Contribution of histone N‐terminal tails to the structure and stability of nucleosomes

1The basic unit of chromatin is the nucleosome core particle, which contains 145–147 base pairs of DNA. The DNA binds on the surface of the histone octamer, composed of two copies of the four core histones, H2A, H2B, H3, and H4. In the nucleosome structure, each core histone contains two common regions, the “histone fold” and the “histone tail”. Histone tails are flexible regions that flank both ends of the histone fold (Fig. 1(A)). In the nucleosome, the histone fold is responsible for the formation of stable H2A–H2B and H3–H4 dimers, and the histone octamer is composed of two H2A–H2B dimers and two H3–H4 dimers. On the other hand, the N-terminal tails of the four core histones and the C-terminal tail of histone H2A protrude from the nucleosome core particle. These histone tails are known to interact with nucleosomal DNA, and the interactions are substantially retained even in the highly acetylated state. In addition, the histone tails not only contact the DNA wrapped around the histone octamer, but also bind to linker DNA and the acidic patches of the neighboring nucleosomes. These interactions between histone tails and DNA may play critical roles in the formation of higher-order chromatin.

2Biochemical studies of “tailless” nucleosomes have revealed the functional importance of histone tails. The N-terminal tails of H2A, H3, and H4 are considered to function in the formation of higher-order chromatin, and the H2B N-terminal tail is specifically required for chromosome condensation. Removal of the histone tails increases the accessibility of the nucleosomal DNA, probably by enhancing the nucleosome dynamics and/or reducing the nucleosome stability. Consistently, deletion and alanine scanning mutation analyses of the N-terminal tail of H3 or deletion of the N-terminal tail of H2B revealed enhanced nucleosome sliding along the DNA. Interestingly, opposite effects have been reported for nucleosomes lacking the N-terminal tail of H2B and/or H4. A molecular dynamics simulation also suggested that the truncation of the histone tail affects the nucleosome structure. The function of each histone N-terminal tail may be redundant, because the removal of the N-terminal tail of each single histone did not have an obvious effect on cell viability, although the simultaneous deletion of two N-terminal tails from the histone pairs H2A/H2B, H3/H4, or H2A/H4 compromised cell survival.

3The histone tails are highly basic, and contain residues that are targets of post-translational modifications, such as acetylation, methylation, phosphorylation, and crotonylation. These histone–tail modifications modulate the histone–DNA and nucleosome–nucleosome interactions, and may play important roles during transcription, replication, recombination, and DNA repair. Therefore, it is important to understand the molecular mechanisms by which the histone tails affect the nucleosome structure and stability.

4In the present study, we determined the crystal structures of four nucleosomes lacking the N-terminal tail of one of the histones, H2A, H2B, H3, and H4. We evaluated the contribution of each histone N-terminal tail to the nucleosome stability and structure.

5Human H2A, H2B, H3.1 and H4 lacking the N-terminal tail regions were overexpressed in Escherichia coli with N-terminal His6-tags. The His6 tag was removed by thrombin treatment during the purification procedure. Four non-native residues, glycine–serine–histidine–methionine from the expression vector, remained at the N-terminus of each histone. Previously reported protocols were used to express, purify, and reconstitute the tailless nucleosomes. A 146 bp palindromic sequence derived from human α-satellite DNA was used for nucleosome reconstitution. Each of these tailless histones was incorporated into nucleosomes by the salt-dialysis method, and the nucleosomes were purified by native polyacrylamide gel electrophoresis. Crystals for all of the tailless nucleosomes were obtained by the hanging drop vapor diffusion method, using 20 mM potassium cacodylate (pH 6.0), 50 mM KCl, and 110–155 mM MnCl2 as the crystallization solution. Drops composed of 1 μl nucleosome solution and 1 μl crystallization solution were equilibrated against 500 μl of reservoir solution, containing 20 mM cacodylate (pH 6.0), 35–40 mM KCl, and 60–80 mM MnCl2, at 20 °C.

6The diffraction data were collected at the beamline AR-NW12A at the Photon Factory (KEK, Tsukuba, Japan), at a wavelength of 1.0000 Å. Crystals were soaked in cryoprotectant solutions containing 20 mM potassium cacodylate (pH 6.0), 40 mM KCl, 60 mM MnCl2, 30% 2-methyl-2,4-pentanediol, and 2% trehalose, and were flash-cooled in a stream of nitrogen gas at 100 K. The data were indexed, integrated, and scaled with HKL2000, and were further processed using the CCP4 suite programs. All structures were solved by the molecular replacement method with the program MOLREP, using the structure of the intact human nucleosome (Protein Data Bank code 3AFA) as the search model. All models were checked using sigma-A-weighted composite omit maps during the modeling. The models were rebuilt with COOT and refined with CNS. The statistics for data collection and refinement are provided in Table 1. All molecular graphics images were generated using PyMOL (pymol.sourceforge.net).

7The exonuclease assay was conducted according to the same method described previously. Briefly, each reconstituted nucleosome, containing tlH2A, tlH2B, tlH3, or tlH4, was treated with 5 units of Escherichia coli exonuclease III (Takara), in 10 μl of 50 mM Tris–HCl (pH 8.0), 5 mM MgCl2, and 1 mM DTT. The reaction was continued for 0, 2, 4, or 8 min at 37 °C, and was stopped by the addition of 55 μl of proteinase K solution (20 mM Tris–HCl (pH 8.0), 20 mM EDTA, 0.5% SDS, and 0.5 mg ml−1 proteinase K (Roche)). After a 15 min treatment at room temperature, the DNA was extracted with phenol/chloroform, precipitated with ethanol, and dissolved in Hi-Di Formamide (Applied Biosystems). The DNA samples were then analyzed by 10% denaturing PAGE, with a gel containing 7 M urea in 0.5× TBE buffer (21 V cm−1 for 1.5 h).

8The nucleosome stability was monitored by the thermal stability assay. SYPRO Orange (SIGMA-ALDRICH) preferentially binds to hydrophobic regions of proteins, but not to nucleic acids. Therefore, thermal denaturation of the nucleosome can be detected, as the fluorescence signals of the SYPRO Orange bound to the denatured histones. Each nucleosome, containing tlH2A, tlH2B, tlH3, or tlH4 (final concentration, 2.25 μM), was prepared in 20 μl of a solution composed of 18 mM Tris–HCl (pH 7.5), 0.9 mM EDTA, 0.9 mM DTT, and SYPRO Orange (final concentration, 5×). The sample temperature was increased by the StepOnePlus™ Real-Time PCR unit (Applied Biosystems), and the fluorescence signals were measured with this system. Since the wavelength at the fluorescence emission maximum of SYPRO Orange is 570 nm, the fluorescence filter ‘filter 3’, which covers the emission wavelength ranges of the TAMRA (580 nm) and NED (575 nm) dyes, was used for detecting the fluorescence of SYPRO Orange bound to the denatured histones. The temperature gradient was from 25 to 95 °C, in steps of 1 °C/min. The fluorescence intensity from the SYPRO Orange probe bound to the denatured histones was automatically converted to the normalized reporter value (Rn), and the Rn value was plotted every minute.

9We bacterially expressed and purified four histone mutants (tlH2A, tlH2B, tlH3, tlH4), which lacked the N-terminal 9, 24, 27, and 15 amino acid residues of H2A, H2B, H3, and H4, respectively (Fig. 1(A) and (C)). These truncated N-terminal regions of the histones contain many previously identified post-translational modification sites. These deletion mutants were designed based on the tailless histones obtained by the trypsin protease treatment, and the previous crystal structures of nucleosomes (Fig. 1(B)).

10We then reconstituted four nucleosomes, each containing one of the tailless histones. Reconstitution was performed by the salt dialysis method, and the nucleosomes were purified by native polyacrylamide gel electrophoresis using a Prep Cell apparatus. The purified nucleosomes containing one of the tailless histones migrated differently on the native polyacrylamide gel (Fig. 1(D)), and contained stoichiometric amounts of the core histones (Fig. 1(E)). The exonuclease treatment assay revealed that all four tailless nucleosomes exhibited similar susceptibility towards ExoIII digestion, as compared to the wt nucleosome (Fig. 2). These results suggested that the DNA ends were similarly wrapped within the nucleosomes, and that tlH2A, tlH2B, tlH3, and tlH4 were properly incorporated into nucleosomes.

11We next crystallized all four tailless nucleosomes, and determined their structures at resolutions ranging from 3.0 to 3.4 Å (Fig. 3, Table 1). All four tlH2A, tlH2B, tlH3, and tlH4 nucleosome crystals belonged to the space group P212121, which is the same as the wt nucleosome crystal (PDB code 3AFA), and the packing arrangements were almost identical among the four tailless nucleosomes. The root mean square deviations (RMSDs) between the structures of the wt nucleosome and each tailless nucleosome were calculated by the superimposition of all Cα atoms of the histones and all phosphorus (P) atoms of the DNA. The overall RMSD values for the Cα and P atoms to the wt nucleosome were 0.499 and 0.703 for tlH2A, 0.667 and 0.897 for tlH2B, 0.502 and 0.627 for tlH3, and 0.568 and 0.749 for tlH4, respectively (Table 1). Therefore, the histone structure and the DNA path in the tailless nucleosomes are not significantly different from those of the wt nucleosome. We then compared the DNA trajectory in the tlH2B nucleosome with that in the wt nucleosome (Fig. 3(E)), since the largest RMSD value for the DNA was observed between the tlH2B and wt nucleosomes (Table 1). The N-terminal tail of H2B directly binds to the DNA, and the N-terminal truncation of H2B slightly, but clearly, affected the DNA trajectory near its binding site (Fig. 3(E)).

12We compared the structure and electron density maps of the tlH3 nucleosome with those of the wt nucleosome. We found that the electron density map of the H3 K37–R42 region of the tlH3 nucleosome, which is a binding site for DNA, was very ambiguous as compared to that of the wt nucleosome (Fig. 4(A) and (B)). The H3 H39–R42 region directly binds to a minor groove of the two DNA gyres in the wt nucleosome: the side chain of the H3 H39 residue is inserted in the minor groove of the entry gyre, and the H3 R40 and R42 side-chains form hydrogen bonds with the atoms of the sugar ring, the adenine base, and the phosphate backbone of another DNA gyre (Figs. 1(B) and 4(A)). In the tlH3 nucleosome, the hydrogen bonds between the DNA and the H3 R40 and R42 residues could be disrupted or weakened (Fig. 4(B)). Alternatively, the H3 K37–R42 binding to DNA may be stably retained but more variable in the tlH3 nucleosome, thus producing the ambiguous electron densities. However, previous analyses, such as single molecule transcription experiments, molecular dynamics simulations, and FRET measurements, indicated that the deletion of the H3 N-terminal tail enhances the transient unwrapping of DNA at the entry/exit regions. Therefore, we prefer the conclusion that some histone–DNA contacts are disrupted in the tlH3 nucleosome, thus reducing its stability.

13The crystal structure of the nucleosome containing the centromeric H3 variant, CENP-A, revealed that CENP-A structurally differs from the conventional histone H3 at the N-terminal region in the nucleosome, and this CENP-A specific N-terminal structure may cause the unwrapping of DNA at the entry/exit regions in the nucleosome. These facts indicated that the H3 N-terminal region contributes to the wrapping of DNA within the nucleosome. Consistently, in the wt nucleosome, the H3 N-terminal H39-R42 region penetrates into the minor groove of the DNA, and the H3 R40 and R42 side-chains directly bind to the DNA. In the present study, we found that the H39–R42 region of histone H3 becomes “less stable” and/or “poorly organized” upon the deletion of the H3 N-terminal amino acid residues 1–27. Such perturbations of the H3–DNA interactions at the N-terminal region may be partly responsible for the instability of the tlH3 nucleosome (see below).

14The H2B N-terminal tail passes between the two gyres of the DNA superhelix (Fig. 1(B)). The H2B R31 and R33 residues interact with one DNA gyre, and K30 and S32 form hydrogen bonds with the other DNA gyre in the wt nucleosome (Fig. 5). The H2B deletion of residues 1–24 affected the structure of the remaining H2B N-terminal region, in which the electron density of the region preceding the S32 residue was missing in the tlH2B nucleosome (Fig. 5(A)). Concomitantly, the H2B–DNA interactions that involve the K30 and R31 residues were disordered in the tlH2B nucleosome (Fig. 5(B)). The electron density corresponding to the side chain moiety of the H2B R33 residue was also missing in the tlH2B nucleosome (Fig. 5(A) and (B)). Therefore, the absence of the H2B–DNA interaction in the tlH2B nucleosome (K30, R31, and R33 residues) may alter the DNA trajectory near the binding site (Fig. 3(E)), and may affect the nucleosome stability. Single molecule experiments revealed three broad regions of strong histone–DNA interactions on the nucleosome core particle. The strongest region is located at the dyad, and the other two strong regions are ∼ ±50 bp from the dyad. The latter regions include the interaction sites with the H2B N-terminal region, suggesting that this region contributes to enhance the affinity between the DNA and the histone octamer in nucleosomes. These findings are consistent with the previous report that the removal of the H2B N-terminal tail modulates nucleosome positioning and promotes uncatalyzed nucleosome sliding.

15It has been proposed that the H2B N-terminal tail plays a role in the structural polymorphism of nucleosomes through DNA untwisting, which could modify the interactions between the distal H3 N-terminal tail and the DNA at the entry-exit site. Consistently, the electron density around the H3 H39–R42 region was ambiguous in the tlH2B nucleosome structure, as compared to the wt nucleosome (Fig. 4(A) and (C)), despite the facts that the full-length H3 was reconstituted in the tlH2B nucleosome, and the H2B N-terminal tails were located far away from those of H3 (Fig. 1(B)). This may not be the only reason for the low-resolution structure, because the electron density of the corresponding region was clearly observed in the tlH2A nucleosome structure at similar resolution (Fig. 4(D)). Actually, the conformations of the DNA entry/exit regions interacting with the H3 N-terminal regions were influenced similarly in both the tlH2B and tlH3 nucleosomes (Fig. 4(E) and (F)). Altogether, the removal of the H2B N-terminal tail seems to destabilize the extensive interactions between the histone and DNA.

16The N-terminal tail of histone H2A is exposed on the disk face of the nucleosome, and is close to the H2B C-terminal tail (Fig. 1(B)). The deletion of the H2A residues 1–9 did not affect the overall nucleosome structure (Fig. 3(A)) and the remaining N-terminal tail structure (Fig. 6(A)). Thus, we concluded that the impact on the nucleosome structure upon the truncation of the H2A N-terminal residues 1–9 is minimal. Similarly, the H4 deletion of residues 1–15 also did not significantly affect the overall and remaining N-terminal tail structures in the nucleosome (Fig. 6(B)). The H4 N-terminal tail directly binds to the acidic patch of the neighboring nucleosome surface, and is reportedly important for the compact folding of chromatin fibers. The H4 1–19 region, especially H4 residues 14–19, has been demonstrated to be essential for chromatin fiber compaction, but the N-terminal tails of H2A, H2B, and H3 were not fully required for the compaction. In the present structure of the tlH4 nucleosome that lacks H4 residues 1–15, no significant structural differences were found in the remaining N-terminal regions of both H4 histones (Fig. 6(B)), or in the crystal packing contacts with the H2A acidic patch. This is consistent with the previous report that the deletion of the H4 residues 1–13 did not influence array compaction. Therefore, the H4 residues 16–19 may function in chromatin fiber compaction.

17Altered stability in the proteolytically generated tailless nucleosomes was observed by the thermal denaturation assay, but not by the salt dissociation assay. Therefore, we performed a thermal stability assay with nucleosomes containing tlH2A, tlH2B, tlH3, or tlH4, in the presence of SYPRO Orange. SYPRO Orange is a dye that preferentially binds to hydrophobic regions of proteins. Therefore, the binding of SYPRO Orange to native, folded proteins is low, whereas the binding increases when proteins are denatured and the internal hydrophobic regions become exposed to the solvent. In fact, small amounts of SYPRO Orange bound to histones that were properly incorporated into nucleosomes. In contrast, the amount of SYPRO Orange bound to histones drastically increased when the histones were denatured, as a consequence of the thermal disruption of nucleosomes (Fig. 7).

18Under the experimental conditions used in this study, the wt nucleosome was disrupted in a two-step manner (with a biphasic curve), which probably reflects the ordered disassembly of H2A–H2B, followed by H3–H4 (Fig. 7). Ausio et al. previously reported that the wt nucleosome exhibited biphasic denaturation, and the first (74.5 °C) and second (∼80 °C) melting transitions are roughly in accordance with our thermal denaturation profile (Fig. 7). In the nucleosomes containing tlH2A and tlH4, the initial phase of the disruption occurred at slightly higher temperatures than that of the wt nucleosome (Fig. 7), suggesting that the nucleosome stability was not affected by the deletion of H2A residues 1–9 or H4 residues 1–15. Since the H4 residues 16–19, which may be important for chromatin fiber compaction, were still retained in the tlH4 nucleosome, further deletion of H4 N-terminal residues may affect the nucleosome stability. By contrast, in nucleosomes containing tlH2B (lacking residues 1–24) and tlH3 (lacking residues 1–27), the initial phase of disruption clearly occurred at lower temperatures, as compared to the wt nucleosome (Fig. 7). These results suggested that the N-terminal tails of H2B (1–24) and H3 (1–27) make important contributions towards the stability of nucleosomes. The tlH2B nucleosome instability revealed by the thermal stability assay is consistent with the previous observation that the deletion of the H2B 3–22 region confers a Sin (SWI/SNF-independent) phenotype, which accompanies nucleosome instability. The deletion of the H3 N-terminal tail reportedly destabilizes the H2A–H2B dimer within the nucleosome. This is consistent with our observation that the first melting transition of the tlH3 nucleosome (Fig. 7) probably reflects the H2A–H2B disassembly occurring at a lower temperature than that in the wt nucleosome.

19These thermal stability data may be explained by our crystal structures of the tlH2B and tlH3 nucleosomes. The histone–DNA contacts were weakened in the tlH2B and tlH3 nucleosomes (Figs. 4 and 5). In addition, alanine point mutations of the H3 H39, R40, and R42 residues reportedly increase nucleosome mobility. These results suggested that the interactions of these residues with DNA may be important for the stable wrapping of the DNA around the histone octamer.

20Coordinates and structure factors have been deposited in the Protein Data Bank (tlH2A nucleosome, PDB: 3W96; tlH2B nucleosome, PDB: 3W97; tlH3 nucleosome, PDB: 3W98; and tlH4 nucleosome, PDB: 3W99).