Site-specific Loss of Acetylation upon Phosphorylation of Histone H3*

Post-translational modification of histones is a central aspect of gene regulation. Emerging data indicate that modification at one site can influence modification of a second site. As one example, histone H3 phosphorylation at serine 10 (Ser10) facilitates acetylation of lysine 14 (Lys14) by Gcn5 in vitro (1, 2). In vivo, phosphorylation of H3 precedes acetylation at certain promoters. Whether H3 phosphorylation globally affects acetylation, or whether it affects all acetylation sites in H3 equally, is not known. We have taken a genetic approach to this question by mutating Ser10 in H3 to fix either a negative or a neutral charge at this position, followed by analysis of the acetylation states of the mutant histones using site-specific antibodies. Surprisingly, we find that conversion of Ser10 to glutamate (S10E) or aspartate (S10D) causes almost complete loss of H3 acetylation at lysine 9 (Lys9) in vivo. Acetylation of Lys9is also significantly reduced in cells bearing mutations in the Glc7 phosphatase that increase H3 phosphorylation levels. Mutation of Ser10 in H3 and the concomitant loss of Lys9acetylation has minimal effects on expression of a Gcn5-dependent reporter gene. However, synergistic growth defects are observed upon loss of GCN5 in cells bearing H3 Ser10 mutations that are reminiscent of delays in G2/M progression caused by combined loss ofGCN5 and acetylation site mutations. Together these results demonstrate that H3 phosphorylation directly causes site-specific and opposite changes in acetylation levels of two residues within this histone, Lys9 and Lys14, and they highlight the importance of these histone modifications to normal cell functions.

The packaging of DNA into chromatin impacts every process that uses DNA as substrate. In recent years, the importance of chromatin remodeling to transcriptional regulation has become especially clear due to the discovery and functional characterization of histone modifying enzymes and ATP-dependent chromatin remodeling complexes (3)(4)(5). Chromatin is built from nucleosomal subunits, which consist of 147 base pairs of DNA spooled around the exterior of an octamer of the four core histone proteins. Post-translational modifications of the his-tones may alter histone-DNA, histone-histone, and histonenon-histone protein interactions (6,7). These changes can affect the placement and stability of individual nucleosomes as well as the formation of higher order chromatin structures, thereby affecting accessibility of particular DNA elements to trans-acting factors.
Histones are subject to a variety of post-translational modifications, including acetylation, phosphorylation, ADP-ribosylation, methylation, and ubiquitination (7)(8)(9). Modulations in histone acetylation levels in particular accompany transcriptional changes. In the last few years, multiple coactivator complexes that house histone acetyltransferase (HAT) 1 activity and corepressor complexes that recruit histone deacetylase (HDAC) activity have been described (10 -12). These HAT and HDAC complexes directly participate in chromatin remodeling at promoter regions. Emerging studies indicate that histone methylases are also involved in both the activation and repression of gene expression (13)(14)(15)(16)(17). In addition, specific kinases phosphorylate histone H3 in response to mitogenic signals or for mitotic events (18 -20).
The sites for these post-translational modifications are clustered within the first 30 amino acids of the core histones. Histone H3, for example, is acetylated at lysines (Lys) 9, 14, 18, and 23, phosphorylated at serines (Ser) 10 and 28, and methylated at Lys 4 and Lys 9 (21). The juxtaposition of these sites provides potential for cross regulation of different modification events. Indeed, deacetylation of Lys 14 in H3 appears to augment methylation of Lys 9 at centromeric heterochromatin in Schizosaccharomyces pombe (17). Phosphorylation of Ser 10 in H3 enhances acetylation of Lys 14 by Gcn5 in vitro and precedes Lys 14 acetylation at specific promoters in vivo (1,2). Ser 10 phosphorylation also inhibits methylation of Lys 9 in vitro, and Lys 9 methylation limits Ser 10 phosphorylation both in vitro and in vivo (15,22). This potential for cross-regulation of different histone modifications, together with findings that particular non-histone proteins selectively bind to histones in specific modification states, has led to the idea of a "histone code" (23). In this model, not only the levels of modifications but also the individual types and sites of modification directly facilitate or antagonize association of regulatory proteins with chromatin. These proteins may themselves contribute to further chromatin organization, or they may recruit additional activator or repressor activities.
The functions of H3 Ser 10 phosphorylation are particularly intriguing, because this modification is associated with two apparently disparate processes, which perhaps illustrate the potential of the histone code. H3 Ser 10 phosphorylation peaks * This work was supported in part by Welch Foundation Grant G1371 and National Institutes of Health Grant GM51189 (to S. Y. R. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  during mitotic chromosome condensation in all eukaryotes examined, and Ser 10 mutations cause defects both in chromosome condensation and segregation in Tetrahymena (18). However, H3 Ser 10 phosphorylation is also linked to chromatin opening and activation of genes in response to certain signals, such as the mitogenic stimulation of c-fos (1,20). In both cases, H3 Ser 10 phosphorylation might lead to a transient opening of the chromatin, allowing binding of either condensation factors during mitosis or activation factors during G 1 . Alternatively, alteration of other modifications upon Ser 10 phosphorylation might create a "code" on the H3 tail that is "read" by binding of different factors at different stages of the cell cycle to aid either chromosome condensation or gene activation. The functions of Ser 10 phosphorylation during mitosis might be redundant with other histone modification states in some organisms, since Ser 10 mutations do not affect chromosome segregation or mitotic progression in yeast (18).
To further probe the function of Ser 10 phosphorylation in yeast and to test the hypothesis that this modification is functionally linked to other H3 modification events, we have the examined the effects of Ser 10 mutations on global levels of H3 acetylation. Surprisingly, we find that conversion of Ser 10 to aspartate (Asp) or glutamate (Glu) specifically and drastically lowers acetylation of Lys 9 . Decreases in Lys 9 acetylation levels are also observed in the presence of glc7 phosphatase mutations (19) that increase H3 phosphorylation levels in vivo. All Ser 10 mutations tested confer synergistic growth defects with loss of the Gcn5 histone acetyltransferase, resulting in delayed progression through G 2 /M. Taken together with previous studies by others that indicate Ser 10 phosphorylation enhances acetylation at Lys 14 (1,2), our data indicate that H3 phosphorylation has opposite effects on two neighboring acetylation sites. Moreover, our results suggest that a proper balance of H3 acetylation and phosphorylation is required for normal cell cycle progression.

EXPERIMENTAL PROCEDURES
Yeast Strains-Strains used for this study are listed in Table I. The glc7 strains have been described previously (24,25). All histone mutations were introduced into the MX1-4C strain (a gift from C. David Allis and M. Mitchell Smith, University of Virginia, Charlottesville, VA) by plasmid shuffle. In brief, MX1-4C contains deletions of both loci that encode histone H3 and H4 (HHT1-HHF1 and HHT2-HHF2). The strain also carries a plasmid with URA3 and HHT1-HHF1. Histone mutations were introduced on a TRP1 and HHT2-HHF2 plasmid, pWZ414-F13 (26). The TRP1 plasmids containing histone mutations were transformed into MX1-4C, and TRP1 ϩ colonies were selected. These colonies were transferred to 5-fluoro-orotic acid media (to select against URA3). Plasmids from the 5-fluoro-orotic acid-resistant colonies were rescued and sequenced to confirm loss of wild type histone alleles and presence of the mutant histone alleles.
Deletion of GCN5-A PCR-based strategy was used to disrupt GCN5. The entire coding region was replaced with the kan r gene as described previously (28). Both PCR and Southern blot analysis were used to confirm the presence of the gcn5::kan r allele.
Immunoblotting-Equivalent numbers of cells (ϳ5-25 ϫ 10 7 ) in exponential growth were harvested from each yeast strain by centrifugation for 5 min at 1000 ϫ g. Yeast pellets were frozen in a dry icemethanol bath and stored at Ϫ80°C. Frozen pellets were resuspended in 1.5ϫ SDS-PAGE loading buffer, mixed with an equivalent volume of acid-washed glass beads, and vortexed at high speed in microcentrifuge tubes for 10 min at 4°C. Extracts were spun through a small hole pierced in the bottom of the microcentrifuge tubes and clarified by centrifugation at high speed in a microcentrifuge for 5 min (4°C). Similar results were obtained from extracts prepared from isolated nuclei and from fresh (never frozen) pellets. For TSA experiments, cultures were treated with 25 M TSA for 3 h prior to harvest. For GLC7 experiments, extracts were prepared as described by Hsu et al. (19).
␤-Galactosidase Assays-Indicated yeast strains were transformed with plasmid pPC97-VP16 413-470 (a gift from S. Berger, Wistar Institute, Philadelphia, PA) and plasmid pLGSD5 (32), which carries the ␤-galactosidase gene under the control of a single Gal4 binding site. Multiple colonies of each strain were assayed at least twice. ␤-Galactosidase assays were performed as described (33) except that cells were frozen as a pellet at Ϫ80°C prior to extract preparation.

RESULTS
Mutation of Ser 10 Lowers H3 Acetylation in Vivo-Phosphorylation of histone H3 peptides at Ser 10 augments acetylation at Lys 14 by Gcn5 in vitro, and conversion of Ser 10 to alanine (S10A) decreases activation of some Gcn5-regulated promoters in vivo (1,2). These results suggest that prevention of phosphorylation by mutation of Ser 10 to alanine weakens Gcn5mediated acetylation of H3, which normally facilitates gene activation. To directly test the effects of the S10A mutation on H3 acetylation levels, we probed whole cell protein extracts from cells bearing wild type or mutated H3 on immunoblots using antibodies specific for particular acetylated H3 isoforms. In cells bearing H3 mutations, both wild type copies of the H3 gene were deleted so that the plasmid-borne, mutated copy provided the sole source for H3 in the cell (34, 35).
Two antibodies were predominantly used in our studies. One commercially available antibody that is often used to evaluate H3 acetylation levels was raised against an H3 peptide containing acetyllysine at positions 9 and 14 (anti-H3 Ac 9,14). A second antiserum was raised against an H3-peptide with acetyllysines at positions 9 and 18 (anti-H3 Ac 9,18) (31). We observed a decrease (ϳ60%) in the amount of H3 acetylation detected by the H3 Ac 9,14 antisera in extracts from cells bearing the H3 S10A mutation ( Fig. 1 and Table II), as expected from the above reports. Less decrease was observed using the anti-H3 Ac 9,18 antisera (Fig. 1).
If the introduction of a negative charge at Ser 10 by phosphorylation is important for augmentation of Gcn5 (or other HAT) functions, then placement of a constitutive negative charge at this position should increase acetylation. However, in contrast to our expectations, S10D and S10E H3 mutations reduced acetylation of H3 detected by the anti-H3Ac 9,14 antibody by 90% or more ( Fig. 1 and Table II). Acetylation detected by anti-H3 Ac9,18 was also reduced, but to a lesser extent (ϳ40%). The persistence in staining with this antibody demonstrates that intact H3 is present in extracts from the mutant cells. Acetylation of H4 was not altered in extracts from the mutant strains, indicating that the effects on acetylation are specific for the mutated H3 histone (Fig. 1). S10D and S10E Mutations Specifically Affect Acetylation of Lys 9 in H3-The apparent loss of acetylation detected by the anti-H3 Ac 9,14 antisera might reflect a decrease in HAT recognition of the mutant H3, increased HDAC activity toward this mutant, or a destruction of the epitope recognized by this antibody. To determine whether the S10D or S10E H3 mutations preclude acetylation in vivo, cells bearing these mutations were grown in the presence of TSA, a specific inhibitor of histone deacetylase activities. Under these conditions, both the S10D and S10E mutant forms of H3 exhibited increased acetylation detected by the H3 Ac 9,14 antibody (Fig. 2, upper panel), confirming the ability of these histones to be acetylated in vivo. Importantly, these results also confirm the ability of the H3 Ac 9,14-specific antisera to recognize these mutant forms of H3, demonstrating that the epitope for the antibody is intact.
To further define the effects of the S10D and S10E mutations on particular acetylation sites, we reinvestigated the specificity of the H3 Ac 9,14 antibody. In previous work, we established that Lys 14 mutations had negligible effects on detection of H3 by this antibody (34). In contrast, mutation of Lys 9 virtually abolished H3 recognition by the H3 Ac 9,14 antisera. These results suggested either that the antibody has a much higher specificity for acetyl-Lys 9 than for acetyl-Lys 14 or that bulk yeast histones are acetylated at Lys 9 much more frequently than at Lys 14 . To distinguish these possibilities, we analyzed the ability of the H3 Ac 9,14 antibody to recognize peptides synthesized with acetyllysine at position 9, position 14, or other sites using a slot blot assay. Equal amounts of peptides were applied to the membrane, as confirmed by staining with India ink (Fig. 3). The H3 Ac 9,14 antibody did not recognize the unmodified H3 peptide, but did recognize all H3 peptides containing acetyl-Lys 9 . In contrast, a peptide acetylated at Lys 14 , Lys 18 , and Lys 23 , but lacking Lys 9 acetylation, was not recognized. Therefore, the H3 Ac 9,14 antibody appears largely specific for H3 acetylated at Lys 9 , although other studies have shown that this antibody can be blocked effectively with a H3 Ac Lys 14 peptide (1). Similar slot blot analysis of the H3 Ac 9,18 FIG. 1. Diminished acetylation of histone H3 Lys 9 in S10D and S10E mutants. Yeast protein extracts were resolved by SDS-PAGE electrophoresis, transferred to polyvinylidene difluoride membrane, and probed with the indicated antibodies. The lower panel shows a Coomassie stain of a parallel gel to illustrate equivalent protein load.
FIG. 2. Histone H3 can be acetylated at Lys 9 in S10D and S10E mutants. Protein extracts were prepared from yeast either treated or not treated with TSA. Histone acetylation was analyzed by immunoblot with the indicated antibodies. Equivalent protein loads were confirmed by Coomassie staining.  as MX1-4C, except plus pRS414-59 instead of pms327 S10A JDY21 as JDY17, except ⌬gcn5ϻKan R S10A JDY18 as MX1-4C, except plus pJD112 instead of pms327 S10D JDY22 as JDY18, except ⌬gcn5ϻKan R S10D JDY19 as MX1-4C, except plus pJD113 instead of pms327 S10E JDY23 as JDY19, except ⌬gcn5ϻKan R S10E BMY1 as MX1-4C, except plus pBM1 instead of pms327 S10D,K9Q BMY2 as MX1-4C, except plus pBM2 instead of pms327 S10D,K9R BMY3 as MX1-4C, except plus pBM3 instead of pms327 S10D,K14Q BMY4 as MX1-4C, except plus pBM4 instead of pms327 S10D,K14R KT1112 MAT a leu2 ura3-52 his3 KT1667-a as KT1112, except glc7-129 KT1638 as KT1112, except glc7-109 KT1640 as KT1112, except glc7-127 antibody confirms its specificity for acetylation events at either Lys 9 or Lys 18 in H3 (Fig. 3). Together with our above experiments, these results indicate that the H3 S10D and S10E mutations likely inhibit acetylation of Lys 9 . We confirmed this effect by probing immunoblots of whole cell extracts with antibodies raised against a peptide acetylated individually at Lys 9 . This monospecific H3 Ac 9 antibody reacted well with wild type yeast H3, but staining was drastically reduced for the H3 S10D or S10E mutants (Fig. 2,  bottom panel). Acetylation of Lys 9 as detected by this antibody was regained upon growth of cells in TSA, again confirming the ability of this residue to be acetylated within the context of the Ser 10 mutations and the integrity of the epitope in the mutant histones. The S10D and S10E mutations, therefore, severely and specifically inhibit acetylation of histone H3 at Lys 9 .
Glc7 Phosphatase Mutations That Increase H3 Phosphorylation Also Inhibit Lys 9 Acetylation-If conversion of Ser 10 to Asp or Glu mimics phosphorylation, then phosphatase mutations that increase Ser 10 phosphorylation in vivo should also decrease acetylation of Lys 9 . The Glc7 phosphatase is an essential gene in yeast. Particular mutant alleles of the GLC7 (glc7-127 and glc7-129) increase H3 Ser 10 phosphorylation levels, whereas other alleles of GLC7 (glc7-109) do not (19) (Fig. 4B). Immunoblots of equal loads of extracts from isogenic wild type or glc7 cells reveal that H3 acetylation detected by the anti-H3 Ac 9,14 antibody is decreased in the presence of those alleles that result in increased phosphorylation at Ser 10 (asterisks in Fig 4a). Thus, high levels of H3 phosphorylation cause a decrease in Lys 9 acetylation just as did conversion of Ser 10 to Asp or Glu. A negative charge at this position in H3 apparently antagonizes acetylation at Lys 9 , even though it stimulates acetylation at Lys 14 .
H3 Ser 10 Mutations Do Not Affect Gcn5-dependent Activation by Gal4-VP16 -Activation of the Gcn5-dependent genes is accompanied by increased acetylation of histones associated with promoter regions (36). Since Gcn5 exhibits an enhanced recognition of phosphorylated H3 peptides in vitro (1,2), phosphorylation of H3 should increase activation in vivo. Accordingly, mutation of H3 Ser 10 to alanine, which prevents phosphorylation, has been reported to limit activation of some Gcn5-dependent genes in yeast (2). We reasoned that the H3 S10D and S10E mutations might enhance expression of Gcn5-dependent genes by facilitating acetylation of Lys 14 by Gcn5. However, since these mutations also abolish acetylation of H3 Lys 9 , they might limit activation. Therefore, we examined expression of a Gcn5-dependent reporter gene in isogenic strains containing wild type H3 or our H3 Ser 10 mutations to determine the relative contributions of Ser 10 phosphorylation, Lys 9 acetylation, or Lys 14 acetylation to the activation of this gene.
Activation of reporter genes containing Gal4 binding sites by the Gal4-VP16 fusion protein requires Gcn5. Previously, we demonstrated that mutation of multiple lysines in H3 and H4 to glutamine can bypass the Gcn5 requirement, supporting an important role for histone acetylation in Gal4-VP16 mediated activation (34). Therefore, we chose this same system to test the effects of the Ser 10 mutations on Gal-VP16-mediated activation. As expected, the reporter was activated in cells containing wild type histones, and this activation was diminished in the absence of Gcn5 (Fig. 5). Surprisingly, none of the Ser 10 mutations affected activation of the reporter in cells containing Gcn5. Reporter activation was decreased in gcn5 cells carrying the H3 Ser 10 mutations. However, the reporter reproducibly exhibited higher expression levels in these cells relative to gcn5 cells containing wild type H3. In our previous work, this reporter gene also exhibited slightly higher activity in gcn5 cells in the presence of H3 Lys 9 mutations (to arginine or to glutamine), although single mutations in H3 at Lys 9 or Lys 14 had little effect on the overall Gcn5-dependence of Gal-Vp16 mediated activation (34). The similar effects we observe in cells carrying the H3 S10A, S10D, S10E, K9R, or K9Q mutations indicate that Ser 10 and Lys 9 do contribute to gene activation, FIG. 3. Specificity of H3 Ac 9,14 and H3 Ac 9,18 antibodies. Slot blots of peptides corresponding to the amino terminus of H3 were probed with the indicated antibody or stained with India ink. The acetylation state of the lysine residues in the indicated position is designated by Ϫ (no acetylation) or ϩ (acetylated).
FIG. 4. H3 acetylation is reduced in GLC7 mutants. Protein extracts were prepared from yeast of the indicated strain. The asterisks indicate GLC7 mutants reported to display increased H3 phosphorylation. Histone acetylation (A) and phosphorylation (B) were analyzed by immunoblot with the indicated antibodies. The signal from the acetylation specific antibody was quantitated using the Apha Innotech Imaging System and normalized to protein load (relative signal). Equal protein loads for the ␣H3 phosphoserine 10 immunoblot were determined by Bio-Rad protein assay of the yeast whole cell extracts (19).
but that the charge of these residues is less important than their overall structure.
Ser 10 Mutations and Loss of Gcn5 Synergistically Cripple Cell Cycle Progression-Previous studies from our laboratory revealed synergistic growth defects upon loss of the Gcn5 acetylase and mutation of predominant acetylation sites in H3 and H4 (34). Mutation of Lys 9 , for example, had little effect on yeast doubling time, but when this mutation (or a mutation of Lys 14 ) was combined with a disruption of GCN5, doubling times were significantly increased. Since Ser 10 mutations globally affect Lys 9 acetylation, we reasoned that these mutations might also exhibit synergistic defects when combined with disruptions of GCN5.
We first determined whether our H3 Ser 10 mutations altered cell growth on their own. Comparison of doubling times indicated no significant difference between strains bearing the S10A, S10D, or S10E mutations or cells bearing wild type H3 (Table III), consistent with previous reports that S10A mutations do not affect the yeast cell cycle (18). Loss of GCN5 alone lengthened doubling times, as we reported previously (34). However, combination of any of the Ser 10 mutations with a GCN5 disruption further lengthened cell doubling times (Ͼ250 min). FACS analyses indicate that a majority of the double mutant cells have a 2N content of DNA, consistent with a delay in progression through G 2 /M (Fig 6).
To determine whether the lengthened cell doubling times observed in the gcn5 strains bearing the Ser 10 mutations are due to specifically to a loss of acetylation of H3 at Lys 9 or Lys 14 , we analyzed the growth of GCN5 strains bearing the S10D mutation together with mutation at either of these lysines (Table III). We did not expect mutation of Lys 9 to mimic the effect of GCN5 loss, since acetylation of Lys 9 is already abolished in the S10D mutant strain. Indeed, mutation of Lys 9 had little effect upon the growth of the S10D mutant. However, since H3 Lys 14 is a preferred substrate of Gcn5, mutation of this site might lead to an increased doubling time in the presence of the S10D mutation. The S10D K14Q and the S10D K14R double H3 mutants both exhibited slightly slower growth than the S10D mutant (Table IV), but these cell doubling times were still significantly shorter than the H3 S10D gcn5 mutant (Table III). These results indicate that loss of acetylation at additional sites and substrates likely contributes to the delayed cell cycle progression observed in the S10D gcn5 strains. Interestingly the similar effects of the K14R and K14Q mutations in the presence of the S10D mutation indicate that the structure of the lysine residue, or acetyllysine, is more important than the charge at this position in H3.
Together, these results demonstrate that H3 Ser 10 is important to normal cell cycle progression in yeast and that the functions of this residue, and its modification, are redundant with multiple acetylation events mediated by Gcn5. Overall, our data suggest that a critical balance in histone acetylation and H3 phosphorylation is required for normal cell division.

DISCUSSION
The histone code hypothesis predicts that specific combinations of histone modifications provide regulatory information through changes in the structure of chromatin and in the association of non-histone proteins with particular nucleosomes (7,23). Inherent in this hypothesis is the idea that modification of one residue in a histone may affect the type and frequency of modifications at other sites. One of the first examples of such cross-regulation was the discovery that phosphorylation of Ser 10 in histone H3 augments recognition of the H3 amino-terminal tail by Gcn5, leading to increased acetylation of Lys 14 (1,2). Here we show that phosphorylation or mutation of Ser 10 has an opposite effect on acetylation of Lys 9 .
Our results indicate that acetylation of each lysine within a histone tail is independently regulated, supporting the idea that each lysine has unique functions. These studies provide further in vivo support for the histone code hypothesis.
A particular arginine in Gcn5 (Arg 164 in yeast Gcn5) provides a pocket for binding-phosphorylated Ser 10 in H3 (1, 2), based on the crystal structure of a ternary Gcn5-H3 peptide-coenzyme A complex (37). These interactions appear to facilitate acetylation of Lys 14 by stabilizing Gcn5-H3 interactions and by aligning Lys 14 in the HAT catalytic center. In support of this model, mutation of Arg 164 in Gcn5 impedes activation of the same subset of Gcn5-dependent genes that are affected by S10A mutations in H3. The inhibition of Lys 9 acetylation that we observe in vivo upon Ser 10 mutation or phosphorylation could reflect a misalignment of this residue within the HAT active site. Unfortunately, Lys 9 (the first residue of the H3 peptide used for crystallization) is highly disordered within the cocrystal, so its location relative to specific active site residues in Gcn5 cannot be discerned from the structures in hand (37).
Although our data indicate that Ser 10 phosphorylation antagonizes Lys 9 acetylation in yeast, they do not rule out the possibility that these modifications might coexist at certain promoters, under specific conditions. Indeed, Ser 10 phosphorylation and Lys 9 acetylation do coexist on H3 molecules associated with the mitogen-stimulated c-fos promoter (38). However, these dually modified H3 molecules occur very rarely in mammalian cells, perhaps reflecting the antagonism we observe here.
We previously reported that mutation of GCN5 causes a delay in G 2 /M progression that is exacerbated upon mutation of specific lysines in H3 or H4 (34). These studies indicated that a certain threshold of acetylation events is required for normal cell cycle progression. Consistent with this idea, concomitant loss of the genes encoding the Gcn5 and Sas3 H3 HAT activities is lethal in yeast (39). The Esa1 H4 HAT is also essential for yeast cell viability (40,41). The cell cycle delays we observe here upon mutation of H3 Ser 10 and loss of GCN5 are highly reminiscent of those observed in our previous studies (21,34) and those observed by Howe et al. (39) upon loss of GCN5 and SAS3. One simple explanation is that the lowered acetylation of Lys 9 that occurs upon Ser 10 mutation becomes critical upon loss of other acetylation events mediated by Gcn5. However, the synergistic phenotype may reflect a more complicated relationship between H3 Ser 10 and Gcn5 functions, since all three Ser 10 mutations tested, S10A, S10D, and S10E, exhibit the same degree of synergy with GCN5 loss, but the S10A mutation causes only a slight (2-fold) change in Lys 9 acetylation levels. The structure of serine or of phosphoserine, not just the charge, at position 10 is apparently important for H3 functions. The G 2 /M delay that we observe could reflect abnormal chromatin folding in the presence of the H3 Ser 10 and GCN5 mutations due to loss of both phosphorylation and acetylation. Alternatively, these changes in H3 modification might result in abnormal expression of a gene(s) required for proper progression through this phase of the cell cycle. Interestingly, mutations in ESA1 or mutations in multiple H4 acetylation sites trigger a RAD9-dependent G 2 /M checkpoint response, likely reflecting abnormal chromatin structures and decreased genome integrity in the face of these mutations (35,41,42).
At least three different kinases phosphorylate H3 Ser 10 in response to specific signals. Ipl1 appears to phosphorylate H3 specifically during mitosis, whereas pp90Rsk-2 and Snf1 phosphorylate H3 in response to mitogenic and nutritional responses, respectively (19,20,43). The involvement of these kinases highlights the potential for H3 as an integrator of multiple signal transduction pathways. Given our data and other recent findings, phosphorylation of H3 Ser 10 may be a nexus for regulation of H3 modifications. In addition to affecting acetylation of both Lys 14 (1,2) and Lys 9 (in opposite directions), Ser 10 phosphorylation inhibits methylation of Lys 9 in S. pombe and Drosophila (15,17,22). H3 Lys 9 methylation in these organisms inhibits Ser 10 phosphorylation, providing potential for feedback regulation (44). Moreover, since Ser 10 augments Lys 14 acetylation, which inhibits Lys 9 methylation, it is easy to envision a cascade of modification events on the H3 tail that might be modulated in response to specific signals to deliver different messages via the histone code. Wild type 151 Ϯ 6 S10D 159 Ϯ 11 S10D,K9R 162 Ϯ 13 S10D,K9Q 165 Ϯ 12 S10D,K14R 177 Ϯ 16 S10D,K14Q 181 Ϯ 18 a Average of three experiments.