The Histone Chaperone Anti-silencing Function 1 Stimulates the Acetylation of Newly Synthesized Histone H3 in S-phase*

Anti-silencing function 1 (Asf1) is a highly conserved chaperone of histones H3/H4 that assembles or disassembles chromatin during transcription, replication, and repair. We have found that budding yeast lacking Asf1 has greatly reduced levels of histone H3 acetylated at lysine 9. Lysine 9 is acetylated on newly synthesized budding yeast histone H3 prior to its assembly onto newly replicated DNA. Accordingly, we found that the vast majority of H3 Lys-9 acetylation peaked in S-phase, and this S-phase peak of H3 lysine 9 acetylation was absent in yeast lacking Asf1. By contrast, deletion of ASF1 has no effect on the S-phase specific peak of H4 lysine 12 acetylation; another modification carried by newly synthesized histones prior to chromatin assembly. We show that Gcn5 is the histone acetyltransferase responsible for the S-phase-specific peak of H3 lysine 9 acetylation. Strikingly, overexpression of Asf1 leads to greatly increased levels of H3 on acetylation on lysine 56 and Gcn5-dependent acetylation on lysine 9. Analysis of a panel of Asf1 mutations that modulate the ability of Asf1 to bind to histones H3/H4 demonstrates that the histone binding activity of Asf1 is required for the acetylation of Lys-9 and Lys-56 on newly synthesized H3. These results demonstrate that Asf1 does not affect the stability of the newly synthesized histones per se, but instead histone binding by Asf1 promotes the efficient acetylation of specific residues of newly synthesized histone H3.

The eukaryotic genome is packaged into a nucleoprotein structure known as chromatin. The basic repeating unit of chromatin, the nucleosome, is made up of 147 base pairs of DNA wrapped around a histone octamer (two molecules each of histone proteins H2A, H2B, H3, and H4) (1). Chromatin provides a formidable obstacle to the cellular machinery gaining access to the DNA. Indeed, it is becoming apparent that chromatin is a highly dynamic structure that tightly regulates all nuclear processes that use DNA as a substrate; including transcription, DNA replication, repair, and recombination (2)(3)(4). Thus, the mechanisms by which chromatin structures are made and modified are fundamental questions of broad interest.
The entire eukaryotic genome is assembled into chromatin following DNA replication, and this occurs in a stepwise manner; a tetramer of histones H3/H4 is deposited first followed by two dimers of H2A/H2B on the outside of the tetramer. Histone-binding proteins, termed histone chaperones, mediate chromatin assembly. It is known that when this process is coupled to DNA replication in vitro, it is mediated by the histone H3/H4 chaperone chromatin assembly factor 1 (CAF-1) 2 (5) together with another H3/H4 chaperone termed anti-silencing function 1 (Asf1) (6). CAF-1 and Asf1 have also been shown to localize to the sites of DNA replication (7,8), supporting a role in assembling chromatin following DNA replication in vivo. The fact that CAF-1 and Asf1 co-purify with the replicationspecific histone variant H3.1 from HeLa cells is further evidence that Asf1 and CAF-1 function together to assemble newly replicated DNA into chromatin (9).
To generate sufficient histones to package both newly replicated daughter genomes into chromatin, the cell produces large amounts of newly synthesized histones at every S-phase. The N-terminal tails of the newly synthesized histones H3 and H4 are characteristically modified on specific residues prior to their deposition onto DNA and are termed deposition-specific modifications. For example, new histone H4 is generally acetylated on lysines 5 and 12 (10), but following deposition onto the DNA, the newly synthesized H4 is deacetylated over the next 30 -60 min (11), which is required for proper chromatin maturation (12). Newly synthesized histone H3 is also acetylated, although the target residues and pattern of acetylation vary between species. In Drosophila, lysines 14 and 23 are the predominant sites of acetylation, whereas in Tetrahymena, lysines 9 and 14 are preferred (10). In budding yeast, new H3 is mostly monoacetylated on lysine 9, although lysines 9, 14, 23, and 27 all show some acetylation (13). It is currently unknown whether the acetylation marks in the N-terminal tails of the newly synthesized histone H3 are removed rapidly following chromatin assembly, as is the case for H4. The available evidence indicates that acetylation of histone H4 lysines at positions 5 and 12 is not essential for chromatin assembly, although acetylation of either lysines 5, 8, or 12 seems to be required for chromatin assembly in yeast when the N-terminal tail of histone H3 is also deleted (14). Despite its requirement for chromatin assembly, the exact purpose of the acetylation of the N-terminal tails of the newly synthesized histones is unclear.
Recent evidence also indicates that lysine 56 in the first ␣ helix of H3 is acetylated on newly synthesized histone H3 (15). This modification is predicted to allow a looser interaction of the histones with DNA, suggesting that it may contribute to chromatin assembly (15,16). Interestingly, recent work has indicated that deletion of the gene encoding the histone chaperone Asf1 greatly reduces the levels of acetylation of H3 on lysine 56 (17,18). Similarly, our previous work had found that yeast lacking the histone chaperone Asf1 has greatly reduced levels of acetylation on lysine 9 of histone H3 (19).
To understand how the histone chaperone Asf1 is leading to normal levels of acetylation on lysines 9 and 56 of histone H3, we further investigated the relationship between Asf1 and histone H3 acetylation. We show that Asf1 is not required for stabilization of the newly synthesized histones per se, but instead Asf1 promotes the Gcn5-mediated S-phase peak of acetylation of lysine 9 on newly synthesized histone H3. Consistent with this result, overexpression of Asf1 results in greatly increased acetylation of lysines 9 and 56. The role of Asf1 in modulating acetylation of nascent H3 is dependent on its ability to bind to histones H3/H4. As such, we propose that Asf1 properly presents the newly synthesized histone H3 for acetylation by Gcn5 and the histone acetyltransferase (HAT) mediating Lys-56 acetylation prior to chromatin assembly.

MATERIALS AND METHODS
Yeast Strains and Growth-All strains were made using standard yeast molecular genetics techniques and are listed in Table 1. All yeast growth was in yeast extract peptone supplemented with 2% dextrose or galactose if stated as such in the figure legend. All analyses were performed in the logarithmic phase of growth. URA3 plasmids were maintained by growth in minimal medium lacking uracil. Cell cycle arrest of BAR1-deleted strains was achieved by addition of ␣-factor to a final concentration of 15 nM for multiple hours. Arrest was assessed by visual inspection at the time of the experiment and confirmed by flow cytometry analysis of propidium iodide strained cells. Release from arrest was achieved by washing the cells and resuspending in fresh medium containing Pronase.

RESULTS
We have previously shown that the steady-state acetylation level on H3 lysine 9, but not on the other residues in the N-terminal tails of H3 and H4, was greatly reduced in asynchronously growing yeast deleted for the gene encoding the histone chaperone Asf1 (19) (Fig. 1A). This reduced acetylation on histone H3 Lys-9 upon deletion of ASF1 was observed in multiple independently derived strains, indicating that it was not because of a second site mutation in the strain background (data not shown). Histone acetylation on lysine 9 can be divided into two types: (a) deposition-specific acetylation that occurs on nascent newly synthesized histones prior to chromatin assembly; and (b) acetylation on chromatin, such as during the processes of transcriptional regulation. To understand whether one or both types of lysine 9 acetylation were reduced in the absence of Asf1, we examined H3 lysine 9 acetylation through the cell cycle. We arrested yeast deleted for ASF1 (asf1⌬) and wild type yeast in G 1 phase of the cell cycle with ␣-factor, then released them from the arrest, and took samples through the subsequent cell cycle (Fig. 1A). We assigned the approximate phase in the cell cycle from flow cytometry analysis of DNA content (Fig. 1A). Upon Western blotting total cell extracts for acetylated H3 lysine 9 through the cell cycle, we observed a striking peak of acetylation correlating with S-phase in the wild type cells (Fig.  1A), which presumably reflected the deposition-specific acetylation of H3 Lys-9. This result shows that that the vast majority of H3 Lys-9 acetylation in the cell occurs in an S-phase-specific peak ( Fig. 1). This pattern closely resembles that of acetylation of lysine 56 on newly synthesized H3 prior to chromatin assembly (15), suggesting that the S-phase peak of H3 Lys-9 acetylation that we observed also reflects the acetylation of newly synthesized H3. The fact that the H3 lysine 9 acetylation levels go down after S-phase indicates for the first time that the deposition-specific acetylation on the N-terminal tail of H3 is removed following replication-dependent chromatin assembly, as is the case for the deposition-specific acetylation marks on the N terminus of histone H4 (11). In contrast to the wild type cells, the peak of S-phase-specific H3 lysine 9 acetylation was absent from the asf1⌬ cells (Fig. 1A). The dependence on Asf1 for lysine 9 acetylation on histone H3 prior to chromatin assembly (Fig. 1) is reminiscent of the requirement of Asf1 acetylation of lysine 56 on newly synthesized histone H3 (17,18). Taken together, these data indicate that Asf1 promotes (or stabilizes) the acetylation of the newly synthesized histones.
To investigate whether Asf1 also influences the acetylation of H3 lysine 9 on chromatin, we performed chromatin immunoprecipitation analysis at the promoters and open reading frames of genomic regions that are known to be enriched in lysine 9 acetylation (20). We found no difference in H3 lysine 9 acetylation levels on chromatin in wild type and asf1⌬ cells (supplemental Fig. 1). The low level of H3 Lys-9 acetylation that persists in asf1 mutant cells may reflect the acetylation on chromatin (Fig. 1A). These data indicate that whereas Asf1 promotes Lys-9 acetylation of the nascent H3, Asf1 does not influence H3 Lys-9 acetylation on chromatin.
To investigate the consequence of the loss of acetylation on newly synthesized histone H3, we combined the asf1⌬ with strains lacking either the H3 or H4 N-terminal tail. We found that yeast lacking both the H4 N-terminal tail and ASF1 was inviable, whereas yeast lacking both ASF1 and the N-terminal tail of histone H3 was viable (Fig. 1B). Although it is possible that the inviability is because of other roles of Asf1 in the cell, this result suggests that loss of H3 lysine 9 and lysine 56 acetylation together with loss of the acetylatable lysines on histone H4 may result in a level of acetylation on newly synthesized histones that falls below the threshold required for chromatin assembly in yeast (14).
There are two possibilities that could explain the reduced level of the deposition-specific acetylations on histone H3 (lysines 9 and 56) in asf1 mutants. First, the ability of Asf1 to act as a histone chaperone may stabilize the soluble pool of histones H3/H4, by either reducing its turn over via degradation or by reducing its rate of assembly onto chromatin. Second, Asf1 may promote the ability of the HATs to perform acetylation on newly synthesized H3. To differentiate between these two possibilities, we examined the deposition-specific acetylation on histone H4. If Asf1 is stabilizing H3 it would also be expected to stabilize H4, as H3 and H4 always exist as heterodimers. We had previously shown that the levels of acetylation of H4 lysines 5 and 12 are not altered in yeast lacking ASF1 (19). However, these analyses examined histone acetylation in asynchronous FIGURE 1. Asf1 is required for the S-phase peak of H3 lysine 9 acetylation. A, acetylation of H3 Lys-9 was examined in WT (JKT0010) and asf1⌬ (JKT0018) yeast. Asy refers to asynchronous cells, whereas mins refers to the number of minutes following release from ␣-factor. Equal amounts of whole cell extracts in the specified part of the cell cycle, determined by the flow cytometry analysis of DNA content on the right, were electrophoresed on 15% SDS gels and transferred to nitrocellulose. Wild type asynchronous samples were loaded onto each gel as a comparison. The membranes were probed with antibodies specific to H3 and acetylated H3 Lys-9. B, schematic shown on the left indicates the locations on plates of strains included in the analysis. H4Tail⌬pASF1(MAY0023), H4Tail⌬asf1⌬pASF1(MAY0009), H3Tail⌬pASF1(MAY0024), and H3Tail⌬asf1⌬pASF1(MAY0010) strains were streaked on medium lacking uracil (ϪUra) as a control for growth and on 5Ј-fluoroorotic acid (5ЈFOA) to test for viability. All strains contain a plasmid that has a copy of the ASF1 gene on a URA3-based plasmid. 5ЈFOA was used to counter-select against the URA3 plasmid.
cultures of cells, and it was unknown whether the transcriptionspecific acetylation may be masking a defect in the depositionspecific acetylation of histone H4. To investigate this possibility, we examined H4 Lys-12 acetylation through the cell cycle to determine whether there is an S-phase-specific peak (Fig. 2), as seen for H3 Lys-9 acetylation (Fig. 1A). We observed a clear S-phase-specific peak of H4 Lys-12 acetylation in wild type cells, indicating that the deposition-specific acetylation of H4 Lys-12 is the most abundant form of this modification in the cell. Importantly, the S-phase-specific peak of H4 Lys-12 acety-lation was also present in asf1 mutant cells, albeit slightly delayed because asf1 mutants had an elongated S-phase (supplemental Fig. 2). Because the deposition-specific acetylation of H3 but not H4 is affected in asf1 mutant cells, and because H3 and H4 are always tightly associated as heterodimers in the cell, this result indicates that Asf1 does not affect histone acetylation indirectly by stabilizing soluble histones H3/H4. Asf1 itself does not have HAT activity (data not shown) indicating that a distinct protein mediates H3 Lys-9 and Lys-56 acetylation with HAT activity. To identify which HAT is responsible for the S-phase peak of acetylation on lysine 9 on newly synthesized H3, we examined yeast deleted for GCN5, HAT1, SPT10, or SAS3 encoding HATs and the YNG2 component of the NuA4 HAT complex (data not shown). When we examined yeast deleted for GCN5 (gcn5⌬), we observed a complete absence of the S-phase peak of acetylation of H3 Lys-9 but not H4 Lys-12 (Fig. 3A). This result indicates that Gcn5 is the HAT for the deposition-specific acetylation of soluble histone H3 in vivo. To rule out an indirect affect of ASF1 deletion on Gcn5 expression or stability we compared Gcn5 protein and transcript levels in asf1⌬ and wild type yeast. We found no significant changes in the transcript levels of any known HAT or histone deacetylase upon the deletion of ASF1 (21). Furthermore, the steady-state levels of Gcn5 protein are indistinguishable between wild type and asf1 mutant yeast (Fig. 3B). Gcn5 however is not the HAT responsible for the S-phase peak of acetylation of histone H3 Lys-56 (supplemental Fig. 3). As such, our data indicate that Asf1 may present the newly synthesized histone H3 to Gcn5 or promote the activity of Gcn5 toward lysine 9.
We were interested to investigate whether any of the mutant phenotypes of yeast deleted for ASF1 (6) may be a consequence of the absence of Gcn5-mediated deposition-specific acetylation of lysine 9 of H3. We found that yeast deleted for GCN5 or ASF1 were equally sensitive to the radiomimetic agent bleomycin that resulted in double-strand DNA damage; the gcn5⌬asf1⌬ strain was even more sensitive to bleomycin than either single mutation (Fig. 3C). The growth defect of the gcn5⌬asf1⌬ strain was also more apparent than that of yeast deleted for either GCN5 or ASF1, as seen by the smaller colony size on rich medium (Fig. 3C). These results indicate that Asf1 and Gcn5 have non-overlapping roles in the cell that are not related to the requirement of both proteins for acetylation of H3 Lys-9.
To further investigate the relationship between Asf1 and H3 acetylation, we asked whether excess Asf1 could promote H3 FIGURE 2. The S-phase peak of H4 Lys-12 acetylation is not affected by deletion of ASF1. Acetylation of H4 Lys-12 was examined in WT (JKT0010) and asf1⌬ (JKT0018) yeast, using the same samples generated as described in the legend to Fig. 1. The membranes were probed with antibodies specific to H3 and acetylated H4 Lys-12. . Gcn5 is the HAT that mediates the S-phase peak of H3 lysine 9 acetylation. A, deletion of GCN5 results in the loss of the S-phase-specific peak of H3 Lys-9 acetylation. Strain gcn5⌬ (MAY0038) was analyzed as described in the legend to Fig. 1, followed by Western blotting for H3, acetylated H3 Lys-9, and acetylated H4 Lys-12. B, Asf1 does not affect the levels of Gcn5 protein in the cell. Total protein extracts from strain WT (Z1466) and asf1⌬ (JCY0011) containing Gcn5-9 Myc were Western blotted for anti-Myc. A portion of the Amido Black staining of the total proteins on the nitrocellulose from the same Western blot is shown as a control to ensure equivalent loading. C, Asf1 and Gcn5 have nonredundant functions for growth and DNA damage resistance. Strains WT (JKT0010), asf1⌬ (JKT0018), gcn5⌬ (MAY0038), and asf1⌬gcn5⌬ (MAY0014) were subjected to 10-fold serial dilution analysis onto control plates (YPD) or plates containing the indicated concentration of bleomycin (BLM). Following 2 days of growth at 30°C, the plates were photographed and the images shown.
acetylation. Although deletion of ASF1 reduced H3 Lys-9 acetylation, we found that the overexpression of Asf1 in yeast greatly increased the levels of acetylated H3 Lys-9 but had no effect on H4 acetylation (Fig. 4A). Notably, overexpression of Asf1 did not increase levels of H3 Lys-9 acetylation on chromatin (supplemental Fig. 1). Similarly, although deletion of ASF1 reduced H3 Lys-56 acetylation, the overexpression of Asf1 greatly increased the levels of acetylated H3 Lys-56 in the cell (Fig. 4B). The increased Lys-9 acetylation was not dependent on Gcn5, suggesting that overexpression of Asf1 caused additional HATs to acetylate this residue (Fig.  4C). Because acetylation on H4 did not increase upon overexpression of Asf1, these results suggest that the increased H3 acetylation was not because of the excess Asf1 stabilizing the pool of soluble acetylated histones H3/H4. Instead, these data are consistent with wild type levels of Asf1 normally being limited in the cell for the optimal presentation of soluble histone H3 to the HATs, such that overexpression of Asf1 results in more H3 Lys-9 and Lys-56 acetylation.
To gain further insight into the mechanism whereby Asf1 promotes acetylation of the free histones on lysines 9 and 56 of H3, we tested mutations in the histone-binding interface of Asf1. We have recently solved the 1.7 Å crystal structure of the Asf1/H3/H4 complex and from this structure we identified mutations that disrupt the interaction between Asf1 and H3/H4 in yeast and in vitro (22). In general, our conservative single amino acid substitutions (S48R, V109M, V146L) reduced the Asf1-H3/H4 interaction to a lesser extent than our double amino acids substitutions or the V94R disruptive mutation (22). Therefore, we tested the effect of these Asf1 substitutions on the ability of Asf1 to promote acetylation of H3 lysines 9 and 56 (Fig. 5A). We found that mutations in either the H3 or H4 binding interface of Asf1 resulted in reduced H3 acetylation (Fig. 5, A and B). Furthermore, although single amino acid substitutions that had reduced H3/H4 binding only had a subtle effect on H3 acetylation, the V94R, and double amino acid substitutions in Asf1 that had a profound effect on H3/H4 binding also had a profound effect on H3 Lys-9 and Lys-56 acetylation (Fig. 5A). These results indicate that the ability of Asf1 to influence the acetylation of Lys-9 and Lys-56 on free histones is intimately related to its histone binding activity.

DISCUSSION
Asf1 Is Likely to Present the Newly Synthesized Free Histone H3 to the HATs-Given that Asf1 only affects acetylation of soluble H3, not soluble H4 or chromatin-bound histones ( Fig. 2 and supplemental Fig. 1), we propose that Asf1 delivers or presents the histones in an appropriate manner to the HATs for H3 Lys-9 and Lys-56 acetylation. Previous evidence has shown that a complex containing Gcn5 purifies with newly synthesized histones, and this was proposed to be the HAT for S-phase-specific acetylation of H3 (23). Consistent with this prediction, we show that the S-phase peak of histone H3 Lys-9 acetylation does not occur in gcn5 mutants (Fig. 3), demonstrating that Gcn5 is the HAT for acetylation of Lys-9 on newly synthesized histone H3 in the cell. It is possible that Asf1 regulates the activity of Gcn5 via a direct interaction. However, no direct interaction between Asf1 and Gcn5 has been reported so far. Asf1 has been reported to bind to the bromodomain-containing proteins Brd1 and Brd2 in yeast and Brahma in Drosophila (24,25). As Gcn5 does contain a bromodomain (26), it may be a potential binding partner of Asf1 that merits further investigation in the future. Notably, no detectable HAT activities toward full-length H3 copurified with Asf1 from yeast cells, although an H4-specific has the ASF1 gene under the pGAL1 promoter with a V5:His 6 epitope tag. WT (MAY0025) has the ASF1 gene under the endogenous promoter and a His 6 epitope. Strains described were grown in 2% galactose, and total cell extracts were Western blotted to detect the His epitopes on Asf1. The slight increase in size of the O/X Asf1 is because of the V5 epitope not found on WT. At the bottom is shown a lighter exposure of the same blot with the Asf1 in the O/X Asf1 in the linear range of detection. Right panel, Western blot analysis of the same protein extracts used above, probing for histone H4, total H4 acetylation, H3, and histone H3 Lys-9 and Lys-14 acetylation levels. B, overexpression of Asf1 leads to increased levels of H3 Lys-56 acetylation. Total extracts from the strains used in A, with the addition of an asf1⌬ (JKT0018) strain, were Western blotted for histone H3 and for H3 acetylated on Lys-56. C, Gcn5 is not required for the increased H3 Lys-9 acetylation that results from overexpression of Asf1. WT (JCY0025), sas3⌬ (JCY0026), gcn5⌬ (JCY0027), and spt10⌬ (JCY0028) strains containing the pGAL1Asf1 plasmid were grown in 2% glucose or 2% galactose to induce the overexpression of Asf1 O/X Asf1. Total protein extracts were Western blotted to determine whether any of these HATs were required for the increased acetylation on Lys-9 of H3 that resulted from the overexpression of Asf1. Western blotting for H3 serves as a normalization control for loading differences.
HAT activity (presumably the SAS complex) did co-purify with Asf1 (data not shown). As such, if there is an interaction between Asf1 and Gcn5, it must be very transient. Although Spt10 is the HAT required for acetylation of H3 Lys-56 on chromatin (16), the identity of the HAT required for acetylation of Lys-56 on the newly synthesized histones prior to chromatin assembly is unclear, but it is not Spt10, Sas3, or Gcn5 (Fig. 4C). The only HAT known to associate with Asf1 to date is the SAS complex, composed of Sas2, Sas4, and Sas5 (27). The SAS complex can acetylate the N-terminal tails of free histones H3 and H4; although somewhat surprisingly, the SAS complex cannot acetylate histones bound to Asf1 (28). Although the HAT for acetylation of soluble Lys-56 is unknown, the SAS complex is an unlikely candidate because inactivation of this complex does not cause sensitivity to DNA damaging agents (27) as is the case for yeast unable to acetylate H3 Lys-56 (15).
It is possible that the binding of histones H3/H4 to Asf1 may induce a conformational change in histone H3 that makes it a more favorable target for acetylation by Gcn5 and the Lys-56 HAT. Because H3 Lys-9 and Lys-56 are not included in the structure of the Asf1⅐H3/H4 complex (22), we do not yet know whether binding to Asf1 induces a conformational change toward the N-terminal region of H3 that may improve the fit into the active site of Gcn5 and Lys-56 HAT. However, this is a possibility, as we observed numerous conformational changes throughout histones H3 and H4 upon binding to Asf1 as compared with their conformation in the nucleosome structure (29). Furthermore, we do not think that the influence of Asf1 on H3 acetylation levels could be because of any potential role for Asf1 in shielding histone H3 from deacetylation following its deposition onto the DNA because the crystal structure of Asf1 bound to H3/H4 (22) indicates that Asf1 must be removed from histones H3/H4 to enable interaction with histones H2A/H2B and the DNA within the nucleosome. Our biophysical and structural analyses have clearly shown that the binding of Asf1 to the histone H3/H4 heterodimer envelops the surface of H3 that would otherwise mediate formation of the H3/H4 heterotetramer (30). Upon deletion of ASF1, this H3-H3 interaction interface of the H3/H4 heterodimer is no longer shielded, and it is likely that the soluble H3/H4 in the cell would associate into H3/H4 heterotetramers prior to chromatin assembly. H3/H4 heterotetramerization in the absence of Asf1 may itself induce conformational changes in the soluble histone H3 that may in turn reduce its ability to be acetylated. This will be difficult to determine until a molecular structure of the free H3/H4 heterotetramer is solved. Alternatively, the proposed heterotetramerization of H3/H4 in the absence of Asf1 may present new binding sites for proteins that block access to the HATs of soluble H3.
The location and pattern of residues on histone H3 that are acetylated in a deposition-specific manner vary largely between different species (10,13). Toward differentiating between some of the possible models above, it would be interesting to determine whether the acetylation of the different deposition-related H3 residues is reduced in other organisms lacking Asf1 function. This may provide insight into whether Asf1 is inducing a conformational change in the newly synthesized histones upon binding to them, or whether the presentation of H3/H4 as a heterodimer by Asf1 make them more favorable for acetylation.
A final possibility is that Asf1 may influence the cellular or sub-cellular localization of the soluble histones, which may affect their delivery to the HATs. It is interesting to note that HAT1, which acetylates Lys-5 and Lys-12 on newly synthesized soluble histone H4, was originally termed the cytoplasmic HAT (31,32), although it is now known to be mostly nuclear (33). However, it is still a matter of debate as to where the acetylation of soluble histones occurs, and there is evidence that H4 is acetylated in the cytoplasm (34). Although Asf1 is not essential for the nuclear import of histones H3/H4 (given the viability of asf1⌬ mutants), it is possible that the absence of Asf1 leads to H3/H4 spending more time in the cytoplasm where they have access to the H4 HAT and less accessibility to the nuclear H3 HATs. Notably, we have not observed increased cytoplasmic staining of green fluorescent protein-tagged histones in yeast deleted for asf1 in budding yeast (data not shown); although it is possible that any cytoplasmic signal would be overpowered by FIGURE 5. The interaction between Asf1 and histones H3/H4 is required for the stimulation of acetylation of histone H3 Lys-9 and Lys-56 by Asf1. A, total protein extracts were made from strain ROY1169 carrying a plasmid expressing the indicated wild type or mutated forms of Asf1 and Western blotted for total H3 levels, acetylated H3 Lys-56, and acetylated Lys-9 and Lys-14. B, representation of the Asf1-H3/H4 crystal structure, with Asf1 in purple, H3 in aqua, and histone H4 in green. Shown in color are locations of Asf1 residues, the substitutions of which are tested above. the intense nuclear staining provided by the chromatinized histones.
Whatever the details of how the histone chaperone Asf1 affects acetylation of lysines 9 and 56 residues on soluble H3, it is clear that the histone binding ability of Asf1 is absolutely required for these acetylation events (Fig. 5). It is important to note that the histone H3/H4 molecules that are bound to Asf1 are seemingly 100% acetylated at every deposition-specific acetylatable lysine in the N-terminal tails (6). By contrast, analysis of all newly synthesized histones H3/H4 shows them to be less than 100% acetylated (10). This saturated acetylation on the newly synthesized H3/H4 bound to Asf1 is fully consistent with our proposal that the binding of H3/H4 to Asf1 potentiates their acetylation. Furthermore, the fact that overexpression of Asf1 greatly increases the levels of Lys-9 and Lys-56 acetylation in the cell (Fig. 4) suggests that the amounts of Asf1 are limiting in the cell, as far as its role in presenting newly synthesized histone H3 to Gcn5 and the Lys-56 HAT.
Finally, having found that acetylation of the soluble histones at H4 Lys-12 and H3 Lys-9 acetylation is apparent by an S-phase peak (Figs. 1-3), we now have a simple assay for the identification of enzymes involved in the acetylation and deacetylation of newly synthesized histones. The HAT for Lys-12 acetylation on newly synthesized H4 is known to be HAT1 (32), and we have now shown the HAT for acetylation of newly synthesized H3 Lys-9 is Gcn5. Similarly, this approach could be applied to identify the histone deacetylases that deacetylate H4 Lys-12 and H3 Lys-9 rapidly after chromatin assembly.