A Strategy to Characterize the Global Landscape of Histone Post-Translational Modifications Within Tissues of Nonmodel Organisms

Abstract

Histone post-translational modifications (PTMs) are epigenetic marks that play a critical role in the expression and maintenance of DNA, but they remain largely uninvestigated in nonmodel organisms due to technical challenges. To begin alleviating this issue, we developed a workflow for histone PTM analysis in Mozambique tilapia (Oreochromis mossambicus), being a widespread and environmentally hardy fish, using mass spectrometry methods. By incorporating multiple protein digestion methods into the preparation of each sample, we reliably quantified 214 biologically relevant histone PTMs. All of these histone PTMs, collectively referred to as the global histone PTM landscape, were characterized in the gills, kidney, and testes of this fish. By comparing the global histone PTM landscape between the three tissues, we found that 91.59% of histone PTMs were tissue-dependent. The workflow and tools for histone PTM analysis described in this study are now publicly available and enable comprehensive investigation into the influence of environmental stress on histone PTMs in nonmodel organisms. Given the functionality and flexibility of histone PTMs, we anticipate that the study of histone PTMs in ecologically relevant contexts will provide ground-breaking insights into comparative physiology and evolution.

Introduction

In the nucleus of eukaryotic cells, DNA wraps around histone proteins that are each decorated with a variety of post-translational modifications (PTMs). Histone PTMs are epigenetic marks that modify the physicochemical properties of chromatin and thereby lead to alterations in the expression, replication, mutagenesis, and repair of DNA.^1,2 Due to such critical functions, histone PTMs have become a focus of human disease research. Mass spectrometry emerged as a powerful tool to study histone PTMs,^3,4 and this technological advance led to discoveries implicating histone PTM dysregulation in cancer and Alzheimer’s disease, among other pathologies.^5,6 Most knowledge of histone PTMs now comes from human disease research;⁷ however, histone PTMs hold the same important functions in plants and animals, and they are influenced by many of the environmental pressures that threaten natural populations of wild species.^8,9 For this reason, there is a pressing need to investigate histone PTMs in nonmodel organisms and in ecological contexts.^10,11

In this study, we first sought to optimize a workflow for histone PTM analysis in Mozambique tilapia (O. mossambicus) tissues. Mozambique tilapia were chosen because they are a eurytopic species of fish, capable of tolerating wide ranges of multiple environmental parameters, including salinity. This presents an opportunity to investigate how various environmental pressures mediate fish physiology through histone PTMs. Currently, there is a very limited knowledge-base of histone PTMs in fishes,¹²⁻¹⁵ let alone Mozambique tilapia. To analyze the most inclusive list of biologically relevant histone PTMs in the most reliable manner, we tailor the mass spectrometry method of quantitative-data-independent acquisition (DIA). To this extent, we aimed to create tilapia-specific DIA assay libraries to serve as lists of all modified and unmodified histone peptides targeted for quantification in every biological sample. Values of the histone peptide abundance would then serve as the basis of histone PTM quantification.

Next, we sought to characterize the presence and relative abundance of all biologically relevant histone PTMs, collectively termed the global histone PTM landscape. We performed this task separately for the gills, kidney, and testes of Mozambique tilapia to determine how cell differentiation into specific tissues influences histone PTMs. The gills and kidney were chosen as environmentally responsive tissues, where histone PTMs may play a role in their malleable physiology. The gills manage homeostasis through the transport of ions such as NaCl, acids and bases, nitrogenous waste, and dissolved gases.¹⁶ Similarly, kidneys respond to environmental changes by regulating urine production, divalent ion transport, and glomeruli function.¹⁷ In this study, testes were characterized because they represent the germ line, where histone PTMs can be passed from one generation to the next.⁸

Experimental Procedures

Experimental Design and Statistical Rationale

Twenty-four Mozambique tilapia with an average weight of 60.64 ± 3.35 g were sacrificed and dissected for their gills, kidneys, and, if male, testes. These dissections yielded n = 24 samples of gills, n = 24 samples of kidney, and n = 18 samples of testes. Each tissue sample was divided into three aliquots so that they could be processed for histone PTM analysis by using three different methods. Four kidney samples were excluded from statistical analyses due to the inadequate recovery of tissue during dissection, which led to an insufficient protein concentration once samples were divided three ways. Therefore, n = 20 kidney samples were used in subsequent statistical analyses. The use of these animals was approved by the UC Davis IACUC under Protocol #21846.

Mechanical Single Cell Suspension

Following dissection, each tissue was broken up into a suspension of single cells to increase the accessibility of the cells to reagents used in the downstream protocols. The methods used for this single cell suspension were based on those reported by Leelatian et al. (2017),¹⁸ with the notable exception of using a mechanical separation of tissues rather than an enzymatic separation.¹⁹ Therefore, freshly dissected tissue samples were submerged in phosphate buffered saline (PBS) without calcium and magnesium (to aid cell dissociation) in a 50 mL conical tube and then centrifuged at 100g for 5 min at room temperature. The supernatant containing dead cells was discarded, and prewarmed (27 °C) Leibovitz’s L-15 medium (Thermo Fisher Scientific, catalog no. 11415064) containing 5% fetal bovine serum and 1% penicillin-streptomycin was added to each sample to cover the remaining tissue pellet. This complete medium was chosen for its compatibility with fish cells.²⁰ Each sample containing tissue and medium was transferred into a 10 cm Petri dish, and tissues were minced into small (1 mm³) pieces by using a razor blade. Each sample was transferred back into a 50 mL conical tube and centrifuged at 100g for 5 min at room temperature. The supernatant was again discarded, and each tissue pellet was resuspended in 5 mL of warmed complete medium.

To mechanically separate tissues into single cells, each sample was pushed with a glass pestle through a cell dissociation sieve containing 40 mesh (Sigma-Aldrich, catalog no. S0770), which has 380 μm pores, into a new 50 mL conical tube. The resulting solution was pushed through a second cell dissociation sieve containing 200 mesh (Sigma-Aldrich, catalog no. S4145), which has 73.7 μm pores, into another new 50 mL conical tube. Samples were centrifuged at 100g for 10 min at room temperature. The supernatant was removed, and then the pellet of cells was resuspended in 5 mL of PBS containing calcium and magnesium. Subsequently, samples were centrifuged at 300g for 10 min, and then the supernatant was again discarded. Samples were flash frozen in liquid nitrogen and then stored at −80 °C until ready for processing through histone acid extraction.

Histone Acid Extraction

Histone acid extraction was used to isolate histone proteins from cells. Continuing from the mechanical single cell suspension, each cell pellet was directly resuspended in 400 μL of 0.4 N H₂SO₄ and incubated with constant rotation at 4 °C for 4 h.²¹⁻²³ Following acid incubation, samples were centrifuged twice at 16,000g for 10 min at 4 °C, and each time, the supernatant containing histones was transferred into a fresh 1.5 mL tube.^21,22 Next, 100% trichloroacetic acid (TCA) was added to each sample in a drop-by-drop manner at a volume equal to 1/4 the estimated sample volume.²¹ Samples were inverted multiple times for mixing and then incubated overnight at 4 °C for the precipitation of histone proteins.²² Histone proteins were collected by centrifuging samples at 3,400g for 5 min at 4 °C. The supernatant was carefully removed and discarded. The histone pellet was rinsed with ice-cold acetone with 0.1% hydrochloric acid and then centrifuged at 3,400g for 2 min at 4 °C. The supernatant was again discarded, and then the remaining histone pellet was rinsed with 100% acetone. Samples were centrifuged again at 3,400g for 2 min at 4 °C, and then the supernatant was discarded.²² Each histone pellet was left to air-dry in a fume hood for 5 min. Samples were then stored at −80 °C until they were ready for use in the in-solution digestion of histone proteins into peptides.

Derivatization and In-Solution Digestion of Proteins into Peptides

Histone proteins were digested into peptides for analysis via liquid chromatography–mass spectrometry (LCMS). This was done according to previously described procedures,^22,24,25 with some modifications. To begin the process, the histone pellet of each sample was brought to room temperature and resuspended in 8 M urea. Next, dithiothreitol (DTT) was added to equal a final concentration of 10 mM DTT. Each sample was briefly vortexed and centrifuged and then left to incubate at 37 °C for 30 min so that proteins would become denatured and reduced. Following this incubation, iodoacetamide (IAA) was added to each sample to equal a final concentration of 30 mM. Samples were again briefly vortexed and centrifuged and then left to incubate in the dark for 30 min at room temperature so that the IAA would alkylate reduced cysteine residues and therefore prevent protein refolding. After this incubation, samples were briefly vortexed and centrifuged again to help further dissolve the proteins. A bicinchoninic acid (BCA) assay (Thermo Scientific, cat. no. 23250) was then performed to determine the protein concentration within each sample. Based on the results, three aliquots of 50 μg of protein were aliquoted from each sample for downstream sample preparation. One aliquot from each sample underwent derivatization followed by digestion with the protease trypsin (Promega, cat. no. V9013) for bottom-up proteomics. The other two aliquots of histones underwent digestion with the protease V8 (Thermo Scientific, catalog no. 20151) for middle-down proteomics. By employing these techniques, we anticipated that a maximal number of histone PTMs would be detected in each tissue.

The bottom-up approach to proteomics relies on creating short peptides (<3 kDa) that can be analyzed through LCMS.³ Trypsin is most commonly used as a protease for this purpose and typically cleaves proteins at the carboxyl end of both lysine and arginine residues. On its own, trypsin is unsuitable for histone PTM analysis because histone proteins contain a high percentage of lysine residues. The peptides generated would therefore be much too short to separate properly along an HPLC column. This is especially problematic because lysine residues are the sites of many important PTMs. To correct for this, we followed a protocol to chemically derive all lysine residues so that any lysine residue that is naturally unmodified or only monomethylated would be propionylated.²² This blocks trypsin from cleaving at lysine so that it cleaves only after arginine. This process produces more appropriately sized histone peptides and therefore enables the analysis of a higher coverage of histones, including the important PTM-containing lysine residues.

Analyzing moderately sized peptides (5–20 kDa) through LCMS is described as the middle-down approach to proteomics.³ Each sample was digested using protease V8 (also known as Glu-C) for this purpose. The cleavage specificity of V8 protease depends on the buffer used. In this workflow, two V8 digestions were performed on every sample: once when V8 was in the buffer ammonium bicarbonate and once when V8 was in the buffer sodium phosphate. V8 in ammonium bicarbonate cleaves proteins at the carboxyl end of glutamate residues. In contrast, V8 in sodium phosphate cleaves proteins at the carboxyl end of both the glutamate and aspartate residues. Following the reduction and alkylation of histones, as described above, samples were diluted in the chosen V8 buffer with a dilution factor of 5.6. V8 was added to samples of histone proteins in a 1:50 ratio. Samples were incubated with the protease for exactly 20 h at 35 °C on a rotator. Following the incubation, samples were centrifuged first at 500g for 2 min and then at 19,000g for 5 min. Each time, the supernatant containing peptides was transferred to a new 1.5 mL tube. Samples were concentrated using a SpeedVac (Thermo Savant, model no. ISS110) until all of the buffer had just evaporated from the tube. Finally, peptides were reconstituted in LCMS-grade water containing 0.1% formic acid to equal a final peptide concentration of 333 ng/μL.

LCMS Analysis

One μL of each sample was injected with a nanoAcquity sample manager (Waters, Milford, MA) and analyzed by online LCMS. A Symmetry trap column (Waters 186003514) was used to remove any remaining contaminants (urea, salt) by flushing with 3% acetonitrile in LCMS water for 1 min at 15 μL/min before loading and separation of peptides on a 1.7 μm particle size BEH C18 separating column (250 mm × 75 μm, Waters 186003545) using reversed phase liquid chromatography. Separation was achieved by eluting peptides from this column with a nanoAcquity binary solvent manager (Waters) over a 125 min linear gradient ranging from 3% to 33% acetonitrile in LCMS water. Upon elution, peptides were forced through a pico-emitter tip (New Objective FS360-20-10-D-20, Woburn, MA), ionized by electrospray, and introduced into a UHR-qTOF mass spectrometer (Impact II, Bruker). CID fragmentation was used for generation of MSMS ions and, therefore, beta-elimination of phosphate groups resulting in loss of H₂O at the PTM site was included as an indicator of phosphorylation on Ser and Thr. This information has been included in section 2.6. Batch processing of samples was controlled with Hystar 4.1 (Bruker), and a quality control standard (34 fmol of BSA peptide mix) was used weekly to monitor instrument performance.

Construction of Histone DIA Assay Libraries and DIA by LCMS

Samples were first acquired by LCMS using a conventional data-dependent acquisition (DDA) approach as previously described.²⁶ After generating peak lists from DDA raw data (DataAnalysis 4.4, Bruker Daltonics), we used PEAKS Suite X Plus (Bioinformatics Solutions Inc., Waterloo, Canada) and MSFragger²⁷ to annotate histone peptides with amino acid sequence and protein ID. The Oreochromis niloticus reference proteome database was used for annotation. This database contained 61,681 proteins and was downloaded from NCBI RefSeq on February 25, 2020. The same number (61,681) of randomly scrambled decoys and 282 common contaminants (e.g., human keratins and porcine trypsin) were included in this database. A first-round database search allowed for cysteine carbamidomethylation, methionine oxidation, and protein N-terminal acetylation. A second-round database search then allowed for a maximum of three PTMs per peptide including the following: acetylation, lactylation/carboxyethylation, and biotinylation on lysine residues; mono-, di-, and trimethylation on lysine and arginine residues; oxidation and dioxidation on proline residues; ubiquitylation and ADP-ribosylation on lysine, threonine, and serine residues; citrullination on arginine residues; and phosphorylation on serine, threonine, and tyrosine residues. Mass tolerance limits were set at 10 ppm for precursors and 0.03 Da for fragment ions in PEAKS. In MSFragger, mass tolerance limits were set at 20 ppm for precursors and 0.05 Da for fragment ions.

For each of the three sets of tilapia histone peptides produced using the three different digestion methods, a spectral library was generated. Each spectral library was created using the peptide-to-spectrum matches and protein annotations generated from the PEAKS and MSFragger DDA data, which were imported into Skyline 20.0²⁸ to construct a nonredundant raw library of MS2 spectra. The raw spectral libraries were then filtered with EncyclopeDIA^29,30 to remove interferences and retain only peptides and transitions that are suitable for DIA quantitation. The resulting filtered spectral libraries were imported into Skyline and used to generate three separate (protease-specific) assay libraries of tilapia histone peptides for DIA quantitation. Transitions were excluded from the DIA assay libraries if they were ambiguous for sharing the same combination of precursor mass-to-charge ratio (m/z), product m/z, and ion type for multiple peptides. This assay represents a tier two assay.³¹

In addition to DDA, each sample was analyzed by a second LCMS acquisition in DIA mode. LC separation parameters and conditions were identical to those used for DDA, but only MS2 spectra were acquired. The mass range for DIA was set to 390–1015 m/z at 25 Hz scan rate with an isolation width of 10 m/z (0.5 m/z overlap, 2.5 s scan interval). Quantitative analyses and visualization of DIA data were performed using Skyline 20.0. Before being filtered for ambiguity, at least four (generally six) transition peaks were detected for each peptide. Their q values were scored using the mProphet algorithm integrated into Skyline (q < 0.01). The mass error threshold for all transitions was set at 20 ppm, and the resolving power was 30,000. The same number of Skyline generated decoy peptides as the number of histone target peptides in the DIA assay list was used in mProphet q-value calculation. Abundance data for each peptide in the assay library represent the sum of peak areas of all transitions belonging to the corresponding peptide. They were normalized with Skyline against the overall median for each sample (normalized area) and then exported in csv format. These values of peptide abundance were used to calculate values histone PTM abundance, as described in the following section. Specific histone PTMs were chosen for quantification if they met both the criteria for being biologically relevant according to their Unimod accession and amino acid residue (Supplemental Table 1, e.g., eliminating methionine oxidation and cysteine carbamidomethylation) and displayed a PTM AScore of at least 10. This resulted in 214 histone PTMs that were of sufficient quality and biological relevance for quantification.

Strategy for Histone PTM Quantification

For each histone PTM chosen for quantification, three values were obtained: (1) relative abundance, (2) beta-value, and (3) M-value. These three values were all calculated using data of histone peptide abundance, called the normalized area, obtained through Skyline. The relative abundance of a histone PTM describes the percent of histones in a sample where the specific amino acid residue on a histone is occupied with the PTM. Because each tissue sample was processed using three different digestion conditions (trypsin, V8 in the buffer sodium phosphate, and V8 in the buffer ammonium bicarbonate), values of relative abundance had to be calculated in two stages, presented in eq 1 and eq 2. As shown in eq 1, the relative abundance based on histone peptides from only one digestion condition (α_d), was calculated as the sum of the normalized area for all histone peptides containing the specific modification (m_x), divided by the sum of the normalized area for all histone peptides containing the specific histone amino acid residue (r_x), multiplied by 100. The value of α_d was calculated for each histone PTM three times and once for each digestion condition. However, not all digestion conditions could detect the histone amino acid residue where a PTM was reported. Therefore, whenever r_x = 0 using one digestion condition, the value of α_d was excluded.

1

The second stage for calculating the relative abundance of a histone PTM takes into account all of the digestion conditions. As shown in eq 2, the relative abundance of a histone PTM (α) is calculated as the sum of all relative abundance values based on each individual digestion condition (α_d,_i), divided by the number of α_d values calculated for the given histone PTM (n).

2

The beta-value of a histone PTM is similar to the relative abundance; however, it is displayed as a number between 0 and 1 rather than a percentage, and a value of 100 is added to the denominator so that the beta-values of histone PTMs located on low-intensity peptides are standardized.^32,33 As with the relative abundance calculation, beta-values must be calculated in two stages, presented in eq 3 and eq 4, to account for multiple digestion conditions used to process each sample. Therefore, as shown in eq 3, the beta-value based on histone peptides from only one digestion condition (β_d) is calculated as the sum of the normalized area for all histone peptides containing the specific modification (m_x), divided by the combination of 100 and the sum of the normalized area for all histone peptides containing the specific histone amino acid residue (r_x). Values of β_d were excluded whenever r_x = 0 using one digestion condition.

3

The second stage for calculating the beta-value of a histone PTM, which accounts for all digestion conditions, is presented in eq 4. The beta-value of a histone PTM (β) was calculated as the sum of the beta-values based on each individual digestion condition (β_d,i) divided by the number of β_d values calculated for the given histone PTM (n).

4

All statistical tests in this study were based on the M-value of histone PTMs. The M-value is a logit transformation of the beta-value (eq 5).³² The reason that M-values were used for statistical purposes rather than relative abundance or beta-values is that the latter values do not conform to a normal distribution: values of relative abundance are bound by 0 and 100%, and beta-values are bound by 0 and 1. The M-value rectifies this issue, but the values themselves are unintuitive. Therefore, we calculated the relative abundance, beta-value, and M-value for each histone PTM in each sample in order to perform statistical tests properly by using the M-value and to report the relative abundance alongside statistical tests for interpretability.

5

The excel workbook we prepared for quickly converting values of histone peptide abundance into the relative abundance, beta-value, and M-value of each histone PTM is available in Supporting Information, Supplemental File 1, along with detailed instructions for its use in Supplemental File 2.

Statistical Analyses

The global histone PTM landscape was compared among the gills, kidney, and testes of Mozambique tilapia to elucidate the influence of tissue type on histone PTMs. First, principal component analysis (PCA) was performed on the tissue samples using the M-values of all histone PTMs as variables. Second, three pairwise comparisons were made between the tissues: (1) gills and kidney, (2) gills and testes, and (3) kidney and testes. For each of these tissue comparisons, t tests were performed using the M-value of each histone PTM. The Benjamini-Hochberg correction was applied to all resulting p-values in order to account for multiple hypothesis testing.³⁴ Alongside this analysis, the mean relative abundance was reported for each histone PTM in each tissue. These values were used to calculate the log₂ fold change of histone PTMs between tissues for each pairwise comparison. In the R programming environment (version 4.2.0),³⁵ volcano plots were produced to visually depict these results. Additionally, histone PTM maps were constructed, containing information about the mean relative abundance of histone PTMs in each tissue. These plots were made using the R packages ggplot2,³⁶tidyverse,³⁷ggrepel,³⁸ and cowplot.³⁹

Results

Reliable Quantification of 214 Biologically Relevant Histone PTMs

Histone peptides were uniformly targeted for quantification within each sample, regardless of any factor that could have affected peptide abundances (e.g., tissue type) through the application of DIA assay libraries. The three DIA assay libraries constructed for this purpose each comprised a unique set of histone peptides due to the method used to digest histone proteins. In the DIA assay library specific for histone proteins digested with V8 in ammonium bicarbonate, a total of 24 proteins, 1,434 peptides, 1,434 precursors, and 10,109 transitions were included for analysis. The DIA assay library built for the histone proteins digested with V8 in sodium phosphate consisted of 20 proteins, 1,401 peptides, 1,405 precursors, and 10,238 transitions. Lastly, the DIA assay library specific for the histone proteins digested using the modified trypsin method targeted a total of 15 proteins, 669 peptides, 672 precursors, and 5,134 transitions. Altogether, the DIA assay libraries captured 25 histone proteins based on their unique protein accession number. The sequence coverage for each of these proteins is provided in Supplemental Table 2. Due to some redundancy in the protein names associated with protein accession numbers, there were only 17 unique protein names identified in our data. We use the protein name (e.g., H1 isoform X1) rather than the protein accession number to describe the histone PTMs in the following sections.

Using the histone peptides from all three DIA assay libraries, a list of histone PTMs was compiled according to the histone protein name, amino acid residue/position, and modification type. Once this list of histone PTMs was filtered for biological relevance, we found that 12 distinct types of modifications remained. These included the highly studied modifications of acetylation, methylation, dimethylation, phosphorylation, and ubiquitylation. It also included less studied modifications of oxidation, dioxidation, amidation, deamidation, biotinylation, and 4-hydroxynonenalation. Additionally, it included lactylation and/or carboxyethylation, which cannot be distinguished from each other in our data set given their identical chemical formula and therefore mass shift (+72.021129) in mass spectrometry data.

Overall, our workflow for histone PTM analysis enabled the reliable quantification of 214 unique and biologically relevant histone PTMs in Mozambique tilapia tissues. The use of three digestion conditions in the in-solution digestion protocol was proven to be especially useful in increasing the number of observed histone PTMs. Individually, the amount of histone PTMs detected was 34 using the trypsin digestion method, 138 using V8 in sodium phosphate, and 137 using V8 in ammonium bicarbonate (Figure 1). The combination of all three digestion conditions not only maximized the total number of histone PTMs; it also increased the diversity of modifications observed.

Figure 1.

Histone PTMs detected using each digestion condition. Samples of histone proteins were digested into peptides using three methods in parallel. Each digestion method produced a distinct set of histone peptides that, when analyzed through LCMS, captured different histone PTMs for quantification. Shown are the number of unique histone PTMs, from each modification type, that can be detected using the three digestion conditions. Abbreviations of Ambic and Napho are used for ammonium bicarbonate and sodium phosphate, respectively.

The methods we developed for histone PTM quantification relied on the stoichiometry of modified to unmodified versions of histone peptides (Figure 2). Therefore, the calculations of relative abundance, beta-value, and M-value for each histone PTM depended on proportions within a given sample rather than overall intensity. This prevented small discrepancies in the overall peptide concentration of samples from interfering with the integrity of the analysis, as all samples were internally standardized.

Figure 2.

Strategy for histone PTM quantification. The strategy used for histone PTM quantification is exemplified with histone H3 arginine 83 methylation (H3R83me1), for which the average relative abundance is displayed for the gills, kidney, and testes (A). The relative abundance of H3R83me1 was calculated by dividing the sum of the normalized area of seven histone peptides that contained the modification of interest (i.e., methylation) by the sum of the normalized area of 86 histone peptides that contained the residue of interest (arginine 83) and then multiplying by 100, as described in eq 1. Panels B–C represent one of the modified peptides that contributed to H3R83me1 quantification, being EIAQDFK[+56]TDLR[+14]. The mass shift of +14 represents methylation at H3R83, and the mass shift of +56 represents propionylation at H3K79. For this modified peptide, library spectrum (B) and an example peak containing the transitions of the most abundant MSMS ions in the library spectrum at distinctive retention times (C) from Skyline are shown. Panels D–E correspond to one of the unmodified peptides used in this calculation, EIAQDFK[+56]TDLR, which has a distinctive library spectrum (D) and an example peak (E).

It is best practice to filter histone PTMs for the PTM AScore after Skyline integration rather than before Skyline integration. The reason is that all peptides included in the DIA assay libraries in Skyline were filtered for representing high-quality unambiguous transitions. What is important in quantifying a biologically relevant histone PTM with a sufficient PTM AScore is how often the amino acid residue was found to contain the modification of interest compared to how often the amino acid residue was found to not contain this modification (see eq 1 and eq 3). Therefore, it does not matter whether high quality peptides in the DIA assay libraries contain chemical artifacts at irrelevant amino acid residues that result from sample processing. Similarly, it does not matter whether these peptides contain modifications at irrelevant amino acid residues that have PTM AScores lower than the set threshold of 10. By including all high-quality peptides in the quantification of the 214 biologically relevant histone PTMs with a PTM AScore of at least 10, regardless of the quality of other PTMs on the peptides, we can be more confident in the accuracy of our quantification for the relative abundance and M-value of histone PTMs due to decreased bias.

In the following sections, we describe histone PTMs in a slightly unconventional way so that additional information can be conveyed for previously undescribed histone PTMs. The name of each histone PTM (e.g., H2A.Z isoform X2 K4 methylation) begins with (1) the full description of the histone protein (e.g., H2A.Z isoform X2), (2) the amino acid residue and sequence position (e.g., K4), and (3) the type of modification observed (e.g., methylation). Amino acid residues are abbreviated using their one letter code; for example, lysine is K, arginine is R, serine is S, threonine is T, tyrosine is Y, and proline is P. For consistency with other studies, we removed the initial methionine (M) before determining the sequence position of subsequent amino acid residues in histone protein sequences. A complete presentation of the global histone PTM landscapes of Mozambique tilapia tissues is provided in Supplemental Table 3. This table includes the average relative abundance of each histone PTM in each tissue. Additionally, it provides the results of three pairwise tissue comparisons for every histone PTM in the form of the p-value, the Benjamini-Hochberg adjusted p-value, and the log₂ fold change of relative abundance.

Characterization of Histone H1

Within each tissue sample from Mozambique tilapia, we analyzed five isoforms of histone H1 proteins, named H1.0-B, H1.10, H1 isoform X1, H1-like, and protamine-like protein. Between these five proteins, 11 types of modifications were detected. This is in contrast to the 12 types of modifications detected across all histone proteins, as described above. The only type of modification found to be absent was dimethylation. In total, 53 biologically relevant histone PTMs were identified and quantified on the histone H1 isoforms. Those located on a representative protein, histone H1-like, are depicted as a histone PTM map (Figure 3A).

Figure 3.

Characterization of the histone PTMs on histone H1. (A) All histone PTMs were mapped on a representative protein, histone H1-like (XP_003459589.1). Each panel represents a different amino acid position, where at least one histone PTM was detected on this protein. The X-axis displays the three tissues characterized in this study: gills (G), kidneys (K), and testes (T). (B–D) Volcano plots depict the differences in PTMs on histone H1 proteins between tissues. These histone PTMs were plotted based on their adjusted p-value and fold change when comparisons were made between gill and kidney samples (B), gill and testes samples (C), and kidney and testes samples (D). Histone PTMs were colored according to their significance in terms of adjusted p-value (blue), fold change (green), both adjusted p-value and fold change (red), or neither (gray). (E) The most highly tissue-dependent PTMs on histone H1 proteins were determined for either their low adjusted p-value or high fold change in the pairwise tissue comparisons. Error bars represent the mean ± the standard error of the mean.

The PTMs on histone H1 proteins were mostly found in low abundance, which we hereby distinguish as having an average relative abundance of less than 50% in each tissue. Only two histone PTMs had an average relative abundance that surpassed this amount, and they both displayed an average relative abundance of exactly 100% in all tissues because the only peptides that contributed to their quantification contained the modification. These histone PTMs were (1) H1-like S1 acetylation and (2) protamine-like protein S1 acetylation.

The gills, kidneys, and testes displayed strong differences in the abundance of PTMs on histone H1 isoforms (Figure 3B–D). Between the three tissues, the gills and kidney were the most similar. Of the 53 histone PTMs quantified, 27 histone PTMs (50.94%) were significantly different between these tissues. The testes were equally dissimilar to the gills and the kidneys. When the 53 histone PTMs were compared between the testes and either of the other tissues, 37 (69.81%) were found to be significantly different. When accounting for all three tissue comparisons, we determined that 45 of 53 histone PTMs (84.91%) on histone H1 isoforms were tissue-dependent.

A list of the most highly tissue-dependent PTMs on histone H1 isoforms was prepared to include the histone PTMs displaying the three lowest adjusted p-values or three highest values of log₂ fold change when comparisons were made between the gills, kidney, and testes (Figure 3E). Only four distinct histone PTMs appeared on this list due to their especially low adjusted p-value and high log₂ fold change when compared between multiple tissues. The instances where histone PTMs generated the three lowest adjusted p-values when compared between any of the tissues were the following: (1) H1-like K44 4-hydroxynonenalation (with an adjusted p-value of 3.34e-28 when compared between the gills and testes), (2) H1-like K44 4-hydroxynonenalation (with an adjusted p-value of 7.08 × 10^–16 when compared between the gills and kidney), and (3) H1-like K44 ubiquitylation (with an adjusted p-value of 1.45 × 10^–15 when compared between the gills and testes). The three instances where histone PTMs had the highest values of log₂ fold change were (1) H1-like K44 4-hydroxynonenalation (with a log₂ fold change of 4.74 between the gills and testes), (2) H1-like K38 ubiquitylation (with a log₂ fold change of 3.75 between the gills and testes), and (3) H1.10 K113 oxidation (with a log₂ fold change of 3.66 between the kidney and testes).

Characterization of Histone H2A

A total of 50 biologically relevant PTMs on histone H2A isoforms were targeted for quantification within our DIA assay libraries. These histone PTMs were found across three proteins: (1) H2A, (2) H2A.Z isoform X1, and (3) H2A.Z isoform X2. On these proteins, 11 types of modifications were observed. Compared to the complete list of 12 modification types across all histone proteins, the modification list on histone H2A isoforms excluded only 4-hydroxynonelation.

Although histone H1 proteins contained a few more biologically relevant histone PTMs than did histone H2A proteins, histone H2A proteins contained more than twice as many highly abundant PTMs. Two histone PTMs were found to have a relative abundance of 100% in every tissue, indicating that only modified versions of the amino acid residue were included in histone PTM quantification. These PTMs were H2A R35 dimethylation and H2A.Z isoform X2 K4 methylation. An additional three histone PTMs had a relative abundance between 50% and 100% in all three tissues. These histone PTMs were (1) H2A.Z isoform X1 T98 phosphorylation, (2) H2A R20 methylation, and (3) H2A Y57 phosphorylation. Many of these PTMs are portrayed in the representative histone PTM map for histone H2A (Figure 4A).

Figure 4.

Characterization of the histone PTMs on histone H2A. (A) All histone PTMs were mapped on a representative protein, histone H2A (XP_003448939.1). Each panel represents a different amino acid position where at least one histone PTM was detected on this protein. The x-axis displays the three tissues characterized in this study: gills (G), kidney (K), and testes (T). (B–D) Volcano plots depict the differences in PTMs on histone H2A proteins between tissues. These histone PTMs were plotted based on their adjusted p-value and fold change when comparisons were made between gill and kidney samples (B), gill and testes samples (C), and kidney and testes samples (D). Histone PTMs were colored according to their significance in terms of adjusted p-value (blue), fold change (green), both adjusted p-value and fold change (red), or neither (gray). (E) The most highly tissue-dependent PTMs on histone H2A proteins were determined for either their low adjusted p-value or high fold change in the pairwise tissue comparisons. Error bars represent the mean ± the standard error of the mean.

When the PTMs on histone H2A proteins were compared between tissues, 46 of the 50 histone PTMs (92.00%) were found to be tissue-dependent. These PTMs had an adjusted p-value less than 0.05 in at least one of the pairwise tissue comparisons. The gills and kidney were the most similar tissues, with 23 of 50 histone PTMs (46.00%) being significantly different. The kidney and testes were the two most dissimilar tissues, with 40 of 50 histone PTMs (80.00%) being significantly different between the two tissues. Between the gills and testes, 33 of 50 histone PTMs (66.00%) were significantly different.

The most highly tissue-dependent PTMs on histone H2A proteins were identified according to their low adjusted p-value and high log₂ fold change (Figure 4E). The three histone PTMs with the highest log₂ fold change were (1) H2A R77 deamidation (with a log₂ fold change of 5.11 when compared between the gills and kidney), (2) H2A R42 deamidation (with a log₂ fold change of 4.02 when compared between the kidney and testes), and (3) H2A.Z isoform X1 K110 biotinylation (with a log₂ fold change of −3.93 when compared between the kidney and testes). The three histone PTMs with the lowest adjusted p-value were (1) H2A P48 dioxidation (with an adjusted p-value of 2.87e-14 when compared between the gills and testes), (2) H2A K95 lactylation/carboxyethylation (with an adjusted p-value of 8.70e-14 when compared between the kidney and testes), and (3) H2A K127 amidation (with an adjusted p-value of 3.57e-13 when compared between the gills and testes).

Characterization of Histone H2B

We identified 74 unique and biologically relevant PTMs across two isoforms of histone H2B, H2B.L4, and H2B 1/2. A full histone PTM map of H2B 1/2 is provided in order to exemplify these PTMs (Figure 5A). Between the two proteins analyzed, the following ten types of modifications were present: acetylation, biotinylation, deamidation, dimethylation, dioxidation, lactylation/carboxyethylation, methylation, oxidation, phosphorylation, and ubiquitylation. All of the PTMs on histone H2B proteins were found at low levels. When the histone PTMs were compared between the gills, kidney, and testes of Mozambique tilapia, it was found that 69 of 74 histone PTMs (93.24%) were tissue-dependent. In this case, the gills and kidney were the most dissimilar in terms of their histone PTMs, and the gills and testes were the most similar. Of the 74 biologically relevant histone PTMs analyzed on histone H2B isoforms, 59 (79.73%) were significantly different between the gills and kidney, 57 (77.03%) were significantly different between the kidney and testes, and 48 (64.86%) were significantly different between the gills and testes (Figure 5B–D).

Figure 5.

Characterization of the histone PTMs on histone H2B. (A) All histone PTMs were mapped on a representative protein, histone H2B 1/2 (XP_003451196.1). Each panel represents a different amino acid position where at least one histone PTM was detected on this protein. The x-axis displays the three tissues characterized in this study: gills (G), kidney (K), and testes (T). (B–D) Volcano plots depict the differences in PTMs on histone H2B proteins between tissues. These histone PTMs were plotted based on their adjusted p-value and fold change when comparisons were made between gill and kidney samples (B), gill and testes samples (C), and kidney and testes samples (D). Histone PTMs were colored according to their significance in terms of adjusted p-value (blue), fold change (green), both adjusted p-value and fold change (red), or neither (gray). (E) The most highly tissue-dependent PTMs on histone H2B proteins were determined for either their low adjusted p-value or high fold change in the pairwise tissue comparisons. Error bars represent the mean ± the standard error of the mean.

Following the comparisons among the three tissues, the PTMs on histone H2B isoforms that displayed the three lowest adjusted p-values and the three highest values of log₂ fold change were selected as the most highly tissue-dependent histone PTMs. The three histone PTMs with the lowest adjusted p-value in any of the three pairwise tissue comparisons were determined to be (1) H2B 1/2 K106 acetylation (with an adjusted p-value of 3.08 × 10^–13 when compared between gills and kidney), (2) H2B 1/2 S76 ubiquitylation (with an adjusted p-value of 7.59 × 10^–13 when compared between gills and testes), and (3) H2B 1/2 S76 ubiquitylation (with an adjusted p-value of 2.73 × 10^–12 when compared between gills and kidney). The three histone PTMs having the highest log₂ fold change when compared between tissues were (1) H2B.L4 T86 ubiquitylation (with a log₂ fold change of 5.26 when compared between gills and kidney), (2) H2B.L4 T86 ubiquitylation (with a log₂ fold change of 4.75 when compared between gills and testes), and (3) H2B.L4 K98 ubiquitylation (with a log₂ fold change of 4.75 when compared between gills and testes). Due to overlap in these two lists, the resulting four histone PTMs exemplify tissue differences on histone H2B isoforms (Figure 5E).

Characterization of Histone H3

In our samples of Mozambique tilapia, the canonical histone H3 was found to contain ten types of modifications: 4-hydroxynonenalation, acetylation, deamidation, dimethylation, dioxidation, lactylation/carboxyethylation, methylation, oxidation, phosphorylation, and ubiquitylation. In total, 22 unique and biologically relevant PTMs on the histone H3 isoforms were identified and found to be in low abundance within all three tissues (Figure 6A). Histone H3 isoforms displayed the highest proportion of tissue-dependent histone PTMs. When these PTMs were compared between the gills, kidney, and testes, we found that all 22 histone PTMs (100.00%) were significantly different between at least two of the three tissues. Of the histone PTMs analyzed, 13 (59.09%) were significantly different between the kidney and testes, 17 (77.27%) were significantly different between the gills and testes, and 20 (90.91%) were significantly different between the gills and kidney (Figure 6B–D). The kidney and testes were therefore the most similar, and the gills and kidney were the most dissimilar.

Figure 6.

Characterization of the histone PTMs on histone H3. (A) All histone PTMs were mapped on histone H3 (XP_005463512.2). Each panel represents a different amino acid position where at least one histone PTM was detected on this protein. The x-axis displays the three tissues characterized in this study: gills (G), kidney (K), and testes (T). (B–D) Volcano plots depict the differences in PTMs on histone H3 proteins between tissues. These histone PTMs were plotted based on their adjusted p-value and fold change when comparisons were made between gill and kidney samples (B), gill and testes samples (C), and kidney and testes samples (D). Histone PTMs were colored according to their significance in terms of adjusted p-value (blue), fold change (green), both adjusted p-value and fold change (red), or neither (gray). (E) The most highly tissue-dependent PTMs on histone H3 proteins were determined for either their low adjusted p-value or high fold change in the pairwise tissue comparisons. Error bars represent the mean ± the standard error of the mean.

As before, histone PTMs were selected as being most highly tissue-dependent based on their low adjusted p-value or high log₂ fold change. The following three histone PTMs were selected for their lowest adjusted p-values: (1) H3 R83 methylation (with an adjusted p-value of 1.67 × 10^–12 when compared between gills and testes), (2) H3 R83 methylation (with an adjusted p-value of 4.67 × 10^–12 when compared between gills and kidney), and (3) H3 R69 deamidation (with an adjusted p-value of 7.29 × 10^–11 when compared between gills and kidney). For their highest values of log₂ fold change between tissues, these additional three histone PTMs were selected: (1) H3 K64 ubiquitylation (with a log₂ fold change of 5.79 when compared between gills and kidney), (2) H3 K64 ubiquitylation (with a log₂ fold change of 3.21 when compared between gills and testes), and (3) H3 R69 deamidation (with a log₂ fold change of 2.83 when compared between gills and kidney). The overlap in these lists resulted in three distinct histone PTMs (Figure 6E).

Characterization of Histone H4

One histone H4 isoform was included in our DIA assay libraries for histone PTM analysis. This protein was named H4-like, and it displayed modifications of acetylation, deamidation, dioxidation, lactylation/carboxyethylation, methylation, oxidation, phosphorylation, and ubiquitylation. We identified and quantified 16 biologically relevant histone PTMs on this protein, which were all found to be in low abundance within each tissue (Figure 7A). Of the 16 histone PTMs analyzed on the histone H4 isoform, 14 (87.50%) were found to be tissue-dependent. Here, the gills and testes were the most dissimilar, and the gills and kidneys were the most similar. Between the gills and testes, 13 histone PTMs (81.25%) were found to be significantly different. Comparisons between the gills and kidney revealed that 9 of the 16 histone PTMs (56.25%) were significantly different. Between the kidney and testes, 12 of the 16 histone PTMs (75.00%) were found to be significantly different (Figure 7B–D).

Figure 7.

Characterization of the histone PTMs on histone H4. (A) All histone PTMs were mapped on histone H4-like (XP_025766521.1). Each panel represents a different amino acid position where at least one histone PTM was detected on this protein. The x-axis displays the three tissues characterized in this study: gills (G), kidney (K), and testes (T). (B–D) Volcano plots depict the differences in PTMs on histone H4 proteins between tissues. These histone PTMs were plotted based on their adjusted p-value and fold change when comparisons were made between gill and kidney samples (B), gill and testes samples (C), and kidney and testes samples (D). Histone PTMs were colored according to their significance in terms of adjusted p-value (blue), fold change (green), both adjusted p-value and fold change (red), or neither (gray). (E) The most highly tissue-dependent PTMs on histone H4 proteins were determined for either their low adjusted p-value or high fold change in the pairwise tissue comparisons. Error bars represent the mean ± the standard error of the mean.

To further characterize tissue differences in the global histone PTM landscape, we depicted the PTMs on histone H4 isoforms that displayed the three lowest adjusted p-values and the three highest values of log₂ fold change when comparisons were made between the gills, kidney, and testes (Figure 7E). The three histone PTMs with the lowest adjusted p-value when compared between any two of the three tissues were (1) H4-like K61 acetylation (with an adjusted p-value of 4.18 × 10^–16 when compared between kidney and testes), (2) H4-like Y74 oxidation (with an adjusted p-value of 4.77 × 10^–15 when compared between gills and testes), and (3) H4-like Y74 dioxidation (with an adjusted p-value of 8.89 × 10^–14 when compared between gills and testes). The histone PTMs selected for their high log₂ fold change when compared between any two of the three tissues were (1) H4-like T56 ubiquitylation (with a log₂ fold change of 1.81 when compared between gills and kidney), (2) H4-like K61 acetylation (with a log₂ fold change of −1.63 when compared between kidney and testes), and (3) H4-like Y74 oxidation (with a log₂ fold change of −1.17 when compared between kidney and testes). Due to the overlap in these lists, a total of four histone PTMs were selected as the most highly tissue-dependent PTMs on histone H4 isoforms.

Tissue-Specific Variation in the Global Histone PTM Landscape

Across all histone proteins analyzed in this study, 196 of 214 histone PTMs (91.59%) were found to be tissue-dependent through pairwise comparisons of the gills, kidney, and testes. Accordingly, a PCA performed on all samples revealed clear distinctions based on tissue type when the 214 histone PTMs were used as variables (Figure 8). In this analysis, PC1 accounted for 28.2% of the variation and PC2 accounted for 19.7% of the variation in the global histone PTM landscape of samples. The top contributors to this variance were (1) H2B 1/2 R97 oxidation, (2) H2B 1/2 K106 dioxidation, 3) H2B.L4 R89 methylation, (4) H2B.L4 T86 phosphorylation, and (5) H3 R83 dimethylation.

Figure 8.

PCA plot of tissue samples. The M-values of all 214 histone PTMs were used as variables to construct the PCA plot of tissue samples. Samples of gills (green), kidney (blue), and testes (purple) are plotted according to PC1 and PC2. A 95% confidence ellipse surrounds the cluster of each tissue type.

To further characterize tissue differences in the global histone PTM landscape, we investigated whether specific types of modification (e.g., acetylation and methylation) were consistently more abundant in certain tissues. For each type of modification, we recorded the number of instances that tissue-dependent histone PTMs had an increased relative abundance in a given tissue when compared to the other two tissues (Supplemental Table 4). Here we highlight the cases where a tissue displayed an increased relative abundance for at least 67% of the histone PTMs from a given modification type when compared with each of the other tissues.

Overall, the gills were found to have the highest levels of histone acetylation, amidation, ubiquitylation, and 4-hydroxynonelation. Of the tissue-dependent histone PTMs containing acetyl groups, 79% had a higher relative abundance in the gills than in the kidney, and 72% had a higher relative abundance in the gills than in the testes. Histone modifications with ubiquitylation showed a similar pattern. When compared to the kidney, the gills displayed an increase in 78% of the tissue-dependent histone PTMs containing ubiquitin. The gills displayed an increase in 71% of the tissue-dependent histone PTMs containing ubiquitin compared to the testes. Of the tissue-dependent instances of histone amidation, 67% had a higher relative abundance in the gills when compared to the kidney and 75% had a higher relative abundance in the gills when compared to the testes. For histone 4-hydroxynonelation, 100% of the tissue-dependent histone PTMs had a higher relative abundance in the gills than in the kidneys or testes. The testes were found to display the highest levels of histone biotinylation between the three tissues analyzed in this study. Of the tissue-dependent histone PTMs containing biotin modifications, 100% were higher in relative abundance in the testes than in the kidney or gills.

To determine whether specific types of modification were consistently less abundant in certain tissues, we recorded the number of instances in which tissue-dependent histone PTMs from each modification type displayed a decreased relative abundance in a given tissue when compared to the other two tissues (Supplemental Table 4). Through this analysis, gills were found to display the lowest levels of histone biotinylation. Of the tissue-dependent biotin modifications, 67% had a lower relative abundance in the gills when compared to the kidneys, and 100% had a lower relative abundance in the gills when compared to the testes. The testes were found to have the lowest levels of histone phosphorylation and 4-hydroxynonelation. Of the tissue-dependent instances of histone phosphorylation, 75% were lower in abundance in the testes when compared to the gills, and 77% were lower in abundance in the testes than in the kidney. Of the tissue-dependent histone PTMs containing 4-hydroxynonelation, 100% were lower in relative abundance in the testes than in the kidney or gills.

Discussion

The Use of Multiple Digestion Methods Maximizes Histone PTM Coverage

To date, approximately 700 biologically relevant histone PTMs have been identified in the scientific literature.⁴⁰⁻⁴² This number, however, far exceeds the amount that has been quantified in individual biological samples. Recent studies have managed to capture up to 200 histone PTMs in samples by using DIA methods.^43,44 Here, we demonstrate that the histone PTM coverage can be further improved by processing samples through multiple digestion methods. By incorporating three parallel digestion methods into our workflow for histone PTM analysis, we quantified 214 biologically relevant histone PTMs in every sample of Mozambique tilapia tissue. The digestion method using V8 in sodium phosphate was most successful for detecting the highest number of histone PTMs at 138; however, V8 in ammonium bicarbonate detected almost as many histone PTMs at 137, and there was a large overlap of 79 histone PTMs that could be detected using either of these two digestion methods. Although the trypsin digestion method captured the fewest number of histone PTMs at only 34, 25 of these histone PTMs could not be detected using the other digestion methods. This study, therefore, confirms that both bottom-up (e.g., trypsin) and middle-down (e.g., V8) approaches to proteomics add value to histone PTM analysis.^3,45,46 Furthermore, it offers a framework for integrating the histone peptides from multiple digestions into histone PTM quantification.

Although the analysis of histone PTMs has been generally limited to model organisms and biomedical research, we show here that over 200 histone PTMs can be simultaneously quantified in a nonmodel organism. Continuing such investigations into the histone PTMs of nonmodel organisms is critical. These epigenetic marks have been shown to drive evolutionary processes when organisms experience environmental stress.⁸ However, histone PTM landscapes do not appear to be conserved across species. For example, a study that used consistent methods for histone PTM analysis found 49 unique histone PTMs in human cells, but only 33 histone PTMs in calf thymus.⁴⁷ Another study found that the conservation of histone PTMs between species even varies depending on the histone protein. When histone PTMs were compared between the sperm of mice and humans, the percent of histone PTMs conserved between the two species was 85% for the PTMs on histone H3, but 0% for the PTMs on histone H1.⁴⁸

The workflow and tools established through this project for histone PTM analysis are now publicly available and can be adopted or modified to serve the needs of the research community. Not all digestion methods need to be completed for a comprehensive analysis of the global histone PTM landscape using our approach; instead, any combination of digestion methods can be chosen, including ones not attempted in this study. Additional proteases that have shown promise for histone PTM analysis include ProAlanase, pepsin, and Asp-N.^47,49,50 ProAlanase, for example, has been shown to increase sequence coverage of histone isoforms beyond what could be achieved through the use of either V8 or trypsin.⁵⁰ In Supplemental Table 3, we specified which histone PTMs were detected using each digestion condition in our study. We encourage others to use this table to select the most informative digestion method for their goals. Regardless of any modifications made to this workflow, the process for calculating the relative abundance, beta-value, and M-value for each histone PTM using histone peptide abundance will remain reliable and unchanged.

The Global Histone PTM Landscape Distinctly Varies by Tissue in Mozambique tilapia

By characterizing the global histone PTM landscape of the gills, kidney, and testes of Mozambique tilapia, we found that the abundances of most histone PTMs are highly dependent on tissue type. Between the three tissues, 91.59% of all quantified histone PTMs were significantly different based on their M-values. This finding supports the notion that histone PTMs uphold patterns of gene expression determined during cellular differentiation and tissue formation.^9,51−54 Overall, the kidney and testes were the two most dissimilar tissues, as they differed in 72.43% of their histone PTMs. The two most similar tissues were the gills and kidney, where 64.49% of histone PTMs was significantly different. Between the gills and testes, 69.16% of histone PTMs was significantly different. Notably, the differences in histone PTMs were so pronounced between Mozambique tilapia tissues that they were detected at a global level, independent of the genomic distribution of histone PTMs.

We expect that histone PTMs vary even more dramatically between tissues at the local level as histone PTMs frequently mediate transcriptional regulation of associated genes. Tools including ChIP-seq and reverse ChIP can be used to answer questions concerning histone PTM distribution, but such analyses are currently constricted to either a limited set of predetermined histone PTMs or a limited set of predetermined genes.^55,56 Still, these tools can be very useful in ascertaining the function of histone PTMs. For example, a recent study used ChIP-seq to demonstrate that the genomic distribution of histone H3 lysine 27 trimethylation (H3K27me3) and histone H3 lysine 4 trimethylation (H3K4me3) enforces the cellular identity of stomatal guard cells in Arabidopsis.⁵⁷ Even at a global level, several histone PTMs associated with transcriptional regulation were observed to be highly tissue-dependent in this study. Histone H3 lysine 14 acetylation (H3K14ac) and histone H3 lysine 79 methylation (H3K79me) are two such histone PTMs associated with transcriptional activation.^40,58 In the case of H3K14ac, we found the average relative abundance to be significantly higher in the gills (25.99%) compared to the kidney (12.19%) or testes (19.58%). Similarly, H3K79me was found to be significantly more abundant in the gills (14.62%) compared to the kidney (10.91%) or testes (10.91%).

Metabolite Concentrations in Tissues May Uphold Global Levels of Histone Modifications

Histone PTMs are written and erased by histone modifying enzymes (HMEs), which require substrates for catalysis. Depending on the type of modification being added to a histone, the required substrate could be acetyl-CoA (acetylation), S-adenosylmethionine (methylation), lactate (lactylation), or biotin (biotinylation), to name a few.^42,59,60 The available concentration of these substrates in the nucleus influences the rate of HME activity.⁶¹ In this study, we evaluated whether specific types of modifications, such as acetylation, were consistently more abundant or less abundant in certain tissues. We anticipated that such trends would exist because tissues balance metabolites in different manners.^62,63 Our results provide evidence to support this hypothesis. Of the 12 types of modifications analyzed in this study, five were found to be differentially abundant in the gills. Histone acetylation, amidation, ubiquitylation, and 4-hydroxynonelation tended to be more abundant, while histone biotinylation tended to be less abundant in the gills when compared to the kidney or testes. The testes tended to have higher levels of histone biotinylation, but they also displayed lower levels of histone phosphorylation or 4-hydroxynonelation than either the gills or kidney.

Based on these results, it is likely that differences between each tissue’s global histone PTM landscape depend, in part, on the tissue’s balance of key metabolites. However, the trends described above are not absolute. The gills, for example, displayed the highest relative abundance for acetylation on histone H2B 1/2 K106, but they also displayed the lowest relative abundance for acetylation on histone H4-like K46 when compared with the three tissues. This observation underscores the importance of quantifying specific histone PTMs (e.g., H3K23ac), instead of relying solely on the overall patterns of abundance for each modification type (e.g., acetylation) when conducting experiments on histone PTMs in nonmodel organisms.

Conclusions

This study presents a strategy to characterize the global histone PTM landscape within tissues of nonmodel organisms. A notable improvement in histone PTM coverage was achieved by processing every biological sample through multiple protein digestion methods in parallel. This resulted in multiple sets of histone peptides, in which each captured different histone PTMs. Consequently, we developed equations to reconcile the values of histone peptide abundance and thereby calculate the values of histone PTM abundance more accurately within every sample. Using these methods, we reliably quantified 214 biologically relevant histone PTMs in the gills, kidneys, and testes in Mozambique tilapia. This result is a significant achievement as Oreochromis mossambicus represents an as yet unannotated species, and a reference genome of a congener (O. niloticus) was used to identify highly conserved histone sequences. In comparing the global histone PTM landscape between the three analyzed tissues, we found that 196 of 214 (91.59%) histone PTMs were tissue specific. The DIA assay libraries and histone PTM quantification pipeline presented here are now publicly available and can be reused for measuring how various factors, such as development and environmental stress, influence the global histone PTM landscape. Although these tools are specific to tilapia, they were designed to be easily modified for application with other nonmodel organisms. Our study breaks new ground for comparative physiological and evolutionary investigations of how global histone PTM landscapes depend on ecological contexts and how they differ for a given species between tissues.

Glossary

Abbreviations

BCA

bicinchoninic acid

DDA

data-dependent acquisition

DIA

data-independent acquisition

HME

histone modifying enzyme

IAA

iodoacetamide

K

lysine

M

methionine

m/z

mass-to-charge ratio

me

methylation

me2

dimethylation

me3

trimethylation

P

proline

PCA

principal component analysis

PTM

post-translational modification

R

arginine

S

serine

T

threonine

TCA

trichloroacetic acid

Y

tyrosine

Data Availability Statement

All DDA and DIA raw data are available at Panorama Public (https://panoramaweb.org/eam01kl.url, DOI: 10.6069/585h-8612) and ProteomeXchange (PXD040536). The three complete DIA assay libraries including all relevant metadata and corresponding DIA data are available at Panorama Public (https://panoramaweb.org/eam01kl.url, doi: 10.6069/585h-8612). The statistical analyses performed in this study are publicly available at https://github.com/emojica2/Histone_PTM_Quantification_Pipeline.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jproteome.3c00246.

• List of available Supporting Information files (PDF)

• Supplemental File 1: Histone PTM Quantification Pipeline (XLSX)

• Supplemental File 2: Detailed Instructions for the Histone PTM Quantification Pipeline (PDF)

• Supplemental Table 1: The Biological Relevance of PTMs Detected on Histone Peptides (PDF)

• Supplemental Table 2: Sequence Coverage of Histone Proteins (XLSX)

• Supplemental Table 3: Complete Characterization of the Global Histone PTM Landscape in the Gills, Kidney, and Testes (XLSX)

• Supplemental Table 4: Tissue-Dependent Variations in the Relative Abundance of Modifications on Histone Proteins (XLSX)

The authors declare no competing financial interest.