IDENTIFICATION OF HISTONE H3 LYSINE 36 ACETYLATION AS A HIGHLY CONSERVED HISTONE MODIFICATION^*

Yeast strains

The wild-type histone H3 and H3K36-to-alanine point mutation (K36A) strains were in the WZY42 background and have been described previously (34). The TAP-tagged GCN5 and SAS3 strains were in the BY4741 background and obtained from Open Biosystems. The wild-type (DY150), gcn5Δ (DY5925) (35), and sas3Δ (DY8179) strains were in the W303 background and were kindly provided by David Stillman (University of Utah).

Histone preparation

Yeast, Tetrahymena and mammalian histones were acid-extracted from purified nuclei as previously described (36,37). For mass spectrometric analyses, Tetrahymena and yeast histone H3 were purified by reverse phase (RP)-HPLC as previously described (38,39). Typically, acid-extracted core histones were separated on a C8 column (220 X 4.6 mm, Aquapore RP-300; PerkinElmer) using a linear ascending solvent B gradient of 35–60% over 75 min at 1.0 ml/min on a Beckman Coulter System Gold 126 Pump Module and 166 Detector (solvent A was 5% acetonitrile/0.1% TFA in water, and solvent B was 90% acetonitrile/0.1% TFA in water). Peak fractions corresponding to H3 were collected, dried and resuspended in water. A portion of each fraction was used to confirm the presence of purified histones by gel electrophoresis followed by Coomassie blue staining. The remainder of the fraction was used for mass spectrometric (MS) analyses.

Mass spectrometric analyses

Mass spectrometric analyses of Tetrahymena and yeast H3 were performed as previously described with slight modifications (36). Briefly, Tetrahymena H3 was digested with trypsin prior to propionylation while yeast H3 lysines were blocked by propionylation followed by digestion with either trypsin or chymotrypsin. Derivatization of H3 with propionylation reagent converts internal lysine residues (monomethylated and endogenously unmodified residues) and the amino terminus to propionyl amides causing the blockage of trypsin cleavage on the C-terminal side of lysines allowing cleavage to occur only C-terminal to arginine (40). Digestion of Tetrahymena H3 prior to propionylation allowed for the generation of peptides distinct from that of yeast H3 improving the chances of detecting posttranslational modifications. Following digestion and propionylation reactions, samples were gradient eluted (Agilent 1100 Series) directly into a Finnigan linear quadrupole ion trap-Fourier transform (LTQ-FT) mass spectrometer (Thermo Electron) at a flow rate of 100 nl/min operated in the MS/MS data-dependent mode. All MS/MS data were manually validated by inspection of band y-type ions.

Electrophoresis and immunoblot analyses

Sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and immunoblot analyses were performed using procedures and reagents from GE Healthcare. The anti-H3K36me3 (α-H3K36me3) rabbit polyclonal antibody was obtained from Abcam (catalog # ab9050) and used at a dilution of 1:2,000–1:5,000. All other rabbit histone modification-specific antibodies were obtained from Upstate Biotechnology and used at the following dilutions: 1:10,000 for H3K18ac (α-H3K18ac, catalog # 07–354), 1:5,000 for H3K14ac (α-H3K14ac, catalog #07–353), and 1:15,000–1:30,000 for the C-terminus of H3 (α-H3, catalog # 07–690). The H3K36ac antiserum (α-H3K36ac, 1:1,000–1:10,000 dilution) was developed by immunizing a rabbit with a synthetic KLH-conjugated peptide specific for acetylation at H3K36 (C-APATGGVKacKPH). Typically, histones or synthetic histone H3 peptides were resolved on SDS-PAGE gels (15% for histones and 10% for peptides), followed by transfer onto polyvinylidene difluoride (PVDF) membranes (Millipore). In some cases, membranes were stained with Ponceau S (Sigma-Aldrich) to ensure proper protein transfer. After incubation with primary antibody and addition of a horseradish peroxidase-conjugated secondary antibody (GE Healthcare), membranes were incubated with ECL-Plus substrate (GE Healthcare), and proteins were detected by exposure to X-ray films.

For peptide competition experiments, the H3K36ac antibody was incubated overnight at 4°C with 5 µg/ml of the following peptides: unmodified histone H3 (amino acids 1–21), K9/14ac histone H3 (amino acids 1–21), K14ac histone H3 (amino acids 1–21), unmodified histone H3 (amino acids 27–46), and K36ac histone H3 (amino acids 27–46). Peptides were either obtained from Upstate Biotechnology or synthesized and verified by mass spectrometry at the University of North Carolina Microprotein Sequencing and Peptide Synthesis Facility.

Chromatin immunoprecipitation (ChIP) and DNA microarray (ChIP-chip) analyses

ChIP assays were performed as previously described (33). Antibodies and amounts used in the immunoprecipitations (IPs) are as follows: α-H3K36ac (2 µl/IP, available from Upstate Biotechnology as immunoaffinity purified antibody), α-H3K36me2 (3 µl/IP, Upstate Biotechnology, catalog # 07–274), and α- H3K9/14ac (3 µl/IP, Upstate Biotechnology, catalog # 06–599). Following DNA recovery, the ChIP-enriched DNA were amplified as described (41). Briefly, two initial rounds of DNA synthesis with T7 DNA polymerase using primer 1 (5'-GTTTCCCAGTCACGATCNNNNNNNNN-3') was followed by 25 cycles of PCR with primer 2 (5'-GTTTCCCAGTCACGATC-3'). Cy3-dUTP or Cy5-dUTP were then incorporated directly with an additional 25 cycles of PCR using primer 2. Direct microarray hybridizations of H3K36ac ChIP vs. H3K36me2 ChIP or H3K9/K14ac ChIP vs. H3K36me2 ChIP were performed using standard procedures (42). This method allowed for the direct comparison between the histone modification patterns, which showed that H3K36ac enrichment was preferentially in the promoter regions of genes while H3K36me2 enrichment was preferentially in the transcribed regions. Additional control experiments in which the reference DNA were from H3 ChIPs (for standard nucleosome occupancy normalization), H3K36ac ChIPs from a H3K36 point mutation strain (K36A), or genomic DNA also demonstrated the H3K36ac enrichment to be promoter specific (data not shown). Following hybridizations, the arrays were scanned with a GenePix 4000B scanner and data was extracted with Genepix 5.0 software. Data were normalized such that the median log₂ ratio value for all quality elements on each array equaled zero, and the median of pixel ratio values was retrieved for each spot. Only spots of high quality by visual inspection, with at least 50 pixels of quality data (regression R² > 0.5) were used for analysis. All data was log-transformed before further analysis. For ChIP-chip data analyses, the log₂ ratio of each spot was transformed to a z-score using the formula z_x = (X-µ)/σ, where X is a retrieved spot value, µ is the mean of all retrieved spots from one array, and σ is the standard deviation of all retrieved spots from that same array. Z-scores from three biological replicates were averaged. Raw data can be obtained from the University of North Carolina Microarray Database at https://genome.unc.edu. Data are also available through GEO (accession number GSE5544).

Purification of native yeast histone acetyltransferase (HAT) complexes

For purification of SAGA and NuA3, 4 L of yeast cells containing Gcn5-TAP or Sas3-TAP were grown to an OD₆₀₀ between 1.0 and 2.0. Cells were disrupted by glass bead-beating in a breaking buffer consisting of 50 mM Tris-HCl (pH 8.0), 350 mM NaCl, 10% glycerol, 0.1% triton-X-100 and protease inhibitors (PMSF, aprotinin, leupeptin, and pepstatin) followed by clarification of the extract by ultracentrifugation at 25,000 rpm for 1 h at 4°C as described (43). TAP-tagged proteins were purified in a one-step procedure from the extracts by directly binding the calmodulin binding peptide (CBP) component of the TAP tag to calmodulin resin (Stratagene; 200 µl beads) in the presence of CaCl₂ (1 mM final concentration) at 4°C for 4 hours. All wash and elution steps were performed in 0.8 × 4 cm Polyprep chromatography columns (BioRad). Protein-bound resin was washed two times with breaking buffer containing CaCl₂ at a final concentration of 1 mM. Following washes, bound proteins were eluted in twelve 250 µl fractions with elution buffer (50 mM Tris-HCl (pH 8.0), 350 mM NaCl, 10% glycerol, 2 mM EGTA, and protease inhibitors). Generally, the peak of complex elution was found in fractions 2 and 3. Purified complexes were analyzed by Coomassie blue staining, immunoblot analysis with an anti-Protein A antibody (Sigma-Aldrich) and in vitro HAT assays (see below). Additionally, an untagged wild-type strain was included in the purification procedure as a control to confirm that there were no contaminating activities due to nonspecific protein interactions with the calmodulin resin.

HAT assays

Acetyltransferase assays were performed as described previously with modification (43,44). Briefly, 3 µl of TAP-purified SAGA or NuA3 complex (fraction 2) was incubated with either 2 µg chicken core histones, 2 µg recombinant mononucleosomes, or 2 µg synthetic H3 peptides along with 30 µM unlabeled acetyl coenzyme A (acetyl-CoA, Sigma-Aldrich) in acetyltransferase buffer (50 mM Tris (pH 7.4), 10% glycerol, and protease inhibitors). Reactions were allowed to proceed for 30 min at 30°C in a total volume of 12 µl and analyzed by SDS-PAGE followed by Coomassie blue staining or immunoblotting. For radiolabeled HAT assays, 3 µl of TAP-purified SAGA or NuA3 complex was incubated with 2 µg synthetic H3 peptides along with 0.125 µCi [³H] acetyl coenzyme A (2–10 Ci/mmol, GE Healthcare) in acetyltransferase buffer. Reactions were allowed to proceed as described above followed by spotting of reaction products onto p81 Whatman paper (Fisher Scientific) and monitoring of ³H incorporation by scintillation counting (filter-binding assay). Recombinant mononucleosomes were a gift from Song Tan (Pennslyvania State University) and consisted of Xenopus core histones that were bacterially expressed, purified, and reconstituted with the NucB region of the MMTV promoter.

Histone H3 is acetylated at lysine 36 in Tetrahymena and yeast

Methylation at H3K36 is a highly conserved modification found in eukaryotes ranging from Tetrahymena to humans, but whether other types of modifications occur at this residue was unknown. To determine if novel modifications occur at H3K36, we analyzed H3 from Tetrahymena by tandem mass spectrometry. This organism, a ciliated protozoan, has proven useful in the discovery of other “ON” histone covalent modifications, notably H3K4me (45). Tetrahymena H3 was purified by RP-HPLC from acid-extracted bulk histones and digested with both trypsin and chymotrypsin followed by chemical derivatization using propionic anhydride reagent (46). Digests were analyzed by on-line nanoflowLC-MS/MS on a linear quadrupole ion trap-Fourier transform (LTQ-FT) mass spectrometer. Using this platform, we determined H3K36 is acetylated.

Shown in Fig. 1A is the MS/MS spectrum of a peptide produced from the propionylated tryptic digest of Tetrahymena histone H3. The [M+2H]²⁺ of the parent ion is shown at m/z 542.3062. The high mass accuracy of the LTQ-FT mass spectrometer can easily distinguish between trimethylation and acetylation on peptides (Δm = 0.0364 Da). The accurate mass of the parent ion recorded was found to be consistent with acetylation on this peptide (+0.18 ppm) and not trimethylation. Importantly, we were able to also detect H3K36 mono-, di-, and trimethylation on other H3 peptides (data not shown), confirming that this residue can be methylated or acetylated.

We additionally surveyed histone H3 purified from budding yeast to determine if H3K36 acetylation might be present in this distinct unicellular organism. Using the approach mentioned above for Tetrahymena H3, we identified that yeast H3K36 was also acetylated (Fig. 1B). H3K36 acetylation was observed on a number of different peptides. Fig. 1B displays the tandem mass spectrum of the [M+2H]²⁺ ion at m/z 823.4552 from a propionylated digest of one such identified H3 peptide. This peptide was identified to be the 27–40 amino acid H3 fragment and it was found to contain acetylation at both H3K27 and H3K36. These results identify a novel acetylation event at H3K36 that is conserved between Tetrahymena and yeast.

To further investigate the occurrence of this modification, we raised an antibody against a synthetic peptide acetylated at H3K36 and analyzed Tetrahymena and yeast histones by immunoblot analysis. As shown in Fig. 2A, the α- H3K36ac antibody efficiently recognized the RP-HPLC purified Tetrahymena H3, originally analyzed by mass spectrometry. We determined the antibody to be specific for H3K36ac, as only a synthetic H3 peptide acetylated at H3K36 could selectively compete away the signal detected by the α-H3K36ac antibody (Fig. 2A). Other unmodified or acetylated H3 peptides did not compete for this antibody’s detection of H3.

Next, we determined if this antibody could specifically recognize acetylation at H3K36 in yeast. Yeast histones isolated by acid-extraction from a wild-type or H3K36 point mutant (K36A) strain were analyzed by immunoblot analysis using the H3K36ac-specific antibody. Similar to Tetrahymena, the α-H3K36ac antibody detected H3 in a wild-type yeast strain but much less so in a strain where H3K36 was mutated to alanine (Fig. 2B). However, we note that high concentrations of histones loaded in these assays results in the ability of the α-H3K36 antibody to weakly recognize the backbone of H3 (Fig. 2B, K36A lane). Nonetheless, these results confirm our mass spectrometry findings that H3K36ac exists in both Tetrahymena and yeast.

H3K36 acetylation is preferentially enriched in the promoters of RNA polymerase II-transcribed genes genome-wide

Next, we used a ChIP-chip approach to determine the genomic distribution of H3K36ac and how it compared to the distribution of methylation found at H3K36 (47,48). The α-H3K36ac-specific antibody was used in ChIP reactions from yeast whole cell extracts. Enriched genomic DNA fragments were treated with RNase, amplified, and labeled fluorescently. DNA enriched from H3K36me2 ChIPs were prepared in a similar manner, and both samples were hybridized on the same array. Three independent sets of ChIPs were performed. Using this method, we directly compared the patterns of H3K36ac and H3K36me2 using arrays that tiled continuously over the entire genome at a resolution of ~1 kb. We observed a preferential enrichment of H3K36ac at 5′ regulatory regions (bidirectional and unidirectional promoters) relative to coding regions (ORFs) and 3′ UTRs (non-coding region downstream of two convergently transcribed genes) (Fig. 3A). Significantly, the H3K36ac pattern was found to be inversely related to that of H3K36me2, which occurs preferentially in the coding region and 3′ UTR of genes (33,48,49).

We then compared the location of the H3K36ac modification to that of other well-characterized acetylation sites on H3; namely H3K9 and H3K14 acetylation (H3K9/14ac). Using the same extracts, ChIPs were performed with an antibody that recognizes diacetylated H3K9/14, and the enriched DNA was amplified, labeled, and hybridized to DNA microarrays as described above. These experiments revealed that H3K9/K14ac was localized to promoters in a pattern strikingly similar to the pattern we observed for H3K36ac genome-wide (data not shown and see Fig. 3B). These data were also consistent with the H3K9/14 acetylation results obtained by others (47,50). We further examined the distribution of H3K36ac, H3K9/K14ac, and H3K36me2 using a high-resolution microarray containing probes that covered all of chromosome III at a resolution of 200 bp (and at 100 bp resolution for 1/3 of the chromosome). For both H3K36ac and H3K9/K14ac, the level of acetylation enrichment drops sharply as a function of distance from translational initiation site while H3K36me2 increases (Fig. 3B). These data are fully consistent with our analysis using lower-resolution arrays and confirm that these acetyl marks occur preferentially upstream of coding regions while H3K36me2 occurs primarily in the coding regions of genes.

We next asked whether H3K36ac associates with genomic regions other than those characteristic of RNA polymerase II (RNAPII) promoters. We found that genomic regions which were transcriptionally silent under our growth conditions, or regions not transcribed by RNAPII including centromeres, telomeres, and silent mating type loci, contained low levels of H3K36ac and H3K9/K14ac (Fig. 3C). These results indicate that H3K36ac is associated strictly with active RNAPII regulatory sequences, and suggest that like H3K36me, H3K36ac may function in RNAPII-mediated gene transcription.

The Gcn5-containing SAGA histone acetyltransferase complex mediates H3K36 acetylation

We next sought to identify the responsible histone acetyltransferase(s) (HATs) that mediate this mark. Given H3K36 acetylation is enriched in the promoter regions of RNAPII-transcribed genes, and has a pattern of acetylation similar to that of H3K9/14 acetylation, we initially focused on the Gcn5-containing SAGA histone acetyltransferase (HAT) complex that mediates acetylation to the N-terminus of H3 (43). Furthermore, we noticed that the amino acid sequence surrounding H3K36 is very homologous to a preferred Gcn5 consensus site of acetylation found at H3K14 (51) (compare STGGK14AP vs. STGGVK36KP; underlined residues indicate homology and bold “K” indicates the residue targeted for acetylation).

To test if SAGA would mediate H3K36ac, we TAP-purified SAGA from yeast whole extracts using a tagged form of Gcn5 and then incubated the complex with either unmodified or modified H3 synthetic peptides along with unlabeled acetyl- CoA (acetyl donor) in a HAT assay. Following the HAT reactions, the products were electrophoresed on 10% SDS-PAGE gels and analyzed by immunoblot with the α-H3K36ac antibody. After incubation with purified SAGA, a previously unmodified H3 27–46 amino acid peptide was recognized by the H3K36ac-specific antibody (Fig. 4A). Importantly, when these same assays were performed using a radiolabeled form of acetyl-CoA, SAGA was readily able to acetylate an N-terminal H3 peptide (residues 1–21 containing H3K9 and H3K14) as well as the unmodified 27–46 peptide, but showed no activity toward a matched 27–46 peptide acetylated at H3K36 (Fig. 4B).

To learn more about the physiological relevance of SAGA-mediated H3K36 acetylation, we next asked if SAGA could acetylate H3K36 in the context of other histone proteins and DNA. Again HAT assays were performed using TAP-purified SAGA in which the complex was incubated with either free chicken core histones or recombinant mononucleosomes. As shown in Fig. 4C, SAGA was capable of acetylating both substrates at H3K36.

We also tested the NuA3 HAT complex for an activity that specifically acetylates H3K36. Like SAGA, NuA3 has been identified as a HAT complex that specifically targets H3 for acetylation (52). Previous work shows that the catalytic subunit of NuA3, Sas3, has overlapping activities with Gcn5 and can target H3K14 for acetylation in vivo (53). Unlike SAGA, however, we found that TAP-purified NuA3 was unable to acetylate H3K36 in the context of H3 synthetic peptides. However, NuA3 could acetylate H3K36 in the context of core histones and nucleosomes (data not shown).

Given the ability of SAGA and NuA3 to catalyze H3K36ac, we next asked whether either one of these HATs was required for H3K36ac in vivo. We therefore purified bulk core histones from GCN5 and SAS3 deletion strains and analyzed them for H3K36ac by immunoblot analysis. As shown in Fig. 4D, acetylation at H3K36 was abolished in the GCN5 deletion, but not in the SAS3 deletion. These results reveal that Gcn5 is the major HAT responsible for H3K36ac in vivo.

H3K36 acetylation is conserved in mammalian cells

Although our studies showed that several unicellular organisms contained H3K36ac, we wanted to determine how conserved this modification would be among several diverse multicellular organisms. We therefore isolated histones by acid-extraction from Tetrahymena, yeast, mouse, and human cells and probed them for the presence of H3K36ac. As shown in Fig. 5A, we found H3K36ac in all of the organisms analyzed, although the relative abundance varied between species. H3K36ac appeared to be most abundant in Tetrahymena and yeast, followed by human and mouse. The mouse embryonic fibroblasts used for this study apparently harbor a very low level of H3K36ac, which is consistent with our mass spectrometry experiments that did not detect H3K36ac in histones isolated from these cells (see 54). In agreement with our human results, a recent mass spectrometry proteomics survey reported the existence of H3K36ac in human HeLa cells (55). Although differences exist in the relative abundance of H3K36ac found in these organisms, and perhaps within different cell types, these results reveal that H3K36ac occurs in organisms as diverse as yeast and humans. Our results also underscore differences in the employment of histone marks in different organism and in the importance of applying independent assays to assess them.