Nitric oxide inhibits ten-eleven translocation DNA demethylases to regulate 5mC and 5hmC across the genome

Abstract

DNA methylation at cytosine bases of eukaryotic DNA (5-methylcytosine, 5mC) is a heritable epigenetic mark that can regulate gene expression in health and disease. Enzymes that metabolize 5mC have been well-characterized, yet the discovery of endogenously produced signaling molecules that regulate DNA methyl-modifying machinery have not been described. Herein, we report that the free radical signaling molecule nitric oxide (NO) can directly inhibit the Fe(II)/2-OG-dependent DNA demethylases ten-eleven translocation (TET) and human AlkB homolog 2 (ALKBH2). Physiologic NO concentrations reversibly inhibited TET and ALKBH2 demethylase activity by binding to the mononuclear non-heme iron atom which formed a dinitrosyliron complex (DNIC) preventing cosubstrates (2-OG and O₂) from binding. In cancer cells treated with exogenous NO, or cells endogenously synthesizing NO, there was a global increase in 5mC and 5-hydroxymethylcytosine (5hmC) in DNA, the substrates for TET, that could not be attributed to increased DNA methyltransferase activity. 5mC was also elevated in NO-producing cell-line-derived mouse xenograft and patient-derived xenograft tumors. Genome-wide DNA methylome analysis of cells chronically treated with NO (10 days) demonstrated enrichment of 5mC and 5hmC at gene-regulatory loci which correlated to changes in the expression of NO-regulated tumor-associated genes. Regulation of DNA methylation is distinctly different from canonical NO signaling and represents a novel epigenetic role for NO.

Figure 1. NO inhibits Fe(II)/2-OG-dependent DNA demethylase activity.

(A–E) TET2 demethylase activity (A) Conversion of 5mC (black) to 5hmC (red) generated when human wt-TET2 was incubated with NO (Sper/NO; 0–500 mM; 3h). Demethylase activity assay was initiated with addition of 5mC-double-stranded DNA substrate. (B) TET2 activity (5hmC formation) measured over a range of Sper/NO concentrations (0 – 1.5 mM), IC₅₀=164.5 mM Sper/NO (~0.5–2 mM NO). (C) % inhibition: TET2 was incubated with 164.5 mM Sper/NO and with different starting DNA substrates for TET2 (5mC, 5hmC, or 5fC) for 3 h. (D) TET2 demethylase activity (product formation (5hmC)) was measured at 2 time points, 1 hour (gray bars) and 3 hours (red bars). Activity was measured on TET2 alone or after incubation with NO for different lengths of time using two NO-donors (rapid NO release: DEA/NO, and (slower NO release: Sper/NO). ELISA, n=4. (E) TET2 activity in the presence of freshly isolate genomic DNA (1 mg) and NO (Sper/NO 0–300 mM), 5hmC measured by dot-blot hybridization using anti-5hmC antibody and total DNA measured by methylene blue. (F–H) ALKBH2 activity. (F) Recombinant ALKBH2 was incubated with Sper/NO (0–500 mM) and the fluorescent probe. Demethylation of the fluorescent probe (ALKBH2 activity) was continuously monitored by fluorescence spectroscopy at 480 nm for 1 h. (G) IC₅₀ determination from data in “F”. (H) Cellular extracts containing ALKBH2 were incubated with the fluorescent probe and the slow NO-releasing donor DETA/NO (t_1/2=22h, 0–150 mM), demethylation was measured in real time for 12 h. Data are Mean ± S.E.M, n≥3.

Figure 2. NO forms a DNIC at the mononuclear non-heme iron atom of TET2.

(A) Representative X-Band (77K) EPR spectra of full length TET2 protein treated with Sper/NO (100 mM) and all substrates and cofactors for 1 min (red line) and 20 min (black line). Control reaction (blue line) is the complete reaction without TET2 after 20 min, n=3. Spectrum is indicative of a non-Heme NO-bound DNIC. (B) DFT computations. Core of the wB97xD/def2-svp optimized geometry of triplet di-nitrosyl complex [Fe^II(OAc)(Im)₂(NO)₂(OH₂)]⁺, OAc = acetate, Im = N-methyl-imidazole. Bond lengths and angles in Å and °, respectively. (C) Core of the wB97xD/def2-svp optimized geometry of [Fe(OAc)(Im)₂(NO)(OH₂)₂]⁺ and (D) [Fe(OAc)(Im)₂(NO)(O₂) (OH₂)]⁺. (E) Crystal structure of TET2 in complex with N-oxalyglycine (OGA), a 2-OG analog (PDB file 4nm6). OGA forms two critical hydrogen bonds with a conserved arginine (Arg 1261 in TET2). (F) Two NO molecules replace the two oxygens from OGA that coordinate with Fe(II).

Figure 3. NO increases 5mC on DNA from cancer cells and tumors.

(A) Relative abundance of 5mC-DNA in MDA-MB-231, −468 (TNBC), PC-3 (prostate), and U251 (glioblastoma) cells that were treated without (light bars) or with (dark bars) NO (DETA/NO 100 mM, 24 h), ELISA. (B) Relative 5mC-DNA and (C) Relative 5hmC-DNA from MDA-MB-231 and MDA-MB-468 cells that were chronically treated with NO for 10 days (DETA/NO 50 and 100 mM added every 48 h), ELISA. (D) Immunoblot for NOS2 protein in MDA-MB-231 empty vector control (VC), MDA-MB-231 NOS2-expressing, and MDA-MB-231 NOS2 + L-NMMA (NOS inhibitor, 1 mM) cells after 24 h. (E) Total NO synthesis as measured by accumulation of nitrite (a stable NO oxidation product) in the media after MDA-MB-231 (VC) and MDA-MB-231 NOS2 +/− L-NMMA cells were cultured for 24 and 48 hours. (F) 5mC-DNA (same conditions as panel “E”), ELISA. (G) Schematic showing how inhibitors cycloleucine (CL), 5-Azacytidine (AZA), and NO interact with DNMT (DNA methyltransferase), MAT (methionine adenosyltransferase), and TET to affect 5mC-, 5hmC-DNA levels. (H)5mC-DNA from MDA-MB-231 cells treated for 24 h +/− NO (100 mM DETA/NO), and +/− 1 mM 5-Azacytidine (AZA), or +/− 1 mM cycloleucine (CL), ELISA. (I, J) MDA-MB-231 and MDA-MB-468 cells were cultured for 10 days with NO (DETA/NO, 100 mM). (I) The expression levels of DNA methyltransferases (DNMT1, 3A, and 3B) and (J) DNA demethylases (TET1,2, and 3, ALKBH2) were measured in both cell lines by immunoblotting. Densitometry for protein expression levels in each cell line relative to paired untreated control cell, n = 3 separate experiments. (thin black vertical line indicates splicing of lanes that were run on the same gel but were noncontiguous). (K) 5mC-DNA levels in MDA-MB-231 xenograft tumors from mice with”NO-producing” tumors (control; dark green bars), and from mice with tumors that did not synthesize NO (treated with a NOS2 inhibitoraminoguanidine (AG) for 37 days, light bars), ELISA, n=7/group. (L) 5mC-DNA from TNBC PDX tumors from control mice with tumors that did synthesize NO (control, vehicle saline injection; dark green bars) and from tumors that did not synthesize NO(administered the pan-NOS inhibitor NG-monomethyl-L-arginine (L-NMMA)), ELISA, n=3/group. Data are represented as mean ± S.E.M., P-values were determined by unpaired two-tailed Student’s t-tests, NS = not significant, n ≥ 3.

Figure 4. NO causes aggressive cancer phenotypes, regulates tumor-permissive gene expression, and induces loci-specific DNA methylation patterns.

(A) Patient survival as a function of NOS2 gene expression and hormone receptor status using “Kaplan-Meier Plotter” (public database includes 7,830 unique breast cancer samples). (B–D) Cell migration and invasion were measured in real time by the xCELLigence^® DP system. (B) MDA-MB-231 cells were plated and allowed to adhere for 20 h before the addition of NO (100 mM DETA/NO). (C) MDA-MB-231 empty vector control (VC) and MDA-MB-231 NOS2-expressing cells were plated and migration was monitored for 40 h. (D) MDA-MB-231 cells were plated on a Matrigel^® matrix and incubated for 20 h before the addition of NO (100 mM DETA/NO), n = 6. (E) NOS2 gene expression and response to chemotherapy using the ROC Plotter platform which analyzes transcriptomic data sets from TNBC patient tumors (n = 164). (F–H) mRNA-Seq was conducted on samples from MDA-MB-231 and MDA-MB-468 cells treated with or without DETA/NO (100 mM; 10 days, n=3 biological replicates). (F) Volcano plots of NO-mediated mRNA changes (mRNA-Seq) in MDA-MB-468 and MDA-MB-231 cells cultured with NO compared to untreated control cells. Significantly up- and down-regulated genes are indicated in red, n = 3 biologically independent samples for each cell type (P-values, and fold change as indicated). (G) Venn diagram demonstrating of the number of significantly NO-regulated genes that overlap between the two cell types. (H) Multidimensional scaling (MDS) analysis of gene expression from mRNA-Seq data sets. (I–M) Oxidative reduced representation bisulfite sequencing (oxRRBS) was conducted on samples from MDA-MB-231 and MDA-MB-468 cells treated with or without DETA/NO (100 mM; 10 days, n=2 biological replicates). (I) Percent of hyper- and hypo-differentially methylated positions (DMP) and hyper- and hypo-differentially hydroxymethylated positions (DhMP) across all annotated sites in both cell types. (J) Annotated CpG sites of NO-mediated hyper-DMP and hyper-DhMP in MDA-MB-231 and MDA-MB-468 cells (DMP/DhMP = P-value < 0.05, mean difference in abs(beta value) of ≥ 0.1 according to RnBeads). (K) Gene regulatory locations of NO-mediated hyper- and hypo-differentially methylated CpG positions (DMPs) and (L)hyper- and hypo-differentially hydroxymethylated CpG positions (DhMPs) in MDA-MB-231 and MDA-MB-468 cells in response to NO. (M) Correlation between the magnitude of DMP and DhMP at functional elements between both cell types. P-values were determined by unpaired two-tailed Student’s t-tests.

Figure 5. NO-mediated changes in 5mC and 5hmC are associated with transcriptional changes in NO-responsive genes.

(A-C) b-values (5mC or 5hmC) at gene-regulatory loci (as determined by RnBeads) are ranked from lowest to highest going from left to right and are paired with the expression changes in their associated genes. (A) Changes in 5mC (blue) and change in gene expression (green) at promoters and gene bodies for MDA-MB-231 and MDA-MB-468 cells. (B)Changes in 5hmC (red) and change in gene expression (green) at promoters and gene bodies for MDA-MB-231 and MDA-MB-468 cells. (C) MDA-MB-468 cells changes in 5mC (blue, top) and 5hmC (red, bottom) and change in gene expression (green) at typical enhancers (In MDA-MB-231 cells the enrichment of 5mC/5hmC at typical enhancer regions did not significantly correlate to gene expression changes). (D) 5mC-associated genes, and (E) 5hmC-associated genes. Left panels “Cellular Responses to NO”, are representative genes shown with their difference in b-value at their promoter region (oxRRBS data) and their expression level (mRNA-seq) in response to NO. Right side “Clinical Correlations”, scatter plot demonstrates correlation between NOS2 expression and specific genes in tumors from patients with aggressive breast cancer. And far right column is correlation between TNBC patient survival and high (red line) or low (black line) expression of the same gene (Kaplan-Meier plot).