Thermodynamics of the ColE1 cruciform

.I. MOZ. Biol. (1984) 180, 179-200

Thermodynamics Comparisons

DAVID

of the ColEl

Cruciform

Between Probing and Topological Using Single Topoisomers M. J.

LILLEY

AND

LAURENCE

R.

Experiments

HALLAM

Department of Biochemistry, Medical Sciences Institute University of Dundee, Dundee DDl 4HN, U.K. (Received 8 May 1984, and in revised form 6 August

1984)

The sensitivity of the ColEl cruciform to four enzyme and chemical probes of secondary structure has been studied as a function of plasmid topology. Purified topoisomers of pColIR515 have been probed with S, nuclease, BnZ31 nucleasr. phage T4 endonuclease VII or osmium tetroxide, and site-specific reaction quantified. Closely similar profiles of reactivity as a function of linking difference were obtained for each probe. Electrophoresis of the pure topoisomers on polyacrylamide/agarose gels revealed a discontinuity in migration as a function of linking difference. Above a threshold linking difference, topoisomers exhibit’ pronounced reduction in mobility. The linking difference at which this band shift is found correlates precisely with that required for site-specific reaction with the four probes. We conclude that both probing and topological methods are valuable in the study of cruciform structure in supercoiled DNA. The band shift has been measured with accuracy to allow the calculation of t,he t,wist, change that accompanies the transition, corresponding to ATw = -3.2+0.1. Using this value t’ogether with the critical linking difference we calculate a free energy of formation for this structure AC = l&4+0.5 kcal mol-’ (1 kcal = 4.184 kJ).

1. Introduction DNA underwinding, or negative supercoiling (Vinograd et al., 1965), is very widespread in nature. This is certainly true for t,he prokaryotes, and their plasmids and bacteriophages, and may be significant in eukaryotes also (Lilley, 1983a). Linkage reduction, characterized by a linking difference (AU), is simply related to deviations in helical twist (ATw) and tertiary writhing ( Wr) according t’o (Vinograd & Lebowitz, 1966; Fuller, 1971):

ALk = ATw+ Wr,

(1)

i.e. DNA underwinding must result in structural changes partitioned according to t,he above equation. The process by which DNA linkage is reduced from its equilibrium value is, by definition, energy-requiring, and this negative supercoiling can be regarded as an energy reservoir for driving structural perturbations. Structural change may be global, writhing the entire duplex axis 00””

2836/84/3:30179-~2%

$03.00/O

179

0

1984 Academic

Press Inc. (London)

Ltd.

180

D. M. J. LILLEY

AND

L. R. HALLAM

for example, or it may be quite localized. It is clear from equation (1) that any process leading to a localized negative ATw value will be favoured in energetic terms by negative supercoiling. This could be a reduction in the helical twist or a local loss of duplex to form a melted “bubble”. An extreme example of the former is the formation of Z-DNA (Wang et al., 1979) which, being left-handed, imparts a considerable negative ATw value, and this transition is promoted by negative supercoiling (Peck et al., 1982; Singleton et al., 1982). An example of the latter process is the premelting that occurs in supercoiled DNA (Vinograd et al., 1968). A more biologically relevant, analogous, process is promoter isomerization to the open complex (Walter et al., 1967), and several promoters have been shown to be sensitive to supercoiling levels (Javor, 1974: Sanzey, 1979; Yang et al., 1979; Wahle & Mueller, 1980; Smith, 1981). The formation of cruciform structures (Platt, 1955; Gierer, 1966) by inverted repeat sequences is topologically very similar to melting, in that a ATuj value is expected corresponding to the effective removal of the entire section of B-DNA. Cruciform formation in supercoiled DNA was therefore predicted as a theoretical possibility (Hsieh & Wang, 1975; Woodworth-Gutai & Lebowitz, 1976; Benham: 1982; Vologodskii & Frank-Kamenetskii, 1982), subject to complete uncertainty with regard to the free energy of formation of the structure. The first experimental demonstrations of cruciform structures resulted from two quite unrelated approaches. Large inverted repeat sequences were constructed and supercoiled in vitro, and topological methods used to demonstrate the resulting ATu) value directly (Gellert et al., 1979; Mizuuchi et al.. 19826). The cruciform structures extruded were large enough, moreover, to visualize by electron microscopy. Short)er inverted repeats const’ructed in plasmids that could be propagated in vuiuo have been used recently in conjunction with two-dimensional gel electrophoresis to estimate free energies for the formation of cruciform from t’hese sequences (Courey Br Wang, 1983; Gellert et al., 1983). The presence of cruciform st,ructures generated from short inverted repeats in natural DNA was first inferred by experiments in which supercoiled plasmids or phage were probed using single-&and-specific nucleases such as S, nuclease (Lilley. 1980; Panayotatos & Wells. 1981). Site-specific cleavage occurred tightly localized to the centres of the inverted repeats, i.e. in the loops of the putative cruciform structures. Probing experiment’s were later ext!ended t,o include other singlestrand-specific enzymes (Dingwall et al., 1981; Lilley. 19835), single-strandselective chemical reagents (Lilley, 1983c; Lilley &, PaleEek. 1984) and a resolving enzyme (Lilley & Kemper. 1984). The Escherichia coli plasmid ColEl (Bazaral & Helinski, 1968) contains a 13 bpt invert,ed repeat, which has proved consistently to be the best substrate for each enzyme and chemical probe employed. In addition, t,opological band shift studies on the small deletant, plasmid pAO3 (Oka et nl., 1979) have revealed writhing changes consistent with the ext)rusion of a small cruciform (Dean et al., 1983: Lyamichev et ab., 1983; Otter et al., 1983). Using a dimer of t’he mini-co1 plasmid pVH51 (Hershfield et aZ., 1976), Singleton & Wells (1982) have demonstrated that a threshold level of supercoiling is required for S, nuclease sensitivity. In the t Abbreviation

used: bp, base-pair(s)

ColEl

CRUCIFORM

THERMODYN$MICS

181

present study we have set out to improve our understanding of the energetics of cruciform formation, and to see if self-consistency of results is possible, using a caombined probing and topological approach. More specifically. we have asked two questions. (1) Are the topological, dependences of enzyme and chemical probing and t,opological bandshift methods identical? Can all these data, irrespective of source. be built into a single model for the ColEI cruciform! (2) What are the thermodynamics of extrusion of the ColEl cruciform, i.e. can w d(4ermine ATw and calculate a AG value for the process? By this approach we have advanced our understanding of the ColEl cruciform from a physical chemical viewpoint

2. Experimental (a) Construction

and

preparation

of pColIR515

pColIR515 (Lilley & Hallam, 1983) was prepared from pColIR215 (Lilley, 1981), a rerombinant plasmid that was constructed by ligation of a 440 bp fragment of ColEl containing the 13 bp inverted repeat between the EcoRI and BamHI sites of pBR322. The size reduction to 2740 bp was achieved by cleavage with BamHI and PwuII, filling in the resulting termini using DNA polymerase I (Bethesda Research Laboratories) and bluntend ligation, before transformation into E. coli K12 HBlOl with selection for Ap’. This procedure resulted in a BamHI site being recreated on ligation. Supercoiled pColIR515 was prepared by chloramphenicol amplification of logarithmically growing cells for 16 h, lysis by lysozyme, EDTA and Triton X100, and isopycnic centrifugation in caesium chloridr and ethidium bromide. Plasmid DNA was recovered by side puncture, extraction with butan-l-01, extensive dialysis against 10 mM-Tris HCl (pH 7.5). 1 mrv-EDTA and finall>1)recipitation with ethanol. (b) Preparation

of topoisomers

of pColIR515

Topoisomer distributions of various mean linking differences (AU) were prepared by incubation of supercoiled pColIR515 with topoisomerase I in the presence of appropriate concentrations of ethidium bromide. Topoisomerase reactions were performed in 10 mM‘I‘ris . HCI (pH &O), 200 m&I-NaCl, 20 mM-EDTA. at 25°C for 30 min. Chicken reticulocyte topoisomerase I was a gift from T. Kimura. Single t,opoisomers were derived from distributions prepared as described above. After rlectrophoresis in 1% (w/v) agarose gels containing appropriate chloroquine concentrations for optimal resolution of the topoisomers required, individual topoisomer bands, visualized by a very brief ethidium bromide staining and exposure to 300 nm light, were excised and electroeluted in dialysis sacs. The buffer contents were then passed through small DEAEc.ellulose (DE52. Whatman) columns in TBE buffer (90 mM-Tris-borate (pH 8.3). 10 mMEDTA), whereupon the DNA bound to the resin. The topoisomer DNA was eluted from the caolumns by 2 M-N&l and precipitated with ethanol. These procedures yield extremely purr 1)S.i. which is readily cleaved by restriction enzymes. (c) Measurement

of linking

differences

Topoisomer linking differences, or mean linking differences in the case of topoisomer distributions, were established by band counting methods (Keller 1975). Ethidium bromide concentrations between 0 and 4.55 pg/ml were used in topoisomerase incubations to derive 25 distributions of different mean linking difference between relaxed and above-nativr

182

1). M. J. LILLEY

AND

L. R. HALLAM

supercoiling, to provide sets of topoisomers for band counting between these points. These were electrophoresed on 7 different agarose gels containing between 0 and 2.5 pg chloroquine/ml, and correspondence between gels was established by microdensitometry of equivalent lanes. In t,his way a totally unambiguous set of bands could be counted over the entire region, and hence linking differences of any topoisomers measured with complete awuracy. For topoisomer distributions. mean linking differences were calculated by the algorithm of Kolb & But (1982). By this method the mean linking difference of native pColIR515 was established to be - 19.2, corresponding to a superhelix density of -0.074. (d) S, n&ease S, nuclease was obtained from Bethesda Researrh Laboratories and diluted to give a 3550 units/ml stock solution. Incubations werp performed in 50 mM-sodium acetate (pH 4.6). 50 mM-N&l. I mM-ZnCl, at 15°C for 15 min. in reaction volumes of 2Opl containing 3 units of S, nuclease. (e)

Bal31

nucleasv

B&31 was obtained from New England Biolabs and diluted to a 30 units/ml stock solution. Incubations were performed in 20 mw-Tris HCl (pH 8-O), 600 mM-NaCl, 12 mMMgCl,, 12 mM-CaCl,, at 15°C for 15 min. Reaction volumes of 30 ~1 contained 0.06 unit of BaZ31. and reactions were terminated by addition of EGTA to a final concentration of 20 rnM. (f) Phage

TJ

endonuclease

VI1

Phage T4 endonuclease VII was a gift from B. Kemper. Incubations were performed in 10 mM$mercaptoethanol, 250 pg bovine serum 50 mM-Tris. HCl (pH %O), 10 mM-MgCl,, albumin/ml at 25°C for 60 min. Reaction volumes of 20 ~1 contained 500 units of T4 endonuclease VII. (g) Osmium

tetroxide

Osmium tetroxide (Johnson-Matthey) was dissolved as an 8 mM stock solution in glassdistilled water. Modification reactions were performed in 50.~1 volumes containing 10 mMTris. HCl (pH 7.8), 1 mM-EDTA, 1% (v/v) py ri d ine and 1.6 mw-osmium tetroxide at 25°C for 10 min. Reactions were terminated by extraction into chloroform. After restriction cleavage, chemically modified DNA was digested by incubation with 5 units of S, nuclease (Bethesda Research Laboratories) in 20 ~1 of 50 mM-sodium acetate (pH 4.6), 50 mM-NaCl. 1 mM-ZnCl, at 37°C for 60 min. (h) Restriction Restriction enzymes were obtained directed by the manufacturer.

enzymes

from Bethesda

Research

Laboratories

and used as

(i) Gel electrophoresis Gel electrophoresis was performed in 1% (w/v) agarose gels at ambient temperature for 16 h in TBE buffer. When resolution of topoisomers was required, gels were prepared and run in TBE buffer containing an optimal concentration of chloroquine diphosphate (Sigma) (Shure et aZ., 1977). Composite 2% (w/v) polyacrylamide/0.5% (w / v ) a g arose gels were prepared according to Peacock & Dingman (1968) and Panyutin et al. (1982). The 0.5 cm thick gels were run in 36 mw-Tris . HCl (pH 7.7), 40 mM-NaH,PO,, 1 mM-EDTA at 7°C with recirculation of buffer. All gels were stained in a solution of 1 pg ethidium bromide/ml. destained in water and

ColEl

CRUCIFORM

THERMODYNAMICS

1x3

photographed under ultraviolet light using Kodak Tri-X Pan film. Densitometry of gt4 negative photographs was achieved using a Joyce-Loebl scanning microdensitometrr. Deliberately under-exposed negatives were selected to ensure linearity of film blackening.

3. Results The plasmid chosen for these studies was pColIR515 (Lilley & Hallam, 1983), a derivative of pColIR215 (Lilley, 1981). pColIR515 contains the region of ColEl that incorporates the inverted repeat that has been shown to undergo conversion to the cruciform conformation, and has an overall size of 2740 hp. which is small enough to facilitate topoisomer manipulation yet large enough to 1)~in the linear range of size dependence of supercoiling free energy (Horowitjz B U’ang, 1984). A map of pColIR515 and the sequence of the ColEl inverted repeat a,re shown in Figure 1. We have adopted two approaches for studying t~he topological dependence of this inverted repeat towards probes of cruciform structure. The first employs topoisomer distributions, whilst the second is based upon t.he purification of individual topoisomers. (a) AS, nuclease

sensitivity

of pColIR515 topoisomer distributions

Topoisomer distributions of varied mean superhelix density were prepared by complete relaxation of supercoiled pColIR515 by chicken topoisomerase I in the presence of appropriate concentrations of ethidium bromide. Figure 2 shows some examples of such topoisomer mixtures resolved on chloroquine-containing gel rlectrophoresis. Each sample contained a Gaussian distribution of topoisomers. for which the mean linking difference could be calculated by the method of Kolb & Hut (1982). Distributions covering the range of mean linking difference from 0 to -20 were examined for site-specific S, nuclease sensitivity. Plasmid samples were equilibrat)ed at 37°C in 10 mM-Tris.HCl (pH 7.5) for 15 minutes. After cooling to ice temperature. S, buffer was added and H, nuclease digestion

PIO. 1. Map of pColIR515, the plasmid used in of the ColEl inverted repeat (filled box) and sequences. Below is shown the sequence (with repeat with the nucleotides related by the 2-fold

these studies. The circular map indicates the position the EcoRI and BarnHI restriction enzyme target hyphens omitted for clarity) of the ColEl inverted axis underscored.

2.25

2.40

2.70

2.55

2.85

300

(b)

Pm. 2. Topoisomer distributions of pCoKR515 produced by topoisomerase incubation in the presence of ethidium bromide. (a) ~hloroquir~~-containing agarose gel electrophoresis of 6 examples of topoisomer distributions. Lanes are lab&d 4~ the concentrations (pug/ml) of ethidium bromide present during the relaxation step. (b) Den&itometer scana through the same gel. Each scan was obtained from one lane of this gel and is labe\\ed witk the ethidium bromide concentration as before. Topoisomer linking differences, indicated at the bottom, were indexed hg reference to a pColIR515 sample whose absolute linking difference had been calibrated by a complefc band counting procedure. The mean linking difference of each sample is shown b;v the position of an arrourhead.

0

-9-3

-10.1

-11.2

-11.8 -12.5

-13.5

-IO*3 -15.2

-16.0

-16.4

-16.4

-l&J

- 19.0

Yrc:. 3. Site-specific S, n&ease cleavage of pCollR525 topoisomer distributions. DNA samples were incubated with S, nuclease followed by BanzHI rritric?iurr ~lra\ag!e. Site-specific cleavage at the ColEl inverted repeat generates a 2400 bp fragment that migrates on agarose gel electrophorrsis as the hand indicatrd hy the arrow shown on the right. Full-length linear pColIR515 migrates at the position indicated by the smaller arrow labellcd L on the left. Lane S c*orrrsponds to an experiment in which native supercoiled pColIR515 was employed. All remaining lanes resulted from identical experiments on topoisomrr distributions, each being labelled by the mean linking difference of the distributions.

N

186

D. M.

J. LILLEF

AND

-0.025

L. R. HALLAM

-005

GO75

8

Flu. 4. Site-specific S, nuclease cleavage of pColIR515 topoisomer distributions as a function of mean linking difference. Data like those of Fig. 3 were quantified by densitometry and the extent of site-specific S, nuclease cleavage calculated. This is plotted as relative extent of cleavage against mean linking difference (AU) and mean specific linking difference (8).

performed at 15°C. After purification by precipitation with ethanol the S, nuclease-cleaved DNA was digested to completion by BamHI, and examined by agarose gel electrophoresis. The basic rationale behind this protocol is the “freezing” of the equilibrated population by rapid cooling. For the ColEl cruciform we have shown (D. M. J. Lilley, unpublished results) that the extrusion process is exceedingly slow below 3O”C, and thus the subsequent assay of cruciform presence by enzyme or other probes is performed using a kinetically trapped equilibrium population. The results are shown in Figure 3. It is immediately clear that the 2400 bp band, which is generated by S, nuclease cleavage at’ the ColEl inverted repeat, is present only above a threshold linking difference of around - 12. The sharp appearance of this site of specific cleavage is revealed even more clearly by densitometry of these results, presented graphically in Figure 4. The transition from intensitivity to sensitivity covers a relatively narrow range of mean linking difference, with a mid-point at ALk = - 14. This behaviour is predicted from the quadratic dependence of supercoiling free energy on linking difference. These results are in close agreement with those of Singleton & Wells (1982), who have studied S, nuclease cleavage of topoisomer populations of dimeric pVH51. (b) I’uri$cation

of individual

pColIR515

topoisomers

Experiments like those in the preceding section are informative, yet partly ambiguous, since mixtures of topoisomers are the inevitable result of the preparative method used. Hence the function sought, i.e. the dependence upon supercoiling, is smeared with a Gaussian curve, i.e. the topoisomer distribution. We therefore set out to remove this shortcoming by purification of each topoisomer individually by gel electrophoresis. Figure 5 shows a chloroquinecontaining agarose gel of the resulting preparations. Each topoisomer is essentially free of contamination by any other form I plasmid. These purified topoisomers have been used in all experiments described subsequently.

ColEI

-10

Fro.

-11

5. Individual

top~istrmcrs.

rwh

-12

CRVCIFORM

-13

-14

-15

-16

topoisomers of pColIR515. Chloroquine-containing labelled with their linking difference.

(c) S, nuclease

1x7

THERMODYSAMIVS

cleavage of pColIR515

-17

-18

aparose

-19

gel of pure

sinpi?

topoisomers

Individual topoisomer species were equilibrated at 37°C for 20 minutes before S, nuclease cleavage at 15°C. Each sample was then precipitated with ethanol, digested to completion with BarnHI and electrophoresed in agarose. Typical results are presented in Figure 6. Sensitivity to S, nuclease is again indicated by the appearance of a 2400 bp fragment. This is absent in the lanes corresponding to linking differences - 13 and below, and there is a clear transition at ALk = - 11. This topoisomer appears to possess a critical free energy of supercoiling to permit’ the extrusion of a stable cruciform. All topoisomers of greater linking difference are specifically cleaved by S, nuclease. (d) Ra131 nuclea,sr

cleavage of pColIR515

topoisonzws

Ha131 is bobh a single-strand-specific endonuclease, and a 3’ and 5’ exonucleasr (Lau & Gray, 1979). We observe that at 15°C this enzyme makes a site-specific cleavage at the ColEl inverted repeat of supercoiled plasmid DNA. The initial cleavage is, to within one or two nucleotides, coincident with that of S, nuclease, although with more extended digestion times the exonucleolysis becomes significant (I~. R. Hallam & D. M. J. Lilley, unpublished results).

18X

D. M. J. LILLEY

-11

-12

-13

Fm. 6. S, nuclease cleavage of single were incubated with S, nuclease followed inverted repeat results in the generation agarose gel electrophoresis is indicated migrates at the position indicated by the linking difference of the topoisomer used.

AND

-14

L. R.

-15

HALLAM

-16

-17

topoisomers of pColIR515. Single topoisomer preparations by BumHI. Site-specific S, nuclease cleavage at the ColEl of a 2400 bp fragment, the migration position of which in by the srrow on the right. Full-length linear pColIR515 arrow labelled L on the left. Each lane is labelled with the

Individual topoisomers, pie-equilibrated as before, were digested with BnZ31 at 15°C before BamHI cleavage and electrophoresis. The results are shown in Figure 7. Fragments due to BaZ31 cleavage appear, on close examination, as doublets due to exonuclease action. Further analysis of this behaviour will be presented elsewhere. As with S, nuclease. a short t’ransition to nuclease sensitivity occurs at a linking difference of - 14. The single-strand-specific nucleases thus generate closely comparable results. (e) Phage

T4 endonuclease

1~11 cleavage of pColIR515

topoisomers

Phage T4 endonuclease VII is a Holliday junction-resolving enzyme or resolvase (Kemper 8; Garabett. 1981). which introduces highly site-specific cleavages at the bases of t’he arms of cruciforms (Mizuuchi et nl.. 1982~~; Lilley &, Kemper, 1984). It is. without’ quest)ion. the most selective enzyme probe for cruciforms, although its limited availability restricts its use to only the most) significant experiments. Tn this instance this means the t’opoisomers around the expected transit’ion region. The experiments were performed analogously to those using the single-strandspecific nucleases. Following the pre-equilibration at 37°C. topoisomers were incubated with T4 endonuclease VII at 25”C, precipitated with ethanol and digested to completion with BumHI. On agarose gel electrophoresis, site-specific

ColEl

N

-10

CRUCIFORM

-11

IN

THERMODYNAMICS

-12

-13

-14

-15

-16

L

FIG. 7. &z/31 cleitvage of single topoisomers of pColIR515. Single topoisomer preparations wtv itlcuhattvl with Ha/31. followed by BumHI. Site-specific &z/31 cleavage results in the generation of thr fr;tymrvt indicated by the avow on the right. Full-length linear pColIR.515 is indicated hy I, 011 the kft. Each laur is labelled with the linking differenre of the topoisomer uxrd except for that labell~i N. it) which native SupercoilPd p(‘o11R515 was employed.

cleavage by the resolvase is indicated by generation of a 2400 bp band. The rcbsults are shown in Figure 8. Just as with the single-strand-specific nucleastas. there is a short t,ransition to nuclease sensitivity at a linking difference of - 11. For rt’asons of enzyme availability this experiment’ was performed only twice. i)ut with ident,ical results. ( f) Osmium tetroxide modification

of pColI R5 15 topoisomers

CWe have employed a number of chemical probes in our studies of cruciform structures. including bromoacetaldehyde (Lilley, 1983c), glyoxal (G. W. Gough & 0. M. J. Lilley, unpublished results) and osmium tet,roxide (Lilley & PaleEek. 1984). These reagents, which all selectively undergo adduct formation wit’h singlestranded DNA, provide valuable complementary data to the enzyme probes, since they are free from any potential artefacts of enzyme-induced structural transition. \VhiM all three reagent)s selectively modify the ColEI inverted repeat, we chose osmium tetroxide for these studies as it is the best-characterized reaction at the loop of the cruciform, and the reaction proceeds at pH 7.8. Following a pre-equilibrium at 37°C analogous to that described above. f)otrntjial site-selective modification was examined by a protocol that differed

190

D. M.

-12

FIG prepa endon right. differs

J. LILLEY

-13

8. Phage T4 endonuclease VII rations were incubated with huclease VII cdeavage results in Full-length Ii near pColIR5l.i is vwe of the to1 ?oisomrr used.

AND

-14

cleavage of single T4 endonuclease the generation of indicated by I, on

L. R. HALLAM

-15

-16

topoisomers of pC‘olIR515. Single top oiso VII, followed by BaraHI. Site- spec :ific the fragment indicated by the ar ‘TOW 011 the Irft. Each lane is labelled wkh the linl

mer T4 the iing

from those of the enzyme probes, because the reagent does not induce breakage of the deoxyribose-phosphodiester backbone. Topoisomers were reacted with 1.6 mllr-osmium tetroxide at 25”C, after which the reagent was removed by extraction in chloroform and the DNA purified by precipitation with ethanol. The modified plasmid was digested to completion with BamHI and purified again by precipitation with ethanol. Finally, the linear plasmid was incubated with S1 nuclease to digest any regions rendered permanently single-stranded by bipyridyl osmium tetroxide &s-ester formation. The details of these procedures have been optimized (Lilley & PaleEek, 1984) for modification and observation of the ColEl cruciform. The results are shown in Figure 9. Site-selective modification at the ColEl inverted repeat is once again indicated by generation of a 2400 bp fragment, visualized after agarose gel electrophoresis. Just as with the enzyme probes, there is a short transition between insensitivity and sensitivity to chemical modification. Once again the least underwound topoisomer to be modified, albeit weakly, is that of linking difference - 14.

ColEl

-11

-12

CRUCIFORM

-13

19 I

THERMODYNAMICS

-14

-15

-16

-17

FIG. 9. Osmium t&oxide modification of single topoisomers of pColIR515. Single topoisomel preparations were reacted with osmium tetroxide/pyridine, followed by BarnHI restriction cleavapr and finally S, nuclease cleavage. Site-selective chemical modification ultimately results in t,he generation of the fragment indicated by the arrow on the right. Full-length linear pColIR515 is indicated by L on the left. Each lane is labelled with the linking difference of the topoisomer used.

(g) Comparison

between probes

The results obtained with each probe were quantitated by densitometry of phot,ographic negatives of agarose gels. The extent of reaction was estimated bp expressing the amount of DNA in the 2400 bp band as a fraction of the total, i.e. full-length linear plus subfragment. It is not meaningful to compare absolute of values of this fraction between experiments using different probes, since extents different reactions may vary. Within a set of reactions using a given probe. however, conditions were carefully controlled for comparability between topoisomers. The results are presented graphically in Figure 10. The most striking result that emerges from this comparison is the close similarity in the positions and widths of the four transitions. Each has a mid-point equivalent to ALk = - 14.5. Small differences in the shape of these curves, such as those to the high supercoiling side of the transitions, may be significant. S, nuclease appears to undergo a slight fall in extent of cleavage at ALk = - 17, and a similar fall may be present in the data from topoisomer distributions shown in Figure 3. Singleton 8: Wells (1982) have previously noted similar behaviour on studying S, cleavage of t,opoisomer distributions of pVH51 dimers. We suspect that this effect may result from adjustment of the tertiary structure of the molecule as its linkage is progressively reduced. The overall similarity of the results between these four very different probes suggests that the effects result from a structural transition in the DNA, which is subsequently recognized by each probe.

192

D. M.

J. LILLEY

I -10

I I -II -12

AND

I I -13 -14

L. R. HALLAM

I I -15 -16

, , -17-18

ALk

FIQ. 10. Extent of enzyme cleavage or chemical modification at the WE1 inverted repeat as a function of linking difference. Relative extents of site-specific cleavage by S, nuclease. BaZ31 T4 endonuclease VII or site-selective reaction modification by osmium tetroxide are plotted against linking difference (AU). The close similarity in transition po\nts is highlighted by the vertical lines drawn through linking differences - 14 and - 15.

(h) Direct observation of writhing changeson cruciform pColIR515 topoisomers

extrusion by

Extrusion of a cruciform results in a local negative change in DNA twist’, which will therefore allow a corresponding relaxation in the torsional stress of the molecule overall. Changes in molecular writhing should affect the frictional properties of the molecule; thus, a relaxation of writhing should be observable directly as a reduction in electrophoretic mobility. This has been demonstrated for some long inverted repeats (Courey & Wang, 1983; Gellert et al., 1983). However, for short inverted repeats this is technically more difficult, since the resulting mobility changes on agarose gel electrophoresis occur at or close to the region in which mobility as a function of writhing has saturated. For the very small ColEl deletant pA03, 1643 bp (Oka et al., 1979), Lyamichev et al. (1983) have shown that the greater resolving power of a composite polyacrylamide/agarose gel enables the observation of the transition on cruciform formation. We therefore decided to see if this technique could be extended to our larger plasmid pColIR515, 2740 bp. Each purified topoisomer was preincubat’ed in 10 mM-Tris * HCl (pH 7.5) for 30 minutes at 37°C and then loaded directly onto a polyacrylamide/agarose gel, which was electrophoresed slowly at 7°C. An example is shown in Figure 11. Topoisomers with linking differences down to - 13 migrate monotonically.

-11

ColEl

CRUCIFORM

-12

-13

THERMODYNAMICS

-14

FIG. 1 1. Composite agarose/polyacrylamide Lanes al re labelled with the linking differences

-15

-16

gel electrophoresis of the topoisomers.

-17

of single

-18

topoisomers

of pColIR515.

FIowrver, a transition is evident at ALk = - 14; a proportion of t,he IIN14 migrates as a band t,hat is considerably retarded. As the linkage of the topoisomers is wduwd further, a corresponding band shift is seen in each case, with the intensity

12

13

I6 -II

-12

-13

-14 -15 ALk

-16

-17

-18

P’w. I%. Relative migration (arbitrary units) of single p(‘olIR515 topoisomers on composite pol~ncr~lamide/agarosr gel electrophoresis as a function of linking difference (A&). The disrontinuit) brought about hy the topology-dependent cruciform extrusion is indicated by the arrow connecting the migration positions of the 2 species of linking difference - 14.

194

D. M. J. LILLEY

AND L. R. HALLAM

of the retarded DNA increasing at each step. Indeed, at ALk = - 17 and below, effectively all the DNA is present as the retarded form. Thus, there is a critical linking difference in the region - 14 to - 15 at which the transition between forms is seen. It is significant that this is precisely in the region in which sensitivity to enzyme and chemical probes becomes apparent. In Figure 12 we have plotted the migration of these species as a function of linking difference. Once again the pronounced retardation of the second structural forms is evident, and the two curves obtained that are connected by a discontinuity arising from a single structural transition at a critical level of supercoiling. The mean separation of these curves corresponds to an average AWr = -3*2+0*1. We therefore estimate the local twist change for ColEl cruciform extrusion as - 3.2 + 0.1.

4. Discussion The transition between total insensitivity and maximal sensitivity to reaction by four probes at the ColEl inverted repeat is a sharp function of superhelix density. This is revealed particularly clearly by the use of purified single topoisomers. Previous investigations have been based upon topoisomeric distributions prepared by topoisomerase relaxation of plasmids in the presence of intercalating molecules. Whilst t,hese experiments have been valuable, there are problems that arise from the fact that the method yields Holtzmann distributions of topoisomers. As can be seen in Figure 2, each sample contains a mixture of at least six topoisomers and the width of this distribution is fixed by the lowest temperature at, which the topoisomerase can be used. Thus, although the data of Figure 4 show that the transition t,o S, nuclease sensit,ivity is narrow, it should be borne in mind that these data are smeared with a Gaussian distribution, and that the true function may be narrower still. Studies using single topoisomeric species confirm this point. These transitions are extremely sharp with only one, or perhaps two, topoisomers of intermediate sensitivity. It’ is significant that the use of each probe results in a virtually identical profile. Table 1 summarizes the four probes employed, their conditions and modes of action, which can be seen to vary widely. Thus, we have employed two single-strand-specific nucleases t’hat, cleave the central (loop) region of the cruciform, yet which are active at acid and alkaline pH, respectively, and at very different ionic strengths. T4 endonuclease VII does not interact with the loop region, but introduces staggered cleavages at t’wo nucleotides in from each 5’ end of the repeat, i.e. it “resolves” diagonally and asymmetrically across the four-way junction (Lilley & Kemper, 1984). Its mechanism is thus quite unlike that of the other nucleases. Finally, osmium tetroxide is a small molecule probe. It has no binding domain and must simply react opportunistically with thymine and cytosine bases possessingsingle-stranded character. These extreme differences in mechanisms make any direct participation by the probes themselves in causing structural change very unlikely. Coupled with the similarity in the dependences upon linking difference, this suggests that the probes reliably reflect a structural

VII

pH pH pH pH

4.6, 8.1, 8.6, 8.0,

Zn2+ Ca*+ Mg* + -

(‘onditions

of the enzyme and chemical

S, nuclease R&l nuclease T4 endonuclease Osmium tetroxide

Probe

Summary

1

Target Single-strand nucleolysis Single-strand nucleolysis Resolution nucleolysis Chemical modification

Mode

used in this study, the conditions

Loop Loop Junction Loop

probes

TABLE

Lilley Lilley

(1984) (1984)

Panayotatos & Kemper & PaleEek

(1980);

Reference & Wells

(1981)

and their target sites and modes of reaction

Lilley

employed

196

D. M. J.

LILLEY

AND

L. R.

HALLAM

transition that is dependent solely on the topological state of the DNA. All available evidence indicates that this transition is the extrusion of the ColEl cruciform. This conclusion is strengthened further by the correlation of probing and bandshift experiments. Bandshift methods involve no interactions with probes of any kind, and are a straightforward consequence of the topological change. The agreement between estimates of the transition point by probing and bandshift methods is excellent, and the self-consistency of all these experiments generates additional confidence in the conclusions, as well as in the methods themselves. In the past, probing experiments have occasionally been somewhat maligned, which is unfortunate for several reasons. Bandshift methods are without doubt very direct and conclusive. However, they can give information only about the overall topological state of the complete molecule. No information is forthcoming about where in the sequence of the DNA the transition occurs, unless mutants are available for comparison. This is precisely the information that is provided by probing data, and thus the use of both methods in parallel is most advisable. In a more complex situation where two or more topology-dependent competing transitions may occur, it may not be possible to gain a full understanding of the system by any other approach. A further advantage of probe studies is that the transitions that they reveal may be associated with relatively small ATw values, whereas the bandshift method requires writhing changes that may be out of the range of resolution by present gel electrophoretic systems. Finally. there is at present’ considerable interest in probing both DNA and chromatin with a large variety of enzymes and chemicals (Wu, 1980: Cartwright & Elgin, 1982: Hentschel, 1982; Jesse et al., 1982; Larsen & Weintraub, 1982; Glikin et al., 1983,1984; Goding & Russell, 1983; Kohwi-Shigematsu et al.. 1983; Nick01 & Felsenfeld, 1983; Mace et al., 1983: Weintraub. 1983: McKeon et al., 1984) in conditions in which the DNA is not’ necessarily under torsional stress or constraint,. It is therefore most’ important t)hat’ their use on well-defined DNA structures be closely studied and understood. Caut)ion is undoubtedly essential in the int’erpretation of probing experiments. Single-strand-specific nucleases in particular may cleave at a variety of st,ructural perturbations including. for example. the B-Z junction (Singleton et al.. 1982). whereas T4 endonuclease VII is considerably more st,ructure-specific. We must emphasize that, structural conclusions should be based upon the maximum amount’ of data from as many sources as possible, both probing and topological. The results of the bandshift experiment, provide a very accurate estimate for the ATw value of ColEl cruciform extrusion. Close inspect’ion of Figure 11 shows t’hat’ the mobility of the retarded form of the t’opoisomer (AI% = - 15) is intermediate bet’ween those of ALk = - 11 and ALk = - 12. Similarly, topoisomer ALk = - 16 lies intermediate between ALk = - 12 and - 13. Estimation of the average A B’r value from Figure 12 gives a value for ATw of cruciform extrusion of - 3.2 f 0.1. Previous estimates based upon writhing changes in pA03 have given ATw values of -4.0 (Lyamichev et aZ., 1983) and - 3.5 (Dean et al.. 1983: Otter et al., 1983). We believe our present value to be highly accurate, since it is based upon one-dimensional gel electrophoresis. Two-dimensional gel electrophoresis

ColEl

CRUCIFORM

THERMODYNAMICS

187

methods (Wang et al., 1983) present a great deal of information in a single gel, but individual topoisomer spots are difficult to estimate densitometrically. This lends a small error to mobility measurements that is not present in one-dimensional gels, where the comb-gel interface provides a leading edge that can be measured with considerable precision. The ColEl inverted repeat contains 31 bp, and T4 endonuclease VII results (Lilley & Kemper, 1984) confirm that the entire sequence is likely, as expected, to participate in cruciform formation. A simple view of t,he formation of a cruciform structure would therefore predict a change in twist

of:

31 ATw = __ = 2.95, 10.5 where the pitch of the DNA is 10.5 (Wang, 1979; Rhodes B Klug, 1981). How may we account for the discrepancy between the measured ATuq of - 3.20 and the caalculated value of -2.952 Two explanations are possible. First, the formation of the junction may involve nucleotides that are not symmetrically related by the P-fold rotational axis. We note that chemical modification experiments using hromoacetaldehyde have hinted at structural change in this region (Lilley, 1983c) a,nd such changes could have an associated ATw value. Second, it, is possible t,hat once the cruciform has extruded completely, optimization of the stereochemistry of the four-way helical junction requires a further quarter-turn rotation of t,he non-cruciform duplexes. We should also note that the cruciform structure ma! itself retard mobility in gels, although this effect seems unlikely to be large. Since we have measured both the critical linking differences for ColEI cruciform cbxtrusion, together with the resulting twist change, we may use these values to calculate the free energy (AG,) of cruciform formation. It has been determined rsperimentally that the free energy of supercoiling is expressed by the function ( I)epew & Wang, 1975; Pulleyblank et al., 1975; Wang et al., 1983): AC=

11 OORT ___ ALk2. N

where R is the gas constant, T the absolute temperature and ilr the plasmid size in base-pairs. The critical linking difference we estimate to be - 14.5, i.e. the linking difference at which cruciform and non-cruciform states would possess equal free energy. This value has to be corrected to take account of the different t*emperatures at which the linking number measurements were performed (20°C) and that at which the samples were equilibrated (37°C). The higher temperature of the latter would result in a relaxation of the molecule due to the temperatIme dependence of the winding angle. Using a value of 0.01 deg./deg.C per bp (Depew & Wang, 1975: Pulleyblank et al., 1975), we calculate an effective critical ALk value of - 13.2 at 37°C. Hence: 11 OORT AC, = ,v[(-13~2)‘-(-10~o~o~1)~]

= l&4&0.5

kcal mol-’

This free energy is quite high and is in good agreement with estimates made for

198

D. M. J. LILLEY

AND L. R. HALLAM

different inverted repeats (Courey & Wang, 1983; Gellert et al., 1983). Some variation between different sequencesis to be expected as the free energy of the structure is likely to be a function of the length and sequence of the central loop sequences,the sequence of the junction region and possibly that of the flanking DNA. In addition, this value has been determined for 37°C in contrast to previous studies, a consequence of the lower extrusion temperature for this particular cruciform (D. M. J. Lilley, unpublished results). The critical linking difference for stability of the ColEl cruciform corresponds to a specific linking difference (a = AU/U’) of -0.051. This is close to the native level (typically -0.06 to -0.07). We calculate that if the symmetrical (stem) length were reduced to 9 bp, the critical specific linking difference would become -0.067. Thus, cruciform struct’ures with stem lengths around 9 bp are likely to be the smallest that can be stabilized in supercoiled molecules of native superhelix density, assuming that the free energy of formation cannot be reduced substantially for other sequences. The smallest inverted repeat detected as a S, nuclease hypersensitive site in supercoiled DNA has a stem length of 9 bp. In summary, we now possess a good thermodynamic understanding of cruciform extrusion by the inverted repeat sequence of ColEl. We intend to extend t,his analysis to systema,tically related sequences in order to dissect the relative energetic contributions of the component features, such as loop and junction sequencesas well as the effects of disruptions to perfect S-fold symmetry. We thank Takeshi Kimura for chicken reticulocyte topoisomerase I and Bijrries Kemper for T4 endonuclease VII. We are grateful to the Medical Research Council and the Royal Society for financial support.

REFERENCES Bazaral, M. & Helinski, D. (1968). J. Mol. Biol. 36, 185-194. Benham, C. J. (1982). Biopolymera, 21, 679-692. Cartwright, I. L. & Elgin, S. C. R. (1982). Nucl. Acids Res. 10. 5835-5852. Courey, A. J. & Wang, J. C. (1983). Cell, 33, 817-829. Dean, F., Krasnow, M. A., Otter, R., Matzuk, M. M., Spengler. S. J. & Cozzarelli, N. R. (1983). Cold Spring Harbor Symp. Quant. Biol. 47, 769-777. Depew, R. E. & Wang, J. C. (1975). Proc. Nat. Acad. Sci., U.S.A. 72, 4275-4279. Dingwall, C., Lomonossoff, G. P. & Laskey, R. A. (1981). Nucl. Acids Res. 9, 2659-2673. Fuller, F. B. (1971). Proc. Nat. Acad. Sci., U.S.A. 68, 815-819. Gellert, M., Mizuuchi, K., O’Dea, M. H., Ohmori. H. & Tomizawa, J. (1979). Cold Spring Harbor Symp. Quark Biol. 43, 35-40. Gellert, M., O’Dea, M. H. & Mizuuchi, K. (1983). Proc. Nat. Acad. Sci., U.S.A. 80, 55455549. Gierer, A. (1966). Nature (London), 212, 1480-1481. Glikin. G. C., Gargiulo, G., Rena-Descalzi, L. & Worcel, A. (1983). Nature (London) 303. 770-774. Glikin, G. C., VojtiBkova, M.. Rena-Descalzi, L. & PaleEek, E. (1984). Nucl. Acids Res. 12. 1725-1735. Goding, C. R. & Russell, W. C. (1983). Nucl. Acids Res. 11, 21-36. Hentschel, C. C. (1982). Nature (London), 295, 714-716. Hershfield, V., Boyer, H. W., Chow, L. & Helinski, D. R. (1976). J. Bacterial. 126, 447-453. Horowitz, D. S. & Wang, J. C. (1984). J. Mol. Biol. 173, 75-91.

(‘olE1

CRUCIFORM

THERMODYNAMI(‘Y

199

Hsieh, H. T.-S. & Wang, J. C. (1975). Biochemistry, 14. 527-535. .Jaror. G. T. (1974). J. Bacterial. 120, 282-286. Jesse, B., Garguilo, G.. Ravzi, F. & Worcel, A. (1982). Nucl. Acids Res. 10. 5823-5834. Keller, W. (1975). Proc. Nat. Acad. Sci., C.S.A. 72, 4876-4880. E;rmper, 13. & Garabett, M. (1981). Eur. J. Biochem. 115, 123-131. Kohwi-Shigematsu. T.. Gelinas, R. & Weintraub. H. (1983). Proc. Na,t. Acad. Sci.. (Y.S.=I. 80. 4389-4393. Kolb. .A. xi Rut. H. (1982). Nucl. Acids Res. 10, 473-485. Larsen, A. & Weintraub. H. (1982). CeZZ,29, 609-622. 1,au. 1’. I’. & Array. H. R. (1979). Nucl. Acids Res. 6. 331-357. I,illry. I). M. ,J. (1980). Proc. Nat. Acad. Sci., C’.S.A. 77, 6468-6472. I,illeg. 1). M. .J. (1981). AVucl. Acids Reu. 9, 1271-1289. I&y. 1). PrI. .J. (1983a). Nuture (London), 305. 276-277. Lillry. 1). M. ,I. (19836). Cold Spring Harbor Symp. f&ant. Biol. 47. 101-l 12. I,illey. I). M. J. (1983c). Nucl. Acids Res. 11, 3097-3112. I,illey. 1). M. ,J. 8: Hallam, L. R. (1983). J. Biomol. Struck Dyn.am. 1. 169-182. l,illey. 1). M. ,J. Br Kemper, B. (1984). Cell, 36, 413-422. I,illey. I). M. ,J. 8r PaleEek. E. (1984). EMBO J. 3, 1187-1192. I,yarnic~hev. V. T.. Panyutin. I. G. & Frank-Kamenetskii. M. D. (1983). FEBS Letters. 153. “!W302. .1lac~. H. A. F.. Pelham, H. R. B. & Travers, A. A. (1983). Nature (London). 304, 555~-557. MlcKron. (‘.. Schmidt, A. & de Crombrugghe, B. (1984). J. Biol. Chem. 259, 6636-6640. Mizuuchi, K.. Kemper, B., Hays, J. & Weisberg, R. A. (1982a). Cell, 29, 357-365. Mizuuchi. K.. Mizuuchi, M. & Gellert, M. (19828). J. Mol. Biol. 156. 229-243. I$ickol, J. M. & Felsenfeld, G. (1983). Cell, 35. 467-477. Oka. ~1.. Komura, i\u’.. Morita. M.. Sugisaki. H., Sugimot’o. K. b Takanami, M. (1979). Mol. Qpn. Genut. 172. 151-159. Ot,ter. R.. Lilley. D. M. ?J. & Cozzarelli, P;. R. (1983). J. Cell Riochem. supl. 7B. Alan R. Liss Inc. Sew York. l’anayotatos, S. & Wells. R. D. (1981). Nature (London), 289, 466-470. l’anyutin, T. G.. Lyamichev, V. I. & Lyubchenko, YuL. (1982). FEBS Letters, 148. 297301. l’earork. A. C. & Dingman, C. W. (1968). Biochemistry, 7, 668-674. Perk. L. J.. Kordhrim, A.. Rich, A. 8: Wang, J. C. (1982). Proc. ‘Vat. Aca,d. Sci., I:.S.A. 79. 4560-4564. l’labt. ,J. R. (1955). Prof. Nut. Acad. Sci., U.S.A. 41, 181-183. Pulleyblank. D. E., Shure, M.. Tang, D., Vinograd, J. & Vosberg, H.-P. (1975). Proc. Na,t. dcad. Sci., I’.S.A. 72, 4280-4284. Rhodes. D. & Klug, A. (1981). Nature (London), 292. 378-380. Sanzey. H. (1979). J. Bucteriol. 138, 40-47. Shure, M.. Pulleyblank. D. E. & Vinograd. J. (1977). Nucl. Acids Res. 5. 1183-1205. Singleton. (1. K. $ Wells, R. D. (1982). J. Biol. Chem. 257, 6292-6295. Singleton. C. K.. Klysik. J., Stirdivant, S. M. & Wells. R. I>. (1982). Nature (London), 299. 312-~316. Smith, G. R. (1981). Cell, 24, 599-600. lTinograd. ,J. & Lebowitz, J. (1966). J. Gen. Physiol. 49. 103-125. Vinograd. J.. Lebowitz, J. Radloff. R.. Watson, R.. & Laipis. P. (1965). Proc. Nat. Acad. Sri.. f’.S.‘4. 53. 1104-1111. Vinograd. J.. Lebowitz, J. & Watson, R. (1968). J. Mol. Biol. 33, 173-197. 1.ologodskii, A. V. & Frank-Kamenetskii, M. D. (1982). FEBS Letters. 143. 257-260. Wahlr. E. & Mueller, K. (1980). Mol. Gen. Genet. 179, 661-667. Walter. (i.. Zillig. W.. Palm. P. & Fuchs, E. (1967). Eur. J. Biochem. 3. 194-201, 12’anp. A. H.-,J.. Quigley. G. J., Kolpak, F. J., Crawford. J. L.. vanBoom. ,J. H.. vanderMare1. G. & Rich, A. (1979). Nature (London), 282. 680-686. \Vang. J. (‘. (1979). Proc. Nat. Acad. Sci., U.S.A. 76, 200-203.

200

D. M. J.

LILLEY

AND

L. R.

HALLAM

Wang, J. C.. Peck, L. J. & Becherer, K. (1983). Cold Spring Harbor Symp. Quant. Biol. 47, 85-91. Weintrauh, H. (1983). Cell, 32, 1191-1203. Woodworth-Gutai, M, & Lebowitz, J. (1976). J. I’irol. 18, 195-204. Wu, C. (1980). Nature (London), 286, 854-860. Yang, H.-L., Heller, K., Gellert, M. & Zubay, G. (1979) Proc. Nat. Acad. Sci., U.S.A. 76, 3304-3308.

Edited by A. Klug