Buoyant Density Studies on Natural and Synthetic Deoxyribonucleic Acids in Neutral and Alkaline Solutions*

SUMMARY Equilibrium buoyant density centrifugation studies in CsCl and C&SO4 solutions (both neutral and alkaline) were performed on 11 DNA polymers and 12 naturally occurring DNA%. For the polymers, no apparent relationship exists between DNA nucleotide composition and the extent of density change on alkaline titration. Thus it is not possible to predict the nucleotide composition of a natural single-stranded DNA from its shift in buoyant density. Large decreases in buoyant density were found on alkaline titration of some DNA polymers in Cs2S04 solution. Also, natural DNA’s show a net density decrement due to alkaline titration. Contrastingly, in CsCl solutions, only buoyant density increases were observed.


ROBERT D. WELLS AND JACQUELYNN E. LARSON
From the Department of Biochemistry, College of Agricultural and Life Sciences, University of Wisconsin, Madison, Wisconsin 53706 SUMMARY Equilibrium buoyant density centrifugation studies in CsCl and C&SO4 solutions (both neutral and alkaline) were performed on 11 DNA polymers and 12 naturally occurring DNA%. For the polymers, no apparent relationship exists between DNA nucleotide composition and the extent of density change on alkaline titration. Thus it is not possible to predict the nucleotide composition of a natural singlestranded DNA from its shift in buoyant density.
Large decreases in buoyant density were found on alkaline titration of some DNA polymers in Cs2S04 solution. Also, natural DNA's show a net density decrement due to alkaline titration. Contrastingly, in CsCl solutions, only buoyant density increases were observed.
Analytical density gradient centrifugation is a standard means of characterizing DNA samples (for a review see Reference 1). Both C&l and Cs2S04 solutions have been widely used since correlations exist between DNA base composition and density values, except for some special cases such as glycosylated phage DNA's or some DNA polymers (14).
Equilibrium density gradient centrifugation in alkaline solution has been employed for the separation of complementary strands of certain natural DNA's (5-10) (however, better methods now have been developed to achieve strand separations (1)). Previous authors have attempted to ascertain the nucleotide composition of single-stranded DNA from the magnitude of the density shift from neutral salt to alkaline salt solution.
It was assumed in these extrapolations that (a) only guanine and thymine moieties are titrated in alkaline solution and thereby acquire a cesium ion which increases the density of the DNA (3,11,12), and (b) that our previously reported density values (3) for the strand separation of poly(dA) .poly-(dT) and poly (dT-dG) .  (3). The parent duplexes were prepared by described procedures and were: poly(dA) . poly(dT) (4)) poly(d1) . poly(dC) and poly(dG) . poly(dC) (13), and poly (dT-dG) .poly(dC-dA) (14). The singlestranded DNA's were characterized by spectral analyses and were shown to be free (less than 5%) of any contaminating DNA by analytical density gradient centrifugation.
Because of the unexpected results observed with poly(dC), this polymer was prepared by strand separation of both poly(dG) .poly(dC) and poly(d1) poly(dC).
Identical results were uniformly observed for all preparations of this DNA. Poly(dT-dC) was prepared by treating poly (dT-dC) .poly(dG-dA) (15) with formic aciddiphenylamine mixture accordin, v to Burton and Petersen as previously described (16). After ether extraction of the reaction mixture, the polymer was exhaustively dialyzed to remove degradation products. Spectral analysis served to characterize the polymer.
Poly(dAdT) . poly(dA-dT), poly(dI-dC) . poly(dI-dC), and poly(dG-dC) . poly(dG-dC) were prepared as described (2,4). Bacteriophage Ml3 DNA (gift of R. W. Sweet) was prepared by extracting the virus (gift of D. Pratt, University of Wisconsin) with phenol and subsequent dialysis (2). Sedimentation velocity studies in 0.9 M NaCl-0.1 M NaOH solution were performed both on the native DNA and on the DNA after mild acid treatment (17) ; the results indicated that the native DNA was completely (within experimental error) in the circular form. The single-stranded DNA's from H-l virus and the minute virus of mice (gifts of L. Crawford, Imperial Cancer Research Fund, London) were isolated from the viruses by the sodium dodecyl sulfate procedure of Crawford (18 (3) and range from 0.5 to 25 x 106. Most of the prior to (or just after) centrifugation was measured as 12.5 to DNA polymers used herein were the actual preparations char-13.3 with a small combination electrode (20) which had been acterized by other determinations including ultraviolet spectra standardized with a calcium hydroxide standard solution (pH (3, circular dichroism spectra (2), absorbance-temperature 12.5 at 25"). At these quite alkaline pH values, it is realized profiles (2,14,20), actinomycin D-binding studies (19,21), that these values simply represent instrument readings.
(11) have inferred that the high concentration of salt which is Density Gradient Centrifugation-Density gradient centrifuga-present does not affect this reading. tion studies were performed in a model E ultracentrifuge modified for four cell operation as previously described (2). The CsCl RESULTS runs were at 52,640 rpm and the Cs.#Oh runs were at 44,770 rpm; Guanine and thymine bases are ionized in alkaline solution; all were performed at 25". All neutral CsCl values are relative thus, it is expected that a DNA containing these bases would to an E. coli DNA density of 1.703 g per ml. The actual become more dense in alkaline solution due to the subsequent density marker was usually poly(dA-dT) . poly(dA-dT) (1.672 g acquisition of cesium ions. In addition, if a DNA contains per ml) (3). In alkaline CsCl solution, poly(dA-dT) . poly(dA-any secondary structure in concentrated salt solution, this dT) was the marker at 1.722 (3). This DNA also served as the ordered configuration will be virtually eliminated in alkaline density marker in neutral and alkaline CseSOd solutions; the solution, thus providing for an additional density increment. determination of the absolute value in neutral solution was However, the latter increment is probably relatively small comreported (19). The alkaline density value of 1.416 g per ml for pared to the former increment (11). this DNA was established by two methods.
First, three sep-CsCl Density Gradients-The buoyant densities of eight singlearate density determinations by the isoconcentration distance stranded DNA polymers in neutral and alkaline cesium chloride method of Ifft, Voet and Vinograd (24) each gave a value of solution are shown in Table I. Three of the single-stranded 1.416 g per ml; second, identical neutral solutions containing DNA's studied contain only bases which are not normally ionpoly(dA-dT) .poly(dA-dT) and poly(dA) were banded at the ized at pH 12 to 12.5 and are poly(dA), poly(dC), and poly(dCsame time in two cells. Both bands in both cells were found to dA). Poly(dA) and poly(dC-dA) do not increase in density as perfectly align and the density of poly(dA) was found to be the pH of the solution is raised but, in fact, show a slight dec-1.379 g per ml, in good agreement with a previously reported rement of 5 and 4 mg per ml, respectively. That these obvalue (25). Next, 0.005 ml of 4 N NaOH was added directly served decrements are genuine was proved by the following into one cell with a Hamilton syringe and 0.005 ml of water was additional experiment. Two identical solutions of polymer in added to the second cell. Upon attainment of equilibrium, the salt solution at pH 7.3 were centrifuged to equilibrium when the poly(dA) bands were still in perfect alignment in the two cells, bands were found to perfectly align in the two cells. Then, indicating that the density of poly(dA) in CS~SO~ solution is not O.OlOmlof water was added to one cell and 0.010 ml of 4 N NaOH was added to the other cell. On recentrifuging the DNA solu- (1 This DNA is not single-stranded at neutral pH but is renaturable. three of these single-stranded DNA's undergo a marked density increase when the cesium chloride solution is made alkaline. However, the density increase was not approximately the same for all three cases. Poly(dT-dG) showed a density increase of 32 mg per ml, whereas poly(dT) and poly(dG) showed density increases of 35 and 26 mg per ml, respectively.
Poly(dT-dG) has no ordered structure in neutral solution as judged by absorbance-temperature studies.' It has been reported that poly(dT) has no secondary structure in neutral solution (25), whereas a highly ordered structure might be expected for poly(dG) under these neutral conditions.
Unexpectedly, poly(dG) shows the smallest buoyant density increase which might be interpreted as indicating that this DNA undergoes the least loss of ordered structure as the solution is made alkaline and hence possesses little (or no) order in neutral solution.
These results suggest that each of these polymers has unique and characteristic properties, which are as yet poorly understood. In addition, it should be noted that the density increase observed for these three single-stranded DNA's, which contain 100% alkalititratable nucleotides, is approximately the same as observed for naturally occurring single-stranded DNA's which contain only 55 to 60% guanine plus thymine (see below).
Poly(dT-dC) shows a density increment of only 7 mg per ml on titration which is much less than found for either poly(dC) or poly(dT).
Poly(d1) shows a density increment of 24 mg per ml on titration.
For comparison, density studies were performed on four naturally occurring single-stranded DNA's (Table II). 4X174, M13, minute virus of mice, and H-l DNA all contain 54 to 57% guanine plus thymine. Accordingly, they have similar buoyant density values in both neutral and alkaline solution.
The satellite (dA-dT) from C. productus, which contains a small amount of guanine + cytosine (28), behaves similarly to biosynthetic poly(dA-dT) + poly(dA-dT) (see below). The three double-stranded DNA polymers, poly(dA-dT) . poly(dA-dT), poly(dG-dC) .poly(dG-dC), and poly(dI-dC) . poly(dI-dC), are also amenable to this investigation due to their repeating, self-complementary nucleotide sequence. Because of the identical nature of the complementary strands of each of the double helices, each DNA exhibits a single band in both neutral and alkaline gradients.
Density increments at least as large as those found for natural single-stranded DNA's were expected since all three DNA's possess highly ordered structures in neutral solution (2,20). In addition, exactly one- 1 R. D. Wells,unpublished work  a Five different authentic preparations of poly(dT-dG) were studed. All showed a single sharp peak in alkaline CstSOl solution at 1.492 g per ml. However, in neutral CSZSO~ solution, four bands (in varying amounts) were observed for all five preparations (1.548, 1.537, 1.524, and 1.494). These densities are f2 mg per ml.
half of the constituent bases of each of these DNA's can be titrated with alkali.
Density increments of 50, 52, and 31 mg per ml, respectively, were found (Table 1110). For natural single-stranded DNA, density increments of 38 to 41 mg per ml are found (Table II) and for natural double-stranded DNA, increments of approximately 62 mg per ml were found (11). Since these three DNA polymers are renaturable (2, 29) a rigorous comparison is not possible.
Cs$01 Density Gradients-In a further effort to establish a relationship between nucleotide composition of single-stranded DNA and buoyant density, analyses were performed in both neutral and alkaline CQO4 solutions.
Unexpectedly, buoyant density decrements induced by titration were found for a variety of DNA's. Table IIIb shows that both poly(dA-dT) .poly-(dA-dT) and poly(dI-dC) . poly(dI-dC) undergo a density decrease, whereas poly(dG-dC) . poly(dG-dC) undergoes a density increase on titration.
Two additional determinations Buoyant Densities of DNA 's Vol. 247,No. 11 are described under "Experimental Procedure" that verify the density decrements. Table IV shows that neither poly(dA) nor poly(dC-dA) undergo dramatic density shifts on titration as expected from Table I. However, poly(dC) undergoes a density decrement of 98 mg per ml. Fig. 1 shows the titration of poly(dC) from pH 7.1 to 12.6. A sharp drop in density is found at pH 7.9. This is consistent with the high pK of poly(dC) (27). The highly ordered protonated structure has a density of 1.520 and the single-stranded coil has a density of 1.422. Poly(dT) Il.420/ * )  a This DNA is not single-stranded at neutral pH but is renaturable.
undergoes a density increase of 19 mg per ml but poly(dG) undergoes a density decrease of 52 mg per ml. Poly(d1) shows a very large density decrement. A similar analysis was not possible for poly (dT-dG) due to aggregation in neutral solution; however, it clearly undergoes a density decrement also. Table V shows that the four single-stranded viral DNA's also undergo a density decrement of 10 to 13 mg per ml on titration. As in CsCl solutions, C. productus (dA-dT) behaves similarly to biosynthetic poly (dA-dT) . poly(dA-dT) . Studies were performed on seven natural double-stranded DNA's (Table VI).
The density values on native DNA's are in excellent agreement with reported values (1). On heat denaturation, density increments of 19 to 28 mg per ml were observed. However, in contrast to the results found in CsCl solution (ll), greater increments were found for DNA's with higher guanine + cytosine contents.
Buoyant density increments were also found on alkaline titration of the DNA's, except for TI DNA (Table VI).
However, in no case was the increment induced by titration as great as that induced by heat denaturation.
A linear relationship was found between base composition and buoyant density increments induced by titration.
Thus, in CszS04 solution, titration of a DNA (with ensuing denaturation) gives rise to a smaller density increment than the loss of secondary structure alone (heat-denatured DNA). Hence, a net buoyant density decrement is found due to titration of the bases. The opposite behavior is found in CsCl solution (see "Discussion").
The reason for this difference in behavior between CsCl and C&O4 solutions is unclear but may be related to the effective hydration of the nucleic acid under different conditions (30). DISCUSSION Vinograd et al. (11) previously reported that the magnitude of the change in buoyant density induced by alkaline titration of a DNA was comprised of at least two components.
First, denaturation of the DNA helix gave a density increment of 13 to 19 mg per ml (in CsCl solution) ; larger increases were found for DNA's with lower guanine + cytosine contents.
Second, alkaline titration of the bases gave an additional density increment of 43 to 49 mg per ml (in CsCl solution) ; larger increases were found for DNA's with higher guanine + cytosine contents In CsCl solutions, alkaline titration produced an over-all density 9. LEFFLER, A. T., II, CRESKOFF, E., LUBORSKY, S. W., MCFAR-LAND, V., AND MORA, P. T. (1970)  106, 23&242 This study was undertaken to attempt to establish a relationship between DNA nucleotide composition and buoyant density change induced by alkaline titration.
DNA polymers were the models used because of their simplicity and because unequivocal conclusions can be drawn since they contain completely defined nucleotide sequences and compositions. The results clearly show that a simple relationship does not exist. A variety of workers have attempted to predict the base composition of natural DNA samples using our previously reported density values on some polymers as standards (see above). The results of this more complete study indicate that such extrapolations are not valid.
Unexpectedly, alkaline titration of all natural DNA's and most polymers in C&SO4 solution elicits a net density decrement, not a density increment as found in CsCl solution.
The reason for this behavior is not clear at present.
An understanding of this, and related buoyant density phenomena may have to await the elucidation of the physical and chemical parameters which govern the observed density of a DNA sample.