Histone Core Phosphorylation Regulates DNA Accessibility*

Background: Transcription and DNA replication are regulated by histone core phosphorylation. Results: Histone phosphorylation near the DNA entry-exit region of the nucleosome increases DNA unwrapping and accessibility, which are further enhanced when combined with histone acetylation. Conclusion: Histone core phosphorylation regulates DNA accessibility. Significance: Histone phosphorylation and acetylation function together to regulate occupancy of DNA regulatory complexes. Nucleosome unwrapping dynamics provide transient access to the complexes involved in DNA transcription, repair, and replication, whereas regulation of nucleosome unwrapping modulates occupancy of these complexes. Histone H3 is phosphorylated at tyrosine 41 (H3Y41ph) and threonine 45 (H3T45ph). H3Y41ph is implicated in regulating transcription, whereas H3T45ph is involved in DNA replication and apoptosis. These modifications are located in the DNA-histone interface near where the DNA exits the nucleosome, and are thus poised to disrupt DNA-histone interactions. However, the impact of histone phosphorylation on nucleosome unwrapping and accessibility is unknown. We find that the phosphorylation mimics H3Y41E and H3T45E, and the chemically correct modification, H3Y41ph, significantly increase nucleosome unwrapping. This enhances DNA accessibility to protein binding by 3-fold. H3K56 acetylation (H3K56ac) is also located in the same DNA-histone interface and increases DNA unwrapping. H3K56ac is implicated in transcription regulation, suggesting that H3Y41ph and H3K56ac could function together. We find that the combination of H3Y41ph with H3K56ac increases DNA accessibility by over an order of magnitude. These results suggest that phosphorylation within the nucleosome DNA entry-exit region increases access to DNA binding complexes and that the combination of phosphorylation with acetylation has the potential to significantly influence DNA accessibility to transcription regulatory complexes.

Nucleosome unwrapping dynamics provide transient access to the complexes involved in DNA transcription, repair, and replication, whereas regulation of nucleosome unwrapping modulates occupancy of these complexes. Histone H3 is phosphorylated at tyrosine 41 (H3Y41ph) and threonine 45 (H3T45ph). H3Y41ph is implicated in regulating transcription, whereas H3T45ph is involved in DNA replication and apoptosis. These modifications are located in the DNA-histone interface near where the DNA exits the nucleosome, and are thus poised to disrupt DNA-histone interactions. However, the impact of histone phosphorylation on nucleosome unwrapping and accessibility is unknown. We find that the phosphorylation mimics H3Y41E and H3T45E, and the chemically correct modification, H3Y41ph, significantly increase nucleosome unwrapping. This enhances DNA accessibility to protein binding by 3-fold. H3K56 acetylation (H3K56ac) is also located in the same DNA-histone interface and increases DNA unwrapping. H3K56ac is implicated in transcription regulation, suggesting that H3Y41ph and H3K56ac could function together. We find that the combination of H3Y41ph with H3K56ac increases DNA accessibility by over an order of magnitude. These results suggest that phosphorylation within the nucleosome DNA entry-exit region increases access to DNA binding complexes and that the combination of phosphorylation with acetylation has the potential to significantly influence DNA accessibility to transcription regulatory complexes.
All eukaryotic genomes are organized into arrays of nucleosomes by tightly wrapping ϳ147 bp of DNA around histone octamer cores (1). In addition to compacting DNA, nucleosomes regulate the occupancy and access of the cellular machinery involved in transcription, replication, and repair to DNA by steric occlusion (2,3). Accessibility of DNA is influenced by ATP-dependent chromatin remodelers (4 -6), histone chaperones (7)(8)(9), and histone post-translational modifications (PTMs) 4 (10,11). Nucleosomes spontaneously partially unwrap due to thermal fluctuations, providing DNA binding complexes transient access to sites within the nucleosome (12,13). This occurs most frequently near the DNA entry-exit region of the nucleosome and can be influenced by histone PTMs (14).
Histone PTMs function by two general mechanisms. One is a signaling function whereby single or combinations of histone PTMs provide binding sites for recruiting specific regulatory complexes (10,11,15). PTMs located within the accessible histone tail regions of the nucleosome appear to primarily function by this signaling mechanism. The second mechanism is the direct alteration of nucleosome stability and dynamics by single or combinations of histone PTMs (14, 16 -18), which in turn regulates DNA accessibility to transcription, replication, and repair complexes.
H3K56ac functions by this second mechanism. This modification is located in the DNA-histone interface, about 10 bp into the nucleosome. This modification is located within promoters (19,20) and at sites of DNA repair (21); it is involved in nucleosome assembly during replication (22) and enhances transcription (23)(24)(25). H3K56ac significantly enhances the probability of DNA to partially unwrap (26) by increasing the unwrapping rate (27), which enhances DNA accessibility to proteins within the nucleosome (28,29).
A number of recently identified histone PTMs are also located within the entry-exit region of the nucleosome (30), including phosphorylation of H3Y41 (31) and H3T45 (32) (Fig.  1A). H3Y41 is phosphorylated by JAK2 in human cells and inhibits binding of HP1␣ (31,33), which is involved in heterochromatin formation. ChIP sequencing data indicate that H3Y41ph is present at transcriptional start sites and correlates closely with H3K4 trimethylation, a mark of active genes (34). Because both H3K56ac and H3Y41ph influence transcription, these modifications could occur within the same nucleosome. H3T45 is phosphorylated in human cells by PKC␦ (32) and DYRK1A (35). PKC␦ phosphorylation is associated with apoptosis, whereas DYRK1A phosphorylation represses HP1 binding similarly to H3Y41ph. In budding yeast, H3T45 is phosphorylated by the S phase kinase Cdc7-Ddf4 (36). The level of this modification peaks during DNA replication, and loss of H3T45ph causes replicative defects.
The addition of a phosphate group at H3Y41 or H3T45 introduces negative charge and steric bulk near the DNA phosphate backbone (Fig. 1A), potentially disrupting DNA-histone inter-actions. These observations, combined with previous results that H3K56ac and H3R42 trimethylation increase DNA unwrapping (26,37), suggest that H3Y41ph and H3T45ph may function to directly increase nucleosome unwrapping.
Here we report the influence of H3Y41 and H3T45 phosphorylation and phosphorylation mimics on nucleosome unwrapping and DNA accessibility. We find that the phosphorylation mimics H3Y41E and H3T45E, and the chemically correct PTM, H3Y41ph, significantly increase nucleosome unwrapping and increase DNA accessibility to transcription factor binding by about 3-fold. We then investigated the combination of H3Y41ph and H3K56ac. Together they increase DNA accessibility to transcription factor binding by 17-fold. Combined, these studies show that phosphorylation in the nucleosome entry-exit region increases nucleosome unwrapping and DNA accessibility similarly to H3K56 acetylation, whereas H3Y41ph in combination with H3K56ac can increase DNA accessibility by over an order of magnitude.

Experimental Procedures
Preparation of DNA Constructs-The nucleosomal DNA ( Fig. 1) used for small angle x-ray scattering (SAXS) (601-147) and micrococcal nuclease (MNase) (601-207) digestion assays was prepared as described in Ref. 38, whereas the nucleosomal DNA used for FRET measurements (DNA-LexA, Fig. 1) was prepared by PCR as described in Refs. 27 and 28. The template for the PCR was a plasmid containing the 601-nucleosome positioning sequence with the 8th through 27th base pairs replaced with the LexA recognition sequence (TACTGTAT-GAGCATACAGTA) (13,28). The oligonucleotides used as primers (Sigma-Aldrich) in the PCR were Cy3-CTG-GAGATACTGTATGAGCATACAGTACAATTGGTCGTA-GCA and ACAGGATGTATATATCTGACACGTGCCTG-GAGACTA. The oligonucleotide containing the LexA site contained an amine attached to the 5Ј end and was labeled with Cy3-NHS (GE Healthcare) and then purified by reverse phase HPLC. The 601-LexA PCR product was phenol-extracted and then purified by anion exchange chromatography. The 601 sequence is asymmetric in its propensity to unwrap (39,40). Here the LexA site has been introduced on the side of the nucleosome that has a reducing probability to unwrap.
Preparation of H3Y41ph with and without H3K56ac-Fully synthetic H3Y41ph and H3Y41ph/K56ac were prepared by sequential native chemical ligation as described previously (28,41) with the following changes. Peptides were synthesized with standard Fmoc-N␣ protection strategies using HCTU activation on an AAPPTec Apex 396 automated peptide synthesizer. All Met residues were substituted with norleucine to eliminate oxidative side products. C-terminal peptide H3(91-135)-A91C,C110A was synthesized on Rink Amide MBHA LL resin (Novabiochem). Peptides H3(47-90)A47Thz with or without K56ac and H3(1-46) with or without Tyr(P)-41 were prepared on Fmoc-Dbz(Alloc)-derivatized resin (where Dbz is 3,4-diaminobenzoic acid and Alloc is allyloxycarbonyl) as described in Ref. 42 and cleaved and purified as the C-terminal N-acylurea derivatives. Synthetic histones were analyzed by RP-HPLC and MALDI-TOF MS (Fig. 2). For FRET measurements, Cy3 is attached at the 5Ј end of the DNA molecule (green), and Cy5 is attached at H2AK119C (orange). The LexA binding site is located from the 8th to the 27th bp of the nucleosome (dark red). The region of the nucleosome containing the three PTMs is enlarged to indicate the residues' orientation relative to the DNA. B, schematic diagrams of the DNA molecules used in these studies. The green star indicates the location of the Cy3 fluorophore. C, EMSA of the nucleosomes (Nuc) used in FRET measurements.
Histone Octamer Preparation-Recombinant histone octamers were refolded and purified as described previously (38). The mutations for H3Y41E, H3T45E, H3K56Q, and H3Y41E/K56Q were introduced into the plasmid expressing H3C110A by sitedirected mutagenesis. Each histone was expressed and refolded into histone octamer following published procedures (38). Refolded histone octamer was purified by gel filtration chromatography. Histone octamer that was to be fluorophore-labeled was refolded with H2AK119C and was labeled before gel filtration purification.
Histone Octamer Cy5 Fluorophore Labeling-Histone octamers used in the FRET measurements included the H3C110A and H2AK119C substitutions and were labeled with Cy5 maleimide as described (28). Briefly, tris(2-carboxyethyl)phosphine (TCEP) was added at 10 mM to refolded histone octamer and incubated for 30 min. TCEP was then removed by dialysis into 5 mM PIPES, pH 6.1, with 2 M NaCl. The sample was then exposed to a stream of argon for 15 min. HEPES, pH 7.1, which was degassed under argon gas, was then added to the sample at a final concentration of 100 mM. Cy5-maleimide, which was resuspended in anhydrous dimethylformamide, was then added at 10 molar excess to histone octamer. The labeling reaction was incubated for 1 h at room temperature and then overnight at 4°C, and then quenched by adding DTT to 10 mM. The excess dye was removed during the histone octamer purification by gel filtration chromatography.
Nucleosome Reconstitution-Nucleosomes were reconstituted with DNA containing the 601-nucleosome positioning sequence by salt dialysis. Nucleosomes used in the SAXS and micrococcal nuclease digestion experiments were prepared as described in Ref. 38. The nucleosomes used in the FRET measurements were prepared as described in Ref. 27 and then purified on a 5-30% sucrose gradient. The quality of all nucleosome samples was analyzed by EMSA on 5% polyacrylamide gels (Fig.  1). The gels of fluorophore-labeled nucleosomes were imaged by detecting Cy3 fluorescence with a Typhoon fluorescence scanner (GE Healthcare).
Micrococcal Nuclease Measurements of Nucleosome DNA Accessibility-Nucleosomes reconstituted with 207-bp DNA that contained a centrally located 147-bp 601-nucleosome positioning sequence were subject to MNase digestion studies. MNase reactions were performed by combining 30 l of nucleosome (20 ng/l) or DNA (20 ng/l) into 2.5 l of BSA (10 mg/ml), 25 l of 10ϫ MNase buffer (New England Biolabs), 2 l (200 units/l) of MNase, and double-distilled H 2 O to bring up the total reaction volume to 250 l. 60 l of reaction mixture was collected at different time points. The reaction was quenched by 5 l of 0.5 M EDTA and stored on ice. 4.2 l of 10% SDS and 1 l of proteinase K (20 mg/ml) were added into each reaction mixture and incubated at 55°C for 30 min. DNA was isolated by phenol-chloroform extraction. The DNA quantity and length following a MNase digestion of nucleosomes with canonical H3, H3Y41E, or H3T45E were analyzed by 6% native PAGE (see Fig. 3). The relative mobility (R f ) of each DNA ladder band and MNase digestion product was measured with ImageQuant TL. The R f and length of each DNA ladder band were correlated and used to calculate the length of each band observed in the MNase reactions.
Size Exclusion Chromatography with Multi-angle Light Scattering (SEC-MALS)-Superdex 200 HR 10/30 column (24 ml total volume, GE Healthcare) was run in-line with the MALS instrument. The flow rate was 0.3 ml/min. 100 l of nucleosomes at 0.3 mg/ml was injected in a buffer containing 20 mM Tris, pH 7.5, 1 mM EDTA, pH 8.0, 1 mM TCEP, 0 mM KCl. The same samples were used in SAXS (at a different KCl concentration). The molecular weight for each sample was calculated using the ASTRA software (Wyatt technologies). SAXS-Nucleosomes containing the 147-bp 601-nucleosome positioning DNA sequence and WT, H3Y41E, or H3T45E octamers were used in the SAXS measurements. SEC-MALS (see Fig. 4) was used to check the purity and monodispersity of each nucleosome sample. All SAXS data collection was done at the SIBYLS beam line (12.3.1) at the Advanced Light Source (Berkeley, CA). Nucleosomes were measured either in the reference buffer (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM DTT) or in the reference buffer with 50 mM KCl to investigate the influence of ionic strength on nucleosome structure. To optimize the data quality and minimize radiation damage, exposure series of 0.5, 1, 2, and 6 s were performed. Data were processed by PRIMUS (44). The dimension of nucleosomes was estimated by GNOM (45). Ten random molecular envelopes were constructed for each nucleosome by DAMMIN (46). They were superimposed by DAMSUP (47). The average molecular envelopes of these 10 random models were calculated by DAMAVER (48). The averaged model was filtered by DAMFILT. Convex shells of all models were built and visualized (see Fig. 5) (49).
For radius of gyration (R g ) determination, data from 1/16th dilutions, exposed for 1 s, were used. This provided strong data at low angle with minimal inter-particle repulsion due to the charged nature of the nucleosomes. A typical example for WT was collected at ϳ0.1 mg/ml. Data from replicate experiments were evaluated simultaneously for each individual mutant. Radius of gyration values were calculated by Guinier analysis (50), with a new algorithm 5 applied to the triplicates of experimental scattering data. The algorithm optimizes a bias-variance tradeoff criterion and allows us to determine R g values at higher precision than previously possible, while at the same time providing statistically well founded uncertainties.
Fluorescence Resonance Energy Transfer Measurements of DNA Unwrapping-FRET efficiencies were determined from fluorescence spectra as described previously (27). Fluorescence spectra were measured with a Fluoromax-4 (Horiba) photoncounting steady-state fluorometer at room temperature (see Fig. 6). The Cy3 donor fluorophore was excited at 510 nm, and the fluorescence emission was measured from 550 nm to 750 nm. The Cy5 acceptor fluorophore was directly excited at 610 nm, and the fluorescence emission was taken from 650 nm to 750 nm. From these fluorescence spectra, the acceptor emissions due to donor and acceptor excitations were determined. Fluorescence emissions (F) were measured by integrating the fluorescence spectrum from 656 to 674 nm after subtracting out the fluorescence spectra of the sample buffer without nucleosomes. The FRET efficiency, E, was then calculated via the (ratio) A method as described previously (51) The superscripts refer to the donor (D) and acceptor (A) fluorophores, and the subscripts refer to the illumination frequencies 510 nm for donor excitation and 610 nm for direct acceptor excitation. A prefactor of 2 reflects the presence of two acceptor molecules per donor molecule. F A 510 is the fluorescence emission of the acceptor after the subtraction of overlapping donor emission when excited at 510 nm. F A 610 is the fluorescence emission of the acceptor when excited at 610 nm. ⑀ A 610 , ⑀ A 610 , and ⑀ D 510 are the molar extinction coefficients of acceptor and donor at 510 and 610 nm. d ϩ is the donor labeling efficiency, which is 1.
Measurements of LexA Binding within the Nucleosome-LexA binding is detected by measuring a reduction in the FRET efficiency that is caused by LexA trapping the nucleosome in a partially unwrapped state. To quantify changes in LexA binding, we performed LexA titrations from 0 to 10 M with 5 nM Cy3-Cy5-labeled nucleosomes in 0.5ϫ Tris-EDTA with 75 mM NaCl. Each 20-l sample was incubated for 3 min before taking an emission spectrum in the fluorometer at room temperature. For each data point, a corresponding blank spectrum was taken with the same LexA concentration to correct for background fluorescence. The FRET efficiency as a function of LexA concentration was fit to a non-competitive binding curve, , where E is the FRET efficiency, S1 ⁄ 2 is the concentration at which the FRET efficiency has decreased by half, and E 0 and E F are the initial and final FRET efficiencies. For each modification and mimic studied, we determined the relative S1 ⁄ 2 ϭ S1 ⁄ 2 mod /S1 ⁄ 2 unmod , which is inversely proportional to the relative change in the probability that the LexA target site is accessible for binding. Each titration was taken in triplicate, and the standard deviation of the three measurements was used as an estimate of the measurement uncertainty.

H3Y41E and H3T45E Increase Nucleosome Unwrapping-To
investigate whether phosphorylation of residues at the DNA entry/exit site affects DNA unwrapping and enhances accessibility within nucleosomes, we first prepared histones with individual H3Y41E and H3T45E substitutions. The glutamic acid introduces a negative charge that mimics certain aspects of phosphorylation. Nucleosomes were reconstituted with histone octamers containing these mutations, and 207-bp DNA molecules where the central 147 bp contain the 601-nucleosome positioning sequence (52). This allows for reconstitution of homogeneously positioned nucleosomes (Fig. 1).
To determine whether H3Y41E and H3T45E increase DNA unwrapping, we first assessed their impact on the rate of MNase digestion of nucleosomal DNA. MNase cleaves linker DNA, but pauses when encountering histone-bound DNA, resulting in the protection of an ϳ150-bp DNA fragment. Transient unwrapping of DNA from the histone octamer surface allows MNase to proceed further into the nucleosomes, resulting in increased rates of cleavage and smaller products.
We analyzed the MNase digestion time course of naked DNA, unmodified nucleosomes, and nucleosomes containing either H3Y41E or H3T45E by polyacrylamide gel electrophoresis (PAGE, Fig. 3). In unmodified nucleosomes, we observe a single 165-bp band after 60 s of MNase digestion, which is further digested to an ϳ140-bp fragment, corresponding to the nucleosome boundaries (12,13). In contrast, MNase digestion of nucleosomes containing H3Y41E or H3T45E did not result in a well defined 150-bp fragment. Instead, a distribution of shorter DNA lengths was produced following a 60-s MNase digestion, whereas a 120-s digestion converged to an ϳ120-bp fragment (Fig. 3B). This reduction in the length of DNA pro-tected from MNase digestion by the histone octamer is a strong indication of increased DNA unwrapping. By comparison, crystallographic analysis of nucleosomes containing the H3 histone variant CenpA shows no changes in structure apart from the disorder observed for the last 10 bp on either end (55). Single molecule measurements have shown that CenpA-containing nucleosomes have increased exposure of terminal DNA and protect ϳ120 bp of DNA from MNase digestions (56). In combination, this indicates that MNase has increased access to DNA within nucleosomes containing either H3Y41E or H3T45E, suggesting that these phosphorylation mimics increase DNA unwrapping from the histone octamer.
To further investigate the influence of H3Y41E and H3T45E on overall nucleosome structure, we carried out SEC-MALS and SAXS measurements. Nucleosomes were reconstituted with 601-147 DNA (Fig. 1B) and unmodified histones or histones containing either H3Y41E or H3T45E. SEC-MALS confirmed that nucleosomes reconstituted both with and without these modification mimics were monodisperse and had the expected molecular mass of ϳ200 kDa (Fig. 4). This verifies that nucleosomes with these amino acid mutations form canonical nucleosomes and not altered DNA histone complexes such as the altosome (53, 54), or nucleosomes lacking histones.
SAXS measurements allow us to determine the molecular envelope and R g , which is defined as the average distance from the center of mass for the ensemble of molecules (Fig. 5A). At low ionic strength, unmodified nucleosomes have an R g of 43.1 Ϯ 0.15 Å, consistent with published data (49), whereas nucleosomes with H3T45E and H3Y41E both have statistically significant increased R g values. This effect is more pronounced for H3T45E at 50 mM salt. This is also apparent in the molecular envelopes calculated from SAXS data collected at low ionic strength, which are consistent with increased DNA unwrap-  To determine the molecular mass and quality of nucleosomes prior to SAXS experiments, SEC-MALS was performed. The molecular mass for each sample was calculated by the ASTRA software (Wyatt Technologies). The calculated mass for each nucleosome is within error of their theoretical mass (200 KDa), and samples are monodisperse. Because the light-scattering signal from the peak that eluted at 7-8 ml in each nucleosome sample is also present in the buffer control (data not shown), it is not due to free DNA or other contaminants. a. u., arbitrary units. ping from the entry-exit region of the nucleosome (Fig. 5B). These results provide additional evidence that phosphorylation of the H3 residues around the nucleosome entry-exit region increases DNA unwrapping to enhance accessibility of proteins to DNA sites within the nucleosome.
H3Y41 Phosphorylation and H3Y41E and H3T45E Each Increase DNA Accessibility within the DNA Entry-Exit Region of the Nucleosome-We next considered whether the increase in nucleosome unwrapping detected by MNase and SAXS measurements might enhance the interaction of proteins with DNA within the entry-exit region of the nucleosome. To investigate this, we used a FRET-based assay developed by Li and Widom (13), modified to assess the impact of histone PTMs on DNA unwrapping (27)(28)(29). In this assay, the recognition sequence of a model transcription factor, LexA, is introduced near the DNA entry-exit region of the nucleosome such that LexA binding is occluded in the wrapped nucleosome state but accessible in the partially unwrapped state. The nucleosome is reconstituted with the 601-LexA DNA molecule, in which the 8th through 27th base pairs of the 147-bp 601-nucleosome positioning sequence are substituted with the LexA recognition sequence (Fig. 1B). The DNA molecule is labeled with the Cy3 donor fluorophore at the 5Ј end near the LexA site, whereas the histone octamer is labeled at H2AK119C with the acceptor fluorophore, Cy5. In the fully wrapped nucleosome state, the Cy3-Cy5 pair exhibits energy transfer, whereas a partially unwrapped nucleosome that is trapped by LexA binding generates a significantly lower energy transfer efficiency. As LexA is titrated from 30 to 3000 nM with a fixed nucleosome concentration of 5 nM, LexA binds its recognition site and traps nucleosomes in a partially unwrapped state. A LexA titration with these Cy3-Cy5-labeled nucleosomes results in a decrease in FRET efficiency. This can be fit to a non-cooperative binding isotherm to determine the S1 ⁄ 2 , which is the concentration of LexA where 50% of the nucleosomes are bound (Fig. 6). A change in the S1 ⁄ 2 is a quantitative measure of a change in the accessibility of the LexA target site, such that a decrease in the S1 ⁄ 2 implies an equal increase in the LexA site accessibility (13,27).
We carried out LexA titrations with nucleosomes containing the phosphorylation mimic H3Y41E and found that it decreased the S1 ⁄ 2 by 2.8 Ϯ 0.4-fold. Because glutamate is chemically distinct from phosphotyrosine, we prepared the modified protein H3Y41ph using sequential native chemical ligation (Fig. 2). We find that H3Y41ph decreases the LexA S1 ⁄ 2 by 3.1 Ϯ 0.4 relative to unmodified nucleosomes. This suggests that increases, which is due to LexA binding and trapping the nucleosome in a partially unwrapped state with lower FRET. au, arbitrary units. C, change in FRET efficiency for increasing concentrations of LexA with unmodified (black), H3Y41ph (green), and H3Y41ph/K56ac (blue) nucleosomes. FRET efficiency changes are normalized to change from 1 to 0 and are fit with a noncooperative binding curve. D, relative S1 ⁄2 reduction (S1 ⁄2 unmod /S1 ⁄2 PTM ) for each single PTM and PTM mimic. Error bars reflect the uncertainty of the S1 ⁄2 mean over three measurements.

TABLE 1
Summary of DNA accessibility to LexA binding within the nucleosome Columns 3 and 4 are included for a comparison between the measurements of nucleosomes containing a single PTM and the measurements of nucleosomes containing two PTMs. If the LexA ⌬⌬G of each individual PTM is additive, then the LexA S 1/2 will be multiplicative. NA, not applicable.
within error H3Y41E accurately mimics the influence of H3Y41ph on the probability of LexA binding to its DNA target site (Fig. 6, Table 1). This change in probability is related to the change in the free energy difference between the exposed and unexposed state by ⌬⌬G PTM ϭ Ϫk B T ln (S1 ⁄ 2 PTM /S1 ⁄ 2 unmod ) and implies that ⌬⌬G Y41ph ϭ 1.1 Ϯ 0.1 k B T ϭ 0.7 Ϯ 0.1 kcal/mol. We next investigated the impact of the phosphorylation mimic H3T45E, which organizes the same DNA minor grove as H3Y41. We found that the glutamate substitution for H3T45 decreased the S1 ⁄ 2 by a factor of 2.2 Ϯ 0.5, which is similar to the reduction observed for H3Y41E, H3Y41ph, H3K56Q, and H3K56ac (Fig. 6, Table 1). We conclude that phosphorylation at H3T45 is likely to increase site exposure for DNA-protein binding within the entry-exit region of the nucleosome similarly to phosphorylation at H3Y41. Furthermore, the similarity between histone modifications in the entry-exit region (Fig. 6, Table 1) suggests that single histone PTMs in this region tend to increase DNA accessibility by about a factor of 3, irrespective of the precise location and nature of the modification.
The Combination of H3Y41ph and H3K56ac Multiplicatively Increases DNA Accessibility-Nucleosomes often contain combinations of histone PTMs (11). We considered the possibility that H3Y41ph and H3K56ac could occur within the same nucleosome because they both are involved in transcriptional regulation and have been identified by ChIP sequencing to occur within nucleosomes around transcription start sites (19 -22, 31, 33, 34). We therefore constructed H3Y41ph/K56ac histone by sequential native chemical ligation. We then prepared Cy3-Cy5-labeled nucleosomes containing the 601-LexA DNA sequence with histone octamer containing H3Y41ph/K56ac. We assessed via FRET the S1 ⁄ 2 of LexA binding to its site in partially unwrapped nucleosomes. We find that the S1 ⁄ 2 of LexA binding to nucleosomes containing H3Y41ph and H3K56ac was reduced by a factor of 17 Ϯ 5 relative to unmodified nucleosomes (Fig. 7, Table 1). This implies that these two modifications in combination dramatically increase the probability that a DNA target site within the entry-exit region is exposed for DNA-protein binding relative to either modification alone.
We compared this S1 ⁄ 2 for LexA binding to nucleosomes containing both H3Y41ph and K56ac to the LexA S1 ⁄ 2 with either individual PTM. If the two PTMs independently influence LexA binding, each individual ⌬⌬G should combine additively, i.e. ⌬⌬G Y41ph ϩ ⌬⌬G K56ac ϭ ⌬⌬G Y41ph/K56ac , and each individual relative S1 ⁄ 2 should combine multiplicatively, i.e., ͩ S 1/ This is because the S1 ⁄ 2 of LexA binding to a modified nucleosome relative to an unmodified nucleosome is related to the ⌬⌬G by a Boltzmann weight, Therefore, the expected multiplicative change in the S1 ⁄ 2 is 10 Ϯ 2 (Fig. 7, Table 1). The observation that the reduction in S1 ⁄ 2 Y41ph,K56ac is greater than this product implies that at minimum H3Y41ph and H3K56ac combine to multiplicatively increase DNA accessibility.
This 17-fold decrease in the S1 ⁄ 2 converts to a ⌬⌬G Y41ph,K56ac of 2.8 Ϯ 0.2 k B T or 1.7 Ϯ 0.1 kcal/mol, which is about 5-10% of the free energy for nucleosome formation (57,58). The sum of the individually measured free energy changes induced by H3Y41ph and H3K56ac is ⌬⌬G Y41ph ϩ ⌬⌬G K56ac ϭ 2.3 Ϯ 0.4 k B T ϭ 1.4 Ϯ 0.2 kcal/mol. This indicates that there is about 0.5 k B T of additional ⌬⌬G that is introduced by combining these PTMs and results in the measured S1 ⁄ 2 Y41ph,K56ac being about a factor of 2 higher than predicted from the product of each individual S1 ⁄ 2 .
Combining H3Y41ph and H3K56ac Amplifies the Differences between Their Mimics-Given the wide use of mutations to mimic PTMs, we investigated the combined influence of the histone PTM mimics H3Y41E and H3K56Q. We find that the combined amino acid substitution H3Y41E/K56Q reduces the S1 ⁄ 2 of LexA binding to nucleosomes by 4 Ϯ 1 (Fig. 6, Table  1). This reduction of the S1 ⁄ 2 is over three times less than the reduction caused by H3Y41ph/K56ac. These results indicate that small differences between individual histone PTMs and their mimics can be magnified when combined together and demonstrates a limitation to using mutations to mimic PTMs.

Discussion
We find that H3Y41ph, H3Y41E, and H3T45E increase DNA accessibility by increasing unwrapping. A similar effect has also been reported for H3R42me2 (37) and H3K56ac (26,28). Together, these results indicate that histone PTMs within the DNA entry-exit region of the nucleosome generally increase DNA accessibility by about 3-fold. These PTMs are involved in transcription activation (21,22,31,33,35,37), and H3Y41ph (34) and H3K56ac (19,20) have both been identified by ChIP sequencing to occur within nucleosomes in the regions around the transcription start site. Therefore, this unwrapping mechanism is consistent with a biological function where these PTMs increase DNA accessibility to transcription regulatory complexes facilitating gene expression.
Our observation that H3Y41ph and H3K56ac in combination increase DNA accessibility by over an order of magnitude, whereas individual modifications increase accessibility by 2-3fold, suggests that different combinations of PTMs can be used to tune DNA accessibility. This could be used to precisely control the level of transcription. Even a modest 2-fold change in FIGURE 7. Relative reduction of the S1 ⁄2 (S1 ⁄2 unmod /S1 ⁄2 PTM ) for the combination of H3Y41ph and H3K56ac. A, H3Y41ph and H3K56ac. B, H3Y41E and H3K56Q. Single mimics shown in green are from Fig. 6D. Double mimics are shown in blue. Red shows the product of the LexA S1 ⁄2 with the nucleosomes containing the single modifications, S1 ⁄2 mod1 ϫ S1 ⁄2 mod2 . If the modifications change in unwrapping free energy are additive, the S1 ⁄2 values should be multiplicative. Error bars indicate the uncertainty of the S1 ⁄2 mean over three measurements.
gene expression is biologically significant as is observed for dosage compensation and haplo-insufficiency diseases. H3Y41, H3T45, and H3K56 all coordinate the same 5-bp minor groove section of DNA (Fig. 1A). The side chain of H3Y41 extends into the minor groove of the DNA double helix at the first (and last) contact between histones and DNA backbone. It does not make any direct interactions with the DNA but rather appears to prevent further compression of the minor groove, thereby determining the entry and exit angle of linker DNA. The addition of a phosphate to this amino acid generates a steric clash in addition to a repulsive charge, thereby shifting the site exposure equilibrium toward partially unwrapped states. Although phosphotyrosine is significantly larger than glutamate, H3Y41ph and H3Y41E induce a similar increase in DNA accessibility. This suggests that electrostatic effects, rather than steric effects, determine DNA site accessibility at this site. In addition, the doubly ionizable phosphotyrosine will introduce more negative charge than the singly ionizable Glu side chain. Our observation that H3Y41E in combination with H3K56Q did not fully replicate the increased unwrapping induced by H3Y41ph in combination with H3K56ac suggests that steric effects and/or differences in electrostatic interactions become more important when H3K56ac is combined with H3Y41ph.
H3K56 is oriented near the phosphate of the 9th base pair of the DNA, and stabilizes the same DNA minor groove as H3Y41 (Fig. 1). Acetylation of H3K56 eliminates this interaction, reducing the binding free energy and increasing the probability for the DNA to partially unwrap. H3K56 and H3Y41 are on opposite sides of this DNA minor groove such that the modification of these residues disrupts separate interactions with the same section of DNA. This should result in H3Y41ph and H3K56ac independently influencing the ⌬⌬G of DNA unwrapping. This is confirmed by our observation that the change in the S1 ⁄ 2 for LexA binding is multiplicative, implying that the ⌬⌬G values of H3Y41ph and H3K56ac are additive. Furthermore, these results suggest that other histone PTM combinations that disrupt separate interactions with the same DNA minor groove could multiplicatively increase the probability for DNA unwrapping.
The hydroxyl group of H3T45 forms a hydrogen bond with the DNA phosphate between the 3rd and 4th base of the 3Ј DNA strand. Either phosphorylation of threonine or substitution with a glutamate mimic as in our H3T45E nucleosomes would disrupt this hydrogen bond. We therefore cannot rule out reduction of a favorable DNA-histone interaction in the context of the H3T45E nucleosomes, in addition to the likely electrostatic repulsion resulting from introduction of negative charge through glutamate substitution or phosphorylation. However, not only are glutamate and phosphothreonine relatively similar in size, but there appears to be space for phosphothreonine to extend between the DNA and the C-terminal tail of H2A in the structure of the wrapped nucleosome, suggesting that glutamate may accurately mimic H3T45ph.
The question of whether mimics can replicate the effect of a naturally occurring chemical modification is highly debated. Glutamine can be an adequate mimic of acetylated lysine, for example in the effect of histone tail acetylation on chromatin self-association (59). However, we previously found that the influence of acetyllysine and phosphothreonine on nucleosome stability is not fully replicated by glutamine (60) and glutamate (53,61), respectively, in the nucleosome dyad region. The effect of mimics in the entry-exit region is less clear. Individual mimics of acetylation or phosphorylation quantitatively replicate the effects of the modification, as demonstrated in our findings here. However, we show here that relatively subtle differences between mimics and modifications are magnified when the residues function synergistically, as for H3Y41ph/K56ac. These findings suggest an additional note of caution regarding interpretation of results acquired using acetylation or phosphorylation mimics in combination in the entry-exit region of the nucleosome.
The kinases that phosphorylate H3Y41 and H3T45 are expected to be sterically occluded from these sites within fully wrapped nucleosomes. This raises a question of how these modifications are introduced. H3K56, which is also inaccessible within fully wrapped nucleosomes, is acetylated before H3 is assembledintonucleosomes (22).Inbuddingyeast,H3T45phosphorylation occurs within newly assembled chromatin (36), suggesting that H3T45 might be phosphorylated before nucleosome assembly. It is also possible that nucleosome unwrapping would provide kinases transient access to both H3Y41 and H3T45.
A large number of additional histone PTMs have been identified that are located within the DNA histone interface (30). Our studies suggest that many of these histone PTMs may function to increase DNA accessibility via enhanced DNA unwrapping. Furthermore, combinations of histone PTMs may function together to more significantly increase DNA accessibility. Future studies that investigate these additional histone PTMs and other histone PTM combinations in the DNA entry-exit region of the nucleosome are essential for understanding the function of the numerous PTMs that are located within this region of the nucleosome.