Skip to main content
Download PDF

Identification of biomarkers and mechanism exploration of ferroptosis related genes regulated by m6A in type 2 diabetes mellitus

Hereditas volume 162, Article number: 24 (2025) Cite this article

Abstract

Purpose

This study is aims to explore the role of ferroptosis genes regulated by N6-methyladenosine (m6A) in Type 2 diabetes mellitus (T2DM).

Material and methods

Firstly, differentially expressed m6A-FRGs (DEm6A-FRGs) were obtained by intersecting the differentially expressed genes (DEGs) and the m6A-related ferroptosis genes (m6A-FRGs). After enrichment analysis of DEm6A-FRGs, artificial neural network (ANN) and nomogram models were constructed using 4 biomarkers. Moreover, the gene set enrichment analysis of biomarkers was performed. Furthermore, the transcription factors (TF)-mRNA and competing endogenous RNAs (ceRNA) regulatory networks were constructed to reveal the potential regulation of biomarkers at molecular level. In addition, the targeted drugs of biomarkers were predicted, and the molecular docking was used to study the inter-molecular interactions between biomarkers and targeted drugs by “AutoDockvina”.

Results

Totals of 10 DEm6A-FRGs were obtained by intersecting the 402 DEGs and 299 m6A-FRGs. Moreover, the ANN model and nomogram model were constructed with 4 biomarkers including CDKN1A, MIOX, MYCN and CD82, among them, CDKN1A was the most important biomarker for forecasting T2DM. Notably, the function of extracellular matrix structural constituent was low expression in CD82 and MIOX, the function of mitochondrial protein-containing complex was high expression in CD82 and CDKN1A. Furthermore, TP63 could regulate CD82, CDKN1A and MIOX, GATA3 could regulate CD82, CDKN1A and MYCN at the same time. The ceRNA network was constructed with 4 mRNAs, 51 miRNAs and 37 lncRNAs, among them, XIST was a key lncRNA that associated with 12 miRNAs, which could influence CDKN1A. In addition, bisphenol A was associated with CD82 and MYCN, CGP 25608 was associated with CDKN1A and MIOX.

Conclusion

This study revealed the potential molecular mechanisms of m6A-related ferroptosis genes in T2DM, which could provide novel insights for the clinical diagnosis and treatment of T2DM.

Introduction

Type 2 diabetes mellitus (T2DM) is a prevalent metabolic disease that is commonly associated with obesity, high blood pressure, high cholesterol, and other metabolic diseases such as cardiovascular disease [1, 2]. It is primarily characterized by insulin resistance and islet beta cells dysfunction [3]. Treatments for T2DM include lifestyle changes (such as modifications in diet and exercise habits), medications (such as oral medications and insulin injections), and surgical treatments (such as gastrointestinal bypass surgery) [3]. Searching for biomarkers could assist doctors in identifying early signs of diabetes, intervening and treating those at high risk early, thus reducing the incidence of complications. Furthermore, searching for biomarkers can also be used to evaluate the effectiveness and prognosis of treatment for T2DM, and provide reference for personalized treatment.

N6-methyladenosine (m6A) modification is a highly prevalent RNA modification that has been shown to play a critical role in numerous biological processes, including post-transcriptional regulation, RNA splicing, RNA stability, and RNA translation [4,5,6].Changes in m6A modification levels have been shown to regulate insulin signaling pathways, glucose metabolism, insulin secretion, and the function of islet beta cells, thereby influencing insulin resistance and the development of T2DM [7,8,9]. Ferroptosis is a newly recognized type of programmed cell death that has gained increasing attention in recent years [10]. Research has demonstrated that ferroptosis plays a crucial role in the pathogenesis of T2DM, and serum iron levels are positively correlated with blood glucose levels in T2DM patients [11, 12]. However, the role of ferroptosis genes regulated by m6A in T2DM has not yet been fully explored.

In this study, we integrated literature-reported m6A regulatory genes, ferroptosis related genes from the FerrDb database, and T2DM transcriptome data from the Gene Expression Omnibus (GEO) database to identify m6A ferroptosis genes. Then we used machine learning algorithms to identify four biomarkers of T2DM. Additionally, we constructed transcription factors (TF)-mRNA regulatory and competing endogenous RNAs (ceRNA) networks of these biomarkers to predict potential drugs, which can provide potential targets for clinical diagnosis of T2DM and a theoretical basis for further understanding of the driving mechanism of T2DM.

Material and methods

Data extraction

T2DM-related datasets GSE76895 (GPL570) and GSE41762 (GPL6244) were extracted from GEO database (https://www.ncbi.nlm.nih.gov/geo/). Among them, GSE76895 was used as the training dataset, which includes pancreatic islet samples of 36 T2DM and 32 healthy controls (HC), GSE41762 was used as the validation dataset to verify the expression levels of biomarkers, which includes pancreatic islet samples of 20 T2DM and 57 HC. 621 ferroptosis-related genes were selected from the FerrDb database (http://www.zhounan.org/ferrdb/current), and 23 m6A methylation regulatory-related genes were downloaded from previous literature [13].

Function analysis of differentially expressed DEm6A-FRGs of T2DM

Firstly, the mRNA expression levels between 36 T2DM and 32 HC samples were compared by the “limma” R package (version 3.52.2) (|log2FC|> 0.5, p < 0.05) [14]. The correlation of 23 m6A methylation regulatory-related genes and 621 ferroptosis-related genes were calculated by “Spearman”, and the m6A-related ferroptosis genes (m6A-FRGs) were obtained with |Cor|> 0.3 and p < 0.05. Then, the differentially expressed m6A-FRGs (DEm6A-FRGs) were obtained by intersecting the differentially expressed genes (DEGs) and m6A-FRGs. Besides, the function analyses of DEm6A-FRGs were conducted by “clusterprofiler” R package (version 4.4.4) [15].

Construction of the diagnostic models

In this study, random forest (RF) and support vector machine recursive feature elimination (SVM-RFE) methods were used to obtain the importance ranking of DEm6A-FRGs and the error rate and accuracy of each iteration combination, respectively. The best combination with lowest error rate was selected, and the corresponding genes were taken as the characteristic/feature genes, respectively [16]. Then, the biomarkers were screened by intersecting the characteristic genes and feature genes for subsequent analysis.

Artificial neural network (ANN) model and nomogram model were constructed for clinical diagnosis of T2DM. On the one hand, the data was standardized using the “neuralnet” R package (version 1.44.2) [17] with the parameters preProcess = c("scale", "center"), and the output was calculated as a probability using the logistic activation function (act.fct = "logistic"). The number of hidden neurons was set to 3, constructing the ANN model. The weight of biomarkers was determined in the training set (GSE76895), and the classification efficiency of the model constructed using the expression and weight of biomarkers was evaluated. Additionally, the performance and stability of the ANN model were evaluated using fivefold cross-validation with number = 5 and repeats = 8, effectively reducing the risk of overfitting caused by insufficient data samples or random splitting. Meanwhile, the model's generalization ability was further validated in the external validation set (GSE41762). Besides, the receiver operating characteristic curves (ROC) were drew to assess the diagnose ability of ANN model. On the other hand, the logistic regression model (LRM) was constructed, and the nomogram model was constructed with these biomarkers by “RMS” R package (version 5.4–1) [18]. Then, the calibration curve, decision curve (DCA) and clinical impact curve were drawn to verify the validity of the nomogram.

The gene set enrichment analysis (GSEA) of biomarkers

The correlation coefficient between each biomarker and all genes in GSE76895 were calculated, and Gene Set Enrichment Analysis (GSEA) of each biomarker was performed by “clusterProfiler” R package [19].

Construction of TF-mRNA and ceRNA regulatory networks

Firstly, the potential TFs were screened in ChEA3 database (https://amp.pharm.mssm.edu/chea3/) and KnockTF database (http://www.licpathway.net/KnockTF/). Then, the targeted TFs were obtained by intersecting these 2 groups of potential TFs. Finally, the TF-mRNA network was constructed by “Cytoscape” (version 3.8.2) [20].

Furthermore, the potential miRNA were predicted in miRwalk database (http://mirwalk.umm.uni-heidelberg.de/) and mirtarBase database (https://mirtarbase.cuhk.edu.cn/). Then, the targeted miRNAs were obtained by intersecting these 2 groups of potential miRNAs. Next, the targeted lncRNAs were predicted in starbase database with clipExpNum > 3 (http://starbase.sysu.edu.cn/). Finally, the ceRNA network was constructed by “Cytoscape”.

Drug prediction

The potential drugs were predicted by biomarkers in Comparative Toxicogenomics Database (CTD, http://ctdbase.org/), and the interaction relationships between potential drugs and biomarkers were visualized by “Cytoscape”. In addition, the targeted drugs which were associated with T2DM were further predicted in CDT. Besides, the interaction scores between biomarkers and T2DM were calculated.

Protein structure prediction of the biomarkers (MIOX, CD82, and MYCN) was performed using Alphafold3, and the structure with the highest prediction score was selected as the structure for subsequent molecular docking and molecular dynamics simulations (MDs). The protein structure of CDKN1A was obtained from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB, https://www.rcsb.org). The 3D structures of the targeted drugs were mind from the PubChem database (https://pubchem.ncbi.nlm.nih.gov), the protein structure of biomarkers were downloaded from the PDB. Water molecules and small molecule ligands were removed with “PyMOL” (v 2.5) [21] and the molecular docking was used to study the inter-molecular interactions between biomarkers and targeted drugs by “AutoDock Vina” (v 4.2) [22]. Specifically, the protein was hydrogenated and the charges were calculated using AutoDockTools (version 1.5.6) [23], and the charges of the small molecules were balanced, along with checking for rotatable bonds. Then, the docking box range was selected based on the receptor's active site. Finally, AutoDock Vina was used to calculate the receptor-ligand docking (num_modes = 20), and the structure with the lowest binding free energy from the output results was chosen (a docking binding energy < -5 kcal/mol indicates a good binding affinity).The “PyMol” was utilized to visualize results.

MDs is a molecular simulation method based on Newton's laws of motion for calculating time-dependent properties of molecular systems. To further validate the plausibility and reliability of the biomarker and drug docking results, in this study we performed 50 ns MDs through GROMACS2024.2 software. Protein sequences were retrieved from NCBI, screened in the database for human species, and AlphaFold3 predicted structures were selected as protein structure files. Parameter and topology files of protein and small molecule ligands were generated from AMBER14SB force field and AMNER gaff force field, respectively. The system was neutralized using NaCL at a concentration of 0.15 mol/L. The steepest descent method was used to minimize the energy consumption of the whole system. Prior to the MDs, the system temperature and pressure were stabilized. Specifically, simulations were performed using the NVT system at 300 K (temperature) and 100 ps to stabilize the temperature and the NPT system at 1 bar (pressure) and 100 ps to stabilize the pressure.

Expression verification of biomarkers

The expression of biomarkers between the T2DM and HC samples were compared by “wilcox.test” in both GSE76895 and GSE41762 datasets.

Western blotting

Primary pancreatic islet cells (CP-R015) were purchased from Procell Life Science&Technology Co.,Ltd. They were cultured with or without 30 mM glucose (high glucose, HG). Protein concentrations were extracted using ice-cold RIPA buffer (Beyotime, Nantong, China) containing protease inhibitor and phosphatase inhibitor (Thermo Fisher Scientific, Waltham, MA, USA). A 20–30 μg protein was subjected to 10%-15% SDS polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes (PVDF), Millipore, Bedford, MA, USA). The PVDF membranes were blocked in 5% skim milk for 1 h and then incubated overnight at 4℃ with CDKN1A (sc-6246, Santa Cruz Biotechnology), MYCN (sc-53993, Santa Cruz Biotechnology), MIOX (sc-376080, Santa Cruz Biotechnology) or CD82 (sc-518002, Santa Cruz Biotechnology). After incubating with appropriate secondary antibodies for 2 h, the band densities were determined using Image J software (NIH, Maryland, USA) and normalized to each internal control.

Results

Totals of 10 DEm6A-FRGs were obtained in T2DM

There were 402 DEGs (324 up-regulated and 78 down-regulated) between 36 T2DM and 32 HC samples (Fig. 1A&B, Supplement Table 1). Then the correlations between 23 m6A methylation regulatory-related genes and 621 ferroptosis-related genes were calculated, and totals of 299 m6A-FRGs were obtained (Fig. 1C, Supplement Table 2). Finally, 10 DEm6A-FRGs (MIOX, MYCN, ARHGEF26-AS1, CD82, ADAM23, YAP1, SOCS1, CDKN1A, NNMT and PSAT1) were obtained by intersecting 402 DEGs and 299 m6A-FRGs (Fig. 1D).

Fig. 1
figure 1

Identification of 10 differentially expressed m6A-FRGs (DEm6A-FRGs) in Gene Expression Omnibus (GEO) datasets. A Volcano plot showed 324 up-regulated (red) and 78 down-regulated genes (blue) in two clusters (|log2FC|> 0.5, adj.P-value < 0.05). B Heatmap of the correlation between T2D group and ND group. C Heatmap of the correlation between m6A-related genes and FRGs. *p < 0.05, **p < 0.01. D Venn diagram for differentially expressed genes (DEGs) and m6A-related ferroptosis genes (m6A-FRGs). E The column plot of the 8 Gene Ontology (GO) Biological Process (BP) terms and top 10 GO Molecular Function (MF) terms were enriched for 10 DEm6A-FRGs. F The bubble plot of the top 10 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were enriched for 10 DEm6A-FRGs

Full size image

Besides, these genes were enriched to 8 Gene Ontology (GO) biological processes and 17 GO molecular functions, including regeneration, pyridine-containing compound biosynthetic and metabolic process, water-soluble vitamin metabolic process, protein kinase inhibitor activity, and etc. In addition, these genes were enriched to 13 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, including p53 signaling pathway, JAK-STAT signaling pathway, type II diabetes mellitus, and etc. (Fig. 1E&F, Supplement Table 3&4).

Construction and evaluation of the diagnostic model of T2DM with 4 biomarkers

Six characteristic genes, including CDKN1A, MIOX, MYCN, CD82, ARHGEF26-AS1 and ADAM23, were screened by RF method, and 7 feature genes, including CDKN1A, MYCN, CD82, ARHGEF26-AS1, MIOX, YAP1 and NNMT were screened by SVM-RFE method (Fig. 2A&B). Then, 4 common genes, including CDKN1A, MIOX, MYCN and CD82 were regarded as the biomarkers for further analysis (Fig. 2C, D).

Fig. 2
figure 2

Screening of 4 biomarkers by Random Forest (RF) method and support vector machine recursive feature elimination (SVM-RFE) method. A-B 6 characterized genes were screened from 10 DEm6A-FRGs by RF method. C 7 characterized genes were screened from 10 DEm6A-FRGs by SVM-RFE method. D Venn diagram of characterized genes

Full size image

Based on it, the ANN model and the nomogram with these 4 biomarkers were constructed, respectively. In ANN model, CDKN1A was the most important biomarker for forecasting T2DM (Fig. 3A,B). Moreover, the areas under ROC (AUC values) of ANN model were great than 0.7 in both GSE76895 and GSE41762 (Fig. 3B), these results showed that the ANN model could be used as an effective diagnostic model. On the other hand, LRM showed the similar results, which CDKN1A was the most important biomarker for forecasting T2DM (odd ratio = 17) (Fig. 4A). The nomogram with these 4 biomarkers were constructed, the calibration curve of the nomogram showed that the slope was the closest to 1 and ROC was 0.937, which indicated that the prediction model could be used as an effective diagnostic model (Fig. 4B, C). The results of the DCA were showed that the benefit rate of the nomogram model was higher than individual biomarker (Fig. 4D), and the results of clinical impact curve were showed that the “Number high risk” curve almost coincided with the “Number high risk with event” curve at the benefit ratio threshold from 0 to 1 (Fig. 4E), all of these indicating that the nomogram model has accurate predictive power.

Fig. 3
figure 3

Construction and validation of an Artificial Neural Network (ANN) model based on 4 biomarkers. A Visualization of the ANN model. The black arrows and associated numbers are weights, which you can think of as the variable's contribution to the next node's the degree of contribution of that variable to the next node. The blue line is the bias weight. B Visualization of the importance of the independent variables on the predicted results of the model. C Confusion matrix and operating characteristic curves (ROC) curve for hub genes in GSE76895 and GSE41762 datasets

Full size image
Fig. 4
figure 4

Construction and validation of logistic regression models for biomarkers. A Forest plot of the multivariable Cox analyses of the biomarkers. B Nomogram model of four biomarkers in the GSE76895 dataset. C Calibration curves of the nomogram for predicting the Odd Ratio of T2DM. The 45-degree line represents the ideal prediction. D Decision curve analyses (DCA) curve of the nomogram. E Clinical impact curves predicted with the nomogram. The number high risk indicates the number of people classified as high risk by the simple model at each threshold, and the number high risk outcome is the number of true positives at each threshold

Full size image

The GSEA of biomarkers

The results of GSEA of 4 biomarkers were shown in the Fig. 5A-D. The results revealed that CD82 was associated with the function of respirasome, CDKN1A was associated with the function of generation of precursor metabolites and energy, MIOX was associated with the function of humoral immune response extracellular matrix structural, MYCN was associated with the function of digestion, proteasome complex, endopeptidase complex, cellular response to copper ion and cadmium ion. Noticeable, the function of extracellular matrix structural constituent lowly expressed in both CD82 and MIOX, the function of mitochondrial protein-containing complex highly expressed in both CD82 and CDKN1A. On the other hands, the pathway of Staphylococcus Aureus infection highly enriched in CD82 and MIOX, the pathway of proteasome highly enriched in CDKN1A and MYCN. In addition, the pathway of huntington disease and parkinson disease highly expressed in CD82 and CDKN1A.

Fig. 5
figure 5

Gene Set Enrichment Analysis (GSEA) analysis of 4 biomarkers in GO terms and KEGG pathways (|NES|> 1, adj. P-value < 0.05, Q–value < 0.2). A-D GSEA analysis of CD82 (A), CDKN1A (B), MIOX (C) and MYCN (D) in GO terms and KEGG pathways

Full size image

Molecular mechanism analysis of 4 biomarkers

A total of 44 TFs associated with biomarkers were downloaded and 59 TF-mRNA relationship pairs were obtained for constructing the TF-mRNA regulatory network (Fig. 6A, Table 1, Supplement Table 5). In this network, TP63 was key TF which could regulate CD82, CDKN1A and MIOX at the same time, GATA3 was key TF which could regulate CD82, CDKN1A and MYCN at the same time, STAT1 and RELA were the common TFs of CD82 and CDKN1A, besides, CDX2, PRDM14, GATA1, HNF1A, HOXB4, SOX2, TAL1 and SETDB1 were the common TFs of CDKN1A and MYCN.

Fig. 6
figure 6

Molecular mechanism analysis of 4 biomarkers. A Construction of transcription factors (TF)-mRNA regulatory network. Blue diamonds represent transcription factors, and orange ovals represent biomarkers. B Construction of competing endogenous RNAs (ceRNA) network. Orange ovals represent biomarkers, azure hexagons represent lncRNAs, and green rectangular shapes represent miRNAs

Full size image
Table 1 Number of predicted transcription factors (TFs) and intersection TFs per database
Full size table

A total of 51 targeted miRNAs were obtained, which includes 3 targeted miRNAs of CD82, 39 targeted miRNAs of CDKN1A, 3 targeted miRNAs of MIOX, and 6 targeted miRNAs of MYCN. Based on it, 37 lncRNAs were predicted for further studies (Table 2, Supplement Table 6). Moreover, the ceRNA network was constructed, which contains 4 biomarkers, 51 miRNAs and 37 lncRNAs (Fig. 6B, Supplement Table 7). In this network, we found that there were 12 common lncRNAs between has-miR-4458 and has-let-7e-5p, 9 common lncRNAs among has-miR-17-5p, has-miR-519d-3p and has-miR-106a-5p. It is worth noting that XIST was a key lncRNA that associated with 12 targeted miRNAs, which could influence CDKN1A.

Table 2 Number of miRNAs screened for 4 biomarkers by miRWalk database
Full size table

Drug prediction

The results of interaction relationship between drugs and 4 biomarkers were showed that totals of 804 drugs and 966 relationship pairs were predicted, among them, 46 drugs were associated with T2DM. It is worth noting that C006780 could increase the expression of CDKN1A, CD82, MYCN and decrease the expression of MIOX at the same time. D019833 could increase the expression of CDKN1A, CD82 and MYCN, C009495 could decrease the expression of CDKN1A, MIOX and MYCN. C023036 could increase the expression of CDKN1A, MIOX, and decrease the expression of CD82. D000077185 could increase the expression of CDKN1A, MYCN, and decrease the expression of MIOX. D004052 could increase the expression of MYCN, MIOX, and decrease the expression of CDKN1A. D005492 could increase the expression of MIOX, and decrease the expression of CDKN1A, MYCN. D003993 could increase the expression of CDKN1A, CD82, and decrease the expression of MIOX. D004041 could increase the expression of MYCN, MIOX, and decrease the expression of CD82 (Fig. 7A, B, Table 3). Besides, the interaction scores and networks between T2DM and the biomarkers were showed in Table 4.

Fig. 7
figure 7

Biomarker-related drug prediction. A Drug prediction for 4 biomarkers. Green circles are biomarkers, blue diamonds are drug small molecules, red edges are up-regulated expression regulation, blue edges are down-regulated expression regulation. B The drug networks of biomarkers in disease. C The Molecular docking results of four biomarkers with key drugs

Full size image
Table 3 Interaction scores of T2D with biomarkers
Full size table
Table 4 Interaction scores and networks between Type 2 diabetes mellitus (T2DM) and biomarkers
Full size table

The results of molecular docking between 4 biomarkers and key drugs were showed in Fig. 7C. The docking affinity between CD82 and bisphenol A was -7.5 kcal/mol. The docking affinity between CDKN1A and Dibutyl Phthalate was -5.0 kcal/mol. The docking affinity between MIOX and perfluorooctanoic acid was -7.1 kcal/mol. The docking affinity between MYCN and bisphenol A was -6.2 kcal/mol. All of these results suggested a good combination of biomarkers and targeted drugs.

To further validate the reliability of molecular docking, MDs were implemented in this study, where root mean square deviation (RMSD), root mean square fluctuation (RMSF) and radius of gyration (RoG) were included in the simulation analysis. The results of this study were presented only for CDKN1A, and the results of MDs for the remaining biomarkers were shown in Supplementary Figs. 1–3. RMSD curves can reveal the positional changes of the protein between the conformation and the initial conformation during the simulation. The RMSD curve fluctuated in the early stages due to the interaction between the complex and the solvent. After that, the protein was in a steady state at 40 ns to 50 ns, and the RMSD value falls between 0.25 and 0.35, indicating that the binding of CDKN1A and dibutyl phthalate was relatively stable (Fig. 8A-B). The RMSF is the average of the atomic position change for time, which can characterize the flexibility and movement intensity of protein amino acids throughout the simulation. The results showed that the protein amino acids were flexible and bind stably to the small molecule drug ligands within the simulation duration (Fig. 8C-D). RoG can characterize the tightness of the protein structure, and similarly can be relied upon to characterize the changes in the degree of peptide chain looseness of the proteins during the simulation. As can be seen in the Fig. 8E-F, CDKN1A-dibutyl phthalate has a stable RoG, which was consistent with the RMSD curve results, indicating a more stable binding. Furthermore, in the simulation process of protein and small molecule ligand, the distance between key amino acids and ligand showed a dynamic change trend. The change of the distance between Dibutyl phthalate and the active site of Serine amino acid residue No. 49 of CDKN1A ligand protein tended to stabilize around 50 ns, and the fluctuation was mainly distributed around 0.6, which could be considered that the binding of dibutyl phthalate to the active site of CDKN1 was stable (Fig. 8G).

Fig. 8
figure 8

MDs of CDKN1A and Dibutyl Phthalate. A-B RMSD curve of CDKN1A protein. C-D RMSF plot of amino acid flexibility and exercise intensity of CDKN1A protein. E–F RoG plot of CDKN1A protein. G Dynamic distance changes of key amino acids and ligands

Full size image

Expression verification of biomarkers

The expression of 4 biomarkers between the T2DM and HC samples were compared, and the results showed that MIOX, MYCN, CD82 were lowly expressed, and CDKN1A was highly expressed in T2DM samples, among them, the expression of CDKN1A and MYCN showed significantly different in both of GSE76895 and GSE41762 (Fig. 9A, B).

Fig. 9
figure 9

Expression verification of biomarkers. A The gene expression level in GSE76895 dataset. B The gene expression level in GSE41762 dataset

Full size image

The alternation of protein expression in primary pancreatic islet cells (CP-R015) under hyperglycemia conditions

To investigate the expression of 4 biomarkers in cells, we cultured primary pancreatic islet cells (CP-R015) using both regular and high glucose medium, and then performed protein blotting. As shown in the Fig. 10A-E, compared with the control group (Ctrl), the high glucose group (HG) showed a significant increase in CDKN1A protein expression, while the expression of MIOX, CD82, and MYCN proteins was significantly reduced.

Fig. 10
figure 10

Protein expression of the biomarkers. A Western blotting results of CDKN1A, MIOX,CD82 and MYCN. B-E The column plot of changes in protein expression of CDKN1A (B), MIOX (C), CD82 (D), and MYCN (E). ***p < 0.001, ****p < 0.0001

Full size image

Discussion

T2DM is a common chronic wasting disease, which is characterized by insulin resistance and insulin secretion deficiency leading to dysglycemia [24]. Iron ions are cofactors of various enzymes and participate in various biosynthetic pathways in the body [25]. Imbalance of iron metabolism is a key factor in the development of endocrine diseases. Abnormal iron metabolism in the body leads to various pathological and physiological changes, such as ferroptosis. Ferroptosis damages to pancreatic beta cells, which leads to a decrease in insulin secretion and induces insulin resistance in the liver, fat, and muscles [26]. The m6A is one of the most common RNA modifications in eukaryotes, and m6A induces ferroptosis by regulating the autophagy signaling pathway [5]. At present, only a few studies have explored the relationship between m6A and ferroptosis in T2DM. Also the specific mechanism of m6A's post transcriptional regulation of ferroptosis is unknown. The study of the key genes of m6A regulating ferroptosis may provide effective therapeutic targets and diagnostic markers for T2DM. Through the analysis of 36 T2DM and 32 healthy controls in dataset GSE76895, the ferroptosis related genes regulated by m6A, which are differentially expressed in T2DM, were determined. Through the comprehensive analysis of random forest (RF) and support vector machine recursive feature elimination (SVM-RFE), four characteristic genes were identified, namely: CDKN1A, MIOX, MYCN and CD82. Then, Artificial neural network (ANN) model was constructed for these four feature genes, and the diagnostic efficacy of these four feature genes and ANN model was tested. We found that these four genes had good efficacy for T2DM diagnosis in the training set GSE76895. In the validation set GSE41762, although the diagnostic efficacy of the four genes decreased, the ANN model could still improve the diagnostic efficacy.

CDKN1A is a cycle dependent kinase inhibitor -1 that mediates cell proliferation, differentiation, and apoptosis by regulating the expression of cyclin [27, 28]. Researches have shown that CDKN1A involved in regulating the p53 signaling pathway to regulate cell differentiation and apoptosis. When CDKN1A is activated by p53 transcription, it upregulates and prevents the release of cell cycle regulatory factors, leading to cell cycle arrest and an increase in pancreatic beta cells apoptosis [29]. In addition, overexpression of p21/CDKN1A controls cell cycle entry into G1 phase, reducing pancreatic function, pancreatic beta cells proliferation and pancreatic beta cells insulin secretion [30]. Our data demonstrated that CDKN1A is highly expressed in T2DM, which is consistent with previous study [31]. In a related study, it was found that the degree of CDKN1A methylation was reduced in the islets of diabetic patients compared to normal subjects, which promoted the enhancement of CDKN1A expression. At the same time, the researchers found that overexpression of CDKN1A in cloned beta cells led to impaired insulin secretion, and overexpression of CDKN1A in cloned alpha cells led to increased glucagon secretion [32]. Therefore, CDKN1A causes T2DM by regulating the cell cycle, affecting insulin secretion and glucagon secretion, and how to promote CDKN1A methylation will become the key to the treatment of T2DM.

CD82 is a transfer inhibitory factor that is a member of the tetraspanin membrane protein family. It interacts with chemokines and integrins, controlling related signaling pathways to regulate cell motility, adhesion, migration, aging, and apoptosis. It has been found that CD82+ cells in human islets can secrete insulin more effectively than CD82− cells, and CD82 can promote the differentiation of endocrine progenitor cells into mature beta cells [33]. In our study, we found a decrease in CD82 in the islets of T2DM patients, which may be related to the enhancement of CD82 methylation, which inhibits its expression.

MYCN is a transcription factor whose main biological effect is to regulate the expression of multiple genes during cell proliferation, growth, aging, metabolism, differentiation, and apoptosis [34]. In a study on the growth cycle of mouse pancreatic islet beta cells, it was found that MYCN is an upstream mediator for pancreatic beta cell expansion, and regulates pancreatic cell growth by mediating the expansion cycle and differentiation of pancreatic beta cells [35]. The overexpression of MYCN makes mouse beta cells enter G1 cell cycle, which is the growth and differentiation stagnation, leading to the growth and differentiation obstacle of beta cells and diabetes [36]. Signe Horn et al. found that MYCN is not only involved in regulating cell growth, survival, metabolism, and death, but also in maintaining cellular redox balance [37, 38]. Overexpression of MYCN promotes the upregulation of TFRC and mediates GPX4 pathway dependent ferroptosis [39]. Meanwhile, overexpression of MYCN increases intracellular iron load and unstable iron pool (LIP) levels, ultimately leading to enhanced lipid peroxidation and ferroptosis [37]. At the same time, the strong proliferative signal overexpressed by MYCN often leads to apoptosis of pancreatic islet beta cells [40]. In the study of neuroblastoma, it was found that MYCN amplification is related to the methylation status of its binding site, and the degree of methylation is negatively correlated with the degree of gene expression [41]. The low expression of MYCN in T2DM may be due to a protective mechanism of the body under hyperglycemia. Ferroptosis, cell apoptosis and methylation play an important role in the progress of T2DM, and MYCN has the co-function of regulating cell proliferation, growth, differentiation and other aspects. Through the biological effects of MYCN, corresponding research will be carried out to clarify new ideas for the treatment of diabetes based on the mechanism of MYCN on methylation and ferroptosis.

Myo-Inositol oxygenase (MIOX) is a non heme ferritin whose transcription is regulated by oxidative stress, free fatty acids, and high glucose environments [42]. In diabetes, the expression of MIOX depends on the demethylation of the MIOX promoter under hyperglycemia [43]. MIOX overexpression promotes the production of reactive oxygen species (ROS), which can oxidize proteins, lipids, carbohydrates and DNA, leading to mitochondrial DNA damage and cell dysfunction [44]. Due to the upregulation of MIOX in T2DM, oxidative stress and endoplasmic reticulum stress are exacerbated, leading to mitochondrial energy metabolism disorders and exacerbating pancreatic islet beta cells damage [45]. MIOX can also regulate inflammatory damage by activating NLRP3 inflammasomes and altering the release of inflammatory mediators [46]. In the state of hyperglycemia, abnormal expression of MIOX activates NLRP3 inflammasomes, leading to increased release of inflammatory mediators, promoting inflammatory response, leading to pancreatic cell damage, and exacerbating insulin resistance. Research has shown that overexpression of MIOX promotes pancreatic islet beta cells ferroptosis, which may be due to the overexpression of MIOX leading to lipid peroxidation and a decrease in GPX4 activity and NADPH levels [47]. NADPH levels are sensitive to ferroptosis and help eliminate lipid hydroperoxides [48]. GPX4 is a key antioxidant enzyme, Ferroptosis can be blocked by inhibiting lipid hydroperoxides that reduce glutathione (GSH) [49]. When MIOX is overexpressed, on the one hand, it accelerates ferroptosis in pancreatic islet cells by downregulating GPX4 activity and intracellular GSH concentration, and on the other hand, it promotes ferroptosis and leads to insulin resistance by regulating intracellular levels of iron [49, 50]. In view of the role that MIOX plays in tissues, it can improve the oxidative stress, inflammatory reaction and ferroptosis of cells by regulating the expression level of MIOX, thus improving the blood sugar level of diabetes and the complications of diabetes.

Our study is the first to identify the key m6A regulated ferroptosis gene in T2DM through bioinformatics and machine learning methods, and conduct experimental validation. However, there are also certain shortcomings in our research. Firstly, the experimental subjects were only at the cellular level and did not undergo in vivo experiments. We will conduct further research in the following studies. In addition, the impact of four key m6A regulated ferroptosis related genes in T2DM and their biological mechanisms need to be further explored to improve their corresponding mechanisms.

Conclusion

In summary, our study integrated literature-reported m6A regulatory genes, ferroptosis related genes from the FerrDb database, and T2DM transcriptome data from the GEO database to identify m6A ferroptosis genes. We then used machine learning algorithms to identify four biomarkers of T2DM. Additionally, we constructed TF-mRNA regulatory and ceRNA networks of these biomarkers to predict potential drugs, which can provide potential targets for clinical diagnosis of T2DM and a theoretical basis for further understanding of the driving mechanism of T2DM. This study revealed the potential molecular mechanisms of m6A-related ferroptosis genes in T2DM, which could provide novel insights for the clinical diagnosis and treatment of T2DM.

Data availability

No datasets were generated or analysed during the current study.

Abbreviations

m6A:

N6-methyladenosine

T2DM:

Type 2 diabetes mellitus

GEO:

Gene Expression Omnibus

DEGs:

Differentially expressed genes

m6A-FRGs:

M6A-related ferroptosis genes

DEm6A-FRGs:

Differentially expressed m6A-FRGs

ANN:

Artificial neural network

RF:

Random forest

SVM-RFE:

Support vector machine recursive feature elimination

TF:

Transcription factors

ceRNA:

Competing endogenous RNAs

GSEA:

Gene Set Enrichment Analysis

HC:

Healthy controls

ROC:

Receiver operating characteristic

LRM:

Logistic regression model

DCA:

Decision curve

PVDF:

Polyvinylidene difluoride membranes

GO:

Gene Ontology

KEGG:

Kyoto Encyclopedia of Genes and Genomes

Ctrl:

Control group

HG:

High glucose group

WBC:

White blood cells

LIP:

Load and unstable iron pool

MIOX:

Myo-Inositol oxygenase

ROS:

Reactive oxygen species

GSH:

Glutathione

References

  1. Saeedi P, et al. Mortality attributable to diabetes in 20–79 years old adults, 2019 estimates: Results from the International Diabetes Federation Diabetes Atlas, 9(th) edition. Diabetes Res Clin Pract. 2020;162:108086.

    Article  PubMed  Google Scholar 

  2. Garber AJ, et al. Consensus Statement by the American Association of Clinical Endocrinologists and American College of Endocrinology on the comprehensive type 2 diabetes management algorithm - 2019 executive summary. Endocr Pract. 2019;25(1):69–100.

    Article  PubMed  Google Scholar 

  3. American Diabetes A. Standards of medical care in diabetes-2019 abridged for primary care providers. Clin Diabetes. 2019;37(1):11–34.

    Article  Google Scholar 

  4. Zhang C, Liu N. N6-methyladenosine (m6A) modification in gynecological malignancies. J Cell Physiol. 2022;237(9):3465–79.

    Article  CAS  PubMed  Google Scholar 

  5. Oerum S, et al. A comprehensive review of m6A/m6Am RNA methyltransferase structures. Nucleic Acids Res. 2021;49(13):7239–55.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Sendinc E, Shi Y. RNA m6A methylation across the transcriptome. Mol Cell. 2023;83(3):428–41.

    Article  CAS  PubMed  Google Scholar 

  7. Zhang BF, et al. M(6)A methylation-mediated elevation of SM22alpha inhibits the proliferation and migration of vascular smooth muscle cells and ameliorates intimal hyperplasia in type 2 diabetes mellitus. Biol Chem. 2022;403(3):317–29.

    Article  CAS  PubMed  Google Scholar 

  8. Benak D, et al. The role of m(6)A and m(6)Am RNA modifications in the pathogenesis of diabetes mellitus. Front Endocrinol (Lausanne). 2023;14:1223583.

    Article  PubMed  Google Scholar 

  9. Chen C, et al. Co-expression network revealed roles of RNA m(6)A methylation in human beta-cell of type 2 diabetes mellitus. Front Cell Dev Biol. 2021;9:651142.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Linkermann A, et al. Synchronized renal tubular cell death involves ferroptosis. Proc Natl Acad Sci U S A. 2014;111(47):16836–41.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Sha W, et al. Mechanism of ferroptosis and its role in type 2 diabetes mellitus. J Diabetes Res. 2021;2021:9999612.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Li D, et al. Quercetin alleviates ferroptosis of pancreatic beta cells in type 2 diabetes. Nutrients. 2020;12(10):2954.

  13. Feng ZH, et al. m6A-immune-related lncRNA prognostic signature for predicting immune landscape and prognosis of bladder cancer. J Transl Med. 2022;20(1):492.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Colaprico A, et al. TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data. Nucleic Acids Res. 2016;44(8):e71.

    Article  PubMed  Google Scholar 

  15. Wu T, et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Camb). 2021;2(3):100141.

    CAS  PubMed  Google Scholar 

  16. Kerker M. Classics and classicists of colloid and interface science 8 Einstein, Albert. J Colloid Interface Sci. 1989;129(1):291–5.

    Article  CAS  Google Scholar 

  17. Cruse H, et al. A modular artificial neural net for controlling a six-legged walking system. Biol Cybern. 1995;72(5):421–30.

    Article  CAS  PubMed  Google Scholar 

  18. Djokic et al. Sensitivity of personal computers to voltage sags and short interruptions. Power Delivery. IEEE Transactions on, 2005.

  19. Yu G, et al. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Shannon P, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Seeliger D, de Groot BL. Ligand docking and binding site analysis with PyMOL and Autodock/Vina. J Comput Aided Mol Des. 2010;24(5):417–22.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  22. Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455–61.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Kang C, et al. Quercetin inhibits the activity and function of dendritic cells through the TLR4/IRAK4/NF-κB signalling pathway. Contemp Oncol (Pozn). 2023;27(3):182–9.

    CAS  PubMed  Google Scholar 

  24. Jun JC, et al. Adipose HIF-1alpha causes obesity by suppressing brown adipose tissue thermogenesis. J Mol Med (Berl). 2017;95(3):287–97.

    Article  CAS  PubMed  Google Scholar 

  25. Marku A, et al. Iron metabolism in pancreatic beta-cell function and dysfunction. Cells. 2021;10(11):2841.

  26. Miao R, et al. Iron metabolism and ferroptosis in type 2 diabetes mellitus and complications: mechanisms and therapeutic opportunities. Cell Death Dis. 2023;14(3):186.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Muhammad SA, et al. Cisplatin’s potential for type 2 diabetes repositioning by inhibiting CDKN1A, FAS, and SESN1. Comput Biol Med. 2021;135:104640.

    Article  CAS  PubMed  Google Scholar 

  28. Chen YS, et al. MiR-1908/EXO1 and MiR-203a/FOS, regulated by scd1, are associated with fracture risk and bone health in postmenopausal diabetic women. Aging (Albany NY). 2020;12(10):9549–84.

    Article  CAS  PubMed  Google Scholar 

  29. Engeland K. Cell cycle regulation: p53–p21-RB signaling. Cell Death Differ. 2022;29(5):946–60.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Wu W, et al. Up-regulation of Fas reverses cisplatin resistance of human small cell lung cancer cells. J Exp Clin Cancer Res. 2010;29(1):49.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Xu EE, et al. SOX4 allows facultative β-Cell proliferation through repression of Cdkn1a. Diabetes. 2017;66(8):2213–9.

    Article  CAS  PubMed  Google Scholar 

  32. Dayeh T, et al. Genome-wide DNA methylation analysis of human pancreatic islets from type 2 diabetic and non-diabetic donors identifies candidate genes that influence insulin secretion. PLoS Genet. 2014;10(3):e1004160.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Watanabe A, et al. CD82 is a marker to isolate β cell precursors from human iPS cells and plays a role for the maturation of β cells. Sci Rep. 2021;11(1):9530.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Otte J, et al. MYCN function in neuroblastoma development. Front Oncol. 2020;10:624079.

    Article  PubMed  Google Scholar 

  35. Horn S, et al. Research resource: a dual proteomic approach identifies regulated islet proteins during beta-cell mass expansion in vivo. Mol Endocrinol. 2016;30(1):133–43.

    Article  CAS  PubMed  Google Scholar 

  36. Fajas L, et al. Impaired pancreatic growth, beta cell mass, and beta cell function in E2F1 (-/- )mice. J Clin Invest. 2004;113(9):1288–95.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. Lu Y, et al. MYCN mediates TFRC-dependent ferroptosis and reveals vulnerabilities in neuroblastoma. Cell Death Dis. 2021;12(6):511.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  38. Sheean ME, et al. Activation of MAPK overrides the termination of myelin growth and replaces Nrg1/ErbB3 signals during Schwann cell development and myelination. Genes Dev. 2014;28(3):290–303.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Jacovetti C, et al. MicroRNAs contribute to compensatory beta cell expansion during pregnancy and obesity. J Clin Invest. 2012;122(10):3541–51.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Zeineldin M, et al. MYCN amplification and ATRX mutations are incompatible in neuroblastoma. Nat Commun. 2020;11(1):913.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Li HL, et al. A review of the regulatory mechanisms of N-Myc on cell cycle. Molecules. 2023;28(3):1141.

  42. Dutta RK, et al. Beneficial effects of myo-inositol oxygenase deficiency in cisplatin-induced AKI. J Am Soc Nephrol. 2017;28(5):1421-36.

  43. Sharma I, et al. High glucose-induced hypomethylation promotes binding of Sp-1 to myo-inositol oxygenase: implication in the pathobiology of diabetic tubulopathy. Am J Pathol. 2017;187(4):724–39.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Wu J, et al. Sources and implications of NADH/NAD(+) redox imbalance in diabetes and its complications. Diabetes Metab Syndr Obes. 2016;9:145–53.

    CAS  PubMed  PubMed Central  Google Scholar 

  45. Sharma I, et al. Myo-inositol Oxygenase (MIOX) overexpression drives the progression of renal tubulointerstitial injury in diabetes. Diabetes. 2020;69(6):1248–63.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  46. Zhou W, Yu C, Long Y. Myo-inositol oxygenase (MIOX) accelerated inflammation in the model of infection-induced cardiac dysfunction by NLRP3 inflammasome. Immun Inflamm Dis. 2023;11(5):e829.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. Deng F, et al. Myo-inositol oxygenase expression profile modulates pathogenic ferroptosis in the renal proximal tubule. J Clin Invest. 2019;129(11):5033–49.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  48. Stockwell BR, et al. Ferroptosis: a regulated cell death nexus linking metabolism, redox biology, and disease. Cell. 2017;171(2):273–85.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Ingold I, et al. Selenium utilization by GPX4 is required to prevent hydroperoxide-induced ferroptosis. Cell. 2018;172(3):409-422 e21.

    Article  CAS  PubMed  Google Scholar 

  50. Kinowaki Y, et al. Glutathione peroxidase 4 overexpression inhibits ROS-induced cell death in diffuse large B-cell lymphoma. Lab Invest. 2018;98(5):609–19.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We express our gratitude to the U.S. GEO database and China FerrDb database for providing open data and sharing relevant research data on type 2 diabetes, iron death and m6A with all participating researchers.

Compliance with ethics Statement

This study did not involve humans directly and the previous studies that utilized the GEO data and FerrDb data used in this study have already obtained ethical approval and patient consent. Therefore, this study does not require additional ethical approval.

Funding

This study was supported by Natural Science Foundation of Ningxia Province (Program No. 2022A1251), the Hainan Province Science and Technology Special Fund (Grant No. ZDYF2024SHFZ124).

Author information

Author notes

  1. Jing Wang and Xuying Li contributed equally to this work.

Authors and Affiliations

  1. Department of Anaesthesiology, Northwest Women’s and Children’s Hospital, Yanxiang Road, Yanta District, Xi’an, 710000, Shanxi Province, China

    Jing Wang & Juan Geng

  2. Department of Anesthesiology, Hainan General Hospital, Hainan Affiliated Hospital of Hainan Medical University, Haikou, Hainan, China

    Xuying Li

  3. State Key Laboratory of Transient Optics and Photonics, Xi’an Institute of Optics and Precision Mechanics, Chinese Academy of Science, Xi’an, 710119, China

    Ruiduo Wang

  4. Department of Anesthesiology, General Hospital of Ningxia Medical University, Yinchuan, China 704 Shengli Street, Yinchuan, 750004, Ningxia, China

    Gang Ma

  5. Department of Neurology, Renmin Hospital, Hubei University of Medicine, Maojian District, No. 39, Chaoyang Middle Road, Shiyan, Hubei, 442000, People’s Republic of China

    Pan Liu

Authors
  1. Jing Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Xuying Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Juan Geng
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Ruiduo Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Gang Ma
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Pan Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

W.J and L.XY share the first author of the article. M.G and L.P are the corresponding authors of this article. L.P proposed the viewpoint of the article, designed experimental ideas, supervised the research, and made revisions, reviews, and proofreads to the article. M.G analyzed the feasibility of the experiment, provided technical support, and made revisions to the initial draft. W.J and L.XY conducted experiments, consulted references, and wrote the first draft. G.J analyzed the data and calibrated it. All authors interpreted the data,made critical revisions to the manuscript, and approved the final version of the manuscript.

Corresponding authors

Correspondence to Gang Ma or Pan Liu.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary Material 1: Supplement Table 1 Differentially expressed genes between T2D groups and HC groups.

41065_2025_385_MOESM2_ESM.xlsx

Supplementary Material 2: Supplement Table 2 Ferroptosis genes significantly associated with N6-methyladenosine (m6A)-regulated genes.

41065_2025_385_MOESM3_ESM.xlsx

Supplementary Material 3: Supplement Table 3 The 8 Gene Ontology (GO), Biological Process (BP) terms, and 17 GO Molecular Function (MF) terms were enriched for 10 differentially expressed m6A-FRGs (DEm6A-FRGs).

41065_2025_385_MOESM4_ESM.xlsx

Supplementary Material 4: Supplement Table 4 The 13 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were enriched for 10 DEm6A-FRGs.

Supplementary Material 5: Supplement Table 5 Construction of 59 pairs of TF-mRNAs in the TF-mRNA regulatory network.

41065_2025_385_MOESM6_ESM.xlsx

Supplementary Material 6: Supplement Table 6 Construction of 150 pairs of regulatory relationships in the competing endogenous RNAs (ceRNA) regulatory network.

41065_2025_385_MOESM7_ESM.xlsx

Supplementary Material 7: Supplement Table 7 Construction of ceRNA network with 4 biomarkers, 51 miRNAs, and 37 lncRNAs.

41065_2025_385_MOESM8_ESM.tif

Supplementary Material 8: Figure S1 MDs of CD82 and bisphenol A. (A-B) RMSD curve of CD82 protein. (C-D) RMSF plot of amino acid flexibility and exercise intensity of CD82 protein. (E–F) RoG plot of CD82 protein. (G) dynamic distance changes of key amino acids and ligands

41065_2025_385_MOESM9_ESM.tif

Supplementary Material 9: Figure S2 MDs of MIOX and perfluorooctanoic acid (A-B) RMSD curve of MIOX protein. (C-D) RMSF plot of amino acid flexibility and exercise intensity of MIOX protein. (E–F) RoG plot of MIOX protein. (G) dynamic distance changes of key amino acids and ligands

41065_2025_385_MOESM10_ESM.tif

Supplementary Material 10: Figure S3 MDs of MYCN and bisphenol A. (A-B) RMSD curve of MYCN protein. (C-D) RMSF plot of amino acid flexibility and exercise intensity of MYCN protein. (E–F) MYCN protein RoG plot. (G) dynamic distance changes of key amino acids and ligands

Supplementary Material 11

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Wang, J., Li, X., Geng, J. et al. Identification of biomarkers and mechanism exploration of ferroptosis related genes regulated by m6A in type 2 diabetes mellitus. Hereditas 162, 24 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s41065-025-00385-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s41065-025-00385-9

Keywords

Hereditas

ISSN: 1601-5223

Contact us