Histone Acetylation and Deacetylation

The acetylation isoforms of histone H4 from butyrate-treated HeLa cells were separated by C4 reverse-phase high pressure liquid chromatography and by polyacrylamide gel electrophoresis. Histone H4 bands were excised and digested in-gel with the endoprotease trypsin. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry was used to characterize the level of acetylation, and nanoelectrospray tandem mass spectrometric analysis of the acetylated peptides was used to determine the exact sites of acetylation. Although there are 15 acetylation sites possible, only four acetylated peptide sequences were actually observed. The tetra-acetylated form is modified at lysines 5, 8, 12, and 16, the tri-acetylated form is modified at lysines 8, 12, and 16, and the di-acetylated form is modified at lysines 12 and 16. The only significant amount of the mono-acetylated form was found at position 16. These results are consistent with the hypothesis of a “zip” model whereby acetylation of histone H4 proceeds in the direction of from Lys-16 to Lys-5, and deacetylation proceeds in the reverse direction. Histone acetylation and deacetylation are coordinated processes leading to a non-random distribution of isoforms. Our results also revealed that lysine 20 is di-methylated in all modified isoforms, as well as the non-acetylated isoform of H4.

The basic structural unit of eukaryotic chromosomes is a DNA⅐protein complex called the nucleosome. The nucleosome consists of a DNA molecule associated with a histone octamer comprised of pairs of the core histones H2A, H2B, H3 and H4. The nucleosomes are joined by linker DNA and histone H1 to form chromatin. Each core histone has a globular region, the histone fold domain, which is involved in histone-histone interactions (1), and the wrapping of DNA around the nucleosome core (2). The N-terminal "tail regions" extend outside of the nucleosome particle where they can interact with DNA and with other regulatory proteins or transcription factors (2)(3)(4). These tails are essential but appear partially redundant. At least one of the two tails for both the H2A-H2B and H3-H4 pairs must be intact to maintain viability of yeast cells (5). The tails account for 28% of the core histone sequences and are extremely basic because of a high proportion of lysine and arginine. In the 2.8-Å nucleosome crystal structure, the electron densities for these tails are largely not observed (6). Presumably their binding sites lie outside of the core particle. Several studies indicate that these tails do not contribute significantly to the primary wrapping of DNA in the nucleosome. Trypsinized nucleosome core particles are just as stable as intact particles with respect to perturbations in temperature, high salt concentrations, and accessibility to DNase I (7,8). Neutron scattering studies have shown that hyperacetylation of histones has little or no effect on core particle structure in solution (9). However, the histone tail sequences are highly conserved, and the reversible acetylation of the ⑀-amino groups of specific lysine residues has been implicated in key regulatory events (3, 10 -15).
Histone acetylation is a very specific phenomenon with various isoforms playing distinct roles (13). Acetylation is a dynamic phenomenon with the steady state mediated by the opposing activities of histone acetyltransferases (HATs) 1 and deacetylases. These activities involve large regulatory complexes that are capable of responding to specific DNA sequences and can contain transcription factors, regulatory ligands, and signal transduction and cell cycle proteins.
Acetylation plays a role in nucleosome assembly. Newly synthesized histone H4 is di-acetylated in the cytoplasm at Lys-5 and Lys-12 by B-type histone acetyltransferase (HAT B) (20). CAF-1 (chromatin assembly factor 1) is a complex of proteins that deposits Lys-5, Lys-12 di-acetylated H4 into chromatin. It has also been found that human CAF-1 readily assembles newly synthesized H3 and H4 onto replicating DNA in vitro. Presumably the specific acetylation patterns of nascent histones are required for their assembly in vitro (21). The acetylation pattern found in mature nucleosomes differs from that of newly incorporated histones.
Increased acetylation is generally correlated with transcriptionally active or poised genes. Histone acetylation within nuclei is effected by A-type histone acetyltransferases (HAT A) that are likely to participate in gene activation. An in-gel HAT assay was used to purify and clone the cDNA for the major macronuclear HAT A, p55, from Tetrahymena (22). The gene encoding this enzyme is homologous to the yeast transcriptional co-activator gene Gcn5. The yeast protein Gcn5p, with ADA2 and ADA3, interacts with enhancer binding factors thereby establishing a direct mechanistic relationship between acetylation and gene activation (23). The yeast protein has also been shown to possess HAT activity. Gcn5p and p55 both contain a bromodomain (12), which has been shown to be important for the assembly and activity of multisubunit transcriptional activation complexes (12). The bromodomain may tether HAT A to specific chromosomal sites, linking histone acetylation and gene activation (12,22,24). The Gcn5p⅐ADA complex interacts functionally with the SWI⅐SNF complex, which is part of the RNA polymerase II holoenzyme (12,25). Further evidence for the connection between acetylation and transcription activation is the finding that H4 shows increased Lys-16 acetylation in the hyperactive X chromosome of male Drosophila (26). In the inactive X chromosome of female flies, H4 is hypoacetylated (27). Immunochemistry experiments demonstrate that H4 in transcriptionally inactive heterochromatin is hypoacetylated predominately in mammals. H4 histones from the heterochromatins of Drosophila (28) and yeast (29) are acetylated only at Lys-12, suggesting that this modification is important in gene silencing in these species.
Histone acetylation and deacetylation may play critical roles in cell cycle procession (3). The natural synchrony of Physarum polycephalum, which has no G 1 phase, enabled Matthews (30) to distinguish between acetylation specific for S phase and G 2 phase. H4 in growing cells was found to be acetylated at Lys-5, Lys-8, Lys-12, and Lys-16. In G 2 phase, however, the turnover was highest at Lys-8 whereas in S phase turnover was highest at Lys-5.
As the above examples illustrate much will be gained by having methodology that is easily able to identify the relative populations of acetylation isoforms as a function of cellular events and gene location. Allis and co-workers (31, 32) used [ 3 H]lysine pulse labeling and Edman microsequencing to determine the acetylation sites of newly synthesized H4 in many organisms including Tetrahymena, Drosophila, and humans. Other microsequencing methods have been used by Couppez et al. (33), Matthews (30), Thorne et al. (34), and Waterborg (35). Couppez et al. (33) observed that the principle tri-acetylated form of cuttlefish H4 is Lys-5/Lys-12/Lys-16 and that the principle di-acetylated and mono-acetylated forms are Lys-5/ Lys-12 and Lys-12 suggesting that Lys-5 is acetylated after The spectrum was calibrated by internal two-point calibration, using bradykinin (20 fmol; peak 1060.569) and trypsin autolysis product (peak 2163.057). Peaks were assigned to peptide sequences of histone H4 by matching the measured masses with the calculated masses (cf. Table I).
Lys-12 (33). In calf thymus H4 (36), Lys-16 is the only monoacetylated position, and all di-acetylated forms also involve Lys-16, implying that Lys-16 is acetylated first. Turner used specific antibodies for acetylated forms of H4 to conclude that Lys-5 and Lys-12 positions are under-used in mono-acetylated H4 from a number of mammalian cell lines and that Lys-8 and/or Lys-16 are the first position(s) to be acetylated (36). Treatment of actively growing cells with histone deacetylase inhibitors such as sodium butyrate shifts the equilibrium in the direction of acetylation of histones (37). Thorne et al. (34) have reported that for H4 from pig thymus and from sodium butyrate-treated HeLa cells Lys-16 is the predominant mono-acetylation site, but further addition of acetyl groups is less specific and progresses though sites Lys-12, Lys-8, and Lys-5 in an N-terminal direction. For example from butyratetreated HeLa cells, the Lys-8/Lys-12/Lys-16 isoform was found to be more than twice as abundant as the Lys-5/Lys-12/Lys-16 and Lys-5/Lys-8/Lys-16 isoforms.
In all the above cases the data were complicated by the kinetics of acetylation-deacetylation reactions and the procedures for histone extraction and separation, which often involved multiple steps of column chromatography. HPLC and/or electrophoresis followed by mass spectrometric characterization of the separated components shows great promise for revealing such patterns. Earlier, Edmonds et al. (38) reported the use of electrospray ionization mass spectrometry and tandem mass spectrometry to evaluate the sequence and modifications of histone H3. We report here the use of mass spectrometric methods to identify the acetylation sites of histone H4 to clarify the complicated and somewhat contradictory picture regarding the relative abundances of H4 acety-lation isoforms. Our mass spectrometric approach shows directly, effectively, and with high sensitivity that the acetylation of H4 in HeLa cell nuclei proceeds from lysine 16 to lysine 5. These results suggest the hypothesis of a "zip" model whereby acetylation of histone H4 proceeds in the direction from Lys-16 to Lys-5, and deacetylation proceeds in the reverse direction. Our results also revealed that lysine 20 is ⑀-amino-di-methylated in all the acetylated isoforms, as well as the non-acetylated isoform of H4.

MATERIALS AND METHODS
Cell Culture and Histone Separation-HeLa S3 (originally from ATCC) were cultured in RPMI 1640 supplemented with 10% calf serum (Invitrogen) as described (39) or purchased from National Cell Culture Center. Hyperacetylated histones were obtained by treating the cultures with 7 mM sodium butyrate (Mallinckrodt) for 24 h prior to harvest. Cells were washed in phosphate-buffered saline containing 10 mM sodium butyrate, and histones were extracted from isolated nuclei with 0.4 N H 2 SO 4 and separated using reverse-phase HPLC Vydac C 4 column (Vydac C 4 TP214; 10-m beads, 300-m pores) on a PerkinElmer Life Sciences Bio 410 system with a diode array detector. A flow rate of 10 ml/min with a gradient of water and CH 3 CN containing 0.1% trifluoroacetic acid was used. Fractions containing H4 (co-eluted with minor H2A) were confirmed by SDS-PAGE (see Fig. 1) and stained using Coomassie Blue (Novex).
In-gel and In-solution Enzymatic Digestion-Histone H4 bands were excised from the gel and digested with trypsin (40). For endoproteinase Asp-N digestion, 5 g of purified histones were dissolved in 20 l of 25 mM NH 4 HCO 3 and incubated at 37°C overnight with 300 ng of Asp-N. After adding 5 l of stop solution (50% acetonitrile, 45% water, 5% trifluoroacetic acid), digested peptides were concentrated using a Speedvac.
Mass Spectrometric Analysis of Peptides-Monoisotopic mass values of all peptides were measured by MALDI using a Voyager DE-STR biospectrometry workstation (PerSeptive Biosystems, Inc.) with de- DNIQGITKPAIRR layed extraction operated in the reflectron mode. A resolution of 5000 to 10,000 full-width at half-maximum was achieved within the mass range of 1000 to 3000. Between 100 and 256 accumulations were analyzed for each spectrum, and peptides were observed as (MϩH) ϩ in the positive mode. Post-source decay (PSD) spectra were recorded for selected ions. Peptides (1/10 of sample) were co-crystallized with equal volumes of a saturated ␣-cyano-4-hydroxycinnamic acid matrix (Hewlett Packard) in 50% acetonitrile, 0.1% trifluoroacetic acid. All MALDI spectra were calibrated internally using trypsin autolysis products and/or calibrated externally by an added peptide standard.

Nano-ESI Tandem Mass Spectrometry Analysis of Peptides-Tan-
dem mass spectra of peptides were obtained on a quadrupole orthogonal acceleration time-of-flight mass spectrometer with a quadrupole collision cell (QSTAR; Sciex, Toronto, Ontario, Canada) with an external nanoelectrospray ion source (Protana A/S, Odense, Denmark). Mass resolution was obtained routinely in the range of 7000 to 12000 (for both conventional mass spectrometric and MS/MS modes of operation), and a mass measurement accuracy of at least 0.02 Da with external calibration was achieved. Approximately 2 l of sample was loaded into a nanoelectrospray tip. Conventional mass spectra were first obtained to measure mass values (parent ions) of trypsindigested peptides and to assign their charge states from observation of stable isotope spacing. Then parent ions of interest were selected for sequence analysis by tandem mass spectrometry.

Identification of Acetylated Histone H4 by MALDI-TOF Mass
Spectrometry-Histone H4 samples were resolved by SDS-PAGE and visualized using Coomassie Blue stain. The single histone H4 band containing all acetylated isoforms was excised and digested with trypsin. The resulting tryptic peptides were analyzed by MALDI-TOF mass spectrometry to determine their individual mass values (Fig. 2). The specific peptide sequences of histone H4 were assigned to the mass values in     Table I, in which the acetylation sites contained in each specific peptide sequence are highlighted.
Whereas unacetylated lysine residues are susceptible to cleavage by trypsin, acetylated lysine residues are resistant. The calculated mass values for the peptide sequences corresponding to the various acetylated isoforms are listed in Table  II. By comparing the measured mass values of the peptide sequences that contain putative acetylation sites (see Fig. 2 and Table I) with the mass values of all possible trypsin-digest peaks of N-terminal portion of histone H4 (see Table II 16 ac R from the tri-acetylated isoform of histone H4 in which the unmodified lysine residue 5 was not cleaved during the trypsin digestion. Upon prolonged tryptic digestion of this component, the peak at 1396.80 eventually disappeared whereas the relative intensity of 1211.70 increased proportionately. The peak at mass value 543. 35 Da is assigned to peptide sequence 20 -23, K Me2 VLR, where lysine 20 is modified by two methyl groups. The sequence was established by MALDI-PSD (see Fig. 3). biguously to acetylated lysine (K ac ). This CID spectrum excludes the possible alternative to peptide fragment 79 -91, KTVTAMDVVYALK, which is isobaric with the tetra-acetylated peptide sequence 4 -17.

Acetylation Sites Established Unambiguously by Nano-ESI/ MS/MS-The
Using a similar nano-ESI tandem mass spectrometry strategy, the nature of precursor ion mass at 606.37 Da was established as peptide sequence 6 -17, 6 GGK ac GLGK ac G-GAK ac R, where lysine residues 8, 12, and 16 are all acetylated (Fig. 5). The other possible assignment of molecular mass 1211.71 Da to the uncleaved peptide sequence 92-102, RQGRTLYGFGG (mono-isotopic mass is 1211.62), is excluded by this CID spectrum. The lack of observation of this particular peptide could be because of digestion at Arg-95 to produce two smaller peptides 92-95 and 96 -102.
The precursor mass at 464.36 Da can be assigned either as peptide sequence 4 -12, 4 GK ac GGK ac GLGK (calculated mono-isotopic mass 927.52), where lysine 5 and 8 are acetylated, or peptide sequence 9 -17, 9 GLGK ac GGAK ac R (calculated mono-isotopic mass 927.53), where lysine 12 and 16 are acetylated. The results from the nano-ESI tandem mass spectrometry experiment (shown in Fig. 6) revealed that this component matches unambiguously the peptide sequence 9 -17 rather than that of the sequence 4 -12. Although the mass of 927.54 Da indicates that lysine 12 and 16 are acetylated, the origin of this peptide could not be confirmed. It could arise from either the di-acetylated isoform of histone H4 (lysine 12 and 16 are acetylated) or the tri-acetylated isoform of histone H4 in which lysine 5, 12, and 16 are acetylated, and lysine 8 is not acetylated and cleaved by trypsin (see Table II). However, the peptide sequence containing acetylated Lys-5 with a mono-isotopic mass of 488.3 Da corresponding to peptide sequence 4 -8 ( 4 GK ac GGK) was not observed (Fig. 2) so that the tri-acetylated isoform of histone H4 at sites Lys-5, Lys-12, and Lys-16 is less likely to be the correct assignment.
Histone H4 Is Di-methylated at Lysine Residue 20 in All Isoforms-To confirm that lysine 20 is modified by two methyl groups in all forms of histone H4, histone H4 was digested with Asp-N, which liberated peptide 1-23. As shown in Fig. 7, masses of 2430.4, 2472.4, 2514.5, 2557.2, and 2598.5 correspond respectively to un-, mono-, di-, tri-, and tetra-acetylated peptide isoforms (42 Da is the incremental mass of an acetyl group). Based on the fact that the measured mass values are 28 mass units higher than mass values of the unmethylated sequence, it is reasonable to conclude that lysine 20 must be modified by two methyl groups. Moreover, di-methylation of lysine residue 20 occurs in all H4 isoforms with different degrees of acetylation. That lysine residues 5, 8, 12, and 16 are acetylation sites has been proved by trypsin digestion and identified/confirmed by MALDI-TOF and nano-ESI tandem mass spectrometry and are therefore excluded as possible sites for methylation. This leaves only lysine residue 20 as the potential methylation site. The di-methylation at lysine residue 20 was confirmed by the identification of the peptide sequence 20 -23, K Me2 VLR, sequenced by MALDI-PSD (see Fig. 3). DISCUSSION We have shown that tandem mass spectrometry can be used to identify the specific sites of modification of individual histone acetylation isoforms from a mixture and that the method is sufficiently sensitive to gain sequence information from samples extracted from electrophoresis gels. This is a remarkable improvement in sensitivity and ease of assignment over earlier microsequencing methods. It also overcomes the problems associated with N-terminal acetylation of H4 Ser-1, which blocks Edman degradation.
These studies have established that there is a clear sequential pattern to the multiple acetylation of H4. For butyratetreated HeLa cells the only detectable acetylated forms are as follows: (i) the tetra-acetylated form is acetylated at lysine 5, 8, 12, and 16; (ii) the tri-acetylated form is acetylated at lysine 6, 12, and 16; (iii) the di-acetylated form is acetylated at lysine 12 and 16; and (iv) the only significant amount of mono-acetylated form is modified only at position 16. This provides ex-  (34), although these results are for total histone extracted from HeLa cells, treated with a deacetylase inhibitor, and thus describe the average acetylation pattern of histones. The levels and positions of acetylation of histones at individual genes may differ considerably and are governed by both HATs and by histone deacetylase. The actions of these two enzymes must be coordinated at most genes if a non-random distribution of isoforms for the entire cell is observed. The HPLC/gel electrophoresis/mass spectrometry approach may be applied to a wider range of cells and to specific chromosomes or chromatin domains and to cell cycle events to clarify such processes. As a working hypothesis we suggest a zip model (see Fig. 8) where histone acetylation starts at lysine 16 and continues until the four sites are acetylated whereas histone deacetylation must proceed in the reverse direction. Our mass spectrometric results (Fig. 7) revealed that lysine 20, in the N terminus of histone H4, is di-methylated and that methylation exists in all isoforms of histone H4 regardless of the degree of acetylation. It would be interesting to see whether this dimethylation site plays any role in acetylation. Does it serves as part of the initial HAT recognition and binding site?
Is there something special about the pattern of acetylation (16 3 12 3 8 3 5)? The acetylation of histones results in the loss of the positive charges on lysine residues located in the core histone N termini, which would weaken the association between histone N-terminal domains and DNA. Thermal denaturation studies revealed that acetylation markedly reduces the binding constant of the H4 N terminus to DNA by six orders of magnitude in dilute buffer containing 5 mM Tris-HCl (41). Protection experiments with a model system consisting of phage DNA as a substrate, and H4 with different degrees of acetylation as the protective agent demonstrated that non-acetylated H4 and mono-acetylated H4 cause similar protection from StuI, whereas di-, tri-, and tetra-acetylation of histone H4 result in cleavage of the DNA (42). Thus, nonacetylated H4 and mono-acetylated H4 have a comparable degree of protection of the DNA whereas di-acetylated, triacetylated, and tetra-acetylated H4 show significantly reduced protection against cleavage. Physical studies are required to determine whether the site of acetylation is as important as the number of positive charges neutralized. It is easy to speculate that the site of histone acetylation plays an important role in the binding of histones with transcription factors, because such specificity in protein-protein interaction is not uncommon. The non-acetylated histones are likely to adopt specific secondary structures when bound to DNA facilitating higher order chromatin condensation but precluding binding transcription factors or other specific chromatinassociated proteins whereas acetylation has been shown to favor transcription factor binding (43)(44)(45)(46).