Biophysics Laboratories, St. Michael's Building, Portsmouth Polytechnic, England. ' Unite 409 .... ing 2 g/I sodium bicarbonate supplemented with 5% new born calf serum just ...... approach with histones from human cells in culture, Turner et al.
Eur. J. Biochem. 193,701 -713 (1990) $>
FEBS 1990
Patterns of histone acetylation Alan W. THORNE ’, Daniel KMICIEK’, Keith MITCHELSON Pierre SAUTIERE’ and Colyn CRANE-ROBINSON’ ’ Biophysics Laboratories, St. Michael’s Building, Portsmouth Polytechnic, England ’ Unite 409, Centre National de la Recherche Scientifique, Institute de Recherche sur le Cancer, Lille, France I,
(Received May 15, 1990)
-
EJB 90 0553
The N-terminal domains of all four core histones are subject to reversible acetylation at certain lysine residues. This modification has been functionally linked to transcription, histone deposition at replication and to histone removal during spermatogenesis. To increase understanding of the significance of this modification we have studied the specificity of site utilisation in the monoacetyl, diacetyl and triacetyl forms of histones H3, H4 and H2B (histone H2A has only a single modification site), using pig thymus and HeLa cells as the source of histones. The HeLa histones were extracted from cells grown both with and without butyrate treatment. It is found that for histone H3 there is a fairly strict order of site occupancy: Lysl4, followed by Lys23, followed by Lysl8 in both pig and HeLa histones. Since the order and specificity is the same when butyrate is added to the HeLa cell cultures, we conclude that addition of the fatty acid does not scramble the specificity of site utilisation, but merely allows more of the natural forms of modified histone to accumulate. For histone H4, the monoacetyl form is exclusively modified at Lysl6, but further addition of acetyl groups is less specific and progresses through sites 12, 8 and 5 in an N-terminal direction. Similar results were obtained for H4 from both pig thymus and butyratetreated HeLa cells. Histone H2B could be studied in detail only from butyrate-treated HeLa cells and a much lower level of site specificity was found: sites 12 and 15 were preferred to the more N- and C-terminal sites at Lys5 and Lys20. The data reinforces the view that lysine acetylation in core histones is a very specific phenomenon that plays several functionally distinct roles. The high degree of site specificity makes it unlikely that the structural effects of acetylation are mediated merely by a generalised reduction of charge in the histone N-terminal domains. Core histone acetylation has been linked to several forms of DNA processing. A large body of evidence correlates the appearance of the modification with transcriptional activation in a wide range of organisms and cell types (for reviews see [l - 31). The link between core histone acetylation and transcriptionally active chromatin has now been demonstrated directly by fractionating chromatin with an antibody that recognises acetylated histones, then probing the DNA of this chromatin with sequences from an active gene [4]. An important question that follows is whether all the core histones are acetylated in transcriptionally active chromatin and to what levels. The antibody-fractionated nucleosomes studied by Hebbes et al. [4], clearly contained extensively, though not fully, acetylated histones H4 and H3, and there was some indication of the modification in histones H2A and H2B. Since this work was carried out on nucleosomes from the mature erythrocytes of chicken blood, it seems unlikely that the antibody-fractionated material would include nucleosomes acetylated for reasons associated with replication. The pattern of histone acetylation should thus be characteristic of transcriptionally active chromatin. Transcriptionally active Correspondence to C. Crane-Robinson, Biophysics Laboratories, St. Michael’s Building, White Swan Road, Portsmouth, Hants. PO1
2DT, England Abbreviutions. Pth, phenylthiohydantoin; HTF, HpaII tiny fragments; H3(1- 50), peptide containing the 50 N-terminal amino acids of histone H3; V8 protease, Staphylococcal serine protease; Aco, Acl, etc. nonacetylated, monoacetylated, etc. Enzymes. Staphylococcal serine protease (EC 3.4.21.19); Arg-C protease (EC 3.4.21.40);Trypsin (EC 3.4.1.24).
nucleosomes from rat liver, fractionated on mercuric acetatecolumns by virtue of exposed sulphydryl groups, likewise show extensive modification of histones H3, H4 and H2B [5]to levels similar to that of the antibody-fractionated chromatin. Indeed, the acetylation of the arginine-rich histones H3 and H4 has previously been studied the most extensively since they typically carry higher levels of acetyl groups [6, 71. The importance of the acetylation of histone H4 has recently been emphasized by genetic experiments in yeast to replace the modified lysines with other types of residue that cannot be acetylated [8]. The mutant yeasts were typically sterile and grew slowly. In experiments involving fractionation of chromatin enriched in transcribed genes, it is not clear if any particular acetylation state is specifically involved. Published acetic acid/ urea protein gels that resolve nonacetylated (Ac,), monoacetylated (Ac,), diacetylated (Ac,), triacetylated (Ac3) and tetraacetylated (Ac4) histones, show a general upward displacement of the mean level of acetylation of all the core histones. In the case of histone H4 from chicken embryo erythrocytes, the antibody-fractionated nucleosomes contained about two sites (out of four) occupied by acetyl groups, compared to only one site in the total chromatin [4]. For chromatin from cells treated with butyrate, the mean level of H4 acetylation is also about two sites occupied/molecule and the chromatin rapidly released by DNaseI has approaching three sites out of four occupied on average [6, 71. Structural studies of nucleosomes prepared by reconstitution with acetylated histones derived from butyrate-treated HeLa cells have shown a partial unwinding of DNA from the core particle [9]. A clear structural change was thus demon-
702 strated with histone that whilst highly modified was not completely acetylated at all sites. Whether further structural changes are induced by fully acetylated core histones has yet to be established. Very recently it has been shown that the chromatin of the so-called HpaII tiny fragment (HTF) islands (that mark out the 5' end of a large proportion of genes in higher eukaryotes) is very highly acetylated, at least regarding histone H4 [lo]. Whilst this suggests that the most highly acet,ylated forms of H4 are specifically associated with a defined region of some transcriptionally active genes, it is still to be demonstrated whether a gene has to be actively transcribing to carry such high levels of acetylation, or merely the possession of an HTF island is sufficient to attract such acetylation. A correlation of high levels of H4 acetylation with transcriptional activity has been claimed for Physarum [I 11, though the correlation has been challenged [12]. If high, but not necessarily specific levels of acetylation are a feature of active chromatin, the question remains as to whether specific sites of acetylation are characteristic of a transcriptional or other functionally active state. For example, in the Ac3 form of histone H4, are three specific sites fully utilized and the fourth not utilized at all? And does co-existing H4(Ac,) use the same sites as H4(Ac3), i.e. is the order of site occupation defined, or is it random? This question has previously been approached using the protozoan tetrahymena, with the advantage of comparison between the transcriptionally inactive micronucleus and the transcriptionally active macronucleus. Partial specificity of acetylation site utilisation was demonstrated and a distinction was drawn between transcriptionally related sites and sites specific for histone deposition following replication [13]. The experimental approach used with tetrahymena involved prior labelling with a pulse of [3H]acetate and detecting acetyllysine in Edman degradations as a peak of radioactivity. In the present work, we have used Edman degradation of unlabelled protein, thereby avoiding any uncertainties arising from non-random take up of radiolabel at the various sites. This approach allows the direct measurement of the ratio of acetyllysine to lysine at each site, and so removes the necessity to allow for the yield at each sequencing step when comparing the degree of acetylation at different sites. Our source of acetylated histones has been pig thymus and HeLa cells in culture. When the latter are harvested after 24 h in butyratecontaining medium, a considerable increase in the amount of the most highly modified histones is observed. This not only allows one to obtain sufficient amounts of the more highly modified histones for sequencing but also to study whether the presence of butyrate (known to inhibit deacetylase activity [6, 14- 161) results in a different and perhaps scrambled pattern of site occupancy. This is important since chromatin from cells treated with butyrate is sometimes used as a model for the effects of acetylation on chromatin structure. MATERIALS AND METHODS
HeLa S3 cell ciilture and harvesting HeLa S3 cells were grown in suspension culture using Joklik modified minimal essential medium (GIBCO) containing 2 g/I sodium bicarbonate supplemented with 5% new born calf serum just prior to use. Cell densities of 5- 10 x lo5 cells/ ml were maintained by daily dilution with fresh medium and serum. Cultures were treated with 7 mM sodium butyrate for 24 h after dilution of the cells to 5 x lo5 cells/ml. Cells were
harvested by centrifugation of the cultures at 2200 g for 15 min at 4°C. Supernatants were removed by aspiration and cell pellets resuspended in 135 mM NaCI, 2.5 mM KCl, 8 mM Na2HP04 and 1.5 mM KH2P04,pH 7.2, containing 25 mM sodium butyrate (butyrate buffer) using a serological pipette. Cell pellets were washed a further twice in butyrate buffer, centrifuging at 540 g for 5 min at 4°C. Cell pellets were frozen and stored in liquid nitrogen until required for use. Nuclear isolation and histone extraction
Nuclei were isolated from HeLa cells and from pig thymus essentially as described by Panyim et al. I171 modified by the inclusion of 50 mM sodium butyrate in all buffers. Histones were extracted from whole nuclei or chromatin using 0.4 M HCI at 4 'C for 4 h. DNA and acid-insoluble proteins were removed by centrifugation at 12000 g for 10 min at 4°C and the histones were precipitated from the supernatants using 10 vol. acetone at - 20°C for 16 h. After washing three times with dry acetone the histones were dried under vacuum and stored at - 20°C. Histone,fractionation Histones were separated into their principal subfractions by gel-filtration chromatography after the method of Van der Westhuzen et al. [18]. 0.8-g batches of total histone extracts were dissolved in 10 ml 8 M urea, 10% mercaptoethanol, 2 mM Tris at NN 50 mg/ml prior to loading. Histones were eluted from a 40 mm x 1400 mm Bio-Rad P60 column (100200 mesh), equilibrated with 20mM HCI, 50 mM NaCl and 0.02% sodium azide, at 30 ml/h. The separated fractions containing histones H1, H2A and H4 were dialysed against 0.5% formic acid (four changes) for 48 h prior to lyophilisation. Fractions containing histones H2A and H3 which coelute from the P60, were similarly recovered and then separated by passage through a 35 mm x 1400 mm Bio-Rad P10 column (200-400 mesh) equilibrated with 10 mM HC1 and 0.02% sodium azide. Histones were dissolved as for the P60 column and eluted in the above buffer at 30 ml/h. Ion-exchange,fractionationof the acetylated forms
The acetylated forms of the intact histones H4, H2B and H3 and a 50-amino-acid N-terminal peptide of histone H3 [H3(1- 50)] were separated using either a low-pressure Whatman CM52 or a CM-3SW HPLC ion-exchange column, both equilibrated with the same buffer: 6 M urea, 50 mM sodium acetate, pH 4.75,2 mM Tris and 0.1 mM EDTA. Proteins or peptides were eluted from both types of column using shallow linear salt gradients in the same buffer. Column dimensions and gradient conditions are indicated in the figure legends. Absorbance was monitored continuously at 230 nm. Fractions containing the acetylated forms were dialysed exhaustively against 0.5% formic acid or desalted by means of a Sephadex G-25 column and finally lyophilised. Residual peptide contaminants present in the various acetylated forms of H3(1- 50) from both pig thymus and HeLa were removed by reversed-phase HPLC on either a C 3 or CIS column using the elution conditions described for the separation of the H4(1- 52)/H4(1 - 53) peptides. Final assessment of purity of all fractions was made by acetic acid/ 6.25 M urea PAGE, as described in [17], and amino acid analysis.
703 Staphylococcal protease digestion
Total histone H3 and the various separated acetylated H4 forms were prepared at 2 mg/ml in 0.1 M ammonium acetate, pH 4.0. Staphylococcal serine (V8) protease (Pierce) was added to give an enzyme/substrate ratio of 1: 25 (by mass) and digestion was allowed to proceed for 48 h at 37°C. Termination was achieved by acidification and lyophilization of the products. Purification ofpeptides H4(1-52)/H4(1-53)
The various V8-protease-digested acetylated H4 forms obtained from butyrate-treated HeLa cells were separately fractionated by HPLC using a Beckman Ultrasphere C3 column (4.6 mm x 75 mm). Samples were dissolved in 8 M urea and 1% formic acid, at 20 mg/ml, and after 5 min of isocratic elution with 0.05% trifluoroacetic acid in water, were separated with a 40-min (0 - 50%) gradient of water/ acetonitrile containing 0.05% trifluoroacetic acid at 1.5 ml/ min. The H4(1- 52)/H4(1- 53) peptides, the major peak in the profile, was identified by acetic acid/urea PAGE analysis, and eluted after about 22 min from the C3 column under these conditions. Peptides resulting from digestion of acetylated pig thymus H4 forms with V8 protease were similarly fractionated employing a Waters CIS y-Bondapak column (3.9 mm x 300 mm) eluting over the same period with a gradient of water/acetonitrile, containing 0.05% trifluoroacetic acid, from 0-70% at a flow rate of 2.0 ml/min. The H4(1-52)/ H4(1- 53) peptides co-eluted after about 21 min. Retention times of these peptides were slightly increased with increasing acetyl content when using both types of column. Citraconylation and trypsin digestion - 53)
of peptides H4 ( 1 -52) /H4( I
Free lysine residues of the various acetylated forms of H4(1- 52)/H4(1- 53) were blocked with citraconic anhydride to allow trypsin cleavage only at arginine residues. Peptides were prepared at 2- 3 mg/ml in 0.5 mlO.1 M NaHC03 on ice and citraconylated by the addition of aliquots of citraconic anhydride until a 100-fold molar excess over the available lysine concentration was reached. The pH was maintained at about 8.3 during the reaction by titration with 1.0 M NaOH. Excess reagent was removed by passage through a Sephadex (3-25 desalting column (600 mm x 6 mm) equilibrated with 20 mM Tris/HCl, pH 8.5. Fractions containing the citraconylated peptides were pooled and digested by the addition of tosylphenylalaninechloromethane-treated trypsin (Sigma, E/S ; 1 :50, by mass), allowing digestion to proceed for 1 h at 37 ’C. The reaction was terminated by the addition of diisopropylfluorophosphate to 0.025%, immediately followed by acidification to pH 2.0 with formic acid to deblock the lysine groups. Samples were then dried by rotary evaporation. Purification of Peptide H4(4 - 17)
Isolation of the peptide H4(4 - 17) from the trypsin digest of each of the separate acetylated forms of both pig thymus and HeLa H4(1- 52)/H4(1- 53) peptides was achieved on a Waters CI8 p-Bondapak column (3.9 mm x 300 mm). Samples were dissolved in 100 pl 5% formic acid and centrifuged at 12000 g for 10 min prior to injection on to the HPLC column, and peptides were eluted for 5 min with 0.05% trifluoroacetic
acid in water then with a 40-min linear gradient of 0 - 50% acetonitrile containing 0.05% trifluoroacetic acid at 2.0 ml/ min. Peaks were identified by amino acid analysis. Purification of peptide H3 ( I
-
50)
Peptide H3(1- 50) of pig thymus and HeLa cells was isolated by digestion of total H3 with V8 protease, followed by gel-filtration chromatography on a Sephadex G-50 superfine column (16 mm x 1 m) equilibrated with 0.5% formic acid (pH 2.0), eluting at 36 ml/h. Absorbance was monitored at 226 nm. Amino acid analysis
Identification and quantification of the peaks in the various HPLC profiles was performed by amino acid analysis. Histones and peptides were hydrolysed in 6 M HCl in vacuo at 11O’C for 24 h. One drop of 1 % phenol was added to avoid excessive degradation of tyrosine. Typically, 5 - 10 nmol peptide was used for analysis on a Beckman 119 CL analyser or 0.5-1 nmol for analysis using a Beckman System 6300 analy ser. Sequence analysis
Acetylated histone H2B samples (90 - 360 nmol) were sequenced on a Beckman 890C liquid-phase sequencer in the presence of polybrene with 0.33 M quadrol as coupling buffer. The phenylthiohydantoin (Pth) derivatives of amino acids were identified and quantified as described by Hermann et al. [19]. Peptides (z1 nmol) of H3 and H4 were sequenced on an Applied Biosystems 470A gas-phase sequencer using the 03RPTH program. The phenylthiohydantoin derivatives were quantified by an Applied Biosystems 120A on-line analyser. Materials
Quadrol was purchased from Beckman. Phenylisothiocyanate, heptafluorobutyric acid and methanol were from Merck. Benzene, heptane and ethyl acetate were purchased from Peypin (France). Dithiothreitol was from Sigma and npropanol and trifluoroacetic acid from Pierce. Acetonitrile for reversed-phase HPLC of peptides was obtained from Rathburn or from Carlo-Erba. All solvents and reagents for gasphase sequencing were from Applied Biosystems. All other reagents were of the highest purity available. RESULTS Separation of histones by total acetyl content Histone H 4 . Fig. 1A and B show the CM52 ion-exchange profiles obtained from the fractionation of intact H4 from pig thymus and HeLa cells treated with butyrate. The accompanying acetic acid/urea PAGE analysis of the peaks in the profile demonstrates the resolution of the Aco, Ac,, Acz, Ac, and Ac4 forms of the H4 histone. The use of a shallow gradient and a long (z70 cm) column was found to be essential to avoid overlap of the acetylated species. The gel of Fig. 1 A also shows that pig thymus H4(Ac4) was contaminated with low-molecular-mass material. Although H4(Ac4) was not used for sequence analysis, these contaminants were readily removed by subsequent gel filtration on Sephadex G-75.
704 M
M
A4
A3
A 2
A1
A0
M
1.2 -
C H1
I
I
I
I
-4
600
400 500 V o l u m e lmll
301
- 3 H4 .2
‘bl
M
A3
A2
A1
A0
M
M , NaCl
A’$
,’
HeLa B u t y r a t e H 4
,,-’
0.20 H1
200
400
300 Volume
lmll -2
-1 H4 -0
M
:;’A 0.8 -
c
A 0 A1
A2
M
M lNeCl Pig T h y m u s H3
0.6 -
-
-H1 -2
51 H 3
0.2 -
