GRAPHICAL ABSTRACT

Highlights

• Our study links DNA methylation to the hyperinsulinemic-euglycemic clamp in men.

• A multiomics approach identified novel genes linked to hyperinsulinemia.

• DNA methylation in gene bodies highlights its previously overlooked role in hyperinsulinemia.

INTRODUCTION

Hyperinsulinemia is a major feature in the pathophysiology of insulin resistance and is associated with various metabolic disorders (e.g., obesity, metabolic syndrome, and type 2 diabetes mellitus [T2DM]) and other related conditions such as chronic inflammation and cancer [1,2]. The interplay between hyperinsulinemia and these conditions is complex, with each condition potentially exacerbating the other [2]. Insulin and insulin-like growth factors (IGFs) activate overlapping signaling pathways in metabolism and growth [1-3]. Insulin binds to its receptor tyrosine kinase insulin receptor (INSR) and to IGF1 receptor (IGF1R) with half-maximal inhibitory concentrations of 0.89 and 30–400 nM, respectively [2]. At high concentrations, insulin can bind to IGF1R, and many cells including immune cells express both receptors which can form hybrid receptors [2,4].

Furthermore, under hyperinsulinemic conditions, insulin resistance is associated with selective alterations in insulin signaling pathways in various tissues, including immune cells [4]. Although the metabolic actions of insulin may be impaired, certain growth-related signaling pathways, such as those involving IGF1 and insulin/IGF1 hybrid receptors, can remain active or become overactive [2]. These selective and different signaling pathways can drive various conditions in cells, such as metabolic dysfunction and cell proliferation or can create an environment conducive to tumorigenesis [2]. Insulin infusion during hyperinsulinemic-euglycemic clamp (HEC) increases the levels of pro-inflammatory cytokines in human serum and adipose tissue [5,6]. INSR mRNA is expressed in multiple immune cell types based on single-cell RNA sequencing data (scRNA-seq) [2], suggesting that at least some effects of hyperinsulinemia are direct.

The role of epigenetic regulators, including DNA methylation, in the progression of T2DM and metabolic dysfunctions has received increased attention [7-9]. Notably, in INSR- versus IGF1R-expressing pre-adipocytes, there was increased basal and insulin-stimulated phosphorylation of proteins involved in chromatin modification, including ARID1A^{S697, S699, S703}, ARID5B^S1002, SMARCA4^S1384, and p300^{S2306, 2322}. ARID1A, ARID5B, and SMARCA4, which are involved in chromatin remodeling and critical for metabolic reprogramming and adaptation to nutritional signals [3].

Prolonged hyperinsulinemia-induced insulin resistance may be sustained through epigenetic changes even after alleviating hyperinsulinemia [8,10]. These findings suggest that hyperinsulinemia can induce long-lasting changes in cellular behavior that may be reversed through targeted epigenetic interventions [10].

However, whether and how short-term hyperinsulinemia, particularly in non-diabetic individuals, affects DNA methylation and gene expression across different genomic regions, such as promoters and gene bodies, remain unclear [8,9,11-13]. Analyzing changes in the epigenome and transcriptome of peripheral blood cells upon exposure to acute prolonged hyperinsulinemia (e.g., HEC) in individuals with normal or near-normal insulin sensitivity could improve the understanding of the action of insulin on the epigenome and transcriptome of human cells [10].

In the present study, we evaluated alterations in the epigenetic and transcriptomic profiles of peripheral blood cells exposed to acute prolonged hyperinsulinemia during an HEC study in healthy men and identified methylome-related changes in gene expression with pathophysiological implications.

METHODS

Study subjects and HEC

The study subjects were selected from a prior HEC study cohort evaluating insulin sensitivity in young and middle-aged Koreans. Of the 23 subjects included in the original clamp study [12], five adult men without diabetes were chosen for this analysis based on the availability of clinical and insulin sensitivity data, as well as blood samples suitable for epigenetic and transcriptomic studies (detailed in Supplementary Methods). Two-stage HEC studies were conducted as previously described (Fig. 1A and B) [12]. The study was approved by the Institutional Review Board of the Gil Medical Center (approval number: GCIRB2016-226) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients.

mRNA sequencing and targeted bisulfite sequencing

mRNA sequencing (mRNA-seq) and targeted bisulfite sequencing (methyl-seq) were performed using standard protocols (detailed in Supplementary Methods). Briefly, mRNA expression was quantified using TruSeq kits followed by Illumina (San Diego, CA, USA) sequencing, processed with FastQC, STAR, and RSEM. DNA methylation was analyzed using Sure-Select^XT (Agilent Technologies, Santa Clara, CA, USA) and Bismark alignment to hg19.

mRNA-seq and methyl-seq data analyses

To analyze differentially expressed genes (DEGs) and quantify methylation, mRNA-seq data (mRNA expression, n=10) and methyl-seq data (DNA methylation, n=10) were utilized. Through methyl-seq of subjects before and after HEC, 2,674,456 common cytosine-phosphate-guanine (CpG) sites were obtained across subjects (Fig. 1C). As a result, 697 DEGs were selected with significant differences (P<0.05) from DEG analysis and 65,389 CpG sites corresponding to these 697 DEGs with DNA methylation values (β-values) were obtained using the methyl-seq data (Fig. 1C).

Selection method 1: methylation and gene expression association

Prior research revealed four biological patterns in which DNA methylation changes in the promoter and gene body regions correspond to gene expression alterations, classified as patterns C1 through C4 (Fig. 1D and E) [9,11,13]. Selection method 1 was developed to systematically identify genes in which methylation changes aligned with significant gene expression shifts according to these patterns. This method first defines specific graphical profiles for each pattern, allowing the visualization of methylation levels across individual CpG sites (graph profiles), along with corresponding gene expression changes (bar graph profiles) (Fig. 1F).

Selection method 1 was conducted using the following criteria: (1) differences between the groups before and after HEC in the average DNA methylation values (β-values) of individual CpG sites within specific regions (i.e., promoter or gene body) and differences in gene expression levels belong to patterns C1 through C4 shown in Fig. 1F; (2) |log₂(fold-change of gene expression after vs. before HEC)| >1, ensuring the biological relevance of transcriptional shifts; and (3) |average methylation differences for a gene region| >2%, ensuring methylation shifts. Selection method 1 enabled identification of genes for which hyperinsulinemia-induced epigenetic modifications were likely to correspond with meaningful transcriptional shifts.

Selection method 2: methylation and gene expression correlation

Selection method 2 was developed to assess the correlation between methylation and expression changes and identify genes with region-specific methylation patterns that may influence gene activity. To perform selection method 2, linear models were constructed using the methylation and expression information for each subject. For a given gene, the differences in methylation and expression before and after HEC for each subject were calculated by dividing the DNA methylation regions of a given gene into two distinct categories: promoter and gene body regions. For each category, the difference in methylation was obtained from the difference between the average DNA methylation values before and after HEC.

Differences in gene expression were calculated as the difference in gene expression values (transcripts per million) before and after HEC for each subject. Expression and methylation differences were set as the response and explanatory variables, respectively, and a linear model was constructed. Genes with a P<0.05 for the coefficient of the explanatory variable were selected. Among these, genes with a positive correlation between methylation and gene expression differences in the gene body region and a negative correlation in the promoter region were identified (Fig. 1G).

Classification and curation of gene sets from the 697 DEGs

Among the 697 DEGs identified via mRNA-seq analysis, we detected 119 DEGs exhibiting methylation changes, obtained using selection methods 1 and 2, as ‘methyl-DEGs.’ These methyl-DEGs were subsequently divided into two subsets: ‘methyl-DEGs-p’ (59 DEGs associated with promoter regions) and ‘methyl-DEGs-g’ (60 DEGs associated with gene body regions). Concurrently, the initial 697 DEGs overlapped with four diabetes-associated traits that we considered important (inflammation, oxidative stress, insulin signaling, and carbohydrate metabolism) using a publicly available gene set database, the Molecular Signatures Database (MSigDB) [14], and relevant curated publications (Supplementary Table 1) [15-17]. Consequently, 45 DEGs (hereafter, ‘public-trait-DEGs’) were aligned with publicly available gene sets associated with these four traits. Furthermore, the intersection of methyl-DEGs and public-trait-DEGs identified six overlapping genes (hereafter, ‘common-DEGs’).

Networks and functional analysis for methyl-DEGs-p and methyl-DEGs-g

First, gene networks were constructed for the methyl-DEG-p and methyl-DEG-g gene sets selected from each region (promoter and gene body) using the STRING protein-protein interaction (PPI) database. Concurrently, ingenuity pathway analysis (IPA; QIAGEN, Hilden, Germany) was used to examine the functional outcomes for the methyl-DEG-p and methyl-DEG-g gene sets. Cytoscape was used to integrate the networks from STRING and functional results from IPA to visualize the interconnections between related genes.

Networks for methyl-DEGs and public-trait-DEGs

To investigate the connectivity between public-trait-DEGs (n=45) and methyl-DEGs (n=119), Cytoscape, and STRING were used.

Experimental validation using THP1 and 3T3–L1 cells

Changes in the expression of specific genes in response to hyperinsulinemic stimulation were validated in THP1 (a human monocyte cell line) and 3T3–L1 cells (a mouse preadipocyte cell line). Cells were cultured as described in Supplementary Methods. Briefly, THP1 cells were cultured in Roswell Park Memorial Institute (RPMI)-1640 supplemented with 10% fetal bovine serum, serum-starved overnight, and treated with 2 nM insulin for 3 hours. Confluent 3T3–L1 pre-adipocytes in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal calf serum were serum-starved and exposed to insulin. Total RNA was isolated, cDNA was synthesized, and mRNA was quantified using quantitative polymerase chain reaction (Supplementary Table 2) and normalized to the ribosomal protein lateral stalk subunit P0 (RPLP0) gene.

Statistics

Baseline demographic and clinical data are presented as the mean±standard deviation. A significance test for the DEGs was performed using Wald test in the DESeq2 package [18]. Statistical significance was determined using F-test with P<0.05, which was considered significant and used in selection method 2. In IPA, Fisher’s exact test was used with P<0.05 considered significant.

RESULTS

Acute hyperinsulinemia exposure in study subjects

Two-stage HEC studies were performed in five adult men without diabetes; the M-values and clinical information of the study subjects are summarized in Table 1. During the HEC study, serum glucose concentrations in the study subjects remained between 89 and 109 mg/dL, with a mean value of 100.3±12.5 mg/dL. Serum insulin concentrations were 27.6±5.8 mU/L at 100 minutes and 237.5±46.6 mU/L at 180 minutes (Fig. 1A and B). Comparison of the mRNA-seq profiles of blood cells exposed to acute hyperinsulinemia for at least 100 minutes during HEC with those at baseline (0 minute) revealed 697 DEGs (details in Methods and Supplementary Methods). Two selection methods were used (detailed in Methods) to identify DEGs with DNA methylation changes, yielding 119 methyl-DEGs (Fig. 1C).

Identifying genes associated with their DNA methylation and gene expression changes in the promoter region

Among the 119 methyl-DEGs, 30 genes selected using selection method 1 were related to the promoter region (Fig. 2A). Of these, 19 genes, including CDX4, CEL, GRID1, and MTUS1 (Fig. 2B-E), showed pattern C1 (pattern C1 in Fig. 1D), in which gene expression increased when CpG site methylation values decreased (Supplementary Table 3, Supplementary Fig. 1A). In contrast, 11 genes, including CDO1, ESR1, OR6M1, and PRSS3 (Fig. 2F-I), showed pattern C2 (pattern C2 in Fig. 1D) (Supplementary Table 3, Supplementary Fig. 1B). Additionally, 28 genes showed a positive association between changes in promoter methylation and changes in gene expression (Supplementary Table 4, Supplementary Fig. 2).

When selection method 2 was applied to the 697 DEGs to identify genes with significant correlation coefficients between changes in promoter methylation values and gene expression, 29 genes were identified (Fig. 2J). Among these 29 genes, 18 exhibited negative associations, including ADCY4, NMNAT1, PRKAR1B, RFX1, RNF5, TRIM50, VBP1, and XIST (Fig. 2K-R and Supplementary Fig. 3).

Overall, 59 methyl-DEGs-p, for which DNA methylation changes in the promoter regions were associated with gene expression changes, were detected using selection methods 1 and 2 (Supplementary Table 3).

Identifying genes associated with their DNA methylation and gene expression changes in the gene body region

Using selection method 1, 30 genes with DNA methylation changes in gene body regions associated with gene expression changes were identified (Fig. 3A). Among these, eight genes, including FGF4 and SERTAD3 (Fig. 3B and C), exhibited decreased expression with hypomethylation (pattern C3 in Fig. 1E) (Supplementary Table 5, Supplementary Fig. 4A). In contrast, 22 genes, including ADAMTS18, GAPT, and GAST (Fig. 3D-F), showed the opposite pattern (pattern C4 in Fig. 1E), with increased gene expression associated with hypermethylation (Supplementary Table 5, Supplementary Fig. 4B). An additional ten genes displayed negative associations between methylation value changes and expression changes in the gene body region (Supplementary Table 6, Supplementary Fig. 5).

Using selection method 2, the correlation between changes in methylation values in the gene body regions and changes in gene expression was examined, revealing 30 genes with significant correlations (Fig. 3G). DNA methylation changes in gene bodies positively regulate associated gene expression changes [9,11,13]. Consistently, 17 of these 30 genes exhibited positive associations, including C19orf54, CEL, OR5P3, TNFSF4, and ZNF677 (Fig. 3H-L, and Supplementary Fig. 6).

Overall, selection methods 1 and 2 detected 60 methyl-DEG-g in which DNA methylation changes in the gene body correlated with mRNA expression changes (Supplementary Table 5). Among the identified genes, some genes like GAPT and GAST influence insulin action and key metabolic pathways [19,20].

Among the 119 methyl-DEGs, OR6B1 appeared in both selection methods within the gene body region, while three genes (TNFSF4, CEL, and CXorf27) were identified by both methods in different regions. Some genes showed opposite methylation patterns between promoter and gene body regions, exemplified by KDM4E (decreased promoter/increased gene body methylation) and RP11-486L19.2 (increased promoter/decreased gene body methylation) (Supplementary Tables 3 and 5).

Relationship between methyl-DEGs and diabetes-related traits

After obtaining methyl-DEG-p and methyl-DEG-g, we functionally annotated these gene sets using PPI analysis and IPA [21]. Among methyl-DEGs-p, eight genes (CDO1, CYP4F2, ESR1, MTUS1, PTPRS, PRKAR1B, TNFSF4, and TNFSF14) were linked to ‘inflammatory response,’ ‘quantity of IGF1 in blood,’ and ‘concentration of D-glucose’ (Fig. 4A). Of the eight genes, CDO1 and ESR1 appeared in all three IPA results, confirming that they are more likely to be associated with insulin action or diabetes than are other genes.

Similarly, the PPI and IPA functional annotations of methyl-DEG-g were enriched for insulin action and diabetes-related terms, such as ‘diabetic nephropathy,’ ‘glucose metabolism disorders,’ and ‘signal transduction.’ Twenty genes (ATOH7, CEL, CPZ, EPOR, LY6G6C, PER1, STRA6, FGF4, GAST, RGS5, TNFSF4, CDKL2, ITGB1BP2, OR5M8, OR5P3, OR6B1, OR10A4, OR11G2, RSPO4, and OR51I1) in methyl-DEGs-g were associated with these terms (Fig. 4B). Of these, EPOR was associated with all three terms. FGF4 and TNFSF4 were associated with two terms, ‘glucose metabolism disorders’ and ‘signal transduction,’ and PER1 was associated with two terms, ‘diabetic nephropathy’ and ‘glucose metabolism disorders’ (Fig. 4B).

As described in the Methods, 697 initial DEGs from mRNA-seq data were overlapped to four diabetes-associated traits (i.e., inflammation, insulin signaling, carbohydrate metabolism, and oxidative stress), yielding 45 ‘public-trait-DEGs.’ The intersection of methyl-DEGs and public-trait-DEGs identified six common-DEGs (B3GALNT1, CDO1, ESR1, PER1, PRKAR1B, and TNFSF4) that serve as potential links between epigenetic modifications and diabetes-associated traits (Supplementary Fig. 7). Network analysis of 45 public-trait-DEGs and 119 methyl-DEGs was then performed using the PPI database STRING, which generated a complex network within the curated gene set, as shown in Fig. 4C. Among these, five of the six common-DEGs were present in the networks, with FGF4 displaying the largest number of interactions among methyl-DEGs and publictrait-DEGs. Consequently, these five common-DEGs, along with FGF4, were considered as key epigenetically regulated genes in response to acute hyperinsulinemia in blood cells, with potential pathophysiological relevance to diabetes-related traits.

In the network shown in Fig. 4C, public-trait-DEGs in the dashed-line area showed increased expression after HEC, whereas some public-trait-DEGs not in the dashed-line area showed decreased expression. Except for TNFSF4, all common-DEGs showed reduced expression in HEC. Among the common-DEGs, ESR1 showed the largest number of interactions with public-trait-DEGs across various categories like ‘insulin signaling’ and ‘inflammation’ (Fig. 4C). Of the 17 methyl-DEGs constituting the PPI network, six genes showed decreased expression after HEC, among which FGF4 with pattern C4 displayed the largest number of interactions.

Validation study using public datasets

As described in the previous section, we confirmed that the five common-DEGs (B3GALNT1, ESR1, PER1, PRKAR1B, and TNFSF4) and FGF4 were strongly associated with genes related to insulin action and diabetes (public-trait-DEGs). To validate the expression patterns of these genes, we leveraged publicly available datasets (GEO accessions GSE9006 [22] and GSE153837 [23]) to examine mRNA expression patterns in peripheral blood cells. The group composition of GSE153837 was compared between fasting and 2 hours after glucose ingestion using blood samples from normal control subjects and patients with prediabetes, while the group composition of GSE9006 was blood samples from normal control subjects and patients with T2DM (Supplementary Fig. 8). The results from the publicly available datasets confirmed that PER1 (Supplementary Fig. 8A), FGF4 (Supplementary Fig. 8B), and TNFSF4 (Supplementary Fig. 8C) genes showed gene expression patterns consistent with our findings, which may be related to postprandial insulin action.

To further address cell type specificity, we utilized a publicly available scRNA-seq dataset (GEO accession: GSE255566) from peripheral blood mononuclear cells of individuals with and without T2DM (Supplementary Fig. 9) [24]. Although this dataset represents different experimental conditions from our study (after vs. before HEC in healthy subjects), we found that PER1 was differentially expressed in three cell types (dendritic cells, monocytes, and natural killer [NK] cells) with statistical significance in monocytes and NK cells (Supplementary Fig. 10).

Gene expression changes in blood cells of study subjects after HEC and in THP1 and 3T3–L1 cells after in vitro insulin treatment

Alterations of the expression levels of the above-mentioned six genes in our mRNA-seq data are shown in Fig. 5A. We tested whether the expression of these confirmed six genes was altered in response to in vitro acute insulin treatment in THP1 and 3T3–L1 cells. Similar expression patterns as in Fig. 5A were observed in these cells (Fig. 5B and C), with slightly variable statistical significance. The gene expression levels of ESR1, PRKAR1B, PER1, and B3GALNT1 were reduced in THP1 cells after insulin treatment (Fig. 5B). In 3T3–L1 preadipocytes, the results were similar, although B3GALNT1 expression was not significantly different (Fig. 5C).

DISCUSSION

In the present study, we examined whether acute hyperinsulinemia during HEC could alter DNA methylation, leading to changes in gene expression in the blood cells of adult men without diabetes. We selected specifically non-diabetic healthy adult subjects to isolate the direct effects of acute hyperinsulinemia on epigenetic modifications, independent of underlying insulin resistance or diabetes-related metabolic dysfunction. The M-value of our subjects (9.35±3.52 mg/kg/min) and homeostatic model assessment for insulin resistance (HOMA-IR; 1.3±0.3) confirmed normal insulin sensitivity, considering suggested cut-off values for insulin resistance using M-value are 4.7 or 5.6 mg/kg/min [25,26]. These parameters were consistent with our previous study reporting an M-value of 9.9±2.7 mg/kg/min in non-diabetic Korean males of similar age [12]. There is also the HOMA-IR, which is a measure of insulin resistance, but generally 2.0 or 2.5 is the cutoff [27,28]. This study design allowed us to examine whether epigenetic changes are caused by hyperinsulinemia alone and whether this leads to changes in mRNA expression, without the confounding effects of chronic insulin resistance that characterize T2DM.

Studies have shown that peak postprandial insulin levels frequently approach or exceed 200 mU/L (approximately 1,389 pmol/L) in individuals with obesity [29]. Serum insulin levels achieved in our clamp study thus could occur in specific clinical settings, suggesting a clinically relevant insight into epigenetic changes under acute hyperinsulinemia from our study results.

Changes in gene expression in response to DNA methylation were found to differ depending on the methylation site of the gene (i.e., promoters and gene bodies) [9,11,13,30]. Specifically, 59 methyl-DEG-p and 60 methyl-DEG-g were identified, resulting in 119 methyl-DEGs. Furthermore, 697 DEGs in response to acute hyperinsulinemia were identified via simultaneous mRNA-seq, further narrowing the DEG list to 45 diabetes-related public-trait-DEGs.

Overlapping genes between our 119 methyl-DEGs and 45 public-trait-DEGs were identified using gene sets related to insulin action and diabetes. Five common-DEGs (B3GALNT1, ESR1, PER1, PRKAR1B, and TNFSF4) and FGF4, showing the largest number of interactions among the methyl-DEGs, were found. The expression of these genes was also reduced after HEC according to our mRNA-seq data, except for that of TNFSF4 (Fig. 5A). These results were consistent with those obtained using the in vitro THP1 and 3T3–L1 cell lines. Interestingly, ESR1 not only constituted many of the networking edges in the PPI network (Fig. 4C) but also showed decreased expression in response to insulin stimulation in both THP1 and 3T3–L1 cells. ESR1 (Fig. 2G) encodes an estrogen receptor involved in insulin action [31], with its deficient models showing impaired insulin sensitivity [32,33] and previous research demonstrating insulin-induced ESR1 methylation and decreased expression [34].

FGF4 (Fig. 3B), along with FGF1, recently gained attention as an important regulator of energy homeostasis and glucose/lipid metabolism beyond its classical function as a growth factor. Paracrine FGF4 acts as a potent antihyperglycemic disease factor in diabetic models [35]. Recombinant FGF4 improved insulin resistance and inhibited adipose macrophage infiltration and inflammation [35]. Therefore, considering the decreased expression of FGF4 after hyperinsulinemia exposure in our study, DNA methylation-mediated regulation of FGF4 expression appears to be an important epigenetic regulatory mechanism.

B3GALNT1 is associated with glycan biosynthesis [36], and its expression in the pancreatic islets of mice with T2DM is decreased compared to that in normal mice [36,37]. PRKAR1B is differentially expressed in β-cells of patients with T2DM and is involved in protein kinase A, androgen, and cardiac hypertrophy signaling [38].

TNFSF4 exhibited a positive correlation between DNA methylation and gene expression. Given the discrepant results observed between human blood cells and mouse preadipocytes after insulin stimulation, as well as evidence suggesting methylation changes of TNFSF4 in inflammation-related diseases such as inflammatory bowel disease [39], further investigation of TNFSF4 is needed to determine its effects on insulin and inflammation in diabetes.

PER1 is a core clock gene that may be dysregulated in older adults and those with metabolic disease genes [40]. In line with our results, PER1 expression has been reported to decrease after meals in both healthy people and those with T2DM, which may be regulated by postprandial insulin [41]. PER1 demonstrated consistent and significant downregulation across our clamp study, the cell lines, and the validation datasets, highlighting it as a more robust and conserved responder to acute hyperinsulinemia [40,41].

The functional implications of intracellular protein changes encoded by B3GALNT1, ESR1, PER1, and PRKAR1B in blood cells may differ from those of secretory factors (FGF4, TNFSF4). Altered intracellular proteins can modify immune cell behavior, activation states, and metabolic responses, which subsequently affect systemic physiology. For instance, PER1 dysregulation can disrupt circadian regulation in immune cells, potentially leading to altered cytokine production timing and inflammatory responses that impact peripheral tissue metabolism [42-44].

In examining systemic blood-based methylation signatures in T2DM, the most recent systematic review published in 2024 [45] confirms that whole-blood methylation profiles are heavily influenced by immune-cell composition—a factor critical for interpreting locus-specific DNA methylation alterations (e.g., TXNIP, ABCG1, PTPRN2) [46,47]. Although our target genes were not among the commonly reported hits in T2DM, this underscores the importance of considering methylation changes as potentially cell-type–driven phenomena.

In addition to the common-DEGs, many genes showed distinct changes in DNA methylation and expression after acute hyperinsulinemia (patterns C1–C4 in Fig. 2). CEL, listed in both methyl-DEGs-p and methyl-DEGs-g, is associated with ‘maturity-onset diabetes of the young, type 8’ [48]. The mRNA expression of CDO1 (Fig. 2F) was negatively correlated with inflammatory markers [49]. In line with this report, decreased CDO1 expression with acute hyperinsulinemia and increased expression of inflammation-related genes among public-trait-DEGs were observed in our study. GAPT (Fig. 3E) binds GRB2 and may be involved in regulating insulin signaling pathways [19]. GAST (Fig. 3F) encodes gastrin, and increased serum gastrin levels are associated with improved glycemic control in diabetes, suggesting a role in glucose regulation [50]. We observed increased GAST expression accompanied by DNA methylation changes after HEC, warranting further investigation of the insulin-gastrin regulatory relationship in glucose homeostasis.

Our study had several limitations. First, this was a single-center study with a limited sample size of five subjects. Second, although we obtained results supporting those of omics analysis through cell experiments, systematic experiments to investigate the functions of the key genes could not be conducted. Third, changes in individuals with severe insulin resistance or T2DM were not considered. Fourth, we did not determine the duration of the epigenetic and transcriptomic changes observed in our post-HEC study. Further studies are needed to determine whether such epigenetic regulation of some important genes occurs only in response to the fasting-feeding cycle or whether it exerts a more durable effect through repeated influences. Despite extensive searches, no cell-type–specific DNA methylation data in blood cells during HEC in healthy or T2DM individuals exist. As the first study in the field, this absence underscores both a limitation and the pioneering nature of our work.

In conclusion, our results showed that DNA methylation was induced in the blood after acute short-term (approximately 100 minutes) hyperinsulinemia, accompanied by altered expression of many genes. These findings provide insights into insulin-mediated epigenetic changes in blood cells, though their impact on metabolism requires further investigation in various insulin-responsive tissues as well.

SUPPLEMENTARY MATERIALS

NOTES

CONFLICTS OF INTEREST

Dae Ho Lee has been an international editorial board of the Diabetes & Metabolism Journal since 2023. He was not involved in the review process of this article. Otherwise, there was no conf lict of interest.

AUTHOR CONTRIBUTIONS

Conception or design: S.N., D.H.L.

Acquisition, analysis, or interpretation of data: M.J., X.T.T.

Drafting the work or revising: M.J., D.S., S.N., D.H.L.

Final approval of the manuscript: all authors.

FUNDING

This study was supported by grants from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health and Welfare, Korea (No. HI14C1135 to Dae Ho Lee) and the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education (No. NRF-2019R1I1A2A02062305 to Dae Ho Lee), a National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. NRF-2021R1 A5A2030333 to Dae Ho Lee), the Gachon University Gil Medical Center (FRD2021-03 to Dae Ho Lee), and a National Research Council of Science & Technology (NST) grant from the Korean government (MSIT) (No. GTL24021-000 to Seungyoon Nam).

ACKNOWLEDGMENTS

The mRNA-seq and methyl-seq data generated in this study were deposited in the BioProject of the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/bioproject; BioProject accessions PRJNA1173800 and PRJNA1175431).

Fig. 1.

Measurement of glucose and insulin during hyperinsulinemic-euglycemic clamp (HEC) and analysis flow using mRNA sequencing (mRNA-seq) and targeted bisulfite sequencing (methyl-seq) to observe changes in gene expression patterns based on methylation changes in different gene regions (i.e., promoters, gene bodies) before and after HEC. (A, B) Time series plot of glucose and insulin concentrations measured during HEC in non-diabetic subjects. (A) Glucose concentration was measured every 5 minutes in plasma with two instruments, and the values were averaged. (B) Insulin was measured using serum at 10 and 5 minutes before performing HEC; 80, 90, and 100 minutes after HEC; and 180, 190, and 200 minutes after HEC. (C) Schematic of the analysis using data from mRNA-seq and methyl-seq. (D) Individual cytosine-phosphate-guanine (CpG) sites in promoter regions and gene body regions showing changes in gene expression patterns based on methylation value changes before and after HEC. Decreased methylation values at CpG sites in promoter regions after HEC lead to increased gene expression (pattern C1), whereas increased methylation values lead to decreased gene expression (pattern C2). (E) Decreased methylation values at CpG sites in gene body regions lead to decreased gene expression (pattern C3), whereas increased methylation values lead to increased gene expression (pattern C4). (F) Selection method 1 predefines the graph profiles corresponding to the patterns in Fig. 1D and E as C1 through C4. Among genes showing significant changes in expression levels before and after HEC, those matching the specific graph profiles are selected. (G) Selection method 2 uses a linear model to identify genes that exhibit positive or negative correlations between regional methylation differences and gene expression differences before and after HEC across subjects (P<0.05), aiming to find genes that correspond to the patterns shown in Fig. 1D and E. DEG, differentially expressed gene; TSS, transcription start site; TPM, transcripts per million.

Fig. 2.

Genes with significant promoter region-related results were selected from the analysis using mRNA sequencing (mRNA-seq) and targeted bisulfite sequencing (methyl-seq). (A) Scatter plot using the average methylation difference (AMD) (x-axis) before and after hyperinsulinemic-euglycemic clamp (HEC) in the promoter region of each gene and log₂(fold-change) (y-axis) of the corresponding genes (i.e., CDX4, CEL, GRID1, MTUS1, CDO1, ESR1, OR6M1, and PRSS3). (B–I) Line graphs (left panel) plotting DNA methylation (β-values; y-axis) at individual cytosine-phosphate-guanine (CpG) sites (x-axis) before (red) and after (blue) HEC for genes identified using selection method 1: (B) CDX4, (C) CEL, (D) GRID1, (E) MTUS1, (F) CDO1, (G) ESR1, (H) OR6M1, and (I) PRSS3. Boxplots (right panel) comparing gene expression levels before and after HEC. Gray rectangles below the x-axis indicate the promoter region range for each gene. Circles represent CpG sites. (J) Scatter plot of the correlation (adjusted R², x-axis) between DNA methylation differences and gene expression differences and –log₁₀(P values) (y-axis) for the linear model results from selection method 2. (K–R) Among the genes indicated in (J), those with a negative correlation: (K) ADCY4, (L) NMNAT1, (M) PRKAR1B, (N) RFX1, (O) RNF5, (P) TRIM50, (Q) VBP1, and (R) XIST. M_after-M_before on the x-axis indicates methylation value (β-value), where M_after is the M after HEC and M_before is the M before HEC. E_after-E_before on the y-axis indicates the gene expression value (transcripts per million [TPM]), where E_after is the E after HEC and E_before is the E before HEC. ^aP<0.05, ^bP<0.01, ^cP<0.001.

Fig. 3.

Genes with significant results related to the gene body regions were selected through mRNA sequencing (mRNA-seq) and targeted bisulfite sequencing (methyl-seq) analysis. (A) Scatter plot using the average methylation difference (AMD) (x-axis) between before and after hyperinsulinemic-euglycemic clamp (HEC) in the gene body region of each gene and log₂(fold-change) (y-axis) of the corresponding genes (i.e., FGF4, SERTAD3, ADAMTS18, GAPT, and GAST). (B–F) For (B) FGF4, (C) SERTAD3, (D) ADAMTS18, (E) GAPT, and (F) GAST discovered using selection method 1, line graphs (left panel) were drawn using DNA methylations (β-values; y-axis) before (red) and after (blue) HEC at individual cytosine-phosphate-guanine (CpG) sites (x-axis), and boxplots of gene expression before and after HEC (right panel) are shown. In the line graphs, the gray rectangles below the x-axis indicate the range of the gene body region of each gene, and the circles represent CpG sites. (G) Scatter plot using the correlation (adjusted R², x-axis) between DNA methylation differences and gene expression differences, which is the result of the linear model of selection method 2, and the log₁₀(P value) (y-axis) of significance of the linear correlation. (H–L) Among the genes indicated in (G), those with a positive correlation: (H) C19orf54, (I) CEL, (J) OR5P3, (K) TNFSF4, and (L) ZNF677. M_after-M_before on the x-axis indicates the methylation value (β-value), where M_after is the M after HEC and M_before is the M before HEC. E_after-E_before on the y-axis indicates the gene expression value (transcripts per million [TPM]), where E_after is the E after HEC and E_before is the E before HEC. ^aP<0.05, ^bP<0.01, ^cP<0.001.

Fig. 4.

Networks of selected genes are associated with insulin action and diabetes. (A) Protein-protein interaction (PPI) networks of methyl-differentially associated genes (DEGs) associated with promoter regions (methyl-DEGs-p) were constructed using STRING and Cytoscape. According to functional analysis using ingenuity pathway analysis (IPA) of the genes belonging to these networks, six genes were associated with ‘inflammatory response’ (beige), four genes were associated with ‘quantity of IGF1 in blood’ (ash gray), and three genes were associated with ‘concentration of D-glucose’ (light purple). (B) PPI networks of methylDEGs associated with gene body regions (methyl-DEGs-g) were constructed using STRING and Cytoscape. According to functional analysis performed using IPA on the PPI networks of methyl-DEGs-g, seven genes were associated with ‘diabetic nephropathy’ (beige), 12 genes were associated with ‘glucose metabolism disorder’ (light purple), and 13 genes were associated with ‘signal transduction’ (ash gray). (C) PPI networks were created using STRING and Cytoscape by integrating methyl-DEGs (n=119) and public-trait-DEGs (n=45). B3GALNT1, ESR1, PER1, PRKAR1B, and TNFSF4 (orange circle line) belonging to common-DEGs and FGF4 (deep blue circle line) with the largest number of connections among methyl-DEGs appeared. Genes connected to these genes are shown with red edges.

Fig. 5.

Effect of insulin treatment on the expression of predicted genes. (A) Gene expression before (n=5) and after (n=5) hyperinsulinemic-euglycemic clamp (HEC) for ESR1, PRKAR1B, PER1, B3GALNT1, FGF4, and TNFSF4 were compared via boxplot. (B) THP1 cells were starved overnight and divided into two groups: unexposed to insulin (control [Con], n=6) and exposed to 2 nM insulin for 3 hours (2 nM, n=6). The expression levels of ESR1, PRKAR1B, PER1, B3GALNT1, FGF4, and TNFSF4 were then measured. (C) 3T3–L1 cells were starved overnight and divided into two groups: unexposed to insulin (Con, n=8) and exposed to 2 nM insulin for 3 hours (2 nM, n=10). The expression levels of Esr1, Prkar1b, Per1, B3galnt1, Fgf4, and Tnfsf4 were then measured. Total mRNA was collected and examined for gene expression by quantitative polymerase chain reaction. Data are presented as the mean±standard deviation. 2 nM, exposed to 2 nM insulin for 3 hours group. ^aP<0.01, ^bP<0.001 vs. Con.

REFERENCES

• 1. Petersen MC, Shulman GI. Mechanisms of insulin action and insulin resistance. Physiol Rev 2018;98:2133-223.ArticlePubMedPMC
• 2. Zhang AM, Wellberg EA, Kopp JL, Johnson JD. Hyperinsulinemia in obesity, inflammation, and cancer. Diabetes Metab J 2021;45:285-311.ArticlePubMedPMCPDF
• 3. Nagao H, Cai W, Wewer Albrechtsen NJ, Steger M, Batista TM, Pan H, et al. Distinct signaling by insulin and IGF-1 receptors and their extra- and intracellular domains. Proc Natl Acad Sci U S A 2021;118:e2019474118.ArticlePubMedPMC
• 4. Makhijani P, Basso PJ, Chan YT, Chen N, Baechle J, Khan S, et al. Regulation of the immune system by the insulin receptor in health and disease. Front Endocrinol (Lausanne) 2023;14:1128622.ArticlePubMedPMC
• 5. Soop M, Duxbury H, Agwunobi AO, Gibson JM, Hopkins SJ, Childs C, et al. Euglycemic hyperinsulinemia augments the cytokine and endocrine responses to endotoxin in humans. Am J Physiol Endocrinol Metab 2002;282:E1276-85.ArticlePubMed
• 6. Westerbacka J, Corner A, Kannisto K, Kolak M, Makkonen J, Korsheninnikova E, et al. Acute in vivo effects of insulin on gene expression in adipose tissue in insulin-resistant and insulin-sensitive subjects. Diabetologia 2006;49:132-40.ArticlePubMedPDF
• 7. Chambers JC, Loh M, Lehne B, Drong A, Kriebel J, Motta V, et al. Epigenome-wide association of DNA methylation markers in peripheral blood from Indian Asians and Europeans with incident type 2 diabetes: a nested case-control study. Lancet Diabetes Endocrinol 2015;3:526-34.ArticlePubMedPMC
• 8. Day SE, Coletta RL, Kim JY, Garcia LA, Campbell LE, Benjamin TR, et al. Potential epigenetic biomarkers of obesity-related insulin resistance in human whole-blood. Epigenetics 2017;12:254-63.ArticlePubMedPMC
• 9. Liu J, Carnero-Montoro E, van Dongen J, Lent S, Nedeljkovic I, Ligthart S, et al. An integrative cross-omics analysis of DNA methylation sites of glucose and insulin homeostasis. Nat Commun 2019;10:2581.PubMedPMC
• 10. Bano S, More S, Mongad DS, Khalique A, Dhotre DP, Bhat MK, et al. Prolonged exposure to insulin might cause epigenetic alteration leading to insulin resistance. FEBS Open Bio 2025;15:81-93.ArticlePubMed
• 11. Weber M, Hellmann I, Stadler MB, Ramos L, Paabo S, Rebhan M, et al. Distribution, silencing potential and evolutionary impact of promoter DNA methylation in the human genome. Nat Genet 2007;39:457-66.ArticlePubMedPDF
• 12. Shin D, Eom YS, Chon S, Kim BJ, Yu KS, Lee DH. Factors influencing insulin sensitivity during hyperinsulinemic-euglycemic clamp in healthy Korean male subjects. Diabetes Metab Syndr Obes 2019;12:469-76.PubMedPMC
• 13. Dhar GA, Saha S, Mitra P, Nag Chaudhuri R. DNA methylation and regulation of gene expression: guardian of our health. Nucleus (Calcutta) 2021;64:259-70.ArticlePubMedPMCPDF
• 14. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545-50.ArticlePubMedPMC
• 15. Cohen P, Peehl DM, Lamson G, Rosenfeld RG. Insulin-like growth factors (IGFs), IGF receptors, and IGF-binding proteins in primary cultures of prostate epithelial cells. J Clin Endocrinol Metab 1991;73:401-7.ArticlePubMed
• 16. Woods AG, Guthrie KM, Kurlawalla MA, Gall CM. Deafferentation-induced increases in hippocampal insulin-like growth factor-1 messenger RNA expression are severely attenuated in middle aged and aged rats. Neuroscience 1998;83:663-8.PubMed
• 17. Scarth JP. Modulation of the growth hormone-insulin-like growth factor (GH-IGF) axis by pharmaceutical, nutraceutical and environmental xenobiotics: an emerging role for xenobiotic-metabolizing enzymes and the transcription factors regulating their expression: a review. Xenobiotica 2006;36:119-218.ArticlePubMed
• 18. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550.ArticlePubMedPMCPDF
• 19. Skolnik EY, Batzer A, Li N, Lee CH, Lowenstein E, Mohammadi M, et al. The function of GRB2 in linking the insulin receptor to Ras signaling pathways. Science 1993;260:1953-5.ArticlePubMed
• 20. Suissa Y, Magenheim J, Stolovich-Rain M, Hija A, Collombat P, Mansouri A, et al. Gastrin: a distinct fate of neurogenin3 positive progenitor cells in the embryonic pancreas. PLoS One 2013;8:e70397.ArticlePubMedPMC
• 21. Kramer A, Green J, Pollard J, Tugendreich S. Causal analysis approaches in Ingenuity pathway analysis. Bioinformatics 2014;30:523-30.ArticlePubMedPDF
• 22. Kaizer EC, Glaser CL, Chaussabel D, Banchereau J, Pascual V, White PC. Gene expression in peripheral blood mononuclear cells from children with diabetes. J Clin Endocrinol Metab 2007;92:3705-11.ArticlePubMed
• 23. Mallu AC, Vasudevan M, Allanki S, Nathan AA, Ravi MM, Ramanathan GS, et al. Prediabetes uncovers differential gene expression at fasting and in response to oral glucose load in immune cells. Clin Nutr 2021;40:1247-59.ArticlePubMed
• 24. Zhao J, Fang Z. Single-cell RNA sequencing reveals the dysfunctional characteristics of PBMCs in patients with type 2 diabetes mellitus. Front Immunol 2025;15:1501660.ArticlePubMedPMC
• 25. Bergman RN, Finegood DT, Ader M. Assessment of insulin sensitivity in vivo. Endocr Rev 1985;6:45-86.ArticlePubMed
• 26. Tam CS, Xie W, Johnson WD, Cefalu WT, Redman LM, Ravussin E. Defining insulin resistance from hyperinsulinemic-euglycemic clamps. Diabetes Care 2012;35:1605-10.PubMedPMC
• 27. Bonora E, Formentini G, Calcaterra F, Lombardi S, Marini F, Zenari L, et al. HOMA-estimated insulin resistance is an independent predictor of cardiovascular disease in type 2 diabetic subjects: prospective data from the Verona diabetes complications study. Diabetes Care 2002;25:1135-41.PubMed
• 28. Yamada C, Mitsuhashi T, Hiratsuka N, Inabe F, Araida N, Takahashi E. Optimal reference interval for homeostasis model assessment of insulin resistance in a Japanese population. J Diabetes Investig 2011;2:373-6.ArticlePubMedPMC
• 29. Mittendorfer B, Patterson BW, Smith GI, Yoshino M, Klein S. Effects of obesity and hyperglycemia on postprandial insulin-mediated and non-insulin-mediated glucose disposal. Diabetes Care 2025;48:84-92.ArticlePubMedPDF
• 30. Jjingo D, Conley AB, Yi SV, Lunyak VV, Jordan IK. On the presence and role of human gene-body DNA methylation. Oncotarget 2012;3:462-74.ArticlePubMedPMC
• 31. Gallagher CJ, Langefeld CD, Gordon CJ, Campbell JK, Mychaleckyj JC, Bryer-Ash M, et al. Association of the estrogen receptor-alpha gene with the metabolic syndrome and its component traits in African-American families: the insulin resistance atherosclerosis family study. Diabetes 2007;56:2135-41.PubMed
• 32. Smith EP, Boyd J, Frank GR, Takahashi H, Cohen RM, Specker B, et al. Estrogen resistance caused by a mutation in the estrogen-receptor gene in a man. N Engl J Med 1994;331:1056-61.ArticlePubMed
• 33. Heine PA, Taylor JA, Iwamoto GA, Lubahn DB, Cooke PS. Increased adipose tissue in male and female estrogen receptor-alpha knockout mice. Proc Natl Acad Sci U S A 2000;97:12729-34.PubMedPMC
• 34. Min J, Weitian Z, Peng C, Yan P, Bo Z, Yan W, et al. Correlation between insulin-induced estrogen receptor methylation and atherosclerosis. Cardiovasc Diabetol 2016;15:156.ArticlePubMedPMCPDF
• 35. Ying L, Wang L, Guo K, Hou Y, Li N, Wang S, et al. Paracrine FGFs target skeletal muscle to exert potent anti-hyperglycemic effects. Nat Commun 2021;12:7256.ArticlePubMedPMCPDF
• 36. Ghai V, Baxter D, Wu X, Kim TK, Kuusisto J, Laakso M, et al. Circulating RNAs as predictive markers for the progression of type 2 diabetes. J Cell Mol Med 2019;23:2753-68.ArticlePubMedPMCPDF
• 37. Yu F, Teng Y, Li J, Yang S, Zhang Z, He Y, et al. Effects of a Ganoderma lucidum proteoglycan on type 2 diabetic rats and the recovery of rat pancreatic islets. ACS Omega 2023;8:17304-16.ArticlePubMedPMCPDF
• 38. Marques ES, Formato E, Liang W, Leonard E, Timme-Laragy AR. Relationships between type 2 diabetes, cell dysfunction, and redox signaling: a meta-analysis of single-cell gene expression of human pancreatic α- and β-cells. J Diabetes 2022;14:34-51.ArticlePubMedPDF
• 39. Cooke J, Zhang H, Greger L, Silva AL, Massey D, Dawson C, et al. Mucosal genome-wide methylation changes in inflammatory bowel disease. Inflamm Bowel Dis 2012;18:2128-37.ArticlePubMed
• 40. Takahashi JS. Transcriptional architecture of the mammalian circadian clock. Nat Rev Genet 2017;18:164-79.ArticlePubMedPDF
• 41. Jakubowicz D, Wainstein J, Landau Z, Raz I, Ahren B, Chapnik N, et al. Influences of breakfast on clock gene expression and postprandial glycemia in healthy individuals and individuals with diabetes: a randomized clinical trial. Diabetes Care 2017;40:1573-9.ArticlePubMedPDF
• 42. Scheiermann C, Kunisaki Y, Frenette PS. Circadian control of the immune system. Nat Rev Immunol 2013;13:190-8.ArticlePubMedPMCPDF
• 43. Hergenhan S, Holtkamp S, Scheiermann C. Molecular interactions between components of the circadian clock and the immune system. J Mol Biol 2020;432:3700-13.ArticlePubMedPMC
• 44. Vieira E, Mirizio GG, Barin GR, de Andrade RV, Nimer NF, La Sala L. Clock genes, inflammation and the immune system-implications for diabetes, obesity and neurodegenerative diseases. Int J Mol Sci 2020;21:9743.ArticlePubMedPMC
• 45. Nadiger N, Veed JK, Chinya Nataraj P, Mukhopadhyay A. DNA methylation and type 2 diabetes: a systematic review. Clin Epigenetics 2024;16:67.ArticlePubMedPMCPDF
• 46. Zhou Z, Sun B, Li X, Zhu C. DNA methylation landscapes in the pathogenesis of type 2 diabetes mellitus. Nutr Metab (Lond) 2018;15:47.ArticlePubMedPMCPDF
• 47. Willmer T, Johnson R, Louw J, Pheiffer C. Blood-based DNA methylation biomarkers for type 2 diabetes: potential for clinical applications. Front Endocrinol (Lausanne) 2018;9:744.ArticlePubMedPMC
• 48. El Jellas K, Dusatkova P, Haldorsen IS, Molnes J, Tjora E, Johansson BB, et al. Two new mutations in the CEL gene causing diabetes and hereditary pancreatitis: how to correctly identify MODY8 cases. J Clin Endocrinol Metab 2022;107:e1455-66.ArticlePubMedPDF
• 49. Latorre J, Mayneris-Perxachs J, Oliveras-Canellas N, Ortega F, Comas F, Fernandez-Real JM, et al. Adipose tissue cysteine dioxygenase type 1 is associated with an anti-inflammatory profile, impacting on systemic metabolic traits. EBioMedicine 2022;85:104302.ArticlePMC
• 50. Rickels MR, Elahi D. Raising serum gastrin to improve glycemic control in (type 2) diabetes: another limb of the enteroinsular axis? J Clin Endocrinol Metab 2012;97:3915-6.Article

Figure & Data

References

Citations

Citations to this article as recorded by  

About this article

Joo M, Shin D, Truong XT, Nam S, Lee DH. Acute Hyperinsulinemia during Hyperinsulinemic- Euglycemic Clamp Influences DNA Methylation and Gene Expression in Peripheral Blood Cells of Adult Men. Diabetes Metab J. 2025 Sep 5. doi: 10.4093/dmj.2025.0072. Epub ahead of print.

Received: Jan 25, 2025; Accepted: Jul 01, 2025

DOI: https://doi.org/10.4093/dmj.2025.0072.

Download citation ↓