High-throughput profiling of histone post-translational modifications and chromatin modifying proteins by reverse phase protein array

Abstract

Epigenetic variation plays a significant role in normal development and human diseases including cancer, in part through post-translational modifications (PTMs) of histones. Identification and profiling of changes in histone PTMs, and in proteins regulating PTMs, are crucial to understanding diseases, and for discovery of epigenetic therapeutic agents. In this study, we have adapted and validated an antibody-based reverse phase protein array (RPPA) platform for profiling 20 histone PTMs and expression of 40 proteins that modify histones and other epigenomic regulators. The specificity of the RPPA assay for histone PTMs was validated with synthetic peptides corresponding to histone PTMs and by detection of histone PTM changes in response to inhibitors of histone modifier proteins in cell cultures. The useful application of the RPPA platform was demonstrated with two models: induction of pluripotent stem cells and a mouse mammary tumor progression model. Described here is a robust platform that includes a rapid microscale method for histone isolation and partially automated workflows for analysis of histone PTMs and histone modifiers that can be performed in a high-throughput manner with hundreds of samples. This RPPA platform has potential for translational applications through the discovery and validation of epigenetic states as therapeutic targets and biomarkers.

1. Introduction

The major epigenomic mechanisms that regulate gene expression independently of DNA sequence include DNA methylation and post-translation modification (PTM) of histone proteins. Methylated DNA is associated with gene silencing either through blocking transcription factor-DNA binding or by recruitment of unique repressor protein complexes, whereas unmethylated DNA is generally associated with active genes [1]. The most common histone PTMs are acetylation and methylation of lysine residues, and methylation of arginine, located in the N-terminal tail domain of core histones. The resultant effect of histone PTMs is to modulate access of transcription factors to target DNA, recruit additional cofactors that assist the assembly of the preinitiation complex, and ultimately license productive transcription elongation. Histone modifications are mediated by epigenetic “writers” including histone acetyltransferases (HATs) and histone methyltransferases (HMTs) and “erasers” such as histone deacetylases (HDACs) and histone demethylases (HDMTs). Acetylation by HATs is generally associated with open chromatin conformation and active transcription, whereas deacetylation by HDACs is associated with tightly compacted and inactive chromatin. The functional effect of histone methylation is more complex, dependent on the amino acid residue modified and the number of methyl groups (mono-, di-, or trimethylation), with some modifications contributing to transcriptional activation (i.e. H3K4, H3K79) and others to repression (i.e. H3K9, H3K27)[2]. Histone readers harbor domains that can site specifically recognize histone PTMs and also facilitate access of transcription factors by recruitment of ATP dependent chromatin remodeling protein complexes that can alter mobility and positioning of nucleosomes and affect higher-order chromatin structure and long-range communication between enhancers and gene promoters.

Alterations in DNA methylation, histone PTMs and chromatin structure are normal events with critical roles during stages of development contributing to tissue specific patterns of gene expression. Aberrant epigenetic alterations are critical drivers of human diseases including initiation and progression of cancer. Hypermethylation of gene promoters resulting in loss of expression of tumor suppressors and DNA hypomethylation leading to overexpression of oncogenes are common occurrences in cancer. These changes can occur through mutations and/or alterations in expression of DNA methyltransferases (DNMTs) and Ten-eleven translocation methylcytosine dioxygenases (TETs) that dynamically regulate the addition and removal of DNA methylation. Dysregulation of histone PTMs affecting chromatin structure and function also occurs during tumorigenesis through mutations or altered expression of histone modifier and reader proteins [3, 4]. Subunits of ATP-dependent chromatin remodeler complexes are also frequently disrupted in cancer. The BRG/BRM and BAF complexes have been identified as tumor suppressors in several cancer types and are mutated with frequencies as high as 20%, including mutations in the central ATPase subunits [5]. Due to the reversible nature of epigenetic states, both DNA and histone/chromatin modifiers have been explored as therapeutic targets in cancer. Inhibitors of DNMTs have been used to reverse DNA hypermethylation and to reactivate expression of tumor suppressor genes. Inhibition of HDACs and stimulation of histone acetylation has also been used as a strategy to induce re-expression of silenced tumor suppressor genes [6, 7]. Compounds that inhibit interaction of bromodomain proteins with acetylated histones have also been developed to reduce expression of oncogenes. Small molecule inhibitors of EZH2 which catalyzes the repressive trimethylation of H3K27, and histone demethylases (LSD1 and LSD2) which remove H3K4 methylation, have also been evaluated in clinical studies [8]. Because of the important role of epigenetic abnormalities in cancer biology, histone modifications have also been explored as potential biomarkers of disease progression and prognosis [9–13].

Sequencing techniques for whole-genome profiling of DNA methylation can be performed with small amounts of tissue/cell samples and in a high-throughput manner suitable for clinical studies and drug screening. Methods are also available for high-throughput genome-wide assessment of chromatin structure such as the assay for transposase-accessible chromatin coupled with high-throughput sequencing (ATAC-seq) that is based on the use of hyperactive Tn5 transposase to simultaneously cut and ligate adaptors into chromatin regions of increased accessibility that can be read using high-throughput sequencing [14]. By contrast histone PTMs are analyzed by lower throughput targeted multiple reaction monitoring (MRM) mass spectrometry [15–17], or by protein biochemistry/immunochemical approaches with PTM selective antibodies including chromatin immunoprecipitation (ChIP)-seq and immunoblot assays. ChIP-seq is valuable for identification of genome-wide loci of histone PTMs, but it is not practical or efficient for analysis of multiple histone PTMs on large numbers of samples. Although targeted MRM mass spectrometry approaches have been reported to be able to achieve increased sensitivity with reduced input sample amounts [18], this approach requires expensive instrumentation not widely available and usually requires assay of one analyte at a time, posting limits on throughput.

In this study, we have adapted and validated reverse phase protein array (RPPA) as an efficient procedure to simultaneously measure changes in multiple histone PTMs and expression levels of proteins that enzymatically modify histones and chromatin structure. RPPA is an antibody-based targeted proteomic platform that involves microarray of cell or tissue protein lysates on replicate glass slides and probes each slide with a different primary antibody to proteins of interest and a secondary detection probe [19–24]. This is a robust platform initially developed to identify alterations in oncoprotein signaling pathways in cancer and has proven to be of value to validate gene expression and other omics data sets, and as a discovery tool for identification of novel protein signaling pathways in cancer progression and potential targets for therapeutic intervention. RPPA is a high-throughput procedure capable of simultaneously analyzing many hundreds and up to thousands of samples arrayed on the same slide. With semi-automated probing of arrayed slides with validated antibodies, up to 350 or more protein targets can be measured [22, 25–32].

2. Materials and methods

2.1. Chemicals and Peptides

Synthetic peptides consisting of histone N-terminal domain PTMs were obtained from JPT Peptide Technologies (Berlin, Germany), except the peptide for H2A.Zac that was synthesized by Thermo Fisher Scientific (Waltham, MA, USA). The sequences and modifications of amino acids residues are listed in Table 1. Inhibitors of histone modifying proteins including Suberoylanilide Hydroxamic Acid (SAHA), EPZ5676, and GSK126 were purchased from Cayman Chemical (Ann Arbor, Michigan, USA). Protease and phosphatase inhibitor tablets were purchased from Roche (Mannheim, Germany)

Table 1.

Synthetic peptides corresponding to histone N-domain PTMs

Peptide	Sequence	Sequence Region
H3S10ph	ARTKQTARK(S-ph)TGGKAPR	H3 1 : 17
H3K4me1	ART(K-me1)QTARKSTGGKAPR	H3 1 : 17
H3K4me2	ART(K-me2)QTARKSTGGKAPR	H3 1 : 17
H3K4me3	ART(K-me3)QTARKSTGGKAPR	H3 1 : 17
H3K9ac	ARTKQTAR(K-ac)STGGKAPR	H3 1 : 17
H3K9me1	ARTKQTAR(K-me1)STGGKAPR	H3 1 : 17
H3K9me2	ARTKQTAR(K-me2)STGGKAPR	H3 1 : 17
H3K9me3	ARTKQTAR(K-me3)STGGKAPR	H3 1 : 17
H3K14ac	ARTKQTARKSTGG(K-ac)APR	H3 1 : 17
H3K18ac	KAPR(K-ac)QLATKAARKSAP	H3 14 : 30
H3K27ac	KAAR(K-ac)SAPATGGVKKPH	H3 23 : 39
H3K27me3	KAAR(K-me3)SAPATGGVKKPH	H3 23 : 39
H3K36me2	KAARKSAPATGGV(K-me2)KPH	H3 23 : 39
H3K36me3	KAARKSAPATGGV(K-me3)KPH	H3 23 : 39
H4K5ac	SGRG(K-ac)GGKGLGKGGAKR	H4 1 : 17
H4K20me1	LGKGGAKRHR(K-me1)VLRDNI	H4 10 : 26
H2AZac	AGG(K-ac)AG(K-ac)DSGKAKTKAVSR	H2AZ 1 : 19
H3K79me1	RKLPFQRLVREIAQDF(K-me1)TDLR	H3 63 : 82
H3K79me2	RKLPFQRLVREIAQDF(K-me2)TDLR	H3 63 : 82
H3K79me3	RKLPFQRLVREIAQDF(K-me3)TDLR	H3 63 : 82

Chemical modifications are indicated in parentheses after the modified amino acid residue. The location of acetylated residues are indicated by (ac), methyl groups are indicated as me1 for monomethyl, me2 for dimethyl and me3 for trimethyl and phosphate residue is ph. Sequence “region” indicates amino acid numbering from the N-terminus of the core histone.

2.2. Antibodies used in RPPA

Antibodies to histone PTMs and to histone and chromatin modifying proteins along with commercial vendors and catalog numbers are provided in Supplementary Tables S1 and S2 respectively.

2.3. Cell culture and inhibitor treatment

Human hepatocellular carcinoma (HepG2) cells were cultured in MEM (Thermo Fisher Scientific) medium supplemented with 10% fetal bovine serum (Sigma, St. Louis, MI, USA), antibiotics (100 U/ml penicillin/streptomycin), 1 mM sodium pyruvate, 2 mM GlutaMAX^™, 10 mM HEPES pH7.4 (Thermo Fisher Scientific) and under a humidified atmosphere with 5% CO₂ at 37°C. Each inhibitor (SAHA, GSK126, or EPZ5676) was dissolved in DMSO, stored at −20°C, and diluted to the final concentrations as specified. Control cells were cultured in medium containing DMSO (the concentration of DMSO obtained when the largest volume of the inhibitor stock solution was added). Cells were seeded into 10-cm dishes and, one day later, were treated with either DMSO or inhibitors at indicated doses (Table 2). The medium was replaced every other day and the inhibitor solution was freshly prepared each time from a stock solution. The incubation periods and various concentrations (doses) of different inhibitors were included in Table 2. After the treatment, the cells were washed by ice-cold PBS twice and harvested by scraping into PBS. Cell suspensions (10 × 10⁶ cells) were centrifuged at 4°C for 5 minutes at 600 g and cell pellets were frozen and stored at −80°C. For mass spectrometry data comparison, SUM159 cells were cultured in 15-cm tissue dishes with Ham’s F-12 media supplemented with 10% USDA fetal bovine serum (Sigma), 100 Units/mL of Penicillin, 100 μg/mL of Streptomycin (Thermo Fisher Scientific), and 2 mM Glutamine (Thermo Fisher Scientific). Cells were grown in a 37°C incubator with 5% CO₂ and 95% relative humidity. Confluent cell cultures were treated with prepared Ham’s F-12 media supplemented with or without 5 μM sodium butyrate, made in-house from Butyric Acid (Acros Organics, Thermo Fisher Scientific) and Sodium Hydroxide (Sigma), for 2 hours at 37°C. The cells were washed with ice-cold PBS twice and harvested by scraping into PBS to prepare pellet for histone purification.

Table 2.

Experiment design of inhibitor treatment: doses and time course

Inhibitors	Enzyme	Modification	Dosage (μM)	Treatment time (days)
SAHA	Class I, II, and IV HDACs	Global histone acetylation	0,1,5,20	1
GSK126	EZH2 methyltransferase	H3K27me3 demethylation	0,1,5,20	3
EPZ5676	DOT1L methyltransferase	H3K79me1,2,3 demethylation	0,1,5,20	5

2.4. Human fibroblast cultures and iPSCs

Primary adult dermal fibroblasts (“Fib”) from a 22-year-old healthy male donor (ZenBio, lot # DFM062509) were used to induce pluripotent stem cells (“iPSCs”) as described previously [33, 34]. Briefly, to initiate reprogramming, the fibroblasts were infected with non-integrating Sendai viruses expressing OCT4, SOX2, KLF4, and c-MYC following the manufacturer’s instructions (CytoTune-iPS 2.0, Thermo Fisher Scientific). Clonal hiPSC colonies were manually picked and expanded under feeder-free conditions using hESC-qualified Matrigel (Corning, Corning, NY, USA) and TeSR-E8 medium (Stemcell Technologies, Vancouver, BC, Canada). The “iPSCs” clonal cell line used in this study (HSCC-003iPS-Sc5 or M22c5) is registered as BCMi001-A in the hPSC Registry at https://hpscreg.eu/.

2.5. Breast cancer models

The MMTV-Wnt-1 transgenic line has been previously described [35, 36]. The FUCGW lentiviral vector carrying KrasG12D (FUCGW-KrasG12D) has been previously described [28]. This virus was injected via up-the-teat route into 9-week-old MMTV-Wnt-1 mice (3.8 ×10⁷ IUs into one #4 mammary gland) as described previously [37]. Mice were palpated for tumors, which were collected for profiling of histone post-translational modifications and chromatin modifying proteins by RPPA when the size reached 2.0 cm in diameter.

2.6. Preparation of total protein extracts

Total protein lysates were prepared from cultured cells or tissue samples with modified Tissue Protein Extraction Reagent (TPER) (Life Technologies Corporation, Carlsbad, CA, USA) supplemented with 300 mM NaCl and a cocktail of protease and phosphatase inhibitors as previously described (Roche, Pleasanton, CA, USA) [24, 25, 28].

2.7. Preparation of histone synthetic peptides for RPPA

The synthetic peptides corresponding to histone N-terminal tail PTMs were dissolved in 0.1% formic acid at a concentration of 1 μg/μL and stored at −80°C. Before each RPPA run, peptides were thawed and diluted to concentration of 0.05 μg/μL in SDS sample buffer and denatured at 100°C for 8 minutes. A 5-point 2-fold serial dilution of the peptides over a range of 0.05 to 0.003125 μg/μL was prepared and arrayed onto glass slides.

2.8. Histone purification and quantification

Core histones were extracted by an acid-extraction method that is optimized for time and sample limited applications but conceptually similar to prior methods [38–40]. Briefly, the ice-thawed cell pellets or tissue powder was suspended in 1 mL of cell lysis buffer (15 mM Tris-HCl pH 7.5, 60 mM KCl, 15 mM KCl, 5 mM MgCl2, 1 mM CaCl2, 0.25 M sucrose, 0.3% NP-40 and a cocktail of protease and phosphatase inhibitors) by gentle pipetting, and left on ice for 10 min. Nuclei were isolated and collected from cell lysates by centrifugation at 600 g for 10 min at 4°C. The supernatant was discarded and nuclei were washed twice by gentle resuspension in 1 mL of nuclear isolation buffer without NP-40 and centrifugation at 600 g. The nuclei were then resuspended in 500 μL of 0.2 M HCl and rotated on an end-over-end rotator for 2 hours at 4°C. After centrifugation at 20,000 g for 10 minutes at 4°C, the supernatants containing histones were precipitated by a slow addition of trichloroacetic acid (TCA) to a final concentration of 20% with continuous and vigorous mixing. With a precipitation on ice for 5 minutes, the histones were collected by centrifugation at 20,000 g for 10 minutes, followed by a wash with 200 μL chilled (−20°C) acetone containing 0.1% (v/v) HCl, and another two washes with ice-cold 100% acetone. Histone pellets were air-dried and then dissolved with 50 μL of water. Purity of the isolated histones was verified by SDS-PAGE analysis with a 4–20% gradient Tris-glycine gels and Coomassie Brilliant Blue staining. The concentration of purified histones was determined by SYPRO Ruby staining after spotting on nitrocellulose-coated slides (Grace Bio-labs, Bend, OR, USA) [24] with bovine serum albumin BSA standard curves (0.0078 to 1 μg/μL) (see method section on RPPA for details).

Histone H3.1 isolation was carried out as described [38]. Fractions containing H3.1 were collected and dried in a Savant SPD131DDA centrifugal vacuum concentrator (Thermo Fisher Scientific), resuspended with 20 μL of water and pooled to a single 1.5 mL tube. Three microliters were used for RPPA analysis, and a portion (~1 μg) was aliquoted for SDS-PAGE analysis with a 4–20% gradient Tris-glycine gels and Coomassie Brilliant Blue staining. The remaining H3.1 was transferred to a 0.25 mL autosampler and dried down in a centrifugal vacuum concentrator for mass spectrometry.

2.9. Immunoblot analysis

Immunoblot analysis was carried out as described [24]. Briefly, whole cell lysates or purified histones were separated by Invitrogen^™ WedgeWell^™ Tris-Glycine gradient gels (Thermo Fisher Scientific) and transferred to a PVDF membrane (Bio-Rad, Hercules, California, USA). Membranes were blocked for 1 hour at room temperature with 5% milk or 5% BSA in 1xTris-buffered saline Tween-20 (TBST) and probed with the indicated primary antibodies at 4°C overnight. After washing with TBST three times for 30 min, membranes were incubated with anti-rabbit IgG or anti-mouse IgG peroxidase-conjugated secondary antibodies (Cell Signaling Technology, Danvers, Massachusetts, USA) at room temperature for 1 hr. The membranes were washed with TBST three times and then immersed in chemiluminescence substrate (Bio-Rad). Chemiluminescent detection was visualized by HyBlot CL^® Autoradiography Film (Thomas Scientific, Swedesboro, New Jersey, USA).

2.10. Reverse phase protein array (RPPA)

RPPA assays were carried out as described previously [20, 28, 41–43]. Synthetic peptides were prepared at five 5-point 2-fold serial dilutions (0.05–0.003125 μg/μL), purified histones were diluted over a range of 0.5–0.03125 μg/μL and whole cell extracts were diluted to a concentration of 0.5 μg/μL, all in SDS sample buffer supplemented with 2.5% 2-mercaptoethanol and denatured by heating to 100°C for 8 min. For mass spectrometry comparison, 4 μg purified histone H3.1 for each sample was diluted into a total of 8 μL in SDS sample buffer supplemented with 2.5% 2-mercaptoethanol and denatured by heating to 100°C for 8 min. The Aushon 2470 Arrayer (Aushon BioSystems/Quanterix, Billerica, MA, USA) with a 40 pin (185 μm) configuration was used to spot samples in technical triplicates onto nitrocellulose-coated slides (Grace Bio-labs, Bend, OR, USA) using an array format of 960 samples/slide (2880 spots/slide with technical triplicates). Slides were pretreated with Re-Blot Plus Strong Antibody Stripping Solution (2504; MilliporeSigma) and IBlock Protein-Based Blocking Reagent (T2015; Thermo Fisher Scientifc). Replicate slides were each incubated at RT for 30 min with a primary antibody specific for 20 different histone PTMs and 40 histone/chromatin modifier antibodies, each at a predetermined optimal concentration. Slides were then incubated with either a goat anti-rabbit or anti-mouse IgG secondary antibodies (NC9256157 or NC9372061; Vector Laboratories) followed by a catalyzed signal amplification system kit and fluorescent IRDye 680 Streptavidin (LI-COR Biosciences, Lincoln, NE, USA) as the detection probe. Immunolabeling was performed with an automated slide stainer (Autostainer Link 48, Agilent Dako, Santa Clara, CA, USA). Fluorescence-labeled slides were scanned on a GenePix 4400 AL scanner, at appropriate photomultiplier tube (PMT) settings to obtain optimal signals for each specific set of antibody-labeled slides. The images were analyzed with GenePix Pro 7.0 (Molecular Devices) and the fluorescence signal intensity (SI) of each antibody labeled spot was obtained after subtraction of local slide background. Selected control slides were incubated with antibody diluent in place of primary antibody as a negative control and with Sypro Ruby Protein blot stain for assessment of total protein content according to manufacturer’s protocol (Molecular Probes, Eugene, OR).

2.11. RPPA data normalization & statistical analyses

For data normalization, a group-based method was carried out [24, 41, 43]. Briefly, the SIs from negative control slides were subtracted from that of the primary antibody and specific SIs for each spot were normalized for variation in Sypro Ruby-stained total protein SI within each defined sample group exactly as described [24]. In this study, each experimental group was defined by individual inhibitor (i.e. SAHA) or experimental model. For purified histones with serial dilution, each protein concentration was considered its own group thus the same dilution point was normalized together in order to observe the effect of serial dilution. For example, the antibody SIs for each dilution of histones for SAHA treatment (along with its corresponding DMSO controls) were normalized for variation in Sypro Ruby-stained total protein at the corresponding concentration on the dilution curve. For synthetic peptides for histone PTMs the input concentration (0.05 μg/μL for the highest concentration) was too low to accurately measure the total protein content by Sypro Ruby. Therefore, total SIs for each the primary antibody were subtracted from that of negative control without total protein normalization.

Analysis of Variance (ANOVA) for multiple group comparison and t-test for two-group comparison were performed for all data. Statistical analyses for purified histone samples were carried out on normalized signals obtained at 0.5 μg/μL concentration because signals are at linear range and not saturated at this dose (Fig. 3B). Prior to statistical analysis, normalized signals across all samples were filtered and SI values below <200 were considered negative protein expression and samples with and high-CV (CV>25%) data were removed from further analysis. Remaining antibody SI values were subjected to respective statistical analyses, with significance achieved at false discovery rate-adjusted P value <0.1, and t-test with significance P value <0.05. In addition, a cut-off of a 1.5-fold difference was applied.

Figure 3. RPPA detects changes in multiple histone PTMs upon treatment of cells with pan HDAC inhibitor SAHA.

(A) RPPA images of serial dilution of histones stained with Sypro Ruby (green) for total protein and with antibodies (red) to five histone PTMs from cells in response to different doses of SAHA. Each treatment has biological and technical triplicates.

(B) Normalized signal intensity values for immunolabeling with H4K5ac antibody at each dose of spotted histones from untreated and SAHA treated cells. The values represent fluorescence signal intensity as mean ± SD of replicates after normalization to total protein.

(C) Normalized signal intensity values for immunolabeling with antibodies to the histone PTMs. Fluorescence signal intensity is presented as the mean ± SD from biological and technical replicates from each SAHA treatment. Unpaired t-tests and a cut-off of a 1.5 fold difference were performed to evaluate differences between treated and untreated SAHA samples. *p < 0.05, **p < 0.01, ***p < 0.001.

(D) Immunoblot validation of changes in histone PTMs in response to SAHA treatment with antibodies to the five targets in A and C above, plus an antibody to total H3 protein as a loading control. Arrow indicates the location of histone protein bands. Numbers under the blot represent relative changes quantified by imaging analyses of the bands. Shown are results of a representative immunoblot experiment of at least duplicate independent immunoblots.

2.12. Mass Spectrometry analysis for peptide validation

Synthetic peptides were resuspended in a 0.5% Formic acid and 50% MeOH solution at a concentration of 100 ng/μL and directly sprayed into Orbitrap Fusion^™ Tribrid^™ Mass Spectrometer by syringe pump at a flow rate of 3 μl/min. The m/z of each peptide was isolated and fragmented by HCD and CID sequentially. Average 100 of MS2 fragment signal per peptide was captured and searched against the pooled target peptides database in Proteome Discoverer 2.1 interface (PD 2.1, Thermo Fisher Scientific) with the Mascot algorithm (Mascot 2.4, Matrix Science). Dynamic modifications of the acetylation, monomethylation, dimethylation, and trimethylation of lysine were allowed. The precursor mass tolerance was confined within 20 ppm with fragment mass tolerance of 0.5 Da. Each peptide’s identification was manually validated.

2.13. Middle-Down Mass Spectrometry analysis for histone PTM profiling

Mass spectrometry analysis for histone H3.1 samples prepared from section 2.8 was performed as previously described with a few modifications [38]. Dried samples were resuspended in Mass Spec resuspension buffer (2% MeCN, 0.1% formic acid in water) to yield 2 μg/μL. Samples were chromatographically separated and introduced into a Lumos Orbitrap Mass Spectrometer (Thermo Fisher Scientific) using an UltiMate 3000 HPLC (Thermo Fisher Scientific). Three technical replicates using 2 μg of H3.1 were performed for each sample.

After digestion with 0.2 μg of Glu-C, H3.1 peptides were chromatographically separated and introduced into a Lumos Orbitrap Mass Spectrometer (Thermo Fisher Scientific) using an UltiMate 3000 HPLC (Thermo Fisher Scientific). A multi-segment linear gradient was used to 4–15% B in 70 minutes; 15–35% from 70–105 minutes; 35–98% in 5 minutes, holding at 98% for 5 minutes before re-equilibration at 0.20 μL/minute (Buffer A: 2% MeCN, 0.1% formic acid; Buffer B (98% MeCN, 0.1% formic acid). MS² scans were in data-dependent mode with the parameters for 2 methods. In Method 1 (0–80 min), Peptides were detected with the orbitrap at a scan range of 505–640 m/z (3 microscans) with a resolution of 60,000. A targeted mass list was used to target only the 34 nominal masses associated with the +9 charge state and proteoforms of the histone H3 1–50AA peptide. Peptides were isolated with an isolation width of 1 m/z and fragmented by ETD. Ion detection was carried out with both the orbitrap (Resolution: 30,000) and ion trap (Scan Rate: Zoom; Scan range: 470–530 m/z). In Method 2 (80–125 min), the orbitrap detection was set to a scan range of 1133–1300 m/z at a resolution of 60,000. Peptides matching the K79 containing peptides were isolated by quadrupole isolation with an isolation width of 1 m/z and fragmented by HCD. Ions were detected with the orbitrap at a resolution of 15,000. The first mass at 200 m/z, AGC of 5.0 × 10⁵, maximum injection time of 22 ms, and 3 microscans were used.

Mass spectrometry data analysis was carried out with a previously described in-house data analysis suite [44, 45]. H3.1 protein sequence with the following variable modifications: K4me1, K4me2, K4me3, K9ac, K9me1, K9me2, K9me3, K14ac, K18ac, K23ac, K27me1, K27me2, K27me3, K36me1, K36me2, and K36me3 were used to generate a fragment ion library to match acquired data. T-test for two-group comparison was carried out, with significance achieved at false discovery rate-adjusted P value <0.1, and P value <0.05, and a cut-off of a 1.5-fold difference, as described above for RPPA analyses for the same samples.

3. Results

3.1. Specificity of RPPA for histone PTMs

A key feature for a reliable RPPA procedure is the quality and validation of each antibody. For this study, commercially available antibodies to 20 different histone PTMs (Table S1) were selected for validation based on specificity and other performance data from the literature plus a database created for assessment of antibodies that recognize histone PTMs (http:www.histoneantibodies.com). Additionally, the antibodies selected are for commonly studied histone PTMs known to be associated with transcriptional activation or repression. Specificity under RPPA conditions was established by the use of synthetic peptides corresponding to N-terminal tail regions of core histones containing chemically modified acetylations, methylations (mono-, di-, and trimethylation) or phosphorylation at the specific amino acid residues shown in Table 1. Also shown are the amino acid sequences, sites, types of modifications and sequence regions of N-terminal tails of histones H3, H4 or H2A.Z (Table 1). Each peptide and modification was confirmed independently by mass spectrometry amino acid sequencing (Fig. S1). Peptides were spotted in triplicate at a five dose serial dilution ranging from 0.05 to 0.003125 μg/μL and each slide was incubated with one of the twenty anti-histone PTM antibodies (Fig. 1A). Although peptide arrays commonly use biotinylated peptides bound to streptavidin plates or slides [46, 47], we determined that non-tagged peptides worked reliably on nitrocellulose membrane-coated glass slides. Additionally, use of carrier protein (1% BSA) for spotting peptides was found to make no difference in the quality of spots and reliability of data (data not shown), thus slides were routinely arrayed with free peptides. Each antibody exhibited a high level of specificity for the corresponding modified histone peptide with little or no cross-reaction with any other peptide (Fig. 1). As examples with antibodies to H3K9ac, H3K9me1, H3K9me2, H3K9me3, H3K27me3, H3K36me2, H4K5ac, and H4K20me1, arrayed spots exhibited strong dose responsive fluorescence signals only with the expected specific peptide on the same slide arrayed with all other histone PTM peptides (Fig. 1A). Antibody specificity was further illustrated as shown by a heat-map of the fluorescence signal intensity (SI) with all peptides at a single 0.0125 μg/μL concentration (Fig. 1B). Quantitative SI values for the H3K9me3 antibody, as an example, shows a high level of differential binding to the corresponding H3K9me3 peptide over all other histone PTM peptides, including H3K9 me1 and H3K9me2 (Fig. 1C). Importantly, the SI values (H3K9me3 as an example) were dose responsive over the range of spotted synthetic peptide concentrations, indicating the ability of RPPA to detect changes in the level of histone PTMs (Fig. 1D). The lack of cross-reaction with other histone PTM peptides further indicates that these antibodies do not recognize epitopes in unmodified histone tails under these conditions and are highly dependent on the specific histone PTM. These synthetic histone PTM peptides were arrayed on all subsequent RPPA assays as specificity controls to assure reliability and consistency between experiments and batches of antibodies.

Figure 1. Specificity of anti- histone PTM antibodies under RPPA conditions.

(A) Map of arrayed histone PTM peptides as serial dilutions on glass slides (top) and fluorescence signals obtained with 8 selected anti-histone PTM antibodies as examples (bottom).

(B) Heat map of fluorescence intensity signals for binding of each of the 20 different antibodies (Y axis) to all the histone PTM peptides (X axis) at concentration of 0.0125 μg/μL.

(C) Specificity of H3K9me3 antibody for binding to the corresponding peptide at concentration of 0.0125 μg/μL.

(D) Fluorescence intensity signals with H3K9me3 antibody against the H3K9me3 peptide spotted as serial dilution.

Single histone molecules in biological systems are commonly modified at multiple sites by different types of PTMs [48]. We have previously shown that combinations of K9 and K27 methylation commonly co-occur on the same molecule (in cis) with both K14 and K23 acetylation [49]. Others have shown that H3K4me3 and H3K9ac frequently occur together and are enriched at gene promoters associated with active transcription [50] Genomic co-localization can result from occurrence in cis (on the same molecule) or trans (in the same nucleosome but not on the same molecule) [51]; however, when they do occur in cis it can lead to epitope occlusion masking detection of either PTM by antibody-based methods [52]. To determine whether antibodies in RPPA assays are able to detect a specific histone PTM in the presence of a second modification, double modified synthetic histone peptides including H3K9me3/H3K14ac, H3K4me3/ H3K9ac, and H3K27me3/ H3K36me3, were analyzed for reactivity with antibodies to corresponding single modifications contained in the peptides. Each antibody exhibited the same specificity and similar sensitivity for the double modified peptides as compared to the corresponding single modified peptides (Fig. S2). These results show at least for these antibodies that the detection of the one histone modification is not blocked by a co-existing modification nearby.

3.2. Validation of histone PTM profiling with inhibitors of enzymatic histone modifying proteins

To further validate the antibodies and to challenge the ability to concurrently profile changes in multiple PTMs of endogenous histones, we performed RPPA analysis of core histones isolated from HepG2 cell cultures treated with and without small molecule inhibitors of histone modifier enzymatic proteins. The targeted histone modifying enzymes, histone PTMs expected to be preferentially affected and experimental conditions for each inhibitor are listed in Table 2. A rapid and efficient method for isolation of core histones was optimized based on an acid-extraction precipitation method from cell nuclei as described [38, 40]. This procedure as outlined schematically yields ~250 μg purified product from 10 × 10⁶ cells, or ~30 mg of tissue, with a high degree of purity of core histones including H1, H2A/B, H3, and H4 as shown by SDS-PAGE analysis (Fig. 2).

Figure 2. Histone purification scheme and analysis by SDS-PAGE and Coomassie blue staining.

A consistent level and degree of purity of isolated core histones was obtained across different experimental treatment groups (SAHA), indicating that total histone protein levels were not altered by inhibitor treatment of cells (Fig. 2). Isolated core histones were resuspended in RPPA sample buffer at a concentration of 0.5 μg/μL and were arrayed in technical triplicates over a two-fold serial dilution range of 0.5 to 0.03125 μg/μL. Each slide was incubated with one of the 20 antibodies to histone PTMs followed by a secondary antibody and a fluorescence detection probe. Fluorescence signal intensities were normalized to total protein in each spot from replicate slides stained with Syrpo Ruby (Fig. 3A).

As shown by fluorescence signals of arrayed spots, five histone acetylation sites (H2A.Z, H3K9, H3K18, H3K27, and H4K5) were observed to be increased the most in HepG2 cells after treating with the pan HDAC inhibitor, SAHA, in a dose dependent manner between 1 to 20 μM of SAHA (Fig. 3A). Quantitative fluorescence signal intensity (SI) values after normalization to total histone protein, increased in a linear manner over the range of arrayed histone concentrations (up to 0.5 μg/μL), as shown with antibody to H4K5ac as a representative example of an PTM affected by SAHA (Fig. 3B). Subsequent statistical analyses for effects of SAHA were carried out on normalized signals obtained at the 0.5 μg/μL concentration of purified histones. SI values were significantly increased for antibodies to H2A.Zac H3K9ac, H3K19ac, HSK27ac and H4K5ac (Fig. 3C). Immunoblot assays confirmed increased acetylation of H2A.Zac, H3K9ac, H3K18ac, H3K27ac, and H4K5ac in response to SAHA treatment without a change in total H3 protein (Fig. 3D). Increases in these acetylation sites have been reported previously to be the predominant effects on histone PTMs upon SAHA treatment [53]. Of the other 15 antibodies to histone PTMs, SAHA also stimulated less substantial but a significant increase in SI for antibodies to H3K4me1, H3K4me3 and H3K27me3 and a decrease in H3pS10, H2K9me1, H3K36me3 and H3K79me2 (Fig. S3A). No changes were detected for antibodies against the other eight histone PTMs (Fig. S3B).

RPPA results with other inhibitors of histone modifying enzymes (Table 2) are shown in Figures 4 and 5 along with immunoblot confirmation. GSK126 is an inhibitor of the EZH2 methyltransferase that methylates H3K27 as a primary target [54]. The most significant effect of GSK126 was a decrease in H3K27 trimethylation as detected by RPPA and confirmed by immunoblot (Fig. 4). GSK126 also had a small but significant effect on some other histone PTMs (H3K4me3, H3K9ac, H3phosS10, H3K27ac, H3K36me2, H4K5ac, and H4K20me1), while no significant change was observed with the other 12 histone PTM antibodies (Fig. S4). We also examined a lysine methyltransferase (DOT1L) inhibitor (EPZ5676) (Fig. 5). The DOT1L protein is a methyltransferase that primarily methylates lysine 79 of histone 3 (H3K79), adding mono-, di-, or trimethyl groups [55]. The DOT1L inhibitor, EPZ5676 (pinometostat), is reported to reduce H3K79 methylation and have moderate clinical activity in patients [56]. The major effect of EPZ5676 treatment was a significant decrease in H3K79 mono-, di- and trimethylation as detected by RPPA analysis that was confirmed by immunoblot assay (Fig. 5). Smaller but significant effects were also observed in four other histone PTMs (H3K4me1, H3K9me3, H3K14ac, and H3K36me2) with no significant effect in the other 13 histone PTMs (Fig. S5). These results demonstrated that RPPA was able to selectively detect global changes in specific histone PTMs known to be major targets of these inhibitors.

Figure 4. Histone PTM profiling upon EZH2 inhibitor GSK126 treatment.

(A) RPPA images of serial dilution of histones stained with Sypro Ruby (green) for total protein and with antibody to H3K27me3 (red) from cells treated with the different doses of GSK126. Each treatment has biological and technical triplicates.

(B) Normalized signal intensity values for immunolabeling with antibody to H3K27me3. Fluorescence signal intensity is presented as the mean ± SD from biological and technical replicates from each GSK126 treatment. Unpaired t-tests and a cut-off of a 1.5 fold difference were performed to evaluate differences between treated and untreated GSK126 samples. *p < 0.05, **p < 0.01, ***p < 0.001.

(C) Immunoblot validation of changes in H3K27me3 in response to GSK126 treatment, plus an antibody to total H3 protein as a loading control. Arrow indicates the location of histone protein bands. Numbers under the blot represent relative changes quantified by imaging analyses of the bands. Shown are results of a representative immunoblot experiment of at least duplicate independent immunoblots.

Figure 5. Histone PTM profiling upon DOT1L inhibitor EPZ5676 treatment.

(A) RPPA images of serial dilution of histones stained with Sypro Ruby (green) for total protein and with antibody to H3K79me1, H3K79me2 and H3K79me3 (red) from cells treated with the different doses of EPZ5676. Each treatment has biological and technical triplicates.

(B) Normalized signal intensity values for immunolabeling with antibody to H3K79me1, H3K79me2 and H3K79me3. Fluorescence signal intensity is presented as the mean ± SD from biological and technical replicates from each EPZ5676 treatment. Unpaired t-tests and a cut-off of a 1.5 fold difference were performed to evaluate differences between treated and untreated EPZ5676 samples. *p < 0.05, **p < 0.01, ***p < 0.001.

(C) Immunoblot validation of changes in H3K79me1, H3K79me2 and H3K79me3 in response to EPZ5676 treatment, plus an antibody to total H3 protein as a loading control. Arrow indicates the location of histone protein bands. Numbers under the blot represent relative changes quantified by imaging analyses of the bands. Shown are results of a representative immunoblot experiment of at least duplicate independent immunoblots.

As a further validation of RPPA to detect changes in histone PTMs, we made comparisons by RPPA and mass spectrometry on purified histones from SUM159 cells treated with or without the pan histone deacetylase inhibitor, sodium butyrate. In these experiments purified histone H3.1 was analyzed instead of all core histones and the same samples were prepared and divided for Coomassie blue SDS-PAGE (to confirm the purity of H3.1, data not shown), RPPA, and mass spectrometry. Among the 20 validated PTM antibodies for RPPA, three are for PTMs on H2 and H4, therefore only the 17 antibodies to H3 PTMs should be potentially reactive in this experiment. As expected, RPPA signal was minimally detectable with antibodies to H2A.Zac, H4K5ac, and H4K20me1 (Fig. 6A). Specific strong signals were detected by RPPA for the 17 antibodies to H3 PTMs with significant increases in acetylation of H3K9ac, H3K14ac, H3K18ac and H3K27ac as the major effect of butyrate treatment. Smaller increases were observed for methylations (H3K4me1, H3K9me1, and H3K36me3) and no changes occurred with the remaining 10 antibodies (Fig. 6A). Mass spectrometry detected 11 of the 17 H3 PTMs with significant increases in acetylation of H3K9ac, H3K14ac, H3K18ac and methylation of H3K4me1 and H3K9me3 (Fig. 6B), commonly identified by RPPA in response to butyrate. Also common to both methods were five H3 methylations showing no changes upon butyrate treatment (Fig. 6). As expected, additional modifications were detected by mass spectrometry for PTMs that antibodies were not available for RPPA including an increase in H3K23ac in response to butyrate treatment (Fig. 6B). These results show a good concordance between the two methods for detection of majority of changes in histone modifications under the experimental conditions. Differences potentially reflect higher sensitivity for selected antibodies to PTMs in RPPA and complementary ability of mass spectrometry in identification of modifications for which antibodies were lacking.

Figure 6. Histone PTM comparison between RPPA and MS.

SUM159 cells cultured in F-12 medium were treated with or without 5μM sodium butyrate for 2 hours (n=3). After the treatment, histones were extracted and H3.1 was isolated for RPPA (A) and MS (B) analysis. RPPA data were normalized and expressed as the mean ± SD. The significant differences between sodium butyrate treated and untreated samples are evaluated by unpaired t-tests. *p < 0.05, ** p < 0.01, *** p < 0.001. (C) Representative RPPA images of six different PTMs.

3.3. RPPA profiling of histone PTMs and epigenetic modifier proteins in experimental model systems

In order to challenge the ability of the RPPA platform to detect simultaneous changes in multiple histone PTMs experimentally, we used an induced pluripotent stem cell model of differentiation and a mouse mammary tumor initiation model. In addition to profiling histone PTMs, we also performed RPPA analysis for expression levels of 40 different epigenetic modifier proteins listed in Supplementary Table S2 that includes histone “writers, erasers and readers”, chromatin remodelers, and regulators of DNA methylation. The workflow for these experiments involves splitting the same cells or tissue samples into total protein lysates for analysis of epigenetic modifier proteins and purified core histones from fractioned nuclei for analysis of histone PTMs (Fig. 7A). The antibodies for epigenetic modifier proteins were validated and analyzed by our established RPPA methods for detection of relative changes in proteins in cell or tissue lysates. Validation criteria include specificity for positive and negative controls and dose responsive signals under RPPA conditions, as well as antibody specificity by immunoblot assays for a single protein band of correct molecular size in appropriate null and positive cell or tissue control samples [24].

Figure 7. Profiling of histone post-translational modifications and expression levels of chromatin modifying proteins during somatic cell reprogramming.

(A) Epigenetic RPPA workflow.

(B) Heat map of changes in histone PTMs detected by RPPA utilizing 0.5 μg/μL histones. Quantitative values were scaled by converting signal intensities to z-score values. Unpaired t-tests p<0.05, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference were performed to evaluate differences.

(C) Bar charts of normalized RPPA signal intensities for selective histone PTM antibodies of active marks (H3K18ac, H3K27ac, H3S10ph) and a repressive mark (H3K27me3) in iPSCs compared to fibroblasts. Data are expressed as means ± SD from four replicates in each experimental group. Unpaired t-tests, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference were performed to evaluate differences between iPSCs and Fibroblast cells. ***p < 0.001.

(D) Immunoblot validation of the RPPA-identified differences of histone PTMs between fibroblasts and iPSCs. Total histone H3 is a loading control. Arrow indicates the location of histone protein bands. Numbers under the blot represent relative changes quantified by imaging analyses of the bands. Shown are results of a representative immunoblot experiment of at least duplicate independent immunoblots.

(E) Heat map of differential expression levels of histone modifier and chromatin remodeling proteins between iPSCs and fibroblasts identified by unpaired t-tests p<0.05, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference.

(F) Bar charts of normalized RPPA signal intensity values for selective chromatin/histone modifier proteins showing increases of BAF155, BAF57 and LSD1 expression after cell reprogramming. Data are expressed as means ± SD from four replicates in each experimental group. Unpaired t-tests, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference were performed to evaluate differences between iPSCs and Fibroblast cells. ***p < 0.001.

(G) Immunoblot validation of selected differentially expressed chromatin/histone modifier proteins identified by RPPA. Arrow indicates the location of histone protein bands. β-Actin is a loading control. Numbers under the blot represent relative changes quantified by imaging analyses of the bands. Shown are results of a representative immunoblot experiment of at least duplicate independent immunoblots.

3.4. Changes in histone PTMs and epigenetic modifier proteins during somatic cell reprogramming

Post-translational modifications are known to play important roles in determining the lineage, developmental trajectory, and fate of cells within a multicellular organism [57]. Induced pluripotent stem cells (iPSCs) are derived from the de-differentiation of somatic cells, by reactivating the pluripotency gene regulatory network and reverting them to a distinctive embryo-like epigenetic state [58–60]. Direct reprogramming of somatic cells to iPSCs provides a favorable system to study the epigenetic features that are prerequisites for pluripotency. Therefore, we set out to quantify epigenetic changes in primary fibroblasts before and after reprogramming by RPPA profiling of histone PTMs and expression levels of epigenomic modifier proteins. Primary human dermal fibroblasts were reprogrammed to iPSCs by transiently overexpressing OCT4, SOX2, KLF4, and C-MYC using Sendai viruses [61]. We observed that stem cells proliferate more rapidly while primary human dermal fibroblasts grow slower and may be partially contact-inhibited but are not completely quiescent. Multiple euchromatin/active histone PTMs (including H3K18ac and H3K27ac) were observed by RPPA to be significantly increased while other repressive histone PTMs (H3K27me3 and H3K79me2/me3) were significantly lower in the pluripotent state than in the differentiated state (Fig. 7B & C). Results were further confirmed by immunoblot assays (Fig. 7D). Reprogramming of somatic cells to iPSCs requires re-opening of chromatin in a process that probably involves some of the same factors that maintain open chromatin. Therefore, we compared the levels of protein expression of several types of histone-modifying and chromatin-remodeling factors in iPSCs and fibroblasts from whole-cell extracts (Fig. 7E & F). Protein expression of histone modifiers and chromatin remodelers were in general significantly higher in iPSCs than in fibroblasts (Fig. 7E), which correlates with more active histones/chromatin in iPSCs (Fig. 7B). An apparent contradictory observation was that iPSCs showed higher levels of KMT6A (EZH2), primarily responsible for methylating H3K27, but have less H3K27me3 than fibroblasts. An explanation might be that various histone lysine demethylases (KDM1A, KDM4B, KDM5A, KDM7B) are also elevated, likely offsetting the effect of EZH2 activity in iPSCs (Fig. 7E).

Mammalian SWI/SNF (SWItch/Sucrose Non-Fermentable) ATP-dependent chromatin remodeling complexes are critical for early embryo development, and are known to form cell type-specific complexes based on subunit composition [62]. Proteomic studies using mouse ES cells showed that they require the activity of distinctive complexes (esBAF), which specifically include Brg (Brahma-related gene), BAF155, and BAF60A, but exclude Brm (Brahma), BAF170, and BAF60C [63]. Indeed, RPPA analysis of histone modifiers and chromatin remodeling proteins in whole cell lysates showed that human iPSCs, which are nearly indistinguishable from human ES cells, have increased levels of SMARCC1 (BAF155), SMARCE1 (BAF57), and KDM1A (LSD1) in comparison to the original fibroblasts (Fig. 7E & F). These results were further confirmed by immunoblots (Fig. 7G).

3.5. RPPA identification of changes in epigenetic markers in histologically distinct mammary tumors driven by the same oncogenes

Breast cancer can be categorized into multiple histological subtypes, including invasive ductal carcinoma, lobular carcinoma, inflammatory breast cancer, and metaplastic breast cancer. While genetic drivers have been shown to have key roles in specifying these subtypes [28, 64–66], the role of epigenetic modifications in the development of histopathologically distinct cancers is not as well defined. Therefore, a previously described MMTV-Wnt1 transgenic mouse model of breast cancer [36] was used here to apply the RPPA as a platform to profile changes in multiple histone PTMs and chromatin modifiers during the transformation of mammary epithelial cells into tumors with different histopathological types. The MMTV-Wnt1 model requires a secondary oncogenic event to progress from pre-cancerous lesions to invasive cancer, which can be achieved by mutation and activation of RAS [67]. The genetic combination of Wnt-1 and a constitutively activated version of Ras is sufficient for the rapid transformation of mammary epithelial cells and the appearance of mammary tumors [28]. However, despite the apparent lack of additional genetic drivers, the resulting tumors are highly heterogeneous varying from typical adenocarcinoma to metaplastic carcinoma with predominantly squamous differentiation (Fig. 8B). Additionally, adenocarcinomas formed more rapidly than metaplastic carcinomas (Fig. 8A). Three adenocarcinomas and 4 squamous-differentiated tumors in this model at 2.0cm in diameter were harvested and processed for histology and extraction of total proteins and histones. Significant expression differences in a subset of histone PTMs (Fig. 8C & D) and in epigenetic modifier proteins (Fig. 8F & G) were observed between adenocarcinomas versus squamous-differentiated tumors. Using immunoblot assays, we confirmed expression alterations in H3K9me3, H3K4me3, and H3K14ac (Fig. 8E) and several histone-modifying proteins including BAF57, CtBP2, and LSD1 (Fig. 8H). Interestingly, the affected histone PTMs and chromatin-modifying proteins were all lower in squamous-differentiated tumors than in adenocarcinomas. This general reduction is not a technical artifact because only a small number of proteins are studied here and increased levels of signaling proteins were also detected in the squamous-differentiated tumors in this MMTV-Wnt1 model when a large panel of antibodies were used (data not shown and [28]). Since the specific histone PTMs reduced in the squamous-differentiated group include markers of both active (H3K4me3) and repressive (H3K9me3) states of chromatin (Fig. 8C), there is likely not a general shift between active/open and repressive/closed chromatin when transformed mammary cells gained metaplastic characteristics. However, the general reduction in histone PTMs and chromatin-modifying proteins suggests a less dynamic state of the epigenome in these metaplastic tumors. A reduction of DNA methyltransferase (DNMT1) was also observed in these metaplastic carcinomas, indicating that alterations in DNA methylation could contribute to the difference between these two tumor histopathological types (Fig. 8F).

Figure 8. Profiling of histone post-translational modifications and chromatin modifying proteins in a mouse breast cancer progression model induced by KrasG12D from MMTV-Wnt1 mice.

(A) Kaplan-Meier tumor-free survival of seven 9-week-old MMTV-Wnt1 transgenic mice injected intraductally with FUCGW-KrasG12D virus. The adenocarcinomas and squamous tumors are marked with blue and green dots, respectively.

(B) The representative images of histopathology of tumors stained with H&E.

(C) Heat map of relative differences in PTMs of histone (0.5 μg/μL) as detected by RPPA between adenocarcinomas and squamous-differentiated mammary tumors. Unpaired t-tests p<0.05, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference were performed to evaluate differences.

(D) Bar charts of normalized RPPA signal intensity values for selected histone PTMs showing that the active marks of H3K9me3, H3K4me3, and H3K14ac are higher in adenocarcinomas (n = 3) than squamous-differentiated mammary tumors (n = 4). Data are expressed as means ± SD from four replicates in each experimental group. Unpaired t-tests, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference were performed to evaluate differences between adenocarcinomas and squamous-differentiated mammary tumors. *p < 0.05, **p < 0.01.

(E) Immunoblot validation of selected histone PTMs detected by RPPA in squamous-differentiated mammary tumors (n = 4) vs adenocarcinomas (n = 3). Arrow indicates the location of histone protein band. Total histone H3 is a loading control. Numbers under the blot represent relative changes quantified by imaging analyses of the bands. Shown are results of a representative immunoblot experiment of at least duplicate independent immunoblots.

(F) Heat map of normalized RPPA data showing relative differential expression of histone modifier and chromatin remodeling between adenocarcinomas and squamous-differentiated mammary tumors identified by unpaired t-tests p<0.05, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference.

(G) Bar charts of normalized RPPA signal intensities for selected histone modifier/chromatin remodeling proteins in adenocarcinomas (n = 3) and in squamous-differentiated mammary tumors (n = 4). Data are expressed as means ± SD from four replicates in each experimental group. Unpaired t-tests, adjusted p value (FDR) =<0.1, and a cut-off of a 1.5 fold difference were performed to evaluate differences between adenocarcinomas and squamous-differentiated mammary tumors. **p < 0.01, ***p < 0.001.

(H) Immunoblot validation of RPPA results in C. Arrow indicates the band location of histone modifiers. β-Actin was used as a loading control. Numbers under the blot represent relative changes quantified by imaging analyses of the bands. Shown are results of a representative immunoblot experiment of at least duplicate independent immunoblots.

4. Discussion

Histone modification-mediated epigenetic regulations are actively involved in physiological and pathological processes, such as cell proliferation, differentiation, dedifferentiation, and malignant transformation. In this report, we describe an adaptation and validation of an antibody-based reverse phase protein array as a high-throughput platform for quantitative profiling of concurrent changes in global histone modifications and levels of regulatory proteins that modify histones and chromatin structure.

Reliable analysis of histone PTMs required the use of purified core histones instead of crude whole cell extracts, as determined by comparative analysis (data not shown). Therefore, a rapid and efficient microscale method for purification of core histones from isolated cell nuclei was included in the procedure. Meanwhile, modifying proteins of histone and chromatin were able to be analyzed from whole tissue/cell protein extracts by our standard RPPA methods [24]. Therefore, a workflow was developed to fractionate tissue/cell samples as total protein lysates and core histones for subsequent spotting on glass slides, followed by probing with corresponding antibodies. In addition to our established RPPA assays [24], novel procedures were introduced in this platform for global analysis of relative quantitative changes in histone PTMs. This included the use of synthetic peptides corresponding to site-specific modifications in the N-terminal domain of histones to validate the specificity and selectivity of each antibody for the histone PTM of interest. Antibody specificity was observed over a wide range of synthetic histone peptide concentrations, and signal intensity varied in a linear pattern, indicating the RPPA assay is capable of detecting changes in relative levels of specific histone PTMs.

To challenge the ability of RPPA to profile relative changes in multiple PTMs on endogenous histones, we first analyzed the effects of chemical inhibitors of “writer” and “eraser” proteins on histone marks in cell culture experiments. The most significant effect of the pan HDAC inhibitor SAHA was observed to be an increase in acetylation of five histone PTMs including H2A.Zac, H3K9ac, H3K18ac, H3K27ac and H4K5ac. Smaller but significant changes in other histone PTMs were observed including an increase in H3K4me1/me3 and H3K27me3, and a decrease in H3S10ph, H3K9me1, H3K36me3 and H3K79me2. Some of these histone PTMs affected by SAHA have been reported previously [68–70]. Multiple different amino acid residues along the length of histone terminal domains can be acetylated or methylated either singly or in different combinations [71, 72]. Extensive cross-talk between sites of modification occurs, which can be either permissive or interfere with regulation of neighboring PTMs. This combinatorial complexity of histone PTMs provides a wide range of determinants capable of regulating multiple active or repressive states of chromatin, and is a possible explanation for effects of SAHA on histone PTMs other than acetylation. Similar results were obtained with other chemical inhibitors with most significant effects observed on the expected target histone PTMs including decreased methylation of H3K27 by the EZH2 methyltransferase inhibitor GSK126 and decreased methylation of H3K79 by the DOT1L methyltransferase inhibitor EPZ5676. RPPA results with inhibitors were confirmed by immunoblots.

To further validate the RPPA platform, we performed a direct comparison of RPPA with mass spectrometry of histone H3 isolated from SUM159 cells treated with or without the pan HDAC inhibitor, sodium butyrate. Both methods detected many of the same changes in H3 PTMs anticipated for the pan HDAC inhibitor [73]. However, RPPA detected an increase in H3K27ac and in methylation at H3K36me3 that were not detected by mass spectrometry. Conversely, mass spectrometry identified four H3 PTMs (H3K27me1, H3K27me2, H3K36me1 and H3K23ac) including an increase in acetylation of H3K23ac by butyrate treatment that were not possible to assess by RPPA, since validated antibodies were not available in the current study.

These results confirm the value and reliability of RPPA to detect changes experimentally in histone PTMs and further support the complementary nature of RPPA and mass spectrometry [22]. Histone PTMs detected by RPPA suggest a greater sensitivity for some modifications due to high affinity of selected antibodies combined with signal amplification, which may offer an advantage for low abundant histone PTMs [74]. In the absence of a validated antibody for histone PTMs, mass spectrometry is the method of choice and as observed here it can complement RPPA to obtain a more comprehensive profile of changes in histone modifications. The RPPA assay is easily scalable to expand the number of PTMs detected beyond the 20 sites in the present study with additional validated antibodies. However in the event that quality antibodies for every PTM of interest are not possible, mass spectrometry could be used as complementary method.

The utility of the RPPA epigenetic platform for experimental applications was illustrated in two model systems including dedifferentiation/reprogramming of somatic cells to iPSCs and mammary tumor progression to different histopathologic tumor types. The global open chromatin epigenetic landscape observed by the RPPA platform here is consistent with previously noted properties of chromatin modifications in iPSCs, including a prominent set of histone modifications, such as H3K18ac, H3K27ac, H3K9ac and H3K4me1, and high levels of chromatin-remodeling protein expression to open chromatin. Such comprehensive characterization can be used to better define stages of developmental pluripotency. Furthermore, the technology described here is potentially valuable in identifying critical changes in epigenetic profiles of pluripotent stem cells, such as comparing profiles before and after directed differentiation into a cell type of interest, as well as in healthy and disease states. Using an MMTV-Wnt1/RAS syngeneic mouse mammary tumor model, the application of the RPPA platform was further demonstrated experimentally with dissected tumor tissues. Differences in multiple histone PTMs and levels of chromatin/histone modifying proteins were detected between adenocarcinoma and squamous-differentiated tumors, suggesting that epigenetic changes may contribute to the progression of mammary tumors to different histopathology types.

A previous study by Partolina and colleagues [75] described an epigenetic RPPA method with an in-house set of highly selective monoclonal antibodies that recognize seven different histone H3 and H4 lysine acetylations and five different H3 lysine methylations. Similar to our study, they determined that reproducible results required use of purified core histones from isolated nuclei as opposed to rapid histone purification kits or use of crude nuclear protein extracts. In cell culture experiments, they found a global increase in acetylation of all the specific histone residues analyzed by RPPA in response to treatment with broad spectrum HDAC inhibitors. Increases in H3 methylation were also observed (H3K4me3 and H3K9me1). Similarly, we observed an increase in H3K4me3 with SAHA treatment but did not detect an increase in H3K9me1 at lower concentration but a decrease in signal intensity at 20 μM concentration. This difference could be due to use of the HDAC inhibitor HC-toxin by Partolina et al. [75] versus SAHA in the present study. In a subsequent report, van Dijk et al. [76] used RPPA for simultaneous measurement of the relative level of expression of 20 different histone modifying proteins plus three histone methylations including H3K4me2, H3K4me3 and H3K27/me3. Histone modifying proteins and histone methylations were analyzed by standard RPPA methods in total cell protein lysates from bone marrow derived CD43+ cells from normal subjects and a cohort of 205 newly diagnosed acute myeloid leukemia (AML) patients. The results showed recurrent expression patterns of histone modifying proteins that significantly correlated with patient overall survival, suggesting that concurrent analysis of multiple histone modifications may be useful for prognosis in AML and for identifying novel epigenetic therapeutic targets.

Compared with these previous reports, the current RPPA epigenetic platform provides a more comprehensive analysis that simultaneously measures relative expression of multiple histone/chromatin modifying proteins and histone PTMs from the same cell or tissue samples. The use of synthetic peptides to histone PTMs provides robust controls for validating specificity of antibodies to histone PTMs that can be applied on all arrays to assure the reliability and consistency of the assay between experiments and batches of antibodies. Our study also showed the complementary feature of RPPA and middle-down mass spectrometry in histone PTM profiling, and established RPPA as a valuable and a reliable tool for profiling changes in histone PTMs. A limitation of RPPA lies in availability of antibody validation for all PTMs which can be compensated by mass spectrometry, while RPPA can potentially identify low abundant modifications that middle-down mass spectrometry may not be able to identify.

In the present study, we also profiled expression levels of 40 proteins that are either responsible for modifying histones or other epigenomic features. These proteins were prepared by standard RPPA methods as crude whole cells extracts, and it requires very small amounts of sample on the order of 18μg of total protein to array enough slides to probe with up to 350 antibodies. The protein arrayer pins deposit 6 nL or ~3 ng of total protein lysate/spot. RPPA assays with whole cell/tissue protein extracts can be performed with even smaller amounts of scarce materials from micro-dissected tissues sections routinely [77, 78]. RPPA of histone modifications requires purification of histones and is thus more of a challenge and requires more starting material than whole cell extracts. However, we adapted a simple, rapid and efficient procedure from Holt et al. [38] for purification of core histones that is amenable to processing large number of samples on the order of sample sets prepared as whole cell lysates. For the purpose of establishing and validating procedures we used 10 × 10⁶ cells, or ~30 mg of tissue as starting material for histone purifications that yielded ~250 μg of purified core histones, which was sufficient for all the experiments for the whole study including RPPA and immunoblots. RPPA platform that has been semi-automated is high-throughput since it is capable of analyzing up to 1,000 or more protein samples for up to 350 protein targets yielding more than 1,000,000 data points. The assay requires specialized equipment and technical expertise that is best provided in a Core Facility setting with a 4–5 week turnaround time for completion including data processing and analysis. The RPPA epigenetic platform described here could be scalable to measure changes in essentially all known histone PTMs and proteins that modify histones and other epigenomic features to which quality validated antibodies are available. This platform that can be used with micro-scale tissue samples may be useful for clinical discovery and validation studies for epigenetic therapeutic targets and biomarkers and for drug discovery studies. The RPPA platform may also serve as a screening tool for higher resolution assays such as ChIP-Seq to identify the specific genomic loci of altered histone modifications and chromatin remodeling events.

5. Conclusion

In this study, we have developed a high-throughput antibody-based RPPA platform for profiling 20 histone PTMs and 40 proteins that enzymatically modify histones and remodel chromatin structure. The platform can be used to determine epigenetic profiles during physiological and pathological conditions on various sample sources, including but not limited to iPSCs and cancer cells, exhibiting the potential to delineate the underlying mechanisms of epigenetic-mediated developmental processes and human diseases.

Abbreviations:

RPPA

reverse phase protein array

PTM

post-translational modification

H3K4me1

monomethylation of histone 3 at lysine 4

H3K4me2

dimethylation of histone 3 at lysine 4

H3K4me3

trimethylation of histone 3 at lysine 4

H3K27ac

acetylation of histone 3 at lysine 27

H3S10ph

phosphorylation of histone 3 at serine 10

HAT

histone acetyltransferase

HMT

histone methyltransferase

HDAC

histone deacetylases

HDMT

histone demethylase

DNMT

DNA methyltransferase

SAHA

suberoylanilide hydroxamic acid

ChIP-Seq

chromatin immunoprecipitation sequencing

iPSCs

induced pluripotent stem cells

Fib

fibroblasts

IDC

invasive ductal carcinoma

Data Availability

Data described in this paper is shown in the figures. The original data making the figures is available upon request to the corresponding author.