Reports

Crystal structure of the overlapping dinucleosome composed of hexasome and octasome

See allHide authors and affiliations

Science  14 Apr 2017:
Vol. 356, Issue 6334, pp. 205-208
DOI: 10.1126/science.aak9867

Nucleosomes in contact

In eukaryotic cells, genomic DNA must be compacted to fit inside the nucleus. A key player in DNA packaging is the nucleosome, which comprises a segment of DNA wrapped around an octamer of histone proteins. During replication and transcription, nucleosomes must reposition themselves on the DNA. In this process, nucleosomes can collide to form a dinucleosome. Kato et al. report a high-resolution crystal structure of a dinucleosome. One of the octamers has lost a histone dimer so that the dinucleosome comprises an octamer and a hexamer. The structure may represent an intermediate during chromatin remodeling.

Science, this issue p. 205

Abstract

Nucleosomes are dynamic entities that are repositioned along DNA by chromatin remodeling processes. A nucleosome repositioned by the switch-sucrose nonfermentable (SWI/SNF) remodeler collides with a neighbor and forms the intermediate “overlapping dinucleosome.” Here, we report the crystal structure of the overlapping dinucleosome, in which two nucleosomes are associated, at 3.14-angstrom resolution. In the overlapping dinucleosome structure, the unusual “hexasome” nucleosome, composed of the histone hexamer lacking one H2A-H2B dimer from the conventional histone octamer, contacts the canonical “octasome” nucleosome, and they intimately associate. Consequently, about 250 base pairs of DNA are left-handedly wrapped in three turns, without a linker DNA segment between the hexasome and octasome moieties. The overlapping dinucleosome structure may provide important information to understand how nucleosome repositioning occurs during the chromatin remodeling process.

Eukaryotic genomic DNA is compacted as chromatin, in which nucleosomes are connected by short linker DNA segments and adopt the “beads-on-a-string” topology (1). Nucleosomes are dynamic entities that slide and reposition along DNA by spontaneous or enzymatic remodeling processes (25). Consequently, a repositioned nucleosome collides with a neighbor to form the intermediate “overlapping dinucleosome,” in which two nucleosomes are intimately associated (6, 7). The overlapping dinucleosome may also be formed in a DNA-sequence–directed manner if two nucleosome positioning sequences are placed at a proper distance along the genomic DNA (710). However, the structure of the overlapping dinucleosome has remained elusive.

The overlapping dinucleosome can reportedly be reconstituted with a 250–base-pair (bp) DNA containing tandem nucleosome positioning sequences derived from Widom 601 sequences (fig. S1A) (7, 11). In this 250-bp DNA (–44 bp), the downstream 22 bp and the upstream 22 bp of each Widom 601 sequence were deleted (fig. S1A). We reconstituted nucleosomes with human histones H2A, H2B, H3.1, and H4, in the presence of the 250-bp DNA segment, by the salt-dialysis method. As a result, mononucleosomes and overlapping dinucleosomes were simultaneously assembled (fig. S1B, lane 1). The reconstituted mononucleosomes and overlapping dinucleosomes were purified by the nondenaturing polyacrylamide gel electrophoretic (PAGE) method (fig. S1B, lanes 2 and 3). All four histones were incorporated into both the purified mononucleosomes and overlapping dinucleosomes (fig. S1C). Electrospray ionization mass spectrometry (ESI-MS) revealed that the molecular mass of the reconstituted overlapping dinucleosome is 350,834 Da (±25), which corresponds very well to the theoretical molecular mass (350 kDa) for the sum of a histone hexamer (one H2A-H2B dimer and two H3-H4 dimers), a histone octamer (two H2A-H2B dimers and two H3-H4 dimers), and 250 bp of DNA (Fig. 1A). Therefore, the overlapping dinucleosome, reconstituted with the 250-bp DNA segment, is composed of a hexasome and an octasome. In contrast, ESI-MS also revealed that the dinucleosomes with the 266, 280, and 294-bp DNAs, in which 28, 14, and 0 bp of the DNA regions were deleted between the two Widom 601 sequences, accommodate two octasomes (fig. S2). These results indicated that the 28-bp DNA deletion between the two Widom 601 sequences is not sufficient to form the overlapping dinucleosome.

Fig. 1 Crystal structure of the overlapping dinucleosome.

(A) Native ESI-MS of the overlapping dinucleosome. Gray arrowheads correspond to multiply charged ions of the overlapping dinucleosome. The charge state of the mainly observed peak is labeled above the peak. (B) The crystal structure of the overlapping dinucleosome. The octasome and hexasome moieties are shown in magenta and light blue, respectively. The electron density of a 131- to 135-bp DNA region from the octasomal DNA edge is not interpretable because of ambiguity (indicated by gray dots). (C) The H2A-H2B dimers are presented in magenta and light blue with the schematic image (right).

We determined the crystal structure of the overlapping dinucleosome at 3.14-Å resolution (table S1 and Fig. 1B). In the crystal structure, the hexasome and octasome moieties are clearly observed (Fig. 1B). One H2A-H2B dimer is missing at the hexasome-octasome interface (Fig. 1C). Surprisingly, the hexasome moiety directly contacts the neighboring octasome and intimately associates without an obvious linker DNA segment between them (Fig. 1B). At the hexasome-octasome interface, the overlapping dinucleosome lacks an H2A-H2B acidic patch of the hexasome moiety and conceals one H2A-H2B acidic patch of the octasome moiety (fig. S3A). The H2A-H2B acidic patches are known to be binding sites for nucleosome binding proteins, such as RCC1 (12). Actually, the binding of the RCC1(22-421) domain to the overlapping dinucleosome was markedly reduced (fig. S3, B and C). Therefore, the nucleosome binding proteins may be precluded from binding to the hexasome-octasome interface. This may prevent the reverse nucleosome positioning, which may be mediated by a nucleosome remodeling factor bound to the hexasome-octasome interface and thus support unidirectional nucleosome remodeling.

Interestingly, the hexasomal H3 αN helix facing toward this area lacking the hexasomal H2A-H2B dimer is drastically shortened (Fig. 2). In contrast, the structures of the other three H3 αN helices are unchanged in the overlapping dinucleosome (Fig. 2).

Fig. 2 The structure of the hexasome H3 αN helix facing the octasome moiety is drastically changed.

(A) Secondary structures of the H3.1 molecules in the overlapping dinucleosome. H3.1 molecules are colored light blue (i and ii) and magenta (iii and iv) for those in the hexasome and octasome moieties, respectively. The sequence from amino acids 38 to 134 is shown at the top. Amino acid residues incorporated into the α helices of the canonical nucleosome are colored red. (B) Locations of the H3 α helices in the overlapping dinucleosome. The H3 α helices labeled with (i), (ii), (iii), and (iv) correspond to the H3.1 molecules shown in (A). (C) Close-up views of four H3 αN helices (i, ii, iii, and iv) in the overlapping dinucleosome. The H2A molecules are shown in orange. (D) The structures of the H3 αN helices (i, ii, iii, and iv). The 2mFo-DFc maps were calculated and contoured at the 1.0 σ level. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

In the hexasome moiety, the H3 Lys56 and H3 Thr80 residues are located close to the DNA backbone of the octasome (Fig. 3A and fig. S4A). The H3 Lys56 residue is located at the C-terminal end of this shortened H3 αN helix (Fig. 2). In the octasome moiety, the H2A Asn68, H2A Arg71, H2B Lys108, H2B Ser112, and H2B Lys116 residues are located close to the DNA backbone of the hexasome (Fig. 3A and fig. S4A). None of these residues directly interacts with the DNA in the mononucleosome, because these residues are at least 5 Å away from the DNA (fig. S4B) (13, 14). Therefore, these residues may potentially form new hydrogen bonds with the DNA at the hexasome-octasome interface of the overlapping dinucleosome.

Fig. 3 The novel histone-DNA interactions in the overlapping dinucleosome.

(A) Locations of the mutated residues. The octasome and hexasome moieties are shown in magenta and light blue, respectively. (B) Purified overlapping dinucleosomes reconstituted with wild-type (lane 1), 7A mutant (lane 2), 7DE mutant (lane 3), or 5DE mutant (lane 4) histones were analyzed by nondenaturing 6% PAGE. (C to F) SAXS profiles (upper graphs) and Guinier plots (lower graphs) of the wild-type (C), 7A mutant (D), 7DE mutant (E), and 5DE mutant (F) overlapping nucleosomes. Solid lines represent the results of the least-square fitting for the SAXS intensity, I(Q), with the Guinier formula, I(Q) = I0exp(–Rg2Q2/3), Q = (4п/λ)sin(θ/2): Q, λ, θ, I0 and Rg are the magnitude of scattering vector, the wavelength of x-ray (1.54 Å), the scattering angle, the zero angle scattering intensity and gyration radius, respectively.

To determine whether these new histone-DNA interactions contribute to the formation of the overlapping dinucleosome structure, we prepared the H2A (N68A, R71A), H2B (K108A, S112A, K116A), and H3(K56A, T80A) mutants, in which those residues were replaced by Ala (A) (fig. S5A). The mutant overlapping dinucleosome was efficiently reconstituted with these mutants (7A mutant overlapping dinucleosome) (Fig. 3B and fig. S5B). The histone stoichiometry of the 7A mutant overlapping dinucleosome was quite similar to that of the wild-type overlapping dinucleosome, as revealed by ESI-MS (fig. S5D). This indicates that these seven amino acid replacements did not affect the formation of the overlapping dinucleosome. Small-angle x-ray scattering (SAXS) analyses revealed that the radius of gyration (Rg) of the mutant overlapping dinucleosome (Rg = 58.1 ± 0.4 Å) was larger than that of the wild-type overlapping dinucleosome (Rg = 57.2 ± 0.5 Å) (Fig. 3, C and D). This suggests that the 7A mutant overlapping dinucleosome may form a more relaxed (or loosened) structure, as compared with the wild-type structure. To confirm this result, we prepared the mutant overlapping dinucleosome containing the H2A (N68D, R71E), H2B (K108E, S112D, K116E), and H3 (K56E, T80D) mutants (7DE mutant overlapping dinucleosome) (Fig. 3B and fig. S5C). The acidic Glu (E) and Asp (D) side chains may repulse the backbone phosphates of the DNA and thus render a more relaxed overlapping dinucleosome structure than that with the Ala (A) side chain in the 7A mutant overlapping dinucleosome. ESI-MS confirmed that the 7DE mutant overlapping dinucleosome was formed with the proper histone stoichiometry (fig. S5D). As expected, SAXS analyses revealed that the radius of gyration of the 7DE mutant overlapping dinucleosome (Rg = 59.6 ± 0.6 Å) was substantially larger than those of the wild-type and 7A mutant ones (Fig. 3E). However, surprisingly, the Rg value (57.1 ± 0.5 Å) of the 5DE mutant overlapping dinucleosome, in which the acidic amino acid replace ments were introduced in H2A and H2B, but not in H3, was almost the same as that of the wild-type one (Fig. 3F). Therefore, the specific histone-DNA interactions of the H3 Lys56 and H3 Thr80 residues between the hexasome and octasome may play pivotal roles to stabilize and compact the overlapping dinucleosome structure, although they may not be essential for the overlapping dinucleosome formation.

In living cells, the linker histone H1 specifically binds to the nucleosomal DNA (1517). To determine whether H1 binds to the overlapping dinucleosome, we reconstituted the overlapping dinucleosome with a 296-bp DNA, which contains extra 23-bp linker DNAs at both ends (fig. S1A and Fig. 4A, lane 1). As a control, a regularly spaced dinucleosome with a 48-bp linker DNA segment was prepared with tandem Widom 601 sequences (Fig. 4B, lane 1). A gel electrophoretic mobility shift assay revealed that two bands corresponding to the histone H1-nucleosome complexes were observed in the experiments with a regularly spaced dinucleosome, indicating that two histone H1s are bound to each nucleosome (Fig. 4B, lanes 2 to 5). In contrast, only a single band corresponding to the complex between H1 and the overlapping dinucleosome was observed, indicating that one molecule of histone H1 binds to an overlapping dinucleosome (Fig. 4A, lanes 2 to 5). Consistently, the purified overlapping dinucleosome-H1 complex contained about half the amount of histone H1, as compared with the purified dinucleosome-H1 complex (fig. S6). To identify the histone H1 binding site, we performed a hydroxyl radical footprinting analysis with the overlapping dinucleosome containing the 296-bp DNA labeled with 5′-Cy5. In this analysis, the DNA wrapped around the histones generated a periodic cleavage pattern with about 10-bp intervals. We found specific DNA protection around the position of base 193 in the overlapping dinucleosome in the presence of histone H1 (Fig. 4C). This region is located near the dyad of the octasome moiety (Fig. 4D). Therefore, we concluded that one histone H1 specifically binds to the DNA region around the dyad of the octasome moiety in the overlapping dinucleosome.

Fig. 4 Linker histone H1 binds to the overlapping dinucleosome.

(A and B) The H1 binding to the overlapping dinucleosome (A) or a regularly spaced dinucleosome (B) was monitored by nondenaturing 4% PAGE. (C) The overlapping dinucleosomes were subjected to hydroxyl radical attack in the absence (lane 3) or presence (lane 5) of H1. Control experiments with mononucleosomes were performed in the absence (lane 2) or presence (lane 4) of H1. Arrows (right) indicate the DNA region protected from hydroxyl radical attack by the H1 binding. (D) The DNA segment that is protected in the presence of H1 is colored red.

Importantly, the linker histone H1 binds to the octasome moiety of the overlapping dinucleosome. This suggests that the overlapping dinucleosome is also formed when the repositioned nucleosome contacts a nucleosome complexed with histone H1, such as the “chromatosome” (fig. S7). The binding of a linker histone generally restricts the nucleosome mobility (1). Therefore, the removal of the linker histone, potentially mediated by a histone chaperone, from the repositioning nucleosome may be a prerequisite for the overlapping dinucleosome formation (fig. S7). The H2A-H2B dimer may be released when the octasome clashes with the chromatosome upon nucleosome remodeling. Because the H2A-H2B dimer loss is also observed during transcription, probably by RNA polymerase passage through nucleosomes (18), the H2A-H2B removal by the action of RNA polymerase may also be a prerequisite for the overlapping dinucleosome formation (fig. S7).

In the present study, we found that the 250-bp DNA was extremely resistant to attack by micrococcal nuclease (MNase), which digests the DNA region detached from the histone surface (fig. S8). Therefore, the 250-bp genomic DNA fragments generated by the MNase treatment may be footprints reflecting the existence of the overlapping dinucleosome in vivo. Our MNase sequencing experiments with extensively MNase-digested HeLa cell chromatin suggested that the 250-bp DNAs mapped at the downstream (+1) regions of transcription start sites (TSSs) (fig. S8). Such substantial MNase digestion to the mononucleosome level may remove the DNA-binding proteins, such as transcription factors. Therefore, the overlapping dinucleosome may predominantly form at the regions downstream of TSSs, which are potential target sites for nucleosome remodelers and RNA polymerases. However, we could not eliminate the possibility that the 250-bp DNA fragments may be protected from the MNase attack by nonhistone DNA-binding proteins, although it is unlikely that DNA-binding proteins covering about 250 bp still remain on the genomic DNA under such extensive MNase digestion conditions. In the future, it will be intriguing to study the correlation between the overlapping dinucleosome formation and the genomic DNA function in cells.

Supplementary Materials

www.sciencemag.org/content/356/6334/205/suppl/DC1

Materials and Methods

Figs. S1 to S8

Table S1

References (1936)

References and Notes

Acknowledgments: We thank F. Adachi, Y. Xie, T. Kujirai, H. Tachiwana (Waseda University), and H. Kimura (Tokyo Institute of Technology) for technical support and suggestions. The synchrotron radiation experiments were performed with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (proposal no. 2016A2537) and the Photon Factory Program Advisory Committee (proposal no. 2014G556). This work was supported in part by JSPS KAKENHI grant nos. JP25116002 (to H.Ku.), JP25116010 (to Y.O.), JP25116003 (to H.Ko.), JP15K18515 (to R.I.), JP15H02042 and JP16H01306 (to M.S.), and CREST, JST (to H.Ku. and Y.O.). This research is partially supported by the Platform Project for Supporting Drug Discovery and Life Science Research (Platform for Drug Discovery, Informatics, and Structural Life Science) from Japan Agency for Medical Research and Development (AMED) (to H.Ku., H.Ko., S.A., and Y.N.). H.Ku. and N.H. were supported by the Waseda Research Institute for Science and Engineering. D.K. was supported by a Research Fellowship from JSPS for Young Scientists (16J06850). This work was also supported by the Project for Construction of the Basis for the Advanced Materials Science and Analytical Study by the Innovative Use of Quantum Beam and Nuclear Sciences in the Research Reactor Institute of Kyoto University (to M.S and R.I.). This work was partly performed in the Cooperative Research Project Program of the Medical Institute of Bioregulation and Advanced Computational Scientific Program of the Research Institute for Information Technology, Kyushu University. The atomic coordinates of the overlapping dinucleosome have been deposited in the Protein Data Bank (PDB), under PDB accession number 5GSE. The MNase sequencing data have been deposited to the DDBJ Sequence Read Archive with accession number DDBJ:DRA005526. D.K., A.O., Y.M., Y.A., N.H., and S.-Y.P. performed biochemical and structural experiments. K.S., A.O., S.A., and Y.N. performed ESI-MS experiments. R.I., D.K., A.M., H.Ko., and M.S. contributed to the solution scattering and computational analyses. D.K., J.N., K.M., and Y.O. performed the genome-wide analysis. H.Ku. conceived, designed, and supervised all of the work and wrote the paper. All of the authors discussed the results and commented on the manuscript.
View Abstract

Navigate This Article