Assembly and Disassembly of Nucleosome Core Particles Containing Histone Variants by Human Nucleosome Assembly Protein I^†

Plasmid DNA.

To express the human histones H2A, H2B, H3, and H4, pET22b-H2A, pET22b-H2B, pET22b-H3, and pET22b-H4, respectively, were used (47). The mouse H3.3 cDNA was amplified from cDNA prepared from NIH 3T3 cells by PCR and subcloned into the NcoI and BamHI sites of pET14b (Novagen). The human H2A.Z, H2A.X, macroH2A1.2, and H2A.Bbd cDNAs were amplified from cDNA prepared from HeLa cells by PCR and subcloned into the NdeI and BamHI sites of pET14b. To prepare the histone fold domain of macroH2A1.2, the original BamHI site in the macroH2A1.2 cDNA was used. The fragment subcloned into the NdeI and BamHI sites of pET14b encodes the histone fold domain (amino acids 1 to 119) of macroH2A1.2. The human NAP-I cDNA was amplified by PCR using cDNA prepared from HeLa cells as a template and subcloned into the NdeI and XhoI sites of pET14b. In this report, the nomenclature NAP-I is used for the human NAP1-like 1 protein. In order to express NAP-I without a hexahistidine tag (His tag), pET21a-NAP-I was constructed by inserting the cDNA encoding NAP-I into the NdeI and BamHI sites of pET21a (Novagen). The sequences of all cDNAs were confirmed by ABI Prism BigDye terminator cycle sequencing (PE Applied Biosystems) using appropriate primers. The sequences of oligonucleotides used are available upon request.

Expression and purification of recombinant proteins.

Recombinant proteins were expressed in BL21(DE3) CodonPlus RIL-pLys S cells (Stratagene). For H2B, H3, H4, and H3.3, cells expressing histones were disrupted by sonication and purified as described previously (26). To purify H2A variants, cells were disrupted by sonication, and the soluble proteins were removed by centrifugation. His-tagged recombinant histones were purified from the insoluble fractions using metal chelating resins (SIGMA) under denaturing condition. Purified His-H2A variant proteins were mixed with purified H2B in 20 mM sodium acetate, pH 5.2, 5 mM β-mercaptoethanol, and 1 mM phenylmethanesulfonyl fluoride (PMSF) containing 8 M urea and then dialyzed against TE buffer (10 mM Tris, pH 7.4, 1 mM EDTA, 5 mM β-mercaptoethanol, and 1 mM PMSF) containing 2 M NaCl for 12 h. To prepare histone H2A-H2B and H2A.Bbd-H2B dimers without the His tag, the refolded dimers (1 mg of total proteins) were dialyzed stepwise with TE buffer containing 1.5, 1.0, 0.5, and 0.1 M NaCl for 3 h at each step and then treated with 3 units of thrombin (Nacalai Tesque) on ice overnight. Histone H2A and its variant proteins contain three additional amino acids (Gly-Ser-His) before the first methionine of the original proteins after thrombin digestion. Thrombin-treated dimers were loaded on a Superdex 200 column (Amersham-Pharmacia) in TE buffer containing 2 M NaCl to remove the His-tag peptide. The histone H2A-H2B and H2A variant-H2B dimers were concentrated by double-stranded DNA-Sepharose column chromatography (Sigma).

Recombinant human NAP-I proteins with or without the His tag were purified from Escherichia coli soluble extracts. His-tagged NAP-I was purified using metal-chelating resins (Sigma) according to the manufacture's protocol. To purify nontagged NAP-I, soluble extracts were fractionated by ammonium sulfate. The soluble proteins in 35% saturation of ammonium sulfate were dialyzed against buffer A (20 mM HEPES-NaOH, pH 7.9, 0.5 mM dithiothreitol [DTT], 0.5 mM PMSF, and 10% glycerol) containing 100 mM NaCl and then loaded on a Mono Q column (1 ml; Amersham-Pharmacia). After extensive washing with the same buffer, the bound proteins were eluted with a linear salt gradient from 100 to 600 mM NaCl. Peak fractions containing NAP-I were dialyzed against buffer A containing 200 mM NaCl and then loaded on a Mini Q column (240 μl; Amersham-Pharmacia). The bound proteins were eluted with a linear salt gradient from 200 to 500 mM NaCl. Peak fractions were collected, and the protein concentration was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) that was stained with Coomassie brilliant blue R250 (CBB).

Nucleosome reconstitution and DNA analyses.

NCPs were assembled with the salt dilution method as described previously (46). Briefly, recombinant histones (2 μg) were mixed with the 5S rRNA gene fragment (2 μg) in 10 μl of 10 mM Tris, pH 7.4, 1 mM EDTA, 0.1 mg/ml bovine serum albumin, 1 mM DTT, and 0.1 mM PMSF in the presence of 2 M NaCl and incubated at 37°C for 10 min. The reaction was serially diluted to 1.5, 1, 0.8, 0.7, 0.6, 0.5, 0.4, 0.25, and 0.2 M NaCl by adding 50 mM HEPES (pH 7.5), 1 mM EDTA, 5 mM DTT, and 0.5 mM PMSF, with 15-min incubations at 30°C for each dilution step. The salt concentration was brought to 0.1 M by adding 100 μl of 10 mM Tris (pH 7.5), 1 mM EDTA, 5 mM DTT, 0.5 mM PMSF, 10% glycerol, and 0.1 mg/ml bovine serum albumin and incubated for 15 min at 30°C. The reconstitutions were confirmed by the nucleoprotein gel analysis.

Nucleoprotein gel analyses, DNase I footprinting, and ExoIII mapping were carried out as described previously (39).

Western blotting.

After the electrophoresis of the NCPs on a nucleoprotein gel, DNA was visualized by ethidium bromide (EtBr) staining. After a brief wash with water, the proteins and DNA on the gel were transferred to a polyvinylidene difluoride (PVDF) membrane at 90 V for 3 h in Tris-glycine buffer (25 mM Tris and 192 mM glycine) containing 20% methanol. Histone H3 and His-tagged H2A proteins were detected by anti-histone H3 (Abcam) and antipolyhistidine (Sigma) antibodies, respectively. To detect H2A.Bbd, antiserum against H2A.Bbd was raised in rabbits by immunizing recombinant full-length His-tagged H2A.Bbd. Recombinant human NAP-I was detected by a monoclonal antibody against NAP-I (a generous gift from A. Kikuchi, Nagoya University).

Reconstitution of NCP containing histone variants.

In order to investigate the stability of NCPs that contained histone variants, recombinant human histones were expressed in bacteria and purified as described in Materials and Methods. Since it has been suggested that the histone variants H3.3 and H2A.Bbd are incorporated into active chromatin, we wanted to clarify the effect of these variant histones on the stability of a nucleosome core particle. We used the expression vector encoding mouse H3.3, the amino acid sequence of which is identical to that of the human H3.3 protein, for the expression of the histone H3.3 protein. Histone H2A and its variants were expressed as N-terminal His-tagged proteins, and the His tag was removed by thrombin digestion (Fig. 1A, lanes 3 and 4). These recombinant histones were first refolded into either H2A-H2B dimers or H3-H4 tetramers and then assembled into histone octamers (Fig. 1A, lanes 7 to 10). As reported previously (4), gel filtration analyses revealed that octamer formation by combination of the H2A.Bbd-H2B dimers with the H3-H4 tetramers was not observed even under high salt concentrations (data not shown). Nevertheless, they were incorporated into the NCPs in the presence of DNA under the condition employed here (see below). The salt dilution method (46) was used to assemble the NCPs on the 196-bp DNA fragment containing the sea urchin 5S rRNA gene with the refolded recombinant histone octamers (Fig. 1B). This 5S rRNA gene-derived sequence was chosen because of its ability to position nucleosomes (10). The assembled NCPs were hereafter designated NCP1, NCP2, NCP3, and NCP4 for NCPs assembled with H2A/H2B/H3/H4, H2A/H2B/H3.3/H4, H2A.Bbd/H2B/H3/H4, and H2A.Bbd/H2B/H3.3/H4, respectively. Nucleoprotein gel analysis revealed that the recombinant histones were assembled into NCPs, and different NCP species, which were designated N1, N2, and N3, were detected (Fig. 1B). Two NCP bands (N1 and N2) were formed in NCP1 and NCP2 due to the presence of a distinct NCP species in which histone octamers occupied distinct positions along the DNA fragments (Fig. 2C) (39). No differences between NCP1 and NCP2 were observed on the nucleoprotein gel, and a single NCP band, N3, was detected in NCP3 and NCP4, which were assembled with histone octamers containing H2A.Bbd (Fig. 1B). Subsequently, the NCPs assembled with recombinant histones on the 5′-end-labeled DNA (Fig. 2A) were subjected to DNase I footprinting analysis (Fig. 2B). DNase I digestion of NCP1 and NCP3 revealed a ladder having the prominent cutting sites that are characteristic for NCPs, whereas DNase I digested naked DNA randomly. It is reported that DNase I digestion of the nonnucleosomal DNA-histone complex also gave rise to a similar ladder (20). To further verify the NCPs assembled with the salt dilution method, the MNase digestion assay was carried out (Fig. 2C). The MNase digestion of NCP1 and NCP3 generated protected DNA fragments of 145 bp and 110 bp, respectively. The size of the DNA fragment generated by the MNase digestion of NCP1 is almost similar to the 147-bp DNA, which is required for the assembly of a mononucleosome. Although a histone octamer containing H2A.Bbd was reported to occupy DNA fragments of approximately 118 bp (4), a slightly shorter DNA fragment of 110 bp was protected from MNase digestion of NCP3 under our experimental conditions (Fig. 2C, bottom panel, lanes 2 to 4). Subsequently, we mapped the positioning of the NCPs along the 5S rRNA gene fragment by digestion with MNase in combination with restriction endonuclease (Fig. 2C). NCP1 and NCP3 were subjected to MNase digestion, and DNA was purified and incubated in the presence of the restriction endonuclease, DraI. As indicated in Fig. 2C, DraI digests the 5S rRNA gene fragment at position +41 and generates DNA fragments of 131 bp and 65 bp (Fig. 2C, left panels). In contrast, DraI digestion of the 5S rRNA gene from NCP1 that was treated with MNase gave rise to mainly two combinations of DNA fragments, 111 bp and 34 bp as well as 96 bp and 49 bp, suggesting that the N1 and N2 nucleosomes, which were assembled with canonical histones, are present between −70 and +75 and between −55 and +90 along the 5S rRNA gene. These findings are harmonious with the results obtained from the ExoIII digestion analysis (see Fig. 4) (39). On the other hand, DraI digestion of the 5S rRNA gene in NCP3 that was treated with MNase mainly gave rise to two DNA fragments of 76 bp and 34 bp, suggesting that NCP3 is located between −35 and +75 along the 5S rRNA gene fragment.

Effect of acidic histone chaperones on the stability of NCPs.

Since histone chaperones are suggested to be responsible for the assembly and disassembly of nucleosomes, we attempted to assess the role of these proteins in the stability of NCPs containing histone variants. We chose three acidic histone chaperones, TAF-I/SET, NAP-I, and B23.1 (Fig. 3A). These proteins have been identified as remodeling factors of adenovirus chromatin, and they show different localization patterns in the cell. TAF-I/SET is present throughout the nucleoplasm, and B23.1 is present in the nucleolus. NAP-I is mainly localized in the cytoplasm but has been demonstrated to shuttle between the cytoplasm and the nucleus (16, 30, 44). The acidic regions of these proteins were shown to be critical for their adenovirus chromatin-remodeling and histone-binding activities (19, 33, 36).

NCPs that were assembled, as shown in Fig. 1, were incubated in either the absence or presence of increasing amounts of acidic histone chaperones, and the reaction was followed by nucleoprotein gel analyses. NCP1 and NCP2, which were assembled with canonical histone octamers and octamers containing H3.3, respectively, were stable even in the presence of excess amounts of TAF-I/SET and B23.1 under our assay conditions (Fig. 3B, lanes 3 to 8). Further, although yeast NAP-I was reported to mediate the nucleosome sliding (40), NAP-I did not affect the stability of NCP1 and NCP2. TAF-I/SET and B23.1 did not significantly affect the structure of NCP3 and NCP4 containing H2A.Bbd (Fig. 3B, lanes 9 to 14). In contrast, the structures of NCP3 and NCP4 were drastically changed after incubation with NAP-I (middle panel of Fig. 3B, lanes 9 to 14). The effect of NAP-I on the structure of NCP4 was slightly greater than that on NCP3 (compare lanes 9 to 11 with lanes 12 to 14 in Fig. 3B). Although TAF-I/SET, NAP-I, and B23.1 bind to histones and transfer them to DNA in a similar manner, only NAP-I induced structural change in NCPs containing H2A.Bbd. The molar ratio of B23.1 relative to DNA used here was lower than that of TAF-I/SET or NAP-I, as shown in Fig. 3B, because B23 presumably functions as a pentamer or decamer in solution (34). However, no significant structural change in NCPs was observed in the presence of B23.1 at the same dose used for Fig. 3B (data not shown). These results prompted us to further analyze the structural change of NCPs containing H2A.Bbd by NAP-I.

Removal of the H2A.Bbd-H2B dimers from NCPs by NAP-I.

In order to investigate the particulars of the NAP-I-induced structural change of the NCPs containing H2A.Bbd, two possibilities were addressed. The first possibility was that NAP-I mediated nucleosome sliding, since rotational positioning of NCPs along a DNA fragment often changes the mobility on the nucleoprotein gel. To test this possibility, we made use of ExoIII mapping analysis to map the positioning of NCPs along the DNA fragment. ExoIII progressively cuts the DNA from the 3′ to the 5′ direction so that when the enzyme reaches the 3′ border of the nucleosome, the digestion of DNA is blocked and strong posing sites are observed. The NCPs were assembled with the 5S rRNA gene fragment in which the 5′ end of the sense strand relative to the direction of transcription was labeled with ³²P and subjected to the ExoIII digestion assay. Consistent with the results shown in Fig. 2C, ExoIII posing sites appeared at positions +75 and +90 and at position +75 when NCP1 and NCP3, respectively, were subjected to ExoIII digestion. If NAP-I mediates nucleosome sliding, novel ExoIII posing sites should appear on incubation with NAP-I. However, as shown in Fig. 4A, the ExoIII digestion patterns of NCP1 and NCP3 that were preincubated without or with NAP-I were not different from each other. The ExoIII posing sites that were observed at positions +75 and +90 for NCP1 and at +75 for NCP3, indicated by bullets in Fig. 4A, were detected regardless of whether NCPs were incubated with or without NAP-I. This suggests that NAP-I did not mediate nucleosome sliding.

Since yeast NAP-I has been reported to transiently remove the H2A-H2B dimers from NCPs, which results in an active exchange of the H2A-H2B dimers (40), we addressed the other possibility that human NAP-I stripped the histone H2A.Bbd-H2B dimers from NCPs. NCPs containing canonical H2A-H2B or H2A.Bbd-H2B dimers (NCP1 and NCP3, respectively) were incubated in the absence or presence of NAP-I, and the reaction was followed by nucleoprotein gel analysis (Fig. 4B). DNA was visualized by EtBr staining, the proteins and DNA fragments were transferred to a PVDF membrane, and histone H3 and H2A.Bbd were detected by Western blotting. When the NCP1 that was assembled with canonical histones was incubated in the absence or presence of increasing amounts of NAP-I, the electrophoretic patterns of DNA and histone H3 did not change significantly (Fig. 4B, top panels). In sharp contrast, as shown in Fig. 3B, NCP3 containing H2A.Bbd was perturbed and a slower-mobility band designated N3* appeared on incubation with increasing amounts of NAP-I. As can be observed in the bottom panel of Fig. 4B, both the N3* and the N3 species contained DNA, H3, and H2A.Bbd. A prominent new band that contained H2A.Bbd appeared on incubation with NAP-I. Since this band did not contain any detectable DNA, it could correspond to either free H2A.Bbd-H2B dimers or a ternary complex having NAP-I. Since H2A.Bbd-H2B dimers cannot enter the gel due to their positive charge, it is likely that the band migrating faster than the NCPs corresponds to a ternary complex having NAP-I. To demonstrate this, NAP-I that was incubated in the absence or presence of free H2A.Bbd-H2B dimers or NCPs was separated on the nucleoprotein gel and analyzed by Western blotting (Fig. 4C). NAP-I alone was distributed throughout the lanes (lanes 1, 5, and 9). This may be because of the possibility that NAP-I alone cannot form a stable conformation under a nondenaturing condition. In fact, it has been reported that yeast NAP-I forms a dimer, and each NAP-I dimer further forms complex oligomers under physiological salt concentrations (28). However, NAP-I incubated with free H2A.Bbd-H2B was concentrated mainly in two bands: one of these bands migrated faster and the other migrated slower than the NCPs. Since these two bands that were detected by an anti-NAP-I antibody also contained H2A.Bbd (Fig. 4C, bottom panel), both correspond to a ternary complex between the H2A.Bbd-H2B dimers and NAP-I. These two different ternary complexes possibly were generated due to the different stoichiometry between the H2A.Bbd-H2B dimers and NAP-I. In addition, the bands migrating faster than the NCPs appeared with incubation with NCP3 and NCP4, which were assembled with histone octamers containing H2A.Bbd. These observations support the idea that the band containing H2A.Bbd and migrating faster than NCPs corresponded to a ternary complex having NAP-I rather than only a free H2A.Bbd-H2B dimer.

Since we prepared the NCPs by the salt dilution method and the assembled NCPs were not purified, it was possible that the free H2A.Bbd-H2B dimers that were not assembled into the NCPs corresponded to the band that migrated faster than the NCP on incubation with NAP-I. However, the same band containing H2A.Bbd appeared when NCP3 purified through a sucrose density gradient was incubated with increasing amounts of NAP-I (see Fig. S1 in the supplemental material). Therefore, it is unlikely that the free H2A.Bbd-H2B dimers that were not assembled into the NCPs cause the band to migrate faster than the NCPs.

We also noted that the ratio of DNA to histone H3 in N3 was similar to that in N3*; however, the amount of H2A.Bbd in N3* was significantly lower than that in N3. Thus, we assumed that N3* is a product that is generated by the removal of one H2A.Bbd-H2B dimer from the NCPs by NAP-I. To test this assumption, we assessed the stoichiometry of histones in N3 and N3* species by using nondenaturing PAGE in combination with SDS-PAGE. The NCPs containing H2A.Bbd incubated with NAP-I were separated by 6% PAGE under nondenaturing conditions, and the lane was cut out and subjected to SDS-PAGE (Fig. 4D). Subsequently, the proteins and DNA were visualized by silver staining. Although an excess of NAP-I was broadly distributed and not detected as a distinct band, the silver-stained gel clearly demonstrated that the H2A.Bbd-H2B dimers were partially stripped on incubation with NAP-I and that the N3* nucleosome appeared simultaneously. The amounts of histones and DNA present in N3 and N3* were quantitatively measured by using NIH Image (Fig. 4D, table). The amounts of the H2A.Bbd-H2B dimers in N3* were distinctly low, and N3* had almost half the amount of dimers of H3 and H4. Therefore, we concluded that N3* contains one H2A.Bbd-H2B dimer and one H3-H4 tetramer. These results revealed that NAP-I mediates removal of the H2A.Bbd-H2B dimers from the NCPs rather than nucleosome sliding.

Since NAP-I removed the H2A.Bbd-H2B dimer from the NCPs, we assumed that this structural change of the NCPs makes the trans-acting factors accessible to nucleosomal DNA. To test this assumption, the NCPs that were preincubated in the presence or absence of NAP-I were subjected to a DNase I digestion assay. The accessibility of DNase I to nucleosomal DNA was much lower than that of naked DNA (Fig. 4E, compare lane 1 with lanes 2 and 4), and the periodic digestion pattern for NCP1 and NCP3 was observed as shown in Fig. 2B. In contrast, the accessibility of DNase I significantly increased when NCP3 was preincubated in the presence of NAP-I (Fig. 4E, lanes 4 and 5). Although distinct changes in the electrophoretic pattern of NCP1 were not observed even in the presence of an excess of NAP-I (Fig. 3 and 4), NAP-I increased the accessibility of DNase I to DNA in NCP1 (Fig. 4D, lanes 2 and 3). Thus, NAP-I may also mediate the transient disassembly of the H2A-H2B dimers from the NCPs, or incubation of NCP1 with NAP-I may result in uncharacterized structural changes of NCPs without removing histone octamers. This could also be true in the cases of TAF-I/SET and B23.1. TAF-I/SET increases the nuclease sensitivity of nucleosomal DNA and stimulates transcription from the chromatin template, whereas significant structural changes of NCPs were not observed on the nucleoprotein gel (Fig. 3B) (12, 38).

Stability of NCPs containing various H2A variants.

It has been reported that a histone variant, H2A.Z, confers rigidity to the nucleosome structure by enhancing the interaction between the H2A.Z-H2B dimers and an H3-H4 tetramer within an NCP (41). On the other hand, H2A.Bbd is suggested to be involved in the formation of a more flexible nucleosome structure than that formed by H2A (Fig. 3 and 4) (4, 13). By examining the stability of NCPs containing various histone variants, using the same biochemical assay system, the effect of histone H2A variants on the stability of NCPs could be investigated in a holistic manner. Therefore, we tested the effect of the histone H2A variants known thus far on the stability of NCPs in the presence of NAP-I. In order to detect the histone H2A variant proteins easily by Western blotting, we used N-terminal His-tagged H2A variant proteins to assemble the NCPs, and His-tagged histone H2A variants were detected by an anti-His-tag antibody. His-tagged H2A, H2A.X, H2A.Z, the histone fold domain of macroH2A1.2 (mH2AN), and H2A.Bbd were expressed in E. coli, purified, and refolded into the dimers with H2B (Fig. 5A); subsequently, histone octamers were prepared. Using these octamers, the NCPs were assembled on the 5S rRNA gene fragment (Fig. 5B). NCPs containing each H2A variant demonstrated differences in electrophoretic mobility in nucleoprotein gels. These differences could be due to the difference in either size or charge among the H2A variants or due to the differential effects of the variants on the position of the octamer along the DNA. The assembled NCPs were incubated in the absence or presence of NAP-I and subjected to nucleoprotein gel analysis and Western blotting with anti-His-tag or anti-histone H3 antibodies (Fig. 5C). As shown in the middle panel of Fig. 5C, each of the His-tagged histone variants was incorporated into the NCPs. The electrophoretic patterns of the NCPs in the absence or presence of NAP-I did not significantly differ from each other except for the NCPs containing H2A.Bbd (lanes 17 to 20). Thus, we concluded that among various H2A variants tested, H2A.Bbd was the most susceptible to dissociation in the presence of NAP-I.

Preferential removal of the H2A.Bbd-H2B dimers from NCPs containing H3.3 rather than those containing canonical H3.

Since the effect of NAP-I on the structure of NCP4 containing H3.3 was slightly more pronounced than that on NCP3 containing canonical H3 (Fig. 3B, middle panel), it is reasonable to hypothesize that the H2A.Bbd-H2B dimers were removed more efficiently from NCP4 than from NCP3. To be more precise, the NCPs assembled with histone octamers containing H2A.Bbd, H2B, H3, and H4 (NCP3) or H2A.Bbd, H2B, H3.3, and H4 (NCP4) were incubated in the absence or presence of increasing amounts of NAP-I and separated on the nucleoprotein gel, and this was followed by EtBr staining and Western blotting with an anti-H2A.Bbd antibody (Fig. 6A). As shown in Fig. 3 and 4, the free H2A.Bbd-H2B dimers appeared on incubation with NAP-I, and NAP-I stripped the H2A.Bbd-H2B dimers more efficiently from NCP4 than from NCP3 (compare lanes 2 to 4 with lanes 6 to 8 in Fig. 6A, bottom panel). Similar results were obtained when the reactions were carried out at 4°C (Fig. 6B). The amounts of H2A.Bbd present at the free dimer position were measured and plotted as a function of increasing amounts of NAP-I (Fig. 6B). As much as 40% and 25% of the H2A.Bbd-H2B dimers were stripped from NCP3 at 37°C and 4°C, respectively, whereas approximately 75% and 50% of the H2A.Bbd-H2B dimers were stripped from NCP4 at 37°C and 4°C, respectively. Further, a new NCP band designated N3** appeared on incubation of NCP4 in the presence of an excess of NAP-I (Fig. 6A, top panel). Since this band contained DNA but not H2A.Bbd and was detected by an anti-H3 antibody (data not shown), it is likely to correspond to the nucleoprotein complexes containing an H3-H4 tetramer and DNA. From these observations, we propose that H2A.Bbd and H3.3 work in combination to make the NCPs flexible.

Reversible assembly and disassembly of the H2A.Bbd-H2B dimers by NAP-I.

Since NAP-I efficiently stripped the H2A.Bbd-H2B dimers from NCPs (Fig. 4 to 6) and yeast NAP-I was reported to mediate the assembly and disassembly of the nucleosome (40), it was assumed that human NAP-I could mediate the reversible assembly and disassembly of the H2A-H2B and H2A.Bbd-H2B dimers. To test this assumption, the His-H2A-H2B or His-H2A.Bbd-H2B dimers that were preincubated with NAP-I were mixed with NCPs assembled with nontagged histones. The NCPs incubated with the His-tagged dimer-NAP-I complexes were analyzed by nucleoprotein gels (Fig. 7A). If NAP-I mediates an exchange of the dimers, the His-tagged dimers would be incorporated into the NCPs and detected at the positions of the NCPs on the nucleoprotein gel. As expected, the nontagged H2A.Bbd-H2B dimers were replaced by the His-H2A-H2B dimers as well as the His-H2A.Bbd-H2B dimers in a NAP-I-dependent manner, and these dimers were detected at the positions corresponding to the N1-, N2-, and N3-NCPs on the nucleoprotein gel (Fig. 7A, lanes 17 to 32). Consistent with the observation shown in Fig. 6, the replacement of the nontagged H2A.Bbd-H2B dimers with the His-tagged dimers in the presence of NAP-I (Fig. 7A, compare lanes 18 to 24 with lanes 27 to 32 of the middle and bottom panels) was higher in NCP4 than in NCP3. In addition, a low but distinct level of the His-tagged H2A-H2B dimers was also detected at the NCP positions when NCP1 and NCP2 were incubated with the His-H2A-H2B dimer-NAP-I complexes (Fig. 7A, lanes 3 to 5 and 11 to 13), suggesting that NAP-I mediates the disassembly and assembly of the H2A-H2B dimers in canonical NCPs. As indicated by the bullets in the top and middle panels of Fig. 7A, the bands having slow mobility appeared during the exchange of H2A-H2B or H2A.Bbd-H2B dimers with His-H2A-H2B dimers. These bands contained DNA and were recognized by an anti-His-tag antibody, thus suggesting that the bands are intermediate NCPs containing one H2A-H2B dimer (with or without His tag) and an H3-H4 tetramer. During the disassembly of the H2A-H2B dimers, the DNA in the NCPs could be transiently exposed, and this structural change in the NCPs could enable the trans-acting factors to access the DNA. Indeed, NAP-I increased the sensitivity of DNase I to nucleosomal DNA in NCP1 (Fig. 4E). It should also be noted that the H2A-H2B dimers in NCP1 and NCP2 were not efficiently replaced by the H2A.Bbd-H2B dimers under these experimental conditions (Fig. 7A, lanes 6 to 8 and 14 to 16), whereas the H2A.Bbd-H2B dimers in NCP3 and NCP4 were efficiently replaced by the His-H2A-H2B dimers in the presence of an excess of NAP-I (Fig. 7A, lanes 19 to 21 and 27 to 29). Once the H2A.Bbd-H2B dimers were replaced by the H2A-H2B dimers, the efficiency of the NAP-I-mediated dimer exchange was significantly decreased. Therefore, the free H2A.Bbd-H2B dimer-NAP-I complexes were accumulated when the H2A.Bbd-H2B dimers were replaced by the H2A-H2B dimers (Fig. 7A, bottom panel, lanes 19 to 21 and 27 to 29). Hence, it is suggested that NAP-I preferentially mediates the replacement of thermodynamically unstable dimers with stable dimers. In order to verify the exchange reaction mediated by NAP-I, we performed the dimer exchange reaction of NCPs containing His-tagged H2A.Bbd-H2B dimers with nontagged H2A-H2B dimers or H2A.Bbd-H2B dimers, and the His-tagged dimers that were removed by NAP-I were quantitatively analyzed (Fig. 7B). Although excess amounts of NAP-I resulted in the efficient removal of the His-tagged H2A.Bbd-H2B dimers from NCP4 in the absence of the free dimers (Fig. 7B, lane 20), the His-tagged dimers were more efficiently accumulated in the presence of the nontagged H2A-H2B dimers than in the absence or presence of nontagged H2A.Bbd-H2B dimers (bottom graphs in Fig. 7B). The replacement of the His-tagged H2A.Bbd-H2B dimers with the nontagged H2A-H2B dimers were evidenced by the appearance of the N1 and N2 nucleosome species on the nucleoprotein gel stained with EtBr (Fig. 7B, top panel, lanes 6 to 10 and 21 to 25). These observations support the idea that NAP-I preferentially mediates the exchange of unstable dimers in NCPs for stable dimers.