Jeffrey J. Hayes is in the Department of Biochemistry Biophysics, University of Rochester School of Medicine, 601 Elmwood Avenue, Rochester New York 14642, USA. jjhs@mail.rochester.edu
Two new biophysical analyses of the folding behavior of model oligonucleosomal arrays have provided interesting new insights into how a histone mutation that mimics a chromatin remodeling activity and the incorporation of a histone variant into chromatin alter chromatin structure and function.
Chromatin is a complex, multifacted hierarchical array of structures that makes up eukaryotic chromosomes. Chromosomes are only 50% (by weight) DNA; most of the remaining mass is comprised of a class of structural proteins known as histones1. The basic repeating subunit of chromatin, the nucleosome, contains 200 bp of DNA, of which 147 bp are wrapped in 1 3/4 turns around an octamer of core histones to form the 'nucleosome core'1,
2. The additional DNA lies between nucleosome cores and is known as linker DNA. Two copies of each of the four core histones, H2A, H2B, H3 and H4, comprise the octamer while an additional 'linker histone' (such as H1, H5, H1°) binds to the core and the linker DNA, which participates in compacting the strings of nucleosomes into a chromatin fiber of 30 nm in diameter1,
3. The histones and several other classes of much lower-abundance proteins promote the assembly of chromatin fibers into as yet poorly defined higher-order chromatin structures1,
3.
Eukaryotic cells have developed multiple strategies to overcome the obvious impediments to the access of DNA provided by chromatin structure. For example, post-translational modification of the core histones such as acetylation, methylation or phosphorylation either provide 'landmarks' for the binding of activator proteins or can lead directly to alterations in the structure of chromatin and accessibility of the packaged DNA4. In addition, molecular machines possessing 'chromatin remodeling' activities and occasional substitution of non-allelic variants of the major types of core histones provide localized functionality within chromatin (see below). Two papers in this issue of Nature Sturctural Biology provide new insights into the mechanisms by which these latter two strategies exert their effects5,
6.
Histones are now perceived as integral components of mechanisms that control gene transcription7. Historically, an important clue to the role of histone proteins in transcriptional control came from genetic experiments in Saccharomyces cerevisiae in which a multicomponent general activator of transcription was identified8,
9. This work was based on the discovery of 'switch' genes (SWI) required for appropriate expression of a gene required for yeast mating type switching, and that of sucrose non-fermentation mutants in several SNF (pronounced 'sniff') genes required for expression of a gene involved in sucrose fermentation8,
9. Later studies showed that the protein products of these genes are contained within a single complex known as the SWI/SNF complex that is required for the up-regulation of transcription of many yeast genes10,
11. Purification of this complex and the identification of homologous complexes in higher eukaryotes led to the discovery that these complexes constitute 'chromatin remodeling' activities12,
13. These activities use the energy of ATP hydrolysis to somehow alter chromatin structure and help alleviate the repressive effects of chromatin structure on transcription14.
The SIN mutations Interestingly, genetic screens in yeast for mutations that allowed transcription in the absence of functional SWI proteins lead to the identification of two SWI-independent (SIN) transcription genes that were found to encode structural components of chromatin15. SIN1 is a nuclear protein that has homology to the group of HMG1/2 nucleosome-binding proteins, whereas SIN2 encodes the core histone H3. More SIN mutations were subsequently identified in the gene encoding core histone H4, the dimerization partner of H3 (ref. 15). Core histone dimerization is mediated by the histone fold domain present in each protein16. Two H2A−H2B and two H3−H4 heterodimers abut end-to-end to form a symmetrical superhelical ramp of protein consisting of (H2A−H2B):(H4−H3): (H3−H4):(H2B−H2A) onto which the DNA is wrapped2,
16 (Fig. 1a). Pairing of protein loops between long and short -helices in the histone fold domains, as well as the pairing of the N-terminal ends of the first helical domain from each of the histones in the heterodimers create repeated DNA-contacting elements. Thus, eight 'paired loops' and four 'paired ends of helices' motifs provide twelve DNA contact sites2,
16 (Fig. 1a). The amino acid changes in histone H3 and H4 that result in the SIN phenotypes are primarily in and near the set of paired loops located closest to the dyad axis of symmetry in the center of the nucleosome (Fig. 1a).
Figure 1. Location of SIN mutations and essential amino acids of H2A.Z within the core histone octamer.
a, Cartoon of nucleosome structure. H4−H3 (red and yellow), and H2B−H2A (green and blue) heterodimer are shown with -helices depicted as columns (based on ref. 2). The paired ends of -helices (PH) and paired loop (PL) motifs involved in DNA binding and the H3−H3 and H4−H2B interfaces between dimers (solid and dashed straight lines, respectively) are indicated. Note that only the top half of the symmetrical nucleosome structure is shown for clarity; this view contains one copy each of the core histones and one turn of DNA. A small amount of H3 (light yellow) emanating from the bottom half of the structure is shown as well. Stars indicate the approximate positions of the SIN2 mutations in one of the two copies of H3 and H4 near the PL region nearest the center (dyad) of the histone octamer5,
15,
17. Red star indicates the location of H4R45. Light blue indicates the region of H2A.Z essential for viability in Drosophila22. The curved dashed line indicates a possible path for linker DNA exiting the core. b, Defined conformational transitions in the model nucleosomal array22. In low salt the array exists in a 29S extended conformation. In solutions in physiological ionic strength, arrays assembled with wild type major core histones at the extended structure equilibrate with a 40S and 55S structure corresponding to a partially folded intermediate and a fully condensed 30 nm diameter chromatin fiber. Note that all three conformations can also undergo cation-induced self-association22.
The location of the amino acid changes associated with the SIN phenotypes suggested that these mutations would result in altered histone interactions with the central turn of DNA in the nucleosome15. Thus, many researchers believed that the SIN mutations would affect structure or stability of the nucleosome itself15,
17. However, while some of the SIN mutants do cause marginal changes in the ability of the core histones to protect the DNA from digestion by nucleases or to constrain DNA supercoiling, other SIN mutations behaved like wild type histones in these assays5,
17. Indeed the study of Horn et al.5 provides evidence that the SIN mutations may affect chromatin fiber dynamics more so than intranucleosomal interactions (see below).
The H2A.Z variant As mentioned above, chromatin structure and function in vivo is also thought to be modulated by incorporation of specialized core histone variants. However, the specialized properties imparted by these variants have not been elucidated. Perhaps the histone variant best characterized structurally is H2A.Z, a protein essential for the viability of several organisms (ref. 5 and refs therein), supplying function(s) in chromatin distinct from the major H2A protein18. In yeast H2A.Z appears to be involved in regulation of gene expression both by facilitating chromatin remodeling activities19 and, paradoxically, by establishing a specialized chromatin structure required for silencing mating-type loci20. Interestingly, sequence-swap experiments and an X-ray crystal structure of an H2A-containing nucleosome indicate that the essential biological function of this histone variant is provided by a portion of the H2A.Z C-terminus that is exposed on the surface of the nucleosome21,
22 (Fig. 1). This suggests that H2A.Z mediates interactions between nucleosomes or with other chromatin-associated proteins.
Properties of the variant nucleosome arrays The molecular bases for the observed biological effects of SIN mutants and H2A.Z in vivo have not been adequately defined. Investigations of structure/function relationships in chromatin are often hindered by the complexity and heterogeneity inherent in these structures. A key to the studies of both Horn et al.5 and Fan et al.6 is the use of a well-defined model system consisting of a nucleosomal array reconstituted from purified core histones and a DNA template of tandemly repeated nucleosome positioning sequences. The conformational dynamics of this system have been extensively characterized by analytical ultracentrifugation, nuclease digestion and analytical gel techniques22. In nonphysiological low-salt solutions, the reconstituted array exists as an extended string of nucleosomes that sediments at 29S23, much like the 'beads-on-a-string' structure of native chromatin viewed in electron micrographs prepared under similar conditions1 (Fig. 1b). In solutions at physiological ionic strength this 29S extended structure equilibrates with a 40S folding intermediate and a more rapidly sedimenting 55S structure, reflecting the hydrodynamic behavior expected for a fully compacted chromatin fiber 30 nm in diameter3,
23. Importantly these model chromatin complexes also undergo divalent cation-mediated self-association, a process thought to mimic the longer-range interactions between chromatin fibers necessary for the formation of higher-order chromatin structures3.
Interestingly, Horn et al.5 demonstrate that model nucleosomal arrays reconstitute as efficiently with core histones containing well-characterized SIN mutations in H4 (H4R45C or H4R45H) as with wild type histones and that the SIN nucleosomes exhibit normal resistance to DNA cleavage by a restriction enzyme5. This result strongly suggests that the mutation itself does not yield unstable nucleosomes or structures resembling nucleosomes already 'remodeled' by the activity of SWI/SNF remodeling complexes. Analysis of the arrays by analytical ultracentrifugation and gel mobility assays suggests that the SIN mutation may cause a slight unwrapping of the DNA from the edges of the nucleosomes but nucleosome positioning appears unchanged by the mutation. Thus the biological properties of this particular SIN mutation cannot be easily explained by remodeling-like alterations of nucleosome structure itself.
However, the SIN arrays do exhibit a striking lack of ability to undergo salt-induced condensation. At salt concentrations that cause wild type arrays to condense into 30−55S structures, the SIN mutant-containing arrays do not sediment greater than 32S. Thus the acquisition of a single SIN mutation in H4 results in a complete inability to form structures resembling the compacted chromatin fiber. Interestingly, the SIN arrays still exhibit normal interarray self-association, suggesting that the mutation does not affect long-range interactions required for the formation of higher-order chromatin structures.
In contrast, substituting the H2A.Z histone variant for the major type H2A within model nucleosomal arrays resulted in quite different behavior than that observed with the H4 SIN mutant. While the arrays reconstituted with H2A.Z as efficiently as with major type H2A, subsequent analysis indicates that the folding behavior of the variant-containing arrays is significantly enhanced compared to the major type. In buffers lacking divalent salts, both arrays exhibited identical hydrodynamic profiles, suggesting both formed similar extended nucleosomal structures with similar nucleosomal stabilities (although a contradicting result regarding nucleosomal stability was reported in ref. 24). However, at all divalent cation concentrations tested, the H2A.Z arrays were on average more highly condensed. Futhermore, and again in striking contrast to the SIN arrays, arrays containing H2A.Z were significantly impaired in their ability to undergo intermolecular self-association compared to those containing major type H2A. Thus, H2A.Z appears to be designed to promote disassembly of higher-order chromatin structures.
Perspectives These studies highlight the multifaceted role of the core histone proteins in directing the assembly of individual nucleosomes, mediating folding of nucleosome arrays and directing interfiber interactions important for assembly of arrays into higher-order structures3. Thus, SIN mutations or the incorporation of a non-allelic variant core histone into the structure does not necessarily imply that the resultant alterations will be evident at the level of the nucleosome itself. Indeed, both of these changes have little effect on nucleosome structure by a number of criteria. Rather, these studies emphasize the multiple subtle ways chromatin structure may be regulated from within. In the case of the SIN mutation, condensation of the chromatin fiber is drastically reduced; in the case of H2A.Z, the condensed fiber is actually stabilized to some extent, however, interfiber interactions are inhibited, presumably affecting the stability of higher-order structures. Thus activities that use DNA as a substrate can be influenced in biologically significant ways by altering the equilibria at multiple levels above the wrapping of DNA about the nucleosome.
The work of Horn et al.5 and Fan et al.6 raises a number of interesting questions. First, it remains to be demonstrated that H2A.Z is present exclusively in long contiguous stretches of nucleosomes. Thus, what is the minimum H2A.Z content required to observe a drastic reduction in interfiber interactions? What is the effect of either SIN mutants or H2A.Z in arrays containing linker histones? Do all SIN mutants result in a drastic reduction in the stability of the condensed chromatin fiber or are there multiple ways in which these mutants can affect chromatin? Analysis of other histone mutants is an obvious target. In addition, the fact that a single amino acid substitution in the H4 SIN mutation causes such a drastic effect on structure suggests that small numbers of specific post- translational modifications may also have similar effects.
200 bp of DNA, of which 147 bp are wrapped in 1 3/4 turns around an octamer of core histones to form the 'nucleosome core'
-helices in the histone fold domains, as well as the pairing of the N-terminal ends of the first helical domain from each of the histones in the heterodimers create repeated DNA-contacting elements. Thus, eight 'paired loops' and four 'paired ends of helices' motifs provide twelve DNA contact sites
