THEJOURNAL BIOLOGICAL OF CHEMISTRY Vol. 257, No. 17, Issue of September 1 ,pp. 10159-10165,1982 0 Printed i U.S A .

DNA Supercoiling Energetic and Structural Inter-relationship between and the Right-to Left-handed Z Helix Transitionsin Recombinant Plasmids*
(Received for publication, February 16, 1982)

Steven M. Stirdivant$, Jan IQysikg, and Robert D. Wells8

From the Departmentof Biochemistry, University of Wisconsin, Madison, College of Agricultural andLife Sciences, Madison, Wisconsin 53706

We have evaluated the B t o Z conformational tran- tract, the type of helix (Z or Z-like) adopted by the tract, the sitions in supercoiled recombinant plasmids containingnature of the conformational interface between the right- and different lengths of (dC-dG) described in the preceding left-handed segments, etc. Previous studies (1,2)verified that paper. The sodium chloride-induced right- t o left- a segment of 58 bp of (dC-dG) (1.3%of the total plasmid) in handed transition in a small segment of the plasmids pRW751 (containing a total of 4516 bp) caused a large relaxcaused a relaxation of (-) supercoils which was moni- ation of superhelical turns, indicating that the (dC-dG)regions tored by electrophoretic mobility changes of individual were adopting a left-handed conformation. The transition also topoisomers on agarose gels containing NaCl at con- occurs with short (dC-dG) sequences in restriction fragments centrations up t o 5.0 M. The number of supercoils re- derived from this plasmid (1, 2). laxed was proportional to the length of the (dC-dG) These determinations required establishing an assay for segment in the plasmid in good agreement with theo- analyzing and identifying individual topoisomers under the retical values. A short B/Z junction region (<5 base high salt (3) or other ionic conditions (2) required to effect pairs) was inferred. stability of the Z conformation the B + Z transition. An agarose gel electrophoretic system The in (dC-dG) segments of the plasmids had a strong length was devised (1) for the purpose. dependency; shorter lengths were less stable. Ten base The capacity to analyze the B + Z transition in a superpairs of (dC-dG) was insufficient to allow a Z conforcoiled plasmid is important since it is a direct indicator of mation under the conditions studied. Supercoiling imparts a substantial favorable free energy to the Z con- helical sense, if the sequence of the plasmid has been deterformation, reducing the NaCl concentration necessary mined. Other assays for the B + Z transition, such as CD (1, the to cause transition. The relationship of supercoiling 2, 3, 6 ) , proton and phosphorus NMR (1, 7-9), and Raman with the NaCl concentration necessary to cause B -f spectroscopy (9, lo), do not provide a direct measure of helical a Z transition suggests that supercoiling alone is suffi- sense. X-ray diffraction on single crystals (5) does give this cient to stabilize the Z conformation at physiological information but is applicable only to short oligomers. In this salt concentrations. These results support the notion paper, the isolation of individual topoisomers of (dC-dG)that left-handed DNA has an important biologicalrole. containing plasmids (11) allowed more precise analysis of the B + Z transition in supercoiled plasmids. We report the structural and energetic properties of the NaC1-induced rightChanges in superhelical density in covalently closed plas- to left-handed transition in (dC-dG)-containing supercoiled plasmids. mids provide a sensitive assay for conformational changes in the primary helix. This change is useful for monitoring B + MATERIALS AND METHODS Z transitions in small cloned segments of recombinant plasStrains and Plasmids-The construction ofpRW751, pRW755, mids (1, 2). Strictly alternating (dC-dG) sequences undergo pRW756, and pRW757was described in the previous paper (11). an ionic condition-mediated cooperative transition (1-3) from PRW451 was described previously (1). pRZ4032was a gift of R. a right-handed B structure with 10.4 bp/turn (reviewed in Johnson and W. Reznikoff (this department) and contained the lac Ref. 4) to aleft-handed Z-like structure, presumably with 12.0 95-bp fragment cloned into pBR322 which had been deleted of the bp/turn (5). If one turn of right-handed primary helix is 375-bp Bum HI-Eco RI segment; other details are given in Ref. 11. converted to one turn of left-handed helix in a covalently All plasmids were isolated from Escherichia coli MO recA- (gift of L. Wray and W. Reznikoff, this department) closed plasmid,two superhelical turns will be lost. Obviously, pRW757 which were isolated from E. coli C600except pRW756 and SF8 (12). the extent of relaxation of superhelical density will be deterPreparation ofDNA-Plasmid DNA was isolated as described mined by several factors such as the length of the (dC-dG) previously (13). A l plasmid preparations were bound to an RPC-5 l * This work was supported by grants from the National Institutes
of Health (CA 20279) and the National Science Foundation (PCM 8002622). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked aduertisernent in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ Present address: The Biological Laboratories, Harvard University, Cambridge, MA 02138. 3 Present address: Department of Biochemistry, University of Aabama-Bhingham, Schools of Medicine and Dentistry, University Station, Birmingham, AL 35294. The abbreviation used is: bp, base pairs. column in 0.5 M KC1, 10 m Tris-HC1,O.l m~ EDTA andwere eluted M with 0.75 M KC1 to remove traces of inhibitors of DNA enzymes. This step was necessary to maximize the activity of certain enzymes (e.g. wheat germ topoisomerase). Relaxation of Plasmids-Plasmids were relaxed with the wheat germ topoisomerase (14) (gift of R. R. Burgess, McArdle Laboratory, University of Wisconsin). The reaction conditions were 100 p g / d of plasmid DNA, 50 m~ Tris-HC1 (pH 8.0), 0.1 m dithiothreitol, 10 M mMMgC12, 0.25 m spermidine, and 1.2-2.0 pg/ml of ethidium broM mide. Sufficient topoisomerase was added to provide the desired population of topoisomers in a I-h reaction at 25 C. In order to generate a full range of topoisomers, ethidium bromide was deleted and 30 m~ NaCl was added. Reactions were stopped by the addition

10159

10160

Left-handedDNA in Plasmids
results obtained were independent of the NaCI04 concentration in which the samples were loaded. Reproducibility of results is excellent in all NaCl concentrations employed. However, the visual quality of the gels declines with higher NaCl concentrations and longer electrophoresis times.
RESULTS

of buffer-equilibrated phenol at an appropriate time interval. A U reaction products were extracted with phenol and dialyzed exhaus~ tively uersus 10 r n Tris-HC1 (pH 8.0), 0.1 m~ EDTA. Isolation of Topoisomers-Individual topoisomers of the various plasmids were isolated by eluting the topoisomer bands from agarose gelsusing the chaotropic agents NaI or KI (15, 16). Topoisomer populations prepared as described above were electrophoresed on a 2% agarose slab gel in a buffer of 40 r n Tris, 20 m sodium acetate, ~ M 2 m EDTA (final pH 8.3). Exposure of the plasmid DNA in the gel M to ethidium bromide and UV irradiation caused significant nicking and linearization. Therefore, the two outside lanes of the gel were cut off, and the DNA was visualized by ethidium bromide staining. The migration positions of the topoisomers were marked and used to align the cutting of the unstained portion of the gel. The DNA was eluted by dissolving the agarose in saturated NaI or KI (3 ml/g of agarose), and the mixture was filtered over Whatman GF/C filters (16). NaI (Mallinkrodt Chemical Works) gave less nicking than KI (Mallinkrodt) in this procedure, but the yields were approximately equivalent (20-509) with both salts.The DNA was eluted from the GF/C filters M with 5 m Tris-HC1 (pH 8.0), 0.5 m EDTA. The topoisomers were M M dialyzed into 10 m~ Tris-HCI, 0.1 m EDTA. Fig. 1shows the results of a topoisomer isolation of pRW751. In general the different samples contain only one topoisomer or one dominant topoisomer and one or two minor contaminants. Nicking and linearization are minimal. 2-5 pg of a single topoisomer can be isolated from one agarose slab gel by this procedure. Agarose Gel Electrophoresis a t High NaCl Concentrations-l2% agarose gels (Sigma Type I) were prepared in 1.0-5.0 M NaCl (as indicated under Results), containing 80 m~ Tris, 40 m~ sodium acetate, 4 m~ EDTA, (final pH 8.3). The gels were 0.5 mm thick to minimize the high current and heat generation which was inherent under these high ionic strength conditions. For gels containing concentrations higher than 3.0 M NaCI, the agarose was prepared in 3.0 M NaCl, and thegels were pre-electrophoresed for 24 h in the desired NaCl concentration before the samples were loaded. The gels with NaCl concentrations below 3.0 M were prepared at thedesired NaCl concentrations and were pre-electrophoresed for 2-3 h before loading. The ionic conditions and pH were maintained by circulating large quantities of buffer (4 liters) through the system; the 4liters of circulating buffer were replaced every 24 h. DNA samples were prepared for loading by lyophilization to dryness and were resuspended in 3 pl of electrophoresis buffer with NaCI04 substituted for NaCl at the indicated molar concentrations. Samples were equilibrated in the loading buffer for 1-2 h before loading. Electrophoresis times ranged from 24 tc 120 h, depending on the NaCl concentration and the concentration of agarose. After the NXE,the gels were stained by desalting with distilled water for 1 h and then stained with 10 pg/ ml of ethidium bromide and visualized under UV irradiation. The

Total Relaxation of (dC-dG)-containing Plasmids-Negatively supercoiled covalently closed plasmids containing lengths of (dC-dG) sequence will undergo a change in the linking difference, and thus a loss in the superhelical density, if the (dC-dG) sequences undergo a conformational change from a B to a Z structure. Theloss of (-) supercoils should be proportional to the lengththe (dC-dG) sequence present of in the plasmid. T o determine the numberof supercoils lost per unit lengthof (dC-dG),individual topoisomers of the plasmids pRW751, pRW755, pRW756, and pRW757 were electrophoresed in low and high NaCl concentrations. The change the in linking difference is determined by comparing the difference in the number of supercoils between the (dC-dG)-containing plasmid and a control plasmid at both salt concentrations low and at concentrations above the cooperative transition. The control plasmids are approximately the same size and have virtually the same sequence (>99.5% identical) as the (dCdG)-containing plasmids but contain no stretches (dC-dG). of Fig. 2 shows the relative migration positions of the (dCdG)-containing topoisomers pRW751, pRW755, pRW756, and pRW757 relative to control topoisomers pRW451 and Na. Fig. 3 shows pRZ4032 when electrophoresed in 20 mM the relative migration positions of the sameplasmid topoisomers electrophoresed in 4.5 M NaC1. Comparison of the difference in the number of supercoils present in the (dC-dG)containing topoisomers relative to the control topoisomera t both high and low NaCl concentrations gives the maximum B Z transition in the (dC-dG) supercoils relaxed due to the segments. The number of supercoils relaxed for pRW751 is 10.5; pRW755, 4.7; pRW756, 5.7; and pRW757, 0 supercoils. At 4.0 M NaCl the same extent of relaxation was observed indicating that themaximum relaxation had been reached. Table I summarizes these data and compares the experi-

FIG. 2. Electrophoretic mobility of plasmids containing different lengths (dC-dG)under normalsalt conditions. Individof ual topoisomers and marker topoisomer populations were electrophoresed on a 1.58 agarose gel a t 25 C in 40m Tris-HCI, 20mM sodium M M acetate, 2 m EDTA (pH 8.3) to determine the difference in number of supercoils in control and (dC-dG)-containing plasmids. The lanes FIG. 1. Electrophoretic analysisof individual topoisomers of contain: A, pRW751 isolated topoisomer; B , pRW451 marker popupRW751. pRW751 topoisomer samples isolated as described under lation; C, pRW451 isolated topoisomer; D,pRW755 isolated topoisoMaterials and Methods were electrophoresed on a 1.5%agarose gel mer; E, G,and I , pRZ4032 marker population; F, pRZ4032 isolated in 40 m~ Tris-HC1, 20 m sodium acetate, 2 r n EDTA (pH 8.3). topoisomer; H , pRW756 isolated topoisomer; J, pRW757 isolated M ~ The two outside lanes contained a f l population of pRW751 topo- topoisomer. Some (+) supercoiled topoisomers are seen between (-) ul isomers as markers. supercoiled ones in Lanes E , G, and I.

Left-handed DNA in Plasmids

10161

I
A B C A

I1
A B B C C D E

111
FIG.3. Maximal relaxations of plasmids containing different lengths of (dC-dG) in high NaCl. The
same plasmids andtopoisomerpopulations shown in Fig. 2 were electrophoresed in 4.5 M NaCl on 1.5% agarosegels at 25 OC. The lanes are: Gel Z,A , pRW751 topoisomer; B , pRW451 marker population; C, pRW451 topoisomer. Gel II, A, pRW755 topoisomer; B , pRZ4032 marker population; C, pRZ4032 topoisomer. Gel ZIZ,A, pRW756 topoisomers;B, pRZ4032 marker population; pR7A032 C, topoisomer; D, pR7A032marker population; E , pRW757 topoisomer.

TABLE I Maximum relaxation for plasmids with different lengths of (dCdG) The reproducibility indeterminingthe experimental relaxation was kO.1 supercoil turn.
Plasmid
Length of (dCdG) tract
bP

TABLE I1 Effecto f temnerature on the maximum relaxation of p R W751

Gel
NaCl concentration
M

Watts

Approximate temperature
"C

Supercoils relaxed

Calculated relaxation

Experimental relaxation

10

effect does not appear to be a simple case of shifting the equilibrium of the B + Z transition. mentally determined relaxation with the theoretical relaxation NaCl Concentration Necessary to Initiate the B + Z based onthe number of (dC-dG) residues in the homopolymer Transition as a Function of Superhelical Density-Rightblocks. The calculated relaxation assumes that all of the (dC- handed (-) supercoils contribute a favorable free energy to dG) residues in the homopolymer segments are converted to any process which involves unwinding of the primary helix. an a Z conformation while all other sequences in the plasmid Therefore, increasing (-) superhelical density should lower remain in a B conformation. With the exception of pRW757, the NaCl concentration necessary to cause the B + Z structhe observed number of supercoils removed by the B -* Z tural transition. To study thiseffect, we have used gel electrotransition agrees with that predicted. This extent relaxation phoresis at different NaCl concentrations to determine the of number of supercoils necessary to facilitate the start of the also rules out the possibility that the relaxation is due to cruciform formation (17, in 18) the (dC-dG) blocks since transition for pRW751 at a given NaCl concentration. Detercruciforms would remove only a little more than half that mining the minimum NaCl concentration necessary to cause number of supercoils. The fact that pRW757, containing only a relaxation for an individual topoisomer is a technical probrequired. Therefore, 10bpof perfectly alternating (dC-dG), does not undergo a lem since a large number of gels would be relaxation at 4.5 M NaCl and at the superhelical densities populations of pRW751 topoisomers were run at different tested demonstrates that shorter (dC-dG) stretches are less NaCl concentrations. At a given NaClconcentration, the more likely to convert from B to Z in a plasmid. highly supercoiled topoisomers will be relaxed while those Effect of Temperature on the Total Relaxation-The total with an insufficient number of supercoils will not undergo a relaxation of pRW751 varied with the temperature of the relaxation. This determines the minimum number of superelectrophoresis gel. Table I1 lists the total relaxation observed coils necessaryto begin the B -* Z transition at a given NaCl M NaCl at four different temperatures for pRW751in4.5 concentration. An example of this type of analysis using populations of obtained by varying the wattage of the electrophoresis or the ambient temperature. The temperatures are only approximate pRW751 is shown in Fig. 4. These data also validate the use for the gels run at 2.8 and 4.0 watts; however, it is clear that of topoisomer populations. Identical DNA samples were elecincreasing the temperature decreases the number of supercoils trophoresed in 2.0 and 2.33 M NaCl. A population of pRW751 relaxed at 4.5 M NaCl. A 5.0 M gel run at 35-40 "C showed the obtained by relaxation of the plasmid by nick-closing enzyme same extent of relaxation as the 4.5 M gel run at the same in 2.0 p g / d of ethidium bromide was electrophoresed in Lane temperature. Furthermore, the topoisomer used in this set of B of both gels. This relaxation results in a Boltzman distriexperiments (16 supercoils at 3.0 M NaCl if no B -* Z transi- bution of 8 observable topoisomers. In the presence of 2.0 or tion) is completely relaxed at 3.0-3.5 M NaCl at 25 "C (see 2.33 M NaCl, the more highly supercoiled topoisomers were Fig. 8). It has been observed that the temperature effect is partially relaxed. We define the start of the B -* Z transition more pronounced at NaCl concentrations close to the mini- as a loss of at least 0.5 supercoil resulting from the presence of the (dC-dG) sequences. In the 2.0 M gel, the most highly mumnecessary to complete the full relaxation. Thus, the

pRW751 pRW755 pRW756 pHW757

58 26 32

A supercoils 10.4 4.7 5.7 1.8

A supercoils at 25 "C 10.5 4.7 5.7 0

B C D E

4.5 4.5 4.5 4.5 5.0

4.0 2.8 2.1 2.0 3.0

45-50 35-40 25 4 35-40

9.7 10.1 10.5 10.6 10.2

10162

Left-handed DNA in Plasmids

2.0 M NaCl
A B C
D

2.33 M N a C l
A B C D
FIG. 4. Determination of the minimum number of supercoils necess a r y to cause a partial relaxation at a specific NaCl concentration. Identical plasmids and topoisomer populan tions were electrophoresed i 2.0 and 2.33 M NaCl at 25 "C. For both gels the lanes are: A, pRW751 topoisomers migrating at 6, 7, and 8 supercoils at these NaCl concentrations in the absence of a B + Z transition; B , a population of pRW751 obtained by reacting with nickclosing enzyme in the presence of 2 pg/ ml of ethidium bromide; C, a complete population of pRW451 topoisomers; D, a pRW751 topoisomer migrating at 8 supercoils at these NaCl concentrations in the absence of a B + Z transition.

supercoiled topoisomer migrated at 8.5 supercoils referenced to the pRW451 marker population. Thus, we deduce that 9 supercoils are the minimum necessary to achieve a partial relaxation at 2.0 M NaC1. At 2.33 M NaCl, the most highly supercoiled topoisomer migrated at 7.5 supercoils. Therefore, 8 supercoils are required to induce a relaxation. To confm that the results obtained with these populations were valid, individual topoisomers were electrophoresed in Lanes A and D on both gels. If no B + Z transition occurred, the major topoisomer in Lane A will migrate a t 7 supercoils, while the minor contaminants migrate at 6 and8 supercoils at these ionic strengths. The topoisomer electrophoresed i Lane n D migrated at 8 supercoils in the absence of a B + Z transition. In 2.0 M NaCl, the topoisomers in Lane A show no relaxation, migrating at 6, 7, and 8 supercoils. However, at 2.33 M NaC1, the number 8 topoisomer was relaxed and co-migrated with topoisomer 7 in the 7 supercoil position. In Lane D at 2.0 M, the topoisomer migrated at 8 supercoils demonstrating that no relaxation occurred. At 2.33 M NaCl, this number 8 topoisomer migrated as having 7 supercoils, clearly confuming the observation in Lane A . Since the individual topoisomers demonstrated that a plasmid with 8 supercoils was partially relaxed at 2.33 M, the result obtained with the population of topoisomers is validated. Fig. 5 shows the results obtained from analyses performed as in Fig. 4 with pRW751 topoisomer populations electrophoresed at different NaCl concentrations from 1.25 to 4.5 M. For the range of superhelical densities examined, there is a linear relationship between the superhelical density and the ln[NaCl] necessary to begin the B + Z transition. The equation for the line is: 7 = 10.6 ln[NaCl] + 16.7 (7 equals the number of supercoils). NaCl Titration of the B + Z Transition-To determine the extent of relaxation due t,o the B to Z transition, single topoisomers of pRW751were electrophoresed at different NaCl concentrations. The extent of relaxation was determined by comparing the migration of pRW751 to themigration of a pRW451 control topoisomer. Typical data atone of the NaCl Fig. The concentrations tested (1.85 M) are shown in 6. pRW751 topoisomer in Lane A , whichwould contain 12.6 supercoils if no B to Z transition had occurred, lost 4.7 supercoils. The topoisomer in Lane E, which would have 9.6 supercoils if no B to Z transition had occurred, lost 0.8 supercoil.

I n [NoCI]

FIG. 5 Relationship of plasmid superhelical density and the . NaCl concentration necessary to initiate the relaxation of pRW751. A series of pRW751 topoisomer populations were electrophoresed at different NaCl concentrations at 25 "C.Analysis of the yields the relationship shown. They axis represents results as in Fig. 4 the minimum numberof supercoils which must be present in orderto obtain a relaxation at the NaCl concentration of the electrophoresis. The line through the points is a least squares fit of the data.

Thus, the topoisomers still migrate as distinct bands in these intermediate states of relaxation although the bands are not quite as sharp as those found whenthe relaxation is at its end point. Similar determinations wereperformed on isolated topoisomers of pRW751 at NaCl concentrations from 1.25-5.0
M.

Fig. 7 shows the number of (-) supercoils present in 1 topoisomer of pRW451 and 4 topoisomers of pRW751 at different NaCl concentrations. The pRW751 topoisomers show a decrease in the number of supercoils as the NaCl concentration increases. The control pRW451 topoisomer has a slight increase in the number of supercoils present between

Left-handed DNA in PLasmids

10163

1.85M N a C l
A B C D E

-N

FIG. 6. Typical electrophoresis of pRW751 topoisomers revealingintermediaterelaxationstates. Two topoisomers of pRW751 and pRW451 control topoisomers were electrophoresed in 1.85 M NaCl at 25 "C. Lane A, pRW751 topoisomer 10 Lanes B and 0, range of pRW451 topoisomers; Lane C, pRW451 topoisomer full 5; Lane E , pRW751 topoisomer 7. The topoisomers are named by the number of (-) supercoils which would be present at 3.0 M NaCl if no B + Z transition had occurred. N designates the migration position of nicked plasmid.

different pRW751 topoisomersis presented in Fig. 8. The zero relaxation data points are derived from Fig. 5. These points represent theNaCl concentration at which a topoisomer with n supercoils would begin the B 4 Z transition if in fact it had n + 1 supercoils. It is assumed that this is the highest NaCl concentration a t which the topoisomer with n supercoils has not begun the transition. The most striking feature of the titration datathe biphasic is nature of the curves. The first plateau of the transitionoccurs after a loss of 5.5-6 supercoils. Thus, we conclude that the 32bp (dC-dG) block undergoes the transition before the 26-bp block the maximumrelaxation due to the 32-bp block in pRW756 is 5.7 whereas the relaxation resulting from the 26bp block in pRW755 is only 4.7. The fact that topoisomers 16 and 13 both have the same maximum relaxation demonstrates that the total relaxation does not vary with the superhelical density. For topoisomers 10 and 7, a full relaxation was not obtained even a t 5.0 M NaCl. Topoisomer 7 does not even begin the second phase of the transition.Fig. 7 shows that both topoisomers 10 and 7 have a small number of supercoils at 5.0 M NaCl. It would appear that at low superhelical densities the free energy of supercoiling is insufficient to obtaina complete transition in both (dC-dG) blocks. These results, along with previous observations (1) with pRW751 topoisomers of even lower superhelical density, demonstrate that the + Z tranB sition w not induce left-handed supercoils in the plasmid. i l (+) pRW751 has never been observed to go from a negatively to a positively supercoiled state as a result of the B + Z transition regardless of the NaCl concentration or superhelical density. This suggests that the energy derived from negative supercoiling is critical to the stability of the Z conformation in a covalently closed plasmid. By the same reasoning, positive supercoiling must be very unfavorable energetically with regard to 2 DNA stability.

\ \

I I
50 .

2.0

3.0

4.0

[NaCI] M

FIG. 8. Extent of relaxation versus NaCl concentration for different topoisomers of pRW751. A series ofpRW751 topoisomers was electrophoresed on agarose gels at 25 "C at different NaCl concentrations. Comparison of the migration positions of these topoisomers relative toa control pRW451 topoisomeras in Fig. 6 yields the results shown. The number of supercoilsrelaxedfor a given pRW751 topoisomer are plotted relative to control topoisomer at the a specific NaCl concentration.0,pRW751 topoisomer 16; A, pRW751 1.0 and 3.0 M NaCl. Beyond 3.0 M, no further increase was topoisomer 13; 0, pRW751 topoisomer 1% 0, pRW751 topoisomer 7. observed. Thus, the number supercoils present for of pRW751 Topoisomers are numbered by the scheme presented in the legend to a t a given NaCl concentration reflects the extent of the B + Fig. 6. Inset, a topoisomer of pRW755 and of pRW756 was electro2 transition in the (dC-dG) blocks as well as the small non- phoresed in different NaCl concentrations and the relaxation of (-) supercoils was analysed in the same manner as pRW751 above. sequence specific ionic effect on the whole plasmid. pRW755 would contain 11.2 (-) supercoilsandpRW75610.7 (-) The number of supercoils relaxed a t various NaCl concen- supercoils at 3.0 M NaClifno B + Z transition hadoccurred. +, trations due to the --., 2 transition in the (dC-dG) blocks in pRW756; 0, pRW755. B

FIG. 7. Number of supercoils in pRW751 and pRW451 topoisomers at different NaCl concentrations. pRW751 topoisomers with different numbersof supercoils were electrophoresed along with pRW451 control topoisomers at 25"C at different NaCl concentrations as in the legend to Fig. 6. The number of (-) supercoils in the individual topoisomers was determined by comparing migration positions to the complete ladder of pRW451 topoisomers. 0, pRW751 topoisomer 16; A, pRW751 topoisomer 13; 0,pRW751 topoisomer 10, 0,pRW751 topoisomer 7; pRW451 topoisomer 5. Topoisomers are numbered by the scheme described inthe legend to Fig. 6.

[ NaCI] M

.,

10164

Left-handed DNA in Plasmids

of the B Z transition (3). The temperature effect cannot result from a change in the free energy of supercoiling since it However, when plotted on a ln[NaCl] scale, the slopes appear increases with increasing temperature. This would favor the roughly equivalent (not shown). The sharpness of the second B * Z-induced relaxation. Thus, the temperature dependence phase of the transition decreases with decreasing superhelical for the plasmid must be a result of the junction regions in density even when plotted on a ln[NaCl] scale. This indicates these molecules. It is possible that high temperature shifts the that there is more than a simple ln[NaCl] versus superhelical equilibrium position of the junction region away from the density relationship that is occurring in this situation. This normal B DNA sequences into the (dC-dG) sequences where complexity may reflect changes in the free energies of the B the junction may be energetically more favorable. and Z conformations, the junctions, and/or supercoiling Over The titration several topoisomers of pRW751 at different of varying NaCl concentrations. NaCl concentrations reveals that supercoiling contributes a The inset in Fig. 8 shows the NaCl titration results for a favorable free energy, that therelaxation is biphasic, and that topoisomer of pRW755 and pRW756. When there is no B states of partial relaxation exist. One possible explanation for Z transition, the pRW755 topoisomer has 0.5 more supercoil the partial relaxation states is that the B + Z transition of than the pRW756 topoisomer. In spite of the fact that the (dC-dG) segments in a plasmid is a cooperative all or none pRW755 topoisomer has a slightly higher superhelical density, transition, as for the polymer (dC-dG), (3). Thus, the partial it undergoes the transition at a higher NaCl concentration relaxation states, in the case of an all or none transition,would than does pRW756. These results demonstrate that the Z represent an averaging of the amount the time a of topoisomer conformation is less stable inthe 26-bp (dC-dG) block than in was relaxed or unrelaxed over the time period of the electrothe 32-bp block. Thus, the biphasic nature of the pRW751 phoresis. The rate of interconversion between the B and Z titration curves can be explained in terms of different stabili- conformation would have to be reasonably fast with respect ties of the Z conformation for the two (dC-dG) blocks. This to the electrophoresis time to give discretebandson the (dC-dG) length effect on Z DNA stability argues in favor of agarose gel. In this situation, the degree of relaxation observed reflect the equilibrium of the B to Z transition an a l or none transition occurring in the blocks (see on the gel would l Discussion). Furthermore, the monophasic transitions for at a given NaCl concentration. pRW755 and pRW756 rule out the possibility that the biphasic The second possibility is that partial B -* Z states exist in relaxation for pRW751 is aresult of cruciform formation which part of the (dC-dG) block exists in a Z conformation and the rest exists in a B conformation with a junctionregion followed by a transition to the Z conformation. in between. A nucleation of the Z conformation would occur DISCUSSION in the (dC-dG)sequence and the conformation would spread Z The isolation of single topoisomers of (dC-dG)-containing outward toward the (dC-dG)/B DNA interfaces as the NaCl plasmids coupled with the ability to electrophorese those concentration was increased. The plasmid would be relaxed topoisomers in high NaCl concentrations has allowed a de- up to the point where the free energy contributed by supertailed study of the properties of left-handed Z DNA in cova- coiling would no longer offset the unfavorable free energy of lently closed plasmids. We chose to study the NaC1-mediated the Z conformation at a particular NaCl concentration. The fmding that thelength of the (dC-dG) block alters the B -+ Z transition since it is the best defined system, although stability of the Z conformation argues strongly for an all or other perturbantshave a similar effect (2). The experimentally determined NaC1-induced relaxation of none transition. If the partial conversion model were true, one plasmids containing different lengths of (dC-dG) agrees well would expect to observe identical extents of partial relaxation + with the calculated relaxation for a B Z transition, assuming occurring at identical NaCl concentrations. For example, if a junction of zero base pairs. A junction of zero base pairs is two supercoils have been relaxed in both pRW755 (26-bp (dCdefined as a DNA with entire (dC-dG) tract in a Z conforma- dG)) and pRW756 (32-bp (dC-dG)),approximately 11 bp of tion while the base pairs abutting the (dC-dG) blocks are in (dC-dG) would have been converted to a Z conformation in a Bconformation. P NMR (1) and laser Raman spectroscopy both plasmids. In such a situation where the B/Z junction (10) on the 157-bpBum HI restriction fragmentfrom pRW751 region is not nearthe ends of the (dC-dG)blocks, the energetic (1) indicates that nearly a l the (dC-dG) sequence is in a Z state of the two plasmids should be nearly identical and the l conformation while nearly all adjoining sequences are in a B same extent of relaxation should occur in both plasmids at or B-like conformation, at least with regard to the phospho- the same NaCl concentration. Since different NaCl concentradiester backbone. These results, coupled with the supercoil tions are needed to obtain the same relaxation for pRW755 relaxation data, argue strongly for a very short B/Z junction, and pRW756, it seems highly likely that the entire (dC-dG) l probably on the order of 5 bp or less. These data also dem- block undergoes an al or none transition. Increasing the onstrate that the relaxation observed cannot be caused by length of the (dC-dG) block probably increases Z conformation stability by increasing the cooperativity parameter and cruciform formation in the (dC-dG) blocks (see Results). pRW757, containing only 10 bp of (dC-dG), did not under the free energy derived from a greater relaxation of superhelgo a relaxation in NaCl concentrations up to 4.5 M and a ical density. If the all or none model holds true, then the intermediate superhelical density of -0.015. It is uncertain why the transition did not occur for this short length of (dC-dG) under these states of relaxation represent the equilibria of the B -* Z conditons. One possible explanation is that the activation transitionin the (dC-dG) blocks. However, since it seems to energy is too high. There may be a minimum number of base likely that conversion of (dC-dG) base pairs next the natural pairs necessary to obtain a nucleation in a (dC-dG) block. B DNA interface may not be as favorable as for base pairs in Alternatively, the Z conformation may not be stable in (dC- the middle of the (dC-dG) block, the equilibria may be more dG) segments of such short length. Crystallographic studies complex than in synthetic polymers. An important question was whether supercoiling is Sufi(19) indicated that a region of 4 bp of (dC-dG) abutting on cient to facilitate the conversion of the (dC-dG) blocks from AATT sequence would not adopt a left-handed structure. An unexpected result was the temperature dependence of B to Z at physiological ionic strengths. Fig. 3 shows a linear the total relaxation observed for pRW751. For synthetic ho- relationship between the number of supercoils present in a mopolymers there is no temperature effect on the equilibrium topoisomer of pRW751 and the ln[NaCl] necessary to begin
8) decreases as the initial superhelical density is decreased.

The cooperativity of the f i t phase of the transitions (Fig.

Left-handed DNA in Plasmids

the B + Z transition over the NaCl range tested. The free energy difference of the B -+ Z transition for polymer (dCdG) appears proportional to the ln[NaCl] (3) while the free energy difference of the B + Z transition as a function of superhelical density is not simply proportional to thenumber of supercoils present (20). Therefore, the linear relationship may befortuitous, a result of the limited range of superhelical density examined. It is reasonable, based on calculations by Benham (20),that the relationship of ln[NaCl] versus number of supercoils might appear linear over a limited range of superhelical densities. The quadratic component of the free energy difference (20)may not become significant until higher superhelical densities are reached. Unfortunately, the resolution limit of the gel system used here does not allow analysis of more highly supercoiled topoisomers. If the relationship shown in Fig. 4 becomes more quadratic with respect to superhelical density as the superhelical density is increased, extrapolation of the linear relationship to physiological conditions should overestimate the number of supercoils necessary to obtain a stable Z conformation. Using the equation of the line of Fig. 4, 34 supercoils or a superhelical density of 0.08 would be suffkient to obtainat least a partial stabilization of the Z conformation at 200 m~ NaC1. This is within the range of in vivo superhelical densities reported for different plasmids in E. coli (21). Thus, the data suggest that supercoiling alone is sufficient to stabilize the Z conformation under physiological conditions. However, it is uncertain whether Z DNA or cruciforms (17, 18) are more stable for (dC-dG)sequences at high superhelical densities and low ionic strengths. A number of studies (reviewed in Ref. 4) have indicated that DNA supercoilingcan modulate DNA replication, recombination, and transcription. Induction of a Z conformation would decrease the superhelical stress of a DNA segment thereby influencing these processes. Thus, the B toZ transition may be an important regulatory mechanism. Also, it is obvious that Z DNA and the B/Z junctions would provide unique protein-binding sites. Furthermore, the Z conformation could be a unique transcription termination sequence, Finally, it has been suggested that alternating purine-pyrimidine sequences could serve as a recombination hot spot (22). Since supercoiled plasmids containing (dC-dG) sequences showed high recombination frequencies (11) and since super-

10165

coiling may be sufficient to induce the Z conformation under physiological conditions, it is possible that the left-handed conformation may be present in E. coli. for these (dC-dG)containing plasmids.
REFERENCES
1. Klysik, J., Stirdivant, S. M., Larson, J. E., Hart, P. A., and Wells, R. D. (1981) Nature 290,672-677 2. Zacharias, W., Larson, J. E., Klysik, J., Stirdivant, S. M., and Wells, R. D. (1982) J. Biol. Chem. 257,2775-2782 3. Pohl, F. M., and Jovin, T. M. (1972) J. Mol. Biol. 67,375-396 4. Wells, R. D., Goodman, T. C., Hillen, W., Horne, G. T., Klein, R. D., Larson, J. E., Muller, U. R., Neuendorf, S. K. Panayotatos, N., and Stirdivant, S. M. (1980) Prog. Nucleic Acid Res. Mol. Biol. 24, 167-267 5. Wang, A. H.-J., Quigley, G. J., Kolpak, F. J., Crawford, J. L., van Boom, J. H., van der Marel, G., and Rich, A. (1979) Nature 282,680-686 6. Behe, M., and Felsenfeld, G. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, 1619-1623 7. Patel, D. J., Canuel, L. L., and Pohl, F. M. (1979) Proc. Natl. Acad. Sci. U. S. A. 76,2508-2511 8. Mitra, C. K., Sarma, M. H., and Sarma, R. (1981) Biochemistry H. 20,2036-2041 9. Pohl, F. M., h a d e , A,, and Stockburger, M. (1973) Biochim. Biophys. Acta 335.85-92 10. Wartell, R. M., Klysik, J., Hillen, W., and Wells, R. D. (1982) Proc. Natt. Acad. Sci.U. S. A . 79,2549-2553 11. Klysik, J., Stirdivant, S. M., and Wells, R. D. (1982) J. Biol. Chem. 257, 10152-10158 12. Struhl, K., Cameron, J. R., and Davis, R. D. (1976) Proc. Natl. Acad. Sci. U. S. A . 73, 1471-1475 13. Hardies, S. C., Patient, R. K., Klein, R. D., Ho, F., Reznikoff, W. S., and Wells, R. D. (1979) J. Biol. Chem. 254, 5527-5534 14. Dynan, W. S., Jendrisak, J. J., Hager, D. A., and Burgess, R. R. (1981) J. Biol. Chem. 256, 5860-5865 15. Vogelstein, B., and Gillespie, D. (1979) Proc. Natl. Acad. Sci. U. S. A. 76,65615-619 16. Yang, R. C.-A., Lis, J., and Wu, R. (1979) Methods Enzymol. 68, 176-182 17. Panayotatos, N., and Wells, R. D. (1981) Nature 289,466-470 18. Lilley, D. M.J. (1980) Proc. Natl. Acud. Sci. U. S. A . 77, 64686472 19. Wing, R., Drew, H., Takano, T., Broka, C., Tanaka, S., Itakura, K., and Dickerson, R. E. (1980) Nature 287, 755-758 20. Benham, C. J . (1981) J. Mol. Biol. 150, 43-68 21. Bauer, W. R. (1978) Annu. Rev. Biophys. Bioeng. 7,287-313 22. Shen, S.-H., Slightom, J. L., and Smithies, 0. (1981) Cell 26,191203