Mass Spectrometric Characterization of Human Histone H3: A Bird’s Eye View C. Eric Thomas,† Neil L. Kelleher,†,‡,§ and Craig A. Mizzen*,‡,| Department of Chemistry, Institute for Genomic Biology, The Center for Top Down Proteomics, and Department of Cell and Developmental Biology, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801 Received August 16, 2005
The modification of H3 in asynchronous HeLa cells was profiled using Top Down Mass Spectrometry. A broad distribution of species differing by 14 Da and containing less than 3% unmodified protein was observed for all three variants. Species of up to +168 Da were observed for H3.1, and fragmentation of all species by Electron Capture Dissociation (ECD) revealed ∼5% methylation of K4 and ∼50% dimethylation of K9. K14 and K23 were major sites of acetylation. H3.3 was slightly hypermodified with the apex of the distribution shifted by ∼+14 Da compared to H3.1. H3.1 (50% and 15%) from colchicine-treated cells was monophosphorylated and diphosphorylated, respectively, with equivalent modification of S10 and S28. Keywords: H3 • histone code • histone • Fourier transform mass spectrometry (FTMS) • electron capture dissociation (ECD) • chromatin • post-translational modification (PTM) • acetylation • methylation • phosphorylation
Introduction The field of histone proteomics is expanding, in large part, due to increasing evidence of the diverse regulatory roles played by this group of proteins.1-4 While it has been known for some time that histones were involved in the packaging of DNA into eukaryotic nuclei, their importance in transcriptional regulation is now widely recognized. With multiple post-translational modifications (PTMs) found largely at basic residues within their N-terminal tails, early models for the effects of acetylation and methylation focused on the ability of histone tails to bind DNA (see refs 5-7 for reviews). However, the discovery of bromo-8 and chromodomains,9,10 which are present in a variety of chromatin regulators and can interact with specific acetylated and methylated residues, respectively, has led to the Histone Code hypothesis to explain the regulation of transcription and chromatin remodeling.6,11 In general, this model proposes that combinations of PTMs can mediate specific phenotypic effects. More recently, the nature and existence of a histone code12-14 has been debated. Beyond the notion that PTM combinations on the same histone tail encode precise regulatory information, modifications have been found to persist through passage of generations.13 This provides an epigenetic mechanism to dictate cellular characteristics based on histone modifications, resulting in a form of “cellular memory”. H3 contains the greatest number of known modification sites of the four core histones.15 As many as 14 residues have been * Corresponding author: Craig A. Mizzen, B107 CLSL, 601 S. Goodwin Avenue, Urbana, IL 61801. Tel, (217) 244-4896; e-mail,
[email protected]. † Department of Chemistry. ‡ Institute for Genomic Biology. § The Center for Top Down Proteomics. | Department of Cell and Developmental Biology.
240
Journal of Proteome Research 2006, 5, 240-247
Published on Web 01/11/2006
Figure 1. (A) Sequence alignment of the human H3 gene family. Nonconserved residues are shown in bold. (B) RPLC separation of histones from HeLa cells. All core histones are observed. H2A variants elute as two resolved peaks. H3 gene products elute last, with incomplete resolution of H3.3 and H3.2. (C) 18+ Charge state from the broadband spectrum of H3.2. (D) 18+ Charge state from the broadband spectrum of H3.3. Note the slightly higher relative abundance of hypermodified forms in the spectrum for H3.3 compared to that of H3.2.
identified to be modified through various in vitro and in vivo studies of multiple organisms; however, the extent to which these residues are modified, singly or in combination, is poorly understood. In humans, three major variants exist: H3.1, H3.2, and H3.3 (Figure 1A). H3.1 and H3.2 differ by only a single 10.1021/pr050266a CCC: $33.50
2006 American Chemical Society
research articles
MS Characterization of Human Histone H3
residue (C96 is replaced by S in H3.2) and have yet to be physiologically differentiated. H3.1 differs by 5 residues from H3.3, it is expressed exclusively during S phase, and its deposition into nucleosomes has been observed to be replication-dependent, involving the CAF-1 chaperone complex. H3.3, however, is expressed throughout the cell cycle and is deposited independent of replication through the chaperone activity of the HIRA complex.16-18 While gene and protein functions can often be conserved across species, histone modification patterns appear to be rather organism-specific. For example, K14 and K23 of H3 are the first acetylated residues in butyrate-treated HeLa cells,19 but K9 and K14 are modified first in the ciliated protozoan Tetrahymena thermophila.20 Of particular relevance, it has also been noted that plant species with larger genomes display uniform methylation of H3 K9, whereas plants with smaller genomes apparently restrict methylation to regions of constitutive heterochromatin.21 Recent analyses of H3 have unveiled modifications and combinations of modifications present in Drosophila,22 Arabidopsis,23 and human cells.24-26 Methylation of K9 has been linked to transcriptional silencing13,27-29 and is of particular interest. Immunochemical characterization has linked K9 trimethylation to pericentric heterochromatin, whereas mono- and dimethylated forms are enriched in silenced domains within euchromatin.30 Still unstudied, however, are the effects of other modifications on methylated K9-directed silencing and on antibody recognition of methylated K9. Given the amino acid sequence- and PTM-heterogeneity of H3, much that is known about its modification states is limited in scope. Antibody probes have revealed much about modification dynamics,9,10,17,22,24,31,32 but little is known about the extent to which the binding of antisera to one PTM is affected by PTMs at nearby sites. Moreover, antisera to specific PTMs, by definition, do not provide detailed information on the complete modification state of a histone. Bottom-up MS is used increasingly in histone PTM characterization,22,23,33-35 as small peptide ions are readily visualized and their modifications quantified. Unfortunately, such approaches hinder assessment of PTM combinations unless they reside on the same peptide. Intact protein analysis can provide a more comprehensive characterization of abundant histone modifications.35,36 Global modification states of intact H3 have been reported at low resolution and without tandem mass spectrometry or significant quantitation.37 Top Down MS has been used in our lab to examine human histone H4,36 and by others interrogating H2B.35 In addition to the inherent advantages of FTMS-driven Top Down analysis, the introduction of Electron Capture Dissociation (ECD) allows nearly complete characterization of target proteins by extensive cleavage of CR-N bonds throughout the protein. Additionally, ECD affords the retention of all PTMs, even if labile to threshold methods for ion fragmentation.38,39 The results reported here describe the global modification pattern of histones H3.1 and H3.3 from asynchronously growing or colchicine-treated HeLa cells that were revealed by characterization in a Top Down fashion. A multiplexed MS/MS procedure, utilizing ECD of all forms en masse (previously utilized in the examination of Tetrahymena H2B35 and bovine H440), allowed several modification states of H3.1 to be characterized. While only 5% of K4 was modified (monomethylated), about half of the K9 residues were dimethylated. Additionally, K14 and K23 were observed to be major sites of acetylation in the H3.1 pool. In colchicine-treated cells, both S10 and S28 were phosphorylated. Assuming a ∼30-fold
dynamic range, this study detected forms of H3 present at greater than ∼3% abundance and provides a perspective for intricate characterization of the myriad chemically distinct H3 molecules produced by human cells.
Experimental Procedures Preparation of Histones. HeLa S3 cells were grown in spinner culture flasks in Joklik’s media supplemented with 5-10% newborn calf serum (Cambrex, East Rutherford, NJ) and maintained at a density of (2-3) × 105 cells/mL. For colchicine arrest, 1 µM colchicine was added to cells at 3 × 105/mL, which were grown for an additional 20 h. Cells were harvested by centrifugation, washed twice in TBS, and resuspended in nucleus isolation buffer (15 mM Tris-HCl (pH 7.5), 60 mM KCl, 15 mM NaCl, 5 mM MgCl2, 1 mM CaCl2, 250 mM sucrose, 1 mM dithiothreitol (DTT), 5 nM microcystin-LR, 500 µM 4-(2aminoethyl)benzenesulfonyl fluoride (AEBSF), and 10 mM sodium butyrate), supplemented with 0.3% NP-40 to afford disruption of the cell membrane. Following two washes in nucleus isolation buffer, histones were extracted from nuclei with 0.4 N H2SO4 and recovered by trichloroacetic acid (TCA) precipitation. About 100 µg of acid-extracted protein was loaded onto a 4.6 mm × 250 mm C18 column (Vydac, Hesperia, CA) for reverse-phase liquid chromatography (RPLC) (System Gold, Beckman Coulter, Fullerton, CA). Histones were separated using a gradient of 30-60% B, with solvents A and B containing 5% acetonitrile (ACN), 0.1% trifluoroacetic acid (TFA), and 90% ACN, 0.092% TFA, respectively. H3.1 eluted as a single peak at 86.5 min, and H3.3 eluted at 78 min, with nearly baseline resolution from the H3.2 peak at 79.5 min (Figure 1B). A second dimension of chromatography on a 4.6 mm × 250 mm CM 300 column (Eprogen, Darien, IL) was used to separate species recovered from RPLC according to acetylation status. A gradient of 0-500 mM NaCl in 23 mM sodium phosphate (pH 7.0) and 8 M freshly deionized urea effected separation of H3.1 into four peaks, eluting at 37.5, 40, 42.5, and 47.5 min. These peaks were subsequently desalted via RPLC. All RPLC protein fractions were dried to completion, washed twice with 20% trichloroacetic acid, and then washed once with 0.1% HCl in acetone and twice with acetone prior to drying. FTMS. Intact histones were resuspended in an electrospray solution consisting of equal parts water and methanol, and 1% formic acid. All data were acquired on a custom 8.5 T Quadrupole-FT ICR MS with a nano-ESI source operated in positive ion mode.41 A NanoMate 100 (Advion BioSciences, Ithaca, NY) was used to automatically establish the nanospray. Collected data sets consist of 100-400 scans with ion accumulation times of up to 4 s for each scan. Entire histone charge states, containing all protein forms, were selectively enhanced using the instrument’s quadrupole and further filtered using Stored Waveform Inverse Fourier Transform (SWIFT).42 ECD was performed with a filament bias of 1.2 V, using 35 loops of 7 ms electron pulses. Data were collected using the modular ICR data acquisition system (MIDAS).43 Data Analysis. Following broadband data acquisition (e.g., Figures 1C, 2A), ECD data analysis involved interpretation of c and z• ions (Figures 2B, 2C) using THRASH.44 Unmodified z• ions were used for internal calibration of the spectra. For exhaustive identification of modified ions and quantitation, some manual analysis was required. Only c-type ions were observed to report on modification states, and their quantitation involved summing abundances of the first two isotopic peaks for each mass, resulting in fragment ion relative ratios,45 Journal of Proteome Research • Vol. 5, No. 2, 2006 241
research articles
Figure 2. (A) 18+ Charge state from the broadband spectrum of H3.1. Numerals correspond to the number of +14 Da shifts from the unmodified ions. (B) ECD spectrum from the 18+ charge state of H3.1 The portion of the spectrum from 460-800 m/z is shown. (C) ECD graphical fragment map of the 18+ charge state of H3.1. The map was generated at 20 ppm tolerance in ProSight’s single protein mode with K9 dimethylated and K23 acetylated.
values were subsequently converted to “percent occupancies” by dividing individual ion abundances by the sum of all values for the ion set. Four independent data sets, from 17+, 18+, 20+, and 23+ parent ions, were averaged to determine the reported values for each species. On the basis of observed signal-to-noise ratios, the limit of detection for these ECD experiments was approximately 5% occupancy. Overall mass accuracy was observed to be within 15 ppm of theoretical. Multiple c ions, each with multiple satellite peaks, were used for quantitation; as many ions exist in these ECD spectra, analysis was performed on fragment ions observed to have the least overlap with other isotopic distributions. The lack of K79 quantitation from ECD data arises from the aforementioned overlap of ion peaks, not from lack of fragment ions reporting on this position.
Thomas et al.
Figure 3. Comparison of c92+ fragments reporting on K9 of H3.1. Unmodified, mono- and dimethyl, and acetyl species are designated by the clear boxes.
Results
Figure 4. Building up the modification profile of H3.1 from the N-terminus. An increase in modification complexity by N f C directional comparison of small c-type fragment ions. The intact profile has been cropped for clarity (ions of up to +168 Da are observed in the spectrum), and a # denotes ions observed with overlapping isotopic distributions.
Acid-extracted HeLa cell H3 eluted in three peaks during RPLC (Figure 1B). Broadband spectra were obtained for H3.1, 3.2, and 3.3 (Figures 2A and 1C,D), followed by manual identification of ECD fragments from H3.1 resulting in extensive coverage of both N- and C-terminal tails (Figure 2B,C). Four sets of seven c fragment ions (collected from four different charge states of H3.1) were used to detect the PTM occupancy of specific sites. The c3 ion exhibited no modifications, indicating no N-terminal acetylation. The c41+ ion cluster was observed to exist as a pair of masses corresponding to unmodified and monomethylated K4. The next ion set used for quantitation was the c9. The most prominent c9 ion observed was the dimethylated form, which accounted for approximately 50% of the ion set (Figure 3). Dimethylation was also the most abundant modification state in the c14, c17, c18, and c23 ion sets, accounting for ∼23% of each. The c10 ions were found to contain the same distribution of modifications as the c9 within 5%, (data not shown), but no phosphorylation of S10 or any other residues was observed in asynchronous material. Because of the presence of more modified residues with increasing fragment ion size, PTM profiles became increasingly
complex (Figure 4).The c14 ion set revealed an apparent acetylation at K14, as the highest observed mass is 70 Da greater than the unmodified mass (and 42 Da greater than the dimethyl form). This +70 species accounted for 15% of the observed c14 ions (Figure 5B). No discernible change was observed in the ratios for ions reporting on R17 (Figure 5C). The c18 ion set also revealed little change in modification state (Figure 5D); however, this set revealed a +84 Da modification state, which is most consistent with minor methylation at K18. Acetylation of K23 was also apparent, as the +84 pool grew to account for 20% in the c23 ion set, while the +42 pool decreased by greater than 2-fold (Figure 5E). Modifications at R26 and/or K27 were apparent from analyses of the c28 ion set, which revealed a significant change compared to the PTM distribution extracted from the c23 ions (Figure 5F). The c28 ion set contains species with modifications of up to +98 Da, with the +56 Da species accounting for 30%. The marked diminution in unmodified and monomethylated species from the c23 ion set to the c28 ion set, combined with the ∼10-fold increase in +56 Da species (Figure 5, panel E vs
242
Journal of Proteome Research • Vol. 5, No. 2, 2006
research articles
MS Characterization of Human Histone H3
Figure 5. Modifications observed in ECD-generated c ions from H3.1. The fractional occupancy of each species was calculated for each of four charge states. Histograms depict the average of these four values and standard deviations are shown as error bars. Table 1. Mass Accuracies for Assigned Modifications of H3.1a ion set c9 c14 c17 c18 c23 c28
c36
∆m 42.044 42.043 42.048 42.044 42.061 84.026 42.039 56.050 70.080 42.052 56.062 70.099 84.082 98.063 112.07
Ac Ac2 Me3 (ppm) (ppm) (ppm) 30.6 20.8 16.3 17.3 6.9 44.0 12.9 17.9 16.4 7.1 11.3 8.4 19.6 31.4 37.0
7.6 3.9 6.3 3.9 9.8 27.6 0.9 4.1 2.7 4.2 0.1 2.8 8.4 20.3 25.9
na na na na na 11.3 na na na na na na 2.7 9.2 14.8
residues
assignment
K9 K14, (S10) R17 K18 K23
acetyl acetyl acetyl acetyl trimethyl and acetyl diacetyl acetyl acetyl + monomethyl acetyl + dimethyl trimethyl or acetyl acetyl + monomethyl acetyl + dimethyl diacetyl diacetyl + monomethyl diacetyl + dimethyl
R26, K27, (S28) K36
a Ions corresponding to species of +42 Da or greater were evaluated to assign +42 Da shifts to either acetylation or trimethylation. For masses of +56 or +70 Da, the assignment of acetylation is in addition to monomethylation or dimethylation, respectively. Additionally, for masses corresponding to +84 Da or greater, assignment of modification included the possibilities of hexamethylation, acetylation + trimethylation, or diacetylation. Errors, in ppm, are reported with respect to theoretical mass of the modified ions; for unobserved species, this error is reported as not applicable (na).
F) is consistent with methylation at R26 and K27 occurring in tandem or alone. The +56 Da species corresponds to a monoacetylated monomethylated species, based on the 4 ppm error versus an 18 ppm error for tetramethylation (Table 1). Inspection of the PTM patterns for the c37 ion set relative to the intact ion pools revealed no significant change (Figure 5G); however, methylation of K36 is possible, as +126 Da species were apparent in the c37 ion set, and species as large as +168 Da were observed in the intact spectrum. C-Terminal fragments reporting on K79 were observed, with no evidence of modification. On the basis of the higher level of chemical noise in the
Figure 6. (A) Comparison of K4-terminal fragments from H3.1 and H3.3. The fraction of unmodified and monomethyl species observed are plotted. (B) Comparison of K9-terminal fragments from H3.1 and H3.3. The fraction of unmodified, monomethylated, dimethylated, and monoacetylated species observed are plotted.
m/z regions where these fragments were found, it is possible that other modified forms were present at