Proc. Nati. Acad. Scf. USA Vol. 75, No. 8, pp. 3649-3653, August 1978 Biochemistry

Histone packing in the nucleosome core particle of chromatin (DNA superhelix/pseudosymmetry/broken symmetry/eukaryote)

C. W. CARTER, JR. Departments of Biochemistry and Anatomy, The University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27514

Communicated by Ernest L. Eliel, May 15,1978

ABSTRACT The chromatin core particle DNA conformation deduced in broad outline by Finch etaL [Finch,J. T., Lutter, L. C., Rhodes, D., Brown, R. S., Rushton, B., Levitt, M. & Klug, A. (1977) Nature 269, 29-36] can be described in detail using other available experimental results. Histone binding sites compatible with the pattern of pancreatic DNase I digestion [Simpson, R. T. & Whitlock, J. P., Jr. (1976) Celi 9, 347-353; Noll, M. (1977) J. MoL. Biol. 116, 49-71; Lutter, L C. (1977)J. Mo). Biol. 117, 53-69] lend to core particle DNA pseudosymmetry characteristic of molecular point group D3. DNA symmetry and pseudosymmetry, in turn, imply equivalence and quasi-equivalence properties of the histone packing arrangement that support the following deductions: (i) One and only one asp2 histone tetramer presumably (H3)2(H4)2, can serve as a stable subassembly witAin the histone octamer. (ii) There is a unique, strand-specific way to assign DNA binding domains to the arginine-rich histones (H3 and H4) (iii) Histones H3 and H4 alone should suffice to impose a supercoiled structure on DNA, as is observed experimentally, because only the tetramer can mimic a screw dislocation and thereby complement the screw symmetry of the DNA supercoil. (iv) The two slightly lysine-rich histones H2A and H2B are probably responsible, each in a different way, for dividing the eukaryotic chromatin fiber into discrete subunits. (v) The proposed arrangement of four distinct proteins appears to be a minimum formal requirement for making nucleosomes; that is, for introducing regularly spaced supercoiled DNA folds without also allowing formation of an indefinitely long (and genetically inert) DNA superhelix. Analyses of chromatin nucleosome core particles by digestion with pancreatic DNase 1 (1-4) and by x-ray crystallography (5) have recently revealed several important aspects of the core particle DNA conformation. For the most part, these features have been incorporated into the proposal of Klug and coworkers (5) that 140 DNA base pairs in a core particle are smoothly deformed into 1.75 turns of a left-handed superhelix approximately 90 A in diameter with a pitch close to 28 A and a screw close to 80 base pairs. There are good theoretical (6, 7) and experimental (8) reasons for believing that DNA can curve smoothly to a 45-A radius without introducing discontinuities. Such a DNA conformation affords a simple structural basis for the observation that DNase I digestion fragments differ in length by multiples of a number close to 10 bases. KMug and coworkers endorse the proposal (1-4) that this pattern arises from the DNA screw repeat, predicted by Levitt (7) to be exactly 10 bases per turn in a uniform DNA supercoil fragment as proposed for the core particle. The radius, pitch, and screw of the DNA supercoil are specified within narrow limits by neutron diffraction measurements (9) and by packing constraints in core particle crystals (5). They are therefore probably correct, whatever modifications become necessary as details of core particle structure emerge (see below). Moreover, they afford a rationale for the relative frequencies for DNase I cutting at different susceptible The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact. 3649

sites (5). Digestion is less efficient near bases 0, 30, 60, 80, 110, and 140, counting from the 5' ends of both strands (1-4), presumably because histones bind to and exclude enzyme from these sites. A superhelical screw repeat close to 80 base pairs effects the close approach of these six presumptive histone binding sites by aligning them in pairs along the superhelix axis (Fig. 1; see also figure 8 of ref. 5). Twofold rotation symmetry is evident in the linear pattern of relative cutting frequencies along the 140 base pair DNA fragment (1-4) and can be identified with a principal dyad axis in the three-dimensional core particle structure (5). This nucleosomal dyad passes through DNA double helix at the midpoint of the 140-base-pair fragment, denoted in Fig. 1 by (70, 70'), and between successive turns at points denoted by (30', 110) and (110', 30). It applies strictly to the deoxyribose-phosphate backbone of the entire chain, and effects interchange of the black and white strands in the figure. As proposed by Kornberg (10), there are two each of four histones-H2A, H2B, H3, and H4-in a nucleosome core particle. How are these eight proteins arranged in space? Oligomeric protein packing arrangements generally tend to approximate either cubic or hexagonal close packing (11). Any core histone configuration with point group symmetry that also forms paired DNA binding sites for successive superhelical turns around its perimeter will exhibit dihedral pseudosymmetry approaching point group D4 for cubic packing and point group D3 for hexagonal packing. This concept of approximate or pseudosymmetry is quite useful in distinguishing between possible octamer models. A model having pseudo-D4 symmetry has been described (12). My purpose here is to develop a model for histone octamer geometry from the basic assumption that it should complement symmetry and pseudosymmetry elements in core particle DNA (Fig. 1). The DNA supercoil fragment has pseudo-D3 symmetry and so, therefore, has the resulting core protein model (Fig. 2). Consequently, only three of the four histones contribute to those six DNA binding sites revealed by the variable pattern of susceptibility to DNase I. This restriction, in turn, implies that an a2#f2 tetramer, presumably (H3)2(H4)2, should be a natural subset of the octamer itself. Intermolecular contacts within this tetramer and between it and the remaining histones suggest why the arginine-rich histones tend to occur as tetramers (14) and why they seem to induce core particle DNA conformation so effectively (15-18). Moreover, there are strong symmetry arguments that the slightly lysine-rich histones H2A and H2B have an essential role in dividing the eukaryotic chromatin fiber into nucleosomes. Histone binding imposes pseudo-D3 symmetry on core particle DNA Screw and local twofold symmetries in double-helical DNA permit similar or identical interactions between successive superhelical turns to be reproduced at intervals along the chain (5). Two additional experimental observations suggest that

3650

Biochemistry:

Carter

FIG. 1. Pseudosymmetry in core particle DNA. Core particle DNA is redrawn from figure 8a of ref. 5, and is represented approximately to scale. Cross-sectional planes intersect the superhelical coil at regions protected from digestion by pancreatic DNase I, passing between base pairs involving the numbered bases (see text). The DNA major groove (semicircle) faces inward at each site, the minor groove (wedge) faces into solution. The descending (0'-140'; 5'-3') strand enters each section at the top and is black. The ascending (0-140; 5' -3') strand is white and enters each section at the bottom.

histone binding to the six DNAse I-resistant regions may exploit such recurrent interactions: (i) Histones protect DNA bases in chromatin from chemical modification, and the major groove is protected exclusively (19). This groove must therefore face the histones at each protected site. (if) DNase I cuts on opposite strands in chromatin are offset by two bases (1-4). These cuts probably occur in the core particle at points where local DNA dyads project radially from the superhelix axis. Only such points can expose staggered sites with equal susceptibility on both strands. A stagger of two bases therefore implies that all DNA dyad axes retained by superhelical DNA in the core particle pass between successive base pairs and not through single base pairs. Studies of Felsenfeld and coworkers show that cuts by several additional nucleases also conform to this type of dyad axis (4). DNA conformation at the six DNase I-resistant regions may now be specified rather precisely. The nucleosomal dyad coincides with a DNA dyad passing between base pairs at (70, 70'). Equivalent dyads must also project radially from the superhelix axis at intervals determined by the DNA screw repeat. The six protected regions occur at such intervals and therefore present the same local DNA conformation to the protecting histones: the major groove faces the histones about a dyad passing between successive base pairs. This condition fixes DNA chain polarities and approximate position numbers of each base as indicated in Fig. 1. The DNA dyad at each protected region is paired with an equivalent dyad along the superhelix axis. This alignment gives rise, in turn, to local dyad axes between successive superhelical turns. Equivalent protected regions near base sets (0, 140') and 0', 140) are therefore related by local twofold symmetry to those near (80, 60') and (80', 60) in a similar manner to the exact, nucleosomal twofold symmetry relating sets (3, 110) and (110', 30) (Fig. 1). Protection of DNA by histones in three regions

Proc. Natl. Acad. Sci. USA 75 (1978)

FIG. 2. An (a#'Y5)2 octamer with a strict twofold axis and pseudo-D3 symmetry. Relative sizes of each subunit represent identification with specific histones (a = H3, is = H4, y = H2A, 6 = H2B) outlined in the text. Solid intersubunit contacts represent strong tendencies to cross-complex, as revealed by the method of continuous variations (13).

separated by approximately 1200 (1350, 1350, 900) lends approximate threefold rotation symmetry to the superhelix axis itself. A threefold axis in combination with three dyad axes perpendicular to it gives point group D3. The nucleosome core particle must therefore have strong elements of pseudo-D3 symmetry in addition to its true twofold symmetry. Both kinds of symmetry should arise from the histone octamer geometry. An octamer that meets this expectation can be constructed by apposing two heterotypic (afry6) tetramers base-to-base at three points and about a twofold symmetry axis (Fig. 2). The distinguishing characteristic of this packing diagram is that the dyad axis passes between two a# heterodimers and through one homodimer, 72. Packing the eight histones in this manner achieves the desired complementarity to core particle DNA (Fig. 1) because the latter can derive its symmetry and pseudosymmetry properties from equivalent and quasi-equivalent (20) contacts involving the histores. Specifically, equivalent histne-DNA interactions should protect sites (110', 30) and (30', 110) because they are related by true twofold symmetry. These sites are covered in this model by the homodimer, 72. Interactions protecting sites (0, 140') and (140, 0') can be at most quasi-equivalent to those at sites (60, 8(Y) and (80, 60'), because the former and latter DNA regions are related only by local symmetry. Heterodimers a,/ and #a can form sets of at most two equivalent contacts. Thus only quasi-equivalent interactions are possible with sites (0, 140') and (80,60'), because thee must be reproduced at their symmetry-related counterparts (140, 0') and (80', 60). Histone packing according to Fig. 2 therefore satisfies the requirement for both true twofold and pseudo-threefold symmetry. Histone octamers complementary to core particle DNA may contain one and only one a&82 tetramer The construction in Fig. 2 reveals one and only one subgroup of points, in this case a2#2, for which isologous a-a and f-f interactions (broken lines) can reinforce heterologous a-,8 in-

Biochemistry:

Proc. Natl. Acad. Sci. USA 75 (1978)

Carter

Clearly, two of the four histones can in principle form a tetrameric complex using these complementatylf associations. It follows that the histone octamer may contain one and only one a2,B2 tetramer as a potentially stable subassembly. By analogy with Fig. 2, a pseudo-D4 octamer would appose two cyclic, afl5y tetramers face-to-face at four points about a dyad axis passing between two pairs of heterodimers, both of which could form stable tetramers. Identification of the a and (8 designations with the arginine-rich histones H3 and H4 is straightforward. It is well known that a complex of these two histones has been isolated by gel filtration chromatography from unfractionated histones and characterized as a tetramer by sedimentation equilibrium and chemical crosslinking (14). Interhistone association parameters of all four histones have been measured by the method of continuous variations (13) and show a strong tendency for histones H3 and H4 to cross-complex. Which slightly lysine-rich histone, H2A or H2B, covers DNA at points to complete the triad of radial spokes around which the core particle DNA winds? Camerini-Otero et al. (15) have found that H2A generated a new protected site when added to a reconstitution mixture containing H3, H4, and DNA, whereas H2B did not. We can infer from this result that protection near bases (30', 110) and (110', 30) arises primarily from interaction of DNA with an (H2A)2 dimer. Two alternative possibilities-that 72 represents either the homodimer (H2B)2 or the heterodimer (H2A-H2B)-seem to be less likely. Neither has as yet been conclusively ruled out experimentally. However, the three possibilities would result in different slightly lysinerich histone dimers in chemical crosslinking experiments, and might therefore be distinguished from one another by such an teractions (11).

experiment. As discussed below, interactions involving the fourth histone, 6, are unique. The construction in Fig. 2 suggests that it may reinforce interactions between the remaining three histones and bind both core and linker DNA.

DNA binding domains of arginine-rich histone molecules are probably strand specific The network illustrated in Fig. 2 cannot adequately represent all of the DNA binding domains of individual histone molecules. The (H3)2(H4)2 tetramer, for example, may cover as many as eight distinct portions of DNA at or near bases 0, 60, 80, and 140 on each strand, in which case each histone would attach twice to DNA. DNA symmetry and pseudosymmetry suggest probable binding domains for histones H3 and H4 in at least three ways: (i) No individual histone can bind to bases n and n' on opposite DNA strands without violating the principal dyad of the

supercoil. (ii) Nor is it likely that an individual histone molecule would bind both DNA strands across any dyad (i.e., to both 60 and 80' across the double helix, or to both 80' and 140 across a local dyad of the supercoil). There is no evidence for internal amino acid sequence homology within either H3 or H4 (21), so their conformations probably lack sufficient intramolecular twofold symmetry to bind thus to DNA. Hence binding sites for DNA double helixes are probably formed by histone dimers. (iii) If the four protected DNA regions are to remain quasiequivalent after they have been crosslinked pairwise by histone dimers, all four binding sites should use heterodimers H3-H4. Use of H4-H4 and H3-H3 contacts would, in accordance with the principal dyad, form inequivalent crosslinks (60, 80') to (60', 80) and (0, 140') to (0', 140). The former crosslink, spanning

3651

only twenty base pairs of DNA, would seem to be superfluous.

Most assignments of arginine-rich histone binding domains either violate the symmetry properties of the DNA coil or require ad hoc assumptions regarding intramolecular symmetry. Of those remaining, only one is consistent with the fact that H3 and H4 have quite different molecular weights. Note from the position number assignments in Fig. 1 that the descending strand (5' -- 3') is on top at each protected site, and that bases 0 and 0' are considerably further apart (about 90 A) than are bases 140' and 140 (about 65 A). The pair of larger histones (H3; molecular weight = 15,300) can cover a larger distance between equivalent points and so bind to the former set 0 and 0'. The H4 (molecular weight = 11,300) pair is smaller precisely in proportion to the shorter distance and thus may attach equivalently to the latter set 140 and 140'. Conditions ii and iii above imply that secondary binding sites probably occur at or near bases 60 and 60' for H3; 80 and 80' for H4. Parallel extension of the histones across the plane of the nucleosome in this manner-H3 from 0 to 60 and 0' to 60', H4 from 140 to 80 and 140' to 80' -assigns both binding sites on a given histone to the same DNA strand. These DNA binding domains are in quite good agreement with recent experimental results obtained by crosslinking histone proteins in core particles to terminally labeled DNA (22, 23). It is widely suspected (12, 24, 25) that a considerable fraction of the histone mass extends from a globular core in order to associate with DNA. Calculations not reported here indicate that for this to be true the remaining, globular, portion of the tetramer must be quite asymmetric, forming a prolate ellipsoid of eccentricity about 3, in order to span the nucleosome at these locations. Prediction methods and physicochemical measurements suggest that there are about 145 a-helical residues in two molecules each of H3 and H4 (21). These could form a bundle approximately 60 A long and 20 A in diameter, containing four roughly parallel, composite a-helixes. Arginine-rich histone tetramer is uniquely suited to supercoil DNA Two properties of the arginine-rich histone tetramer help to explain why these proteins can form supercoiled DNA structures similar to nucleosomes without participation by lysine-rich histones (15-18). First, as already noted above (see also ref. 5), it is the largest potentially stable subassembly of the octamer, and the vertical contacts (shaded in Fig. 2) can have the appropriate pseudosymmetry to bind DNA at four quasi-equivalent points in the nucleosome. Second, and perhaps more fundamentally, its a2fl composition permits, within the tetramer and between tetramer and 72 dimer, precisely those nonequivalent lateral contacts formally required by a screw dislocation in the succession of vertical latches suggested by planes in Fig. 1 and by cross-hatched columns in Fig. 2. In contrast, the dimer 72 alone has no lateral contacts and cannot discriminate effectively between two successive superhelical turns and two stacked rings of DNA. If the histone octamer has point group symmetry, then it must mimic a screw dislocation in order to complement the screw symmetry of the DNA superhelix. To clarify this point, compare the effects of screw-dislocating an ideal 76 hexainer having true D3 symmetry with lateral contacts in the a218272 hexamer of Fig. 2. Two diagonal y-y contacts change substantially upon dislocation. One becomes shorter and one longer, according to the handedness of the screw. Formally, the f-ft and a-a contacts in Fig. 2 are homologous to new short and new long contacts. Twofold symmetry from the parent D3 group,

3652

Biochemistry:

Proc. Nati. Acad. Sci. USA 75 (1978)

Carter

required by double-helical DNA, is retained at the dislocation site (70, 70') but nowhere else. Thus, strained contacts between the two dislocated dimers and the third create two different interfaces, formally homologous to f3-y and y-a in Fig. 2. Resolving this conflict between point group symmetry in the core protein and screw symmetry in the DNA supercoil is analogous to assembling a spherical virus capsid from more than the formal limit of 60 identical protein subunits allowed by icosahedral symmetry (20). Quasi-equivalent protein-protein contacts appear in both cases. It is evidently more difficult to design a nucleosome using identical protein subunits. Thus, three different subunits are necessary to prescribe a fixed, superhelical path for a discrete length of double-helical DNA. Convincing evidence for these conclusions can be adduced from the x-ray diffraction studies of Kornberg on reconstituted nucleohistone preparations (14). Complexes of DNA and (H3)2(H4)2 tetramers show only a shoulder at 37 A. This feature is also characteristic of chromatin and has been ascribed to the second maximum of a Bessel function corresponding to the diameter of a DNA supercoil (5). Chromatin itself and reconstituted complexes containing all four nucleosomal histones show another maximum at 55 A ascribed to two different interparticle spacings that require a well-defined disk (5). Tetramers alone can therefore probably induce DNA supercoils with a characteristic diameter, but that flap about their common hinge [i.e., the segment containing (70, 70')]. Addition of slightly lysine-rich histones would then clamp successive turns together opposite this hinge at (30, 110') and (110, 30') to form a disk. How and why do nucleosomes arise? A large, diverse body of evidence has accumulated that the nucleosome is a well-defined and discrete subunit in eukaryotic chromatin (ref. 26 contains an excellent discussion of this point). The simple and regular DNA fold in the core particle is somewhat surprising in light of this evidence: there is no obvious hint of the fundamental discontinuity that divides chromatin fibers into nucleosomes. On the contrary, the core particle DNA conformation itself seems poised to nucleate an indefinitely long superhelix. The flexible, beaded structure of chromatin fibers contrasts sharply with, for example, the rigid, helical tobacco mosaic virus particle. Favorable protein-protein and protein-RNA interactions act cooperatively in virus assembly to promote elongation so that a regular superhelix is the thermodynamically stable product (27). What prevents cooperativity of histone-histone and histone-DNA interactions between adjacent core particles so that nucleosomes and not a superhelix arise when DNA complexes with histone octamers? Reconstitution experiments using purified intranucleosomal histones yield a beaded fiber (18), so no additional proteins, such as histone H1, are necessary to prevent superhelix formation. The discrete nature of adjacent nucleosomes seems therefore to arise from an inherent property of the octamer itself. This property may, of course, involve factors such as steric hindrance between adjacent octamers, which can be elucidated only with detailed stereochemical knowledge of histone-DNA interactions. Nevertheless, it is worthwhile to point out that D3 pseudosymmetry of the proposed histone packing arrangement introduces structural discontinuities in chromatin more decisively than models with D4 pseudosymmetry. Hypothetical chromatin fibers constructed from these two limiting structures are compared schematically in Fig. 3 to illustrate this point. Cooperative octamer binding and superhelix formation may result if successive octamers can nest smoothly on top of one another such that the first attachment site of one lies within the

70

A B FIG. 3. Hypothetical chromatin fibers built from repeating pseudo-D4 (A) and pseudo-D3 (B) histone octamers. The first 140base-pair fragment of each fiber is black, showing the length of core particle DNA. Beveled vertical edges represent histone-DNA contacts. Pseudo-D4 octamers in A are intended to be identical; the alternating pattern of cross-hatching is only for clarity. Dyad axes of successive octamers in A occur at 90-base-pair intervals as indicated by their position numbers. Core particles in B represent schematic superposition of the DNA and octamer structures in Figs. 1 and 2. The linker DNA in B is drawn with broken lines to emphasize that its conformation and the relative orientation of core particles are hypothetical.

supercoil fragment defined by the previous core particle. Octamers of either symmetry type can follow homologous ones starting as close as position 90-i.e., within a 140-base-pair fragment-without occupying sites already taken by the previous one. Each octamer in Fig. 3A starts at base 90 of the DNA segment spanned by the previous one to emphasize the accidental complementarity of pseudo-D4 octamers with the 80base-pair superhelical screw. The DNA path around adjacent pseudo-D4 octamers (Fig. 3A) need not contain separate linkers or even distinct core particles. Such octamers would resemble a simple stack of building blocks. All four histones have three nearest neighbors and three next-nearest neighbors, so no pair

has a unique packing environment. Moreover, the frequency of their attachment sites, 20 base pairs apart, exactly matches the 80-base-pair superhelical screw of DNA, and divides it into four equal segments. DNA in such a superhelix would contain an almost continuous succession of 20-base-pair segments containing links to the previous octamer at base 10 and to the succeeding octamer at base 20. A supercoil so bound together by vertical crosslinks could prove difficult to uncoil and hence deleterious. There are several reasons why it is much more difficult to envision such an arrangement in the case of pseudo-D3 oc-

Biochemistry:

Proc. Nat!. Acad. Sci. USA 75 (1978)

Carter

tamers. Stacked pseudo-D3 core particles cannot provide a continuous superhelical path for DNA and must the&rfe create distinct linkers This is because points 6 (black in Fig. 3B) occupy unique positions: they can form only three histonehistone contacts, not six as can the others, and do not form intranucleosomal 6-6 contacts. Internucleosomal DNA segments must interact in a new way with the 62 dimer [(H2B)2 in this case], and these contacts separate adjacent core particles into

beads. Similar conclusions follow from the fact that 80 is not an integral multiple of 3, so one turn of DNA around a pseudo-D3 octamer must divide into unequal segments-S0, 30, and 20 base pairs. The five intervals in a complete core particle-O0, 30,20,30,30-compose a 140-base-pair segment that in no way matches the superhelical screw. Its two halves are related by rotation symmetry (about 70), not translation symmetry. Appropriate ionic conditions render the DNA at this symmetry axis susceptible to cutting by DNase II, both in chromatin and in nuclei (28). Thus, the screw symmetry of superhelical DNA may actually be broken within the core particle itselfl Any meaningful analysis of histone packing in the nucleosome core particle should help us to see its biological significance. I believe that this model suggests why there are four different histone proteins in a nucleosome, why they bind DNA where they do, and how nucleosome structure may be related to possible functions. (i) Four distinct histone proteins are necessary to achieve several delicate compromises between point group and screw symmetry. The two arginine-rich histone proteins must form an a2fi2 tetramer in order to define the locus of a screw dislocation. The 7Y2 dimer, (H2A)2, must bind opposite the tetramer to stabilize successive superhelical turns of DNA. The specific attachment sites at (50, 110') and (110, 50') preclude histone binding at intervals compatible with the 80-base-pair superhelical screw. The fourth histone, 6 (H2B), is necessary to create a distinct linker. The two slightly lysine-rich histones therefore both act, each in a different way, to break the potential screw symmetry of the DNA supercoil and thereby to divide the eukaryotic chromatin fiber into nucleosomes. (ii) The beaded architecture creates two prominent structural elements-dyad axes at the center of core particles and linkers between them-that can serve a variety of recognition purposes. Long-range recognition mechanisms may exploit the fact that the central segment containing the dyad is shorter than the others. Precise, sequence-specific recognition mechanisms may utilize the fact that dyad axes retained in the core particle pass between successive base pairs, because recognition sequences in double-stranded DNA-e.g., for restriction endonucleases-are almost invariable related by such dyads (29). (I) The nucleosome itself is a metastable transition state, both structurally and functionally, between melted or active and condensed or inactive DNA conformations. Kornberg (10, 26) has pointed out that discrete bundling of DNA may create a flexibly jointed chain that can build higher-order structures and condense enormous amounts of DNA into chromosomes. Nucleosome structure must be equally important in facilitating the converse process of melting chromosomes. For, just as DNA

3653

helixes are structurally rigid for the purpose of packaging, so siperhelixes are functionally inert with regard to gene expression. Many colleagues provided helpful discussions. Special thanks are due to L. D. Kuntz, J. Hermans, J. D. Griffith, and a referee for criticism of the manuscript; and to M. Blackburn for help with the illustrations. This work was supported by National Institutes of Health Grant GM 21991, University of North Carolina Medical Faculty Grant VB 169, and a Jefferson-Pilot Fellowship in Academic Medicine.

Simpson, R. T. & Whitlock, J. P., Jr. (1976) Cell 9,347-353. Noll, M. (1977) J. Mol. Biol. 116,49-71. Lutter, L. (1977) J. Mol. Biol. 117,53-69. Felsenfeld, G. (1978) Nature 271, 115-122. Finch, J. T., Lutter, L. C., Rhodes, D., Brown, R. S., Rushton, B., Levitt, M. & Klug, A. (1977) Nature 269,29-36. 6. Sussman, J. L. & Trifonov, E. N. (1978) Proc. Natl. Acad. Sci. USA 75, 103-107. 7. Levitt, M. (1978) Proc. Nat!. Acad. Sci. USA 75,640-644. 8. Harrington, R. (1977) Nucleic Acids Res. 4,3519-3535. 9. Pardon, J. F., Worcester, D. L., Wooley, J. C., Cotter, R. I., Lilley, D. M. J. & Richards, B. M. (1977) Nucleic Acids Res. 4,31993214. 10. Kornberg, R. D. (1974) Science 184,868-871. 11. Matthews, B. W. & Bernhard, S. A. (1973) Annu. Rev. Blophys. Bloeng. 2,257-317. 12. Weintraub, H., Worcell, A. & Alberts, B. (1976) Cell 9, 409417. 13. D'Anna, J. A. & Isenberg, I. (1974) Biochemistry 13, 49924997. 14. Thomas, J. 0. & Kornberg, R. D. (1974) Science 184, 865868. 15. Camerini-Otero, R. D., Sollner-Webb, B. & Felsenfeld, G. (1976) Cell 8,333-S347. 16. Camerini-Otero, R. D. & Felsenfeld, G. (1977) Nucleic Acids Res. 4, 1159-1181. 17. Bina-Stein, M. & Simpson, R. T. (1977) Cell 11, 609-618. 18. Oudet, P., Gross-Bellard, M. & Chambon, P. (1975) Cell 4, 281-00. 19. Mirzabekov, A. D., San'Ko, D. G., Kolshinsky, A. M. & Melnikova, A. F. (1977) Eur. J. Biochem. 75,379-389. 20. Caspar, D. L. D. & Klug, A. (1962) Cold Spring Harbor Symp. Quant. Biol. 27,1146. 21. Fasman, G., Chou, P. Y. & Adler, A. J. (1976) Biophys. J. 16, 1201-1238. 22. Simpson, R. T. (1976) Proc. Nat!. Acad. Sci. USA 73, 44004404. 23. Mirzabekov, A. (1977) Cold Spring Harbor Symp. Quant. Biol. 42, 149-156. 24. Richards, B., Cotter, R., Lilley, D., Pardon, J., Wooley, J. & Worcester, D., (1976) in Current Chromosome Research, eds. Jones, K. & Brandhaur, P. E. (Elsevier, Amsterdam), pp. 7-16. 25. Church, G. M., Sussman, J. L. & Kim, S. H. (1977) Proc. Nat!. Acad. Sci. USA 74,1458-1462. 26. Kornberg, R. D. (1977) Annu. Rev. Biochem. 46, 931-954. 27. Klug, A. & Durham, A. T. C. (1970) Cold Spring Harbor Symp. Quant. Biol. 36,449-460. 28. Altenburger, W., H6rz, W. & Zachav, H. G. (1976) Nature 264, 1. 2. 3. 4. 5.

517-522. 29. Roberts, R. (1976) Crit. Rev. Biochem. 4, 123-164.

Histone packing in the nucleosome core particle of chromatin.

Proc. Nati. Acad. Scf. USA Vol. 75, No. 8, pp. 3649-3653, August 1978 Biochemistry Histone packing in the nucleosome core particle of chromatin (DNA...
1MB Sizes 0 Downloads 0 Views