Human Centromere Protein B Induces Translational Positioning of Nucleosomes on α-Satellite Sequences*

The human centromere proteins A (CENP-A) and B (CENP-B) are the fundamental centromere components of chromosomes. CENP-A is the centromere-specific histone H3 variant, and CENP-B specifically binds a 17-base pair sequence (the CENP-B box), which appears within every other α-satellite DNA repeat. In the present study, we demonstrated centromere-specific nucleosome formation in vitro with recombinant proteins, including histones H2A, H2B, H4, CENP-A, and the DNA-binding domain of CENP-B. The CENP-A nucleosome wraps 147 base pairs of the α-satellite sequence within its nucleosome core particle, like the canonical H3 nucleosome. Surprisingly, CENP-B binds to nucleosomal DNA when the CENP-B box is wrapped within the nucleosome core particle and induces translational positioning of the nucleosome without affecting its rotational setting. This CENP-B-induced translational positioning only occurs when the CENP-B box sequence is settled in the proper rotational setting with respect to the histone octamer surface. Therefore, CENP-B may be a determinant for translational positioning of the centromere-specific nucleosomes through its binding to the nucleosomal CENP-B box.

In the present study, we reconstituted the centromere-specific nucleosome containing histones H2A, H2B, H4, and CENP-A and studied the role of CENP-B binding in nucleosome formation on ␣-satellite DNA in vitro.

EXPERIMENTAL PROCEDURES
Recombinant Proteins-The recombinant human histones H2A, H2B, H3, H4, and CENP-A were purified as described previously (20). The DNA fragments encoding histones H2A, H2B, H3, and CENP-A were inserted into the pHCE vector (BioLeaders) (46), and that encoding histone H4 was inserted into the pET15b vector (Novagen). The codons of the histone H4 and CENP-A genes were optimized for expression in Escherichia coli cells (20). These recombinant proteins were expressed as N-terminal hexahistidine (His 6 )-tagged proteins. The His 6 -tagged histones and CENP-A were recovered in the insoluble fraction. The pellets were dissolved in 20 mM Tris-HCl buffer (pH 8.0), containing 500 mM NaCl, 5% glycerol, and 6 M urea. The recombinant histones and CENP-A were purified by nickel-nitrilotriacetic acid (Ni 2ϩ -NTA)-agarose (Qiagen) chromatography under denaturing conditions. The H2A/H2B dimer, the H3/H4 tetramer, and the CENP-A/H4 tetramer were reconstituted by dialysis against 20 mM Tris-HCl buffer (pH 8.0), containing 2 M NaCl, 5 mM dithiothreitol, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 5% glycerol. After the NaCl concentration was reduced, the His 6 tags of these recombinant histones and CENP-A were removed by a treatment with thrombin protease. The recombinant human CENP-B-(1-129) was expressed in E. coli cells and was purified in the presence of 6 M urea, as described previously (37).
The ␣-Satellite DNA Fragments-The human ␣-satellite DNA fragment (satellite-4) (20) was amplified from HeLa genomic DNA by the PCR, and was ligated into the pGEM-T easy vector (Promega). The 186-base pair fragment was prepared by HindIII digestion of the plasmid containing the satellite-4 sequence, and its sequence is as follows: 5Ј-CTTGCTAGCAATCTGCAAGTGGATATTTGGACCGCATTG-AGGCCTTCGTTGGAAACGGGATTTCTTCATTTCATGCTAG-ACAGAAGAATTCTCAGTAACTTCTTTGTGCTGTGTGTATTC-AACTCACAGAGTGGAACGTCCCTTTGCACAGAGCAGATTT-GAAACACTCTTTTTGTAGTCGACAAG-3Ј. The CENP-B box sequence is underlined.
The 192-base pair fragment was amplified by PCR with the following primers (Fw, 5Ј-AGC TTG CTA GCA ATC TGC AAG T-3Ј; Rv, 5Ј-AAG CTT GTC GAC TAC AAA AAG AGT G-3Ј). This 192-base pair fragment contains three extra bases at both ends of the 186-base pair fragment. These ␣-satellite DNA fragments were purified by the electroelution method from an agarose gel.
The 12 mutant ␣-satellite DNA fragments were constructed by the PCR mutagenesis method, and were ligated into the pGEM-T easy vector (Promega).
The nucleosomes reconstituted with the recombinant histones and CENP-A were analyzed by nondenaturing 6% polyacrylamide gel electrophoresis in 0.5ϫ TBE buffer (45 mM Tris base, 45 mM boric acid, and 1.25 mM EDTA). The gel (20 ϫ 20 ϫ 0.1 cm) was run at 10 V/cm for 90 min. Bands were visualized by autoradiography and were quantitated with a BAS2500 image analyzer (Fuji).
DNase I Footprinting-The ␣-satellite DNA was labeled by T4 polynucleotide kinase (New England Biolabs) in the presence of [␥-32 P]ATP at the NheI end (Fig. 1B). The reconstituted nucleosomes containing the 32 P-labeled ␣-satellite DNA were dialyzed against 10 mM Tris-HCl buffer (pH 8.0), containing 0.1 mM EDTA, 1 mM 2-mercaptoethanol, and 0.1 M NaCl, for 16 h at 4°C. After the dialysis, the Mg 2ϩ concentration was adjusted to 3 mM, concomitantly with the addition of DNase I. Naked DNA and the CENP-B-(1-129)-DNA complex were treated with 0.2 units of DNase I (Takara, Japan)/g of DNA. Nucleosomes were treated with 0.5 units of DNase I/g of DNA. DNase I reactions were carried out at room temperature for 2 min and were terminated by the addition of EDTA to 25 mM. For the footprinting by the gel-purified method, the nucleosomes and the CENP-B-(1-129)nucleosome complexes were separated by nondenaturing 6% polyacrylamide gel electrophoresis in 0.5ϫ TBE buffer (45 mM Tris base, 45 mM boric acid, and 1.25 mM EDTA), and the DNA fragments were isolated from the gel (47). DNA fragments from these complexes were isolated and analyzed by denaturing 8% polyacrylamide gel electrophoresis.
Micrococcal Nuclease (MNase) Protection and Translational Positioning Assays-Reconstituted nucleosomes, containing 10 g of DNA, were dialyzed against 10 mM Tris-HCl buffer (pH 8.0), containing 0.1 mM EDTA, 1 mM 2-mercaptoethanol, and 0.1 M NaCl, for 16 h at 4°C. After the dialysis, the Ca 2ϩ concentration was adjusted to 1 mM, concomitantly with the addition of MNase. Naked DNA was digested with 2.5 and 1.25 units/ml MNase (Worthington) for 5 min at 22°C. The nucleosomes were digested with 1.5, 0.75, and 0.25 units/ml MNase, and the reactions were terminated by the addition of EDTA to 25 mM. Then the DNA fragments were extracted with phenol/chloroform and were precipitated by ethanol. The purified DNA fragments were resuspended in 100 l of TE buffer, and a 10-l aliquot was used for the labeling reaction by T4 polynucleotide kinase in the presence of [␥-32 P]ATP. Then the 32 P-labeled DNA fragments were analyzed by nondenaturing 8% polyacrylamide gel (20 ϫ 20 cm) electrophoresis.
In the translational positioning assay, a 90-l aliquot of the MNasetreated DNA fragments was subjected to nondenaturing 8% polyacrylamide gel (20 ϫ 40 cm) electrophoresis, and the 147-base pair DNA fragments were extracted from the gel. The extracted 147-base pair fragments were labeled at both ends and were cleaved with EcoRI and DraIII. The resulting 32 P-labeled DNA fragments were analyzed by denaturing 8% gel electrophoresis.

CENP-A Is Properly Incorporated into the Nucleosome Core
Particle with the Human ␣-Satellite Sequence-The nucleosome core particle is composed of a histone octamer (containing two H2A/H2B dimers and an H3/H4 tetramer) and a 147-base pair DNA fragment, which is tightly FIGURE 1. Human CENP-A forms nucleosomes. A, schematic representation of the nucleosome reconstitution assay by the salt dialysis method. B, the human ␣-satellite DNA fragment used in the present study. The end of the DNA fragment close to the CENP-B box is defined as the proximal edge, and the other DNA end is defined as the distal edge. The numbers correspond to the bases from the proximal edge of the 186-base pair DNA fragment. The blue box indicates the CENP-B box sequence, which is located from base pair 44 to 60. The arrows indicate the sites for the HindIII, NheI, EcoRI, and DraIII restriction enzymes. C, nucleosomes were reconstituted with the recombinant human histones and CENP-A by the salt dialysis method and were analyzed by nondenaturing 6% polyacrylamide gel electrophoresis in 0.5ϫ TBE buffer. Lane 1, the 186-base pair ␣-satellite DNA fragment. Lanes 2 and 3, the H3 nucleosomes and the CENP-A nucleosomes, respectively. D, the MNase assay. The reconstituted nucleosomes were treated with MNase, and the resulting DNA fragments were analyzed by nondenaturing 6% PAGE. Lanes 1,4,8,9, 12). CP, the DNA fragments incorporated into the nucleosome core particle. DECEMBER 16, 2005 • VOLUME 280 • NUMBER 50 wrapped around the histone octamer (48,49). This 147-base pair fragment is protected from MNase digestion, and therefore, it can be detected after MNase digestion if the ␣-satellite DNA is properly wrapped around the histone octamer (MNase assay). To test whether CENP-A properly forms nucleosomes, we reconstituted the CENP-A nucleosome with recombinant histones H2A, H2B, H4, and CENP-A in the presence of a 186-base pair human ␣-satellite DNA fragment (Fig. 1,

Nucleosome Positioning Induced by CENP-B
As shown in Fig. 1D, when the CENP-A nucleosomes, formed with the 186-base pair human ␣-satellite DNA fragment, were digested with MNase, a 147-base pair DNA fragment was detected by nondenaturing polyacrylamide gel electrophoresis (lanes 10 and 11). As controls, a 147-base pair DNA fragment was also detected when H3 nucleosomes were digested with MNase (Fig. 1D, lanes 5 and 6), but it was not obvious when the naked DNA fragment was used as a template for this assay (Fig. 1D, lanes 2 and 3). Therefore, these results indicate that the CENP-A nucleosome wraps 147 base pairs of the ␣-satellite sequence within its nucleosome core particle, like the canonical H3 nucleosome.
CENP-B Can Bind to Nucleosomal DNA-Next, we performed the nucleosome reconstitution assay in the presence of the DNA-binding domain of human CENP-B (CENP-B-(1-129)) (32,37,50). CENP-B specifically binds the CENP-B box (a 17-base pair DNA sequence) with this DNA-binding domain. The 186-base pair ␣-satellite DNA fragment used in this experiment contains a CENP-B box, and it is located in the region from base pair 44 to 60 (Fig. 1B). Since the nucleosome core particle contained 147 base pairs of DNA (Fig. 1D), the CENP-B box sequence could be located within the nucleosome core particle, even if the nucleosome was formed on an edge of this ␣-satellite DNA fragment (Fig. 1B).
As shown in Fig. 2A, CENP-B-(1-129) bound to the H3 and CENP-A nucleosomes. The nucleosome-CENP-B-(1-129) complexes migrated more slowly than the nucleosomes without CENP-B-(1-129) on a nondenaturing 6% polyacrylamide gel ( Fig. 2A). A footprint of CENP-B-(1-129) binding to the CENP-B box sequence was detected in both the H3 and CENP-A nucleosomes by DNase I probing (Fig. 2B, lanes 5 and 7, respectively). However, the CENP-B footprint of the nucleosomal DNA was very weak as compared with that of the naked DNA (Fig. 2B, lanes  3, 5, and 7). To ensure that the histones and CENP-B-(1-129) did not dissociate during the DNase I treatment and to confirm that the footprint did not result from two distinct complexes with or without CENP-B-(1-129), we gel-purified the CENP-B-(1-129)-nucleosome complexes and the nucleosomes after DNase I treatment, but the intensity of the CENP-B footprint did not change (Fig. 2C). Therefore, CENP-B-(1-129) specifically binds to the CENP-B box sequence within the nucleosomal DNA, although the footprint of the nucleosomal CENP-B box was not clear, as compared with that of the naked CENP-B box DNA. The CENP-B binding to the nucleosomal DNA may make the CENP-B box susceptible to DNase I, or the binding may not be stable, as compared with that with the naked CENP-B box DNA.
CENP-B Binding Does Not Alter the Rotational Setting of the Nucleosomal DNA-When the nucleosomal DNA is rotationally phased on the histone octamer surface, a 10-base pair periodicity of DNase I-hypersensitive sites is observed (51). In both H3 and CENP-A nucleosomes without CENP-B-(1-129), the 10-base pair periodicity was observed by DNase I probing (Fig. 2B, lanes 4 and 6, respectively), indicating that the ␣-satellite DNA was rotationally phased in both H3 and CENP-A nucleosomes. When the nucleosome-CENP-B-(1-129) complexes were probed by DNase I, the 10-base pair periodicity of the DNase I-hypersensitive sites was not different in the H3 and CENP-A nucleosomes, except around the CENP-B box (Fig. 2B, lanes 5 and 7, respec-tively). These results indicate that the rotational setting of the ␣-satellite nucleosomal DNA is not affected by the CENP-B binding.
CENP-B-(1-129) Binding Induces Translational Positioning of Nucleosomes-Next, we tested whether CENP-B binding affects the translational position of the histone octamer along the ␣-satellite DNA. The H3 and CENP-A nucleosomes were reconstituted with or without CENP-B-(1-129) on the 186-base pair ␣-satellite DNA fragment, and the nucleosomal DNAs were treated with MNase. As shown in Fig. 3A, a 147-base pair protected DNA fragment was still observed when CENP-B-(1-129) bound to the nucleosomal DNA. To map the translational positioning of these nucleosomes, the resulting 147-base pair DNA fragments were excised from nondenaturing polyacrylamide gels. Then the extracted 147-base pair fragments were cleaved by a restriction enzyme, such as EcoRI, and were analyzed by 8% denaturing polyacrylamide gel electrophoresis (Fig. 3B).
These mutant ␣-satellite DNA fragments (192 base pairs) were amplified by polymerase chain reaction and were purified by agarose gel electrophoresis. As shown in Fig. 4B, CENP-B-(1-129) bound to all of these mutant ␣-satellite DNAs, except for the ϪCB ␣-satellite DNA. The CENP-B-(1-129)-DNA complexes showed different mobilities on a nondenaturing 6% polyacrylamide gel (Fig. 4B), probably due to the intrinsic curvature of the ␣-satellite sequence and the local DNA kinks (about 60°) induced by the CENP-B binding (37,53).

CENP-B Binds to CENP-A Nucleosomes Containing Mutant ␣-Satel-
lite DNAs-We tested CENP-B binding to the centromere-specific nucleosome, the CENP-A nucleosome, containing these mutant ␣-sat-ellite DNAs. To do so, the CENP-A nucleosomes were reconstituted with the mutant ␣-satellite DNAs (192 base pairs, Fig. 4A) in the presence or absence of CENP-B- (1-129), and the resulting nucleosomes  1 and 3, respectively), indicating that CENP-B-(1-129) binding to these mutant ␣-satellite nucleosomal DNAs depends on the presence of the CENP-B box sequence.
In order to confirm the CENP-B-(1-129) binding to the mutant ␣-satellite nucleosomes, we performed a Ni 2ϩ -NTA pull-down assay (Fig. 6A). In this assay, the mutant ␣-satellite nucleosomes containing His 6 -tagged histones and CENP-A were reconstituted in the presence of CENP-B-(1-129). CENP-B-(1-129) bound to the nucleosomal CENP-B box DNA was co-precipitated with His 6 -tagged histones and CENP-A by the Ni 2ϩ -NTA-agarose beads (Qiagen) and was detected on an SDS-16% polyacrylamide gel stained with Coomassie Brilliant Blue as a band just above the histone H4 band. CENP-B-(1-129) was successfully co-precipitated with the histones in the presence of the wild-type ␣-satellite DNA (Fig. 6C, lane 2), but was not co-precipitated in the presence of the ϪCB ␣-satellite DNA (Fig. 6B, lane 2). These results indicate that CENP-B-(1-129) bound to the CENP-B box DNA but not to histones and nonspecific DNA. No proteins were detected in control experiments without His 6 -tagged histones (Fig. 6, B and C), indicating that the CENP-B-(1-129) bound to the nucleosome-free CENP-B box DNA was not precipitated.
Consistent with the gel shift analysis, the CENP-B-(1-129) binding was detected with the Ϫ20, Ϫ10, Ϫ9, Ϫ6, Ϫ3, ϩ3, ϩ9, and ϩ10 nucleosomes (Fig. 6, B (lanes 4, 6, 8, 10, and 12) and C (lanes 4, 8, and  10)). However, the CENP-B-(1-129) binding could not be detected with the ϩ6, ϩ20, and dyad nucleosomes (Fig. 6C, lanes 6, 12, and  14), although their interactions with CENP-B-(1-129) were detected in the gel shift analysis (Fig. 5A, lanes 19, 25, and 27). We repeated this entire experiment twice and confirmed that the results were consistent. Therefore, CENP-B may bind to the nucleosomes containing the ϩ6, ϩ20, and dyad mutant ␣-satellite, but its affinity for them could be very low.  4, 7, 10, and 13). CP, the DNA fragments incorporated into the nucleosome core particle. B, mapping of the translational positioning of nucleosomes. The 147-base pair DNA fragments generated by the MNase digestion were excised from the polyacrylamide gel. The extracted 147-base pair DNA fragments were cleaved by the EcoRI restriction enzyme and were analyzed by 8% denaturing polyacrylamide gel electrophoresis. Lane 1, molecular mass markers (10-bp ladder marker). C, translational positioning of the nucleosome core particle. Two major translational positions, A and B, are shown in red and blue circles, respectively.
The Ϫ20 nucleosome may contain the CENP-B box in a linker region, not in the nucleosome core particle, if the nucleosome was located at the distal edge of the DNA (position A). However, part of the CENP-B box of the Ϫ10 nucleosome could be located within the nucleosome core particle, if the nucleosome formed at position A. Furthermore, the entire CENP-B box could be located within the nucleosome core parti-  DECEMBER 16, 2005 • VOLUME 280 • NUMBER 50 cle in the ϩ9 and ϩ10 nucleosomes (Fig. 4A). Since the average pitch of nucleosomal DNA is about 10 base pairs/turn (48,54), the CENP-B box sequences in the Ϫ20, Ϫ10, ϩ9, and ϩ10 nucleosomes may have similar rotational settings to that of the wild-type ␣-satellite DNA with respect to the histone octamer surface. Therefore, the restriction of the nucleosome mobility by CENP-B may occur only when the CENP-B box is located with the proper rotational setting, with respect to the surface of the histone octamer within the nucleosome. CENP-B may be a determinant for the translational positioning of centromere-specific nucleosomes by its functional binding to the nucleosomal CENP-B box sequence.

DISCUSSION
In the present study, we showed that CENP-B has the ability to bind to the CENP-B box sequence located within the nucleosome core particle. This suggests that CENP-B binds to nucleosomal DNA without seriously disrupting the contacts between the histone octamer and DNA. Actually, CENP-B binding did not disrupt the nucleosome. We tested two types of nucleosomes: a centromere-specific CENP-A nucleosome containing histones H2A, H2B, H4, and CENP-A and a canonical H3 nucleosome containing histones H2A, H2B, H3, and H4. Genetic analyses with CENP-A-depleted mouse cells revealed that the absence of CENP-A caused significant dispersion of CENP-B (28), suggesting that CENP-B binding to the centromere may depend on the presence of CENP-A. However, our in vitro analyses did not show any preference for CENP-B binding to the CENP-A nucleosome rather than the H3 nucleosome. Additional factor(s) or an unrevealed mechanism may be involved in the functional links between CENP-A and CENP-B.
According to the crystal structure of CENP-B-(1-129) complexed with CENP-B box DNA, CENP-B-(1-129) directly contacts three essential sites (sites 1-3) within the 17-base pair CENP-B box sequence (37). All of these direct interactions between CENP-B-(1-129) and DNA are formed on only one side of the DNA helix and induce local DNA kinks of about 60° (Fig. 7A). Therefore, we constructed a docking model between the nucleosome core particle and the CENP-B-(1-129) structures (Fig. 7B). In this model, we supposed that the nucleosome is formed at position A, because CENP-B binding induced preferential nucleosome formation at position A. As shown in Fig. 7B, CENP-B-(1-129) could bind to the nucleosomal DNA without serious steric hindrance in this model. This model structure does not contain the DNA kinks induced by CENP-B binding. No significant difference in MNase susceptibility was observed between the nucleosome and the nucleosome-CENP-B-(1-129) complex, indicating that the nucleosome core particle somehow accommodates the CENP-B-induced  1-129). B, translational positions of the CENP-A nucleosomes containing the mutant ␣-satellite DNAs. The 147-base pair DNA fragments generated by the MNase digestion were excised from the polyacrylamide gel. The extracted 147-base pair DNA fragments were cleaved by the EcoRI restriction enzyme and were analyzed by 8% denaturing polyacrylamide gel electrophoresis. Lanes 1,4,9,14,21,26, and 31 indicate molecular mass markers (10-bp ladder marker). The asterisks indicate the bands corresponding to position A, as shown in Fig. 3B.
DNA kinks and maintains nucleosome stability upon CENP-B binding. Further structural studies of the nucleosome-CENP-B complex are required to understand the functional accommodation of CENP-B within the nucleosome core particle.
The CENP-B box sequence reportedly functions as a cis-element for nucleosome assembly on ␣-satellite DNA (45). Consistent with this observation, in the present study, we directly showed that CENP-B binding to the nucleosomal CENP-B box induced translational positioning of the nucleosome core particles. Nucleosome reconstitution experiments with the mutant ␣-satellite DNAs showed that inappropriate localization of the CENP-B box within the nucleosomal ␣-satellite DNA still allowed CENP-B binding to the nucleosomal CENP-B box sequence but did not induce translational positioning of the nucleosome core particle. These facts indicate that the proper rotational setting of the CENP-B box within the nucleosome is essential for the CENP-B-induced translational positioning, which may function in the centromere-specific chromatin formation. The DNase I footprinting experiment showed that the ␣-satellite DNA sequence was rotationally phased in both the H3 and CENP-A nucleosomes, probably due to its intrinsic curvature (55)(56)(57). This intrinsic rotational setting ability of the ␣-satellite sequence is important to determine the CENP-B box rotational setting, which regulates the CENP-B-induced translational positioning of the nucleosome. Therefore, the rotational setting and translational positioning of the centromere-specific nucleosomes are tightly linked in the presence of CENP-B and may be important for the formation of a functional centromere-specific chromatin structure.

Nucleosome Positioning Induced by CENP-B
It has been reported that demethylation of centromeric satellite DNA sequences, accomplished by growing cells in the presence of a DNA methyltransferase inhibitor, resulted in the redistribution of CENP-B (58). Intriguingly, the CENP-B box sequence contains two CpG dinucleotides within its essential sites for CENP-B binding (37). We previously found that CENP-B preferentially binds to an unmethylated CENP-B box, and the DNA-binding ability of CENP-B is reduced nearly to the level of nonspecific binding by CpG methylation at both sites (50). In the present study, we showed that CENP-B binding to the nucleosomal CENP-B box induces the translational positioning of nucleosomes on the ␣-satellite sequence. Therefore, CpG methylation at the CENP-B box may be an important regulator of centromere-specific chromatin formation, through the CENP-B-induced nucleosome positioning on the ␣-satellite DNA repeats.