Volume 6 Number 4 April 1979
Nucleic Acids Research
31p NMR studies of the solution strcture and dynamics of nucleosones and DNA Leonard Klevan+, Ian M.ArmitagetM.Cpthers+' and Donald Departments of Chemistry+ and Molecular Biophysics and Biochemistryt, Yale University, New Haven, CT 06520, USA Received 24 January 1979 ABSTRACT
31p NMR studies of 140 base pair DNA fragments in nucleosomes and free in solution show no detectable change in the internucleotide 31p chemical shift or linewidth when DNA is packaged into nucleosomes. Measurements of 31p spin-lattice relaxation times T1 and 31p_@-} nuclear Overhauser enhancements revealed internal motion with a correlation time of about 4 x 10-10 sec in double helical DNA, both free in solution and bound to nucleosomal core proteins. This result implies greater dynamic mobility in double helical DNA than has previously been supposed.
INTRODUCTIQW Recently, 31p NMR has been used to probe the conformational state of the phosphodiester backbone of 140 base pair segments of DNA, both free in solution and coiled (or kinked) into a nucleosome core particle 1,2 . Previous studies have shown that the greatest degree of flexibility in DNA resides in the 0(3')-P(w') and P-O(5')(w) bonds324 with strong coupling between these dihedral angles and the O-P-0 bond angle 5 . Alterations in the phosphodiester bond geometry are accompanied by large changes in the phosphorus chemical shift6'7. For example, chemical shift changes have been observed for the helix-coil transition in DNA8'9'10, intercalation of actinomycin D into deoxyoligonucleotides8 , tertiary interactions in t-RNA and melting of small oligonucleotides . By analogy, 31p chemical shift changes might be expected for kinked sites13 within nucleosomes, since variation in sugar puckering and base stacking are coupled to the local geometry of the phosphodiester backbone 4. In agreement with earlier studies1'2 we found no appreciable difference in the 31 chemical shift between free DNA in solution and DNA in nucleosome particles. Also, we found no specific phosphate binding sites for lanthanide shift reagents in nucleosomes. The predominantly dipolar spin lattice relaxation time (T1) observed for nuC) Information Retrieval Limited 1 Falconberg Coun London WI V 5FG England
1607
Nucleic Acids Research cleosomes in solution 2 is significantly shorter than would be expected for particles with dimensions of 110 x 110 x 55 A 15. We have found clear evidence from measurement of the T1 relaxation times and 31P-fHl nuclear Overhauser enhancements (NOE) that internal motion which is sufficient to dominate the 31p nuclear relaxation rate occurs on a time scale between 10-9 and 10-10 sec in DNA and nucleosome core particles. Internal motion of approximately 10-9 sec has been shown to dominate the nuclear relaxation processes of single stranded poly A16, but the more rigid character of double helical DNA did not lead us to expect the same phenomenon in this case. Speculation on the origin of this internal motion is presented.
MATERIALS AND ]METHODS Ouir preparation of nucleosomal core particles, and their physical pro17 perties, have been described earlier . Micrococcal nuclease digestion of histone Hl-depleted chromatin yields a population of nucleosome particles having a uniform length of DNA. The particles are characterized by a sedimentation coefficient S20,w of 10.3, a sedimentation equilibrium molecular weight of 204,800, and a molar ellipticity at 282 nm of 1800 degree cm /decimole phosphate. No single strand nicks were observed for these 140 base pair DNA fragments when analyzed on polyacrylamide-urea gels or during Tm measurements. The particles appear stable in NMR buffer (10 MM Tris, 5 M Na2EDTA, pH 7.6) at 4 0C for at least 1-2 weeks, as judged by the 2 . Nucleosomes were concentrated for 3 p NMR criteria of electric dichroism~~~~~~18 by ultrafiltration on Amicon PM30 membranes just prior to each experiment. In one case ultrafiltration was used to replace the H20-containing NMR buffer by a buffer of the same composition in 99.8% D20 (Merck). The integrity of the protein-DNA complex was not altered during the NMR experiments, as demonstrated by electrophoresis of the particles on 5% polyacrylamide gels17 at the conclusion of the NMR run. DNA was extracted from nucleosome particles by incubation with 100 pg/ml of proteinase K (E. Merck) in 1M NaCl for several hours at 370C, followed by precipitation with two volumes of 95% ethanol. The precipitate was redissolved in Tris buffer, extracted with buffered phenol and concentrated by a second
ethanol precipitation.
31p NMR
spectra were recorded on a Bruker HFX-90 spectrometer interfaced to an IBM 1800 computer and operated in the Fourier transform mode at 36.4 MHz. 1.0-1.5 ml samples containing 4-6 mg/ml DNA were placed in a Wilmad
1608
Nucleic Acids Research 10 mm NMR tube (513-3PP) with vortex plug. The temperature
standard Wilmad coaxial (WGS-lOBL) and Teflon was maintained at 8 + 1 0C using a Bruker B-ST a
100/700 temperature controller. prior to
use
All tubes
were
to remove paramagnetic impurities.
washed with 50% nitric acid Chemical shifts
are
quoted
per million relative to external 85% phosphoric acid. Longitudinal relaxation times T1 were determined by the progressive saturation method 9 using delay times in the ratio of 1:4. Nucleosome T 20 values were also obtained by the inversion recovery method with nearly identical results. The 31PP-$1H3 nuclear Overhauser enhancement was determined as described elsewhere
in parts
RESULTS AND DISCUSSION Conformational Studies. P NMR spectra for nucleosomes and nucleosomal DNA are shown in Figures la and b, respectively. It is evident that only one peak is present in each spectrum, corresponding to the DNA phosphodiester linkage with a chemical shift of -1.85 ppm and a linewidth of approximately 35 Hz (Table I). Signal-to-noise ratios of up to 58:1 have been observed for the nucleosome resonance, so that regular kinks every 10 base pairs would be apparent if they were accompanied by a substantial change in the phosphodiester dihedral angles which was reflected in the 31P chemical shift. Several nucleosome preparations gave identical results. We have also attempted to use trivalent lanthanide cations as NMR shift probes, based on the possibility that the binding constant to "kinked" sites may be different from that of unperturbed phosphates in nucleosomes. Previous proton NMR studies have shown that ions such as Pr3+ and Eu3+, which have very short electron relaxation times, give rise to large shift perturbations with little line broadening when bound to the phosphates of 2 23. EuCl 3 was found to precipitate nucleosomes even at exnucleotides but we were able to measure spectra in the prelow concentrations, tremely sence of PrCl3. 10 pl samples of 400 mM PrCl3 were added to 1.2 ml of nucleosomes, 20.5 mM in phosphate, in 10 mM Tris-5 mM Na2EDTA, pH 7.6. We
Pr3
could be added up to 21 mM in excess of the EDTA which the found that solution contained before precipitation occurred. However, the phosphorous
signal-to-noise ratio was significantly reduced upon each addition of an aliquot of the praseodymium solution suggesting the formation of large nucleosome aggregates. Prior to large scale precipitation, no chemically shifted resonance of the internucleotidal phosphate was observed.
1609
Nucleic Acids Research
a)
b)
115
_ 10
I
5
0
-5
-10
-15
ppm
Proton decoupled 31P NMR spectra of (a) nucleosomes containing Figure 1. 4 mg DNA/ml in 10 t*M Tris-5 mM Na2EDTA-98% D20, pH = 7.6. 800 transients were collected with a delay time of 5.6 sec and convoluted by an exponential function equivalent to 3 Hz line broadening (b) 140 base pair DNA, 4 mg DNA/ml in 10 mM Tris-5 mM Na2EDTA, pH = 7.6. 600 transients were collected with a delay time of 5.6 sec and convoluted by an exponential function equivalent to 3 Hz line broadening. All spectra were recorded at 281°K.
TABLE I
T (sec,)
P-
JH}NOE
b12 (Hz)
T( C)
2.8
1.7
33
8
140 Base pair DNA 10 mM Tris-5 mM EDTA, pH = 7.6
2.8
1.6
37
8
Nucleosomes - 98% D20 10 liM Tris-5 utM EDTA, pD
3.6
1.6
37
8
Nucleosomes 10 mM Tris-5 mM EDTA, pH
1610
=
7.6
= 7.6
Nucleic Acids Research Dynamic NMR Measurements. The spin-lattice nuclear relaxation rate l/T1 is the sum of the rate l/T DD due to dipolar interactions and the rate l/T1 other due to the sum of all other processes, or 1
=
1
(1) 1 T1 1 other
+
T1
T1DD
When the only motion present is isotropic rotation of the whole molecule, described by a single correlation time TR) the dipolar relaxation rate for 31
nucleus is given by
a
24
2
1 1n _ C T
lDD
YP YH
R 6
1
0r.i
i1
+
(WH -
2 2
WP)TR
R 2 222 1 + WPTR 1 + (%+ WP) TR
(2) in which n protons are assumed to be dipolar coupled to a phosphorous nucleus, with r. the distance from 31P to the i'th proton; yp, and H, u% are the P and respectively, and gyromagnetic ratio and resonance frequency for i is Planck's constant divided by 2Tn. nuclear Overhauser enhancement The has the following dependence on TlDD in the extreme narrowing limit, (wp + %)TR
Nucleic Acids Research
31p NMR studies of the solution strcture and dynamics of nucleosones and DNA Leonard Klevan+, Ian M.ArmitagetM.Cpthers+' and Donald Departments of Chemistry+ and Molecular Biophysics and Biochemistryt, Yale University, New Haven, CT 06520, USA Received 24 January 1979 ABSTRACT
31p NMR studies of 140 base pair DNA fragments in nucleosomes and free in solution show no detectable change in the internucleotide 31p chemical shift or linewidth when DNA is packaged into nucleosomes. Measurements of 31p spin-lattice relaxation times T1 and 31p_@-} nuclear Overhauser enhancements revealed internal motion with a correlation time of about 4 x 10-10 sec in double helical DNA, both free in solution and bound to nucleosomal core proteins. This result implies greater dynamic mobility in double helical DNA than has previously been supposed.
INTRODUCTIQW Recently, 31p NMR has been used to probe the conformational state of the phosphodiester backbone of 140 base pair segments of DNA, both free in solution and coiled (or kinked) into a nucleosome core particle 1,2 . Previous studies have shown that the greatest degree of flexibility in DNA resides in the 0(3')-P(w') and P-O(5')(w) bonds324 with strong coupling between these dihedral angles and the O-P-0 bond angle 5 . Alterations in the phosphodiester bond geometry are accompanied by large changes in the phosphorus chemical shift6'7. For example, chemical shift changes have been observed for the helix-coil transition in DNA8'9'10, intercalation of actinomycin D into deoxyoligonucleotides8 , tertiary interactions in t-RNA and melting of small oligonucleotides . By analogy, 31p chemical shift changes might be expected for kinked sites13 within nucleosomes, since variation in sugar puckering and base stacking are coupled to the local geometry of the phosphodiester backbone 4. In agreement with earlier studies1'2 we found no appreciable difference in the 31 chemical shift between free DNA in solution and DNA in nucleosome particles. Also, we found no specific phosphate binding sites for lanthanide shift reagents in nucleosomes. The predominantly dipolar spin lattice relaxation time (T1) observed for nuC) Information Retrieval Limited 1 Falconberg Coun London WI V 5FG England
1607
Nucleic Acids Research cleosomes in solution 2 is significantly shorter than would be expected for particles with dimensions of 110 x 110 x 55 A 15. We have found clear evidence from measurement of the T1 relaxation times and 31P-fHl nuclear Overhauser enhancements (NOE) that internal motion which is sufficient to dominate the 31p nuclear relaxation rate occurs on a time scale between 10-9 and 10-10 sec in DNA and nucleosome core particles. Internal motion of approximately 10-9 sec has been shown to dominate the nuclear relaxation processes of single stranded poly A16, but the more rigid character of double helical DNA did not lead us to expect the same phenomenon in this case. Speculation on the origin of this internal motion is presented.
MATERIALS AND ]METHODS Ouir preparation of nucleosomal core particles, and their physical pro17 perties, have been described earlier . Micrococcal nuclease digestion of histone Hl-depleted chromatin yields a population of nucleosome particles having a uniform length of DNA. The particles are characterized by a sedimentation coefficient S20,w of 10.3, a sedimentation equilibrium molecular weight of 204,800, and a molar ellipticity at 282 nm of 1800 degree cm /decimole phosphate. No single strand nicks were observed for these 140 base pair DNA fragments when analyzed on polyacrylamide-urea gels or during Tm measurements. The particles appear stable in NMR buffer (10 MM Tris, 5 M Na2EDTA, pH 7.6) at 4 0C for at least 1-2 weeks, as judged by the 2 . Nucleosomes were concentrated for 3 p NMR criteria of electric dichroism~~~~~~18 by ultrafiltration on Amicon PM30 membranes just prior to each experiment. In one case ultrafiltration was used to replace the H20-containing NMR buffer by a buffer of the same composition in 99.8% D20 (Merck). The integrity of the protein-DNA complex was not altered during the NMR experiments, as demonstrated by electrophoresis of the particles on 5% polyacrylamide gels17 at the conclusion of the NMR run. DNA was extracted from nucleosome particles by incubation with 100 pg/ml of proteinase K (E. Merck) in 1M NaCl for several hours at 370C, followed by precipitation with two volumes of 95% ethanol. The precipitate was redissolved in Tris buffer, extracted with buffered phenol and concentrated by a second
ethanol precipitation.
31p NMR
spectra were recorded on a Bruker HFX-90 spectrometer interfaced to an IBM 1800 computer and operated in the Fourier transform mode at 36.4 MHz. 1.0-1.5 ml samples containing 4-6 mg/ml DNA were placed in a Wilmad
1608
Nucleic Acids Research 10 mm NMR tube (513-3PP) with vortex plug. The temperature
standard Wilmad coaxial (WGS-lOBL) and Teflon was maintained at 8 + 1 0C using a Bruker B-ST a
100/700 temperature controller. prior to
use
All tubes
were
to remove paramagnetic impurities.
washed with 50% nitric acid Chemical shifts
are
quoted
per million relative to external 85% phosphoric acid. Longitudinal relaxation times T1 were determined by the progressive saturation method 9 using delay times in the ratio of 1:4. Nucleosome T 20 values were also obtained by the inversion recovery method with nearly identical results. The 31PP-$1H3 nuclear Overhauser enhancement was determined as described elsewhere
in parts
RESULTS AND DISCUSSION Conformational Studies. P NMR spectra for nucleosomes and nucleosomal DNA are shown in Figures la and b, respectively. It is evident that only one peak is present in each spectrum, corresponding to the DNA phosphodiester linkage with a chemical shift of -1.85 ppm and a linewidth of approximately 35 Hz (Table I). Signal-to-noise ratios of up to 58:1 have been observed for the nucleosome resonance, so that regular kinks every 10 base pairs would be apparent if they were accompanied by a substantial change in the phosphodiester dihedral angles which was reflected in the 31P chemical shift. Several nucleosome preparations gave identical results. We have also attempted to use trivalent lanthanide cations as NMR shift probes, based on the possibility that the binding constant to "kinked" sites may be different from that of unperturbed phosphates in nucleosomes. Previous proton NMR studies have shown that ions such as Pr3+ and Eu3+, which have very short electron relaxation times, give rise to large shift perturbations with little line broadening when bound to the phosphates of 2 23. EuCl 3 was found to precipitate nucleosomes even at exnucleotides but we were able to measure spectra in the prelow concentrations, tremely sence of PrCl3. 10 pl samples of 400 mM PrCl3 were added to 1.2 ml of nucleosomes, 20.5 mM in phosphate, in 10 mM Tris-5 mM Na2EDTA, pH 7.6. We
Pr3
could be added up to 21 mM in excess of the EDTA which the found that solution contained before precipitation occurred. However, the phosphorous
signal-to-noise ratio was significantly reduced upon each addition of an aliquot of the praseodymium solution suggesting the formation of large nucleosome aggregates. Prior to large scale precipitation, no chemically shifted resonance of the internucleotidal phosphate was observed.
1609
Nucleic Acids Research
a)
b)
115
_ 10
I
5
0
-5
-10
-15
ppm
Proton decoupled 31P NMR spectra of (a) nucleosomes containing Figure 1. 4 mg DNA/ml in 10 t*M Tris-5 mM Na2EDTA-98% D20, pH = 7.6. 800 transients were collected with a delay time of 5.6 sec and convoluted by an exponential function equivalent to 3 Hz line broadening (b) 140 base pair DNA, 4 mg DNA/ml in 10 mM Tris-5 mM Na2EDTA, pH = 7.6. 600 transients were collected with a delay time of 5.6 sec and convoluted by an exponential function equivalent to 3 Hz line broadening. All spectra were recorded at 281°K.
TABLE I
T (sec,)
P-
JH}NOE
b12 (Hz)
T( C)
2.8
1.7
33
8
140 Base pair DNA 10 mM Tris-5 mM EDTA, pH = 7.6
2.8
1.6
37
8
Nucleosomes - 98% D20 10 liM Tris-5 utM EDTA, pD
3.6
1.6
37
8
Nucleosomes 10 mM Tris-5 mM EDTA, pH
1610
=
7.6
= 7.6
Nucleic Acids Research Dynamic NMR Measurements. The spin-lattice nuclear relaxation rate l/T1 is the sum of the rate l/T DD due to dipolar interactions and the rate l/T1 other due to the sum of all other processes, or 1
=
1
(1) 1 T1 1 other
+
T1
T1DD
When the only motion present is isotropic rotation of the whole molecule, described by a single correlation time TR) the dipolar relaxation rate for 31
nucleus is given by
a
24
2
1 1n _ C T
lDD
YP YH
R 6
1
0r.i
i1
+
(WH -
2 2
WP)TR
R 2 222 1 + WPTR 1 + (%+ WP) TR
(2) in which n protons are assumed to be dipolar coupled to a phosphorous nucleus, with r. the distance from 31P to the i'th proton; yp, and H, u% are the P and respectively, and gyromagnetic ratio and resonance frequency for i is Planck's constant divided by 2Tn. nuclear Overhauser enhancement The has the following dependence on TlDD in the extreme narrowing limit, (wp + %)TR
