DNA interactions andnucleosome positioning. J. Mol. Biol.338, 695–709.

14. Virstedt, J., Berge, T., Henderson, R.M.,Waring, M.J., and Travers, A.A. (2004).The influence of DNA stiffness uponnucleosome formation. J. Struct. Biol.148, 66–85.

15. Davey, C.S., Pennings, S., Reilly, C.,Meehan, R.R., and Allan, J. (2004). Adetermining influence for CpGdinucleotides on nucleosome positioningin vitro. Nucleic Acids Res. 32,4322–4331.

16. Wiggins, P.A., Phillips, R., and Nelson,P.C. (2005). Exact theory of kinkablepolymers. Phys. Rev. E 71, 021909.

17. Yan, J., and Marko, J.F. (2004). Localizedsingle-stranded bubble mechanism forcyclization of short double helix DNA.Phys. Rev. Lett. 93, 108108.

18. Roychoudhury, M., Sitlani, A., Lapham,J., and Crothers, D.M. (2000). Globalstructure and mechanical properties of a10-bp nucleosome positioning motif.Proc. Natl. Acad. Sci. USA 97,13608–13613.

19. Payet, D., and Travers, A. (1997). Theacidic tail of the high mobility groupprotein HMG-D modulates the structuralselectivity of DNA binding. J. Mol. Biol.266, 66–75.

20. Churchill, M.E., Changela, A., Dow, L.K.,and Krieg, A.J. (1999). Interactions ofhigh mobility group box proteins withDNA and chromatin. Methods Enzymol.340, 99–133.

MRC Laboratory of Molecular Biology,Hills Road, Cambridge CB2 2QH, UK.

DOI: 10.1016/j.cub.2005.05.007

Dispatch R379

Christopher T. Richie and Andy Golden

For more than three decades,phosphorylation of histone H3has been used as a marker forentry of a eukaryotic cell into

meiosis or mitosis: it correlateswith the condensation ofchromosomes that is aprerequisite for their movementand thus proper segregation.Over the years, numerous kinaseshave been implicated in histone

Chromosome Segregation: AuroraB Gets Tousled

Aurora B kinases play important roles during mitosis in eukaryoticcells; new work in Caenorhabditis elegans has identified the Tousledkinase TLK-1 as a substrate activator of the model nematode’s AuroraB kinase AIR-2 which acts to ensure proper chromosome segregationduring cell division.