Proc. Nat. Acad. Sci. USA Vol. 72, No. 3, pp. iO43-1045, March 1975
Quaternary Structure of Chromatin (neutron scattering/new coiled-coil model)
STANLEY BRAM*, GILL BUTLER-BROWNE*, PIERRE BAUDY*, AND KONRAD IBELt * Departement de Biologie Molculaire, Institut Pasteur, Paris 75015, France; and t Institut Laue-Langevin, Grenoble, France Communicated by Franiss Jacob, November 29, 1974 Neutron scattering experiments on chroABSTRACT matin solutions have been carried out at Bragg spacings
several times larger thani in any previous x-ray work. A new series of very low angle reflections at about 400, 200, and 140 A- were observed. These reflections demonstrate the existence of a higher order structure of the 100 A unit fibril. A coiled-coil model with a major pitch of 500 A and a radius of center to fibril center of 130 A is proposed to explain the results.
eter and monochromatization, instrumental smearing of the scattering patterns are considerably less than with the x-ray diffractometers used for previous small angle x-ray scattering (2). Even so, the' maxima in the scattering profile are somewhat smeared out. For the discussion that follows,, it is worth discussing some of the features of neutron scattering. The neutron scattering amplitude of each atom or isotope is dependent only on the nuclear structure and not oh the atomic number, as is the case for x-rays. Thus, the scattering from a molecule in solution at zero angle is a function of the sum of the scattering lengths of its atoms or of the atoms in its compl8hent molecular parts, minus that in the same volume of solvent. H20 has a .scattering length density of -0.6 X 1010/cm3, while pure D20 has a value that is 10 times larger and positive. Because the density and atomic abundances of DNA and protein are dissimilar, the neutron scattering from each will vary rather differently with the solvent scattering length density, which is only a function of the H20/D20 content of the solvent. At a D20 content of 38%o by volume, the scattering from histone will be almost zero, whereas DNA will scatter very weakly in the neighborhood of 65% D20. For the results of this paper, it suffices to point out that if a reflection or a scattering ring is prominent in D20 concentrations less than about 45% and absent in the neighborhood of 65%, it may be attributed to scattering from DNA. It must be stressed that this contrast matching treatment is only valid at low concentrations and small angles, and if the structure is invariant in D20/H20
Until recently, structural studies of chromatin have been oriented towards determining the secondary or tertiary coiling of chromosomal DNA (1-3). Consequently, mainly those features smaller than about 110 A have been investigated. Evidence that suggests that the tertiary structurebof DNA and protein consists of globular units approximately 80-100 ! in diameter arranged like a string of beads on a thread has appeared. This evidence has come mainly from nuclease digestion studies and electron microscopy of chromatin and nuclei (4-6). However,, except for the work df Ris, which implies that the 100 A unit thread doubles up in the presence of calcium ions (7), the highei-order folding of the unit thread has been somewhat neglected. Previous x-ray scattering studies on chromatin, although directed at tertiary structure of the order of 100 A, often showed weak inflections in the scattering profile at spacings corresponding to about 400 and 200 A (see, for example, Fig. 2 of ref. 2),. It now seems that the weakness of these low angle x-ray reflections was due to slit smearing from the x-ray diffractometer. mixtures. This paper presbnts the results obtained from neutron investigations of the higher-order arrangement of the 100 A RESULTS unit thread. Very low angle reflections are indeed found to be Neutron Scattering. Solutions and gels of chromatin were quite prominent in the neutron scattering at spacings of up studied over a concentration range of 3-25 mg/ml, between to 460 A. It is shown that a quaternary chromatin structure Bragg spacings of 1500-80 A. Between 8 and 10 equivalent does exist, and it is suggested that a coiling of the 110 A unit H20/D20 contents were examined on more than eight inthread with an average pitch of about 500 A and a radius of preparations. Figs. 1 and 2 show some of the results dependent 130 A is consistent with these results. of typical scattering curves at different H20/D20 contents. A METHODS distinct shoulder or peak is seen in the scattering profiles at a Bragg spacing equivalent to about 400 A when the D20 Calf thymus chromatin was isolated by the method of Panyim content is less than about 45%$T. On the other hand, this very et al. (8), which is based on the isolation of nuclei, their purilow angle reflection is absent at higher D20 contents. Maxima fication, and their subsequent gentle disruption. Chromatin at smaller spacings-about 220 and 140 A-are often seen in samples were concentrated by 'centrifugation, and all samples against dialyzed were controls with the exception of a few three changes of the chosen H20/D20 mixture, with a known t The spacing varies between 400 and 460 A, which is consistent amount of salt. with experimental error. It seems more prudent to refer to a Neutron scattering experiments were done with the NILS spacing "at about 400 A" than to give an exact figure which may diffractometer built by K. Ibel (9). Exposure times were on change as our precision improves. Furthermore, this spacing varies with the solution environment. the order of 20 min using a AX/X of 8%. With this diffractom1043
1044
Biochemistry: Bram et al.
Proc. Nat. Acad. Sci. USA 72
(1975)
50 20 k
0
VVXIV IV V V V V V
0
x
xx
Vv r- 25 % D20 V
x
0 0
'-o c
*
o 0
0
Oc
5 _
0)
a
xx
A
0
X
V V
Ux
xXx
AA0 0 00
V
xx
0
AA
5
0
Vx
~~0
AA
Chromati n
0
x
A
00
A
0
.01
.
.01
VV
Vxxx V
x
0 35% D20 AD
J
0o00 0
.00
&A
' 10 0%
0
o 0
cn
Ao o O o o A
"j, 0HHi
0
-J
Xx 4 Coil
0
.
.02
.03 h
(A
FIG. 1. Very small angle neutron scattering patterns from solutions of total calf thymus chromatin at a concentration of 5 mg/ml and from chromatin (3 mg/ml) depleted of H1 (13). The solvent was H20/1.5 mM NaCl/0.15 mM sodium citrate at pH 7. (The total counter ion concentration is dependent on the amount of chromatin present, and this may be on the order of 20 mM for some of the higher chromatin concentrations.) A maximum at Bragg spacings between 400 and 480 is prominent in both profiles. The Bragg spacing equals 2wr/h.
the scattering profiles, yet no reflections corresponding to larger spacings have been detected. (Fig. 1 shows a profile into 1200 .) The scattering profiles seem to be relatively free of concentration effects, at least up to 25 mg/ml. We previously reported (3) that H1 (very lysine-rich histone)-depleted chromatin shows very low angle spacings at Like chromatin, Hi-depleted material shows about 200 very small angle reflections near 400 A and at about 140 A. These Hi-depleted chromatin rings probably have an origin similar to those of total chromatin. The scattering profiles are appreciably smeared by the AX/X of 8%, which is usually used with the NILS diffractometer. To improve the peak resolution, we carried out some experiments with a double crystal monochromator diffractometer at Saclay, France, having, a ,A/X of 1%. The very low angle reflections show up more distinctly in these patterns. A.
Model Calculations. The scattering profiles expected from various models were calculated with the Debye equation and a computer program (2). Coils of finite thickness were computer-simulated by placing spheres of various radii at points on a discontinuous regular helix. We first determined the following: pitch: about 500 A from the 400 spacing; radius: 120 from the very small angle radius of gyration (2); fiber thickness: 100 from the second cross-section radius of gyration and electron microscopy (2). The first model curve agreed well with experimental values, except that the reflection at about 140 was absent and the scattering fell off too slowly. Our second try with a radius of 135 A gave us the model curve of Fig. 2. All the features of the experimental curve in H20 agree very well with the model. Subsequent models with 10-20% larger or smaller radii or pitch generally gave poorer fits to the data. At angles larger than those shown in Fig. 2 there are other
.02
.03
.04
.05
h
(4-
)
FIG. 2. The scattering profiles from three chromatin samples compared to a model calculation described in the text. The chromatin concentration was 15 mg/ml. A, 65% D20; 0, 35% D20; v, 25% D20. All data points were corrected for electronic noise in the detector but not for wavelength or instrumental smearing. The model curve (X) was calculated for a coil with a pitch of 500 A, fiber center radius of 135 A, and a translation per helix point in the direction of the axis of 60 A; a sphere of 100 A diameter was placed at each of 25 helix points. Prominent maxima or inflections are observed at low D20 contents and for the model at about 400, 200, and 140 A.
peaks, notably one at about 55 A, which arises from the secondary maxima of a sphere With a 50 X radius. Thus, the 55 A chromatin reflection could well be due to the geometrical shape of the cross section. DISCUSSION The existence of these very large reflections at a Bragg spacing equivalent to about 400 A1 shows that a structural ordering of this extent exists in chromatin. Given that the conformation of 100 A diameter fiber has been referred to as tertiary structure, we attribute these very low angle spacings to what we will call the quaternary structure. The observed disappearance of the 400 A ring at high D20 content may have one of two explanations: (i) Since this reflection is absent when the DNA scattering is weak (65% D20), it is due mostly to a DNA higher order structure that is not the same as that of protein. (ii) The ring is due to both protein and DNA, but the higher order structure of chromatin is modified by D20. The second explanation is supported by our observation that the cross-section radius of gyration at about 500 A spacings, which would correspond to the quaternary or higher order structure, is larger by 30-50% in pure D20. Furthermore, the 400 A reflection is absent in pure D20. Although the DNA scattering is 30% smaller than that of protein in D20, one would still expect to see a weakened maximum. In any case, this question remains unresolved, and it should be kept in mind that this higher order structure that we discuss may pertain only to DNA. Reflections of 400 A spacing cannot be due to a regular separation of globular units or beads, which appear to be separated by about 100 A in the electron microscope (5, 6). The 400 A reflection is also inconsistent with the interstrand packing of fibers, whose electron microscope and x-ray scattering diameters are only 100 A (2, 5, 6). We are thus left with the conclusion that the very low angle reflections result from a long range ordering of the 100 A unit fibril.
Quaternary Structure of Chromatin
(1975)
Proc. Nat. A cad. Sci. USA 72
"'500
A
1045
plainable by the inclusion of quaternary super coiling in the second measurement. The total compaction expected from the 500 A pitch coiled-coil model is about seven, and this agrees with that. determined from the intercept of the first linear region of the x-ray scattering (2); The quaternary folding may result from torsional strains introduced by the tertiary coiling of DNA. If this is the case, the quaternary coiling will have the opposite handedness to the tertiary coiling of the DNA. A functional role for the very tight folding of DNA has previously been suggested (11). Since such a compacted DNA is sterically inaccessible to recognition proteins and polymerases, it would consequently be inative. On the other hand, unwound stretches of up to several thousand angstroms in length are seen in the electron micrographs of chromatin. These regions, which comprise a few percent of the total mass, would be readily accessible to recognizer molecules and, therefore, marked for recognition (11).
a1
/
/
4; FIG. 3. A representation of the coiled-coil model used for the computer calculation of the scattering profile given in Fig. 2. The 100 A spheres used in the calculation would contain both protein and DNA. We do not know the exact arrangement of protein and DNA inside the spheres, but the DNA must be contracted about 4-fold. The eight 100 A spheres per turn represent the most compact structure consistent with the radius and pitch, but as few as three units might agree if a bead on a string model is chosen (6, 14).
Based upon our neutron and x-ray scattering results, we propose a super coiling, which may be flattened, of the 100 A unit fiber. The radius of this coil, deduced from the very small angle scattering cross-section radius of gyration and model calculations, is about 130 A and the pitch determined from the largest spacing is about 500 A. Our calculations of scattering curves for this model imply that it is consistent with all of the features of small angle scattering profile, including the presence of the two linear cross-section regions and their slopes and intercepts. (A sketch of the structure used for these calculations is shown in Fig. 3.). With this super coil in mind, it is worthwhile to mention that a value for the contraction of the DNA tertiary structure by total histone minus H1, of about 3- to 4-fold has been recently determined by electron microscopy. This value also conforms well with that found by x-ray scattering for the 100 A fibril (2). Griffith, on the other hand, finds a total contraction of DNA in a simian virus 40 DNA-histone complex of 7 to 8 (10). The 2-fold difference in these determinations is ex-
Electron micrograph views approximating the coiled-coil model can be selected from photographs of critical point-dried chromatin (for example, plates I and II of ref. 2). However, until wet chromatin can be viewed, one must treat the electron micrographs with caution. Dupraw has proposed that the 250 A diameter fibril found in whole chromosomes is made up of a single super coil with a pitch of 250-300 A (12). But the work presented here is on chromatin not containing calcium, and the same methods of electron microscopy as used by Dupraw show a fibril diameter of 100 A (2). Also, the DNA contraction ratio for the 250 A fiber is much larger than that expected from our model. We propose that a further winding of the tertiary or quaternary structure, perhaps by calcium ions, may wrap the coiled 100 A unit fibril into a double chromatin helix to give the 250 A diameter fiber. 0
We thank the staff of the Institut Laue-Langevin for their assistance and Drs. W. W. Beeman and H. Ris for their critical comments on the manuscript. We are pleased to acknowledge financial assistance from CNRS, DGRST, and NATO. 1. Pardon, J. F., Wilkins, M. H. F. & Richards, B. M. (1967) Nature 215, 508-509. 2. Brain, S. & Ris, H. (1971) J. Mol. Biol. 55, 325-336. 3. Bras, S., Butler-Browne, G., Bradbury, E. M., Baldwin, J. P., Reiss, C. & Ibel, K. (1974) Biochimie 56, 987-994. 4. Sahasrabuddhe, C. G. & Van Holde, K. E. (1973) J. Biol. Chem. 249, 152-156. 5. Woodcock, &' L. F. (1973) J. Cell Biol. 54, 368. 6. Olins, D. E. & Olins, A. L. (1974) Science 183, 330-332. 7. Ris, H. & Chandler, B. (1963) Cold Spring Harbor Symp. Quant. Biol. 28, 1-8. 8. Panyim, S., Bilik, N. & Chalkley, R. (1971) J. Biol. Chem. 246, 4206-4215. 9. Schmatz, W., Springer, T., Schelten, J. & Ibel, K. (1974) J. Appl. Crystallogr. 7, 96-116. 10. Griffith, J. (1974) Eighth Int. Cong. Electron Microsc. 2, 254-255. 11. Bram, S. (1972) Biochimie 54, 1005-1011. 12. Dupraw, E. J. (1966) Nature 209, 577-581. 13. Bolund, L. A. & Johns, E. W. (1973) Eur. J. Biochem. 35, 533-546. 14. Kornberg, R. D. (1974) Science 184, 868-871.
Quaternary Structure of Chromatin (neutron scattering/new coiled-coil model)
STANLEY BRAM*, GILL BUTLER-BROWNE*, PIERRE BAUDY*, AND KONRAD IBELt * Departement de Biologie Molculaire, Institut Pasteur, Paris 75015, France; and t Institut Laue-Langevin, Grenoble, France Communicated by Franiss Jacob, November 29, 1974 Neutron scattering experiments on chroABSTRACT matin solutions have been carried out at Bragg spacings
several times larger thani in any previous x-ray work. A new series of very low angle reflections at about 400, 200, and 140 A- were observed. These reflections demonstrate the existence of a higher order structure of the 100 A unit fibril. A coiled-coil model with a major pitch of 500 A and a radius of center to fibril center of 130 A is proposed to explain the results.
eter and monochromatization, instrumental smearing of the scattering patterns are considerably less than with the x-ray diffractometers used for previous small angle x-ray scattering (2). Even so, the' maxima in the scattering profile are somewhat smeared out. For the discussion that follows,, it is worth discussing some of the features of neutron scattering. The neutron scattering amplitude of each atom or isotope is dependent only on the nuclear structure and not oh the atomic number, as is the case for x-rays. Thus, the scattering from a molecule in solution at zero angle is a function of the sum of the scattering lengths of its atoms or of the atoms in its compl8hent molecular parts, minus that in the same volume of solvent. H20 has a .scattering length density of -0.6 X 1010/cm3, while pure D20 has a value that is 10 times larger and positive. Because the density and atomic abundances of DNA and protein are dissimilar, the neutron scattering from each will vary rather differently with the solvent scattering length density, which is only a function of the H20/D20 content of the solvent. At a D20 content of 38%o by volume, the scattering from histone will be almost zero, whereas DNA will scatter very weakly in the neighborhood of 65% D20. For the results of this paper, it suffices to point out that if a reflection or a scattering ring is prominent in D20 concentrations less than about 45% and absent in the neighborhood of 65%, it may be attributed to scattering from DNA. It must be stressed that this contrast matching treatment is only valid at low concentrations and small angles, and if the structure is invariant in D20/H20
Until recently, structural studies of chromatin have been oriented towards determining the secondary or tertiary coiling of chromosomal DNA (1-3). Consequently, mainly those features smaller than about 110 A have been investigated. Evidence that suggests that the tertiary structurebof DNA and protein consists of globular units approximately 80-100 ! in diameter arranged like a string of beads on a thread has appeared. This evidence has come mainly from nuclease digestion studies and electron microscopy of chromatin and nuclei (4-6). However,, except for the work df Ris, which implies that the 100 A unit thread doubles up in the presence of calcium ions (7), the highei-order folding of the unit thread has been somewhat neglected. Previous x-ray scattering studies on chromatin, although directed at tertiary structure of the order of 100 A, often showed weak inflections in the scattering profile at spacings corresponding to about 400 and 200 A (see, for example, Fig. 2 of ref. 2),. It now seems that the weakness of these low angle x-ray reflections was due to slit smearing from the x-ray diffractometer. mixtures. This paper presbnts the results obtained from neutron investigations of the higher-order arrangement of the 100 A RESULTS unit thread. Very low angle reflections are indeed found to be Neutron Scattering. Solutions and gels of chromatin were quite prominent in the neutron scattering at spacings of up studied over a concentration range of 3-25 mg/ml, between to 460 A. It is shown that a quaternary chromatin structure Bragg spacings of 1500-80 A. Between 8 and 10 equivalent does exist, and it is suggested that a coiling of the 110 A unit H20/D20 contents were examined on more than eight inthread with an average pitch of about 500 A and a radius of preparations. Figs. 1 and 2 show some of the results dependent 130 A is consistent with these results. of typical scattering curves at different H20/D20 contents. A METHODS distinct shoulder or peak is seen in the scattering profiles at a Bragg spacing equivalent to about 400 A when the D20 Calf thymus chromatin was isolated by the method of Panyim content is less than about 45%$T. On the other hand, this very et al. (8), which is based on the isolation of nuclei, their purilow angle reflection is absent at higher D20 contents. Maxima fication, and their subsequent gentle disruption. Chromatin at smaller spacings-about 220 and 140 A-are often seen in samples were concentrated by 'centrifugation, and all samples against dialyzed were controls with the exception of a few three changes of the chosen H20/D20 mixture, with a known t The spacing varies between 400 and 460 A, which is consistent amount of salt. with experimental error. It seems more prudent to refer to a Neutron scattering experiments were done with the NILS spacing "at about 400 A" than to give an exact figure which may diffractometer built by K. Ibel (9). Exposure times were on change as our precision improves. Furthermore, this spacing varies with the solution environment. the order of 20 min using a AX/X of 8%. With this diffractom1043
1044
Biochemistry: Bram et al.
Proc. Nat. Acad. Sci. USA 72
(1975)
50 20 k
0
VVXIV IV V V V V V
0
x
xx
Vv r- 25 % D20 V
x
0 0
'-o c
*
o 0
0
Oc
5 _
0)
a
xx
A
0
X
V V
Ux
xXx
AA0 0 00
V
xx
0
AA
5
0
Vx
~~0
AA
Chromati n
0
x
A
00
A
0
.01
.
.01
VV
Vxxx V
x
0 35% D20 AD
J
0o00 0
.00
&A
' 10 0%
0
o 0
cn
Ao o O o o A
"j, 0HHi
0
-J
Xx 4 Coil
0
.
.02
.03 h
(A
FIG. 1. Very small angle neutron scattering patterns from solutions of total calf thymus chromatin at a concentration of 5 mg/ml and from chromatin (3 mg/ml) depleted of H1 (13). The solvent was H20/1.5 mM NaCl/0.15 mM sodium citrate at pH 7. (The total counter ion concentration is dependent on the amount of chromatin present, and this may be on the order of 20 mM for some of the higher chromatin concentrations.) A maximum at Bragg spacings between 400 and 480 is prominent in both profiles. The Bragg spacing equals 2wr/h.
the scattering profiles, yet no reflections corresponding to larger spacings have been detected. (Fig. 1 shows a profile into 1200 .) The scattering profiles seem to be relatively free of concentration effects, at least up to 25 mg/ml. We previously reported (3) that H1 (very lysine-rich histone)-depleted chromatin shows very low angle spacings at Like chromatin, Hi-depleted material shows about 200 very small angle reflections near 400 A and at about 140 A. These Hi-depleted chromatin rings probably have an origin similar to those of total chromatin. The scattering profiles are appreciably smeared by the AX/X of 8%, which is usually used with the NILS diffractometer. To improve the peak resolution, we carried out some experiments with a double crystal monochromator diffractometer at Saclay, France, having, a ,A/X of 1%. The very low angle reflections show up more distinctly in these patterns. A.
Model Calculations. The scattering profiles expected from various models were calculated with the Debye equation and a computer program (2). Coils of finite thickness were computer-simulated by placing spheres of various radii at points on a discontinuous regular helix. We first determined the following: pitch: about 500 A from the 400 spacing; radius: 120 from the very small angle radius of gyration (2); fiber thickness: 100 from the second cross-section radius of gyration and electron microscopy (2). The first model curve agreed well with experimental values, except that the reflection at about 140 was absent and the scattering fell off too slowly. Our second try with a radius of 135 A gave us the model curve of Fig. 2. All the features of the experimental curve in H20 agree very well with the model. Subsequent models with 10-20% larger or smaller radii or pitch generally gave poorer fits to the data. At angles larger than those shown in Fig. 2 there are other
.02
.03
.04
.05
h
(4-
)
FIG. 2. The scattering profiles from three chromatin samples compared to a model calculation described in the text. The chromatin concentration was 15 mg/ml. A, 65% D20; 0, 35% D20; v, 25% D20. All data points were corrected for electronic noise in the detector but not for wavelength or instrumental smearing. The model curve (X) was calculated for a coil with a pitch of 500 A, fiber center radius of 135 A, and a translation per helix point in the direction of the axis of 60 A; a sphere of 100 A diameter was placed at each of 25 helix points. Prominent maxima or inflections are observed at low D20 contents and for the model at about 400, 200, and 140 A.
peaks, notably one at about 55 A, which arises from the secondary maxima of a sphere With a 50 X radius. Thus, the 55 A chromatin reflection could well be due to the geometrical shape of the cross section. DISCUSSION The existence of these very large reflections at a Bragg spacing equivalent to about 400 A1 shows that a structural ordering of this extent exists in chromatin. Given that the conformation of 100 A diameter fiber has been referred to as tertiary structure, we attribute these very low angle spacings to what we will call the quaternary structure. The observed disappearance of the 400 A ring at high D20 content may have one of two explanations: (i) Since this reflection is absent when the DNA scattering is weak (65% D20), it is due mostly to a DNA higher order structure that is not the same as that of protein. (ii) The ring is due to both protein and DNA, but the higher order structure of chromatin is modified by D20. The second explanation is supported by our observation that the cross-section radius of gyration at about 500 A spacings, which would correspond to the quaternary or higher order structure, is larger by 30-50% in pure D20. Furthermore, the 400 A reflection is absent in pure D20. Although the DNA scattering is 30% smaller than that of protein in D20, one would still expect to see a weakened maximum. In any case, this question remains unresolved, and it should be kept in mind that this higher order structure that we discuss may pertain only to DNA. Reflections of 400 A spacing cannot be due to a regular separation of globular units or beads, which appear to be separated by about 100 A in the electron microscope (5, 6). The 400 A reflection is also inconsistent with the interstrand packing of fibers, whose electron microscope and x-ray scattering diameters are only 100 A (2, 5, 6). We are thus left with the conclusion that the very low angle reflections result from a long range ordering of the 100 A unit fibril.
Quaternary Structure of Chromatin
(1975)
Proc. Nat. A cad. Sci. USA 72
"'500
A
1045
plainable by the inclusion of quaternary super coiling in the second measurement. The total compaction expected from the 500 A pitch coiled-coil model is about seven, and this agrees with that. determined from the intercept of the first linear region of the x-ray scattering (2); The quaternary folding may result from torsional strains introduced by the tertiary coiling of DNA. If this is the case, the quaternary coiling will have the opposite handedness to the tertiary coiling of the DNA. A functional role for the very tight folding of DNA has previously been suggested (11). Since such a compacted DNA is sterically inaccessible to recognition proteins and polymerases, it would consequently be inative. On the other hand, unwound stretches of up to several thousand angstroms in length are seen in the electron micrographs of chromatin. These regions, which comprise a few percent of the total mass, would be readily accessible to recognizer molecules and, therefore, marked for recognition (11).
a1
/
/
4; FIG. 3. A representation of the coiled-coil model used for the computer calculation of the scattering profile given in Fig. 2. The 100 A spheres used in the calculation would contain both protein and DNA. We do not know the exact arrangement of protein and DNA inside the spheres, but the DNA must be contracted about 4-fold. The eight 100 A spheres per turn represent the most compact structure consistent with the radius and pitch, but as few as three units might agree if a bead on a string model is chosen (6, 14).
Based upon our neutron and x-ray scattering results, we propose a super coiling, which may be flattened, of the 100 A unit fiber. The radius of this coil, deduced from the very small angle scattering cross-section radius of gyration and model calculations, is about 130 A and the pitch determined from the largest spacing is about 500 A. Our calculations of scattering curves for this model imply that it is consistent with all of the features of small angle scattering profile, including the presence of the two linear cross-section regions and their slopes and intercepts. (A sketch of the structure used for these calculations is shown in Fig. 3.). With this super coil in mind, it is worthwhile to mention that a value for the contraction of the DNA tertiary structure by total histone minus H1, of about 3- to 4-fold has been recently determined by electron microscopy. This value also conforms well with that found by x-ray scattering for the 100 A fibril (2). Griffith, on the other hand, finds a total contraction of DNA in a simian virus 40 DNA-histone complex of 7 to 8 (10). The 2-fold difference in these determinations is ex-
Electron micrograph views approximating the coiled-coil model can be selected from photographs of critical point-dried chromatin (for example, plates I and II of ref. 2). However, until wet chromatin can be viewed, one must treat the electron micrographs with caution. Dupraw has proposed that the 250 A diameter fibril found in whole chromosomes is made up of a single super coil with a pitch of 250-300 A (12). But the work presented here is on chromatin not containing calcium, and the same methods of electron microscopy as used by Dupraw show a fibril diameter of 100 A (2). Also, the DNA contraction ratio for the 250 A fiber is much larger than that expected from our model. We propose that a further winding of the tertiary or quaternary structure, perhaps by calcium ions, may wrap the coiled 100 A unit fibril into a double chromatin helix to give the 250 A diameter fiber. 0
We thank the staff of the Institut Laue-Langevin for their assistance and Drs. W. W. Beeman and H. Ris for their critical comments on the manuscript. We are pleased to acknowledge financial assistance from CNRS, DGRST, and NATO. 1. Pardon, J. F., Wilkins, M. H. F. & Richards, B. M. (1967) Nature 215, 508-509. 2. Brain, S. & Ris, H. (1971) J. Mol. Biol. 55, 325-336. 3. Bras, S., Butler-Browne, G., Bradbury, E. M., Baldwin, J. P., Reiss, C. & Ibel, K. (1974) Biochimie 56, 987-994. 4. Sahasrabuddhe, C. G. & Van Holde, K. E. (1973) J. Biol. Chem. 249, 152-156. 5. Woodcock, &' L. F. (1973) J. Cell Biol. 54, 368. 6. Olins, D. E. & Olins, A. L. (1974) Science 183, 330-332. 7. Ris, H. & Chandler, B. (1963) Cold Spring Harbor Symp. Quant. Biol. 28, 1-8. 8. Panyim, S., Bilik, N. & Chalkley, R. (1971) J. Biol. Chem. 246, 4206-4215. 9. Schmatz, W., Springer, T., Schelten, J. & Ibel, K. (1974) J. Appl. Crystallogr. 7, 96-116. 10. Griffith, J. (1974) Eighth Int. Cong. Electron Microsc. 2, 254-255. 11. Bram, S. (1972) Biochimie 54, 1005-1011. 12. Dupraw, E. J. (1966) Nature 209, 577-581. 13. Bolund, L. A. & Johns, E. W. (1973) Eur. J. Biochem. 35, 533-546. 14. Kornberg, R. D. (1974) Science 184, 868-871.
