Histone H4 Proteoforms and Post-translational Modifications in the Mus musculus Brain with Quantitative Comparison of Ages and Brain Regions

Abstract

Histone proteins are essential to the regulation of the eukaryotic genome. Histone post-translational modifications (PTMs) and single-molecule combinations of these modifications (proteoforms) allow for the regulation of many DNA-templated processes, most notably transcription. Histone H4 is a part of the core histone octamer, which packages DNA into nucleosomes. Top-down proteomics allows for the inquiry of the epigenetic landscape with proteoform-level specificity. Although these approaches are well-demonstrated ex vivo, our knowledge of in vivo histone proteoform biology remains sparse. Here we demonstrate the first in vivo quantitative top-down analysis of histone H4 and analyze the forebrains and hindbrains of differently aged mice. This reveals novel differences between the mouse forebrain and hindbrain and region-specific changes during adolescence in histone H4 PTMs and proteoforms. At 25 days of age (P25), histone H4 of the hindbrain is more acetylated than the forebrain. At 47 days of age (P47), there are fewer significant differences in histone H4 PTMs and their combinations between regions. Histone H4 of the forebrain is more acetylated in P47 than in P25 forebrains. Hindbrains exhibit the opposite difference with histone H4 of the P25 hindbrain being more acetylated than P47 hindbrains. These differences are mainly driven by less abundant hyperacetylated proteoforms. Transcription of histone acetyltransferases such as p300, CBP, and HAT1 are known to be higher in cortical neurons, consistent with the observed acetylation levels. Lysine 20 methylation (K20me1, K20me2, and K20me3) is notably invariant with brain region and age difference.

Graphical Abstract

Introduction

Histone proteins package DNA into nucleosomes, compacting and regulating access to eukaryotic genomes. Nucleosomes are composed of approximately 146 base pairs of DNA wrapped slightly less than twice around a core histone octamer comprised of two copies of each of the members of the H2A, H2B, H3, and H4 histone protein families (1). Histones H2A and H2B form two dimers (2xH2A & 2xH2B) while histones H3 and H4 form a tetramer (2xH4–2xH3). These two dimers and tetramer collectively form the core histone octamer. The core nucleosomes are often further complexed by linker histone H1 and packed into higher-order chromatin structures. Histone H4 is one of the most conserved sequences in eukaryotes and, in humans, is encoded by 14 genes on multiple chromosomes that encode identical amino acid sequences. Histone H4 genes are intronless and the transcribed mRNAs are not polyadenylated but regulated by a stem-loop structure. The first five amino acids of histone H4 are identical to histone H2A. Histone H4 variants are rarely seen (2). Due to their crucial role in the regulation of the genome, changes in histones can have dramatic effects on the physiology of the genome. Most prior work has focused on small absolute changes with large fold changes due to the technological limitations of antibody-based technologies. However, the absolute scale of these changes is also important and is represented in this work.

Post-translational modifications (PTMs) of histones such as acetylation, methylation, and phosphorylation are well-known to affect the binding of other factors and can have a wide range of effects. We refer to the biological state of the protein with a specific single-molecule combination of these PTMs as a “proteoform” (3,4). Histone H4 modifications occur near the N-terminus, which is constitutively co-translationally acetylated and is in close contact with the histone H3 N-terminus, allowing for trans-histone regulation. In histone H4 for example, K20me2 is required for efficient acetylation of nearby H4 lysine residues and the resulting acetylation events are also hierarchical (5). Histone H3 and H4 modifications are known to work together to license productive transcription elongation. H3S10ph begins a cascade involving FOSL1, MOF, 14-3-3, BRD4, and P-TEFb where MOF writes H4K16ac. Both H3K9ac and H4K16ac are needed for direct BRD4 binding (6).

Combinations of histone modifications are known to act in concert to cause different effects than a single PTM (6–11). As PTMs add specific masses, mass spectrometry methods are well suited. However, the only way to obtain full information about all PTMs on a given protein molecule requires no enzymatic digestion or a “top-down” method. This is challenging and requires optimized chromatography, fragmentation, and sophisticated data analysis. Rigorous, robust, sensitive, and high-throughput top-down quantitation of histone proteoforms has been an ongoing aspiration (12–16). Despite progress in capability, our in vivo knowledge of the biological dynamics, spatial distributions, function, and mechanisms of histone proteoforms remain limited. This is mostly due to a lack of methods with reproducible proteoform quantitation and challenges in the efficient and reproducible protein isolation from tissues. In prior work we established the capacity reproducibly quantitate H4 proteoforms ex vivo (17); however, this is the first demonstration in vivo as well as the first to reveal spatial and temporal differences in mammalian brain tissue. An alternative middle-down approach that uses limited digestion to preserve as much proteoform information as possible is useful when top-down analysis presents unsolved challenges, such as with the histone H3 family (18–21). However, these approaches provide incomplete proteoform information and some aspects of the biology may be lost, notably here H4K31ac. Careful consideration of the objectives of each line of inquiry is necessary to design an experiment to decipher the influence of PTMs and proteoforms in each context (17). In this study, we have developed a histone extraction method for in vivo use, particularly to isolate histones from Mus musculus brains. Once histones are isolated, we use our established top-down mass spectrometry histone H4 method to quantitate discrete PTMs and proteoform abundances of histones H4 in mouse brain regions and at different ages (17). This allows interrogation and quantification of combinations of histone modifications.

While top-down proteomic analysis reveals the true physiological state of proteins, a reductionist interpretation is often useful to decipher the mechanisms by which proteoforms function. We refer to the presence and abundance of a particular PTM independently of the other modifications on the histone molecule as “discrete” PTMs. This is the level of information obtained with typical antibody use, which is also non-quantitative. “Binary” refers to single-molecule combinations of two of these discrete PTMs, with deliberate ignorance of modifications not listed in the combination. A proteoform takes into account all possible covalent modifications on the molecule. For example, we can measure the abundance of the proteoform <Nα-acK5unK8unK12unK16acK20me1K31ac>. This refers to a histone H4 molecule that is N-terminally acetylated, acetylated on lysines 16 and 31, and monomethylated at lysine 20. For conciseness, we refer to this proteoform as <K16acK20me1K31ac>. The angled brackets denote a definitive absence of modifications at all other sites except for the histone H4 N-terminus, which is definitively acetylated. This proteoform, <K16acK20me1K31ac>, influences the discrete abundance of {K16ac}, {K20me1}, and {K31ac} and the related binary combinations, such as but not limited to {K16ac, K20me1}, {K5un, K31ac}, and {K16ac, K31ac}. To sift through all the data that is obtained from top-down analysis of histone proteoforms, we often marginalize to look at discrete or binary PTMs to inform us of the importance of these modifications, and their mutual interplay.

Our methods also provide unique quantitative insights that require the use of precise terminologies to disambiguate different metrics of abundance and change. Our proteoform data is presented as the absolute percent of the protein that is modified in that specified manner. That is, we know the percent abundance of each proteoform relative to the total H4 protein. Thus, we also know approximately how much of the genome is decorated by this modification state (save how the same proteoforms might co-occupy nucleosomes). From this, we compute metrics of abundance at the discrete and binary combination levels, also in percent of the total H4 protein in the cell. Because we start with percent abundance, we can measure a change in two different ways: fold change and absolute change. Fold change is the prevailing metric of change and is more familiar; however, it does not capture a complete picture. Insight is also gained through understanding the absolute change in abundance because this reflects how much of the genome is changing. We compute the absolute change as the arithmetic difference in the percent abundance and express this as a percentage point (pp) difference. The pp difference is also representative of the percent of the genome that has changed. For example, if the proteoform <K16acK20me2> is 30% abundant in one condition but 42% abundant in another, there is only a 0.4-fold increase (40 percent change) but this is a 12 pp absolute change (percentage point difference), which represents a change to a considerable 12% of the epigenome.

Substantial efforts have been made toward understanding the function of Mus musculus brains and by extension human brains. Recent studies have focused on mapping the transcriptome and proteome in a cell-type-specific manner (22–26). While efforts toward understanding transcriptional regulation through histone proteoforms ex vivo are informative, few quantitative approaches have been established in the brain (5,8,16,20,27,28). Histone H4 proteoforms in the brain are largely unknown as most studies focus on histone H3 variants. We expect that through quantitating histone H4 proteoforms in the brain, we will observe differences in the abundance of proteoforms in various contexts. In this study, we analyze forebrains (comprised of the cerebrum, thalamus, and hypothalamus) and hindbrains (comprised of the medulla, pons, and cerebellum) from mice at 25 and 47 days of age (P25 and P47, respectively). At P25, mice were just weaned and are beginning their adolescent phase (early postnatal). At P47, mice are approximately 2/3 of the way through adolescence and are reaching sexual maturity (29). While the relative volume of brain regions remains similar between P20-P45, the total brain volume increases (30). However, differences with development become more pronounced as one focuses on specific brain regions or specific targets of interest. For example, the hypothalamus, of the forebrain, increases in density dramatically compared to other brain regions (31). Neurological studies observe remarkable specific changes that correspond to mouse development; however, most observe the greatest differences within the first 25 days of age (29–32). Mus musculus behavior is still developing between 25 and 47 days, yet few studies have been able to pinpoint a genetic link to these changes (29). Histone post-translational modifications could yield insight into this gap in knowledge. We expect to observe significant changes in histone H4 proteoforms during development that could shed light on developmental epigenetic regulation. Additionally, the histone H4 proteoform landscape is not known in Mus musculus brains at any age. Determining relative transcriptional activity can point to ages or brain regions that require additional interrogation to determine specificity and significance. In addition to establishing a baseline by quantitating histone H4 proteoforms, interrogating histone H4 proteoforms during adolescence will allow us to observe differences in epigenetic regulation.

Materials and Methods

Mice and dissection

Wildtype CD1 mice (Crl: CD1(ICR)) were used at 25 days of age (P25) and 47 days of age (P47). Our dataset included n=4 females for each age, producing one forebrain and one hindbrain datapoint from each organism. Mice were stored at −80 °C immediately after sacrifice and then dissected after thawing. The brain was removed, washed with cold PBS, then separated into brain regions. Each mouse produced one right forebrain, left forebrain, and hindbrain that were processed separately. Left and right forebrain samples of each histone were combined for top-down LC-MS/MS analysis as they are not epigenetically different (Fig.S1).

Histone isolation

Histones were isolated as previously described with further optimization (33). After brain regions were separated, each sample was homogenized with 1 mL NIB + NP-40 and a cocktail of epigenetic inhibitors using a Dounce homogenizer. Nuclei isolation and acid extraction were performed with the changes of doubling the NIB + NP-40 incubation time to 10 minutes and centrifuge duration to 10 minutes at 20,000 x g (33). Isolated histones were resuspended in 85 μL 5% acetonitrile 0.2% trifluoroacetic acid then histone families were separated by offline high-performance liquid chromatography (HPLC) with standard lab protocol, as described in Holt et al. (33).

Histone H4 MS method

The histone H4 fractions (34–36.5 minutes) from the offline high-performance liquid chromatography (HPLC) were combined if necessary and concentrated to dryness with a vacuum centrifuge concentrator (Savant^™ SPD131 SpeedVac, Thermo Scientific, Waltham, MA). Dried fractions were diluted using μg H4 = (Peak area – 6.0114)/31.215 to calculate the dilution to 200 ng H4/ μL MS buffer A (2% acetonitrile, 0.1% formic acid). 1 μL (200 ng) histone H4 was loaded onto a 10 cm, 100 μm inner diameter C3 column (ZORBAX 300SB-C3 300 Å 5 μm) self-packed into fused silica pulled to form a nanoelectrospray emitter. Online HPLC was performed on a Thermo U3000 RSLCnano Pro-flow system. The 70-minute linear gradient using buffer A: 2% acetonitrile and 0.1% formic acid, B: 98% acetonitrile and 0.1% formic acid is described in Table 1. The column eluant was introduced into a Thermo Scientific Orbitrap Fusion Lumos by nanoelectrospray ionization. A static spray voltage of 1800 V and an ion transfer tube temperature of 320 °C were set for the source.

Table 1:

Online HPLC parameters used to quantitate histone H4 proteoforms.

Sample Volume	1 μl
Temperature	4.0 [°C]
Buffer A	2% ACN 0.1% FA
Buffer B	98% ACN 0.1% FA
Equilibration	Duration = 5.000 [min]
Flow Rate	0.200 [μl/min]
Column	10 cm × 100 μm self-packed C3
Linear Gradient Program
Time:	%B:
0.0	29.0 [%]
0.5	29.0 [%]
70.0	35.0 [%]
75.0	98.0 [%]
80.0	98.0 [%]
85.0	5.0 [%]

The orbitrap MS1 experiment used a 60k resolution setting in positive mode. An AGC target of 5.0e5 with 200 ms maximum injection time, three microscans, and a scan range of 700–1400 m/z were used. An intensity threshold of 1e5 was set for the selection of ions for fragmentation. The target precursor selected for MS2 fragmentation included all major H4 peaks and chose the top 20 most abundant m/z. ETD fragmentation at a 14 ms reaction time, 5.0e5 reagent target with 200 ms injection time was used. MS2 acquisition was performed using the orbitrap with the 60k resolution setting, an AGC target of 5.0e5, a max injection time of 200ms, a ‘normal’ scan range, and three microscans.

Histone H4 proteoform quantitation

Raw files were converted to mzXML. Data processing was performed by a custom analysis suite as previously described using a 3.4 Da cutoff for precursor mass, 10.0 ppm error for matched fragments, and maximum +10 peptide charge (TDMS 7.04, Interp 2v15) (17,34).

Data analysis

Each sample was run twice using the MS methods described above for a total of two technical replicates and averaged to give the measured value for each biological data point. Each data point has a biological n=4, apart from the P47 forebrain (n=3). A Welch’s two-tailed t-test was used for significance. The datasets generated during and/or analyzed during the current study are available in the MassIVE repository, ftp://massive.ucsd.edu/MSV000090249/.

Statement of Human and Animal Rights

All animal studies have been approved by the appropriate ethics committee and have therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. The mice used were sacrificed consistent with standard practices and donated for our use as spare material from approved experiments.

Results

Histone H4 can be efficiently isolated and analyzed from mouse brains with only minor modifications to previous methods.

Our previously described histone extraction method, which is optimized for speed without efficiency compromise, is suitable for brain tissue with only minor modifications to increase yield (33). These modifications include homogenizing the brain in NP-40 alternative detergent and centrifuging the sample during nuclei isolation for 10 minutes instead of the usual five minutes. The nuclei isolation and acid extraction steps show the largest influence on histone yield and are optimized for both time and yield. Once histones were isolated with high-performance liquid chromatography (HPLC), the histone fractions were treated normally according to established protocols and analyzed by our previously described quantitative top-down proteomic analysis (17). Representative spectra are shown in Fig.1, which shows the clean isolation of histone H4 with proteoform identification using electron transfer dissociation fragmentation. This method was used to obtain all histone H4 discrete modification, binary combination, and proteoform abundances from samples of different ages and brain regions. Important comparisons and qualities of this method are described in detail in (17). This level of data could not be obtained without the use of a quantitative, top-down mass spectrometry method.

Fig.1.

Representative spectra of our histone H4 mass spectrometry-based method show low signal: noise and clear proteoform identification. (a) MS1 average spectra from 30 to 70 minutes show clear +16 to +9 charge states matching m/z of histone H4. (b) MS1 spectrum of 11305.5153 Da species (m/z = 754.7083, charge +15) matches a mass of H4 with N-terminal acetylation + 2 methylations + 2 acetylations + 0 phosphorylations with a 6.5 ppm error (0.07 Da). (c) The ion map from MS2 fragmentation of species from (b) shows unambiguous localization information for the PTMs from a mass change at less than 10 ppm error. (d) Annotated MS2 spectra show the most abundant peaks are from the same proteoform, <K16acK20me2K31ac >, with less than 10 ppm error. Our mixture model would be applied to this spectrum and is described in detail here (34) (e) Annotated MS2 spectra from (d) showing 500 to 700 m/z. (f) Annotated MS2 spectra from (d) showing 750 to 1000 m/z.

Histone H4 Discrete Post-translational Modifications (PTMs)

Histone H4 modifications of Mus musculus brains differ greatly from common cancer cell lines.

Histone modification and histone proteoform analyses have most commonly been applied ex vivo in rapidly proliferating cancer cells (6,8,20). While these studies are valid and mechanistically insightful, the transcriptional regulation of somatic tissue is expected to differ greatly from highly proliferative human cells. These differences are apparent at even the single (discrete) PTM level, shown in Fig.2a. While some differences are observed between human breast cell lines SUM159 and MCF7, mouse brains are always strikingly different from both cell lines. K16 acetylation {K16ac} and K20 di-methylation {K20me2} are markedly lower in the mouse brain, two modifications that are known to have combinatorial effects (5,8). We do not focus here on describing or investigating the differences between ex vivo cells and in vivo tissues; however, it is important to recognize the in vivo context as distinct.

Fig.2.

Discrete PTM analysis of histone H4 from Mus musculus brain shows few significant changes in acetylation. (a) Rapidly proliferating cancer cell lines (SUM159 and MCF7) compared to an average Mus musculus brain region (P47 hindbrain) show significant differences between the discrete abundance of most PTMs. (b) Comparing epigenetic changes between the forebrain and hindbrain histone H4 in P25 mice uncover higher amounts of acetylation in the hindbrain. (c) Comparing epigenetic changes between P25 and P47 forebrains uncover a general increase in histone H4 acetylation with age with K8ac significantly increasing with age. (d) Comparing epigenetic changes between P25 and P47 hindbrains uncover a general decrease in histone H4 acetylation with age. (e) A table lists the absolute abundance (average ± standard error) of each PTM for both brain regions at each age. For all panels, the cutoff for significance is 0.05 (*) with a two-tailed t-test, and error bars represent standard error between biological replicates (n=4 for each group, except for P47 forebrains with n=3).

Histone H4 discrete PTM abundances exhibit only a few specific changes spatially and temporally.

Discrete PTM data is inherently present in proteoform data and may be obtained through marginalization, a procedure that simply adds the abundance of the proteoforms containing the PTM to be marginalized. This allows for a useful if incomplete, reductionist perspective and convenient comparison to more commonly available methods that operate at the discrete PTM level. While we know PTMs often act in concert and proteoforms are the species present in vivo, discrete PTMs are sometimes also insightful. When ignoring the presence of all other modifications to focus on a single PTM abundance, there are few differences between Mus musculus brain regions. However, we observe significant changes to discrete histone PTMs between brain regions of P25 mice and between forebrains and hindbrains of different ages, summarized in Fig.2b–e. In P25 mice brain regions, K5 and K8 acetylation is higher in the hindbrain (n=4, p=0.080 and 0.056 respectively). In mouse forebrains at P25 and P47 ages, {K8ac} significantly increases with age (n=3 and 4, p=0.007). In mouse hindbrains at P25 and P47 ages, there are no significant changes observed; however, acetylation of lysines 5, 8, 12, and 16 decreases with age. There are no significant differences observed between P47 brain regions, K20 methylation events, and K31 acetylation. It is common for discrete PTMs to be unchanged while key proteoforms or combinations of modifications are changing. No significant differences were observed between the left and right hemispheres of the forebrain, as shown in Fig.S1a, confirming that the hemispheres are epigenetically similar. While these PTM changes are modest, they are linked to more complex signals that exhibit greater differences, such as hyperacetylation, as we discuss in more detail throughout the manuscript (5).

Histone H4 Binary Combinations

P25 mice show differences in binary combinations of acetylations.

Mus musculus brains at 25 days of age are considered prepubertal and are still in development. Marginalizing to analyze changes in binary (2-PTM) combinations allows us to observe PTMs that change or function in concert. Binary combinations of discrete histone H4 PTMs were calculated by adding the abundance of proteoforms that contain both PTMs to determine the percent abundance of each two-PTM combination. They are not mutually exclusive by nature. The results are shown in Fig.3a. Significant (p<0.05, two-tailed t-test) and greater than 1.5-fold changes including the modification K31ac represent a change to approximately 12 percent of the genome or 31 percent of the total changes seen in binary combinations.

Fig.3.

Volcano plots showing histone H4 binary PTM changes (35). (a) Comparing epigenetic changes in the forebrain and hindbrain of P25 mice uncover fewer significant PTM combination differences, including K31ac. Significant changes in 2-PTM combinations account for 31 percent of the total 2-PTM changes seen between the forebrain and hindbrain of P25 mice. (b) Brain regions of P47 mice show a higher abundance of the two PTM combination {K5ac, K31ac}. (c) Comparing epigenetic changes between P25 and P47 forebrains uncover a significant increase in 2-PTM combinations including K31ac and K8ac. Significant changes in 2-PTM combinations including K31ac account for 40 percent of the total changes seen between forebrains of P25 and P47 age. Significant changes in 2-PTM combinations including K8ac account for 23 percent of the total changes seen between forebrains of P25 and P47 age. (d) Benchmark point size for ≤ 0.5 to 6 absolute percentage point (pp) difference, respectively. Absolute change below 0.5 pp is represented as 0.5 pp. (e) A table lists the absolute abundance (average ± standard error) of the top 20 (by abundance) PTM combinations for both brain regions at each age. The cutoff for significance is 0.05 with a two-tailed t-test (n=3 for P47 forebrain, n=4 for P25 forebrain and hindbrain, P47 hindbrain). Grey dotted lines show p-value = 0.05 and fold change = 1.5. Red points represent both p < 0.05 and greater than 1.5-fold change, blue points represent p < 0.05 and less than 1.5-fold change, grey points represent p > 0.05 and greater than 1.5-fold change, and black points represent p > 0.05 and less than 1.5-fold change. For red points, the absolute pp difference is represented by the radius of the point.

Binary PTM combinations of P47 mice show few but specific changes.

Mus musculus at 47 days of age are considered adolescent, and the epigenetic landscape at this age is helpful to see differences in the regulation of brain regions (forebrain and hindbrain). The results are shown in Fig.3b. When looking at binary combinations at P47, there is a significant (p=0.04) and 1-fold change and 1.2 percentage point (pp) absolute change in the combination {K5ac, K31ac} when comparing the forebrain to hindbrain.

Binary PTM combinations of mouse forebrains exhibit differences in multiple acetylations.

While multiple ongoing studies hope to decipher particular functions of the mouse forebrain brain regions, it is generally understood that these regions are similar to that of humans and control processes such as sensory processing and reasoning. With age, it is expected for the transcriptional regulation of the forebrain to change. The results are shown in Fig.3c. Significant and greater than 1.5-fold changes including the modification K31ac account for a change in the epigenetic conditions of 7.0 percent of the genome or 40 percent of the total changes in binary combinations. Significant (p<0.05) and greater than 1.5-fold changes including the modification K8ac account for a change in the epigenetic conditions of 3.9 percent of the genome or 23 percent of the total changes in binary combinations. There are no significant changes observed between the hindbrains of P25 and P47 mice and greater than a 1.5-fold change in left and right forebrains, shown in Fig.S1b.

Histone H4 Proteoforms

Histone H4 proteoforms of mouse brains show strong differences in hyperacetylated proteoforms with age and brain region.

Full proteoform analysis represents the most complete picture of the histone H4 landscape and allows for the comparison of specific histone H4 molecules. As mentioned previously, Mus musculus brains at 25 days of age are considered prepubertal and are still in development. Comparing brain regions when divided into forebrain and hindbrain, P25 mice show a significant (p=0.01), 4.9-fold change in abundance of the hyperacetylated proteoform <K5acK8acK12acK16acK20me2K31ac> accounting for a 0.8 percent change in the histone H4 landscape. Additionally, the very similar proteoform <K5acK8acK12acK16acK20me1K31ac> shows a significant (p=0.008), 5.1-fold greater abundance, and changes 0.24 percent of the histone H4 landscape. Forebrains and hindbrains of P47 mice show a combined 0.5 pp histone H4 difference with the proteoforms <K8acK12acK16acK20me2> and <K8acK12acK16acK20me2K31ac> (0.42- and 0.37-fold change, p=0.007 and p=0.04, respectively) alongside an insignificant (p=0.13) but large absolute difference of 1.7 pp, <K16acK20me3>. Forebrain comparisons show similar hyperacetylated + K20me2 proteoforms <K5acK8acK12acK16acK20me2K31ac> and <K8acK12acK16acK20me2> which account for a one pp difference in histone H4 (fold change= 4.5 and 1.9, p=0.03 and 0.056, respectively). Comparing hindbrains of P25 versus P47 mice, <K12acK16acK31ac> decreases significantly (p=0.03) with a 0.2-fold change and 0.08 pp difference, and there are fewer differences with large absolute change. The results are shown in Fig.4. Only one significant and greater than 1.5-fold change is observed between the left and right hemispheres of the forebrain: <K12acK16acK20me1K31ac>, shown in Fig.S1c. Due to distinct changes in multiply acetylated proteoforms, further marginalization based on the acetylation state could help uncover biochemical trends with age and brain region.

Fig.4.

Volcano plots show full histone H4 proteoform changes are significant with respect to age and brain region (35). (a) Comparing epigenetic changes between the forebrain and hindbrain of P25 mice uncover a significant overexpression of multiply acetylated proteoforms. Of note, some proteoforms increase approximately one percentage point (pp), a large absolute change when looking at the proteoform level. (b) Comparing epigenetic changes between the forebrain and hindbrain of P47 mice uncover a significantly lower amount of hyperacetylated proteoforms with small absolute change. (c) Comparing epigenetic changes between P25 and P47 forebrains uncover a significant increase in hyperacetylated proteoforms. (d) Comparing epigenetic changes between P25 and P47 hindbrains uncover a significant decrease in hyperacetylated proteoforms. (e) Benchmark point size for ≤ 0.1, 0.25, and 0.75 absolute pp difference (subtraction of the percent abundance between the two groups) respectively. (f) A table lists the absolute abundance (average ± standard error) of the top 20 (by abundance) proteoforms for both brain regions at each age. The cutoff for significance is 0.05 with a two-tailed t-test (n=3 for P47 forebrain, n=4 for P25 forebrain and hindbrain, P47 hindbrain). Grey dotted lines show p-value = 0.05 and fold change = 1.5. Red points represent both p < 0.05 and greater than two-fold change, blue points represent p < 0.05 and less than two-fold change, grey points represent p > 0.05 and greater than two-fold change, and black points represent p > 0.05 and less than two-fold change. For significant points, the absolute percent difference is represented by the radius of the point. Values less than 0.1 pp were represented with the 0.1 pp dot size. Points are labeled with the respective proteoform with unmodified residues removed for conciseness (e.g. <K20me1K31ac> is <Nα-acK5unK8unK12unK16unK20me1K31ac>). Points with ±5-fold change generally occur in just one condition.

Multiply acetylated proteoforms differ by age and brain region.

Marginalizing proteoforms to the number of acetyl groups on the histone H4 molecule, we can determine the degree of acetylation for each biological replicate. The relative abundance of histone H4 with no acetylation (other than constitutive N-terminal acetylation, Nα-ac) and up to 5 acetylation events were combined. When comparing the number of total acetylation events on a histone H4 molecule, we see significant increases in multiply acetylated molecules, defined as two or more acetylation events. Consistent with discrete acetylation events, we see that P25 acetylation is higher in the hindbrain, increases with age in the forebrain, and decreases with age in the hindbrain. Results are summarized in Fig.5a–c. Specifically in P25 mice, histone H4 molecules with two and five acetylation events are significantly higher (n=4, p=0.01 and 0.009 respectively) with a significant decrease in only N-terminal acetylation (n=4, p=0.02, two-tailed t-test). Forebrain comparisons show a significant increase in histone H4 molecules with five acetylation events (n=3 4, p=0.01). Hindbrain comparisons show a significant decrease in histone H4 molecules with two acetylation events (n=4, p=0.03). There is no significant change in hyperacetylation in the forebrains and hindbrains of P47 mice.

Fig.5.

Percent histone H4 of different acetylation states. (a) A comparison of the forebrain and hindbrain of P25 mice shows significantly higher histone H4 that contains 2 (p=0.01) and 5 (p=0.009) acetylation events in the hindbrain, with significantly lower N-terminal acetylated only histone H4 in the hindbrain (p=0.02). (b) A comparison of P25 and P47 forebrains show an increase of multiply acetylated (2+ acetylation events) histone H4 with age. Only histone H4 with 5 acetylation events is significantly increasing with age (p=0.01). (c) A comparison of P25 and P47 hindbrains show a decrease of multiply acetylated histone H4 with age. Histone H4 with 2 acetylation events shows a significant (p=0.03, two-tailed t-test) decrease with increasing age. (d) The weighted sum of acetylations per histone H4 molecule reflects the average number of acetylation events per biological condition. Histone H4 is significantly more acetylated in the hindbrain of P25 mice (p=0.03). (e) Percent multiply acetylated histone H4 shows the percent of histone H4 that has more than two acetylation events. P25 mice show a significantly higher amount of multiply acetylated histone H4 in the hindbrain (p=0.04). Mouse forebrains at P47 age show a significant increase in the percent of multiply acetylated histone H4 compared to P25-aged mice (p=0.02). The degree of acetylation refers to the total number of acetylation events on the histone H4 molecule (two acetylation events = 2 ac). As histone H4 is constitutively N-terminally acetylated, this is noted as appropriate and N-terminal acetylation + one acetylation event is counted as one acetylation event. Error bars show standard error for biological replicates. For line graphs, N-terminal acetylation and 1 acetylation event were removed for scale and did not significantly change. A two-tailed t-test was used for significance with n=4 for each group, except for P47 forebrains with n=3. One asterisk (*) refers to a p-value < 0.05, two asterisks (**) refer to a p-value < 0.01, and three asterisks (***) refer to a p-value < 0.001.

The weighted sum of acetylated histone H4 changes within P25 mice.

The weighted sum of acetylation, shown in Fig.5d, is a helpful tool to see broad acetylation changes in each biological context- the percent abundance of each proteoform times the number of acetylations of that proteoform (disregarding the constitutive N-terminal acetylation). A significant difference is only observed between P25 forebrains and hindbrains (n=4, p=0.03) with significantly higher acetylation in the hindbrain.

Percent multiply acetylated histone H4 changes within P25 mice and forebrain region.

Multiply acetylated histone H4 molecules, results shown in Fig.5e, have more than two acetyl events (disregarding constitutive N-terminal acetylation). Both P25 brain regions and forebrain comparisons show significant differences. Percent multiply acetylated histone H4 is significantly higher in hindbrains at P25 (n=4, p=0.04). Percent multiply acetylated histone H4 significantly increases with age in the forebrain (n=3 4, p=0.02).

Discussion

Mus musculus brain histone proteoforms are an important, albeit overlooked level of epigenetic regulation.

The brain is a vital organ that influences many aspects of whole organism biology. Mus musculus models are often used to gain information on human neurobiology and are useful tools in determining biological mechanisms in specific contexts. In-depth analyses of Mus musculus brain regions have been performed with a variety of approaches; however, few consider histone modifications and even fewer consider histone proteoforms (16,22,23,25–27). We are the first to quantitate histone H4 proteoforms in the Mus musculus brain. Regulation of gene expression and more specifically transcription is central to cellular function. Approaches such as RNA-seq reveal the output of transcriptional regulation but reveal only some aspects of the mechanisms of how the transcription of genes is differentially regulated. Histones are proteins whose modifications greatly influence the transcription of genes by controlling the packaging of DNA and thus accessibility. More importantly, they also serve as a platform for post-translational modifications (PTMs). Histone PTMs in turn recruit factors for transcription activation or repression often through modulating protein-protein interactions. Our lab has developed robust methods of quantitating histone H4 proteoforms that reveal the abundance of all combinations of modifications (17). The molecular state of histones gives clues to the regulation of transcription at different ages and brain regions of Mus musculus. This quantitative, top-down proteomic analysis of histones is the first to obtain this type of analysis in a Mus musculus brain. The further optimized method for the isolation of histone families described allows for effective probing of histone proteoform abundance in various biological states.

Proteoform-level data is essential to deciphering epigenetic regulation; however, it is not a complete picture of biological regulation.

While the quantitative, proteoform-level data we have obtained is informative of the fundamental biochemistry of neuroepigenetics, it does not localize this activity to specific genomic loci. Experiments such as chromatin immunoprecipitation-sequencing (ChIP-seq) would allow for the specificity of discrete marks to particular genes; however, it is incapable of probing for multiple acetylation states or multiple PTMs at co-occurring on single molecules. In addition, antibody-based approaches often struggle to distinguish between methylation states such as the H4 K20 methylation events we observe (7). It has been established in our prior work and the work of others, that biologically important histone modifications rarely exist in isolation and that many of the mechanisms that govern chromatin regulation of transcription function through multivalent interactions and complex interplay between sometimes distal PTMs.

Our top-down mass spectrometry method obtains high-quality, proteoform-level data.

Top-down analysis of histone H4 preserves all regulatory residues, including K31, while having quantitative power in part through identical ionization of H4 proteoforms. Enzymatic digestion with proteases (trypsin, Lys-N/C, Arg-C, Asp-N/C, Glu-C) does not keep amino acids 1–31 on the same peptide, losing crucial regulatory information. As we are keeping all 102 amino acids together, one modification does not have as great of an effect on ionization as one modification on a smaller peptide (36). Our quantitative method is possible due in part to consistent ionization (17). While histone isolation from mouse brains can be difficult with established protocols, increased washing and centrifugation time is sufficient for brain tissue (33). After analyzing histone H4 discrete (1-PTM), binary (2-PTM), and proteoforms (all PTMs) of different regions and aged mouse brains, the most important changes were at the proteoform level where we observed significant changes in multiply acetylated histone H4. These trends would not be discovered if not for our quantitative top-down method.

Histone H4 proteoform trends observed in Mus musculus brains are similar to those from cultured cells.

In Mus musculus brains, the top 60 (of 120) proteoforms by abundance account for 99 percent of total histone H4, and the top 20 proteoforms account for 90 percent of total H4 Fig.S2b. Interestingly, the percent abundance of the top half of H4 proteoforms in P25 forebrains and hindbrains is almost significant (99.1 and 96.8 percent respectively, p=0.09). For reference, the top half of histone H4 proteoforms isolated from cell culture comprise 99.94 percent of the total H4 in SUM159 and MCF7 cells, see Fig.S2a (8). The three most abundant proteoforms observed in the Mus musculus brain are <K20me2>, <K20me3>, and <K20me1> (approximately 31, 13, and 7 percent abundance respectively) and all do not significantly change between brain regions or age. The most abundant H4 proteoforms observed in cell culture are somewhat similar to those in the brain (8). In SUM159 cells, the three most abundant proteoforms are <K20me2>, <K16acK20me2>, and <K20me1> (approximately 51, 19, and 7 percent abundance respectively). In MCF7 cells, the three most abundant proteoforms are <K20me2>, <K16acK20me2>, and <K20me3> (approximately 51, 22, and 7 percent abundance respectively) (8). These trends show that histone proteoforms are both highly regulated and proteoform specific. This proteoform specificity implies that there is information in these proteoforms that is distinct and that biology operates on the single molecule level of histones (proteoforms). These minimally modified proteoforms allow for the writing of more PTMs, giving a specific signal. These minimally modified, high-abundance proteoforms could be considered poised for multiple signals through the writing of other PTMs. Additionally, abundant histone acetylations, such as H4K16ac, may also serve as reservoirs of acetate for ACSS2-mediated acetyl-CoA production (37).

Few changes are observed between P47 forebrains and hindbrains; however, many significant differences in hyperacetylated histones are uncovered between P25 forebrains and hindbrains.

Multiply acetylated histone H4 proteoforms are higher in the hindbrain than forebrain in P25 mice. There are few changes in histone H4 PTMs and proteoforms between brain regions of P47 mice. Forebrains show an increase in multiply acetylated histone H4 proteoforms between P25 and P47 ages. Hindbrains show a decrease in multiply acetylated histone H4 proteoforms between P25 and P47 ages. Generally, hyperacetylation of histones is accepted as a transcriptionally active state (38). This suggests that P25 forebrains are the most transcriptionally active with the differences between brain regions decreasing with increasing age. This is consistent with diminishing transcriptional diversity in neuron subtypes with age in Drosophila (39) and developing mouse brains (24). Morphological studies on developing mouse brains help explain these differences. There is decreasing cortical thickness from the rostral to the caudal direction and the cortex and whole brain volume significantly increases from 3 to 8 weeks of age (~P21, ~P56). The myelin content increases dramatically between 3 and 8 weeks (~P21, ~P56) (30). These findings indicate dramatic morphological development in the cortex in this time frame. RNA-seq experiments on P25 mouse brains also uncover increases in histone PTMs such as H3{K36me3} that is indicative of recent transcription (40). This is consistent with the increased levels of H4 acetylation that we observe in P25 mice, which are also indicators of active transcription (38). These findings support a model of high transcriptional activity and developmental change in the forebrain of P25 mice. The reduced acetylation in P47 is likely due to important developmental machinery being silenced as these mice reach sexual maturity.

Our model of histone H4 proteoform changes is consistent with previous findings.

While it is well known that the relative abundance of histone H3 variants accumulate with age (41,42), histone H4 also turns over at a rate equal to the sum of all H3 variants (41). It was also observed that histone H3.3 accumulation is delayed in glial cells compared to neurons where histone H3.3 becomes the most abundant H3 variant over H3.2 in non-neuronal cells (41). In neuronal cells, histone H3.3 becomes the most abundant H3 variant before 3 weeks of age (41). With lenti-miR (H3fa/b)/ H3.3 knockdown of primary cortical neurons, the gene expression of histone acetyltransferases p300, CBP, and HAT1 are higher with less H3.3 (p = 1E-6, 6E-4, 0.06; log2fold change = 0.7, 0.7, 1.2, respectively) (41). This is consistent with lower levels of acetylation in the cortex (forebrain) at P25 compared to P47 and points to p300/CBP as responsible. Combinations of acetylations of H4 are signals to particular factors. For example, bromodomain-containing proteins BRPF1 require H4K5ac and H4K8ac to bind, ATAD2 requires H4K5ac and H4K12ac to bind, and BRD2 requires H4K5ac and H4K12ac to bind, (7). BRPF1 and ATAD2 are both subunits of the MOZ complex and are typically implicated in early development (43). BRD2 enforces chromatin boundaries (44). Interestingly, BRD2 expression should decrease in the forebrain with age, extrapolating from changes observed with histone H3.3 (41). This supports p300/CBP acetylation of histone H4 that causes increased activity of BRD2 in the forebrains of P25 mice, perhaps the more important time that chromatin boundaries are enforced during adolescence.

Important expansions of this study include an expanded age range and additional histone families.

This work is an important first step in the study of histone proteoforms in brain regions for future studies to build upon. These could include an expanded dataset to look at older mice brain proteoforms at particular developmental milestones as well as studying brain regions with greater specificity (i.e. is there a particular brain region that deviates from or contributes most greatly to the broader forebrain and hindbrain categories). Proteoform-level histone data has not been obtained for many other organs and would be influential in future organ-specific studies. This study explores only histone H4, leaving histones H3, H2A, H2B, H1, and the myriad of PTMs, variants, and proteoforms thereof for future study. Further analysis of all histones and their modifications at more diverse ages, and possibly with more spatial resolution in future work will provide insights into the neuroepigenetics of the brain. Epigenetic processes are mediated by readers, writers, and erasers which change the transcription of certain genes. We consistently see feedback loops in the transcription of epigenetic genes and biologically (developmentally) important genes. The field of epigenetics hopes to fully understand this system and our findings of the epigenetic footprint through histone proteoforms help inform this broader goal. Our laboratory is working toward expanding this study to obtain a more complete view of epigenetic changes that occur with development and age, as well as the finer spatial resolution of brain regions.