In this report, we examined the stability of NCPs containing histone variants. Among the various histone H2A variants tested thus far, H2A.Bbd was shown to generate the most flexible NCP structure in the presence of human NAP-I (Fig. ). Interestingly, the NCPs containing H2A.Bbd were more flexible in combination with the histone H3.3 (Fig. and ), which is a histone H3 variant reported to be preferentially incorporated in the active chromatin (1
). These results increase the possibility of the histone variants H2A.Bbd and H3.3 being involved in the formation of the active chromatin structure without the involvement of posttranslational modifications. Interestingly, TAF-I/SET and B23.1, which demonstrate significant similarities in sequence and function to NAP-I and nucleoplasmin, respectively, did not show efficient assembly and disassembly activity of the H2A.Bbd-H2B dimers (Fig. ). Thus, the acidic nature and the histone-binding activity of NAP-I are not sufficient to explain the mechanism of NAP-I-mediated assembly and disassembly of the NCPs.
Several genetic studies have clearly demonstrated that histone variant proteins play important roles during mammalian development and various cellular processes (7
). However, the function of each histone variant in NCPs has not been studied extensively. H2A.Bbd is the most recently identified histone variant and shows the lowest sequence similarity to the canonical histone H2A among the various histone H2A variants known thus far (8
). The exogenously expressed H2A.Bbd is preferentially incorporated at the active chromosome loci and almost excluded from the “Barr Body” that is formed by the inactive X chromosome (8
). Bao et al. demonstrated that the NCPs containing H2A.Bbd are less rigid than canonical NCPs and the docking domain of the H2A.Bbd is responsible for this effect (4
). In agreement with this finding, the NCPs containing H2A.Bbd were exceptionally flexible among the NCPs containing various H2A variants in the presence of NAP-I, as shown in Fig. . DNase I digestion assays of NCPs containing H2A.Bbd (NCP3) in the absence of NAP-I also demonstrated that in addition to the periodic cutting sites with about 10-bp distances, spontaneous cutting sites were generated (Fig. and ). This suggests that H2A.Bbd weakens the interaction not only between the H2A.Bbd-H2B dimers and H3-H4 tetramers but also between the histone octamers and DNA. The exposure of DNA sites is proposed to occur via the spontaneous transient dissociation of short stretches of DNA from the surface of the histone octamer beginning at one end and extending progressively inwards (3
). This exposure is likely to occur more efficiently in the NCPs containing H2A.Bbd than in the canonical NCPs, thereby allowing DNase I to access the DNA in the NCPs containing H2A.Bbd.
When incubated with NAP-I, H3.3, a histone H3 variant, generated a more-flexible nucleosome structure in combination with H2A.Bbd (Fig. and ). Evidence from several reports demonstrated that H3.3 is preferentially deposited at the active chromatin independently of DNA synthesis (1
). It is well established that once H3.3 is deposited at the active chromatin, histone modification enzymes mark the specific amino acids, such as lysine 4 and lysine 9, in order to maintain the active chromatin structure (29
). However, it is unclear whether H3.3 generates a less-rigid chromatin structure than canonical H3 without these modifications. H3.3 and H3 differ from each other with regard to only four amino acids, and three of these amino acids are located at the α2 helix of H3, where the solvent-accessible site are suggested to be located (25
). At this point in time, it is unknown whether these amino acids alter the interaction between H2A.Bbd-H2B dimers and H3.3-H4 tetramers or between NAP-I and H2A.Bbd-H2B dimers in the NCPs. The biological relevance of NCPs containing H2A.Bbd-H2B dimers and an H3.3-H4 tetramer in vivo is an important issue to be addressed.
A previous report demonstrated that yeast NAP-I mediates the exchange of the histone H2A-H2B dimers for variant dimers (40
). Our results clearly demonstrated that NAP-I preferentially mediated the exchange of H2A.Bbd-H2B dimers for H2A-H2B dimers, and this efficient exchange reaction mediated by NAP-I was not observed when the exchange reaction was reversed (Fig. ). This suggests that the disassembly of the dimer from NCPs is a rate-limiting step of the dimer exchange reaction. Since the NCPs containing the H2A-H2B dimers are more stable than those containing H2A.Bbd-H2B dimers, the disassembly of the H2A-H2B dimers by NAP-I is much slower than that of H2A.Bbd-H2B dimers. Assembly of the dimers by NAP-I would be equally efficient for both canonical dimers and dimers containing H2A.Bbd. This assumption explains the efficient exchange of the H2A.Bbd-H2B dimers for H2A-H2B dimers (Fig. ). Since NAP-I alone cannot efficiently remove the H2A-H2B dimers from the NCPs by NAP-I alone, other factors, such as the ATP-dependent chromatin remodeling machineries, could be required for this process. Indeed, ATP-dependent histone exchange complexes (22
) have been shown to mediate the exchange of the dimers. The stability of the NCPs is also presumed to be regulated by posttranslational modifications, including the acetylation, of histones. In fact, the Tip60 acetyltransferase activity and an ATP-dependent chromatin remodeling protein, Domino, were demonstrated to be required for the efficient execution of a dimer exchange reaction (22
). The p300-mediated acetylation of histones in the nucleosome has also been reported to facilitate the transfer of the H2A-H2B dimers from the nucleosome to NAP-I (17
). The effect of histone modifications on the stability of NCPs is, therefore, a matter of concern to be investigated in the future.
NAP-I was originally identified as a factor that facilitates nucleosome formation in vitro (15
). NAP-I is conserved from yeast to humans, although the biological function of NAP-I has not been completely established. Several lines of genetic evidence revealed that NAP-I is involved in the regulation of a distinct set of genes (23
). From these observations, one can conclude that NAP-I, at least in part, is likely to be involved in chromatin remodeling by the disassembly of histone H2A-H2B dimers in vivo. In addition to NAP-I, several acidic histone-binding proteins have been identified. In a manner similar to that of NAP-I, TAF-I/SET and B23.1 bind to histones and transfer them to DNA to assemble the nucleosome (19
). Unlike NAP-I, however, TAF-I/SET and B23.1 did not demonstrate efficient dimer stripping activity (Fig. ). The carboxyl-terminal acidic region of yeast NAP-I was shown to be critical for the stripping of the H2A-H2B dimers and for remodeling of the adenovirus chromatin (19
); however, this region is dispensable for histone binding and nucleosome assembly (11
). Therefore, the C-terminal acidic region may be required to compete with DNA for the removal of the basic proteins from DNA. However, the mechanism of dimer stripping by NAP-I is more complex, because TAF-I/SET, which has a similar, long acidic stretch at its C terminus, did not demonstrate efficient dimer stripping activity. Thus, the acidic region and the other functional domain(s) yet to be identified are important in order for NAP-I to demonstrate its complete histone chaperone activity.