Yng2p-dependent NuA4 Histone H4 Acetylation Activity Is Required for Mitotic and Meiotic Progression*

In all eukaryotes, multisubunit histone acetyltransferase (HAT) complexes acetylate the highly conserved lysine residues in the amino-terminal tails of core histones to regulate chromatin structure and gene expression. One such complex in yeast, NuA4, specifically acetylates nucleosome-associated histone H4. Recent studies have revealed that NuA4 comprises at least 11 subunits, including Yng2p, a yeast homolog of the candidate human tumor suppressor gene, ING1 . Consistent with prior data, we find that cells lacking Yng2p are deficient for NuA4 activity and are temperature-sensi-tive. Furthermore, we show that the NuA4 complex is present in the absence of Yng2p, suggesting that Yng2p functions to maintain or activate NuA4 HAT activity. Sporulation of diploid yng2 mutant cells reveals a defect in meiotic progression, whereas synchronized yng2 mutant cells display a mitotic delay. Surprisingly, genome-wide expression analysis revealed little change from wild type. Nocodazole arrest and release relieves the mitotic defects, suggesting that Yng2p may have a critical function prior to or during metaphase. Rather than a uniform decrease in acetylated forms of histone H4, we find striking cell-to-cell heterogeneity in the loss of acetylated histone H4 in yng2 mutant cells. Treating yng2 using a Zeiss AxioSkop microscope equipped with epifluorescence optics, High Q fluorescein, and DAPI filter sets (Chroma Technologies) and a Sensys CCD camera (Photometrix) controlled with IP-Lab Spectrum software (Signal Analytics). Cells carrying GFP- YNG2 (SKY2303) were grown to A 600 nm (cid:1) 0.2– 0.4 in synthetic media at 30 °C and washed with water. To examine GFP-Yng2p localization through the cell cycle, cells were subjected to an (cid:1) f arrest-release regimen. Samples were collected at 30-min inter- vals after release and imaged using a High Q fluorescein filter set (Chroma) and the microscope system described above. Cells carrying GFP-Yng2p were grown in the presence of DAPI (2.5 (cid:2) g/ml) in synthetic media for 3–4 h for colocalization studies. regulation cell

In all eukaryotes, multisubunit histone acetyltransferase (HAT) complexes acetylate the highly conserved lysine residues in the amino-terminal tails of core histones to regulate chromatin structure and gene expression. One such complex in yeast, NuA4, specifically acetylates nucleosome-associated histone H4. Recent studies have revealed that NuA4 comprises at least 11 subunits, including Yng2p, a yeast homolog of the candidate human tumor suppressor gene, ING1. Consistent with prior data, we find that cells lacking Yng2p are deficient for NuA4 activity and are temperature-sensitive. Furthermore, we show that the NuA4 complex is present in the absence of Yng2p, suggesting that Yng2p functions to maintain or activate NuA4 HAT activity. Sporulation of diploid yng2 mutant cells reveals a defect in meiotic progression, whereas synchronized yng2 mutant cells display a mitotic delay. Surprisingly, genomewide expression analysis revealed little change from wild type. Nocodazole arrest and release relieves the mitotic defects, suggesting that Yng2p may have a critical function prior to or during metaphase. Rather than a uniform decrease in acetylated forms of histone H4, we find striking cell-to-cell heterogeneity in the loss of acetylated histone H4 in yng2 mutant cells. Treating yng2 mutants with the histone deacetylase inhibitor trichostatin A suppressed the mitotic delay and restored global histone H4 acetylation, arguing that reduced H4 acetylation may underlie the cell cycle delay.
Histone acetyltransferases (HATs) 1 are related enzymes that mediate regulated acetylation of histones on their highly conserved amino-terminal lysines. HATs have been isolated from yeast, nematode, Drosophila, mouse, and human (1,2). Recent work in budding yeast has demonstrated that multi-protein HAT-containing complexes acetylate chromatin-associated histones (3). Histone acetylation has important roles in both growth and cell division. Mutations in the HAT catalytic subunits, as well as mutant histones lacking their amino-terminal tails, confer marked cell cycle delays in mitosis and other growth defects (4 -6, 14, 15).
NuA4 (nucleosomal acetyltransferase of H4) is probably responsible for most, if not all, histone H4 acetylation in chromatin during vegetative growth (9,11). This complex consists of at least 10 subunits in addition to Esa1p, virtually all of which are essential for life (4,5,11,12). Although biochemical properties of NuA4 are well characterized, its regulation and function in cell division is less well understood. In wild-type yeast cells, most histone H4 is fully or partially acetylated via modification of lysines at positions 5,8,12, and/or 16 in the histone H4 amino terminus (13). Although mutating the four histone H4 lysines to arginine confers dramatic growth defects (14), a single lysine is sufficient to restore regulated growth (15). Chromatin containing a high proportion of polyacetylated histones is considered to be more "open" or accessible to regulatory factors, whereas unacetylated chromatin is more compact and may be less active (16). In turn, acetylation may facilitate large scale conformational change that is impaired by the tighter packing of unacetylated chromatin (17). Recent data showing that Esa1p has a relatively limited role in transcriptional regulation at specific loci but is important for large scale acetylation of chromatin (18,19) suggest that NuA4 has a primary role in global histone H4 acetylation and thereby in regulating changes in large scale chromatin structure and function.
That recombinant Esa1p alone acetylates only free histone H4 suggests that other NuA4 subunits target this HAT to its chromatin-associated nucleosomal substrates (9). Thus, further insight into the function of NuA4 in cellular processes may be provided by genetic and biochemical analysis of the noncatalytic subunits. In this report, we further characterize the NuA4 subunit Yng2p, previously identified as a yeast homolog of the candidate human tumor suppressor gene ING1 (12,20,21).

EXPERIMENTAL PROCEDURES
Yeast Strains and Manipulations-All yeast strains used in this study derive from W303-1A (22). Yeast culture and genetic techniques were essentially as described (23). Media were obtained from United States Biological, molecular biology reagents from New England Bio-Labs, and chemical reagents from Sigma, unless otherwise noted. Yeast were cultured in YPD (1% yeast extract, 2% peptone, 0.3 mM adenine, and 2% glucose) or SC (synthetic complete media with 2% glucose) lacking the appropriate amino acids.
Plasmid and Strain Construction-A diploid strain heterozygous for a knockout of YNG2 (SKY2302) was constructed by PCR-based gene disruption as described (24) to replace the entire coding sequence of one allele of YNG2 in a wild-type diploid with the kanMX6 gene conferring G418 resistance. Transformants were sporulated and dissected to isolate YNG2 disruptants (yng2::kanMX6). YNG2 and EPL1 were epitopetagged at their carboxyl terminus with 13-Myc or 3-Ha, respectively, using a PCR-based strategy (24). All strain constructions were confirmed by PCR.
Cell Cycle Analysis-Cells of the relevant genotypes were grown in YPD overnight at 22°C then diluted to an A 600 nm of ϳ0.05-0.1 in fresh YPD and allowed to grow for 3-4 h in the presence of 5 M alpha mating peptide (␣f, Research Genetics) or 15 g/ml of nocodazole at 22°C. Alternatively, cells were incubated at 22°C in the presence of ␣f or nocodazole for 2 h and then shifted to 37°C for 1-1.5 h. Trichostatin A (TSA) in methanol (5 mg/ml) was added directly to cultures to a final concentration of 30 g/ml. Cells were released from ␣f or nocodazole arrest by centrifugation, a YPD wash, and then resuspension in liquid YPD at 22°C or in YPD prewarmed to 37°C. Cells were collected at 20-min intervals (or as described under "Results") for flow cytometry and Northern and Western analyses (see "Protein Techniques"). For flow cytometry, cells were fixed in 70% ethanol, resuspended in 50 mM Tris-HCl (pH 7.5), sonicated, treated with 1 mg/ml RNase A at 50°C for 1 h, and stained with 0.05 mg/ml propidium iodide. Approximately 2-3 ϫ 10 4 cells were analyzed using CellQuest software and a FACS-Calibur flow cytometer (Becton Dickinson) for each time point. For Northern analysis, 25 ml of culture were collected, and RNA was extracted as described (27), separated on a 1% formaldehyde-agarose gel, and transferred onto a nylon membrane (Osmonics). Probes labeled with [ 32 P]dATP or -dCTP (ICN) were generated with PCR or Klenow and used to detect YNG2, CLN2, CLB5, CLB1, and CLB2 mRNA. A PhosphorImager detection system and ImageQuant software (Molecular Dynamics) were used for analysis of blots.
Microarray Analysis-Strains were grown in YPD media to mid-log phase and then harvested. RNA was collected from cells using a Qiagen Rneasy kit (Qiagen). 24 g of total RNA was used as template for double-stranded cDNA synthesis using a Life Technologies SuperScript Double Strand cDNA synthesis kit and protocol described by Affymetrix (Affymetrix GeneChip Expression Analysis Technical Manual). Briefly, total RNA is synthesized into double-stranded cDNA by means of the SuperScript Choice system (Life Technologies, Inc.) and a T7-(dT) 24 Primer (Genset Corp.). Samples were then extracted with phenol:chloroform:isoamyl alcohol (25:24:1) and precipitated with 7.5 M ammonium acetate and ethanol. An Enzo Bioarray Highyield RNA transcript labeling kit (Enzo Diagnostics, Inc.) was used according to manufacturer's instructions to transcribe biotin-labeled cRNA, which was purified using an RNeasy kit (Qiagen). These products were fragmented into 50 -200-nucleotide pieces with a Trizma base solution as described by Affymetrix. 5 g of fragmented cRNA was hybridized on an Affymetrix Test 2 array. After washing, the array was stained with streptavidinphycoerythrin (Molecular Probes), amplified by biotinylated antistreptavidin (Vector Laboratories, Inc.), and then scanned on an HP Gene-array scanner. The intensity for each feature of the array was captured with Affymetrix GeneChip software according to standard Affymetrix procedures. All 3Ј-5Ј ratios fell between normal parameters. 15 g of fragmented cRNA was assayed on Yeast Genome S98 arrays (Affymetrix) utilizing the same protocol and equipment. Data were analyzed on the Affymetrix MicroArray Suite data analysis package. Of a total of 9159 probe sets representing ϳ7000 genes, including nonannotated open reading frames (ORFs) from SAGE analysis, 7616 (83.2%) were detected and exceeded the probe pair threshold determined by Affymetrix GeneChip software. The 54 ORFs that decreased includes 18 transcripts scored by Affymetrix MicroArray Suite as "no change." Nonetheless, these were included for completeness. Molecular function and gene names for each open reading frame was obtained from the Saccharomyces Genome Data base web page (at genome-www. stanford.edu/Saccharomyces).
Protein Techniques-For protein extracts, 25-50 ml of culture at A 600 nm ϳ0.1 was pelleted, washed, and resuspended in buffer containing protease inhibitors (50 mM Tris-HCl, pH 7.5, 10% glycerol, 0.5% Nonidet P-40, 2 mM EDTA, 150 mM NaCl, 500 M benzamidine HCl, 10 g/ml aprotinin, 1 g/ml leupeptin, 1 g/ml pepstatin A, 1 mM phenylmethanesulfonyl fluoride). An equal volume of 0.5-mm glass beads was added, and the mixture was vortexed at high speed for 30 min at 4°C and centrifuged at 20,000 ϫ g at 4°C to clear the lysate. For Western analysis, 50 g of protein, determined by Bradford assay (Bio-Rad), was loaded per lane on a 12% SDS-polyacrylamide gel and then transferred to Hybond C nitrocellulose (Amersham Pharmacia Biotech). Blots were processed as described by the manufacturer for enhanced chemiluminescence (Amersham Pharmacia Biotech) and exposed to Hyperfilm ECL (Amersham Pharmacia Biotech). The following concentrations of antibodies were used to detect Yng2p-Myc, Epl1p-Ha, Esa1p, and Clb2p: 1/1000 of A14 anti-Myc polyclonal (Santa Cruz), 1/500 of 16B12 anti-Ha polyclonal (Covance), 1/200 of anti-Esa1p (Santa Cruz), and 1/2000 anti-Clb2p (28). For immunoprecipitation reactions, 3 g of a monoclonal anti-Myc antibody (9E10, Covance) or anti-Ha antibody (12CA5, Roche Molecular Biochemicals) was added to 400 l of total yeast extract (0.2-1.0 mg/ml) and incubated on a rocker at 4°C for 2-3 h or overnight. Then, 50 l of a 50% slurry of protein A-Sepharose beads (Amersham Pharmacia Biotech), washed once with lysis buffer, was added and incubated on a mixer at 4°C for 2-3 h. Immune complexes were collected by centrifugation at 800 ϫ g for 2 min, washed three times with 1 ml of lysis buffer, and then resuspended in 1ϫ SDSpolyacrylamide gel electrophoresis sample buffer (62.5 mM Tris, pH 6.8, 10% glycerol, 2% SDS, 16 mM dithiothreitol, 0.004% bromphenol blue) or washed once with 1 ml of HAT buffer for HAT assays (see below).
Histone Acetyltransferase Assays-HAT assays were performed essentially as described (29). Immunoprecipitates were incubated at 30°C for 45 min with 50 g of calf thymus histones (type IIA, Sigma), 0.3 Ci of [ 3 H]acetyl-CoA (ICN) in HAT buffer A (50 mM Tris-Cl, pH 8.0, 10% glycerol, 10 mM sodium butyrate, 1 mM phenylmethanesulfonyl fluoride, 1 mM dithiothreitol, and 1.5 M acetyl-CoA in a total volume of 50 -80 l. Reactions were terminated by the addition of 4ϫ SDSpolyacrylamide gel electrophoresis sample buffer and separated on a 15 or 18% SDS-polyacrylamide gel, transferred onto polyvinylidene difluoride (Millipore), treated for fluorography using En3Hance (PerkinElmer Life Sciences) as described by the manufacturer, and exposed to Hyperfilm ECL (Amersham Pharmacia Biotech) at Ϫ80°C for 1 to 2 weeks.
Cells carrying GFP-YNG2 (SKY2303) were grown to A 600 nm ϳ0.2-0.4 in synthetic media at 30°C and washed with water. To examine GFP-Yng2p localization through the cell cycle, cells were subjected to an ␣f arrest-release regimen. Samples were collected at 30-min intervals after release and imaged using a High Q fluorescein filter set (Chroma) and the microscope system described above. Cells carrying GFP-Yng2p were grown in the presence of DAPI (2.5 g/ml) in synthetic media for 3-4 h for colocalization studies.

Yng2p Is Required for Full NuA4 HAT Activity-Recently,
Loewith et al. (12) reported that Yng2p is associated with both Tra1, an ataxia-telangiectasia mutated (ATM)-like subunit of NuA4, and Esa1p, the catalytic subunit of NuA4 (9). Independently, by two-hybrid screening and co-immunoprecipitation (data not shown), we found that Yng2p physically interacts with Epl1p, a stable subunit of the NuA4 HAT complex (11). Moreover, by Western analysis of total histone protein, Loewith et al. (12) observed decreased levels of acetylated forms of histone H4 in YNG2 deletion mutants. These results suggest that Yng2p is a bona fide NuA4 subunit and is important for H4 acetylation in vivo. However, Yng2p may be important for NuA4 HAT catalytic activity, may act to target NuA4 to its chromosomal histone H4 substrate, or may serve some other role. Given that YNG2 is not essential for vegetative growth at 22°C (12), we sought to assay the NuA4 HAT activity in YNG2deficient cells. Because Epl1p is a stable subunit of NuA4 (11) and the expression of Epl1p-Ha in yng2 mutants was comparable with that in wild-type cells (Fig. 1A), immunoprecipitation of Epl1p-Ha provided a means of assaying NuA4 HAT activity in vitro. Epl1p-Ha immune complexes from wild-type cells possessed ϳ5-fold greater H4 HAT activity than Epl1p-Ha immune complexes isolated from Yng2p-deficient cells (Fig.  1B). To determine whether Yng2p is necessary for full NuA4 activity and/or assembly in vivo, we examined Epl1p-Ha immune complexes for the presence of Esa1p in wild-type cells versus yng2 mutant cells. Strikingly, we found that Esa1p coimmunoprecipitated with Epl1p-Ha to a similar degree in both wild-type and yng2 mutant cells (Fig. 1C). These data suggest that Yng2p is not necessary for NuA4 complex formation but is required for wild-type levels of NuA4 H4 HAT activity.
yng2 Mutant Cells Exhibit Slow Mitotic Progression-The physiological processes that Yng2p may mediate remain unclear. Previous studies by Megee et al. (15) demonstrate that yeast cells harboring a non-acetylatable mutant histone H4 allele delay in G 2 and mitosis. Moreover, Clarke et al. (5) showed that cells carrying a temperature-sensitive allele of ESA1 display a G 2 /M arrest. To examine whether a similar delay might underlie the slow growth at 22°C and temperature sensitivity at 37°C of yng2 mutant cells (12), we performed flow cytometry on YNG2 wild-type and mutant cells synchronized by ␣f-induced G 1 arrest. To limit growth to a single cell cycle, ␣f was added back after cells entered S phase, trapping cells in the subsequent G 1 phase. Virtually all wild-type cells released at 22°C completed DNA synthesis by 60 min, progressed through mitosis, and reentered G 1 as indicated by the complete return to a 1 N DNA peak within 140 min ( Fig. 2A,  left). Under the same conditions, yng2 mutants completed S phase with near wild-type kinetics. However, cells with 2 N DNA content persisted and only began to reenter G 1 at 160 min (compare Fig. 2B, left with Fig. 2A, left). Furthermore, 45% of yng2 mutant cells did not exit from the 2 N DNA content fraction for the duration of the experiment (Fig. 2B, left). At 37°C, wild-type cells completed DNA replication by 40 min, promptly entered mitosis, and began to return to G 1 by 80 min (Fig. 2A, right). yng2 mutants exhibited a brief delay at the onset of S phase but then completed DNA replication by 60 min (Fig. 2B, right). However, a long mitotic arrest ensued. Al-  1 and 2), whereas a dramatic loss in HAT activity is observed in yng2 mutant cells (lanes 3 and 4). 5X indicates 5-fold more extract for immunoprecipitation reactions than in 1X lanes. C, Epl1p-Ha immunoprecipitated from wild-type and yng2 mutant cells is associated with the NuA4 HAT Esa1p (lanes 9 -12). Control immunoprecipitation reactions (lanes 5-8) were performed with protein A beads alone. Western analysis of whole cell extracts (WCE) demonstrates that Esa1p is expressed at similar levels in wildtype and yng2 mutant cells (lanes 1-4). Extracts were prepared from four independently derived strains expressing Epl1p-Ha in a wild-type (lanes 3, 7, and 10 and 4, 8, and 12) or yng2 mutant (lanes 1, 5, and 9 and 2, 6, and 11) background.

FIG. 2. Yng2p-deficient cells display a mitotic delay and fail to sporulate.
A and B, wild-type and yng2 mutant cells were collected every 20 min after ␣f-induced G 1 arrest. ␣f was added back after cells entered S phase in order to arrest cells in the subsequent G 1 phase. Flow cytometry reveals that wild-type cells complete one cell cycle by 140 min at 22°C and 100 min at 37°C. In contrast, yng2 mutant cells delay with a persistent 2 N DNA peak (arrows) at 22°C; this was exacerbated at 37°C. C, diploid wild-type or yng2 homozygous mutant cells were sporulated on solid media for 3 days. Only ϳ1.5% of yng2 mutant cells formed four spore asci compared with ϳ30% in wild-type cells. Cells stained with DAPI revealed only a single nucleus in yng2 mutant cells that failed to sporulate, suggesting an early meiotic defect.

Yng2p-dependent NuA4 Activity in Mitosis and Meiosis
though some yng2 mutant cells were able to reenter G 1 by 120 min, ϳ48% of yng2 mutant cells persisted with a 2 N DNA content for the duration of the experiment (compare Fig. 2B, right with 3A, right). These data suggest that the growth defect observed in yng2 mutant cells is largely because of slow mitotic progression.
Loss of Yng2p-dependent NuA4 HAT Activity Inhibits Meiosis-Previous studies by Park and Szostak (30) showed that sporulation efficiency is dramatically reduced in amino-terminal histone H4 mutants, suggesting that yng2 mutants might also be defective for meiosis. To address this possibility, we determined sporulation efficiency for diploid wild-type or yng2 homozygous mutant cells. After a 3-day incubation on sporulation media, only ϳ1.5% of yng2 mutant cells formed four-spore asci compared with ϳ30% in wild-type cells (Fig. 2C). Staining the yng2 mutant cells with DAPI revealed that, except for the few cells that successfully formed four spores, yng2 mutant cells arrested unbudded and with a single nuclear mass (Fig.  2C), suggesting that Yng2p-dependent NuA4 activity is critical for early meiotic progression.
Transcription Is Largely Unchanged in yng2 Mutants-Histone acetylation has been shown to be important for transcriptional activation (31,32). In general, increased histone acetylation is associated with highly expressed genes, whereas deacetylation is coupled with transcriptional repression (33). We hypothesized that hypoacetylation, leading to decreased expression of mitotic cyclins, might underlie the cell cycle delay in yng2 mutants. We performed Northern analyses using wildtype and yng2 mutants synchronously released from ␣f-mediated G 1 arrest at 22°C. The G 1 cyclin CLN2 appeared with kinetics similar to the wild type but persisted longer, the S phase cyclin CLB5 is expressed at nearly normal levels and with similar kinetics to wild-type cells, and the expression of CLB2 in yng2 mutants began later and peaked at a lower level but persisted longer than in wild-type cells (Fig. 3A). In addi-tion, transcription of CLB1 was not detectable in yng2 mutants but occurred with similar kinetics and abundance as CLB2 in wild-type cells. These changes in timing and abundance correspond to the kinetics of S phase entry and mitotic progression revealed by flow cytometry (Fig. 3B).
With a view toward a comprehensive and quantitative examination of transcription in yng2 mutants, we used Affymetrix DNA oligonucleotide microarray analysis to compare genome-wide expression levels between asynchronously growing wild-type and yng2 mutant cells at 22°C. 83% of all probe sets were detected in this experiment. Surprisingly, only 54 annotated ORFs (Table I), 9 non-annotated ORFs from SAGE analysis, the rRNAs (5, 25, and 35 S), and 1 snRNA (SNR31) were decreased by 2-fold or more. Nearly 87% of transcripts were unchanged (Ͻ2-fold difference) relative to the wild type. Suspected targets of Esa1p identified by chromatin immunoprecipitation experiments (18) all fell in the unchanged class. Of the genes having significantly decreased expression in a yng2 mutant, the only known mitotic regulator is CLB1, confirming the Northern analysis. Expression of the other cyclins and known cell cycle regulators was Ͻ2-fold different between wild type and yng2 mutants. A number of genes important in biosynthetic pathways also fell within the 54 down-regulated ORFs (Table I). However, we observed similar growth defects in yng2 mutants grown on rich or synthetic media, suggesting that down-regulation of these genes is unlikely to contribute to the yng2 mutant phenotypes (data not shown).
The Mitosis-promoting Cyclin, Clb2p, Persists during the Mitotic Delay in yng2 Mutant Cells-Previous studies have shown that CLB2 is the major mitosis-promoting cyclin (34). In the absence of CLB2, cells perform G 1 and S phase functions on schedule but then delay at mitosis (34,35). To test whether the lower peak CLB2 mRNA expression in yng2 mutants led to a corresponding under-expression of Clb2p protein, we performed Western analysis on cells collected at 15-min intervals after release from an ␣f-induced G 1 block at 37°C. To limit growth to a single cell cycle, ␣f was added after S phase onset, trapping cells in the subsequent G 1 . From G 1 and through onset of replication, as indicated by a rightward shift in the flow profile, Clb2p is absent from both yng2 and wild-type cells (Fig. 4, A and B). By 30 min after release, wild-type cells replicated their genome, as indicated by a 2 N DNA peak in the flow profile, and Clb2p began to accumulate. By 45 min, wildtype cells displayed maximal levels of Clb2p protein and entered mitosis, and by 60 min Clb2p protein was destroyed (Fig.  4A). Similarly, yng2 mutant cells expressed Clb2p protein upon completion of S phase at 60 min and displayed a peak in Clb2p 15 min later. However, unlike the wild type, Clb2p protein in yng2 mutant cells was not rapidly destroyed but remained high through 105 min (Fig. 4B). Most yng2 mutant cells persisted with a 2 N DNA content during this interval. Importantly, both 2 N cells and Clb2p protein remained through 170 min (Fig. 4B).
Yng2p Function Is Required prior to Mitosis-The temperature-sensitive mitotic delay in yng2 mutants suggests at least two possibilities for the time at which Yng2p performs its function(s). Yng2p activity may be necessary during anaphase for normal mitotic progression. Alternatively, advancement through mitosis may depend on pre-mitotic events mediated by Yng2p. To distinguish between these two possibilities, we analyzed the mitotic delay by applying an execution point test used by Hartwell and colleagues (36) to characterize cdc mutants. When synchronously growing CDC mutant cells are shifted to nonpermissive temperature before reaching their execution point, these cells arrest homogeneously with a characteristic terminal morphology within the first cell cycle. How-

FIG. 3. Transcription of the mitotic cyclins is decreased, whereas G 1 and S phase cyclins are normal in yng2 mutant cells.
A, Northern analysis was performed on RNA isolated from wild-type (YNG2) and yng2 mutant cells collected after ␣f-induced G 1 block and release at 22°C. B, cells used in A were also analyzed by flow cytometry to determine cell cycle position. ␣f was added back after cells entered S phase (ϳ60 min) to trap cells in the subsequent G 1 phase (ϳ120 min). ever, if mutant cells pass the execution point prior to the temperature shift, cells will complete the first cell cycle and then arrest in the second cell cycle.
To determine whether the temperature-sensitive mitotic delay reflected a requirement for Yng2p during or prior to mitosis, we performed nocodazole arrest and release experiments at 22°C or 37°C. Wild-type and yng2 mutant cells were arrested with nocodazole at 22°C and then left at 22°C or shifted to 37°C before release from nocodazole into ␣f to trap cells in the subsequent G 1 . We found that when yng2 mutants are released from a nocodazole block at 37°C, they progress through mitosis and reenter G 1 with nearly identical kinetics to wild-type cells (Fig. 5, A and B). Importantly, Clb2p protein accumulated to similar levels in both wild-type and yng2 mutant cells during the nocodazole arrest and was then destroyed with similar kinetics after release (Fig. 5C). These results suggest that Yng2p performs its critical function prior to anaphase and that delaying mitotic progression can relieve the defect.

Yng2p Is Constitutively Expressed and Remains Nuclear
Localized throughout the Cell Cycle-Suppression of the mitotic delay in yng2 mutants after nocodazole arrest-release suggested a requirement for Yng2p-dependent NuA4 activity during or prior to mitosis. To determine whether expression or activity of Yng2p might be cell-cycle regulated, we examined YNG2 mRNA and protein abundance. In each of the following analyses, ␣f was added back at 40 min after cells entered S phase, trapping cells in the subsequent G 1 by 120 min. Northern analysis of YNG2 expression in ␣f synchronized cultures revealed uniform levels of YNG2 mRNA throughout the cell cycle (data not shown), confirming the results of microarray analysis (37). Western analysis performed with extracts collected from ␣f-synchronized cells expressing a single integrated copy of YNG2-13Myc revealed that Yng2p-Myc protein was present at relatively constant levels throughout the cell cycle (Fig. 6A). Alternatively, Yng2p-associated NuA4 HAT activity might be up-regulated during or before mitosis. In vitro NuA4

Yng2p-dependent NuA4 Activity in Mitosis and Meiosis
HAT activity reactions performed with Yng2p-Myc immune complexes isolated from cells synchronized in an identical manner revealed relatively similar levels of activity throughout the cell cycle (Fig. 6B). Although Loewith et al. (12) previously observed that GFP-Yng2p was nuclear localized in asynchronously growing cells, it remained possible that Yng2p might shuttle between the nucleus and cytoplasm during a specific cell cycle stage. To test this possibility we followed cells expressing GFP-Yng2p that were arrested with ␣f and then allowed to synchronously enter the cell cycle to more precisely determine whether any changes occurred in GFP-Yng2p localization. Fluorescent microscopic analysis of these cells revealed that GFP-Yng2p protein remained exclusively in the nucleus throughout the cell cycle (Fig. 6C). We found that GFP-Yng2p colocalized with DAPI staining in asynchronously growing cells, demonstrating that GFP-Yng2p is nuclear localized (Fig. 6D).
Immunofluorescence Reveals Cell-autonomous Differences in Acetylated Histone H4 Isoforms in yng2 Mutants-The measured decrease in total acetylated histone H4 in yng2 mutants (12) may reflect a proportional loss of acetylated histone H4 in every cell or a more dramatic loss in only a subpopulation of cells. To distinguish between these two possibilities we performed indirect immunofluorescence on asynchronously growing wild-type and yng2 mutant cells with polyclonal rabbit antibodies specific to amino-terminally acetylated isoforms of histone H4. In wild-type cells immunostaining for histone H4 tetraacetylated at lysines 5, 8, 12, and 16 (tH4) or acetylated at lysine 8 (H4K8) or lysine 12 (H4K12) was observed in each nucleus at all points in the cell cycle at 22°C (Fig. 7, A-F) or 37°C (Fig. 7, M-R). However, qualitatively stronger immunostaining was observed for each acetylated histone H4 isoform in G 1 or S phase cells (i.e. unbudded, small, and medium budded cells) compared with G 2 or mitotic cells (i.e. large budded cells). When an identical set of experiments was performed with yng2 mutants, we observed dramatic differences in immunostaining for each acetylated histone H4 isoform in different cells in the population with no strict correlation to cell cycle position. At 22°C, ϳ3% of yng2 mutant cells displayed a complete absence of H4K12 immunostaining, whereas others displayed qualitatively reduced or wild-type levels of immunostaining (Fig. 7, H   FIG. 4. Clb2p protein is expressed on schedule and accumulates to wild-type levels in yng2 mutant cells. A and B, using a polyclonal anti-Clb2p antibody, Clb2p protein was analyzed from wildtype and yng2 mutant cells treated exactly as described in the legend for Fig. 2. Expression of Clb2p in yng2 mutants occurs on schedule and at similar levels as in wild-type cells. However, unlike wild-type cells, Clb2p protein levels persist in yng2 mutant cells during the latter half of the time course and begin to decline only when cells reenter G 1 . Below each Western analysis are flow profiles that show the cell cycle position of the cells at each time point. Dots below each flow profile denote DNA content; a single dot is 1 N, and double dots are 2 N. ␣f was added back after cells entered S phase (ϳ60 min), to trap cells in the subsequent G 1 phase.

FIG. 5.
A nocodazole-mediated delay relieves the M/G 1 delay in yng2 mutants, suggesting that Yng2p functions prior to or during metaphase. A and B, cells were first arrested in nocodazole (15 g/ml) at 22°C and released into YPD containing ␣f (5 M) at 22°C or 37°C. Both wild-type cells and yng2 mutant cells return to G 1 by 120 min at 22°C or 60 min at 37°C. C, Clb2p protein is rapidly destroyed in wild-type and yng2 mutant cells released from nocodazole arrest at 37°C.
FIG. 6. Yng2p protein expression, associated NuA4 HAT activity, and nuclear localization are largely unchanged throughout the cell cycle. A, wild-type cells (Yng2p-Myc) were collected after ␣f-induced G 1 block and release. ␣f was added back after cells entered S phase (ϳ40 min), to trap cells in the subsequent G 1 (ϳ120 min). Western analysis demonstrates that Yng2p-Myc protein abundance remains relatively similar throughout the cell cycle. B, Yng2p-Mycassociated HAT activity assayed from cells treated as described in A remains relatively unchanged throughout the cell cycle. C, GFP-Yng2p localization was followed in synchronously (␣f arrest/release) growing cells. In each case, GFP-Yng2p remained exclusively nuclearly localized. DIC, differential interference contrast. D, DAPI staining of asynchronously growing cells expressing GFP-Yng2p demonstrates that GFP-Yng2p is nuclearly localized.

Yng2p-dependent NuA4 Activity in Mitosis and Meiosis
and K). When yng2 mutants were grown at 37°C, ϳ28% of cells displayed an absence of H4K12 immunostaining, and the remainder of the cells exhibited a relative reduction in H4K12 signal compared with wild-type cells (Fig. 7, T and W). Similarly, immunostaining for tH4 in yng2 mutant cells grown at 22°C (Fig. 7, I and L) and 37°C (Fig. 7, U and X) was significantly compromised. Approximately 40% of yng2 mutants at 37°C displayed no tH4 immunostaining, whereas the remaining cells contained tH4 immunostaining that was reduced relative to wild-type cells (compare Fig. 7, O and R with U and X). Surprisingly, the distribution of H4K8 immunostaining in yng2 mutants was less heterogeneous than that of H4K12 or tH4. Virtually all yng2 mutants grown at 22°C (Fig. 7, G and J) or 37°C (Fig. 7, S and V) possessed H4K8 immunostaining, al- FIG. 7. Cell-autonomous levels of acetylated forms of histone H4 are observed in yng2 mutant cells. A-X, indirect immunofluorescence performed on wild-type or yng2 mutant cells grown at 22 or 37°C using polyclonal antibodies specific for tetraacetylated histone H4 at lysines 5, 8, 12, 16 (tH4), specific for acetylated histone H4 at lysine 8 (H4K8), and specific for acetylated histone H4 at lysine 12 (H4K12). DIC (differential interference contrast) indicates visualization by Nomarski optics, and IF indicates immunofluorescence. Immunostaining for all three isoforms of histone H4 was present in wild-type cells at all stages of the cell cycle when grown at 22°C (A-F) or 37°C (M-R). In contrast, cell-to-cell differences in immunostaining for each H4 isoform in yng2 mutants were observed. A complete absence or severe reduction of H4K12 immunostaining was observed in a small population of yng2 mutants at 22°C and this increased at 37°C. Qualitatively reduced or even wild-type levels of H4K12 immunostaining was observed in the remaining cells. H4K8 immunostaining in yng2 mutants was less heterogeneous than that of H4K12 in that virtually all cells contained H4K8 at 22°C (G and J) and 37°C (S and V), although it was qualitatively reduced relative to wildtype cells. A reduction in immunostaining of tH4 was also observed in yng2 mutants at both 22°C and 37°C, with a large population of cells displaying an absence of tH4 immunostaining (arrows). The exposure times for photographs of immunofluorescence were identical for wild-type and yng2 mutant cells and have not been manipulated.
though it was qualitatively reduced relative to wild-type cells.
When identical antibody immunostaining experiments were performed with cells carrying only nonacetylatable mutants of histone H4 (6), no immunostaining was observed, demonstrating that the apparent differences between yng2 mutant and wild-type cells are specific (data not shown). We also performed immunofluorescence to detect acetylated forms of histone H3 in wild-type and yng2 mutant cells, because defects in NuA4 should not affect H3 acetylation. Indeed, identical staining patterns for acetylated H3 were observed in both wild-type and yng2 mutants (data not shown). These findings demonstrate that in the absence of Yng2p, loss of specific isoforms of acetylated histone H4 is highly cell autonomous.
The Deacetylase Inhibitor Trichostatin A Suppresses the H4 Acetylation Defect and Mitotic Delay of yng2 Mutants-Abundance of acetylated isoforms of histone H4 in vivo is determined by competing acetylation and deacetylation activities (38). Considering that yng2 mutants have reduced NuA4 HAT activity, we tested whether inhibiting the antagonistic deacetylase activity might restore levels of acetylated isoforms of H4. Toward this end, we examined the levels of tH4 immunoreactivity in yng2 mutant cells grown in the presence of the deacetylase inhibitor TSA (39). Whereas yng2 mutant cells grown at 37°C displayed a marked decrease in tH4 immunostaining (Fig. 7, U and X), TSA treatment significantly suppressed this defect (Fig. 8A, right). Although 40% of yng2 mutant cells incubated at 37°C displayed an absence of tH4, TSA treatment decreased this proportion to 4%. To determine whether inhibiting deacetylation might also relieve the mitotic delay, we performed flow cytometry on synchronized wild-type and yng2 mutant cells collected after ␣f-induced G 1 block and release in the presence of TSA. Wild-type cells treated with TSA progressed through all stages of the cell cycle with kinetics similar to untreated cells (compare Fig. 2A with 8B). Remarkably, TSA-treated yng2 mutant cells no longer delayed in mitosis but reentered G 1 by 160 min at 22°C and 120 min at 37°C, yielding kinetics nearly identical to those of wild-type cells (compare Fig. 2B to Fig. 8B).

DISCUSSION
The NuA4 HAT complex comprises at least 11 subunits (9), and yet functional studies have been challenging as most subunits are individually essential for viability. We took advantage of the conditional viability of mutants lacking YNG2 to examine the role of NuA4-mediated histone H4 acetylation in cell cycle progression. We find that full NuA4 activity depends on Yng2p and is required for both mitotic and meiotic progression. In addition we discovered cell-autonomous differences in histone H4 acetylation in yng2 mutant cells rather then a uniform decrease in acetylated histone H4.
We found that Yng2p physically interacts with the NuA4 subunit Epl1p, confirming work by Loewith et al. (12) showing that Yng2p physically associates with Tra1, another NuA4 subunit. Yng2p-deficient cells display decreased total acetylated histone H4 in vivo (12), and in this report we find significantly reduced NuA4 HAT activity in vitro. Furthermore, we show that Epl1p-Ha and Esa1p coimmunoprecipitated in the presence or absence of Yng2p, suggesting that Yng2p is a component of the NuA4 complex that is necessary for normal NuA4 activity but not for complex formation. Analysis of cells synchronously released from an ␣f-mediated G 1 arrest revealed a mitotic delay associated with persistent Clb2p protein levels in yng2 mutants, indicating that full NuA4 activity is required for the onset of metaphase and/or anaphase. To determine the time at which Yng2p functions during the cell cycle, we applied the execution point paradigm used to characterize cdc mutants (36) to analyze the origin of the mitotic delay. Importantly, we show that a nocodazole-mediated metaphase arrest relieves both the mitotic delay and the accumulation of Clb2p protein observed in yng2 mutants when released from G 1 . This result suggests that the time at which Yng2p normally serves its function(s) most critical for cell cycle progression is likely to be at or before metaphase. We infer that the cell cycle defects may be a direct result of inadequate histone H4 acetylation rather than a NuA4-independent function of Yng2p. Perhaps NuA4mediated H4 acetylation of histone H4 reassembled onto nascent DNA during S phase is a critical event that allows mitosis to occur in a timely manner.
The loss of both acetylated histone H4 and NuA4 HAT activity in extracts from yng2 mutant cells would predict a proportional loss of acetylated histone H4 in every yng2 mutant cell. Unexpectedly, we discovered a heterogeneous distribution of acetylated histone H4 in yng2 mutant cells. Apparently, normal immunoreactivity for acetylated histone H4 was present in some cells, whereas it was completely absent from others. In asynchronously growing cells we did not observe a strict correlation between cell cycle stage or morphologically aberrant cells and a complete lack of histone H4 acetylation. However, following release at 37°C of synchronized yng2 mutant cells, we observed that the anaphase/telophase cells at late time points are particularly deficient for immunoreactivity of tetraacetylated histone H4. 2 One inference is that the heterogeneous mitotic delay and morphology of yng2 mutants may be determined on a cell-by-cell basis by the degree of histone H4 acetylation. Indeed, treatment of yng2 mutant cells with the deacetylase inhibitor TSA restores histone H4 acetylation to all cells and returns the population to nearly normal mitotic progression. Future studies to understand the basis for the cell autonomous differences in H4 acetylation may provide new insights into the roles and regulation of NuA4 through the cell cycle.
We envision at least three potential roles for Yng2p-dependent NuA4 histone H4 acetylation in mitotic and meiotic progression. First, Yng2p might participate in NuA4-dependent histone acetylation to promote transcription of specific genes. Second, Yng2p-dependent NuA4 HAT activity might facilitate genome-wide acetylation that ensures proper chromatin structure during mitotic and meiotic progression. Finally, Yng2pdependent NuA4 HAT activity may mediate acetylation of nonhistone substrates that regulate cell cycle progression or checkpoints.
Numerous studies have established that histone acetylation/ deacetylation modulates transcription (32,40). Indeed, NuA4 HAT activity stimulates transcription on nucleosomal array templates in vitro (41,42). However, to date, Esa1p has been implicated in the regulation of only a surprisingly small group of genes (11,18). The growth defect and other phenotypes of yng2 mutants may well derive from repression of critical genes. Suggesting that the mitotic delay might originate from a transcriptional defect, expression of the mitotic cyclin CLB1 was undetectable by Northern analysis. However, cells deficient for CLB1 demonstrate no vegetative growth phenotypes (34). We also observed a delay in the onset of transcription and lowered peak expression level of the mitotic cyclin CLB2. Despite the altered CLB2 transcription pattern, Western analysis demonstrated that Clb2p appeared on schedule after completion of S phase and accumulated to normal abundance in yng2 mutant cells. Potentially, other cell cycle regulators under similar transcriptional regulation to CLB2 (e.g. members of the Clb2 cluster (37)) are also NuA4 targets, and their delayed expression may underlie the cell cycle defects. Nonetheless, Yng2p deficiency confers surprisingly little effect on transcription. Microarray analysis revealed that expression of most genes, including nearly all members of the Clb2 and Sic1 clusters, differs by less than 2-fold between wild-type and yng2 mutant cells. Of the 54 genes repressed 2-fold or more in the yng2 mutant, CLB1 was the only characterized cell cycle regulator.
The meiotic arrest in yng2 mutant diploids may arise from a specific defect in Clb1p expression, a gene that is essential for proper meiotic progression (43,44). However, no obvious transcriptional defect explains the observed cell cycle delays in vegetative growth. Perhaps the genome-wide loss in H4 acetylation per se contributes to the mitotic delay in yng2 mutants. Indeed, histone H4 acetylation in large regions of chromatin is greatly reduced in esa1 mutants, suggesting that NuA4 activity is normally responsible for acetylation over large chromatin domains (19). Our studies provide further evidence for a NuA4 role in genome-wide acetylation. Total acetylated histone H4 is reduced in yng2 mutant cell extracts (12), and immunofluorescence reveals a complete loss of tetraacetylated histone H4 in many yng2 mutant cells. Furthermore, GFP-Yng2p is localized throughout the nucleus rather then to specific loci. Regulation of chromatin structure by histone acetylation is likely a critical process during mitosis and meiosis, as each require dramatic structural rearrangements such as chromosome pairing, condensation, and decondensation. Future studies of the role of acetylation in chromosome structure and function during mitosis and meiosis may shed light on the defects seen in yng2 mutants.
Finally, although we have no evidence for non-histone substrates of NuA4, this remains more than a formal possibility. Studies from metazoans suggest that the function of key cell cycle regulators such as p53 is modulated by acetylation (45). Because the human homolog of Yng2p, ING1, associates with p53 in vivo (46), one possibility is that ING1 is a subunit of a multisubunit HAT complex in humans that also participates in regulation of or by p53.