A Method for Genetically Installing Site-Specific Acetylation in Recombinant Histones Defines the Effects of H3 K56 Acetylation

1The posttranslational acetylation of chromatin on the ɛ-amine of lysine residues in histone proteins defines the epigenetic status of human embryonic stem cells and is a crucial regulator of DNA replication, chromosome condensation, transcription, and DNA repair in model organisms. Acetylation may alter nucleosome or chromatin structure and function directly or may act to recruit other factors to the genome via interaction with bromodomain-containing proteins and other potential acetyl-lysine binding modules.

2It is emerging that modifications in the globular core of histones play crucial roles in regulating the structure and function of chromatin and controlling biological function. H3 K56 acetylation is a particularly important modification in the globular core of H3 that is conserved from yeast to humans. Numerous reports have demonstrated its role in DNA repair and replication, regulation of transcription, and chromatin assembly. K56 is located on the first α helix of H3 and makes a water-mediated contact close to the DNA at the entry-exit point on the nucleosome. This has led to the proposal that K56 acetylation modulates the binding of the DNA in the nucleosome. Though H3 K56 acetylation is clearly important in defining epigenetic status and regulating transcription, replication, and repair, it has not been possible to experimentally and quantitatively test the mechanistic proposals for how K56 acetylation might affect these complex cellular phenomena because it has not been possible to synthesize homogeneously acetylated H3 K56.

3Current methods to introduce site-specific acetylation into recombinant histones include enzymatic posttranslational modification and native chemical ligation. Core histones consist of a structural domain and a long N-terminal tail, and chemical ligation provides an approach suitable for introducing modifications into histone tails. For example, the role of H4 K16 acetylation in antagonizing chromatin compaction was demonstrated with histones modified at the N-terminal tail produced by native chemical ligation. This method remains challenging, as it requires the synthesis of large quantities of peptide thioester, has not been demonstrated for acetylation in the structured core of histones, and yields small quantities of acetylated protein. Enzymatic posttranslational acetylation of histones has also proved challenging. The acetyl-transferase enzymes are large complexes that are difficult to purify. Moreover, enzymatic acetylation does not yield homogeneous samples, as acetylation at the desired site is not quantitative and rarely exceeds 30%, and in vitro acetylation at other sites leads to heterogeneous samples. We recently reported proof-of-principle experiments in which we demonstrated the production of a homogeneously acetylated protein using an aminoacyl-tRNA synthetase and tRNACUA pair that we created by directed evolution. This pair directs the incorporation of acetyl-lysine in response to an amber codon in the gene of interest encoded on a plasmid in E. coli. While this system is, in principle, applicable to the production of homogeneously and site-specifically acetylated histones, its utility has previously only been demonstrated in a single case with MnSOD, a small nonhistone enzyme.

4We report an improved method for genetically encoding site-specific and homogeneous acetylation of histones in E. coli. This method has permitted the production of histones modified in the core of the protein, enabling us to produce recombinant histone H3 that is acetylated on K56 in milligram quantities. We have assembled histone octamers, nucleosomes, and chromatin arrays that bear this modification, and we have investigated the effect of H3 K56 acetylation on nucleosome and chromatin structure and function. We have examined the effect of H3 K56 acetylation on nucleosome stability, transient unwrapping of DNA from single nucleosomes, chromatin compaction, and nucleosome remodeling by SWI/SNF and RSC to test the mechanistic hypotheses on the role of this modification that have been proposed on the basis of cellular experiments.

5We recently described an acetyl-lysyl-tRNA synthetase (AckRS)/tRNACUA pair that is derived from the M. barkeri (Mb) pyrrolysyl-tRNA synthetase/tRNACUA pair. The AckRS/tRNACUA pair directs the incorporation of acetyl-lysine in response to the amber codon with high translational efficiency and fidelity to produce homogeneously acetylated protein. To improve the efficiency of this system further, we made a library that randomizes residues L266, A267, L270, Y271, and L274 in acetyl-lysyl-tRNA synthetase (Figure 1), and we selected for improved efficiency of acetyl-lysine incorporation as previously described. These selections yielded an improved synthetase, which contains a single L266M mutation with respect to AcKRS-1. We named this synthetase AcKRS-3. Expression of myoglobin-his6 incorporating acetyl-lysine at position 4 from a myoglobin gene bearing an amber codon at position 4 demonstrates the improvement in protein expression using the AckRS-3/tRNACUA system.

6In order to synthesize acetylated histone H3, we further created a vector that contains the MbtRNACUA gene and an N-terminally hexahistidine-tagged histone H3 downstream of a T7 promoter (Figure S1 available online). We introduced an amber codon at position 56 and transformed this vector into BL21 E. coli bearing AcKRS-3. Cells were supplemented with 10 mM acetyl-lysine and 20 mM nicotinamide (to inhibit the E. coli deacetylase CobB) at OD 0.7, and protein expression was induced by addition of IPTG 30 min later. Like unmodified histone H3, the H3 K56Ac is overexpressed and found in inclusion bodies (Figure 2B). Histone H3 K56Ac was purified by denaturing Ni-NTA chromatography with a yield of 2 mg per liter of culture. Subsequent cleavage with TEV protease cleanly removed the N-terminal His6 tag. Electrospray ionization mass spectrometry (Figure 2C) demonstrates the homogeneous incorporation of a single acetyl-lysine residue, and MS/MS confirms that the amino acid is incorporated at the genetically encoded site. By simply moving the position of the amber codon in the H3 gene, we have made several other important acetylated variants of H3, including H3 K14Ac, K23Ac, and K27Ac. To further demonstrate the generality of the method, we have expressed and characterized other histones containing acetylated lysines at specific positions H2B K5, H2B K20, and H2A K9 (Figures 2B and Figure S2).

7We assembled H3 K56Ac into histone octamers with H2A, H2B, and H4 using standard methods of refolding with a comparable efficiency to unmodified H3, indicating that acetylation does not affect octamer formation in vitro (Figure 2D). We used these H3 K56Ac-containing histone octamers to assemble nucleosomes with DNA. The efficiency of octamers' formation with unacetylated H3 and H3 K56Ac was comparable (Figure 2E), again indicating that acetylation does not affect the efficiency of nucleosome formation.

8The structure of the nucleosome core particle shows that H3 K56 is in an α helix that binds the last 10 bp of DNA at the entry/exit site, and H3 K56 itself makes a water-mediated contact between H3 K56 and the phosphate backbone of the DNA. Several groups have proposed that K56 acetylation affects the stability of the nucleosome or DNA breathing on the nucleosome and have suggested that this provides a structural, mechanistic, and energetic basis for observed cellular phenomena. However, there are no experimental data on the effect of this modification on nucleosome stability or DNA breathing. To investigate the effect of H3 K56 on nucleosome stability, we first compared the equilibrium stability as a function of NaCl concentration for nucleosomes containing H3 K56Ac and unacetylated H3 by fluorescence resonance energy transfer (FRET) using previously established assays and fluorophore positions.

9We placed a Cy3 FRET donor on the 5′ end of a 147 bp DNA nucleosome-positioning sequence and quantitatively labeled a K119C mutant of H2A with a Cy5 maleimide to create a FRET acceptor (Figure S3). We assembled nucleosomes with the Cy3-labeled DNA and octamers that contained the Cy5-labeled H2A and either H3 K56Ac or unacetylated H3 (Figures 3A and 3B). In each nucleosome, there are two Cy5 fluorescently labeled H2A molecules; however, only one of these H2A molecules is close enough to the Cy3 on the DNA to contribute significantly to FRET. At high NaCl concentrations in which the nucleosome is dissociated, excitation of the Cy3 donor leads to strong emission centered on 565 nm, consistent with Cy3 fluorescence and negligible acceptor emission centered on 670 nm, as expected in the absence of FRET (Figure 3C). At low NaCl concentrations in which the nucleosome is intact, excitation of the Cy3 donor leads to a decreased donor emission at 565 nm and an increased Cy5 acceptor emission at 670 nm, consistent with a FRET signal. To assess the stability of H3 K56 acetylated nucleosomes, we followed the emission of donor and acceptor fluorophores as a function of [NaCl] for nucleosomes bearing H3 K56 acetylation and nucleosomes bearing unmodified H3 (Figure 3D). We found that acetylated and nonacetylated nucleosomes show comparable stability to NaCl through a range of concentrations that cover partial unwrapping of the DNA, dissociation of H2A/H2B dimers, and dissociation of H3/H4 dimers. These data indicate that acetylation of H3 K56 does not have a substantial effect on nucleosome stability, but the error in the assay does not allow us to distinguish small effects in partial unwrapping of the DNA that result from DNA breathing.

10To investigate the partial unwrapping of the DNA resulting from the spontaneous transient breathing of DNA from the histone core, we used a recently developed combination of single-pair FRET (spFRET), alternating excitation (ALEX) selection, and native PAGE. Nucleosomes reconstituted on a nucleosome-positioning element DNA containing a Cy3B-ATTO647N FRET pair were separated from free DNA using native PAGE. The nucleosome-containing band was excised from the gel and imaged in a confocal microscope using rapidly alternating green and red excitation. Resulting photon bursts were separated into a green and a red channel. The fluorescence of each nucleosome that diffuses through the focus was analyzed for both FRET efficiency and fluorescent label stoichiometry. Finally, a distribution of FRET efficiencies was generated from individual nucleosomes that have both the donor and acceptor label. Using these stringent selections, we identified complexes that were folded into nucleosomes and contained both fluorescent labels. By examining this population, we were able to accurately assess the transient DNA conformations of individual nucleosomes.

11To measure the effect of H3 K56 acetylation on DNA breathing, we used two label pairs. The first label pair was positioned 2 bp from the end of the DNA on the nucleosome and within 7 bp of the K56 location (end label, −2 position) and was located near the positions labeled in our bulk FRET experiments. The second label pair was positioned 27 bp from the end of the nucleosomal DNA and 20 bp internally to the K56-binding position (internal label, −27 position) (Figure 4A).

12For the unmodified nucleosomes, 11% of the DNA is unwrapped (FRET efficiency < 0.3) at the internal position, and 13% of the DNA is unwrapped at the end position (Figures 4B and 4C) in the first turn of DNA. For H3 K56 acetylated nucleosomes, the fraction of nucleosomes in which the DNA is unwrapped at the end position is doubled to 28%, whereas the fraction of nucleosomes in which the DNA is unwrapped at internal position only increases 3%, from 11% to 14%. These data clearly indicate that H3 K56 acetylation is sufficient to cause a local increase in spontaneous DNA breathing at the entry exit point of the DNA on the nucleosome. This effect may increase the accessibility of nucleosomal DNA to other factors, such as nucleosome-remodeling complexes. Assuming that unwrapping of the end positions is required for unwrapping the internal pair, only 2% (13%–11%) of unacetylated nucleosome cores are unwrapped only at the end label position in the first turn of the DNA. In contrast, acetylated nucleosome cores are 14% (28%–14%) unwrapped only at the end label position. Comparison of the fraction of nucleosomes unwrapped only at the end label position suggests that acetylation increases DNA breathing within the last turn of DNA on the nucleosome core 7-fold.

13Compaction is a prerequisite for heterochromatin formation. Mutation of H3 K56 to a noncharged residue causes defects in silencing at telomeres, in which K56 acetylation is normally less abundant. Moreover, failure to deacetylate K56 may lead to defective silencing at telomeres. These experiments suggest that H3 K56Ac may mediate, directly or indirectly, the compaction state of chromatin.

14To investigate the direct effect of H3 K56Ac on chromatin compaction, we reconstituted nucleosome arrays. To do this, we assembled histone octamers containing H3 K56Ac or unmodified H3 on 61 repeats of the 197 base pair 601 nucleosome-positioning DNA sequence with increasing amounts of H5 linker histone (Figure 5A). These chromatin arrays were then folded in 1 mM MgCl2 and 10 mM TEA (pH 7.4). Sedimentation velocity analysis (Figure 5B) of the H3 K56Ac arrays reveals a compaction profile that is essentially indistinguishable from that of the arrays bearing wild-type H3. Our results suggest that H3 K56Ac has little effect on the compaction of the chromatin fiber and support the view that its effect on compaction suggested from in vivo experiments is either mediated by additional factors or is context dependent.

15Chromatin immunoprecipitation (CHIP) experiments demonstrated a correlation between H3 K56Ac and SWI/SNF recruitment to activated promoters. Because SWI/SNF contains a bromodomain, we investigated the effect of K56 acetylation on the direct recruitment of SWI/SNF. We did not detect any difference in binding of K56 acetylated nucleosomes and nonacetylated nucleosomes to SWI/SNF (using electrophoretic mobility shift assays; data not shown), indicating that recruitment of SWI/SNF to K56-acetylated nucleosomes is either context dependent or is mediated by another factor. Similarly, we did not observe enhanced binding of the acetylated nucleosomes to RSC or Bdf1, which also contain bromodomains (data not shown). These experiments demonstrate that H3 K56Ac is not sufficient to directly recruit these remodelers.

16It was proposed previously that H3 K56Ac modulates chromatin remodeling by SWI/SNF by facilitating access to the DNA at the entry-exit gate. An H3 K56R mutant failed to recruit SWI/SNF as judged by ChIP analysis and failed to activate histone gene transcription. To investigate the effect of H3 K56Ac on remodeling by SWI/SNF and RSC, we compared the remodeling of nucleosomes containing H3 K56Ac and unmodified nucleosomes (Figures 6A and 6B). SWI/SNF repositioned H3 K56Ac-containing nucleosomes 1.4 fold ± 0.2 faster than unmodified control nucleosomes (Figure 6A), whereas RSC repositioned the acetylated nucleosomes 1.2 fold ± 0.1 faster than unmodified nucleosomes (Figure 6B). This observation is also consistent with the data obtained using H3 K56Q mutated nucleosomes, which were repositioned ∼1.3-fold faster than wild-type nucleosomes. Collectively, these observations show that histone H3 K56 acetylation has a small effect (∼20%) on nucleosome redistribution by RSC and SWI/SNF. However, given that the effects of H3 K56Ac on in vivo transcriptional activation are 2- to 3-fold, it is possible that the modest increase in acetylation-dependent nucleosome repositioning contributes to the observed activation.

17In addition to moving nucleosomes in cis along DNA, the Snf2 subfamily members have also been shown to displace H2A/H2B dimers from nucleosomes. We therefore performed ATP-dependent H2A/H2B dimer transfer assays (Figures 6C and 6D) using H3 K56Ac nucleosomes to investigate whether this modification was able to influence this specific type of remodeling behavior. In brief, wild-type or H3 K56Ac nucleosomes were assembled with Cy5-labeled H2A onto Cy3-labeled 54A18 DNA fragments. A chromatin acceptor of wild-type H3/H4 tetramer assembled onto 0W0 DNA fragments was added at ∼3-fold molar excess. Following remodeling, the samples were resolved on a native PAGE gel, and the Cy5 histone dimer fluorescence was monitored. ATP-dependent dimer transfer for both RSC and SWI/SNF was not greatly affected by the H3 K56Ac donor nucleosomes relative to wild-type nucleosome controls (Figures 6C and 6D). Quantification of the data revealed that both RSC and SWI/SNF were no more efficient at H2A/H2B dimer transfer from H3 K56Ac nucleosomes than from unmodified nucleosomes, indicating that H3 K56Ac has a minimal effect on dimer transfer.

18We have developed a method for the production of large quantities of histones bearing site-specifically defined and homogeneous acetylations. This method has allowed us to prepare histone H3 with 100% acetylation at K56 in the globular core of the histone. Using nucleosomes assembled with H3 K56Ac, we have measured the effect of H3 K56 acetylation on nucleosome stability and DNA breathing at the entry-exit points of the nucleosome. In summary, we find that, whereas H3 K56Ac does not affect the compaction of chromatin fibers, it affects the nucleosome core structure in a number of subtle ways: it increases DNA breathing at the entry-exit point of the nucleosome and has a small effect on remodeling of mononucleosomes. It is possible that these effects may be amplified within chromatin, leading to large-scale changes in accessibility and structure.

19Understanding the effects of lysine acetylation on transcription, DNA replication, or DNA repair requires biochemical analysis of the effect of this modification on chromatin structure. Peptide models have allowed the interactions of histone tail modifications to be investigated, but these experiments cannot address the direct or indirect effects of modifications on nucleosome structure, the higher-order structure of chromatin, or the effect of modifications on the interaction with other factors and remodelers within the context of intact chromatin. Moreover, peptide models cannot be used to probe the role of the emerging modifications in the globular core of histone proteins, the most prominent of which is, perhaps, H3 K56 acetylation. From the outset, H3 K56 acetylation was hypothesized to strongly destabilize this DNA-histone interaction at the entry-exit point. This was assumed to affect the structure of chromatin, especially during compaction, and has been implicated in silencing at telomeres. The acetylation might directly affect compaction or act to recruit factors that affect the remodeling and compaction of chromatin. Our data demonstrate that H3 K56Ac is not sufficient to cause the 2- to 3-fold changes in compaction observed for H4 K16. The effect of H3 K56 acetylation on silencing is, therefore, more subtle and must depend on the simultaneous presence of other modifications or on the modification-dependent recruitment or action of other factors.

20Single-molecule FRET experiments on acetylated nucleosomes, which are possible with the homogeneous material produced by our method, demonstrate that H3 K56 acetylation is sufficient to cause a 7-fold increase in DNA breathing on the nucleosome. These results are consistent with the observed increase in MNase sensitivity of yeast chromatin bearing an H3 K56Q mutation. Our data also support the proposal that deacetylation of H3 K56Ac by Sir2 acts to close the entry-exit gates of DNA around the histone octamer. Furthermore, the 7-fold increase in DNA breathing observed in our single-molecule experiments may well explain a number of phenotypes reported for mutations in H3 K56 or the enzymes involved in its modification. For example, the Grunstein lab reported (using ChIP assays) that wild-type yeast recruited the remodeler SWI/SNF 2- to 3-fold more efficiently to the HTA1 promoter than an H3 K56R mutant. Changes seen in gene expression levels caused by deletion of Spt10 (an enzyme required for the acetylation of H3 K56 on histone genes) are of the same magnitude and overlap with the changes seen for mutations in H3 K56. H3 K56Ac is also involved in the activation of the PHO5 promoter. The loss of nucleosomes from the PHO5 promoter is slowed ∼2 fold by a H3 K56R mutation. H3 K56Q mutants also affect telomeric chromatin structure, leading to a 4- to 6-fold increase in the mRNA levels of a telomere proximal gene. The magnitude of the physical changes resulting from H3 K56Ac are of the same order of magnitude as the increases in, for instance, transcription, suggesting that the biological processes are directly regulated by the enhanced nucleosome plasticity.

21The method that we present for genetically encoding acetyllysine will facilitate investigation into the roles of histone acetylation within the same nucleosome, as well as the roles of multiple acetylations on a single histone. We have recently extended our method for genetically encoding methylation. Our method of acetylation is compatible with native chemical ligation, as well as methods for installing methylation analogs, and it will therefore be possible to introduce acetylation in combination with other histone modifications, such as ubiquitylation and methylation on individual histones and nucleosomes. Combining the full arsenal of methods, synthetic and genetic, for installing histone posttranslational modifications in chromatin will be increasingly important for defining the combinatorial role of modifications and providing a biochemical understanding of biological phenomena.

22Information on library design and selection, mass spectrometric analysis of recombinant histones, and the purification of remodeling complexes can be found in the Supplemental Data. Standard methods and assays, such as the reconstitution of histone octamers and nucleosome arrays and their analysis by sedimentation velocity, bulk FRET, and remodeling assays, can also be found there.

23BL21 DE3 (for H3) or Rosetta DE3 (for H2A and H2B) cells were transformed with plasmid pAcKRS-3 and pCDF PylT-1 carrying the ORF for the histone with amber codons at the desired sites. The cells were grown overnight in LB supplemented with 50 μg/ml kanamycin and 50 μg/ml spectinomycin (LB-KS). One liter of prewarmed LB-KS was inoculated with 50 ml overnight culture and was incubated at 37°C. At OD600 of 0.7–0.8, the culture was supplemented with 20 mM nicotinamide (NAM) and 10 mM acetyl-lysine (AcK). Protein expression was induced 30 min later by addition of 0.5 mM IPTG. Incubation was continued at 37°C, and cells were harvested 3–3.5 hr after induction, washed with PBS 20 mM NAM, and stored at −20°C.

24The pellet was resuspended in 30 ml PBS supplemented with 20 mM NAM, 1 mM PMSF, 1 × PIC (Roche), 1 mM DTT, 0.2 mg/ml lysozyme, and 0.05 mg/ml DNase I and was incubated for 20 min with shaking at 37°C. Cells were lysed by sonication (output level 4 for 2 min on ice). Extracts were clarified by centrifugation (15 min, 18,000 rpm, SS34), and the pellet was resuspended in PBS supplemented with 1% Triton X-100, 20 mM NAM, and 1 mM DTT. The samples were centrifuged as above and washed again, once with the same buffer and then twice with the same buffer without Triton X-100. The pellet was macerated in 1 ml DMSO and incubated for 30 min at room temperature. Twenty-five milliliters of 6 M guanidinium chloride, 20 mM Tris, and 2 mM DTT (pH 8.0) were used to extract the histone proteins from the pellet. The samples were incubated for 1 hr at 37°C with shaking, centrifuged as above, and loaded onto a pre-equilibrated Ni2+-NTA column (QIAGEN). The column was washed with 100 ml 8 M urea, 100 mM NaH2PO4, and 1 mM DTT (pH 6.2). Bound proteins were eluted with 7 M urea, 20 mM sodium acetate, 200 mM NaCl, and 1 mM DTT (pH 4.5).

25The eluates containing the protein were combined and dialysed at 4°C against 5 mM β-mercaptoethanol (two times against the 100-fold volume). The solution was made up of 50 mM Tris/HCl (pH 7.4) and supplemented 1:50 with 4 mg/ml TEV. The reaction was incubated for 5 hr at 30°C. Afterward, salts were removed by dialysis as above, and the protein was lyophilized.

26Mononucleosomes were reconstituted on a fluorescently labeled 155 bp DNA template containing a 601 nucleosome-positioning sequence as described. In brief, the template DNA was prepared by PCR and was labeled with Cy3B (donor) and ATTO647N (acceptor) by incorporation of fluorescently labeled, HPLC-purified primers (IBA GmbH, Göttingen, Germany). Nucleosome reconstitutions were analyzed with 5% native poly-acrylamide gel electrophoresis (PAGE). A sample of 0.1–1 pmol was loaded on the gel (29:1 bis:acrylamide, 0.2× TB). The gel was run at 19 V/cm at 4°C for 80 min and visualized with a gel imager (Typhoon 9400, GE, Waukesha, WI, USA). The band corresponding to reconstituted nucleosomes was excised and put on a home-built confocal microscope equipped with a 60× water immersion microscope objective (NA = 1.2, Olympus, Zoeterwoude, The Netherlands) and two single-photon avalanche photodiodes (SPCM AQR-14, Perkin-Elmer [EG&G], Waltham, MA, USA). The photodiodes were read out with a TimeHarp 200 photon counting board (Picoquant GmbH, Berlin, Germany). A 515 nm diode-pumped solid-state laser (Cobolt, Solna, Sweden) and a 636 nm diode laser (Power Technology, Little Rock, AR, USA) were alternated at 20 kHz for excitation. In a typical experiment, data were collected for 10 min, and 1000–5000 bursts of fluorescence were detected.

27Photon arrival times in the donor and acceptor channel were sorted according to excitation period, resulting in four photon streams: I515D, donor emission during green excitation; I515A, acceptor emission during green excitation; I636D, donor emission during red excitation; and I636A, acceptor emission during red excitation. The total fluorescence emission was analyzed with a burst detection scheme as described. A burst was selected if a minimum of 100 photons arrived subsequently, with a maximum interphoton time of 100 μs. For each burst, we calculated the apparent FRET efficiency E (also known as proximity ratio): and the apparent label stoichiometry S:, in which N515D, N515A, and N636A are number of photons in the burst from the different photon streams, and γ is a parameter to correct for photophysical properties of the dyes, in our case equal to unity. The excitation powers were chosen such that N515A + g N515D ≈N636A, resulting in S∼0.5 for doubly labeled nucleosomes. Only nucleosomes with 0.2 < S < 0.8 were selected for FRET analysis.