Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jun 29.
Published in final edited form as: Metabolomics. 2024 Aug 3;20(5):91. doi: 10.1007/s11306-024-02159-2

Metaboepigenetic regulation of gene expression in obesity and insulin resistance

Swapan K Das 1,*, Mary E Comeau 2, Carl D Langefeld 2,3
PMCID: PMC13310568  NIHMSID: NIHMS2188976  PMID: 39096438

Abstract

Introduction:

Variation in DNA methylation (DNAm) in adipose tissue is associated with the pathogenesis of obesity and insulin resistance. The activity of enzymes involved in altering DNAm levels is dependent on several metabolite cofactors.

Objectives:

To understand the role of metabolites as mechanistic regulators of epigenetic marks, we tested the association between selected plasma metabolites and DNAm levels in the adipose tissue of African Americans.

Methods:

In the AAGMEx cohort (N=256), plasma levels of metabolites were measured by untargeted liquid chromatography-mass spectrometry; adipose tissue DNAm and transcript levels were measured by reduced representation bisulfite sequencing, and expression microarray, respectively.

Results:

Among the 21 one-carbon metabolism pathway metabolites evaluated, six were associated with gluco-metabolic traits (PFDR <0.05, for BMI, SI, or Matsuda index) in AAGMEx. Methylation levels of 196, 116, and 180 CpG-sites were associated (P<0.0001) with S-adenosylhomocysteine (SAH), cystine, and hypotaurine, respectively. Cis-expression quantitative trait methylation (cis eQTM) analyses suggested the role of metabolite-level-associated CpG sites in regulating the expression of adipose tissue transcripts, including genes in G-protein coupled receptor signaling pathway. Plasma SAH level-associated CpG sites chr19:3403712 and chr19:3403735 were also associated with the expression of G-protein subunit alpha 15 (GNA15) in adipose. The expression of GNA15 was significantly correlated with BMI (β= 1.87, P = 1.9×10−16) and SI (β= −1.61, P = 2.49 × 10−5).

Conclusion:

Our study suggests that a subset of metabolites modulates the methylation levels of CpG sites in specific loci and, in turn, regulates the expression of transcripts involved in obesity and insulin resistance.

Keywords: Metabolite, DNA methylation, cis-eQTM, African American, Obesity, Insulin resistance

1. Introduction:

Metabolites in plasma and tissue are small molecules and represent intermediate phenotypes linking genetic and environmental factors to complex disorders like obesity and insulin resistance (Newgard, 2017; Pillon et al., 2021). However, the molecular mechanisms connecting altered metabolite level and gluco-metabolic disease remain poorly elucidated. Several lines of evidence, including studies by our laboratory, support the important contribution of altered DNA methylation (DNAm) in the pathogenesis of insulin resistance, obesity and type 2 diabetes in humans (Ling and Rönn, 2019; Sharma et al., 2020). Chromatin-modifying enzymes acts as writers, readers and erasers, and can induce genomic and cellular responses by altering epigenetic marks, including DNAm (Gut and Verdin, 2013). Activity of these enzymes are dependent on the availability of and fluctuations in the level of several key metabolites, as these small molecules act as cofactors for their catalytic activities (Haws et al., 2020a; van der Knaap and Verrijzer, 2016). For example, DNA and histone methylation reactions are catalyzed site-specifically by DNA-methyltransferase and histone methyltransferase enzymes, both of which requires S-adenosylmethionine (SAM), a core one-carbon metabolite as co-factor (Haws et al., 2020a). In vitro and in vivo studies in animal models suggested that altered abundance of SAM determine epigenetic states by altered methylation of lysine residues of histone (H3Kme) and subsequently regulate gene expression (Haws et al., 2020b; Shyh-Chang et al., 2013). Additionally SAM is also shown to be involved in preserving chromatin state stability during metabolic stress (Haws et al., 2020b). A limited number of mechanistic studies similarly shows link between SAM abundance and methylation of the DNA base cytosine at carbon 5 (5mC) (Maddocks et al., 2016; Mattocks et al., 2017). Thus, a subset of metabolites, termed “epimetabolites”, are likely to be principal and mechanistically direct regulators of epigenetic marks involved in complex disorders like obesity and insulin resistance (Dai et al., 2020; Donohoe and Bultman, 2012; Sharma and Rando, 2017). However, the role of plasma epimetabolite levels in disease process-associated epigenetic regulation of gene expression remain poorly understood.

Mechanistic studies in cell and animal models, confirmed the causal link between metabolite-mediated modulations of the activity of DNA methyltransferases (DNMTs) and the ten-eleven translocation (TET) family enzymes, which resulted in altered gene expression via methylation and demethylation of cytosine residues of CpG-sites in regulatory regions of the genome (Betto et al., 2021; Fluhr et al., 2021; Lechner et al., 2023). However, only a few studies have investigated the association between metabolites and DNAm levels in human cohorts (Suhre and Zaghlool, 2021). Epigenome Wide Association Study (EWAS) by Petersen et al in a European ancestry cohort first tested if methylation levels of ~457,000 CpG-sites in blood DNA are associated with plasma levels of 649 metabolites (Petersen et al., 2014). A few other studies have tested association between methylation and metabolite levels by focusing on a small number of gluco-metabolic trait-associated CpG sites (Agarwal et al., 2022; Christiansen et al., 2022; Goodrich et al., 2020; Jones et al., 2023; Zaghlool et al., 2018). These studies in European, Hispanic, and Asian ancestry cohorts are limited primarily to DNAm levels in the blood.

Differential methylation of CpG sites in regulatory regions of genes encoding enzymes, transporters, or transcription factors can modulate their expression and in turn, determine the levels of metabolites in plasma or tissues by regulating its synthesis, degradation and/or transport (Petersen et al., 2014; Suhre and Zaghlool, 2021). Conversely, specific metabolite co-factors can modulate the activity of enzymes involved in the epigenetic regulation and can alter gene expression by differential methylation of CpG sites in regulatory regions of key genes involved in gluco-metabolic phenotypes (Donohoe and Bultman, 2012; Haws et al., 2020a; Sharma and Rando, 2017). Thus, the association between DNAm and metabolites are bidirectional. However, biochemical lines of evidence suggest that metabolites in one-carbon metabolism (a series of interlinked metabolic pathway including folate cycle, methionine remethylation and transsulfuration pathways) are involved in methylation of DNA (Clare et al., 2019; Lin et al., 2022), whereas several metabolites in TCA cycle (e.g. α-ketogluterate, fumarate, and succinate) act as activator or repressor of TET DNA demethylases (Kohli and Zhang, 2013; van der Knaap and Verrijzer, 2016). Thus, we hypothesize that plasma metabolites may causally contribute to obesity and insulin resistance by epigenetically regulating gene expression via locus-specific alteration of DNA-methylation levels in adipose tissue.

To test this hypothesis, we first evaluated the association of plasma metabolites with gluco-metabolic traits in a well characterized African American Cohort. Next we performed EWAS for a subset of gluco-metabolic trait-associated putative “epimetabolites”, which tested for association of these plasma metabolites with the CpG-site DNAm levels in adipose tissue. Our analyses primarily focused on metabolites in the one-carbon metabolism pathway. To establish a mechanistic role for these metabolite-associated CpG sites in regulating gene-expression, we further performed cis-expression quantitative trait methylation (cis-eQTM) analyses. Our study suggests an important role for S-adenosylhomocysteine (SAH), cystine, and hypotaurine in obesity and insulin resistance by epigenetic regulation of gene expression.

2. Methods:

2.1. Human Subjects and physiologic phenotyping:

To address our hypothesis, multiomics data including plasma metabolite, adipose tissue gene expression and DNA-methylation data from participants in the African American Genetics of Metabolism and Expression (AAGMEx) cohort, recruited previously by us, was used in this study. The AAGMEx is a Cohort of 256 metabolically well-characterized individuals without diabetes from North Carolina. Clinical, anthropometric, and physiological characteristics of the AAGMEx cohort have been previously described (Sajuthi et al., 2016; Sharma et al., 2019; Sharma et al., 2016). Briefly, cohort participants were healthy, self-reported AA men and women (53.9% male) residing in North Carolina, aged 18 to 60 years (mean age =40.7 ± 11.5 years), with a BMI between 18 and 42 kg/m2 (mean BMI =28.9 ± 5.6 kg/m2). Fasting blood samples were drawn for biochemical analyses and analysis of plasma metabolites. Abdominal subcutaneous adipose biopsies were obtained near the umbilicus under local anesthesia. Tissue biopsies were collected by Bergstrom needle from participants after an overnight fast. Tissues were immediately rinsed in sterile saline, quick-frozen in liquid nitrogen, and stored at −80°C. All participants provided written informed consent under protocols approved by the Institutional Review Boards at Wake Forest School of Medicine.

2.2. Physiologic phenotyping:

A standard 75-g oral glucose tolerance test (OGTT) was used to evaluate insulin sensitivity and exclude individuals with diabetes. Plasma glucose levels in the AAGMEx cohort were analyzed by glucose oxidase methods at a CLIA-certified commercial laboratory (LabCorp, NC) (Sharma et al., 2016). Plasma insulin was measured using an immuno-chemiluminometric assay (Invitron Limited, Monmouth, UK). Plasma glucose and insulin data from five OGTT time points (0, 30, 60, 90, and 120 min) were used to calculate the Matsuda insulin sensitivity index (http://mmatsuda.diabetes-smc.jp/MIndex.html; in AAGMEx, mean Matsuda index =6.2 ± 6.7). Detailed evaluation of insulin sensitivity was performed by the insulin-modified (0.03 U/kg) frequently sampled intravenous glucose tolerance tests (FSIGT). The MINMOD Millennium program was used to analyze FSIGT data to determine insulin sensitivity index (SI) by Minimal model analysis (in AAGMEx, mean SI =4.0 ± 3.3).

2.3. Plasma metabolomic profiling:

Plasma samples isolated from EDTA-anticoagulated blood collected at overnight fasting conditions and stored at −80 °C since collection from 253 AAGMEx cohort participants (M/F =136/117) with <50% European admixture (computed from genome-wide genotype data) was available for metabolomic profiling. Untargeted metabolite profiling (Discovery-HD4 panel /Metabolon Global Metabolomics mVIEW+) of these samples was performed together with a larger set of plasma samples (1518) at a Metabolon service facility (Metabolon, Morrisville, NC). Details of Metabolon’s robust quality assurance and quality control are available at their web site (https://www.metabolon.com/why-metabolon/validation/). Ultra high-performance liquid chromatography/tandem accurate mass spectrometry methods were used for detection and quantification of metabolites. Metabolon utilized a Waters ACQUITY ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-Exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35,000 mass resolution. Different control samples, including samples generated from a pool of human plasma (served as technical replicate throughout the data set), were analyzed in concert with the experimental samples. Experimental samples and controls were randomized across the platform run. The median relative standard deviations for the instrument variability (determined by internal standard) and total process variability (determined by endogenous biochemical) were 6% and 10%, respectively. Metabolites were identified by automated comparison of the ion features in the experimental samples to a reference library of chemical standard entries curated at Metabolon. The Discovery HD4 panel identifies and provides relative quantitation for known chemical compounds (validated by purified standards) among amino acid, carbohydrate, energy, lipid, nucleotide and peptide super-pathways. Additionally, it also identifies many xenobiotic, and “unknown” metabolites. Peaks were quantified using area-under-the-curve. Quality controlled metabolite quantification data was block corrected for run day and normalized by batch. For data normalization, each metabolite was rescaled to set the median equal to 1 (median-scaled using the MED method). Following normalization, data was log-transformed and missing values for each metabolite were imputed to the minimum, as described earlier (Das et al., 2021). Plasma Metabolite data from Metabolon for AAGMEx included 1245 known and novel metabolites. After excluding metabolites with high missing data (missing in >75% of the samples in median normalized original scale data), we used good quality data for 1124 metabolites for analyses.

2.4. Adipose tissue DNA-methylation profiling:

Subcutaneous adipose tissue (SAT) biopsy and high-quality insulin sensitivity data from both OGTT and FSIGT were available for a subset of 230 individuals (male/female =124/106) in AAGMEX and were selected for epigenome-wide profiling of DNA methylation (DNAm). Genomic DNA from ~100 mg frozen SAT biopsies was isolated with the Qiagen DNeasy tissue kit (cat. no. 69504) and as reported earlier, profiling of DNAm levels was performed by Reduced Representation Bisulfite Sequencing (RRBS) (Sharma et al., 2020). In brief, the DNAm profiling and bioinformatics analysis of RRBS data was conducted by Diagenode RRBS service (Diagenode, Belgium). RRBS libraries were prepared using the Premium RRBS Kit (Diagenode Cat# C02030033). RRBS library pools were sequenced on a HiSeq3000 (Illumina) using 50 bp single-read sequencing (SR50) to obtain at least 30 million reads/sample. The adapter-trimmed sequence reads were aligned to the Homo sapiens reference genome (Genome Reference Consortium 37, hg19) and cytosine methylation values were called using Bismark v0.20.0 with default parameters (Krueger and Andrews, 2011). Further, methylKit was used to filter the CpG dataset for low coverage and for extremely high coverage to discard reads with possible PCR bias (Akalin et al., 2012). The data was then normalized for read coverage distribution between samples with the methylKit default settings (median normalization). On average, 40.8 ± 5.8 million sequencing reads were obtained per sample. Alignment of this data to the human genome (hg19) resulted in an average of 21.4 ± 3.2 million uniquely aligned reads per sample, and 15.1 ± 2.3X average coverage. We filtered our dataset for CpGs with at least 10X coverage, and present in at least 75% of the samples, corresponding to 1,073,614 CpGs. Coverage tables and normalized methylation values for the CpG sites covered more than 10X in at least 75% of the samples were generated with methylKit. We filtered out non-variable and non-autosomal CpG sites and used data on 1,054,719 CpG sites for analyses.

2.5. Transcriptomic profiling:

As reported previously, extraction of total RNA from SAT biopsies was performed using miRNeasy Mini Kit (Qiagen) (Sajuthi et al., 2016; Sharma et al., 2016; Xu et al., 2023). Quantities of RNA samples were determined by ultraviolet spectrophotometry (Nanodrop, Thermo Scientific), and the quality of RNA was determined by electrophoresis (Experion nucleic acid analyzer, BioRad Laboratories, Inc.). Genome-wide expression data for 256 AAGMEx participants were generated using HumanHT-12 v4 Expression BeadChip (Illumina, San Diego, CA; Data submitted to Gene Expression Omnibus: GEO id #GSE95674). Participants were block-randomized by age, gender, and BMI, and their RNA samples were assigned to a group, totaling 12 samples for hybridization per BeadChip. Chips were scanned in the Illumina HiScan Reader using Illumina iScan Control Software. Genome-wide gene expression data (probe level) were extracted using Illumina GenomeStudio V2011.1. Expression levels were log2 transformed, robust multi-array average normalized (RMA, includes quantile normalization), and batch-corrected using ComBat. The HumanHT-12 v4 Expression BeadChip includes 47,231 probes annotated to transcripts; however, data on transcript probes encompassing common SNPs (based on ReAnnotator, or SnpInProbe annotation, and UCSC SNPv141) and transcripts that were not significantly expressed (p-value<0.05) in ≥25% of the samples were excluded.

2.6. Statistical analyses:

The analysis strategy of our study is shown in Figure 1.To test for an association between glucose homeostasis or obesity traits and plasma metabolite level, a linear regression model was computed with the trait (transformed) as the outcome and plasma metabolite level (normalized and median scaled) as the predictor. Models included age, sex, and ancestry proportion (admixture) as covariates. Admixture estimates were computed using the ADMIXTURE program as presented in our published studies (Sajuthi et al., 2016; Sharma et al., 2019). Matsuda Index and SI were natural logarithm transformed and BMI was square root for BMI to better approximate model assumptions. A false discovery rate (FDR) corrected p-value <0.05 in linear regression analysis was considered as evidence for significant metabolite-wide association. However, in this study we focused on a subset of metabolites in one-carbon metabolism-related pathway (annotated to Methionine, Cysteine, SAM and Taurine Metabolism). Additionally, for better biological interpretation of effect size, association between glucose homeostasis or obesity traits and plasma metabolite level were also tested by partial correlation (Pearson) analysis, which adjusts both the metabolite and the outcome variable by the same covariates to produce the partial-r.

Figure 1:

Figure 1:

Open in a new tab

Study Workflow

To test the primary hypothesis, the association between adipose tissue CpG DNA-methylation levels and selected plasma metabolite levels were evaluated by linear regression analysis. In this Epigenome-wide association study (EWAS) DNA-methylation levels (1,054,719 CpG sites) were the outcome, plasma levels of the metabolites were predictors, and age, sex, and admixture were included as covariates in the model. For this analysis we computed Box-Cox power transformations on the metabolites, and selected the best transformation (natural log for S-adenosylhomocysteine and cystine; and square root for hypotaurine) to improve model fit to the data. The CpG sites with sufficient variability (CpG-methylation level interval between the 5th and 95th percentile > 0.1) and FDR corrected p-value <0.1 in linear regression analysis was primarily considered as evidence for significant epigenome-wide association. Additionally, EWAS for plasma metabolite level were also computed by Pearson’s and Spearman’s partial correlation analysis (parametric and non-parametric test, respectively) using same transformed data and covariates.

To determine the putative functional role of metabolite-associated CpG sites in modulating expression of transcripts, we performed cis-expression quantitative trait methylation (cis-eQTM) analysis. A linear regression model was computed with the adipose tissue transcript expression (log2 transformed) as the outcome and CpG methylation level (normalized) as the predictor. Models included age, sex, and ancestry proportion (admixture) as covariates. The cis-eQTM analysis was implemented in the R package matrixEQTL and was used to test association between each CpG site and any expressed transcript within 1Mb around these sites. Additionally, cis-eQTMs for metabolite-associated CpG sites were also computed by Pearson’s and Spearman’s partial correlation analysis. The CpGs associated with a transcript at p<0.0001 in either linear regression or partial correlation analysis were considered significant.

2.7. Bioinformatic annotation:

Known metabolites identified in plasma samples were annotated to identifiers in several chemical databases including PUBCHEM (https://pubchem.ncbi.nlm.nih.gov/), CAS (https://commonchemistry.cas.org/), KEGG (https://www.genome.jp/kegg/) and HMDB (https://hmdb.ca/). These plasma metabolites were further classified into super and sub-pathways. Considering the putative role of one-carbon metabolism in DNA methylation, in this study we focused on 21 metabolites classified in “methionine, cysteine, SAM and taurine metabolism” sub-pathway under “amino acid” super-pathway. Metabolite-associated CpG sites were annotated with the R/Bioconductor package ‘annotatr’ (DOI: 10.18129/B9.bioc.annotatr). The annotation consisted of two categories: (i) distance to a CpG island and (ii) annotation to gene region. The distance-related annotations identify whether CpG sites overlap a known CpG island, 2000bp of the flanking regions of the CpG islands (shores), 2000bp of the flanking regions of the shores (shelves), or outside these regions (cpg_inter or open sea). The gene region analysis classifies methylation sites in groups in the context of genes, namely, exons, UTRs, introns, promoters, and intergenic regions. Annotations were derived using hg19 data from the built-in annotations for 'cpgs', 'basicgenes', 'genes_intergenic', and 'genes_intronexonboundaries'. Biological pathway enrichment analysis was performed using Database for Annotation, Visualization and Integrated Discovery (DAVID; https://david.ncifcrf.gov/home.jsp; DAVID Knowledgebase v2023q4) and g:GOSt functional profiler (https://biit.cs.ut.ee/gprofiler/gost).

3. Results:

3.1. Association of one-carbon metabolism-related pathway metabolites with Insulin sensitivity and BMI in AAGMEx:

Among the 1124 plasma metabolites evaluated in AAGMEx participants, 166, 153 and 60 were significantly associated (PFDR<0.05) with BMI, Matsuda index and SI, respectively (Supplementary Figure-1 and Supplementary Table-1). However, focusing on the 21 metabolites in one-carbon metabolism-related pathway (classified in methionine, cysteine, SAM and taurine metabolism sub pathway) we identified association of six of these metabolites with gluco-metabolic traits (FDR<0.05, for SI, Matsuda index, or BMI, Table-1). Plasma level of cystine in this pathway was strongly positively and inversely associated with BMI (β= 1.09, P= 7.53 × 10−9) and Matsuda index (β= −1.06, P= 2.96 × 10−4), respectively. The SAH, a key one-carbon metabolism pathway metabolite was also significantly associated with BMI (β= 0.21, P= 0.005). However, alpha-ketoglutarate and succinate, two TCA cycle metabolites and key cofactor of TET DNA demethylases were not significantly associated with gluco-metabolic traits in AAGMEx.

Table 1:

Association of One-Carbon metabolism-related pathway metabolites with Insulin sensitivity (SI) and body mass index (BMI) in African American Genetics of Metabolism and Expression (AAGMEx) cohort.

BIOCHEMICAL COMP ID BMI Matsuda Index SI
β SE(β) P-value rpartial β SE(β) P-value rpartial β SE(β) P-value rpartial
Cystine 56 1.09 0.18 7.53E-09 0.39 −1.06 0.29 2.96E-04 −0.25 −0.90 0.29 0.003 −0.22
Hypotaurine 590 −0.37 0.07 4.58E-08 −0.36 0.46 0.10 1.26E-05 0.28 0.49 0.10 3.24E-06 0.31
2,3-dihydroxy-5-methylthio-4-pentenoate (DMTPA)* 62805 0.58 0.18 0.002 0.22 −0.91 0.28 0.002 −0.24 −0.57 0.28 0.046 −0.17
Alpha-ketobutyrate 4968 0.19 0.06 0.002 0.19 −0.24 0.09 0.011 −0.19 −0.20 0.09 0.029 −0.16
S-adenosylhomocysteine (SAH) 42382 0.21 0.07 0.005 0.16 −0.23 0.11 0.040 −0.13 −0.37 0.12 0.002 −0.18
S-methylcysteine sulfoxide 43378 −0.11 0.06 0.067 −0.11 0.25 0.09 0.006 0.18 0.16 0.09 0.090 0.10
N-acetyltaurine 48187 −0.09 0.09 0.322 −0.03 0.32 0.13 0.018 0.19 0.16 0.14 0.234 0.07
Methionine sulfone 44878 −0.17 0.09 0.066 −0.10 0.31 0.14 0.024 0.11 0.31 0.14 0.025 0.15
Cysteine sulfinic acid 37443 −0.12 0.07 0.107 −0.11 0.23 0.11 0.044 0.13 0.14 0.11 0.199 0.10
Lanthionine 42002 −0.13 0.07 0.073 −0.09 0.03 0.11 0.801 0.02 0.13 0.11 0.259 0.04
S-methylcysteine 39592 −0.13 0.08 0.090 −0.11 0.16 0.11 0.165 0.06 0.09 0.12 0.437 0.02
Cysteine s-sulfate 22176 0.05 0.04 0.229 0.08 −0.10 0.06 0.097 −0.13 −0.05 0.06 0.443 −0.04
Cystathionine 15705 0.01 0.06 0.925 −0.05 −0.16 0.10 0.101 −0.05 −0.04 0.09 0.641 0.04
Taurine 2125 −0.27 0.18 0.138 −0.10 0.18 0.28 0.526 0.06 0.05 0.29 0.856 0.01
Cysteine 1868 −0.16 0.15 0.284 −0.08 0.29 0.23 0.198 0.09 0.14 0.23 0.556 0.04
2-hydroxy-4-(methylthio)butanoic acid 63739 −0.02 0.07 0.782 0.07 −0.11 0.11 0.290 −0.07 0.06 0.11 0.599 0.00
5-methylthioribose** 62100 0.07 0.07 0.295 0.06 −0.03 0.10 0.773 −0.01 0.00 0.10 0.971 −0.01
Methionine sulfoxide 18374 0.01 0.16 0.954 −0.03 0.06 0.24 0.800 0.06 0.22 0.24 0.364 0.08
N-formylmethionine 2829 −0.13 0.18 0.479 −0.06 0.14 0.27 0.604 0.04 0.04 0.28 0.874 0.02
N-acetylmethionine 1589 0.09 0.15 0.547 0.02 0.05 0.22 0.814 0.01 −0.15 0.23 0.521 −0.04
Methionine 1302 0.06 0.24 0.806 0.01 −0.05 0.37 0.885 0.02 0.10 0.37 0.777 0.04
Open in a new tab

Metabolites significant at FDR-p<0.05 are shown in bold.

We reviewed published studies for replication of association of one-carbon metabolism pathway metabolites in independent cohorts. Plasma levels of cystine (β= 0.049, P= 1.99 ×10−28), hypotaurine (β= −0.043, P= 4.23 ×10−19), and alpha-ketobutyrate (β= 0.032, P= 1.54 ×10−10) was significantly associated with BMI in Insulin Resistance Atherosclerosis Family Study (IRASFS) Mexican Americans (N= 1108) (Das et al., 2021). Plasma levels of hypotaurine was also strongly associated with BMI in European ancestry individuals in TwinsUK3 (N= 861, P= 4.77 ×10−10) and Health Nucleus (N= 427, P= 0.016) cohorts (Cirulli et al., 2019). The levels of SAH (P= 3.04 ×10−5) and cystine (P= 0.004), but not DMPTA and alpha-ketobutyrate, were associated with BMI in the Health Nucleus cohort (Cirulli et al., 2019). The SAH was also associated with disposition index in IRASFS African Americans (N= 571, β= −2.79, P= 0.024) (Okut et al., 2023). Considering the known biological role of SAH as a central one-carbon metabolite in DNA methylation and the strong association and replication of cysteine and hypotaurine with the gluco-metabolic phenotype, we prioritized these three metabolites for EWAS analysis in the next step.

3.2. Adipose tissue DNA methylation levels of CpG sites are associated with plasma metabolites:

We performed EWAS to determine if plasma metabolites can modulate the DNA methylation levels at specific CpG sites in adipose tissue. EWAS was performed for three selected gluco-metabolic traits-associated one-carbon metabolism-related pathway metabolites namely, SAH, cystine, and hypotaurine. At PFDR<0.1, methylation levels of 136, 0, and 2 CpG-sites were associated with SAH, cystine, and hypotaurine, respectively. However, in a less stringent threshold (P<0.0001 in linear regression adjusted for age, sex, and admixture, or in partial correlation analysis) methylation levels of 196, 116, and 180 CpG sites were associated with SAH, cystine, and hypotaurine, respectively (Figure-2 and Supplementary table-2). DNA methylation levels in a total of 492 CpG sites were associated with these three epimetabolites. CpG sites chr22:31002992 (β=0.04, P=2.96×10−8, in promoter of TCN2), chr11:89135566 (β=0.19, P=4.92×10−7, in exon of NXO4), and chr21:38120202 (β=0.05, P=5.53×10−8, in exon of SIM2) were most significantly associated with plasma SAH, cystine, and hypotaurine levels, respectively (Table-2).

Figure 2: Manhattan plot for EWAS showing association of selected One-Carbon metabolism-related pathway plasma metabolites with adipose tissue CpG-DNA methylation.

Figure 2:

Open in a new tab

713,100 CpG sites with the methylation value interval (P95 – P05 > 0.1) in the plots. PFDR=0.1 is marked by horizontal black line.

Table-2: Association of selected plasma metabolites with adipose tissue CpG-DNA methylation.

Top five CpG sites for three one carbon metabolites ranked by minimum p-value from linear regression and partial correlation analysis and are shown.

CpG Name (chr: position) Linear Regression Pearson partial correlation COMPID BIOCHEMICAL
β βSE P-value PFDR
chr22:31002992 0.047 0.008 2.96E-08 0.002 0.409 42382 S-adenosylhomocysteine (SAH)
chr7:4765052 0.077 0.014 1.64E-07 0.003 0.356 42382 S-adenosylhomocysteine (SAH)
chr7:100942958 0.069 0.013 1.92E-07 0.003 0.358 42382 S-adenosylhomocysteine (SAH)
chr7:1286857 0.054 0.010 3.35E-07 0.003 0.370 42382 S-adenosylhomocysteine (SAH)
chr5:134871118 0.050 0.010 4.71E-07 0.004 0.365 42382 S-adenosylhomocysteine (SAH)
chr21:38120202 0.054 0.009 5.53E-08 0.058 0.400 590 hypotaurine
chr6:109017744 0.072 0.013 1.16E-07 0.061 0.378 590 hypotaurine
chr13:49752530 −0.034 0.007 1.53E-06 0.269 −0.313 590 hypotaurine
chr2:21266700 −0.078 0.016 3.37E-06 0.272 −0.305 590 hypotaurine
chr9:114362358 −0.135 0.028 3.58E-06 0.272 −0.327 590 hypotaurine
chr11:89135566 0.191 0.037 4.92E-07 0.299 0.329 56 cystine
chr7:97877803 −0.106 0.020 5.67E-07 0.299 −0.337 56 cystine
chr2:119599911 −0.098 0.020 2.65E-06 0.698 −0.306 56 cystine
chr11:8284947 −0.108 0.022 1.46E-06 0.512 −0.350 56 cystine
chr22:38379711 0.307 0.066 6.47E-06 0.703 0.310 56 cystine
Open in a new tab

3.3. Epimetabolite associated CpG sites regulate expression of local transcript:

A large majority of these metabolite-associated CpG sites (341, 69.3 %) were located in or within 5Kb of a protein coding gene. Also, 237 (48.1%) of these CpG sites were within or close (CpG shelves and shores) to CpG islands. We performed cis-eQTM analyses to obtain evidence in support of the functional role for these 492 metabolite-associated CpG sites in regulating adipose tissue gene expression. The cis-eQTM analyses identified 15 pairs of CpG-transcript probe association (P< 0.0001), suggesting association of 13 CpG-sites with the transcript levels of 12 protein coding genes in adipose tissue (Table-3 and Supplementary table-3). Based on DAVID pathway enrichment analysis, these target genes for one-carbon metabolite-associated CpG sites were significantly enriched for GTPase activity (GO:0003924; 4 genes, P= 5.36 ×10−4, PFDR=0.017) and are members of G-protein coupled receptor signaling pathway. For example, plasma cystine level-associated CpG site chr9:116930188 was associated with the transcript level of Regulator of G-protein Signaling 3 (RGS3; Spearman rpartial = −0.27, P= 5.99 × 10−5). In adipose tissue, expression of RGS3 was strongly associated with BMI (β = −1.67, p= 8.84 × 10−12). Similarly, plasma SAH level-associated CpG site chr19:3403712 (rpartial = 0.32, P= 5.50 × 10−7) and chr19:3403735 was associated with the transcript level of G protein subunit alpha 15 (GNA15) gene. Integration of the EWAS results for glucometabolic traits in AAGMEx cohort (Sharma et al., 2020) suggested that CpG site chr19:3403712 is associated with BMI (β= 1.02, P= 1.17 × 10−6) and Matsuda index (β= −1.21, P= 7.72 × 10−5). We also identified a strong association of adipose tissue transcript levels of GNA15 with BMI (β= 1.87, P= 1.90 ×10−16) and Matsuda index (β= −2.83, P= 5.18 ×10−16) (Figure-3).

Table 3:

Plasma Cystine, Hypotaurine and S-adenosylhomocysteine (SAH) associated CpG-DNAm sites regulate expression of transcripts in adipose tissue.

CpG Name (chr:position) Region Annotation* Position Annotation* EWAS eQTM analysis
Linear Regression Spearman partial correlation Metabolite Transcript Linear Regression Spearman partial correlation
β SE(β) P-value rpartial P-value Probe_ID Symbol β SE(β) P-value rpartial P-value
chr9:116930188 island 5’UTR 0.21 0.05 2.13E-05 0.25 2.52E-04 Cystine ILMN_1724349 RGS3 −0.15 0.05 3.27E-03 −0.27 5.99E-05
chr22:38379690 island exon 0.29 0.07 8.15E-05 0.27 8.68E-05 Cystine ILMN_1709795 RAC2 0.54 0.13 6.21E-05 0.22 1.17E-03
chr6:168983447 open sea intron 0.12 0.03 9.88E-05 0.25 1.34E-04 Cystine ILMN_1676679 KIF25 −1.17 0.27 2.70E-05 −0.27 4.48E-05
chr6:168983447 open sea intron 0.12 0.03 9.88E-05 0.25 1.34E-04 Cystine ILMN_1665903 KIF25 −1.23 0.28 1.37E-05 −0.26 4.81E-05
chr9:114362358 shore promoter −0.13 0.03 3.58E-06 −0.33 3.50E-06 Hypotaurine ILMN_1778161 DNAJC25 −0.16 0.04 3.43E-05 −0.27 1.21E-04
chr9:114362354 shore promoter −0.12 0.03 2.80E-05 −0.28 6.06E-05 Hypotaurine ILMN_1768197 ROD1 0.28 0.06 8.92E-07 0.35 6.65E-07
chr7:6206061 shore intron −0.15 0.04 3.42E-05 −0.30 1.94E-05 Hypotaurine ILMN_3250614 EIF2AK1 0.08 0.02 1.93E-04 0.28 7.13E-05
chr22:50319331 shelf exon 0.11 0.03 6.07E-05 0.22 8.24E-04 Hypotaurine ILMN_1694810 PANX2 −0.13 0.03 7.66E-05 −0.23 4.99E-04
chr10:121129167 open sea intron −0.16 0.04 9.97E-05 −0.28 1.64E-04 Hypotaurine ILMN_1733538 RGS10 0.31 0.08 7.90E-05 0.31 2.69E-05
chr10:121129167 open sea intron −0.16 0.04 9.97E-05 −0.28 1.64E-04 Hypotaurine ILMN_1668559 RGS10 0.14 0.04 5.84E-04 0.31 2.67E-05
chr2:60251202 open sea intergenic 0.07 0.02 1.47E-04 0.26 5.28E-05 Hypotaurine ILMN_2255133 BCL11A −0.18 0.04 1.91E-05 −0.31 1.27E-06
chr4:8509523 open sea intergenic −0.08 0.02 1.25E-03 −0.25 9.31E-05 Hypotaurine ILMN_1756595 SH3TC1 0.13 0.04 9.38E-04 0.26 7.24E-05
chr11:64022778 shelf exon −0.03 0.01 3.25E-05 −0.03 7.17E-01 SAH ILMN_2408748 SLC22A12 −0.72 0.16 1.61E-05 −0.14 4.20E-02
chr19:3403712 open sea intron 0.08 0.02 6.79E-04 0.27 4.27E-05 SAH ILMN_1773963 GNA15 0.27 0.05 6.81E-07 0.32 5.50E-07
chr19:3403735 open sea intron 0.07 0.02 2.50E-03 0.26 5.62E-05 SAH ILMN_1773963 GNA15 0.24 0.05 3.60E-06 0.30 3.70E-06
Open in a new tab

EWAS, Epigenome-wide association study (Metabolite → CpG-DNAm); cis-eQTM, cis-expression quantitative trait methylation (CpG-DNAm → transcript; Selected cis-eQTM within +/− 1Mb of a transcript with p< 0.0001 are shown); SAH, S-adenosylhomocysteine;

*

, annotatr bioinformatics tool derived annotation of CpG sites based on distance to a CpG island and physical location on a gene region.

Figure 3: DNA methylation level of CpG site at chr19:3403712 is associated with plasma S-adenosylhomocysteine (SAH) level, and may epigenetically determine the BMI by regulating the expression of GNA15 transcript in adipose tissue.

Figure 3:

Open in a new tab

BMI, Body mass index; GNA15, Guanine Nucleotide-binding Protein (G-protein) subunit Alpha 15.

4. Discussion:

The extent to which metabolites and other co-factors determine the glucometabolic phenotype by changing the epigenome in humans is not well established. To our knowledge, this is the first study in African Americans that provides evidence that a subset of metabolites in the one-carbon metabolism-related pathways may modulate the DNA methylation levels of CpG-sites at specific loci and, in turn, regulate the expression of adipose tissue transcripts involved in obesity and insulin resistance. Future studies in multiethnic cohorts will be required to define common and ancestry-specific mechanisms of metaboepigenetic regulation of gene expression relevant of obesity and insulin resistance.

One-carbon metabolism is involved in the transfer of one-carbon moieties and methyl groups for nucleotide synthesis, methylation reactions, and reductive metabolism. Biochemical evidence suggest that metabolites in this and other pathways are required for epigenetic regulation, including DNA methylation-mediated regulation of gene expression. Thus, metabolites may serve as more than just downstream products in the biological information flow from DNA to disease (Costa Dos Santos et al., 2021). Among the 21 one-carbon metabolism pathway metabolites evaluated, six were associated with gluco-metabolic traits in AAGMEx. Among these metabolites plasma levels of cystine and hypotaurine were strongly associated with gluco-metabolic traits in AAGMEx, and was replicated in published studies on other ethnicities. We also identified replication of association for the key one-carbon pathway metabolite SAH, but association for DMPTA, S-methylcysteine sulfoxide, and alpha-ketobutyrate was not consistent in other cohorts. Integrating plasma metabolite and adipose tissue DNA methylation and gene expression data from our well-phenotyped African American cohort showed that metabolites in the one-carbon metabolism-related pathway, namely SAH, cystine, and hypotaurine, may alter the DNA methylation level of nearly 500 CpG sites and should be considered as epimetabolites in humans. This observation supports our hypothesis and suggests a putative mechanistic role for metabolites in shaping the human epigenome in a key tissue involved in insulin resistance and obesity. However, a limitation of our study is the lack of full coverage of metabolites in the one-carbon metabolism-related pathways. The untargeted metabolomic profiling method in our study covers only a small subset of the metabolites in this pathway and thus lacks information on several putatively key epimetabolites, including S-adenosylmethionine (SAM) (Knight et al., 2018; McCabe et al., 2022). SAM is synthesized from methionine and is considered the main methyl-donor substrate for DNMT-catalyzed DNA methylation reactions (Haws et al., 2020a). Future metaboepigenetic studies with targeted profiling of these metabolites in plasma and in the nuclei of cells in tissues will be crucial to bridging the gap in our knowledge.

Integrating adipose tissue methylation and transcriptome data into cis-eQTM analysis improved our understanding of the target genes for these metabolite-associated CpG sites. Interestingly, 12 genes identified as target genes for these epimetabolite-regulated CpG sites were enriched for the GPCR pathway. Adipose tissue expression levels of all four of these GPCR pathway genes namely, RGS3 (P= 8.44 ×10−12), RAC2 (P= 1.49 ×10−9), RGS10 (P= 1.15 × 10−10) and GNA15 (P= 1.9 ×10−16) were significantly associated with BMI in AAGMEx cohort. Expression of these genes in subcutaneous adipose tissue were also significantly associated with BMI and Matsuda insulin sensitivity index in a METSIM cohort of Finnish Europeans (Civelek et al., 2017). Our cis-eQTM analysis suggests that the SAM-associated CpG sites chr19:3403712 and chr19:3403735 regulate the expression of GNA15. Adipose tissue single-nuclei RNA-seq data suggested that there is higher expression of GNA15 in non-adipocytes (monocytes, macrophages and dendritic cells; Supplementary figure-2) (Emont et al., 2022). We examined our previously published EWAS for BMI to check if these CpG sites were influenced by the adipose tissue cell type heterogeneity. The CpG region chr19:3403001–3404000 encompassing these two CpG sites were also associated with BMI (β = 1.4, P= 3.81 × 10−6), and these associations remain significant (P= 0.0011) even after additionally adjusting for heterogeneity of adipose tissue cell types (Sharma et al., 2020). Thus, cell type heterogeneity had a minor impact on our findings. DNA methylation-mediated epigenetic regulation of GNA15 in adipose tissue was previously reported in MuTHER, a cohort of European ancestry female twins (Allum et al., 2019). Our study suggests a novel molecular link between the SAH level and regulation of GNA15. The SAH is a key determinant of BMI putatively via the DNA-methylation mediated epigenetic transcriptional regulation of GNA15. The GPCRs are cell surface receptors expressed in many tissues including adipose (Im et al., 2021). These receptors are activated by a wide range of extracellular ligands, have critical physiological functions in glucose and energy balance regulation, and contribute to the pathophysiology of metabolic disorders such as diabetes and obesity.

Despite the strong supporting data, this cross-sectional study precludes us from inferring a causal mechanism, necessitating functional validation experiments. A recent study showed that diet and gut microbiota are the strongest predictors of serum metabolites in human, explaining over 50% of the observed variance in some cases (Bar et al., 2020). Diet and microbiota based interventions thus can be applied to manipulate the level of circulating and tissue metabolites. Metabolite mediated epigenetic regulation of GPCRs may be targeted as a therapeutic option for treating obesity in the future.

Supplementary Material

Supplementary Text
Supplementary Tables

Acknowledgement:

We thank the investigators, staff, and participants of the AAGMEx cohort for their valuable contributions.

Funding:

NIH/NIDDK grant R01 DK090111 and R01 DK118243 (PI: SK Das)-Clinical and genomic studies in the AAGMEx cohort. American Diabetes Association Research Grant #1-18-ICTS-113 (PI: SK Das)- DNA methylation data generation and analysis. NIH/NIDDK grant U01 DK105556 and R01 DK066358- Metabolite data generation. North Carolina Diabetes Research Center Grant P30 DK124723 (CD Langefeld, NCDRC data analysis core). Statistical analysis were completed using the Wake Forest University High Performance Computing Facility (DEAC cluster).

Data Availability:

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

References:

  1. Agarwal P, Wicklow BA, Dart AB, Hizon NA, Sellers EAC, McGavock JM, Talbot CPJ, Fonseca MA, Xu W, Davie JR, Jones MJ, Acharjee A and Dolinsky VW (2022) Integrative analysis reveals novel associations between DNA methylation and the serum metabolome of adolescents with type 2 diabetes: A cross-sectional study. Front Endocrinol (Lausanne) 13, 934706. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Akalin A, Kormaksson M, Li S, Garrett-Bakelman FE, Figueroa ME, Melnick A and Mason CE (2012) methylKit: a comprehensive R package for the analysis of genome-wide DNA methylation profiles. Genome Biol 13, R87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Allum F, Hedman Å K, Shao X, Cheung WA, Vijay J, Guénard F, Kwan T, Simon MM, Ge B, Moura C, Boulier E, Rönnblom L, Bernatsky S, Lathrop M, McCarthy MI, Deloukas P, Tchernof A, Pastinen T, Vohl MC and Grundberg E (2019) Dissecting features of epigenetic variants underlying cardiometabolic risk using full-resolution epigenome profiling in regulatory elements. Nat Commun 10, 1209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bar N, Korem T, Weissbrod O, Zeevi D, Rothschild D, Leviatan S, Kosower N, Lotan-Pompan M, Weinberger A, Le Roy CI, Menni C, Visconti A, Falchi M, Spector TD, Adamski J, Franks PW, Pedersen O and Segal E (2020) A reference map of potential determinants for the human serum metabolome. Nature 588, 135–140. [DOI] [PubMed] [Google Scholar]
  5. Betto RM, Diamante L, Perrera V, Audano M, Rapelli S, Lauria A, Incarnato D, Arboit M, Pedretti S, Rigoni G, Guerineau V, Touboul D, Stirparo GG, Lohoff T, Boroviak T, Grumati P, Soriano ME, Nichols J, Mitro N, Oliviero S and Martello G (2021) Metabolic control of DNA methylation in naive pluripotent cells. Nat Genet 53, 215–229. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Christiansen C, Tomlinson M, Eliot M, Nilsson E, Costeira R, Xia Y, Villicaña S, Mompeo O, Wells P, Castillo-Fernandez J, Potier L, Vohl MC, Tchernof A, Moustafa JE, Menni C, Steves CJ, Kelsey K, Ling C, Grundberg E, Small KS and Bell JT (2022) Adipose methylome integrative-omic analyses reveal genetic and dietary metabolic health drivers and insulin resistance classifiers. Genome Med 14, 75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cirulli ET, Guo L, Leon Swisher C, Shah N, Huang L, Napier LA, Kirkness EF, Spector TD, Caskey CT, Thorens B, Venter JC and Telenti A (2019) Profound Perturbation of the Metabolome in Obesity Is Associated with Health Risk. Cell Metab 29, 488–500.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Civelek M, Wu Y, Pan C, Raulerson CK, Ko A, He A, Tilford C, Saleem NK, Stančáková A, Scott LJ, Fuchsberger C, Stringham HM, Jackson AU, Narisu N, Chines PS, Small KS, Kuusisto J, Parks BW, Pajukanta P, Kirchgessner T, Collins FS, Gargalovic PS, Boehnke M, Laakso M, Mohlke KL and Lusis AJ (2017) Genetic Regulation of Adipose Gene Expression and Cardio-Metabolic Traits. Am J Hum Genet 100, 428–443. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Clare CE, Brassington AH, Kwong WY and Sinclair KD (2019) One-Carbon Metabolism: Linking Nutritional Biochemistry to Epigenetic Programming of Long-Term Development. Annu Rev Anim Biosci 7, 263–287. [DOI] [PubMed] [Google Scholar]
  10. Costa Dos Santos G Jr., Renovato-Martins M and de Brito NM (2021) The remodel of the “central dogma”: a metabolomics interaction perspective. Metabolomics 17, 48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Dai Z, Ramesh V and Locasale JW (2020) The evolving metabolic landscape of chromatin biology and epigenetics. Nat Rev Genet 21, 737–753. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Das SK, Ainsworth HC, Dimitrov L, Okut H, Comeau ME, Sharma N, Ng MCY, Norris JM, Chen YI, Wagenknecht LE, Bowden DW, Hsu FC, Taylor KD, Langefeld CD and Palmer ND (2021) Metabolomic architecture of obesity implicates metabolonic lactone sulfate in cardiometabolic disease. Mol Metab 54, 101342. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Donohoe DR and Bultman SJ (2012) Metaboloepigenetics: interrelationships between energy metabolism and epigenetic control of gene expression. J Cell Physiol 227, 3169–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Emont MP, Jacobs C, Essene AL, Pant D, Tenen D, Colleluori G, Di Vincenzo A, Jørgensen AM, Dashti H, Stefek A, McGonagle E, Strobel S, Laber S, Agrawal S, Westcott GP, Kar A, Veregge ML, Gulko A, Srinivasan H, Kramer Z, De Filippis E, Merkel E, Ducie J, Boyd CG, Gourash W, Courcoulas A, Lin SJ, Lee BT, Morris D, Tobias A, Khera AV, Claussnitzer M, Pers TH, Giordano A, Ashenberg O, Regev A, Tsai LT and Rosen ED (2022) A single-cell atlas of human and mouse white adipose tissue. Nature 603, 926–933. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Fluhr L, Mor U, Kolodziejczyk AA, Dori-Bachash M, Leshem A, Itav S, Cohen Y, Suez J, Zmora N, Moresi C, Molina S, Ayalon N, Valdés-Mas R, Hornstein S, Karbi H, Kviatcovsky D, Livne A, Bukimer A, Eliyahu-Miller S, Metz A, Brandis A, Mehlman T, Kuperman Y, Tsoory M, Stettner N, Harmelin A, Shapiro H and Elinav E (2021) Gut microbiota modulates weight gain in mice after discontinued smoke exposure. Nature 600, 713–719. [DOI] [PubMed] [Google Scholar]
  16. Goodrich JM, Hector EC, Tang L, LaBarre JL, Dolinoy DC, Mercado-Garcia A, Cantoral A, Song PX, Téllez-Rojo MM and Peterson KE (2020) Integrative Analysis of Gene-Specific DNA Methylation and Untargeted Metabolomics Data from the ELEMENT Cohort. Epigenet Insights 13, 2516865720977888. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gut P and Verdin E (2013) The nexus of chromatin regulation and intermediary metabolism. Nature 502, 489–98. [DOI] [PubMed] [Google Scholar]
  18. Haws SA, Leech CM and Denu JM (2020a) Metabolism and the Epigenome: A Dynamic Relationship. Trends Biochem Sci 45, 731–747. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Haws SA, Yu D, Ye C, Wille CK, Nguyen LC, Krautkramer KA, Tomasiewicz JL, Yang SE, Miller BR, Liu WH, Igarashi K, Sridharan R, Tu BP, Cryns VL, Lamming DW and Denu JM (2020b) Methyl-Metabolite Depletion Elicits Adaptive Responses to Support Heterochromatin Stability and Epigenetic Persistence. Mol Cell 78, 210–223.e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Im H, Park JH, Im S, Han J, Kim K and Lee YH (2021) Regulatory roles of G-protein coupled receptors in adipose tissue metabolism and their therapeutic potential. Arch Pharm Res 44, 133–145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Jones AC, Ament Z, Patki A, Chaudhary NS, Srinivasasainagendra V, Kijpaisalratana N, Absher DM, Tiwari HK, Arnett DK, Kimberly WT and Irvin MR (2023) Metabolite profiles and DNA methylation in metabolic syndrome: a two-sample, bidirectional Mendelian randomization. Front Genet 14, 1184661. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Knight AK, Park HJ, Hausman DB, Fleming JM, Bland VL, Rosa G, Kennedy EM, Caudill MA, Malysheva O, Kauwell GPA, Sokolow A, Fisher S, Smith AK and Bailey LB (2018) Association between one-carbon metabolism indices and DNA methylation status in maternal and cord blood. Sci Rep 8, 16873. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kohli RM and Zhang Y (2013) TET enzymes, TDG and the dynamics of DNA demethylation. Nature 502, 472–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Krueger F and Andrews SR (2011) Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics 27, 1571–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lechner L, Opitz R, Silver MJ, Krabusch PM, Prentice AM, Field MS, Stachelscheid H, Leitão E, Schröder C, Fernandez Vallone V, Horsthemke B, Jöckel KH, Schmidt B, Nöthen MM, Hoffmann P, Herms S, Kleyn PW, Megges M, Blume-Peytavi U, Weiss K, Mai K, Blankenstein O, Obermayer B, Wiegand S and Kühnen P (2023) Early-set POMC methylation variability is accompanied by increased risk for obesity and is addressable by MC4R agonist treatment. Sci Transl Med 15, eadg1659. [DOI] [PubMed] [Google Scholar]
  26. Lin JG, Kourtis S, Ghose R, Pardo Lorente N, Kubicek S and Sdelci S (2022) Metabolic modulation of transcription: The role of one-carbon metabolism. Cell Chem Biol. [DOI] [PubMed] [Google Scholar]
  27. Ling C and Rönn T (2019) Epigenetics in Human Obesity and Type 2 Diabetes. Cell Metab 29, 1028–1044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Maddocks OD, Labuschagne CF, Adams PD and Vousden KH (2016) Serine Metabolism Supports the Methionine Cycle and DNA/RNA Methylation through De Novo ATP Synthesis in Cancer Cells. Mol Cell 61, 210–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Mattocks DA, Mentch SJ, Shneyder J, Ables GP, Sun D, Richie JP Jr., Locasale JW and Nichenametla SN (2017) Short term methionine restriction increases hepatic global DNA methylation in adult but not young male C57BL/6J mice. Exp Gerontol 88, 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. McCabe CF, LaBarre JL, Domino SE, Treadwell MC, Baylin A, Burant CF, Dolinoy DC, Padmanabhan V and Goodrich JM (2022) Maternal and neonatal one-carbon metabolites and the epigenome-wide infant response. J Nutr Biochem 101, 108938. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Newgard CB (2017) Metabolomics and Metabolic Diseases: Where Do We Stand? Cell Metab 25, 43–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Okut H, Lu Y, Palmer ND, Chen YI, Taylor KD, Norris JM, Lorenzo C, Rotter JI, Langefeld CD, Wagenknecht LE, Bowden DW and Ng MCY (2023) Metabolomic profiling of glucose homeostasis in African Americans: the Insulin Resistance Atherosclerosis Family Study (IRAS-FS). Metabolomics 19, 35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Petersen AK, Zeilinger S, Kastenmüller G, Römisch-Margl W, Brugger M, Peters A, Meisinger C, Strauch K, Hengstenberg C, Pagel P, Huber F, Mohney RP, Grallert H, Illig T, Adamski J, Waldenberger M, Gieger C and Suhre K (2014) Epigenetics meets metabolomics: an epigenome-wide association study with blood serum metabolic traits. Hum Mol Genet 23, 534–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Pillon NJ, Loos RJF, Marshall SM and Zierath JR (2021) Metabolic consequences of obesity and type 2 diabetes: Balancing genes and environment for personalized care. Cell 184, 1530–1544. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Sajuthi SP, Sharma NK, Chou JW, Palmer ND, McWilliams DR, Beal J, Comeau ME, Ma L, Calles-Escandon J, Demons J, Rogers S, Cherry K, Menon L, Kouba E, Davis D, Burris M, Byerly SJ, Ng MC, Maruthur NM, Patel SR, Bielak LF, Lange LA, Guo X, Sale MM, Chan KH, Monda KL, Chen GK, Taylor K, Palmer C, Edwards TL, North KE, Haiman CA, Bowden DW, Freedman BI, Langefeld CD and Das SK (2016) Mapping adipose and muscle tissue expression quantitative trait loci in African Americans to identify genes for type 2 diabetes and obesity. Hum Genet 135, 869–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Sharma NK, Chuang Key CC, Civelek M, Wabitsch M, Comeau ME, Langefeld CD, Parks JS and Das SK (2019) Genetic Regulation of Enoyl-CoA Hydratase Domain-Containing 3 in Adipose Tissue Determines Insulin Sensitivity in African Americans and Europeans. Diabetes 68, 1508–1522. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Sharma NK, Comeau ME, Montoya D, Pellegrini M, Howard TD, Langefeld CD and Das SK (2020) Integrative Analysis of Glucometabolic Traits, Adipose Tissue DNA Methylation, and Gene Expression Identifies Epigenetic Regulatory Mechanisms of Insulin Resistance and Obesity in African Americans. Diabetes 69, 2779–2793. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Sharma NK, Sajuthi SP, Chou JW, Calles-Escandon J, Demons J, Rogers S, Ma L, Palmer ND, McWilliams DR, Beal J, Comeau ME, Cherry K, Hawkins GA, Menon L, Kouba E, Davis D, Burris M, Byerly SJ, Easter L, Bowden DW, Freedman BI, Langefeld CD and Das SK (2016) Tissue-Specific and Genetic Regulation of Insulin Sensitivity-Associated Transcripts in African Americans. J Clin Endocrinol Metab 101, 1455–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Sharma U and Rando OJ (2017) Metabolic Inputs into the Epigenome. Cell Metab 25, 544–558. [DOI] [PubMed] [Google Scholar]
  40. Shyh-Chang N, Locasale JW, Lyssiotis CA, Zheng Y, Teo RY, Ratanasirintrawoot S, Zhang J, Onder T, Unternaehrer JJ, Zhu H, Asara JM, Daley GQ and Cantley LC (2013) Influence of threonine metabolism on S-adenosylmethionine and histone methylation. Science 339, 222–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Suhre K and Zaghlool S (2021) Connecting the epigenome, metabolome and proteome for a deeper understanding of disease. J Intern Med 290, 527–548. [DOI] [PubMed] [Google Scholar]
  42. van der Knaap JA and Verrijzer CP (2016) Undercover: gene control by metabolites and metabolic enzymes. Genes Dev 30, 2345–2369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Xu P, Wang M, Sharma NK, Comeau ME, Wabitsch M, Langefeld CD, Civelek M, Zhang B and Das SK (2023) Multi-omic integration reveals cell-type-specific regulatory networks of insulin resistance in distinct ancestry populations. Cell Syst 14, 41–57.e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Zaghlool SB, Mook-Kanamori DO, Kader S, Stephan N, Halama A, Engelke R, Sarwath H, Al-Dous EK, Mohamoud YA, Roemisch-Margl W, Adamski J, Kastenmüller G, Friedrich N, Visconti A, Tsai PC, Spector T, Bell JT, Falchi M, Wahl A, Waldenberger M, Peters A, Gieger C, Pezer M, Lauc G, Graumann J, Malek JA and Suhre K (2018) Deep molecular phenotypes link complex disorders and physiological insult to CpG methylation. Hum Mol Genet 27, 1106–1121. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Text
NIHMS2188976-supplement-Supplementary_Text.pdf (273.2KB, pdf)
Supplementary Tables
NIHMS2188976-supplement-Supplementary_Tables.xlsx (2.7MB, xlsx)

Data Availability Statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

RESOURCES