polyoma DNA (generously provided by R. D. Wells, WI). The absence of ..... especially thank Profs. lruno Zimm and Robert Wells for many helpful discussions ...
Proc. Nati. Acad. Sci. USA Vol. 76, No. 9, pp. 4165-4169, September 1979
Chemistry
1H nuclear magnetic resonance investigation of flexibility in DNA (correlation times/dipolar interactions/DNA backbone)
THOMAS A. EARLY AND DAVID R. KEARNS Department of Chemistry, Revelle College, University of California at San Diego, La Jolla, California 92093
Communicated by Bruno H. Zimm, May 10, 1979
EXPERIMENTAL Salmon sperm (or calf thymus) DNA (Worthington) was dissolved (10 mg/ml) in 100 ml of 0.1 M NaAcO/2 mM ZnCl2 at pH 4.6, and S1 nuclease (1000 international units) was added. This solution was dialyzed against buffer containing 0.1 M NaAcO and 2 mM ZnC12 at pH 4.6 for >36 hr followed by 5fold concentration to 500 OD/ml. The concentrated DNA was further decreased in size by digestion with Worthington deoxyribonuclease II (100 units/ml) and S1 nuclease (500 units/ ml). Enough S1 nuclease was present in the reaction mixture to completely degrade the entire amount of DNA in 100 min if it was present as single-stranded material. The course of the digestion was followed by removing aliquots of the digestion mixture for examination by NMR, and sized DNA was prepared by fractionation on an Ultrogel AcA 34 or AcA 22 column or a Sepharose 6B column. Molecular weights of the digested DNA were determined by using Bio-Rad polyacrylamide gels pre-equilibrated with Tris/glycine buffer at pH 8.9. The gels were calibrated by using Hpa II restriction enzyme digests of polyoma DNA (generously provided by R. D. Wells, WI). The absence of nicks in the double-stranded DNA was demonstrated by electrophoresis of the DNA after denaturation by NaOH. Spectra were obtained by using a Varian HR 300 proton spectrometer, modified to operate as a correlation spectrometer. With 3 mg of DNA in 100-1.l Wilmad 508 cp microcells, a typical lowfield spectrum required 15 min of signal averaging, whereas only 5 min were required to obtain a reasonable signal-to-noise ratio in the aromatic (9-6 ppm) region of the spectrum. Selective pulse experiments were carried out on a Varian HR 220 FT spectrometer. Resonance positions are in ppm downfield relative to the standard, sodium 3-trimethylsilylpropionate. The viscosities of the NMR samples were measured by using a viscometer that was calibrated with glycerol/water solutions of known viscosity.
ABSTRACT In this paper we report successful observations of proton NMR spectra of native double helical salmon sperm and calf thymus DNA of various lengths. Measurements of the linewidths arising from proton-proton dipolar interactions are used to obtain information about the dynamic behavior of DNA helices in solution. Depending upon which protons are used to monitor the local internal motions of the DNA, different results are obtained. The lowfield resonances from hydrogen-bonded imino protons in the base pairs indicate that the correlation time for reorientation of base pairs is less than 3 X 10-7 sec, whereas correlation times for motion of neighboring sugar protons relative to the aromatic protons are 2000
0
15
14
13
12
ppm
FIG. 1. Lowfield 300-MHz NMR spectra of native calf thymus DNA obtained at different stages of digestion by DNase II and S1 nuclease (for reaction conditions see text). The total time (in minutes) of reaction is indicated at the right of each spectrum and the median molecular weight, in bps, is indicated to the left. All spectra were obtained at 280C in a buffer solution containing 100 mM NaAcO and 1.0 mM ZnCl2 at pH 4.6.
double helical, and this was confirmed by electrophoresis of the denatured DNA. Furthermore, the only resonances observed in the lowfield region are due to hydrogen-bonded imino protons in native Watson-Crick bps. After these preliminary digestion experiments were carried out, new samples were subjected to different periods of digestion and then fractionated. Spectral measurements of these "sized" samples were then carried out with the results shown in Fig. 2 for the lowfield region and in Fig. 3 for the aromatic (9-6 ppm) region. Theory predicts that the rotational relaxation times of rigid DNA helices should vary with the viscosity of the solvent, and therefore the viscosities of the samples were measured; the results are shown in Table 1. The viscosities of some of the high molecular weight samples are quite large [ne23 centipoise (cP); 1 CP = 1 X 10-3 Pa-sec], and two of these samples were diluted to see what effect this might have. Rather surprisingly we observed that dilution of two high molecular weight samples decreased viscosity to 1/6th to 1/4th of its original value, but produced no change in the spectral envelope. Obviously the
macroscopic viscosity in this case is not the appropriate one to use to predict the rotational behavior of the concentrated solutions of the higher molecular weight material. Because a broad range of molecular weights was present in the higher molecular weight samples, it was possible that the resonances observed in the NMR spectra were due entirely to a low molecular weight fraction that was present. This possibility is unlikely because integrated intensities correspond to at least 80% of the expected intensity, based on the spectral range (11-15 ppm) examined, the breadth of the lines observed, and the integrated intensities observed with lower molecular weight samples. Furthermore, the spectra of 140-bp DNA from nucleosome core particles fit well with the series of DNA lengths presented here (24). The broad collection of lowfield resonances centered around 13.5 ppm are from A-T bps and those around 12.5 ppm are from G-C bps (23). Individual resonances are not resolved due to 16 different possible ring current shifts from neighboring bps and 16 additional different possible contributions from second neighbors, yielding a total of 256 different possible positions for A-T or G-C bp resonances (25, 26). The results of Figs. 1 and 2 and Table 1 indicate that, as the length of the DNA is increased from 20 to 200-500 bps, the lowfield spectrum broadens, but, with further increase in length, there is relatively little increase in the linewidth. A number of factors can contribute to the linewidth of individual resonances in the lowfield spectrum, including proton-proton and proton-nitrogen dipolar interactions, chemical shift anisotropy, proton-nitrogen scalar coupling, and quadrapolar relaxation via the nitrogen (27, 28). To estimate the protonproton dipolar contributions to broadening of the imino proton resonances (11-15 ppm) by neighboring amino group protons and other exchangeable protons, spectra obtained in H20 and in 2H20/H20 mixtures were compared (Fig. 2). In this way it is possible to evaluate separately the contribution of nearest neighbor exchangeable protons to dipolar broadening, because other sources of broadening remain unchanged. The spectra shown in Fig. 2 were used in two ways to estimate the linewidths of individual A-T and G-C resonances. In the first method, it is assumed that the increase in the width of the lowfield envelopes observed in going from 66% 2H20 to H20 can be equated to the increase in individual linewidths (see Table 1). In the second, the Nicolet 1180 line broadening program was used to broaden the individual lines in the 2H20/H20 spectra to fit the spectra in H20. Results obtained by the two methods were comparable. Aromatic Spectra. Resonances in the aromatic region of the spectrum (9-6 ppm) arise from nonexchangeable C-H protons [A-H2, A-H8, G-H8, C-H6, and T-H6 (23)]. With low molecular weight material (20 bp average), the individual linewidths are less than 30 Hz [judging from other synthetic DNA samples we have studied (29, 30)], and therefore the broad envelope between 7 and 8 ppm is attributed to a wide distribution of different intrinsic chemical shifts for the five different types of protons involved. In order to more accurately determine individual linewidths, a selective pulse experiment was done at 220 MHz on this aromatic envelope. First a homogeneous, lowpower selective 1800 pulse was applied to the aromatic envelope followed by the usual 90° observation pulse and data acquisition. Centering the selective pulse at 7.5 ppm results in "burning a hole" in the envelope. Difference spectra (data not shown) indicate that for DNA less than 200 bps long, the linewidth of the resulting hole was less than 30 Hz. A T2 measurement on the low molecular weight DNA (30-50 bps), using the standard Carr-Purcell method, yielded a value of 12 msec, also consistent with a 30-Hz linewidth. The spectra of the low molecular
Chemistry: Early and Kearns
Proc. Natl. Acad. Sci. USA 76 (1979)
II 29
I
16
I
750
154
I
,
I
12
14
16
I
14
ppm
4167
I
I
16
12
'
14
ppm
12
ppm
FIG. 2. Lowfield 300-MHz NMR proton spectra of salmon sperm DNA samples with nine different average molecular weights (shown in bps). Spectrum observed in 100% H20; -, spectrum in mixtures of 2H20 and H20. (See Table 1 for H20-to-2H20 ratios.) Samples i-iii were separated on an Ultrogel AcA 34 column, samples iv-vi were separated on an Ultrogel AcA 22 column, and samples vii-ix were separated on a Sepharose 6B column. For a description of the molecular weight distribution, see Table 1. -,
- -
developed by Woessner (31) for the motion of rigid prolate ellipsoids. In B-form DNA, two in-plane amino protons are located 2.4 A from the central imino proton in a G-C bp (one in the case of an A-T bp), and two additional protons on neighboring bps are located approximately 3.4 A above and below (32). Other protons in the molecule are located at still larger distances. When the broadening contributions from two closest in-plane amino protons and the two nearest out-of-plane protons are combined, we obtain the results shown in Fig. 4 and in Table
weight samples can therefore be used as a reference to compare with the higher molecular weight samples to estimate the increases in the intrinsic linewidths that occur when the length of the DNA is increased (see Table 1). DISCUSSION Analysis of Linewidths. If DNA is assumed to behave as a rigid rod, then the measured proton-proton dipolar contributions to the linewidths (Table 1) can be converted into correlation times (Tc) for local rotational motion by using the theory
Table 1. NMR linewidths and other properties of salmon sperm DNA helices
Samplea i ii iii iv v vi vii viii ix
Helix length, bpsb
