Nitric oxide generates a coordinated histone modification signature consistent with chromatin compaction, linked to methyl-modifying enzyme dysregulation and tumor-permissive gene silencing Authors: Sara Shayan, Divya Vasudevan, Douglas D Thomas Published: 2026 ABSTRACT AbstractOur laboratory has established that the free radical signaling molecule nitric oxide (NO) is an endogenous regulator of epigenetic methylation through direct inhibition of the Fe(II)/2-oxoglutarate-dependent dioxygenase (2-ODD) family, including histone demethylases (KDMs)[1, 2], mRNA demethylases (FTO, ALKBH5) [3], and DNA demethylases (TETs, ALKBH2) [4]. We previously showed that NO alters 38 histone post-translational modifications (PTMs) and genome-wide transcription in breast cancer cells [3]. Here, we comprehensively quantify 76 histone modifications, including 30 combinatorial marks on adjacent residues not resolved in the original analysis, and systematically profile methyl-modifying enzyme expression.We identify four coordinated histone PTM axes converging on a chromatin compaction signature: (i) H3K9 methylation accumulation with selective K9me1K14ac remodeling, (ii) H4K20me1→me2/me3 conversion, (iii) a novel H3K27–K36 combinatorial landscape shift, and (iv) preferential depletion of hyperacetylated H4 species. A composite Chromatin Compaction Score integrating all 76 PTMs with literature-based valence assignments confirms progressive compaction, validated independently by PCA. ChIP-seq reanalysis reveals that H3K9me2 accumulates predominantly at intergenic regions while K9ac undergoes activity-dependent redistribution governed by baseline chromatin state, with near-perfect reciprocal H3K9me2 gain at silent loci. Systematic enzyme profiling shows coordinated upregulation of 13/18 detected KDMs alongside methyltransferase downregulation, a compensatory response that fails to overcome direct enzymatic inhibition by NO.This compaction signature is associated with tumor-permissive gene silencing: DNA repair genes are downregulated, chromatin remodeling complexes suppressed, and IDH1 downregulation creates a feed-forward loop compounding 2-ODD inhibition. These findings establish NO-mediated KMD inhibition as a driver of coordinated chromatin compaction linked to pro-tumorigenic gene silencing.Keywords: Nitric oxide, Histone post-translational modifications, Histone demethylases, Chromatin compaction signature, Tumor suppressor silencing, Epigenetics Introduction Nitric oxide (NO) is a pleiotropic signaling molecule with established physiological roles in vascular homeostasis, neurotransmission, immune defense, and cancer biology [1]. Beyond classical signaling through soluble guanylate cyclase activation and S-nitrosation, our laboratory has established NO as an endogenous regulator of epigenetic methylation through direct inhibition of the Fe(II)/2-oxoglutarate-dependent dioxygenase (2-ODD) family of demethylases. We demonstrated that physiological NO concentrations directly and reversibly inhibit 2-ODD enzymes responsible for removing methyl groups from histones (e.g., KDM3A) [2,3], mRNA (e.g., FTO, ALKBH5) [4] [3], and DNA (e.g., TET enzymes, ALKBH2) [5], leading to substrate-specific hypermethylation and altered gene expression. Independent of 2-ODD inhibition, NO has also been shown to regulate chromatin state through S-nitrosylation of chromatin-modifying enzymes. Bhatt and colleagues demonstrated that S-nitrosylation of HDAC2 at Cys262/274 triggers its release from chromatin, promoting histone acetylation and gene activation in neurons [6], a process mediated by transnitrosylation from SNO-GAPDH to nuclear targets including HDAC2 and SIRT1 [7]. More recently, S-nitrosylation of the NuRD chromatin remodeling complex was shown to increase DNA accessibility during cellular reprogramming [8]. These findings establish NO as a multifaceted epigenetic regulator that can alter chromatin state through both S-nitrosylation-dependent mechanisms and the 2-ODD inhibitory pathway that is the focus of the present study. Our mechanistic studies revealed that NO, which has structural and bonding similarities to O2, inhibits 2-ODD catalysis not by replacing its substrate O2 but by competing with 2-oxoglutarate at the catalytic iron center [2,3]. Structural analysis showed that two NO molecules bind to form a dinitrosyl iron complex (DNIC; [Fe(NO)2]9) on the demethylases [5]. Although we previously determined that DNICs are the most abundant NO-derived adducts formed within cells exposed to NO [9], the discovery that DNIC formation at the 2-ODD active site directly inhibits catalytic function represents one of few examples of functional consequences attributable to cellular DNIC formation. Beyond their role at 2-ODD catalytic centers, DNIC formation has broader consequences for cellular iron homeostasis. We previously demonstrated that NO-mediated iron sequestration into DNICs depletes the chelatable iron pool, which independently contributes to histone methylation increases and upregulates the metastasis suppressor NDRG1 through an iron deficiency signaling pathway [2,10]. Intracellularly, DNICs are stored by glutathione S-transferase P1-1 (GSTP1) and actively exported by multidrug resistance protein 1 (MRP1), constituting a coordinated system that regulates NO and iron availability [11,12]. NDRG1, which is regulated by cellular iron levels and functions as a metastasis suppressor across multiple cancer types [13,14], appears among the NO-activated loci in the present study (see Results, Fig. 4A), suggesting convergent regulation through both iron deficiency signaling and chromatin remodeling. Representative Browser Track Patterns Illustrate Five Categories of Locus-Specific K9 Remodeling. H3K9ac and H3K9me2 ChIP-seq signal tracks (RPGC-normalized bigWig files) displayed at representative gene loci. Control (untreated) and NO-treated (24 h DETA/NO, 500 μM) tracks shown for both marks at each locus. Gene expression log2 fold-changes (NO/UT, from Affymetrix microarray [3]) indicated for each gene. (A) K9ac Gain at NO-Activated Genes (K9me2 Unchanged): HMOX1 (+1.6), NDRG1 (+2.1), ADM (+2.4), and LOX (+2.8) show increased K9ac signal at their promoters. (B) K9ac Loss at NO-Silenced Genes: LUM (−3.2) and FGFBP1 (−1.7) show K9ac depletion without K9me2 gain. (C) Dual Compaction (K9ac Loss + K9me2 Gain): SEMA3D (−1.7), HCP5 (−1.6), and SPINK4 (−1.2) show simultaneous K9ac loss and K9me2 accumulation, representing direct evidence of a permissive-to-repressive chromatin switch. (D) K9me2 Loss + K9ac Gain at 2-ODD Target Gene: EGLN2/PHD1 (+0.4) shows K9me2 loss consistent with chromatin opening for compensatory enzyme upregulation. (E) K9me2 Gain at NO-Silenced Genes: CLIC3 (−1.7) and HSD17B8 (−1.6) show promoter K9me2 accumulation. Gain and loss of histone modifications in Fig. 4 browser tracks are assessed by comparing RPGC-normalized (Reads Per Genomic Content) ChIP-seq signal between NO-treated and untreated conditions at the promoter region (defined as TSS ± 2 kb). For each locus, the integrated signal area under the curve in the NO-treated track was compared to the corresponding untreated control. Loci were classified as showing gain (NO > UT) or loss (NO < UT) when the fold-change exceeded 1.5-fold at the promoter. These browser track observations are consistent with the genome-wide quantitative analyses presented in Fig. 3 and Supplemental Fig. S2. Expression fold-changes (parenthetical values) are from Affymetrix microarray data. Blue traces: untreated control; red traces: NO-stimulated (24 h DETA/NO, 500 μM). Arrows indicating transcription direction have been enlarged for improved readability. Among the 2-ODD family, the histone demethylases (KDMs) are of particular interest because they directly regulate the histone code, the combinatorial pattern of post-translational modifications governing chromatin structure and gene accessibility. Over 20 KDM family members target methylated lysine residues across histones H3 and H4 [15]. We found that these enzymes exhibit markedly different sensitivities to NO and are differentially expressed across tissues and cellular states, suggesting that selective inhibition by NO may result in cell-type-specific epigenetic reprogramming [5]. Our initial studies demonstrated that NO inhibition of the histone demethylase KDM3A leads to H3K9me2 accumulation and, in breast cancer cells, correlates with genome-wide transcriptional changes affecting hundreds of genes [2,3]. ChIP-seq confirmed redistribution of H3K9me2 and H3K9ac across genomic loci [2]. However, if NO broadly inhibits multiple KDM family members, as the shared 2-ODD catalytic mechanism predicts, then methylation changes should extend across multiple histone residues targeted by distinct KDM subfamilies, including H3K4 (KDM5), H3K27 (KDM6), H3K36 (KDM2/KDM4), and H4K20 (PHF8) [2,5]. A critical dimension of this question that has not been addressed is the behavior of combinatorial histone modifications, the co-occurrence of multiple marks on adjacent residues of the same histone peptide. Mass spectrometry uniquely resolves these combinatorial states because tryptic peptides spanning multiple lysine residues preserve the relationship between adjacent modifications. For example, the H3 peptide spanning residues 9–17 resolves four distinct K9–K14 states (e.g., K9me1K14ac vs. K9me3K14ac), and the K27–40 peptide resolves the full K27 × K36 modification matrix. These adjacent marks cannot be resolved by antibody-based approaches and were not analyzed in our prior study [2]. Additionally, the expression of chromatin-modifying enzymes as a coordinated functional class, particularly the simultaneous behavior of histone demethylases and their cognate methyltransferases, was not systematically examined in our original analysis [16,17]. With new understanding of how NO can globally inhibit numerous 2-ODD enzymes (beyond KDM3A), we return to the histone mass spectrometry and gene expression microarray datasets from Vasudevan et al. [3] to perform new analyses not included in the original publication. First, we quantify 76 histone H3 and H4 modifications including 30 combinatorial marks across adjacent residues in the K9–K14, K27–K36, and H4 multi-acetylation regions, revealing coordinated state transitions invisible at the single-mark level. Second, we systematically analyze expression changes across all detectable 2-ODD demethylases and cognate methyltransferases, revealing a coherent compensatory transcriptional program that nonetheless fails to prevent methylation accumulation. Third, we develop a composite Chromatin Compaction Score (CCS) integrating all 76 PTMs with literature-based valence assignments, providing quantitative evidence that these four axes converge on a unified compaction signature. Fourth, we analyzed H3K9ac and H3K9me2 ChIP-seq data to determine the genomic distribution of K9 modification changes, reconciling bulk mass spectrometry measurements with locus-specific chromatin remodeling. Together, these analyses demonstrate that NO generates a histone modification pattern consistent with chromatin compaction across four independent modification axes through coordinated regulation of methyl-modifying enzymes, and that this signature manifests as locus-specific redistribution of activating and repressive marks at gene promoters. Materials and methods Dataset and study design All data in this study are derived from the experiments described in our original study Vasudevan et al. [3]. Briefly, MDA-MB-231 triple-negative breast cancer cells were treated with the NO using the NO donor DETA/NO (500 μmol/L; ∼250 nmol/L [NO]ss) for 24 or 48 h, with matched untreated controls at each timepoint. A “wash” condition was included in which the NO-source (DETA/NO) was removed after 24 h and cells were maintained in NO-free medium for an additional 24 h (48 h total). A baseline (0 h) condition served as reference. All conditions were performed in biological triplicate (n = 3). Cell culture, histone extraction, mass spectrometry, and microarray methods are described in detail in Ref. [2]. Gene expression microarray data and ChIP-seq data are available through GEO (accession: GSE66324). Adjacent histone PTM quantification The mass spectrometry data from our study [3] were analyzed to quantify combinatorial modifications on tryptic peptides bearing modifications at adjacent lysine residues. In addition to the 38 individual modification states reported previously, we quantified 30 combinatorial marks including: H3 K9–K14 peptide (4 species), H3 K27–K36 peptide (8 species), H3.3 K27–K36 peptide (7 species), H4 K5–K8–K12–K16 multi-acetylation (10 species), and H3 K18–K23 (1 species). A total of 76 individual and combinatorial modifications were quantified across 6 conditions (18 samples). Statistical analysis Pairwise comparisons between conditions were performed using Welch's two-sample t-test on biological triplicates (n = 3 per condition). The six experimental conditions: baseline (0 h), 24 h untreated, 24 h NO, 48 h untreated, 48 h NO, and wash (24 h NO followed by 24 h recovery), yielded 12 biologically informative condition-pairs, each constituting one comparison family: (i) primary treatment effects at matched timepoints (24 h NO vs. 24 h UT; 48 h NO vs. 48 h UT); (ii) NO and culture effects relative to baseline (24 h NO vs. 0 h; 48 h NO vs. 0 h; 24 h UT vs. 0 h; 48 h UT vs. 0 h); (iii) progressive effects within treatment groups (48 h NO vs. 24 h NO; 48 h UT vs. 24 h UT); and (iv) reversibility (Wash vs. 48 h NO; Wash vs. 48 h UT; Wash vs. 24 h NO; Wash vs. 0 h). Within each family, all ∼76 PTMs were tested simultaneously and Benjamini–Hochberg false discovery rate (FDR) correction was applied independently, controlling for multiple testing across PTMs while preserving sensitivity to condition-specific effects. This per-family correction strategy was chosen over a single global correction across all 912 tests because each condition-pair addresses a distinct biological question. Three cross-design comparisons (e.g., 24 h UT vs. 48 h NO) were excluded as they confound treatment and time effects. Of 912 total comparisons, 207 were nominally significant (p < 0.05), 74 survived per-family FDR at q < 0.05, and 147 at q < 0.10. Chromatin compaction score (CCS) To quantitatively assess the net direction of chromatin remodeling across all 76 PTMs, we developed a composite Chromatin Compaction Score. Each modified PTM state was assigned a literature-based valence of +1 (repressive/compaction-promoting), −1 (permissive/compaction-opposing), or 0 (ambiguous), based on established functional associations in the histone modification literature (see Supplemental Table S2 for complete assignments and citations). For each biological replicate, PTM abundances were z-score normalized across all samples, then multiplied by the assigned valence and summed to yield a per-sample CCS. Higher (more positive) CCS values indicate a net shift toward repressive chromatin marks; lower (more negative) values indicate permissive chromatin. The CCS was also decomposed into four axis-specific sub-scores (H3K9–K14, H4K20, H3K27–K36, H4 acetylation) to assess the contribution of each regulatory axis independently. Statistical comparisons of CCS between conditions were performed using Welch's t-test. Principal component analysis (PCA) validation PCA was performed on z-score normalized PTM abundances across all 18 biological replicates (6 conditions × 3 replicates) to provide an unsupervised validation of the CCS valence assignments. Spearman rank correlation was computed between PC1 loadings and the literature-based valence assignments to assess concordance between data-driven and literature-derived chromatin state classifications. Sensitivity analysis To assess the robustness of the CCS to ambiguous valence assignments, three scenarios were tested: (1) all ambiguous marks assigned +1 (compaction), (2) all ambiguous marks assigned −1 (permissive), and (3) all ambiguous marks excluded from the calculation (primary analysis). CCS trajectories were compared across scenarios and were nearly identical, confirming that the compaction signature is insensitive to ambiguous mark handling (Fig. 2E). Chromatin Compaction Score Integrates Four PTM Axes into a Unified Compaction Signature. (A) Composite Chromatin Compaction Score (CCS) computed by assigning each PTM a literature-based valence: compaction-promoting (+1), compaction-opposing (−1), or ambiguous (0). CCS increases progressively with NO treatment (24 h: p = 0.034; 48 h: p = 0.004) and persists through washout (p = 0.031). Error bars: SEM (n = 3). (B) Axis-specific CCS sub-scores showing each axis contribution. (C) PCA of the full 76-PTM matrix; PC1 (40.7% of variance) separates NO-treated from untreated conditions. (D) PC1 loadings vs. literature-based valence assignments (Spearman rho = −0.56, p less than 0.00001), providing unsupervised data-driven validation. (E) Sensitivity analysis: CCS recomputed under three ambiguous-mark scenarios; all trajectories are nearly superimposable (less than 10% divergence). (F) Schematic overview of the four axes converging on the unified compaction signature. All data from mass spectrometry (n = 3 biological replicates; Supplemental Table S2). Methyl-modifying enzyme expression analysis The gene expression microarray data (Affymetrix GeneChip PrimeView Human Gene Expression Arrays; GEO accession: GSE66324) from Ref. [3] were examined with targeted analysis of all detectable 2-ODD demethylases and histone methyltransferases. Expression changes are reported as linear fold-change (DETA/NO vs. control at 24 h) with associated p-values. For histone PTM data, fold-changes are reported as log2 values throughout (Table 1, Table 2 and text). Table 1Expression of methyl-modifying enzymes in response to •NO (24h).Table 1GeneFunctionFCp-valueSig.Axis2-ODD Demethylases (↑) — 13 of 18 upregulatedKDM3AH3K9me1/2 demeth.+3.330.0005∗∗∗K9JMJD1C2-ODD, K9-related+2.030.015∗K9KDM5BH3K4me2/3 demeth.+1.600.003∗∗K4KDM6BH3K27me2/3 demeth.+1.440.243nsK27KDM5AH3K4me2/3 demeth.+1.430.067nsK4KDM5CH3K4me2/3 demeth.+1.370.001∗∗K4KDM6AH3K27me2/3 demeth.+1.310.003∗∗K27KDM2AH3K36me1/2 demeth.+1.300.012∗K36KDM2BH3K36me1/2 demeth.+1.250.082nsK36PHF8H3K9me2/H4K20me1+1.160.054nsK9/K20KDM4DH3K9/K36 demeth.+1.080.048∗K9/K36KDM4BH3K9/K36 demeth.+1.070.063nsK9/K36KDM5DH3K4me2/3 demeth.+1.070.093nsK42-ODD Demethylases (↓)PHF2H3K9me2/H4K20me1−1.750.003∗∗K9/K20KDM4AH3K9/K36 demeth.−1.200.272nsK9/K36KDM4CH3K9/K36 demeth.−1.120.102nsK9/K36KDM3BH3K9me1/2 demeth.−1.080.197nsK9MINARibosomal demeth.−1.020.660nsOtherMethyltransferases (↓) — 6 of 7 downregulatedNSD2/WHSC1H3K36me2 writer−4.07<0.0001∗∗∗K36EHMT2/G9aH3K9me1/2 writer−1.680.095nsK9EHMT1/GLPH3K9me1/2 writer−1.580.022∗K9SETDB1H3K9me3 writer−1.520.046∗K9SUV39H1H3K9me3 writer−1.310.002∗∗K9KMT5A/SET8H4K20me1 writer−1.180.016∗K20Methyltransferases (↑)SUV4–20H1H4K20me2/3 writer+1.280.004∗∗K20Table 1. Expression changes of methyl-modifying enzymes in response to 24h •NO treatment. Fold-change (FC) values represent •NO vs. untreated control from microarray data (GEO: GSE66324). Among 2-ODD histone demethylases, 13 of 18 detected were upregulated, consistent with a broad compensatory transcriptional response to enzymatic inhibition. Cognate methyltransferases show the opposite pattern: 6 of 7 were downregulated, with the sole exception of SUV4–20H1, which catalyzes the H4K20me1→me2/me3 conversion consistent with the observed K20 redistribution. Significance: ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, ns = not significant. Expression of methyl-modifying enzymes in response to •NO (24h). Table 1. Expression changes of methyl-modifying enzymes in response to 24h •NO treatment. Fold-change (FC) values represent •NO vs. untreated control from microarray data (GEO: GSE66324). Among 2-ODD histone demethylases, 13 of 18 detected were upregulated, consistent with a broad compensatory transcriptional response to enzymatic inhibition. Cognate methyltransferases show the opposite pattern: 6 of 7 were downregulated, with the sole exception of SUV4–20H1, which catalyzes the H4K20me1→me2/me3 conversion consistent with the observed K20 redistribution. Significance: ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, ns = not significant. Table 2Key histone PTM changes by regulatory axis.Table 2Modification24h FC48h FCp (48h)RecoveryReversibilityH3K9 AxisK9me1−0.56−1.380.00333%PersistentK9me2+0.76+0.580.038105%ReversibleK9me3+0.33+1.000.057−1%†WorsenedK9me1K14ac−0.95−1.64<0.00134%PersistentK9me3K14ac+1.23+1.720.14772%PartialH4K20 AxisK20me1−0.70−1.61<0.00118%PersistentK20me2+1.25+2.07<0.00129%PersistentK20me3+1.87+2.89<0.00143%PartialH3 K27–K36 AxisK27me1K36me1−2.16−3.080.04017%PersistentK27me1K36me3+1.01+0.910.01415%PersistentK36me1 (total)−3.81−5.120.00622%PersistentH4 Acetylation AxisK8acK12acK16ac−1.00−3.020.00313%PersistentK5acK8acK16ac−1.01−2.360.0083%PersistentK5acK12acK16ac−1.72−3.250.06429%PersistentTable 2. Key histone post-translational modification changes organized by regulatory axis. FC = log2 fold-change vs. matched NT control. p (48h) = raw p-value for 48h •NO vs. NT comparison. Recovery% = percent reversal of 48h •NO effect upon 24h washout (48h Wash vs. 48h NO, relative to 48h NT). Most changes show <35% recovery, indicating persistent chromatin remodeling. †K9me3 continued to accumulate after •NO removal (negative recovery), consistent with self-reinforcing heterochromatin spreading. Key histone PTM changes by regulatory axis. Table 2. Key histone post-translational modification changes organized by regulatory axis. FC = log2 fold-change vs. matched NT control. p (48h) = raw p-value for 48h •NO vs. NT comparison. Recovery% = percent reversal of 48h •NO effect upon 24h washout (48h Wash vs. 48h NO, relative to 48h NT). Most changes show <35% recovery, indicating persistent chromatin remodeling. †K9me3 continued to accumulate after •NO removal (negative recovery), consistent with self-reinforcing heterochromatin spreading. ChIP-seq analysis H3K9ac and H3K9me2 ChIP-seq data generated in the original study [3] were reprocessed from raw FASTQ files (GEO accession: GSE66324) using the nf-core/chipseq pipeline (v2.1.0) with default parameters for single-end data. Reads were aligned to the human genome (GRCh38) using BWA-MEM, followed by duplicate removal (Picard MarkDuplicates), quality filtering (MAPQ ≥10), and blacklist region exclusion. Peaks were called using MACS2 (narrow peaks for H3K9ac; broad peaks for H3K9me2). Signal tracks (bigWig files) were generated and normalized to reads per genomic content (RPGC). For promoter-level quantification, H3K9ac and H3K9me2 signal was computed across promoter regions (TSS ± 2 kb) for all annotated genes using pyBigWig, yielding per-gene mean signal values for each replicate. H3K9me2 gene body signal was additionally quantified from TSS to TES. Fold-changes were computed as log2(NO/UT) for each replicate pair and for replicate means. Quality control of the H3K9ac ChIP-seq data revealed a signal magnitude imbalance between the two untreated replicates (mean promoter signal: UT-R1 = 0.381, UT-R2 = 0.189; 2.01:1 ratio), whereas the NO-treated replicates were well-matched (NO-R1 = 0.457, NO-R2 = 0.505; 0.90:1 ratio). Despite this absolute signal difference, within-condition correlations were high for both pairs (UT: r = 0.928; NO: r = 0.983), indicating that gene-level rankings were preserved. Because averaging the imbalanced UT replicates would systematically inflate the apparent NO/UT fold-change, all quantitative K9ac promoter analyses were performed using the replicate 1 (R1) pair, in which both conditions were well-matched in signal magnitude. The replicate 2 pair corroborated gene-level trends (inter-replicate fold-change correlation: r = 0.75; 77.5% directional agreement) and is presented in Supplemental Fig. S1. H3K9me2 replicates were well-matched across both conditions and were analyzed as replicate means. For expression-stratified analyses, genes were classified by microarray log2 fold-change as upregulated (>0.5), downregulated (<−0.5), or unchanged (−0.5 to +0.5). Results Adjacent modification analysis reveals widespread NO-dependent PTM changes Our original analysis [3] quantified 38 individual histone PTM states and identified substantial NO-dependent changes including H3K9me2 accumulation, H3K27me2 increase, H3K36me3 increase, and H4K20me3 increase. We analyzed the mass spectrometry data to include modifications at adjacent residues, for example, resolving H3K9me1 alone from H3K9me1K14ac on the same tryptic peptide, expanded the total to 76 quantified PTM states (66 modified states across 10 tryptic peptides, plus 10 unmodified baselines). Of these, the top 40 ranked by effect size are shown in Fig. 1A; the majority are combinatorial marks not previously reported. Three features of the expanded dataset are immediately apparent. First, the changes are overwhelmingly directional: repressive methylation marks increase while permissive marks (acetylation, low-state methylation) decrease, consistent with a global shift toward heterochromatin (Fig. 1A). Second, combinatorial marks dominate the most extreme changes; the single strongest hit, H3K9me1K14ac (log2FC = −1.64, p < 0.001), was invisible to single-mark analysis because the decrease in K9me1 and the persistence of K14ac cancel when measured independently. Third, the changes are progressive and largely persistent through washout (Fig. 1B), indicating that NO establishes a self-reinforcing chromatin state rather than a transient signaling response. Volcano plot analysis confirmed statistically significant changes spanning all four regulatory axes, with the most statistically significant single hit, H3K9me1K14ac, being a combinatorial mark not resolved in the original analysis (Fig. 1C). These axes: H3 K9–K14, H4K20, H3 K27–K36, and H4 acetylation, are described in detail below. NO Drives a Coordinated Histone Modification Signature Across Four Independent Axes. MDA-MB-231 cells were treated with NO (DETA/NO, 500 μM; ∼250 nM [NO]ss) for 24 or 48 h, with matched untreated controls and a 24 h washout condition (n = 3 biological replicates per condition). Histone PTMs were quantified by mass spectrometry, resolving 76 modification states including 30 combinatorial marks on adjacent residues. (A) Heatmap of the top 40 PTMs ranked by effect size (log2 fold-change, 24 h and 48 h NO vs. untreated). Repressive methylation marks increase while permissive marks decrease. (B) Temporal trajectories of key modifications across baseline (0 h), untreated controls (24 h, 48 h), NO-treated (24 h, 48 h), and washout conditions, demonstrating progressive and largely persistent changes. (C) Volcano plot (log2FC vs. -log10 p-value) confirming statistically significant changes across all four modification axes; the most significant single hit (H3K9me1K14ac) is a combinatorial mark not resolved in the original single-mark analysis [2]. (D) The H3 K9–K14 axis: H3K9me1K14ac decreases progressively (48 h: log2FC = −1.64, p less than 0.001), while higher K9 methylation states (K9me2/3K14ac) increase. (E) The H4K20 axis: H4K20me1 decreases (48 h: log2FC = −1.61) while H4K20me2 (+2.07) and H4K20me3 (+2.89) accumulate, indicating a coordinated me1-to-me2/me3 conversion cascade. (F) The H3 K27–K36 axis: adjacent modifications on the same tryptic peptide reveal coordinated state transitions. (G) The H4 acetylation axis: all multi-acetylated H4 species decrease with NO, with effect size scaling with acetylation number. All values from mass spectrometry quantification (n = 3 biological replicates per condition). Statistical testing by two-tailed Student t-test. The H3K9–K14 axis: K9me1K14ac→K9me3K14ac transition reveals selective chromatin remodeling The single strongest hit in the expanded dataset was the combinatorial mark H3K9me1K14ac (Fig. 1D), which decreased progressively with NO treatment (24h: log2FC = −0.95; 48h: log2FC = −1.64, p < 0.001; Table 2). This dual mark represents monomethylated K9 coexisting with acetylated K14 on the same histone H3 tail and is associated with transcriptionally permissive chromatin. Concurrently, K9me2/3K14ac increased, indicating that NO drives K9 methylation upward specifically on nucleosomes retaining K14 acetylation. This finding has mechanistic implications beyond what bulk single-mark measurements reveal. KDM3A, the primary H3K9me1/me2 2-ODD demethylase and established NO target [2], normally maintains low K9 methylation at active, acetylated chromatin. When NO inhibits KDM3A [5], relative methylation levels increase at these permissive loci as catalytic demethylation is inhibited. The result is K9 methylation drifts progressively upward, shifting from me1 to me2 and me3, even while K14 acetylation persists. This represents a mechanism for NO-driven transcriptional silencing specifically at previously active genes. The H4K20 axis: methylation state redistribution Although individual H4K20 methylation changes were noted in our prior report [3], the coordinated nature of the redistribution becomes apparent when the full methylation cascade is examined (Table 2; Fig. 1E). H4K20me1 decreased progressively (24h: log2FC = −0.70; 48h: log2FC = −1.61, NO-specific), while H4K20me2 (48h: log2FC = +2.07) and H4K20me3 (48h: log2FC = +2.89) accumulated, indicating accelerated conversion of the monomethyl substrate to higher methylation states. H4K20me2 serves as the binding platform for 53BP1 and is essential for DNA double-strand break repair pathway choice. Importantly, 53BP1 recognition of H4K20me2 requires both concurrent H2AK15 ubiquitination (via RNF168) and the absence of H4K16 acetylation [18,19]. The simultaneous accumulation of H4K20me2 and the loss of H4 multi-acetylation described below (including species bearing K16ac) would synergistically enhance 53BP1 binding, mechanistically linking two of the four PTM axes at the level of DNA damage response regulation. The dominant change here is the significant K20me1→me2 conversion (me1 82% → 28% and me2 18% → 71%, respectively), consistent with enhanced 53BP1 recruitment and altered repair pathway selection. Although K20me3 is a hallmark of constitutive heterochromatin and is also increased in fold-change terms, its absolute abundance remained below 1%, making functional relevance at this mark state uncertain. The mechanistic basis for the K20me1→me2 shift likely involves crosstalk with H3K9 methylation: H3K9me3 recruits HP1 proteins, which in turn recruit SUV4–20H1/H2 methyltransferases to convert H4K20me1 to me2/me3. Consistent with this model, SUV4–20H1 was the only histone methyltransferase upregulated in our gene expression data (+1.28-fold; Table 1; see enzyme expression analysis below and Fig. 5B). Coordinated Methyl-Modifying Enzyme Dysregulation and Pro-Tumorigenic Gene Silencing. Gene expression from Affymetrix microarray (24 h DETA/NO vs. untreated, MDA-MB-231 cells [2]). (A) 2-ODD histone demethylases (KDMs): 13 of 18 detected KDMs are upregulated, representing a coordinated compensatory transcriptional response to NO-mediated enzymatic inhibition across all four modification axes: KDM3 (K9), KDM4 (K9/K36), KDM5 (K4), KDM6 (K27), and KDM7 (K9/K27). (B) Histone methyltransferases (HMTs): cognate methyltransferases are predominantly downregulated, consistent with a homeostatic attempt to reduce active methylation. (C) Schematic illustrating the compensatory enzyme paradox: the cell detects the methylation imbalance and mounts the correct homeostatic response (KDMs up, KMTs down), but direct 2-ODD enzymatic inhibition dominates. (D) DNA repair gene suppression: 29 of 37 DNA repair genes with detectable changes are downregulated, spanning homologous recombination, Fanconi anemia, mismatch repair, nucleotide excision repair, base excision repair, PARP signaling, and checkpoint signaling. (E) Cancer hallmark pathways enriched among differentially expressed genes. (F) Key tumor-associated genes affected by NO-mediated chromatin remodeling, with expression fold-changes and pathway annotations. The K27–K36 axis: combinatorial landscape shift The most novel finding concerns the H3 K27–K36 modification landscape. To our knowledge, this represents one of the first systematic quantifications of K27–K36 combinatorial states in response to a signaling molecule. These two methylation sites, separated by nine residues on the same H3 tail (aa 27–40), reside on the same tryptic fragment and are resolved simultaneously by mass spectrometry. We observed a coordinated state transition (Table 2; Fig. 1F): the “ground state” K27me1K36me1 was significantly depleted by NO (24h: log2FC = −2.16, p = 0.029; 48h: log2FC = −3.08, p = 0.040), while K27me1K36me3 accumulated (24h: log2FC = +1.01, p < 0.001; 48h: log2FC = +0.91, p = 0.014). Total K36me1 showed the largest single-mark depletion among canonical H3 marks (48h: log2FC = −2.42, p = 0.006). This redistribution is functionally significant because the K36me1/me2 demethylases KDM2A/2B, and the K36me2/me3 demethylases KDM4A-C, are all 2-ODD enzymes, directly implicating them as NO targets beyond the established KDM3A paradigm. With demethylation blocked, K36 methylation accumulates, most strikingly as K27me1K36me3, which doubled with NO treatment (from 4.4% to 10.0% of H3 species). This combinatorial state is noteworthy: while K27me3 and K36me3 are well established as mutually exclusive marks maintained by PRC2-SETD2 antagonism [20,21], K27me1 is a distinct and more permissive modification that can co-exist with K36me3 at actively transcribed gene bodies. The K27me1K36me3 accumulation therefore likely reflects continued SETD2-mediated co-transcriptional K36 trimethylation at loci where K27 demethylation has stalled at the monomethyl state. K27me1K36me3 showed only 15% recovery upon NO washout (Table 2), the lowest recovery of any mark in the dataset, suggesting that once established, this combinatorial state is highly stable. The H4 acetylation axis: hyperacetylated species are preferentially depleted All multi-acetylated H4 species decreased with NO treatment, with the effect scaling with the number of acetyl groups (Table 2; Fig. 1G). Tri-acetylated species showed the largest changes: H4 K8acK12acK16ac (48h: log2FC = −3.02, p = 0.003), K5acK8acK16ac (48h: log2FC = −2.36), and K5acK12acK16ac (48h: log2FC = −3.25). This axis is particularly significant because, unlike histone methylation, which can be either gene activating (e.g., H3K4me3, H3K36me3) or repressive (e.g., H3K9me3, H3K27me3) depending on the residue and methylation state, histone acetylation is universally associated with transcriptional activation. The coordinated loss of every multi-acetylated H4 species therefore represents an unambiguous shift toward transcriptional silencing that cannot be explained by compensatory mechanisms. The preferential loss of these marks is consistent with, and likely secondary to, the repressive modification pattern driven by methylation changes along the K9, K20, and K27–K36 axes: as repressive methyl marks accumulate, HP1 and other heterochromatin factors are recruited, excluding histone acetyltransferases (HATs) from condensed chromatin. Notably, classical HDACs are not upregulated (most are in fact downregulated, Supplemental Table S1), suggesting that the acetylation loss reflects structural exclusion of acetyltransferases rather than active deacetylation. Chromatin compaction score: quantitative convergence across four axes To quantitatively assess whether the four PTM axes converge on a unified chromatin remodeling signature, we computed a composite Chromatin Compaction Score (CCS) integrating all histone PTMs with literature-based valence assignments (Fig. 2A; Supplemental Table S2). Each PTM was classified as compaction-promoting (+1; e.g., H3K9me2/me3, H4K20me2/me3, H3K27me3), compaction-opposing (−1; e.g., H3K9me1K14ac, multi-acetylated H4 species, H3K4me3), or ambiguous (0), based on established functional associations in the histone literature. Per-sample abundances were normalized and multiplied by the assigned valence to yield a single composite score. In this framework, loss of a permissive mark (valence −1) contributes positively to the CCS, functionally equivalent to gain of a repressive mark, such that coordinate depletion of multi-acetylated H4 species and accumulation of repressive methylation marks both drive the score in the same compaction-promoting direction. The CCS reveals a progressive and statistically significant shift toward chromatin compaction with NO treatment (Fig. 2A). Baseline (0h) and untreated conditions cluster near zero, while NO-treated conditions show dose-dependent increases: 24h NO vs. 24h NT (p = 0.034), 48h NO vs. 48h NT (p = 0.004), and washout vs. 48h NT (p = 0.031). The persistence of an elevated CCS in the washout condition is consistent with the mark-level observations of incomplete reversibility described above. Decomposition into axis-specific sub-scores confirmed that the H4K20 and K27–K36 axes contribute most strongly to the compaction signature at 48h, while the H4 acetylation axis shows the most dramatic NO-dependent shift (Fig. 2B). PCA validation of compaction score To validate the literature-based valence assignments independently, we performed principal component analysis (PCA) on the full 76-PTM matrix across all 18 biological replicates (6 conditions × 3 replicates). PCA is an unsupervised dimensionality reduction method that identifies the axes of greatest variance in a dataset without any prior knowledge of sample groupings or PTM function. PC1 captured 40.7% of total variance and separated NO-treated from untreated conditions along a single axis (Fig. 2C), indicating that NO exposure is the dominant source of variation in the histone modification landscape. PC2 captured an additional 15.3% of variance and distinguished the time-dependent progression of NO effects, with 48 h NO samples separating further from baseline than 24 h NO samples. Critically, the PC1 loadings, which represent each PTM's contribution to this dominant axis and are entirely data-driven, requiring no prior valence assumptions, showed strong concordance with the literature-based valence assignments used to compute the CCS (Spearman ρ = −0.56, p < 0.00001; Fig. 2D). The negative sign reflects an arbitrary axis orientation: PTMs assigned repressive valence (+1) load negatively on PC1, and vice versa. In practical terms, this means that the PTMs most responsible for separating NO-treated from untreated samples are the same PTMs that the published literature classifies as chromatin-compacting. This concordance demonstrates that the dominant axis of variation in the histone PTM data aligns with the repressive-versus-permissive distinction from the published literature, providing unsupervised, orthogonal validation of the CCS framework and its underlying valence assignments. Sensitivity analysis Because 10 of the 76 PTM states were classified as ambiguous (valence = 0), including H3K36me2, K27me1K36me2, K27me1K36me3, K9me2K14ac, and several minor species, we tested whether the CCS is sensitive to these assignments by recomputing the score under three scenarios: all ambiguous marks reassigned as repressive (+1), all reassigned as permissive (−1), or all excluded (primary analysis). The resulting CCS trajectories were nearly superimposable (Fig. 2E), with all three scenarios producing the same qualitative pattern of progressive compaction with NO treatment and persistence through washout. The maximum divergence between any two scenarios was <10% of the total CCS range at any condition, and the NO-treated vs. untreated separation was preserved in every case. This confirms that the compaction signature is driven by the well-characterized repressive and permissive marks, whose valence assignments are unambiguous, rather than by the subset of marks with context-dependent functions (numerical values in Supplemental Table S2). This robustness reflects the fact that the compaction signature is overdetermined: each of the four major axes is internally consistent and strongly directional, such that reclassifying a small subset of ambiguous marks (∼10% of total) cannot overcome the dominant convergent signal from the well-characterized repressive and permissive marks. A schematic overview of how these four axes converge on a unified compaction signature is presented in Fig. 2F. Baseline Chromatin State and Expression Response Predict Locus-specific K9 mark redistribution The mass spectrometry analyses described above measure bulk (global) histone modification levels and therefore cannot assign PTM changes to specific genomic loci. To determine how the K9 methylation accumulation and acetylation loss detected by mass spectrometry are distributed across the genome, we reanalyzed the H3K9ac and H3K9me2 ChIP-seq data from the original study [3] using the nf-core/chipseq pipeline (see Methods). Initial characterization confirmed that K9me2 accumulation is predominantly intergenic rather than promoter-localized: despite a 69% global increase by mass spectrometry, the median promoter log2FC was +0.005 (Supplemental Fig. S2A–B). K9me2 gene body signal showed a modest but significant inverse correlation with gene expression (Spearman ρ = −0.114, p = 6 × 10−55; Supplemental Fig. S2C). To dissect the locus-specific redistribution of both marks in greater detail, we performed systematic decile analyses stratifying genes by their baseline chromatin state and by their expression response to NO. Baseline chromatin state predicts K9ac redistribution. To characterize how pre-existing chromatin state influences NO-driven K9ac changes, we ranked all genes by their untreated K9ac signal level and divided them into ten equal-sized groups (deciles), where Decile 1 contains the 10% of genes with the lowest K9ac (transcriptionally silent, minimal promoter acetylation) and Decile 10 contains the 10% with the highest K9ac (actively transcribed, strong promoter acetylation). This stratification revealed a striking monotonic relationship: K9ac fold-change scaled almost perfectly with baseline signal level (Spearman ρ = 0.915, p < 0.001; Fig. 3A). Genes in the lowest baseline decile showed the most dramatic K9ac loss in response to NO, while genes in the highest decile were largely protected, consistent with the activity-dependent pattern observed when genes were stratified by expression response (Supplemental Fig. S2D). The near-perfect rank correlation demonstrates that baseline chromatin state, rather than gene identity per se, is the primary determinant of K9ac redistribution. Baseline Chromatin State and Expression Response Predict Locus-Specific K9 Mark Redistribution. H3K9ac and H3K9me2 ChIP-seq data were processed using nf-core/chipseq (v2.1.0; see Methods). K9ac uses the R1 replicate pair due to UT replicate signal imbalance (see Methods and Supplemental Fig. S1); K9me2 uses replicate means. Genes were ranked by the indicated parameter and divided into ten equal-sized groups (deciles). (A) K9ac fold-change (log2 NO/UT) by baseline K9ac signal decile. Genes with the lowest baseline K9ac (silent loci) show the greatest K9ac loss, while genes with the highest baseline K9ac (active loci) are relatively protected (Spearman ρ = 0.915, p < 0.001). (B) K9me2 fold-change by baseline K9ac signal decile. K9me2 gain is concentrated at loci with the lowest baseline K9ac, forming an anti-correlated pattern with K9ac loss (ρ = −0.964, p < 0.001). (C) K9ac fold-change by expression response decile (log2 FC NO/UT microarray). K9ac loss is widespread but shows no significant monotonic relationship with expression change (ρ = 0.212, n.s.). (D) K9me2 fold-change by expression response decile. K9me2 gain is preferentially concentrated at the most downregulated genes (ρ = −0.467, p < 0.05; D1–D2 vs. D5–D6 Mann–Whitney p = 5.03 × 10−37), demonstrating that K9me2 accumulation preferentially targets expression-responsive loci. Reciprocal K9me2 gain mirrors K9ac loss across baseline deciles. If the NO-driven compaction signature reflects coordinated chromatin remodeling, K9me2 changes should inversely mirror K9ac redistribution at the same loci. Strikingly, when genes were stratified by the same baseline K9ac deciles used above, K9me2 fold-change showed an almost perfectly anti-correlated pattern (Spearman ρ = −0.964, p < 0.001; Fig. 3B): loci that lost the most K9ac (low-baseline deciles) gained the most K9me2, while loci that retained K9ac (high-baseline deciles) showed minimal K9me2 change. This reciprocal relationship provides direct evidence for coordinated mark switching at individual genomic loci, the deacetylation-then-methylation sequence predicted by the enzymatic model, in which NO-mediated KDM inhibition allows K9me2 to accumulate specifically where K9ac has been removed. Expression response reveals preferential K9me2 targeting of downregulated genes. To determine whether the coordinated mark redistribution has functional consequences for gene expression, we next stratified genes by their expression response to NO and examined both marks across expression deciles. K9ac fold-change showed a positive but modest correlation with expression response (Spearman ρ = 0.212, n.s.; Fig. 3C), consistent with the known association between promoter acetylation and transcriptional activity. In contrast, K9me2 fold-change showed a significant inverse correlation with expression response (ρ = −0.467, p < 0.05; Fig. 3D): genes in the most downregulated deciles (D1–D2) gained substantially more K9me2 than genes in the moderately changed deciles (D5–D6; Mann–Whitney p = 5.03 × 10−37). This preferential K9me2 accumulation at downregulated loci indicates that the repressive mark is not deposited randomly but is targeted to genes undergoing transcriptional silencing. Dual-mark override: K9me2 gain at loci retaining K9ac. A critical feature of the decile analyses is the dissociation between K9ac retention and K9me2 gain at expression-responsive loci. Among the 1902 downregulated genes that retained detectable K9ac, the median K9me2 fold-change was +0.052, indicating that these loci simultaneously carry both the permissive mark (K9ac) and the newly deposited repressive mark (K9me2). This dual-mark state suggests an override mechanism in which K9me2 gain is sufficient to silence gene expression even at loci where K9ac has not been fully removed, consistent with a model in which the repressive function of K9me2 dominates the permissive signal of K9ac at co-marked promoters. Reconciliation of bulk and locus-specific measurements. Together, the decile analyses reconcile the bulk mass spectrometry findings with locus-specific chromatin remodeling. The global K9ac loss detected by mass spectrometry (−78%) reflects widespread deacetylation concentrated at low-signal, transcriptionally silent loci, while actively transcribed genes are protected and NO-responsive genes actively gain K9ac (Supplemental Fig. S2E). The global K9me2 gain (+69%) reflects predominantly intergenic accumulation, but decile analysis reveals a functionally significant locus-specific component: K9me2 is preferentially deposited at loci losing K9ac and at genes undergoing transcriptional repression. The near-perfect anti-correlation between K9ac loss and K9me2 gain across baseline deciles (Fig. 3A and B), combined with preferential K9me2 targeting of downregulated genes (Fig. 3D), establishes that NO drives a coordinated, locus-specific chromatin switch from permissive to repressive K9 marks, with the resulting dual-mark state at expression-responsive loci providing a mechanistic basis for gene silencing. Representative Browser Track Patterns Illustrate Five Categories of Locus-specific K9 remodeling To illustrate the locus-specific K9 modification changes described above, we examined H3K9ac and H3K9me2 ChIP-seq browser tracks at representative gene loci, identifying five distinct categories of K9 remodeling (Fig. 4). These categories integrate the promoter-level quantification presented in Fig. 3 with visual inspection of signal distribution across entire gene loci. Category A: K9ac gain at NO-activated genes. Genes transcriptionally upregulated by NO, including HMOX1 (expression log2FC = +1.6), NDRG1 (+2.1), ADM (+2.4), and LOX (+2.8), showed clear K9ac signal gain at their promoters following NO treatment (Fig. 4A). These loci exemplify the activity-dependent K9ac redistribution pattern quantified in Supplemental Fig. S2D: actively transcribed genes not only resist the global K9ac depletion but gain acetylation, consistent with transcription factor-driven HAT recruitment overriding the heterochromatic compaction program. Category B: K9ac loss at NO-silenced genes. Genes downregulated by NO, such as LUM (expression log2FC = −3.2) and FGFBP1 (−1.7), displayed K9ac signal loss at their promoters without corresponding K9me2 gain (Fig. 4B). Unlike the promoter-focused K9ac changes in Category A, the K9ac depletion at these loci extends broadly across the gene body and flanking regions, consistent with a domain-wide shift in chromatin environment. This pattern is consistent with the predominant K9ac loss observed at transcriptionally inactive promoters (Supplemental Fig. S2D; Fig. 3A). Category C: Dual compaction, K9ac loss with concurrent K9me2 gain. A subset of downregulated genes, including SEMA3D (expression log2FC = −1.7), HCP5 (−1.6), and SPINK4 (−1.2), showed simultaneous K9ac loss and K9me2 accumulation at their loci (Fig. 4C). This dual mark switch from permissive to repressive chromatin at the same locus represents the most direct evidence of chromatin compaction at individual gene promoters and is consistent with the convergent mass spectrometry findings across the K9 axis. Category D: K9me2 loss at 2-ODD target gene. EGLN2 (PHD1), itself a 2-ODD enzyme transcriptionally upregulated by NO (expression log2FC = +0.4), showed K9me2 loss at its locus (Fig. 4D). This pattern is noteworthy because EGLN2 encodes prolyl hydroxylase domain protein 1, a member of the same 2-ODD enzyme family inhibited by NO. The K9me2 loss at this locus is consistent with chromatin opening to facilitate the compensatory upregulation of 2-ODD enzymes described below. Category E: K9me2 gain at NO-silenced genes. Genes such as CLIC3 (expression log2FC = −1.7) and HSD17B8 (−1.6) showed K9me2 accumulation at their promoters (Fig. 4E), representing locus-specific instances of the predominantly intergenic K9me2 gain detected by mass spectrometry (Supplemental Fig. S2A–B). Although the genome-wide median promoter K9me2 change was negligible, these individual loci demonstrate that promoter-level K9me2 accumulation does occur at specific genes and is associated with transcriptional silencing. Together, these five categories demonstrate that the bulk K9 modification changes detected by mass spectrometry manifest as diverse but coherent patterns of locus-specific chromatin remodeling. The predominant pattern is K9ac redistribution (Categories A and B), consistent with the quantitative analysis in Fig. 3, while a subset of loci show direct evidence of K9me2 changes at gene regulatory regions (Categories C, D, and E). Coordinated compensatory response: 2-ODD demethylases up, methyltransferases down Across all four axes (Fig. 1), the combinatorial modifications reveal a consistent directionality. To assess this, we examined all 25 adjacent-residue modifications quantified across the K9–K14 (4 marks), K27–K36 (11 marks on canonical H3 and H3.3), and H4 multi-acetylation (10 marks) axes (Table 2). Of these, 14 decreased and 8 increased at 48h NO (3 unchanged). The 14 marks that decreased were uniformly permissive chromatin species: all 10 H4 multi-acetylated species, K9me1K14ac, and 3 low-methylation K27–K36 ground states (e.g., K27me1K36me1). The 8 marks that increased were uniformly repressive, high-methylation species: K9me2K14ac, K9me3K14ac, and 6 higher-methylation K27–K36 combinations (e.g., K27me1K36me3, K27me3K36me1). Because these are measured on the same tryptic peptide, these transitions are occurring on the same histone tail rather than reflecting population averages, providing direct evidence of coordinated chromatin state transitions at individual nucleosomes. Next, we examined the gene expression microarray data from Ref. [3] (GEO: GSE66324) with focused attention on methyl-modifying enzymes which revealed a striking coordinated pattern not previously described (Table 1, Fig. 5). 2-ODD demethylase upregulation. Among 2-ODD histone demethylases, 13 of 18 detected were upregulated (Fig. 5A), including KDM3A (+3.33-fold, p = 0.0005), JMJD1C (+2.03-fold, p = 0.015), KDM5B (+1.60-fold, p = 0.003), KDM5C (+1.37-fold, p = 0.001), KDM6A (+1.31-fold, p = 0.003), KDM2A (+1.30-fold, p = 0.012), and KDM2B (+1.25-fold, p = 0.082). This is consistent with another of our studies where we measured (RT-PCR) NO-mediated increases in 8 KDMs [2,22]. This near-universal upregulation spans KDM families targeting H3K4 (KDM5), H3K9 (KDM3), H3K27 (KDM6), and H3K36 (KDM2/KDM4), consistent with the cell sensing NO inhibition across multiple 2-ODD family members and mounting a broad compensatory transcriptional response. Methyltransferase downregulation (Fig. 5B). Conversely, cognate methyltransferases were broadly downregulated. All four detectable H3K9 methyltransferases decreased: EHMT2/G9a (−1.68-fold, p = 0.095), EHMT1/GLP (−1.58-fold, p = 0.022), SUV39H1 (−1.31-fold, p = 0.002), and SETDB1 (−1.52-fold, p = 0.046). NSD2/WHSC1, a major H3K36me2 writer, was the most dramatically altered chromatin modifier (−4.07-fold, p = 0.0001). This coordinated downregulation represents the second arm of the compensatory response: reducing active methylation. H4K20 axis validation. SUV4–20H1, the primary H4K20me2 writer, was upregulated (+1.28-fold, p = 0.004), consistent with active promotion of the me1→me2 conversion observed at the protein level. KMT5A/SET8, the H4K20me1 writer, was modestly downregulated (−1.18-fold, p = 0.016), consistent with me1 depletion. Enzymatic inhibition dominates over transcriptional compensation. The critical interpretive point is that despite this coordinated compensatory effort, demethylases upregulated to restore demethylation capacity, methyltransferases downregulated to reduce methylation input-the histone PTM data show that methylation nonetheless accumulates progressively from 0 to 24 to 48 h. This suggests that direct enzymatic inhibition of 2-ODD demethylases by NO [5,9] overrides the transcriptional compensation. The cell detects the problem and mounts the correct homeostatic response but cannot overcome the biochemical blockade. Alternative explanations for the compensatory pattern should be noted: KDM upregulation could reflect a direct transcriptional consequence of altered histone marks at KDM gene promoters rather than a sensor-mediated response, and KMT downregulation could reflect feedback from accumulated methyl marks on their own loci. Regardless of the upstream sensing mechanism, the functional outcome, failed compensation, remains the same. The resulting paradox, in which the cell mounts the correct homeostatic response but cannot overcome the biochemical blockade, is illustrated in Fig. 5C. Functional consequences: DNA repair suppression and a pro-tumorigenic gene expression signature To assess whether the histone modification signature consistent with chromatin compaction is associated with functionally coherent gene expression changes, we examined the microarray data for curated gene sets representing cancer hallmarks. The most striking finding was broad suppression of the DNA damage repair machinery. DNA repair gene suppression. Of 37 DNA repair genes with detectable expression changes (|FC|>1.2), 29 were downregulated and only 8 upregulated (Fig. 5D). The downregulated genes span every major repair pathway: homologous recombination (RAD51, RAD51C, RAD51L1/L3, BRCA1, MRE11A, NBN, RAD50), Fanconi anemia (FANCL, FANCI, FANCC, FANCD2, FANCF, FANCG; six of the thirteen FA complementation group genes), mismatch repair (MSH2, MSH6), nucleotide excision repair (ERCC2, XPA, ERCC6), base excision repair (XRCC1, APEX1, OGG1, MUTYH), PARP signaling (PARP1, PARP4, PARP9), and DNA damage checkpoint signaling (ATR, CHEK1, CHEK2). This coordinated suppression across seven repair pathways is consistent with global chromatin compaction restricting transcriptional access to repair gene loci. The ChIP-seq data presented above provides initial locus-specific context: the global K9me2 accumulation at intergenic regions and widespread K9ac loss at silent promoters are consistent with a chromatin environment that restricts access to repair gene loci, although direct validation of K9 modification changes at individual DNA repair gene promoters would strengthen this interpretation. Cancer hallmark bias. Extending this analysis across additional cancer-relevant gene sets revealed a systematic pro-tumorigenic bias in the expression changes (Fig. 5E). DNA repair genes showed the strongest skew (78% down), followed by angiogenesis inhibitors (78% down) and chromatin remodeling complexes (74% down). Among chromatin remodelers, 14 of 19 SWI/SNF and chromodomain helicase genes were downregulated (Supplemental Table S1), including SMARCA2 (−2.32-fold, p = 0.0017), SMARCA4 (−1.27-fold), and ARID1A (−1.13-fold), representing frequently mutated tumor suppressors across multiple human cancers. EMT and invasion markers. CDH1/E-cadherin (−2.32-fold), the master gatekeeper of epithelial identity, was significantly downregulated, while MMP1 (+41.4-fold) and MMP3 (+5.86-fold) were among the most dramatically upregulated genes in the entire dataset. The anti-invasion factors TIMP2 (−1.44-fold), TIMP3 (−1.95-fold), and TIMP4 (−1.65-fold) were all suppressed, and thrombospondin-1 (THBS1; −2.04-fold), a major endogenous anti-angiogenic factor, was downregulated (Fig. 5F). This pattern is consistent with NO-mediated activation of invasion and metastatic potential. IDH1 downregulation: a feed-forward loop (Fig. 5F). Isocitrate dehydrogenase 1 (IDH1) was downregulated (−2.81-fold, p = 0.002). IDH1 produces 2-oxoglutarate (α-ketoglutarate), the essential co-substrate for 2-ODD demethylases. Reduced IDH1 expression could diminish intracellular 2-OG availability, thereby compounding the direct enzymatic inhibition of 2-ODD demethylases by NO and creating a feed-forward loop that reinforces the chromatin compaction signature. It should be noted that IDH1 is one of three isocitrate dehydrogenases (IDH1, IDH2, IDH3), and the TCA cycle continuously produces 2-OG through IDH3 and via glutaminolysis. Whether IDH1 downregulation alone significantly reduces total cellular 2-OG would require direct metabolite measurements; however, the cytosolic compartmentalization of IDH1, in contrast to the mitochondrial IDH2 and IDH3, suggests that nuclear 2-OG availability for chromatin-modifying enzymes may be preferentially affected. Discussion By analyzing the histone mass spectrometry, ChIP-seq, and gene expression data [3] with attention to adjacent PTM states, genomic distribution of K9 modifications, and methyl-modifying enzyme expression as coordinated systems, we reveal a coherent histone modification signature consistent with chromatin compaction driven by broad 2-ODD demethylase inhibition that was not apparent from the original single-mark analysis. This study uses the same datasets as Vasudevan et al. [3], which established NO as a regulator of 38 histone PTMs and H3K9me2/K9ac redistribution in MDA-MB-231 cells. The present work asks a fundamentally different question: what do adjacent PTM patterns reveal about the integrated mechanism of chromatin remodeling, and how do bulk-level modifications map to specific genomic compartments? Novel PTM targets: H4K20 and K27–K36 adjacent states The H4K20 methylation redistribution emerges as a coherent cascade in the present analysis. The K27–K36 adjacent modification landscape represents a unique strength of mass spectrometry-based histone profiling: these co-occurring modifications on the same peptide cannot be resolved by Western blot or ChIP-seq. The shift from K27me1K36me1 to K27me1K36me3 indicates K36 trimethylation occurring specifically at K27me1-marked active gene bodies. The established K36me3–PRC2 antagonism means that NO effectively reinforces active gene body identity at these loci. The persistence of K27me1K36me3 upon washout (15% recovery; Table 2) suggests self-reinforcing stabilization through K36me3-dependent DNMT3B recruitment [20,21]. The integration of ChIP-seq data with bulk mass spectrometry measurements strengthens the chromatin compaction interpretation. The 69% global K9me2 increase detected by mass spectrometry does not manifest at gene promoters (median promoter log2FC = +0.005), indicating that K9me2 accumulation is concentrated at intergenic and heterochromatic regions where demethylase-dependent turnover is highest. Concurrently, K9ac undergoes activity-dependent redistribution with global loss concentrated at silent promoters and protection at active loci, explaining how 78% bulk acetylation loss coexists with maintained acetylation at NO-responsive genes (ADM, LOX, HMOX1, VEGFA). The modest but highly significant K9me2 gene body correlation with expression (Spearman ρ = −0.114, p = 6 × 10−55) confirms functional relevance at individual loci. Coordinated regulation of methyl-modifying enzymes A central finding is the cell's coordinated but ultimately insufficient attempt to compensate for NO-mediated 2-ODD inhibition. The simultaneous upregulation of 2-ODD demethylases and downregulation of methyltransferases represents a homeostatic response involving both arms of the methyl-modifying enzyme network; that methylation nonetheless accumulates demonstrates the dominance of direct enzymatic inhibition. Notably, 2-ODD enzymes exhibit markedly different sensitivities to NO [4], and differential KDM sensitivity may contribute to the axis-specific kinetics observed here, such as rapid K36me1 depletion versus more gradual K9me2 accumulation [2,23,24]. Pro-Tumorigenic Consequences: from chromatin compaction to functional gene silencing The observation that NO-driven chromatin remodeling preferentially silences DNA repair genes and tumor suppressors provides a mechanistic link between NO signaling and its established pro-tumorigenic effects [2,3,25]. The suppression of six Fanconi anemia genes, the core HR machinery (RAD51, BRCA1, MRE11A, NBN), and mismatch repair (MSH2, MSH6) would collectively impair the cell's ability to maintain genomic integrity [26,27]. In the context of MDA-MB-231 cells, which already harbor TP53 and BRCA1 mutations and exhibit a mesenchymal phenotype, further suppression of these repair pathways may exacerbate genomic instability [28,29]. Whether these effects would be more dramatic in less-transformed cell types with intact DNA repair capacity is an important question for future investigation, as MDA-MB-231 cells represent a highly aggressive baseline from which further suppression may have diminishing functional consequences. The coordinated downregulation of SWI/SNF chromatin remodeling complexes is particularly significant [30,31]. SMARCA2, SMARCA4, and ARID1A are among the most frequently mutated genes across all human cancers, and their loss promotes tumor development by preventing chromatin accessibility at tumor suppressor loci [32]. NO-mediated transcriptional suppression of these remodelers would compound the repressive histone modification signature, creating a self-reinforcing chromatin state [33,34]. While some histone methylation changes may be activating ([2]), histone acetylation is unambiguously associated with activation regardlessof the residue or histone involved [35,36]. The coordinated loss of every multi-acetylated H4 species, with tri-acetylated marks showing the largest decreases (up to 8-fold), establishes that the cell is unambiguously shifting toward a transcriptionally silent chromatin state. Combined with the DNA repair gene suppression and SWI/SNF silencing described above, these data demonstrate that NO-driven chromatin remodeling has direct functional consequences for gene expression [37,38]. Indeed, the compensatory pattern extends to deacetylases: most classical zinc-dependent HDACs are downregulated (8 of 11), mirroring the KMT downregulation, while the acetylation loss persists, further evidence that HAT exclusion from condensed chromatin, rather than HDAC activity, drives the deacetylation (Supplemental Table S1). This conclusion is further supported by the observation that NO is known to S-nitrosylate HDAC2 at cysteine residues, directly inhibiting its deacetylase activity [6]. Similarly, S-nitrosation of HDAC6 and HDAC8 has been reported to suppress their deacetylase activity [39,40]. However, we cannot exclude the possibility that NO-mediated S-nitrosation modulates HDAC activity post-translationally, which could contribute to the acetylation changes observed here independently of the 2-ODD pathway. The IDH1 downregulation (−2.81-fold) is mechanistically important because IDH1 produces 2-oxoglutarate, the essential co-substrate for all 2-ODD enzymes [31]. Reduced IDH1 expression would diminish 2-OG availability, compounding the direct NO-mediated KDM inhibition with reduced substrate availability [41]. This creates a feed-forward loop: NO inhibits 2-ODD enzymes directly → repressive chromatin changes reduce IDH1 transcription → less 2-OG available → further impaired demethylation. Notably, IDH1 gain-of-function mutations that consume 2-OG (producing the oncometabolite 2-hydroxyglutarate) are a well-established driver of epigenetic dysregulation in glioma and AML [42]. Washout experiments in which the NO source was removed after 24 h reveal that the histone modification signature consistent with chromatin compaction is largely persistent (Table 2). Of 16 key modifications examined, only H3K9me2 showed full reversibility (105% recovery), while most marks, including K20me1/me2, K27me1K36me3, K36me1, and all multi-acetylated H4 species, showed less than 30% recovery. Most strikingly, H3K9me3 continued to accumulate after NO removal (−1% recovery, i.e., worsened), consistent with the self-reinforcing nature of heterochromatic H3K9me3 through HP1-SUV39H1 positive feedback loops [23,24]. Connections to mRNA and DNA methylation We recently demonstrated that NO inhibits the mRNA demethylase FTO, leading to altered N6-methyladenosine (m6A) patterns [4]. The gene expression data in the present study show coordinated changes in m6A pathway components (METTL3, ALKBH5, FTO, YTHDC1), consistent with NO simultaneously regulating both histone and RNA methylation landscapes through 2-ODD inhibition. Additionally, our laboratory demonstrated that NO inhibits the 2-ODD family of DNA demethylases (TET1,2,3 and ALKBH2), leading to genome-wide accumulation of 5-methylcytosine (5 mC) and 5-hydroxymethylcytosine (5hmC) [5]. TET inhibition would compound the histone-level changes described here by stabilizing DNA methylation at CpG islands and gene regulatory regions, reinforcing transcriptional silencing through a parallel epigenetic axis [5,33]. The convergence of histone hypermethylation, DNA hypermethylation, and altered m6A-mRNA patterns through 2-ODD inhibition via DNIC formation [5]. Clinical context These findings are relevant to clinical scenarios in which tumor-associated NO production is elevated. Inducible nitric oxide synthase (iNOS/NOS2) is highly expressed by tumor-associated macrophages and by the tumor cells themselves in several cancer types, including triple-negative breast cancer [22,43]. MDA-MB-231 cells [3], are triple-negative (ER−/PR−/HER2−), and TP53-mutant, making them a clinically relevant model for aggressive breast cancer. The chromatin remodeling program identified here provides a mechanistic basis for the established association between high tumor NOS2 expression and poor clinical outcomes. Integrated Model We propose that NO generates a histone modification signature consistent with chromatin compaction through a hierarchical mechanism (Fig. 6). NO forms DNICs at 2-ODD catalytic iron centers, broadly inhibiting multiple KDM families while sequestering cellular labile iron (step 1–2) [1]. With demethylation blocked, a coordinated shift in 76 histone modification states emerges (step 3; Fig. 1), organized into four convergent compaction axes, H3K9me2/K14ac remodeling, the H4K20me1→me2→me3 cascade, H3K27me1K36me3 combinatorial landscape shift, and H4 hyperacetylation depletion—quantified by a composite chromatin compaction score and validated by PCA (step 4; Fig. 1D–G, 2). ChIP-seq analysis reveals that the K9 signature manifests in a locus-specific manner: K9me2 accumulates predominantly at intergenic and heterochromatic regions (+69% global, negligible promoter change; step 5a; Supplemental Fig. S2A–C), while K9ac undergoes baseline-dependent redistribution in which genes with the lowest initial acetylation suffer the greatest loss, K9me2 accumulates reciprocally at these silent loci, and the lowest-expression deciles show K9me2 gain that overrides residual K9ac (step 5b; Fig. 3A–D, Supplemental Fig. S2D–E). Browser track inspection validates five distinct categories of locus-specific K9 remodeling (step 6; Fig. 4). Despite compensatory upregulation of 2-ODD demethylases and downregulation of methyltransferases, enzymatic inhibition by NO dominates, maintaining the repressive mark imbalance (step 7; Fig. 5A–C). The resulting chromatin compaction suppresses DNA repair genes (29 of 37 pathway members), SWI/SNF remodelers, and tumor suppressors, while IDH1 downregulation reduces 2-OG availability, creating a feed-forward loop that compounds enzymatic inhibition (step 8; Fig. 5D–F) [41,44]. Positive feedback through HP1–SUV39H1 recruitment may convert reversible K9me2 into stable K9me3 heterochromatin; note that SUV39H1 transcript is modestly downregulated (−1.31-fold) at 24 h, suggesting that residual protein-level activity, rather than new synthesis, sustains this feedback, while H3K36me3 accumulation and NO-mediated TET inhibition could further stabilize this epigenetic state through DNA methylation [5,45]. Integrated Model: NO Generates a Hierarchical Histone Modification Signature Consistent with Chromatin Compaction and Pro-Tumorigenic Consequences. Schematic illustrating the proposed mechanism: (1) NO exposure via DETA/NO (500 μM, 24 h) generates sustained NO flux; (2) NO forms dinitrosyl iron complexes (DNICs) at 2-oxoglutarate-dependent dioxygenase (2-ODD) catalytic iron centers, broadly inhibiting multiple KDM families; (3) a coordinated histone PTM signature emerges across 76 modification states (Fig. 1); (4) four independent compaction axes converge—H3K9me2/K14ac remodeling, H4K20 conversion cascade, H3K27me1K36me3 combinatorial landscape shift, and H4 hyperacetylation depletion (Fig. 1D–G), quantified by a composite chromatin compaction score and validated by PCA (Fig. 2); (5a) K9me2 spreads selectively at intergenic and heterochromatic regions (+69% global, minimal promoter change; Supplemental Fig. S2A–C); (5b) K9ac undergoes baseline-dependent redistribution: monotonic loss from silent to active loci (ρ = 0.915), reciprocal K9me2 accumulation at silent loci (ρ = −0.964), and dual-mark override at the lowest-expression deciles despite K9ac retention (Fig. 3A–D; Supplemental Fig. S2D–E); (6) browser track inspection validates five categories of locus-specific K9 remodeling (Fig. 4); (7) methyl-modifying enzyme dysregulation: KDMs are upregulated (KDM3A, 4A/B, 7A/B) while KMTs are downregulated (EHMT2, SETDB1/2), representing an insufficient compensatory response (Fig. 5A–C); (8) the resulting chromatin compaction drives tumor-permissive gene silencing, including DNA repair suppression (29/37 pathway genes), SWI/SNF remodeler loss, and an IDH1 feed-forward loop that reduces 2-OG availability and compounds enzymatic inhibition (Fig. 5D–F). Positive feedback through HP1–SUV39H1 recruitment converts reversible K9me2 into stable K9me3 heterochromatin; note that SUV39H1 transcript is downregulated (−1.31-fold at 24 h), consistent with residual protein-level activity sustaining this feedback. Importantly, DNIC formation at 2-ODD iron centers also depletes the broader cellular labile iron pool [2], which may independently reinforce the chromatin remodeling program through iron deficiency signaling. Consistent with this, NDRG1, an iron-regulated metastasis suppressor whose NO-dependent upregulation we previously linked to chelatable iron depletion and DNIC formation [10], shows K9ac gain at its promoter in the present ChIP-seq data (Fig. 4A), indicating convergent regulation through both iron-dependent and chromatin-level mechanisms. The relationship between K9ac loss and K9me2 gain at individual loci raises the question of whether deacetylation alone is sufficient for gene silencing, or whether additional methylation is required. Our data suggest that these modifications serve complementary but mechanistically distinct functions. K9ac loss neutralizes the positive charge on the lysine side chain, reducing electrostatic repulsion between the histone tail and DNA and permitting tighter nucleosome-DNA association. However, this represents a passive, readily reversible state. K9me2, by contrast, serves as a binding platform for HP1 (heterochromatin protein 1), which recruits the methyltransferase SUV39H1 to convert K9me2 to K9me3, establishing a self-reinforcing positive feedback loop that propagates stable heterochromatin. Consistent with this model, K9me2 accumulation shows a stronger correlation with gene expression changes (Spearman ρ = −0.467, p < 0.05) than K9ac loss alone, and the persistence of K9me2 through the washout condition suggests that the HP1–SUV39H1 feedback loop sustains silencing even after the initial NO stimulus is removed. Limitations Several limitations should be acknowledged. First, the histone PTM data represent global (bulk) measurements; however, the ChIP-seq reanalysis presented here provides initial locus-specific context by demonstrating that K9me2 accumulation is predominantly intergenic while K9ac redistribution is activity-dependent, partially bridging the gap between bulk and locus-specific measurements. Validation of the full compaction signature at individual loci by chromatin accessibility assays (ATAC-seq or MNase-seq) would further strengthen this interpretation. Second, beyond KDM3A we have not directly demonstrated NO inhibition of KDM2, KDM4, KDM5, or KDM6 family enzymes; this is inferred from their conserved catalytic mechanism, convergence of PTM accumulation, and compensatory transcriptional upregulation, and browser track visualization at representative loci confirms five distinct categories of K9 remodeling consistent with the quantitative analysis (Fig. 4). In vitro activity assays with recombinant KDMs would provide direct validation. Third, the gene expression data were obtained at a single timepoint (24 h). Fourth, while we observe coordinated changes consistent with a unified mechanism, we cannot exclude well-known NO-dependent pathways independent of 2-ODD inhibition, including sGC activation and S-nitrosation of chromatin-associated proteins (e.g., HDAC2 [6]). Fifth, the correlation between the histone modification signature and gene expression changes, while functionally coherent and now supported by locus-specific ChIP-seq data, remains correlative at the individual gene level; additional locus-specific studies are needed to establish direct causal links between specific PTM changes and individual gene silencing event. Lastly, although this analysis was conducted on data sets from one cell type treated with NO (MDA-MB-231), we have tested over 10 different diverse cancer cell types and found that in all cases NO increases global histone methylation and always decreases global histone acetylation [2,3], suggesting that this is a conserved mechanism. Conclusions By applying adjacent histone modification analysis, systematic methyl-modifying enzyme expression profiling, and ChIP-seq reanalysis, we identify NO-mediated 2-ODD inhibition as a broad-spectrum mechanism that generates a coordinated histone modification signature consistent with chromatin compaction across four independent modification axes [2]. A composite Chromatin Compaction Score integrating all PTMs with literature-based valence assignments quantitatively confirms the progressive shift toward repressive chromatin with NO treatment, validated by principal component analysis. Analysis of H3K9ac and H3K9me2 ChIP-seq data reveals that the bulk compaction signature manifests as locus-specific redistribution: K9me2 accumulation is concentrated at intergenic and heterochromatic regions while K9ac shows activity-dependent redistribution with global loss at silent promoters and protection of transcriptionally active loci. The resulting repressive chromatin state is associated with silenced DNA repair genes, SWI/SNF chromatin remodelers, and established tumor suppressors, while browser track analysis at representative gene loci reveals five distinct categories of K9 remodeling (Fig. 4). IDH1 downregulation suggests a feed-forward loop compounding enzymatic inhibition. These findings demonstrate the integrated chromatin and transcriptional consequences of the pan-2-ODD inhibition mechanism our laboratory has established across histone [2,3], mRNA [3], and DNA [5] demethylases, and extend the mechanistic framework linking NO-mediated epigenetic dysregulation to NO's pro-tumorigenic effects. Declaration of generative AI use Generative AI tools (Anthropic) were used during the preparation of this manuscript to assist with data visualization and statistical analysis. The authors reviewed and edited all AI-generated content and take full responsibility for the content of the publication. CRediT authorship contribution statement Sara Shayan: Data curation, Formal analysis, Writing – review & editing. Divya Vasudevan: Data curation, Formal analysis, Investigation, Methodology, Writing – review & editing. Douglas D. Thomas: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Project administration, Supervision, Validation, Writing – original draft. Declaration of competing interest D.V. is currently employed by Bristol Myers Squibb, Lawrenceville, NJ. Acknowledgements This work was partially supported by NIH, USA 1R01GM155946-01 (DDT). Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.redox.2026.104170. Supplementary data The following are the Supplementary data to this article. Multimedia component 1 Multimedia component 2 K9ac replicate quality assessment. (A) UT replicates show 2:1 signal ratio despite high correlation (r = 0.928). (B) NO replicates are well-matched (ratio = 0.90). (C) Per-replicate fold changes correlate (r = 0.75) with 77.5% direction agreement. (D) R1-only vs R2-only FC distributions are shifted by ∼1.1 log2 units. This figure supports the use of R1-only K9ac analysis described in Methods. Genomic Distribution of H3K9 Modifications at Gene Promoters and Gene Bodies. H3K9me2 and H3K9ac ChIP-seq data from Ref. [3] were reprocessed using nf-core/chipseq (v2.1.0; see Methods). K9me2 uses replicate means; K9ac uses R1 pair due to UT replicate signal imbalance (see Methods and Supplemental Fig. S1). (A) H3K9me2 signal at gene promoters (TSS ± 2 kb) shows minimal change (median log2FC = +0.005) despite +69% global increase by mass spectrometry; only 0.5% of genes exceed |log2FC| > 0.5, indicating predominantly intergenic accumulation. (B) H3K9me2 gene body signal (TSS to TES) similarly unchanged at median (log2FC = 0.000). (C) K9me2 gene body fold-change vs. expression fold-change: modest but highly significant inverse correlation (Spearman ρ = −0.114, p = 6 × 8−55). (D) H3K9ac log2FC (NO/UT) stratified by gene expression response: upregulated genes show K9ac gain, unchanged genes show loss. Boxes: median and IQR; Wilcoxon p-values shown. (E) K9ac changes at known NO-responsive genes (ADM, LOX, NDRG1, HMOX1, BNIP3, VEGFA): all show promoter K9ac gain consistent with transcriptional activation. Data availability ChIP-seq and microarray data sets used in this study are available in the NCBI Gene Expression Omnibus under accession number (GEO: GSE66324). References Associated Data Supplementary Materials Multimedia component 1 Multimedia component 2 Data Availability Statement ChIP-seq and microarray data sets used in this study are available in the NCBI Gene Expression Omnibus under accession number (GEO: GSE66324).