Purification of closed circular lambda deoxyribonucleic acid and its sedimentation properties as a function of Sodium chloride concentration and ethidium binding.

The sedimentation of circular lambda DNA suggests that the molecular undergoes significant changes in shape and super-coiling as the NaC1 concentration increases. Closed circular lambda DNA, species I, isolated and purified from superinfected immune bacteria, sediments in sucrose gradients of low ionic strength at a rate 2.0 times faster than linear lambda DNA, species III. The addition of ethidium causes the sedimentation rate of species I DNA to decrease until enough dye is bound to remove 121 supercoils per molecule. At this point, species I co-sediments with nicked and nonsupercoiled species II. Futher additions of ethidium cause the sedimentation rate to increase until the relative rate of species I is again at least twice that of species III. This classical behavior is altered when NaC1 is present in the buffer. In 1.0 M NaC1 the changes in S are complex. Initially, species I sediments 1.55 times faster than species III. Titration with ethidium caused a decrease in S to an early minimum value, than an increase to a first maximum, followed by a decrease to the S of species II. At this point enough dye has intercalated to remove 208 superhelical turns. Further additions of dye introduce supercoils and cause S to increase again. In 0.1 to 0.4 M NaC1 the relative S of species I is 1.69 and 1.59, respectively. If titrated with ethidium, S first increases to a maximum value then decreases to the minimum rate when enough dye is bound to remove 158 and 183 supercoils, respectively. The results indicate an increase in the superhelix density from 0.026 turns per 10 base pairs in buffer alone to 0.045 in the same buffer with 1.0 M NaC1. If this change in superhelix density results from a concomitant change in the average rotation angle between base pairs in the Watson-Crick helix, the addition of 1.0 M NaC1 alters the rotation angle by 0.68 degrees per base pair.

of circular X DNA suggests that the molecule undergoes significant changes in shape and supercoiling as the NaCl concentration increases.
Closed circular X DNA, species I, isolated and purified from superinfected immune bacteria, sediments in sucrose gradients of low ionic strength at a rate 2.0 times faster than linear X DNA, species III.
The addition of ethidium causes the sedimentation rate of species I DNA to decrease until enough dye is bound to remove 121 supercoils per molecule.
At this point, species I co-sediments with nicked and nonsupercoiled species II. Further additions of ethidium cause the sedimentation rate to increase until the relative rate of species I is again at least twice that of species III.
This classical behavior is altered when NaCl is present in the buffer.
In 1.0 M NaCl the changes in S are complex. Initially, species I sediments 1.55 times faster than species III.
Titration with ethidium causes a decrease in S to an early minimum value, then an increase to a first maximum, followed by a decrease to the S of species II. At this point enough dye has intercalated to remove 208 superhelical turns.
Further additions of dye introduce supercoils and cause S to increase again.
In 0.1 to 0.4 M NaCl the relative S of species I is 1.69 and 1.59, respectively.
If titrated with ethidium, S first increases to a maximum value then decreases to the minimum rate when enough dye is bound to remove 158 and 183 supercoils, respectively. The results indicate an increase in the superhelix density from 0.026 turns per 10 base pairs in buffer alone to 0.045 in the same buffer with 1.0 M NaCl.
If this change in super-* This work was supported by Grants AI 06493 and GM 18182 from the United States Public Health Service and Grants GB 6961 and GB 2793 from the National Science Foundation. Much of the material is taken from a thesis presented to the University of Maryland bv D. M. H. in martial fulfillment of the reauirements for the-Ph.D. degree. Prelfminary reports of this work' were presented at meetings of the Federation of American Societies for Experimental Biology (20,21 helix density results from a concomitant change in the average rotation angle between base pairs in the Watson-Crick helix, the addition of 1.0 M NaCl alters the rotation angle by 0.68" per base pair. Increasing the ionic strength causes a decrease in the sedimentation rate of circular X DNA (1, 2) but has relatively little effect on the sedimentation rate of linear or nicked circular X DNA.: Since the molecular weight is constant, shape or flexibility changes in the supercoiled molecule must account for the rather pronounced changes in the sedimentation velocity. Experiments reported in the preceding paper suggest that the number of potential supercoils, as estimated from ethidium dye binding affinity, increases as the sodium chloride concentration is raised from 0.1 to 1.0 M (3). Since we had not expected an increase in supercoiling to cause a lowering of the S value, the sedimentation velocity of circular X DNA was examined in greater detail as a function of both ionic strength and the number of supercoils.
The sedimentation rate of purified circular X [3H]DNA is measured relative to a linear X [V]DNA marker as a function of bound ethidium and (thereby the number of supercoils) at four NaCl concentrations, 0.0, 0.1, 0.4, and 1.0 M in 0.01 M Tris-HCl plus 10 mM sodium-EDTA buffer. In each solution and at each dye concentration, the number of potential supercoils present in a circular X DNA molecule is calculated from the nurnber of ethidium molecules bound relative to the number which are required to remove all supercoils, i.e. the number bound when 1 We refer to circular DNA molecules whose component single strands form two intertwined but unbroken circles of phosphodiester bonds as simply circular DNA molecules (frequent synonyms : species or form I, covalently closed circular DNA).
Molecules in which the chains of phosphodiester bonds have been broken by hydrolysis or other means are referred to as nicked circular DNA (frequent synonyms: species or form II, open circles). The latter is not a homogenous collection since the number a,nd location of chain interruptions may vary within the population of molecules.
Species or form III is linear duplex DNA. Single-stranded circular molecules do not enter this discussion. Step 1-E. coli 1100, an endonuclease I deficient mutant (7) was lysogenized for X to serve as an immune host for superinfection. Cultures of E. coli 1100 (X) were grown to a density of 5 X lo8 cells per ml at 37" in tryptone broth supplemented with 0.02 mg per ml of thiamin.
The cells were sedimented 9,000 rpm, Sorvall GSA rotor, 1 hour, 5". The cell paste from l-to 3-liters of culture was suspended in 500 ml of cold X-dil.
(The protocol will accommodate a maximum of 3 liters or 1.5 X 1Or2 total cells). Purified XCSO phage suspended in 400 ml of cold X-dil were added at a multiplicity of 10 and allowed to adsorb for 10 min at 0". The complex was shaken for 10 min in a 6-liter flask in a water bath at 37" to allow infection (almost 70010 of the radioactivity in the input phage is injected).
One In this manner as little as 0.01 pg per ml of free ethidium was measured.

RESULTS
Step Q-After dialysis, solid CsCl and ethidium bromide were added to give a density of 1.65 g per ml (720 = 1.3870) and a final ethidium bromide concentration of 100 pg per ml in a total volume of 29 ml.
(When more than 1 liter of cells was infected, the DNA pool was divided into two 29 Fig. 1. The X [3H]DNA molecules must be in the species I form to band in the more dense peak. Based on the counts per min per ml to A260 ratio, or the presence of labeled linear marker DNA, the peak fraction is contaminated by no more than 5% with molecules from the less dense band of linear and nicked circular molecules.
Nevertheless, after removal of ethidium and dialysis, only 70 to 807, of them sediment as species I. The remaining 20 to 30% are nicked circles. The molecular weight of X DNA is 10 times greater than SV40 or polyoma DNA. Therefore, a circular X DNA molecule is 10 times more sensitive to conversion to species II by treatments that cause or permit single strand breaks and more care must be taken in its purification.

Sedimentation
Velocity as Function of Bound Ethidium-Several studies, using various circular DNA molecules and ethidium intercalation, have employed sedimentation velocity to monitor the unwinding and winding of superhelices (9-15). Although measured directly in a few critical instances, the latter was estimated routinely using the Scatchard equation2 (16).
The ethidium binding isotherms at the four ionic strengths used in this study were measured in several concentrations of sucrose. The inclusion of sucrose did not alter the value of K or n. Table II lists the values for these constants determined in 25% sucrose. The corresponding values determined in the absence of sucrose can be found in Table II of the preceding  paw (3). Contrary to the results reported by Le Pecq and Paoletti (17), sucrose did affect the fluorescence of free ethidium. Fluorescent intensity increased about 15% when the buffer contained 25% rather than 5% sucrose. This was accompanied by a slight red shift in the emission spectrum.
Sucrose did not change the fluorescent yield from ethidium when it was bound to DNA. Table III summarizes the conditions used in 12 gradients with varying amounts of ethidium and no added NaCl.
Routinely, the amount of X DNA sedimented was less than 2 I.rg, and the amount of bound dye was calculated from the free dye concentration in the gradients using the Scatchard equation.
To check the validity of the methods, larger amounts of marker DNA or circular DNA were added to samples 6 through 12. This permitted the direct measurement of bound ethidium by fluorescence. Since the DNA concentrations are known from the counts per min per ml in each fraction, r can be determined directly and compared with the value calculated using the free dye concentration only. The two gradients displayed in detail in Fig. 2 present results from a detailed study of the gradients.
The results indicated that even under these conditions, where greater than normal amounts of DNA moved through the upper portion of the linear sucrose gradient, the free dye concentration remained constant at the initial value throughout the tube during centrifugation.
As verified by the direct measurement of bound dye, Fig. 2 and Table III, the amount of ethidium which binds and sediments with X DNA is accurately calculated using the free dye concentration in the sucrose gradient and the previously determined binding constants. The 3H counts were associated with purified circular X DNA (5660 cpm per pg) which contained 157, nicked circular molecules.
The 32P counts, O-O, were associated with a linear X DNA marker.
Unlabeled linear X DNA was added to the marker before sedimenting, yielding a specific activity of 600 cpm per pg. The fluorescence, A---A, at 590 nm (530 nm activation) is primarily due to the dye bound to linear DNA.
In the peak fraction of the s2P profile there was 6.60 pg per ml of linear X DNA and 0.36 pg pes ml of bound ethidium bromide. Therefore, r was 0.046. Srei of circular X DNA with respect to linear is 1.30 under these conditions. Table III, sample 9, gives further details on this gradient.
Sedimentation: 4 hours, 20", 25,000 rpm, SW 25.1 rotor. Bottom, the detailed conditions are given in Table III, sample  Examples of more typical gradients performed to determine the relative sedimentation rate, L&i, as a function of salt concentration and ethidium bound are shown in Fig. 3. Srer is the distance sedimented by the X 3H-labeled species I divided by the distance sedimented by the linear h [32P]DNA marker present in the same tube.
In the one gradient (top) enough dye was bound to remove 17 of the 208 negative supercoils believed to exist when no dye is present.
In the other gradient, the DNA examined had been Under all of the conditions examined in this study, one observes only two peaks of X 3H-labeled circular molecules, the rapidly sedimenting supercoiled form and the nicked circles sedimenting 157, faster than the linear marker. The species I molecules sediment as a single sharp band suggesting that,, at least after time-averaging, they are homogeneous with respect to dye intercalation and supercoiling. The species I used for gradients where &,I was near minimal values, i.e. all supercoils removed, was further purified in a Buffer A sucrose gradient just prior to sedimentation in the ethidium containing gradient.
This yielded a preparation that was 90% species I and avoided complications in interpreting gradients containing significant amounts of species II whose band pattern would overlap with that of unwound species I.
Relative Sedimentation Rate in Buffer A as Function of Bound Ethidium-The variation of Srel as dye is bound and supercoils are removed is seen in Fig. 4. The value of Srel decreases from a value of 2.0 (2, 18) to a value of 1.12 as Cf increases from 0 to 0.38. Binding more dye introduces positive supercoils and causes Srel to increase.
A slight decrease in the s20,w value of species II or linear DNA as ethidium is bound was reported (9, 11, 19)  The changes in Srei are attributed predominantly to changes in the rate at which species I sedimented.
Sedimentation Velocity as Function of Bound Ethidium in 1 .O M NaCl-A complex and nonmonotonic variation in Srei was observed when 1 M NaCl was included in the Buffer A-ethidium sucrose gradients (20,21).
The results of 48 gradients performed with several different species I preparations are shown in Fig. 5. The intercalation of enough ethidium to remove 22 supercoils caused a decrease in Srei to a first minimum.
This was followed by an increase in Srei as more ethidium bound. About 112 supercoils were removed at the midpoint of a first maximum. With additional ethidium, Srei decreased again to a second minimum, lower than the first, whose value is characteristic of molecules lacking supercoils.
At this point enough ethidium was bound to remove 208 supercoils.
In higher concentrations of ethidium, Srei increased to a second maximum, presumably due to the introduction of supercoils of the opposite (positive) handedness.
Although it was not as well characterized as the first minimum, a third minimum in Srei appeared when enough dye was bound to both remove 208 negative supercoils and introduce 124 positive ones. The value of Srei then increased again and (not shown in the figure) was 1.95 at Cf equal to 20 pg per ml.
The Srei versus C; graph (Fig. 5) is the raw data and includes no assumptions or corrections related to dye binding affinity or binding capacity. Plots of Cf versus r for circular X DNA prepared from data in the previous paper (3)  dye concentration.
Values for the superhelix parameters at each of the maximums and minimums mentioned above were then calculated and are presented in Table IV.
Sedimentation Velocity of Species I as Function of Ethidium Concentration at Intermediate Salt Concentrations-The shape of the Srei versus C, curve at intermediate iYaC1 concentrations (Fig. 6) is intermediate between that seen for Buffer A alone and Uuffer A plus 1 M salt. S rei in the absence of ethidium is lowered by the addition of salt. Binding ethidium causes an initial increase in the sedimentation rate of circular molecules, but as more dye is bound the loss of potential supercoils eventually leads to a minimal relative rate equal to 1.1. The relative sedimentation rates as a function of ethidium bound at the four ionic strengths examined are compared in Fig. 7. The values of C/ and r for the sedimentation minimum, where nicked and closed circular molecules co-sediment and the latter has no supercoils, are shown in Table V. Assuming a constant unwinding angle due to ethidium intercalation with a value equal to -12" (22), the superhelix parameters indicated in the table were  calculated for circular molecules in the four different ionic environments.

1)ISCUSSION
As reflected by the sedimentation rate, the number of ethidium molecules which must intercalate to remove all supercoils from a circular X DNA is a function of the NaCl concentration.
It was not possible to obtain good fluorometric data for ISuffer A alone, but for the other salt concentrations, the free ethidium concentration at the major minimum in relative sedimentation rate is identical with the value of Cf at the point of equivalent dye affinity determined fluorometrically (3). Although plots of r  Table III  verSus Cf determined fluorometrically are used to convert these values of Cf to the corresponding binding ratios for estimating the number of supercoils, the end point determinations themselves are independent of each other. Their coincidence demonstrates that equivalent dye affinity and the co-sedimentation of nicked and closed circles are achieved by intercalating the same number of dye molecules, i.e. by unwinding the primary helix the same number of turns.
Both methods support the conclusion that increasing salt concentration increases r" and therefore must We use the term unwinding for the effect of ethidium on the primary helix and winding for the effect of increasing the NaCl concentration.
This is in keeping with the sense and handedness of supercoils as originally discussed (9) and with the view that ethidium intercalation unwinds the helix by 12" (22) rather than winding it by about the same amount (23). Our data, however, only demand that the effects of salt and ethidium be of opposite -There have been a number of reports on the linear expansion of duplex DNA at low ionic strength as visualized by electron microscopy (24, 25). Expansion at low salt is believed to occur by a uniform partial unwinding of the normal B configuration helix. A salt-dependent winding at high ionic strength and unwinding at low is consistent with the results obtained in these studies.
If circular X DNA is supercoiled because the primary helix already is underwound (as circular molecules from natural sources all appear to be), further winding of the Watson-Crick helix in high salt will lead to an increase in the number of supercoils.
Increased temperature also causes an unwinding of the helix which is measurable with circular molecules (15). Therefore in this and the previous paper a constant temperature at 20" was maintained during sedimentation and dye affinity measurements. We use the term potential supercoils in relating the number of ethidium molecules intercalated to the number of supercoils removed because we cannot be certain that this number physically exists. The physical and energetic requirements of supercoiling at high superhelix densities might induce changes in the primary helix which reduce the average rotation angle between base pairs (15,26). The ethidium titration would yield the same end point as if the supercoils existed since it would require an equivalent unwinding to both return the helix to its normal parameters and remove the residual negative supercoils.
One reason for initiating this study was the conflict between the number of supercoils in X DNA in high salt as estimated by electron microscopy (1) and by ethidium intercalation (15). When circular X DNA was spread from a cytochrome c solution of low (0.06) ionic strength, most of the visualized molecules had more than 90 duplex helix crossings. The average number, 117, is very similar to the 7' value of 121 obtained here for circular  DNA in Buffer A. When the DNA was spread from cytochrome c solution in 2 M NaCl, a majority of the molecules had fewer than 40 supercoils with an average of 12. Not only is this quantitively very different from the 208 potential supercoils obtained here by ethidium intercalation and sedimentation studies, but it suggests an ionic effect of opposite sign.
Initially we were misled by the early minimum in &I (Fig. 5) and submitted an abstract indicating that the dye binding sedimentation data also suggested a lower number of supercoils at high salt concentrations (20). Fortunately, before the work was actually presented, the complete picture, as seen in Fig. 5, had emerged but the abstract remains in error.
Since that report, the nonmonotonic sedimentation behavior in ethidium titrations of circular DNA with potentially high superhelix density has been observed by a number of other workers with circular DNA from several sources (13,(27)(28)(29).
The shape and stiffness transitions reflected in the sedimentation rate pattern can also be detected by viscosity measurements (30).
From their electron micrographs of circular SV40 DNA,

1079
(3 x lo6 daltons) with various superhelix densities and from theoretical consideration of the hydrodynamic properties of circular DNA, Upholt et al. (27) proposed a model for changes in DNA tertiary structure as a function of superhelix density. It explains the nonmonotonic sedimentation patterns (similar to those presented here), on the basis of changes in looping, branching, and stiffness of the supercoiled DNA as a function of u.
Since it is clear that to remove all of the negative supercoils more dye must be bound at high salt concentrations than at low, the electron micrographs prepared from 2 M NaCl-cytochrome c solutions reported by Bode and MacHattie (1) remain as a discordant result.
Those prepared from low ionic strength gave a reasonably consistent value for 7'. Although additional experimental support would be necessary to establish the electron microscopy-result as other than an artifact of the method, other possibilities cannot be completely eliminated by the existing data.
A number of reports have suggested DNA may exist in solutions in other than the 1~ form (e.g. Refs. 31-33) and a forced transition in the secondary structure to a helix form with more than 10 base pairs per turn could explain the ethidium binding, microscopy, and sedimentation results in 1 M NaCl. When circular X DNA supercoils more tightly, as it does in high salt, it might reach a state where it is energetically more favorable to alter the B form of the primary helix than to supercoil.
As more bases per turn are accommodated in the distorted form of the primary helix, fewer supercoils are demanded and physically present.
Nevertheless, removal of all supercoils by ethidium intercalation would require a complete reversal of the process as indicated in the discussion of actual and potential supercoils.
The properties of circular DNA can be acquired by any DNA under special conditions. When linear molecules are restricted at two points so that single strands of the duplex cannot rotate about each other, the region of DNA between the points behaves like a circle. In bacteria, topological division into a number of loops with independent supercoiling has been reported for the large folded chromosome of E. coli (34).
The shape transitions suggested by the change in Srei about the first local minimum (Fig. 5) occur over a relatively narrow range of superhelix density. Therefore, a topologically restricted region of linear DNA could be forced to undergo this change by a rather small variation in the local environment.
It will be of interest to learn whether cells ever utilize this possibility in a biologically significant way to regulate the shape or function of their chromosomes.