GRAPHICAL ABSTRACT

INTRODUCTION

Diabetes mellitus (DM) is a metabolic disease that substantially contributes to cardiovascular complications, which remain a leading cause of mortality among affected patients [1]. Among these complications, coronary artery disease (CAD) is the most prevalent in individuals with diabetes. When diabetes and CAD coexist, a condition referred to as DM-CAD, disease progression is often accelerated and accompanied by increased clinical complexity [2]. Clinical studies have demonstrated that patients with DM-CAD exhibit more pronounced intimal thickening, heightened plaque vulnerability, increased rates of restenosis, and a greater incidence of adverse cardiovascular events [3,4]. Although current therapeutic strategies, including glycemic control, lipid-lowering agents, and antiplatelet therapy, are widely implemented, their efficacy in mitigating the immune-inflammatory milieu of DM-CAD remains limited, and disease recurrence or progression is frequently observed [5,6]. Accumulating evidence indicates that, beyond classical metabolic abnormalities, immune dysregulation and immunometabolic reprogramming play central roles in the pathogenesis of DM-CAD [7,8]. Notably, macrophages represent the predominant immune cell population involved in the initiation and progression of atherosclerotic plaques, and their functional states directly shape the inflammatory microenvironment and influence plaque stability [9,10]. Accordingly, elucidating the mechanisms governing macrophage metabolism and inflammation in DM-CAD has emerged as a critical focus in both basic cardiovascular research and clinical intervention.

Macrophages play a pivotal role in atherosclerotic lesions, and their functional states are tightly regulated by cellular metabolic pathways [11,12]. Traditionally, macrophages have been regarded primarily as lipid scavengers that differentiate into foam cells, thereby promoting atherogenesis. However, accumulating evidence demonstrates that metabolic processes, including glycolysis and oxidative phosphorylation, directly modulate macrophage pro-inflammatory and anti-inflammatory phenotypes [13,14]. In the diabetic milieu, chronic hyperglycemia and insulin resistance lead to sustained activation of glycolytic pathways in macrophages, resulting in lactate accumulation, increased production of reactive oxygen species (ROS), and enhanced secretion of pro-inflammatory cytokines, which collectively exacerbate inflammatory responses [15,16]. In addition, macrophage metabolism is closely linked to essential cellular functions such as phagocytosis, migration, and intercellular communication, and dysregulation of these processes contributes to the persistence of immune imbalance within atherosclerotic plaques [17,18]. A growing body of research underscores the bidirectional relationship between macrophage metabolic reprogramming and chronic inflammation, suggesting that metabolic pathways themselves may represent viable therapeutic targets for anti-inflammatory intervention [19,20]. Despite these advances, the specific molecular drivers responsible for sustained metabolic and functional abnormalities in macrophages during DM-CAD remain poorly characterized [21,22]. Identifying upstream regulators of macrophage metabolism and inflammatory responses is therefore essential for advancing mechanistic insight into DM-CAD pathophysiology.

In recent years, RNA methylation has emerged as a critical layer of epigenetic regulation, with N⁶-methyladenosine (m⁶A) modification receiving increasing attention for its roles in immune cell development, function, and metabolic regulation [23,24]. m⁶A modification influences multiple post-transcriptional processes, including RNA splicing, stability, and translation efficiency, thereby contributing to the regulation of diverse pathological conditions [25,26]. YTH N6-methyladenosine RNA binding protein F1 (YTHDF1), a key m⁶A ‘reader’ protein, recognizes and binds m⁶A-modified RNA sites and facilitates RNA stabilization and translational regulation. This protein has been implicated in a range of inflammation-associated diseases, including tumor immunity, neurodegenerative disorders, and autoimmune conditions [27,28]. In parallel, long non-coding RNAs (lncRNAs), as important components of the regulatory RNA landscape, have been shown to play broad roles in inflammatory and metabolic processes. Within atherosclerosis, several lncRNAs have been closely linked to macrophage phenotype switching and lipid metabolism [29,30]. Emerging studies further suggest that lncRNAs represent key targets of m⁶A modification, with their stability, subcellular localization, and biological activity subject to regulation by the m⁶A–YTHDF1 axis [31-33]. Nevertheless, the specific mechanisms through which this regulatory network operates in DM-CAD remain largely unexplored. In particular, whether m⁶A modifications contribute to lncRNA-mediated macrophage functional remodeling in the diabetic atherosclerotic context lacks direct experimental evidence. Accordingly, systematic investigation of m⁶A-dependent lncRNA regulation in macrophage immunometabolism may reveal previously unrecognized regulatory networks underlying chronic metabolic-inflammatory diseases.

In our preliminary bioinformatics analysis based on single-cell RNA sequencing (scRNA-seq) and bulk RNA-seq datasets derived from the Gene Expression Omnibus (GEO) database using a DM-CAD mouse model, we identified a significant up-regulation of the lncRNA Opa interacting protein 5 antisense RNA 1 (OIP5-AS1) in macrophages, which showed a positive correlation with pro-inflammatory gene expression. Although OIP5-AS1 has previously been implicated in cell cycle regulation and tumor metabolism [34,35], its role in immune-inflammatory diseases remains largely unclear. Further screening using least absolute shrinkage and selection operator (LASSO) regression and random forest algorithms revealed that YTHDF1 and OIP5-AS1 exhibit highly consistent expression patterns and shared cell-type specificity, suggesting a potential m⁶A-dependent regulatory relationship. A literature review combined with bioinformatic prediction analyses indicated that OIP5-AS1 harbors multiple putative m⁶A modification sites with a high probability of YTHDF1 binding. These observations collectively raise a novel hypothesis that YTHDF1 may stabilize OIP5-AS1 through m⁶A modification, thereby activating metabolic pathways and promoting inflammatory responses in macrophages. Systematic validation of this regulatory axis may provide critical insight into the epigenetic mechanisms underlying immunometabolic reprogramming in DM-CAD and offer a theoretical foundation for future targeted therapeutic strategies.

Building on the aforementioned background and preliminary findings, this study aims to systematically elucidate how YTHDF1-dependent m⁶A modification regulates the stability and function of lncRNA OIP5-AS1 and to define its critical role in driving macrophage metabolic reprogramming and inflammatory responses in DM-CAD. Through integrative multiomics analyses, single-cell transcriptomic mining, differential gene screening, and machine-learning approaches, we identified OIP5-AS1 and YTHDF1 as key candidate regulators. The m⁶A-dependent regulatory mechanism was further validated using RNA immunoprecipitation followed by quantitative polymerase chain reaction (RIP-qPCR), site-directed mutagenesis, and actinomycin D (ActD)-based RNA stability assays. In addition, metabolomic profiling, Oil Red O (ORO) staining, and ROS assays were employed to comprehensively assess the role of OIP5-AS1 in glycolytic activation, foam cell formation, and pro-inflammatory processes in macrophages. Collectively, our findings reveal a critical contribution of the m⁶A–YTHDF1–OIP5-AS1 axis to the pathogenesis of DM-CAD and provide a novel epigenetic perspective on immune-metabolic reprogramming. From a clinical standpoint, the molecular pathways and potential therapeutic targets identified in this study may offer a theoretical basis and strategic direction for early intervention and personalized treatment of DM-CAD, thereby advancing precision medicine in cardiometabolic diseases.

METHODS

Single-cell RNA sequencing data acquisition and processing

The single-cell transcriptomic dataset (accession number: GSE211216) was obtained from the GEO database. Data processing and downstream analyses were conducted using the Seurat package in R (R Foundation for Statistical Computing, Vienna, Austria). Quality control (QC) filtering criteria included retaining cells with 200< nFeature_RNA <5,000 and mitochondrial gene content (percent.mt) <20%. For downstream analyses, the top 2,000 highly variable genes, identified based on variance, were selected.

All animal experiments were approved by the Animal Ethics Committee (No. 2025-DWSYLLCZ-64).

Single-cell transcriptomic analysis

To reduce the dimensionality of the scRNA-seq dataset, principal component analysis (PCA) was performed using the top 2,000 highly variable genes identified based on variance. The first 20 principal components (PCs) were selected for downstream analysis using the ElbowPlot function implemented in the Seurat package. Cell clustering was performed using the FindClusters function in Seurat, with the resolution parameter set to the default value (res=1). Nonlinear dimensionality reduction was subsequently conducted using the uniform manifold approximation and projection (UMAP) algorithm. Marker genes for each cell cluster were identified using Seurat, and cell-type annotation was performed based on established lineage-specific markers in combination with the CellMarker database and the ‘SingleR’ R package.

Pseudotime trajectory construction

To reconstruct state transition pathways of immune cells in the DM-CAD and control-CAD groups, pseudotime trajectory analysis was performed using the Monocle2 package (Trapnell Lab, Seattle, WA, USA) on preprocessed and annotated single-cell expression matrices. Based on prior UMAP analysis, specific cell subpopulations, including macrophages and endothelial cells, were extracted, and highly variable genes were selected for dimensionality reduction. The discriminative dimensionality reduction tree (DDRTree) algorithm was then applied for low-dimensional embedding and trajectory inference. By designating endothelial cells as the root of the trajectory, we examined the distribution patterns and state transitions of different cell types along pseudotime, thereby characterizing the developmental branching of macrophages under pathological conditions and their degree of coupling with endothelial cells. Finally, trajectory plots were generated to visualize dynamic changes in cell states and to compare structural differences between the DM-CAD and control-CAD groups.

Cell–cell communication analysis

To systematically investigate intercellular communication patterns among cell subpopulations in the DM-CAD and control-CAD groups, the CellChat R package was employed to infer cell–cell communication networks based on single-cell transcriptomic data. Using previously annotated cell clusters and normalized expression matrices, CellChat objects were constructed, and the built-in ligand–receptor (L–R) interaction database was applied to identify potential communication events. Subsequently, the number and strength of signaling interactions between cell types were quantified to evaluate overall communication activity and to identify significantly altered signaling pathways. Further analyses focused specifically on macrophage-endothelial cell interactions, from which cell-type–specific enriched L–R pairs were extracted. Key signaling pathways, including visfatin, tumor necrosis factor (TNF), and macrophage migration inhibitory factor (MIF), were visualized using network diagrams, heatmaps, and bubble plots to illustrate signaling sources and target responsiveness. These analyses revealed aberrant activation of pro-inflammatory signaling networks under diabetic conditions.

Bulk RNA-seq and differential expression analysis

To induce DM, 6-week-old male apolipoprotein E–deficient (Apoe^−/−) mice (strain code: 221; Beijing Vital River Laboratory Animal Technology Co. Ltd., Beijing, China) were intraperitoneally injected with streptozotocin (55 mg/kg; S0130, Sigma-Aldrich, St. Louis, MO, USA) or citrate buffer (control; 77-92-9, Sigma-Aldrich) once daily for 5 consecutive days, thereby establishing a DM-CAD mouse model. Mice were maintained on a standard chow diet and monitored for 10 weeks. After this period, mice were euthanized by CO₂ inhalation, and aortic tissues were collected for subsequent analyses.

Total RNA (≥1 μg) was extracted from aortic tissues and purified using DNase I treatment and silica membrane columns (RNeasy Mini Kit, 74004, Qiagen, Hilden, Germany). RNA concentration was measured using the Qubit RNA HS Assay Kit (Q32852, Thermo Fisher Scientific, Waltham, MA, USA) and diluted to 100 ng/μL. RNA integrity was assessed using a Fragment Analyzer (Agilent 5400, Agilent, Santa Clara, CA, USA), and samples with RNA quality scores ≥6 were selected for library preparation. Library construction was performed under International Organization for Standardization (ISO)/International Electronical Commission (IEC) 17025–certified conditions using the TruSeq RNA Library Prep Kit v2 (RS-122-2302, Illumina, San Diego, CA, USA). Messenger RNA was enriched using oligo(dT) magnetic beads (MS04T, VDO Biotech, Suzhou, China), fragmented, reverse-transcribed into complementary DNA (cDNA), and subjected to adapter ligation and PCR amplification.

Sequencing was performed on an Illumina HiSeq 2500 V4 platform (Illumina), generating 126-bp paired-end reads with a minimum output of 12.5 Gbp per sample (approximately 50 million read pairs). Base calling, image analysis, and quality verification were conducted using the Illumina data analysis pipeline (RTA v1.18.64 and Bcl2fastq v1.8.4). RNA-seq reads were exported in compressed Sanger FASTQ format.

Differentially expressed genes (DEGs) were identified using the ‘edgeR’ package in R, with thresholds set at |log₂ fold change| >1 and an adjusted P value <0.05. Functional enrichment analyses, including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses, were conducted using the ‘clusterProfiler’ package and visualized with ‘ggplot2.’

LASSO regression

LASSO regression is a commonly used approach for variable selection and regularization in regression modeling. By imposing a penalty on the absolute size of regression coefficients, LASSO shrinks some coefficients toward zero, thereby eliminating less informative variables and enhancing model interpretability. The optimal regularization parameter (λ) was determined by cross-validation, during which models were fitted across a range of λ values and the value yielding optimal performance was selected. Under the optimal λ, coefficients for certain variables were reduced exactly to zero, facilitating feature selection. In this study, LASSO regression analysis was conducted using the ‘glmnet’ package in R.

Identification of key genes using the random forest algorithm

Using RNA-seq data from all samples, gene importance was evaluated with the ‘randomForest’ package in R. The number of decision trees was set to 500, and model error was visualized as a function of tree number. The optimal model was selected by identifying the minimum error point using the which.min() function. Gene importance was quantified using the MeanDecreaseGini metric, with higher values indicating greater importance [36].

Immune infiltration

Based on bulk RNA-seq expression profiles derived from aortic tissues, the immune cell infiltration landscape was inferred using the Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts (CIBERSORT) algorithm. All analyses were performed in R (v4.2.1) using the e1071 and preprocess-Core packages. The LM22 signature matrix was employed as the reference dataset, and species adaptation was conducted by incorporating mouse-specific immune cell marker genes reported in the literature to ensure analytical validity for murine samples. The resulting CIBERSORT output scores were used to estimate the relative proportions of 22 immune cell types across the different experimental groups. Statistical analyses were conducted within the R environment, and immune infiltration patterns were visualized using the ggplot2 package.

Untargeted metabolomics and differential analysis

Cells from the vector group and the overexpression OIP5-AS1 (oe-OIP5-AS1) group (n=6 per group) were collected for untargeted metabolomic analysis. For each sample, 300 μL of cell lysate was transferred into a 1.5-mL polypropylene tube, followed by the addition of 900 μL of 80% methanol (CAS No. 67-56-1, Sigma-Aldrich) containing 0.1% formic acid (CAS No. 64-18-6, Sigma-Aldrich). The mixture was vortexed for 2 minutes and centrifuged at 12,000 ×g for 10 minutes. The resulting supernatant was collected and transferred into autosampler vials for subsequent analysis.

Metabolite detection was performed using an LC20 ultra-high-performance liquid chromatography system (Shimadzu, Kyoto, Japan) coupled to a Triple TOF 6600 mass spectrometer (AB Sciex, Framingham, MA, USA). Chromatographic separation was achieved using a Waters ACQUITY UPLC HSS T3 C18 column (100×2.1 mm, 1.8 μm), maintained at 40°C with a flow rate of 0.4 mL/min. Mobile phase A consisted of water containing 0.1% formic acid, whereas mobile phase B consisted of acetonitrile (CAS No. 75-05-8, Sigma-Aldrich) supplemented with 0.1% formic acid. The gradient elution program was set as follows: 0.0–11.0 minutes, B=5%; 11.0–12.0 minutes, B=90%; and 12.1–14.0 minutes, B=5%. The eluate was introduced directly into the mass spectrometer, and metabolite signals were acquired in both positive and negative ion modes to ensure comprehensive metabolic coverage.

During the analytical process, one QC sample was injected after every six experimental samples to monitor the stability of the liquid chromatography–mass spectrometry (LC–MS) system. The relative standard deviation (RSD) was calculated for all detected peaks, and all RSD values were below 20%, satisfying standard QC criteria for LC–MS-based metabolomic analyses. Raw data were subjected to peak extraction and normalization, followed by half-minimum imputation to address missing values. Metabolites with a missing rate exceeding 50% across all samples were excluded to minimize bias in subsequent statistical analyses.

Differential metabolites were identified using thresholds of fold change ≥2 or ≤0.5 in combination with a Student’s t test P<0.05. Metabolite annotation was conducted by referencing multiple spectral databases, including Human Metabolome Database (HMDB), METLIN, and KEGG, supplemented by accurate mass-retention time (AMRT) matching. Only metabolites with matching scores ≥80% were retained for downstream analyses. Functional enrichment analysis of the differential metabolites was performed using MetaboAnalyst version 5.0 (https://www.metaboanalyst.ca).

PCA and orthogonal partial least squares discriminant analysis

To evaluate the global metabolic impact of OIP5-AS1 overexpression in macrophages, untargeted metabolomic data were initially subjected to PCA. This unsupervised dimensionality reduction method was used to extract the primary sources of variance and to assess sample distribution patterns and potential outliers. PCA was performed separately under both positive and negative ionization modes, and two-dimensional score plots were generated to visualize metabolic differences between the oe-OIP5-AS1 and control groups.

Subsequently, orthogonal partial least squares discriminant analysis (OPLS-DA) was conducted as a supervised classification approach to enhance group discrimination while removing irrelevant intra-group variation. Cross-validation was applied to assess model robustness and predictive performance. The resulting score plots demonstrated clear separation between the oe-OIP5-AS1 and vector groups in metabolic feature space, confirming that OIP5-AS1 overexpression induces statistically significant alterations in the macrophage metabolic profile.

Cell culture and transfection

The human monocytic cell line THP-1 (CL-0233, Wuhan Pricella Biotechnology Co. Ltd., Wuhan, China) was differentiated into adherent macrophages by treatment with 100 ng/mL phorbol 12-myristate 13-acetate (PMA; P8139, Sigma-Aldrich) for 48 hours. Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Waltham, MA, USA) supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin/streptomycin (Invitrogen, Carlsbad, CA, USA) under standard incubation conditions. Both differentiated macrophages and corresponding control cells (CC) were exposed to high-glucose medium (40 mM) for 48 hours to establish the experimental condition, while isotonic mannitol treatment was used in the control group to account for osmotic effects. Transfection was performed using Lipofectamine 3000 reagent (Invitrogen) with either an oe-OIP5-AS1 or an empty vector control (vector: short hairpin [sh]-negative control [NC]+oe-NC). To generate a loss-of-function model, a short hairpin RNA targeting OIP5-AS1 (sh-OIP5-AS1; sequence: 5′-ACCTCGGCTGGAGTGATGCG TGATTTTTCAAGAGAAAATCACGCATCACTCCAGCCTT-3′) was also transfected. Cells were harvested 48 hours after transfection for downstream analyses. To assess whether the effects of OIP5-AS1 were dependent on glycolytic metabolism, selected experiments included treatment with the glycolysis inhibitor 2-deoxyglucose (2-DG; D8375, Sigma-Aldrich) at a final concentration of 10 mM for 6 hours following transfection.

Prediction analysis of m⁶A modification sites

To evaluate potential m⁶A modification sites on the lncRNA OIP5-AS1, the full-length transcript sequence was analyzed using the sequence-based RNA adenosine methylation site predictor (SRAMP) online prediction platform. The analysis was performed in ‘full transcript mode,’ and predicted m⁶A sites were ranked according to their combined confidence scores.

RIP-qPCR assay

RIP-qPCR was performed using the EZ-Magna RIP RNA-binding protein immunoprecipitation kit (17-701, Millipore, Burlington, MA, USA). Experimental groups included a magnetic bead-only control, a non-specific immunoglobulin G (IgG) control, an RNase pre-treatment group, and a YTHDF1 antibody group. Cells were lysed to release RNA–protein complexes. Specific antibodies were incubated with magnetic beads to generate antibody–bead conjugates, which were subsequently incubated with cell lysates to enrich target RNA–protein complexes. After multiple washing steps to remove non-specific binding, the complexes were treated with proteinase K buffer to digest proteins and release the bound RNA. RNA was then purified by phenol:chloroform:isoamyl alcohol extraction followed by ethanol precipitation and subsequently reverse-transcribed into cDNA for downstream analysis. Immunoprecipitation was carried out using an anti-YTHDF1 antibody (ab220162, Abcam, Cambridge, UK) or a rabbit IgG isotype control antibody (ab172730, Abcam). The enriched RNA was extracted, reverse-transcribed, and subjected to qPCR analysis. Primers used for reverse transcription quantitative polymerase chain reaction (RT-qPCR) were as follows: OIP5-AS1-F, 5′-AGCCGGATTTCTCTCATGCT-3′; OIP5-AS1-R, 5′-TGTGAGCTGTGTGGTTCCTC-3′.

Construction and functional validation of YTHDF1 mutants

Site-directed mutagenesis of YTHDF1 was performed to target two key residues within its functional domain, K395 and Y397, using the KOD-Plus-Mutagenesis Kit (SMK-101, TOYOBO, Osaka, Japan). A mutant expression plasmid, mutant YTHDF1-Flag (YTHDF1-MUT-Flag), was generated by introducing a Cterminal Flag tag (DYKDDDDK) to facilitate subsequent immunoprecipitation assays. THP-1–derived macrophages were transfected with either wild-type YTHDF1-Flag (YTHDF1-WT-Flag) or YTHDF1-MUT-Flag constructs using Lipofectamine 3000 (Invitrogen). At 48 hours post-transfection, RIP assays were performed to evaluate the binding affinity of the different YTHDF1 variants to OIP5-AS1.

RNA stability assay

To assess the regulatory effect of YTHDF1 on OIP5-AS1 RNA stability, cells from each experimental group were treated with ActD (A1410, Sigma-Aldrich) at a final concentration of 5 μg/mL to inhibit nascent RNA synthesis. Samples were collected at 0, 2, 4, and 6 hours after treatment, with three biological replicates obtained at each time point. Total RNA was extracted and subjected to RT-qPCR analysis. Reverse transcription was performed using the HiScript III RT SuperMix for qPCR (R323-01, Vazyme, Nanjing, China), followed by qPCR using the ChamQ Universal SYBR qPCR Master Mix (Q711, Vazyme). Nonlinear regression analysis was conducted using GraphPad Prism version 10.1.2 (GraphPad Software Inc., San Diego, CA, USA), and intergroup comparisons were evaluated by one-way analysis of variance (ANOVA).

Foam cell model construction

THP-1 cells were treated with 100 ng/mL PMA for 48 hours to induce differentiation into macrophages. Foam cell formation was subsequently induced by incubation with oxidized low-density lipoprotein (ox-LDL; final concentration, 50 μg/mL; YB-002, Yiyuan Biotechnologies, Guangzhou, China) for 24 hours [37].

Foam cell formation assay

Macrophages were co-incubated with 25 μg/mL of native low-density lipoprotein, ox-LDL, or CC for 24 hours to induce foam cell formation, with assessments performed every 8 hours during incubation. After treatment, non-adherent cells were removed, and adherent cells were fixed with neutral buffered formalin for 30 minutes. Following fixation, cells were washed with endotoxin-free water (Sigma-Aldrich) and stained with 0.3% ORO for 45 minutes. Cells were subsequently washed again with endotoxin-free water, and intracellular lipid accumulation was evaluated under a light microscope. Images were acquired using a Zeiss Axio microscope (Carl Zeiss Microscopy GmbH, Jena, Germany) at 20× magnification, capturing 10 non-overlapping fields per well while avoiding the well edges. The percentage of ORO-positive foam cells was calculated relative to the total cell number per well. For quantitative analysis, cells were decolorized with 60% isopropanol for 15 seconds, washed three times with endotoxin-free water, and then treated with 99% isopropanol for 1 minute to extract intracellular ORO. Absorbance was measured at 405 nm using a Promega GloMax Discover microplate reader (Promega, Madison, WI, USA), as described previously [38].

Oil red O staining for lipid accumulation

Lipid accumulation in ox-LDL–treated cells was assessed using an ORO staining solution (G1262, Solarbio, Beijing, China). Cells were fixed with 4% paraformaldehyde (P0099, Beyotime, Haimen, China), pre-treated with 60% isopropanol, and stained according to standard protocols. Excess dye was removed, and images were captured using an inverted microscope (Carl Zeiss AG). Quantification of lipid deposition was performed by measuring the stained area using ImageJ software v1.53 (National Institutes of Health, Bethesda, MD, USA).

Determination of intracellular cholesterol content

Total cholesterol (TC) and cholesteryl ester (CE) levels were measured using a commercial cholesterol assay kit (A111-1-1, Nanjing Jiancheng Bioengineering Institute, Nanjing, China) according to the manufacturer’s instructions. Cellular lipids were extracted, and absorbance values were recorded using a multifunctional microplate reader (SpectraMax iD5, Molecular Devices, San Jose, CA, USA). TC and CE concentrations were calculated and normalized to total protein content, with results expressed as μg/mg protein.

Reverse transcription quantitative polymerase chain reaction

Total RNA was extracted from cells using TRIzol Reagent (15596026, Invitrogen). Reverse transcription was performed using the HiScript III RT SuperMix for qPCR (R323-01, Vazyme), followed by quantitative real-time PCR using the ChamQ Universal SYBR qPCR Master Mix (Q711, Vazyme). All samples were analyzed in triplicate. Briefly, total RNA was isolated following the TRIzol protocol, including cell lysis, chloroform phase separation, isopropanol precipitation, and ethanol washing. RNA concentration and purity were assessed spectrophotometrically, with optical density [OD]₂₆₀/OD₂₈₀ ratios between 1.8 and 2.0 considered acceptable. cDNA was synthesized from purified RNA using reverse transcriptase and gene-specific primers. qPCR reactions contained cDNA templates, gene-specific primers, and SYBR Green Master Mix, and no-template controls were included to exclude contamination. Amplification was performed using a real-time PCR system under standard cycling conditions, and cycle threshold (Ct) values were automatically recorded. Relative gene expression levels were calculated using the 2^{−Δ ΔCt} method to compare expression differences among experimental groups. Target genes included cholesterol uptake receptors (CD36, class A1 scavenger receptor [SR-A1]), cholesterol efflux transporters (ATP-binding cassette transporter A1 [ABCA1], ATP-binding cassette subfamily G member 1 [ABCG1]), and inflammatory cytokines (interleukin 1β [IL-1β], TNF-α, and IL-6). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the internal control. All primers were synthesized by Tsingke Biotechnology (Beijing, China), and primer sequences are provided in Supplemental Table S1.

Western blot analysis

Cells were lysed in radio-immunoprecipitation assay buffer (Thermo Fisher Scientific) supplemented with protease and phosphatase inhibitors (Roche, Basel, Switzerland). Protein concentrations were determined using a bicinchoninic acid protein assay kit (Thermo Fisher Scientific). Equal amounts of protein (30 μg per lane) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis on 12% gels and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore). Membranes were incubated overnight at 4°C with primary antibodies against YTHDF1 (ab220162, 1:1,000, Abcam) and GAPDH (ab9485, 1:2,500, Abcam), which served as a loading control. After washing, membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (Cell Signaling Technology, Danvers, MA, USA) and developed using an enhanced chemiluminescence (ECL) detection system (Thermo Fisher Scientific). Images were captured using a chemiluminescence imaging system (ChemiDoc XRS+, Bio-Rad, Hercules, CA, USA) and quantified with Image Lab software v6.1 (Bio-Rad).

DOT blot assay

Appropriately sized membranes, including nitrocellulose (NC) or PVDF, were prepared for dot blot analysis. PVDF membranes were pre-activated by immersion in methanol for approximately 1 minute and subsequently rinsed with distilled water or Tris-buffered saline with Tween-20 (TBST) buffer, whereas NC membranes were soaked in distilled water prior to use. Processed samples, including protein solutions or denatured nucleic acids, were directly spotted onto the membranes at volumes of 1–5 μL per sample and allowed to air-dry at room temperature or baked at 37°C to immobilize the samples. Membranes were then incubated in blocking buffer containing 5% non-fat milk or bovine serum albumin at room temperature for 1 to 2 hours with gentle agitation to prevent non-specific antibody binding. Blocked membranes were incubated with appropriately diluted primary antibodies in blocking buffer for 1 to 2 hours at room temperature or overnight at 4°C. After incubation, membranes were washed three to five times with TBST buffer for 5 minutes per wash to remove unbound primary antibodies, followed by incubation with HRP-conjugated (or otherwise labeled) secondary antibodies diluted in blocking buffer for 1 to 1.5 hours at room temperature. Membranes were subsequently washed an additional three to five times with TBST to eliminate unbound secondary antibodies. Signal detection was performed using chemiluminescent substrates, such as ECL, or chromogenic substrates, such as 3,3′-diaminobenzidine (DAB), depending on the labeling strategy. Signals were visualized and documented using an imaging system, as described previously [39].

Lactate concentration measurement

To assess the impact of OIP5-AS1 overexpression on glycolytic activity, culture supernatants were collected from THP-1–derived macrophages. L-lactate concentrations were quantified using a YSI 2900 biochemical analyzer (YSI Incorporated, Yellow Springs, OH, USA). All samples were analyzed in triplicate, and lactate concentrations were normalized using a standard calibration curve. The lactate production rate was calculated as the ratio of lactate concentrations measured at 1 and 6 hours, using the formula lactate_1h/lactate_1h.

Intracellular ROS detection

To determine whether OIP5-AS1 overexpression induces oxidative stress, intracellular ROS levels were measured using the fluorescent probe 2′,7′-dichlorofluorescin diacetate (DCFDA). THP-1–derived macrophages were washed with Hank’s Balanced Salt Solution and incubated at 37°C with 100 μM DCFDA (D399, Invitrogen) for 30 minutes. After incubation, cells were rinsed with phosphate-buffered saline (PBS) and counterstained Hoechst 33342 to label nuclei. Fluorescence images were captured using a Zeiss fluorescence microscope (Carl Zeiss) equipped with a 40× objective. Fluorescence intensity was quantified using Zeiss Zen Blue software. For each experimental group, five randomly selected fields were imaged, and all experiments were performed in triplicate.

Mitochondrial reactive oxygen species detection

Mitochondrial ROS (mtROS) levels were assessed using Mito-SOX Red (catalog no. M36008, Invitrogen). THP-1-derived macrophages were washed with PBS and incubated at 37°C with 2 μM MitoSOX Red for 30 minutes. Following incubation, cells were washed with PBS and counterstained with Hoechst 33342 to label nuclei. Images were captured using a Zeiss fluorescence microscope (Carl Zeiss) equipped with a 40× objective, and red fluorescence intensity was quantified using Zen Blue software. Three biological replicates were included for each group, with five random fields imaged per replicate.

Mitochondrial membrane potential assessment

Mitochondrial membrane potential (ΔΨm) was assessed using the 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbo cyanine iodide (JC-1) staining kit (C2006, Beyotime Biotechnology). THP-1–derived macrophages were washed with PBS and incubated with JC-1 dye (5 μg/mL) at 37°C for 20 minutes. After incubation, cells were washed with PBS and counterstained with Hoechst 33342 to label nuclei. Fluorescence images were acquired using a Zeiss fluorescence microscope (Carl Zeiss) equipped with a 40× objective. The ratio of red-to-green fluorescence intensity was used to evaluate changes in ΔΨm, and quantitative image analysis was performed using Zeiss Zen Blue software. For each group, three biological replicates were analyzed, with five random fields imaged per replicate.

Statistical analysis

All data are presented as mean±standard deviation (SD) and are derived from at least three independent experiments. Statistical analyses were performed using GraphPad Prism version 10.0 (GraphPad Software). Comparisons between two groups were conducted using Student’s t test, whereas comparisons among multiple groups were performed using one-way or two-way ANOVA, as appropriate for the experimental design, followed by Bonferroni’s post hoc test. A P≤0.05 was considered statistically significant. Statistical significance is indicated in the figures as follows: P<0.05, P<0.01, P<0.001, and P<0.0001.

RESULTS

Increased developmental trajectory coupling between macrophages and endothelial cells in DM-CAD

To investigate intercellular regulatory dynamics within the immune microenvironment of DM-CAD, particularly the potential influence of macrophages on endothelial cell function, we first obtained the scRNA-seq dataset GSE211216 from a publicly available database. This dataset comprises arterial tissue derived from a DM-CAD mouse model. A comprehensive single-cell transcriptomic analysis was subsequently performed (Fig. 1A). During preprocessing, the dataset was integrated using the Seurat package. High-quality cells were retained based on standard QC criteria, including nFeature_RNA >200 and <5,000, nCount_RNA >1,000 and <20,000, and mitochondrial gene content (percent.mt) <20%, yielding an expression matrix containing 14,008 genes across 6,882 cells (Supplemental Fig. S1A). Correlation analysis demonstrated a moderate negative correlation between nCount_RNA and percent.mt (r=−0.2) and a strong positive correlation between nCount_RNA and nFeature_RNA (r=0.93), indicating that the filtered dataset was of high quality and suitable for downstream analysis (Supplemental Fig. S1B).

Following quality filtering, the data were normalized, and highly variable genes were selected for dimensionality reduction using PCA (Supplemental Fig. S1C). The distribution of cells along PC_1 and PC_2 suggested the presence of batch effects among samples (Supplemental Fig. S1D). To address this issue, batch correction was performed using the Harmony package. PC SDs were ranked using the ElbowPlot function (Supplemental Fig. S1E, F), and post-correction analyses confirmed effective mitigation of batch effects (Supplemental Fig. S1G). Based on the top 20 PCs obtained after Harmony correction, a k-nearest neighbor graph was constructed to model cell-to-cell relationships. Graph-based clustering was then performed using the Louvain algorithm, enabling identification of transcriptionally distinct cell populations. The clustree package was used to visualize cluster stability and hierarchical relationships across multiple resolution parameters (Supplemental Fig. S2A). Ultimately, cells were partitioned into 20 distinct clusters, and UMAP was applied for nonlinear dimensionality reduction and visualization (Supplemental Fig. S2B, C).

Building on the Louvain clustering results, cell identity was further characterized using both t-distributed stochastic neighbor embedding and UMAP for dimensionality reduction and visualization. Through this integrative approach, eight major cell subpopulations were identified, including macrophages, endothelial cells, smooth muscle cells, and additional cell types. A small subset of cells was annotated as neuron-associated cells (NACs), which were likely attributable to sample heterogeneity or minor contamination (Fig. 1B). Cell-type annotation was performed based on the expression of well-established lineage-specific marker genes, including fibroblasts (collagen type I alpha 1 chain [Col1a1], podoplanin [Pdpn]), smooth muscle cells (myosin heavy chain 11 [Myh11], actin alpha 2, smooth muscle [Acta2]), macrophages (Cd68, Cd14), endothelial cells (cadherin 5 [Cdh5], platelet and endothelial cell adhesion molecule 1 [Pecam1]), T cells (CD3 delta subunit of T-cell receptor complex [Cd3d], CD3 epsilon subunit of T-cell receptor complex [Cd3e]), neutrophils (S100 calcium binding protein A9 [S100a9], lymphocyte antigen 6 family member G [Ly6g]), B cells (Cd19, Cd79a), and NACs (killer cell lectin like receptor K1 [Klrk1], killer cell lectin-like receptor, subfamily A, member 8 [Klra8]). DotPlot visualization (Supplemental Fig. S2D) illustrated the expression patterns of these canonical markers across clusters, thereby confirming the accuracy and reliability of cell-type annotations.

Cell composition analysis revealed that, compared with the control group, the DM-CAD group exhibited a significant increase in smooth muscle cell abundance and a marked reduction in macrophage proportions (Fig. 1C). These shifts suggest that the diabetic state may substantially alter immune and stromal cell distribution within the vascular microenvironment.

To explore cellular state transitions, pseudotime trajectory analysis was conducted using Monocle2. In the control-CAD group, endothelial cells were predominantly positioned at early stages of the trajectory, consistent with a relatively homeostatic vascular environment, whereas macrophages appeared at later stages with limited overlap along the trajectory, suggesting localized inflammatory interactions. In contrast, within the DM-CAD group, macrophages were distributed along trajectory branches closely connected to endothelial cells, exhibiting a high degree of developmental coupling and forming a continuous, tightly aligned state transition gradient (Fig. 1D). This pattern indicates that under diabetic pathological conditions, proinflammatory macrophages engage in more extensive and sustained interactions with endothelial cells, potentially contributing to the amplification of vascular inflammation.

Collectively, scRNA-seq analysis demonstrated that within the immune microenvironment of DM-CAD, macrophages not only undergo pronounced alterations in relative abundance but also display enhanced developmental coupling with endothelial cells along functional trajectories. These findings suggest that under hyperglycemic conditions, pro-inflammatory macrophages may perpetuate vascular inflammation and tissue injury through close interaction with endothelial cells, thereby accelerating atherosclerotic progression.

Single-cell communication analysis reveals increased macrophage-endothelial interactions mediated by inflammatory L–R pairs

To further elucidate intercellular communication mechanisms between macrophage subpopulations and vascular endothelial cells within the immune microenvironment of DM-CAD, we employed the CellChat computational framework to systematically infer cell–cell communication networks across all cellular subtypes in the GSE211216 single-cell transcriptomic dataset. This analysis was designed to identify macrophage-driven inflammatory signaling pathways that may contribute to DM-CAD pathogenesis.

Comparative analysis between the control-CAD and DM-CAD groups revealed that overall intercellular signaling activity was markedly elevated in the DM-CAD group, as evidenced by significant increases in both the number and strength of signaling interactions (Fig. 2A). Pathway enrichment analysis showed that signaling activity in the control-CAD group was primarily associated with secreted phosphoprotein 1 (SPP1), ANNEXIN, and neurotrophin (NT) pathways, whereas the DM-CAD group exhibited pronounced activation of inflammatory and growth factor–related pathways, including TNF, pleiotrophin (PTN), transforming growth factor-β, and insulin-like growth factor (IGF) (Fig. 2B). These results suggest that the arterial microenvironment under diabetic conditions is more susceptible to proinflammatory signaling and vascular remodeling processes.

Further examination of L–R pair distributions and communication intensities revealed significantly elevated interaction scores between macrophages and other cell populations, particularly endothelial and neutrophil subsets, in the DM-CAD group (Fig. 2C). This finding suggests that macrophages under diabetic pathological conditions may remain in a persistently activated state, thereby reinforcing inflammatory interactions with endothelial cells through specific signaling pathways.

Subsequently, we constructed a comprehensive signaling pathway interaction map for the DM-CAD group (Supplemental Fig. S3A) and identified key intercellular communication networks. Heatmap analysis of cell–pathway correlations demonstrated that, in the outgoing signaling mode, macrophages predominantly communicated with neighboring cells via the visfatin, IGF, TNF, galectin, C-C motif chemokine ligand (CCL), and MIF pathways. In contrast, in the incoming signaling mode, endothelial cells showed strong responsiveness to the visfatin, semaphorin 3 (SEMA3), TNF, CXC motif chemokine ligand (CXCL), angiopoietin-like protein (ANGPTL), midkine (MK), and MIF pathways (Supplemental Fig. S3B). Notably, visfatin, TNF, and MIF signaling were highly active in both macrophages and endothelial cells, indicating their potential role as central mediators of macrophage-endothelial communication.

To further characterize macrophage-endothelial communication, we identified specific L–R pairs between these two cell types. Among the most prominent pairs with high signaling strength were Tnf–TNF receptor superfamily member 1A (Tnfrsf1a), nicotinamide phosphoribosyltransferase (Nampt)–(integrin subunit alpha 5 [Itga5]+integrin subunit beta 1 [Itgb1]), and Mif–atypical chemokine receptor 3 (Ackr3), all of which are closely associated with inflammatory signaling, chemotaxis, and cell adhesion processes (Supplemental Fig. S3C).

Finally, differential pathway analysis of visfatin, TNF, and MIF signaling revealed enhanced directional communication from macrophages to endothelial cells in the DM-CAD group. These findings indicate that, within the diabetic vascular microenvironment, macrophages may promote endothelial activation and injury through multiple pro-inflammatory signaling axes (Fig. 2D–F).

In summary, single-cell communication network analysis demonstrates that, within the DM-CAD microenvironment, macrophages markedly enhance their interactions with vascular endothelial cells through multiple inflammation-associated pathways, including visfatin, TNF, and MIF. This aberrant pro-inflammatory communication may exacerbate endothelial dysfunction and thereby contribute to the progression of atherosclerosis.

OIP5-AS1 and YTHDF1 are significantly upregulated and involved in inflammatory immune regulation

To investigate molecular differences between DM-CAD and isolated CAD (control-CAD), with a particular focus on non-coding RNAs and epigenetic regulators within the immune microenvironment, we established DM-CAD and control-CAD mouse models and performed transcriptomic sequencing of arterial tissues (Fig. 3A).

Differential expression analysis identified 174 genes that were upregulated and 26 genes that were downregulated in the DM-CAD group compared with the control-CAD group (Fig. 3B). GO enrichment analysis indicated that these DEGs were primarily associated with processes such as the ‘immune system process,’ ‘nicotinamide adenine dinucleotide (NAD) biosynthetic process,’ ‘peptide antigen-transporting adenosine triphosphatase (ATPase) activity,’ and ‘oxidized nicotinamide adenine dinucleotide (NAD⁺) ADP-ribosyltransferase activity,’ suggesting coordinated dysregulation of metabolic and immune functions in DM-CAD arterial tissue (Fig. 3C). Consistently, KEGG pathway analysis demonstrated significant enrichment of pathways including the ‘tumor necrosis factor (TNF) signaling pathway,’ ‘glycolysis/gluconeogenesis,’ ‘hypoxia-inducible factor 1 (HIF-1) signaling pathway,’ and ‘fructose and mannose metabolism’ in the DM-CAD group, reflecting prominent metabolic reprogramming and heightened inflammatory signaling under diabetic conditions (Fig. 3D).

To identify key regulatory genes, we applied LASSO regression for feature selection, with cross-validation used to determine the optimal lambda value. This analysis identified four signature genes strongly associated with the DM-CAD condition: Opa interacting protein 5, opposite strand 1 (OIP5OS1; also known as OIP5-AS1), hexokinase 2 (HK2), C-C motif chemokine ligand 5 (CCL5), and YTHDF1 (Fig. 3E). Box plot analysis further confirmed that all four genes were significantly upregulated in the DM-CAD group, with OIP5OS1 exhibiting the most pronounced difference (Fig. 3F).

Subsequently, a random forest algorithm was used to assess the relative importance of the identified genes. OIP5-AS1 and YTHDF1 displayed the highest MeanDecreaseGini values within the classification model, indicating their potential roles as key contributors to DM-CAD pathogenesis (Fig. 3G).

Previous studies have shown that OIP5-AS1 plays a regulatory role in inflammatory responses and metabolic reprogramming [40], particularly through modulation of macrophage-mediated inflammation and oxidative stress [41]. In parallel, YTHDF1 has been reported to sustain M1 macrophage polarization by promoting glycolytic activity [42].

To further validate the immunoregulatory relevance of these genes, we applied the CIBERSORT algorithm to evaluate immune cell infiltration patterns. In the DM-CAD group, pro-inflammatory immune cell populations, including M1 macrophages and CD8⁺ T cells, were significantly increased, whereas anti-inflammatory or immunoregulatory populations, such as M2 macrophages and activated dendritic cells, were markedly reduced, indicating a strongly pro-inflammatory immune microenvironment (Supplemental Fig. S4A, B).

Correlation analysis between OIP5-AS1, YTHDF1, and 22 immune cell subsets revealed that OIP5-AS1 expression was negatively correlated with M0 macrophages, plasma cells, and CD4⁺ memory resting T cells, but positively correlated with activated mast cells, M1 macrophages, CD8⁺ T cells, and regulatory T cells (Tregs) (Supplemental Fig. S4C). A similar pattern was observed for YTHDF1, which showed negative correlations with M0 macrophages, neutrophils, plasma cells, and CD4⁺ memory resting T cells, and positive correlations with activated mast cells, M1 macrophages, and Tregs (Supplemental Fig. S4D). These results suggest that both OIP5-AS1 and YTHDF1 may function as critical regulators of pro-inflammatory responses, macrophage polarization, and T-cell–mediated immunity.

Collectively, integrative transcriptomic and immune infiltration analyses identified OIP5-AS1 and YTHDF1 as key regulatory factors in DM-CAD. Their upregulation in DM-CAD tissues and positive associations with pro-inflammatory immune cell populations suggest their involvement in metabolic-inflammatory signaling crosstalk that drives endothelial injury and immune imbalance. These findings not only enhance our understanding of DM-CAD pathogenesis but also provide a theoretical foundation for the development of targeted therapeutic strategies involving OIP5-AS1 and YTHDF1.

YTHDF1 increases the stability of lncRNA OIP5-AS1 through m⁶A recognition and binding

To determine whether lncRNA OIP5-AS1 harbors m⁶A modification potential, a series of analyses and validation experiments were conducted (Fig. 4A). Using the SRAMP prediction tool, we analyzed the full-length OIP5-AS1 sequence and identified 15 putative m⁶A modification sites. Among these, only a single site located at position 112 achieved high confidence, with a combined score of 0.606. Notably, the local sequence flanking this site (‘GGAGGCCGAGUGCAUCUGUUGGACCGUGCGAGGAGAAGAAAAAAA’) conforms to the canonical DRACH consensus motif (D=A/G/U, R=A/G, H=A/C/U) for m⁶A modification (Fig. 4B, Supplemental Table S2).

To validate the interaction between YTHDF1 and OIP5-AS1, RIP-qPCR was performed in THP-1-derived macrophages (Fig. 4C). The results demonstrated significant enrichment of OIP5-AS1 RNA in YTHDF1 immunoprecipitates, indicating a direct binding interaction. To further assess the requirement of m⁶A recognition, a mutant YTHDF1 construct (YTHDF1-MUT) was generated to disrupt the m⁶A-binding domain (Fig. 4D), and successful transfection of both WT and mutant proteins was confirmed by Western blotting using an anti-Flag antibody (Fig. 4E). Subsequent RIP-qPCR analysis showed that enrichment of OIP5-AS1 by YTHDF1-MUT was markedly reduced compared with that observed for YTHDF1-WT (Fig. 4F). In parallel, RIPqPCR using an m⁶A-specific antibody revealed that YTHDF1-WT significantly enhanced m⁶A enrichment of OIP5-AS1, whereas YTHDF1-MUT failed to do so, yielding signals comparable to those of the IgG NC. Collectively, these findings indicate that YTHDF1 promotes OIP5-AS1 binding and stability through m⁶A-dependent recognition (Fig. 4G).

To further confirm that YTHDF1 enhances OIP5-AS1 stability through m⁶A recognition, YTHDF1 expression was silenced using shRNA, and knockdown efficiency was verified (Fig. 4H). Transcription was subsequently inhibited using ActD, and total RNA was collected at multiple time points to assess OIP5-AS1 expression by RT-qPCR. The results showed that both YTHDF1 knockdown and disruption of its m⁶A-binding capability significantly reduced the stability and half-life of OIP5-AS1 (Fig. 4I, J).

Collectively, these data demonstrate that YTHDF1 enhances the stability of OIP5-AS1 by recognizing specific m⁶A modification sites, thereby providing a mechanistic basis for its downstream regulatory functions (Fig. 4K). In addition, a dot blot assay was performed to assess the specificity of the m⁶A antibody. Methylated and unmethylated RNA probes were included as controls to ensure assay reliability. The results confirmed that the antibody exhibited high specificity for m⁶A-modified RNA, producing strong signals exclusively in methylated probes (Fig. 4L).

OIP5-AS1 enhances glycolytic pathway activity in macrophages

To determine whether OIP5-AS1 contributes to metabolic reprogramming in macrophages through modulation of glycolytic pathways, we established an OIP5-AS1 overexpression model (oe-OIP5-AS1) in THP-1–derived macrophages, with empty-vector-transfected cells serving as controls. Untargeted metabolomic profiling was subsequently performed to comprehensively characterize OIP5-AS1–associated metabolic alterations (Fig. 5A).

A total of 1,939 metabolic feature peaks were detected using the LC–MS platform. Through AMRT matching combined with secondary spectral comparison against multiple databases, including HMDB, METLIN, and KEGG, a broad spectrum of metabolites was identified and annotated. PCA revealed a clear separation between the oe-OIP5-AS1 and vector groups, indicating substantial differences in global metabolic profiles (Fig. 5B). This distinction was further validated by OPLS-DA, which confirmed statistically significant metabolic separation between the two groups (Fig. 5C).

Differential metabolite analysis revealed significant enrichment and upregulation of key glycolytic intermediates in the oe-OIP5-AS1 group, including lactate, glucose-6-phosphate (G6P), 6-phosphogluconic acid, and glucose-1-phosphate. All identified metabolites exhibited fold changes greater than 1.5 and P values below 0.05 (Fig. 5D). Metabolite annotations were validated by AMRT and MS/MS confirmation, ensuring high confidence in metabolite identification. These findings suggest that OIP5-AS1 promotes macrophage metabolic reprogramming by increasing glycolytic flux, potentially through the accumulation of glycolysis-associated intermediates.

To further delineate the metabolic pathways involved, pathway enrichment analysis of differential metabolites was performed using MetaboAnalyst 5.0 under both positive and negative ion modes. KEGG pathway analysis demonstrated significant enrichment of glycolysis/gluconeogenesis, the tricarboxylic acid (TCA) cycle, pyruvate metabolism, and glycerophospholipid metabolism (Fig. 5E). Complementary analysis using the Small Molecule Pathway Database identified enrichment of the Warburg effect, gluconeogenesis, and mitochondrial acetyl-co-enzyme A transport pathways (Fig. 5F), further supporting a role for OIP5-AS1 in regulating macrophage energy metabolism.

In summary, metabolomic profiling demonstrates that OIP5-AS1 promotes metabolic reprogramming in macrophages by enhancing glycolysis- and TCA cycle–associated pathways. This metabolic shift may contribute to macrophage pro-inflammatory activation and underscores the potential pathogenic role of OIP5-AS1 in inflammatory conditions such as DM-CAD.

OIP5-AS1 promotes foam cell formation and upregulates pro-inflammatory cytokine expression, contributing to plaque instability

To assess whether OIP5-AS1-mediated metabolic reprogramming influences macrophage susceptibility to foam cell–inducing stimuli, an ox-LDL–induced foam cell model was established using THP-1-derived macrophages. Foam cell formation was examined under conditions of OIP5-AS1 overexpression (oe-OIP5-AS1) and knockdown (sh-OIP5-AS1) (Fig. 6A). ORO staining revealed substantial intracellular lipid droplet accumulation in vector-transfected macrophages following ox-LDL exposure. This lipid accumulation was further increased in the oe-OIP5-AS1 group, whereas the sh-OIP5-AS1 group exhibited markedly reduced lipid deposition (Fig. 6B). Quantitative analysis showed that TC and CE levels were significantly elevated in the oe-OIP5-AS1 group and reduced in the sh-OIP5-AS1 group compared with controls (Fig. 6C, D), indicating that OIP5-AS1 enhances macrophage susceptibility to foam cell formation.

To elucidate the underlying mechanism, we examined the expression of genes involved in cholesterol uptake and efflux. qPCR analysis demonstrated that overexpression of OIP5-AS1 significantly upregulated the cholesterol uptake receptors CD36 and SR-A1, while concomitantly downregulating the cholesterol efflux transporters ABCA1 and ABCG1. In contrast, OIP5-AS1 knockdown resulted in reduced expression of CD36 and SR-A1 and increased expression of ABCA1 and ABCG1 (Fig. 6E–H). When integrated with our prior findings of glycolytic pathway enrichment, these results suggest that OIP5-AS1 promotes lipid accumulation by enhancing cholesterol uptake and impairing cholesterol efflux through metabolic reprogramming, thereby facilitating foam cell formation.

To evaluate the impact of foam cell-associated metabolic reprogramming on the local inflammatory microenvironment, we further quantified the expression of pro-inflammatory cytokines IL-1β, TNF-α, and IL-6. Expression of all three cytokines was significantly increased in the oe-OIP5-AS1 group and markedly reduced in the sh-OIP5-AS1 group (Fig. 6I-K). These findings indicate that OIP5-AS1 not only promotes lipid accumulation and foam cell formation but also enhances transcription of inflammatory mediators, thereby exacerbating chronic inflammation within atherosclerotic plaques (Fig. 6L).

OIP5-AS1 activates glycolytic pathways to exacerbate macrophage foam cell formation and oxidative stress

To determine whether OIP5-AS1–mediated foam cell formation and pro-inflammatory cytokine expression depend on activation of glycolytic pathways, we introduced the glycolysis inhibitor 2-DG into THP-1–derived macrophages overexpressing OIP5-AS1 and established a corresponding experimental workflow (Fig. 7A).

ORO staining revealed that, under ox-LDL stimulation, intracellular lipid accumulation was markedly increased in the oe-OIP5-AS1 group. In contrast, treatment with 2-DG significantly reduced staining intensity and lipid droplet area, indicating that OIP5-AS1-induced foam cell formation is dependent on sustained glycolytic activity (Fig. 7B). Consistently, measurements of TC and CE levels showed significant elevation in the oe-OIP5-AS1 group, which was substantially attenuated following 2-DG treatment (Fig. 7C), further linking enhanced foam cell formation to glycolytic metabolism.

Gene expression analysis of cholesterol metabolism–related markers showed that OIP5-AS1 overexpression significantly upregulated the cholesterol uptake receptors CD36 and SR-A1 while downregulating the cholesterol efflux transporters ABCA1 and ABCG1. Conversely, 2-DG treatment reversed these expression patterns, resulting in decreased CD36 and SR-A1 levels and increased ABCA1 and ABCG1 expression (Fig. 7D). These findings indicate that inhibition of glycolysis effectively ameliorates OIP5-AS1-induced dysregulation of cholesterol metabolism in macrophages.

Analysis of pro-inflammatory cytokine expression further demonstrated that OIP5-AS1 significantly enhanced transcription of IL-1β, TNF-α, and IL-6, whereas 2-DG treatment markedly suppressed their expression (Fig. 7E), supporting a role for OIP5-AS1 in glycolysis-driven inflammatory activation. To confirm effective inhibition of glycolytic flux, lactate concentrations in the culture supernatant were measured and found to be significantly reduced in the 2-DG–treated oe-OIP5-AS1 group (Fig. 7F).

To further evaluate the impact of OIP5-AS1–induced glycolytic reprogramming on mitochondrial function, mtROS production and membrane potential were assessed. MitoSOX and DCFDA staining revealed significantly elevated mitochondrial and total ROS levels in the oe-OIP5-AS1 group compared with vector controls, indicating enhanced oxidative stress. Notably, 2-DG treatment substantially reduced ROS accumulation to near-control levels (Fig. 7G–I). In parallel, JC-1 staining demonstrated a marked reduction in ΔΨm, reflected by a decreased red-to-green fluorescence ratio, in the oe-OIP5-AS1 group, indicative of mitochondrial dysfunction. This reduction was partially reversed following 2-DG treatment (Fig. 7J, K).

Collectively, these findings demonstrate that OIP5-AS1 activates glycolytic pathways, leading to increased mtROS production and loss of ΔΨm, thereby inducing mitochondrial stress. This stress response likely acts in concert with metabolic reprogramming to promote foam cell formation and inflammatory activation. Together, these results support a model in which OIP5-AS1 amplifies inflammatory signaling by driving glycolysis-dependent mitochondrial dysfunction.

DISCUSSION

YTHDF1, a canonical ‘reader’ protein of m⁶A modification, has been extensively studied for its roles in regulating protein translation in tumor cells, embryonic development, and immune responses [43,44]. However, its function in metabolic cardiovascular diseases remains poorly understood, particularly in the context of DM-associated atherosclerosis [45]. In this study, we demonstrate for the first time that YTHDF1 is highly expressed in DM-CAD and stabilizes the lncRNA OIP5-AS1 in macrophages through m⁶A-binding activity, thereby modulating downstream metabolic and inflammatory pathways. These findings not only broaden the functional scope of YTHDF1 but also indicate its pathological relevance beyond malignant diseases. In contrast to YTHDF2, which primarily mediates RNA degradation, YTHDF1 promotes RNA stability and translation and may exhibit target selectivity within macrophages. Mechanistically, this study provides direct evidence that YTHDF1 binds a specific lncRNA target, thereby filling a critical knowledge gap in inflammatory cardiovascular disease research and laying the groundwork for the future development of YTHDF1-targeted inhibitors or modulators.

OIP5-AS1, a lncRNA, has previously been implicated in cancer progression, including roles in cell proliferation, apoptosis, and drug resistance, supporting its oncogenic potential [35,46]. In contrast, its function in cardiovascular diseases—particularly in immune cell metabolism—has remained largely unexplored [47,48]. In the present study, we identified significant upregulation of OIP5-AS1 in DM-CAD models, with strong positive correlations to pro-inflammatory gene expression, indicating its involvement in inflammatory processes. More importantly, we demonstrated that OIP5-AS1 overexpression enhances glycolysis, foam cell formation, and ROS accumulation, thereby directly contributing to macrophage metabolic reprogramming and phenotypic modulation. Compared with previously studied cardiovascular lncRNAs such as metastasis associated lung adenocarcinoma transcript 1 (MALAT1) and nuclear paraspeckle assembly transcript 1 (NEAT1), OIP5-AS1 not only exerts inflammatory regulatory effects but also plays a pivotal role in metabolic control, potentially serving as a regulatory hub at the intersection of metabolism and inflammation. Collectively, these findings extend the functional characterization of OIP5-AS1 beyond tumor biology to a novel role in immune-metabolic regulation, representing a significant conceptual advance.

In the field of RNA modification research, the functional implications of m⁶A sites have emerged as a major focus. However, many existing studies rely predominantly on bioinformatic predictions [49,50]. In contrast, we systematically validated the presence of m⁶A modification sites on OIP5-AS1 and demonstrated their dependence on YTHDF1 binding through multiple experimental approaches, including RIP-qPCR, site-directed mutagenesis, and ActD-based RNA stability assays. Compared with studies based solely on computational inference, our work provides robust experimental confirmation. Mechanistically, we propose that YTHDF1 protects OIP5-AS1 from degradation by binding to its m⁶A-modified sites, thereby prolonging its half-life and enhancing its biological activity. This finding not only expands current understanding of m⁶A-mediated regulation of lncRNAs but also introduces an m⁶A–lncRNA regulatory axis into the pathogenic framework of DM-CAD. Furthermore, we observed that OIP5-AS1 may influence transcription factors involved in glycolytic regulation, suggesting that its regulatory effects may extend beyond RNA-level modulation to upstream transcriptional control.

The metabolic state of macrophages is tightly linked to their immunological function, with glycolytic activation commonly associated with a pro-inflammatory phenotype [51,52]. As a prototypical chronic inflammatory condition, DM-CAD is characterized by dysregulated glucose metabolism in lesion-resident macrophages; however, the causal relationship between metabolic disturbance and inflammatory amplification remains incompletely defined [53,54]. Our findings demonstrate that OIP5-AS1 markedly upregulates glycolysis-related gene expression, accompanied by lactate accumulation, increased ROS production, and enhanced foam cell formation. Foam cell accumulation contributes to reduced plaque stability [55], indicating that OIP5-AS1 plays a critical role in amplifying inflammatory metabolism. Moreover, under sustained ox-LDL stimulation, foam cell formation progressively increased, further compromising plaque stability [56]. Unlike prior studies that have directly targeted glycolytic enzymes [57,58], our approach indirectly modulates metabolic pathways through lncRNA regulation, potentially offering greater specificity and regulatory depth. Notably, glycolytic inhibition with 2-DG effectively reversed OIP5-AS1-induced pathological phenotypes, further substantiating the pathogenic role of glycolytic reprogramming. Together, these findings strengthen the conceptual basis for targeting immunometabolism in DM-CAD and highlight RNA modification as a previously underappreciated layer of metabolic regulation.

Despite its widespread use, 2-DG is a relatively non-specific glycolytic inhibitor with well-documented off-target effects. In addition to inhibiting glycolysis, 2-DG perturbs multiple intracellular processes, sometimes producing effects unrelated to its intended metabolic inhibition. In macrophages, 2-DG has been shown to suppress both glycolysis and oxidative phosphorylation, leading to intracellular ATP depletion and disruption of Janus kinase (JAK)-signal transducer and activator of transcription 6 (STAT6) signaling. In astrocytes, 2-DG treatment induces endoplasmic reticulum stress-related genes, including heat shock protein family A (Hsp70) member 5 (HSPA5)/glucose-regulated protein 78 (GRP78) [59]. Furthermore, 2-DG has been reported to inhibit neutrophil extracellular trap formation and alter macrophage polarization states. These effects likely reflect broad metabolic suppression rather than pathway-specific modulation, rendering the overall impact of 2-DG context-dependent. Accordingly, there is a clear need to employ more selective glycolytic inhibitors in future studies. Because replacing 2-DG with more specific compounds would require substantial additional experimentation, such optimization will be addressed in subsequent in-depth investigations.

Through analysis of the GSE211216 single-cell transcriptomic dataset, we identified a convergent developmental trajectory between macrophages and endothelial cells in DM-CAD, suggesting coordinated alterations in cellular architecture and signaling under inflammatory conditions. We further observed marked activation of inflammatory signaling pathways, including VISFATIN, TNF, and MIF, within intercellular communication networks, indicating the presence of a complex immune amplification system in the DM-CAD microenvironment. In contrast to prior studies that have primarily emphasized T-cell infiltration and Th1/Th2 balance [60,61], our work highlights macrophage-endothelial interactions as a central pathogenic axis. Activation of these communication pathways may represent a key mechanism sustaining persistent plaque inflammation. We propose that macrophage metabolic state may modulate the intensity of macrophage-endothelial signaling, forming a ‘metabolism–communication amplification loop.’ This conceptual framework, proposed here for the first time in the context of diabetic cardiovascular complications, offers both mechanistic insight and potential therapeutic relevance.

This study integrates single-cell transcriptomics, bulk RNA-seq, and untargeted metabolomics, in combination with LASSO regression and random forest modeling, to robustly identify key regulatory factors. Compared with conventional approaches that rely solely on differential expression analysis [62,63], machine-learning algorithms enable the identification of the most predictive variables in high-dimensional datasets, thereby reducing data redundancy and false discovery rates. Moreover, cross-validation across multiple omics platforms ensured consistency in the expression patterns of key factors, strengthening biological credibility. This integrative strategy is particularly well-suited for complex pathological conditions such as DM-CAD, in which multifactorial and multi-pathway mechanisms coexist. Accordingly, the multi-omics framework employed here may serve as a methodological reference for investigating other chronic inflammatory metabolic disorders and may facilitate the advancement of translational medicine strategies.

Previous studies investigating the mechanisms underlying DM-CAD have primarily focused on classical inflammatory mediators, dysregulated lipid metabolism, and endothelial dysfunction, whereas investigations into epigenetic regulation remain relatively limited [64,65]. This study addresses this gap by exploring the regulatory roles of RNA modifications and non-coding RNAs, thereby extending mechanistic inquiry beyond phenotypic observations to upstream regulatory networks. We identified a pathogenic role for the lncRNA OIP5-AS1 and demonstrated that its stability is regulated by YTHDF1-mediated m⁶A modification, establishing a coherent upstream regulatory axis. These findings reinforce the theoretical framework of RNA modification–mediated immunometabolic regulation, enrich the etiological understanding of DM-CAD, and provide new perspectives for the development of non-traditional therapeutic strategies. In addition, other m⁶A reader proteins may participate in the regulation of OIP5-AS1 expression. For example, YTHDF3 has been reported to coordinate with YTHDF1 and YTHDF2 to promote mRNA translation and degradation, thereby influencing the expression dynamics of target transcripts, including OIP5-AS1 [66]. Importantly, our findings were validated using both murine models and THP-1-derived human macrophages; nevertheless, further confirmation in clinical samples is required to strengthen the translational relevance of these results.

Despite these strengths, this study has several limitations. First, the sample size was relatively small, with bulk RNA-seq performed on three samples per group (n=3) and scRNA-seq conducted on one sample per group (n=1), which may limit statistical power. Consequently, the present findings should primarily be interpreted as mechanistic hypotheses that require validation in larger and multicenter cohorts. Second, functional validation relied predominantly on cellular and murine models. In addition, whether other m⁶A reader proteins or demethylases, beyond YTHDF1, contribute to OIP5-AS1 regulation remains to be determined. Future studies should incorporate analyses of human atherosclerotic plaque tissues and utilize OIP5-AS1 knockout mouse models to further evaluate associations between OIP5-AS1 and clinical DM-CAD progression. Expanding functional analyses to additional immune cell populations and metabolic disease contexts may further enhance translational potential. Moreover, clinical validation in this study was based on a single patient cohort without independent replication; therefore, increased sample sizes and multicenter validation will be essential to improve the robustness of the conclusions.

In this study, we integrated scRNA-seq, bulk RNA-seq, and metabolomics data (Fig. 8) to systematically elucidate the molecular mechanism by which YTHDF1-dependent m⁶A modification stabilizes the lncRNA OIP5-AS1, thereby driving macrophage metabolic reprogramming and exacerbating inflammatory imbalance in DM-CAD. Our findings demonstrate that upregulation of OIP5-AS1 significantly enhances glycolytic activity, promotes foam cell formation and mtROS accumulation, and subsequently activates pro-inflammatory pathways that contribute to immune dysregulation within the atherosclerotic plaque microenvironment. As an m⁶A reader protein, YTHDF1 directly binds OIP5-AS1, enhancing its stability and providing an epigenetic foundation for this pathogenic regulatory axis (Fig. 9).

This study is the first to define the functional role of OIP5-AS1 in accelerating atherosclerosis under diabetic conditions and to propose that its regulatory mechanism depends on YTHDF1-mediated m⁶A methylation. These findings expand current understanding of the m⁶A–lncRNA interaction network in cardiovascular inflammatory metabolism. The stable expression of OIP5-AS1 is closely associated with glycolytic activation, highlighting its potential utility as an early diagnostic biomarker and therapeutic target for metabolic inflammation. Collectively, these insights offer new opportunities for precision-based strategies to treat cardiovascular complications in diabetes.

Supplementary Material

Article information

CONFLICTS OF INTEREST

No potential conflict of interest relevant to this article was reported.

ACKNOWLEDGMENTS

This study was supported by the National Natural Science Foundation of China (No. 82170826).

All data generated or analyzed during this study are included in this article and/or its supplementary material files. Further enquiries can be directed to the corresponding author.

All raw data generated and analyzed in this study have been deposited in Figshare and are publicly available under the DOI 10.6084/m9.figshare.30547946.

AUTHOR CONTRIBUTIONS

Conception or design: H.W., L.L. Acquisition, analysis, or interpretation of data: Z.Z. (Zhen Zhen), Z.Z. (Zheqi Zhang), Y.S., B.L. Drafting the work or revising: H.W., L.L., Z.Z. (Zhen Zhen), Z.Z. (Zheqi Zhang), Y.S., B.L. Final approval of the manuscript: H.W., L.L., Z.Z. (Zhen Zhen), Z.Z. (Zheqi Zhang), Y.S., B.L.

Fig. 1

Single-cell transcriptomic analysis reveals immune and vascular cell heterogeneity in the aortas of mice with diabetes mellitus (DM)-coronary artery disease (CAD). (A) Overview of the single-cell RNA sequencing (scRNA-seq) dataset (GSE211216) used in this study, including aortic tissues from mice with control-CAD and DM-CAD. (B) Uniform manifold approximation and projection (UMAP)-based dimensionality reduction and clustering plot showing the distribution of distinct cell types in the control-CAD and DM-CAD groups. (C) stacked bar plot comparing the proportions of major cell populations between the two groups. (D) Pseudotime trajectory analysis. The left panel is colored by pseudotime progression, whereas the middle and right panels are colored by cell type, illustrating the distribution of individual cell populations along developmental trajectories (control-CAD, n=1; DM-CAD, n=1). t-SNE, t-distributed stochastic neighbor embedding; NAC, neuron-associated cell.

Fig. 2

Enhanced inflammatory communication networks between macrophages and endothelial cells revealed by CellChat analysis. (A) Predicted number of intercellular communications (left) and mean interaction strength (right) inferred by CellChat analysis. (B) Comparison of relative information flow (left) and signaling strength (right) across key signaling pathways between groups, with pathways including visfatin, tumor necrosis factor (TNF), macrophage migration inhibitory factor (MIF), transforming growth factor-β (TGFβ), cerebrospinal fluid, and insulin-like growth factor (IGF) exhibiting markedly increased activity in the diabetes mellitus (DM)-coronary artery disease (CAD) group. (C) Intercellular interaction network among major cell types, in which node size represents cell-type abundance and edge thickness indicates communication strength. (D–F) Cell–cell communication network structures of the visfatin (D), TNF (E), and MIF (F) signaling pathways in the two groups. NAC, neuron-associated cell; SPP1, secreted phosphoprotein 1; NT, neurotrophin; MK, midkine; CCL, C-C motif chemokine ligand; CXCL, CXC motif chemokine ligand; ANGPTL, angiopoietin-like protein; PROS, protein S; GAS, growth arrest-specific; PDGF, platelet-derived growth factor; PTN, pleiotrophin; BMP, bone morphogenetic protein; CSF, colony-stimulating factor; SEMA3, semaphorin 3.

Fig. 3

Transcriptomic analysis reveals molecular features of metabolic dysregulation and immune activation co-driving diabetes mellitus (DM)-coronary artery disease (CAD) pathogenesis. (A) Schematic overview of the animal model and high-throughput RNA-sequencing workflow. (B) Volcano plot of differentially expressed genes (DEGs) between the DM-CAD and control-CAD groups, identifying 174 up-regulated and 26 downregulated genes (criteria: P<0.05, |log₂FC| >1). (C) Gene Ontology (GO) enrichment analysis of DEGs. (D) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs. (E) Construction of the least absolute shrinkage and selection operator (LASSO) regression model to identify key feature genes based on minimal cross-validation error (left, lambda path plot; right, L1 norm plot), used for downstream risk model development. (F) Expression levels of four selected feature genes (Opa interacting protein 5, opposite strand 1 [Oip5os1], hexokinase 2 [Hk2], C-C motif chemokine ligand 5 [Ccl5], and YTH N6-methyladenosine RNA binding protein F1 [Ythdf1]) in the DM-CAD and control-CAD groups. (G) Variable importance ranking of the four genes using the random forest algorithm (control-CAD, n=3; DM-CAD, n=3). Apoe, apolipoprotein E; NAD, nicotinamide adenine dinucleotide; MHC, major histocompatibility complex; ATPase, adenosine triphosphatase; ADP, adenosine diphosphate; BP, biological process; CC, cellular component; MF, molecular function; HIF-1, hypoxia-inducible factor 1; IL, interleukin.

Fig. 4

YTH N6-methyladenosine RNA binding protein F1 (YTHDF1) increases the binding affinity and RNA stability of Opa interacting protein 5 antisense RNA 1 (OIP5-AS1) through N⁶-methyladenosine (m⁶A) recognition. (A) Schematic overview of the proposed mechanism by which YTHDF1 regulates OIP5-AS1 stability. (B) Predicted m⁶A modification sites within the OIP5-AS1 sequence identified using the sequence-based RNA adenosine methylation site predictor (SRAMP) prediction tool. (C) RNA immunoprecipitation followed by quantitative polymerase chain reaction (RIP-qPCR) analysis of YTHDF1 binding to long non-coding RNA (lncRNA) OIP5-AS1. (D) Diagram of YTHDF1 mutant construction, highlighting key mutations at the m⁶A recognition residues K395A and Y397A. (E) Western blot (WB) analysis confirming expression of wild-type and mutant YTHDF1 proteins using an anti-Flag antibody. (F) RIP-qPCR analysis demonstrating reduced binding capacity of mutant YTHDF1 to OIP5-AS1. (G) m⁶A-RIP-qPCR quantification of OIP5-AS1 methylation levels under different YTHDF1 expression conditions. (H) WB validation of YTHDF1 knockdown efficiency following shRNA transfection. (I, J) Reverse transcription quantitative polymerase chain reaction (RT-qPCR) analysis of OIP5-AS1 expression at multiple time points following actinomycin D (ActD) treatment to assess RNA stability. (K) Proposed mechanistic model illustrating how YTHDF1 enhances OIP5-AS1 stability via m⁶A recognition. (L) Dot blot analysis demonstrating the specificity of the m⁶A antibody for methylated RNA. All data are presented as mean±standard deviation (SD). Experiments were independently repeated three times. Statistical analysis was performed using analysis of variance followed by the Bonferroni post hoc test. IgG, immunoglobulin G; WT, wild-type; MUT, mutant; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; NC, negative control; ns, not significant. ^aP<0.05; ^bP<0.01; ^cP<0.001; ^dP<0.0001; ^eP<0.01.

Fig. 5

Opa interacting protein 5 antisense RNA 1 (OIP5-AS1) overexpression remodels glycolytic metabolic features in macrophages. (A) Schematic of the metabolomics workflow used in this study, illustrating liquid chromatography–tandem mass spectrometry (LC–MS/MS)–based analysis of metabolites extracted from macrophages overexpressing OIP5-AS1 (oe-OIP5-AS1) and vector control cells. (B) Principal component analysis (PCA) demonstrating clear metabolic separation between oe-OIP5-AS1 (red) and Vector (blue) groups, with PC1 and PC2 cumulatively explaining 47.5% of the variance. (C) Orthogonal partial least squares discriminant analysis (OPLS-DA) plot confirming strong discriminative separation between the two groups. (D) Volcano plot showing the distribution of differential metabolites. (E, F) Functional enrichment analysis of differential metabolites using MetaboAnalyst, with pathway results derived from the Kyoto Encyclopedia of Genes and Genomes (KEGG) (E) and Small Molecule Pathway Database (F) databases. oe-OIP5-AS1, n=6; Vector: n=6. GC, gas chromatography; LC, liquid chromatography; EC, electron capture; TOF, time-of-flight; Q, quadrupole; IT, ion trap; TCA, tricarboxylic acid.

Fig. 6

Opa interacting protein 5 antisense RNA 1 (OIP5-AS1) promotes foam cell formation and increases pro-inflammatory cytokine expression, contributing to plaque destabilization. (A) Schematic illustration of the experimental workflow. (B) Oil Red O (ORO) staining showing intracellular lipid accumulation in macrophages under different treatment conditions (scale bar=100 μm). (C, D) Quantification of total cholesterol (TC) and cholesteryl ester (CE) levels across experimental groups; (E–H) Reverse transcription quantitative polymerase chain reaction (RT-qPCR) analysis of cholesterol uptake-related genes (CD36, class A1 scavenger receptor [SR-A1]) and cholesterol effluxrelated genes (ATP-binding cassette transporter A1 [ABCA1], ATP-binding cassette subfamily G member 1 [ABCG1]). (I–K) qPCR assessment of inflammatory cytokine expression (interleukin 1β [IL-1β], tumor necrosis factor-α [TNF-α], and IL-6) in macrophages to evaluate the impact of OIP5-AS1 on local plaque inflammation. (L) Schematic diagram summarizing the mechanism by which OIP5-AS1 enhances foam cell formation through coordinated modulation of cholesterol uptake, efflux, and glycolytic metabolism. All data are presented as mean±standard deviation from three independent experiments. Statistical significance was determined using one-way analysis of variance followed by the Bonferroni post hoc test. THP-1, human monocytic leukemia cell line; ox-LDL, oxidized low-density lipoprotein. ^aP<0.05; ^bP<0.01; ^cP<0.001; ^dP<0.0001.

Fig. 7

Glycolysis inhibitor 2-deoxyglucose (2-DG) reverses Opa interacting protein 5 antisense RNA 1 (OIP5-AS1)-induced foam cell formation and pro-inflammatory cytokine expression. (A) Schematic diagram illustrating the experimental workflow for oxidized low-density lipoprotein (ox-LDL) and 2-DG treatment in OIP5-AS1-overexpressing macrophages. (B) Oil Red O (ORO) staining assessing intracellular lipid accumulation. (C) Colorimetric quantification of total cholesterol (TC) and cholesteryl ester (CE) levels in macrophages. (D) Reverse transcription quantitative polymerase chain reaction (RT-qPCR) analysis of cholesterol uptake genes (CD36, class A1 scavenger receptor [SR-A1]) and efflux genes (ATP-binding cassette transporter A1 [ABCA1], ATP-binding cassette subfamily G member 1 [ABCG1]). (E) qPCR quantification of mRNA expression levels of inflammatory cytokines interleukin 1β (IL-1β), tumor necrosis factor-α (TNF-α), and IL-6. (F) Measurement of lactate concentration in culture supernatants as an indicator of glycolytic activity. (G, H) Fluorescence staining with 2′,7′-dichlorofluorescin diacetate (DCFDA) and MitoSOX (Invitrogen) to detect total and mitochondrial reactive oxygen species (ROS) levels in macrophages. (I) 5,5′,6,6′-Tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide (JC-1) fluorescence microscopy to assess mitochondrial membrane potential (ΔΨm). (J, K) Quantitative analysis of the red-to-green fluorescence intensity ratio from JC-1 staining to evaluate mitochondrial function. All data are presented as mean±standard deviation from three independent experiments. Statistical significance was determined using analysis of variance followed by the Bonferroni post hoc test. THP-1, human monocytic leukemia cell line; ox-LDL, oxidized low-density lipoprotein; DCFH-DA, 2′,7′-dichlorodihydrofluorescein diacetate. ^aP<0.01; ^bP< 0.001; ^cP<0.0001.

Fig. 8

Integrated multi-omics strategy reveals YTH N6-methyladenosine RNA binding protein F1 (YTHDF1)-N⁶-methyladenosine (m⁶A)-Opa interacting protein 5 antisense RNA 1 (OIP5-AS1) axis in macrophage-driven inflammation in diabetes mellitus (DM)-coronary artery disease (CAD). t-SNE, t-distributed stochastic neighbor embedding; MSC, mesenchymal stem cell; HSPC, hematopoietic stem and progenitor cell; NK, natural killer; pDC, plasmacytoid dendritic cell; FC, fold change; DEG, differentially expressed gene; LC–MS, liquid chromatography–mass spectrometry; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Fig. 9

YTH N6-methyladenosine RNA binding protein F1 (YTHDF1)-mediated N⁶-methyladenosine (m⁶A) stabilization of long non-coding RNA (lncRNA) Opa interacting protein 5 antisense RNA 1 (OIP5-AS1) promotes glycolytic reprogramming and inflammation in macrophages in diabetes mellitus (DM)-coronary artery disease (CAD). ROS, reactive oxygen species; IL, interleukin; TNF-α, tumor necrosis factor-α; ox-LDL, oxidized low-density lipoprotein; SR-A1, class A1 scavenger receptor; ABCA1, ATP-binding cassette transporter A1; ABCG1, ATP-binding cassette subfamily G member 1.

References

• 1. Varughese GI, Tomson J, Lip GY. Type 2 diabetes mellitus: a cardiovascular perspective. Int J Clin Pract 2005;59:798–816.ArticlePubMed
• 2. Nicholls SJ, Tuzcu EM, Kalidindi S, Wolski K, Moon KW, Sipahi I, et al. Effect of diabetes on progression of coronary atherosclerosis and arterial remodeling: a pooled analysis of 5 intravascular ultrasound trials. J Am Coll Cardiol 2008;52:255–62.PubMed
• 3. Kornowski R, Mintz GS, Kent KM, Pichard AD, Satler LF, Bucher TA, et al. Increased restenosis in diabetes mellitus after coronary interventions is due to exaggerated intimal hyperplasia: a serial intravascular ultrasound study. Circulation 1997;95:1366–9.ArticlePubMed
• 4. Burgmaier M, Hellmich M, Marx N, Reith S. A score to quantify coronary plaque vulnerability in high-risk patients with type 2 diabetes: an optical coherence tomography study. Cardiovasc Diabetol 2014;13:117.ArticlePubMedPMCPDF
• 5. Johny E, Bhaskar P, Alam MJ, Kuladhipati I, Das R, Adela R. Platelet mediated inflammation in coronary artery disease with type 2 diabetes patients. J Inflamm Res 2021;14:5131–47.ArticlePubMedPMCPDF
• 6. Wu W, Jia C, Xu X, He Y, Xie Y, Zhou Y, et al. Impact of platelet-to-HDL-cholesterol ratio on long-term mortality in coronary artery disease patients with or without type 2 diabetes: insights from a Chinese multicenter cohort. J Inflamm Res 2024;17:2731–44.ArticlePubMedPMCPDF
• 7. Xu X, Xie Y, Gu X, Zhou Y, Kang Y, Liu J, et al. Association between systemic immune inflammation level and poor prognosis across different glucose metabolism status in coronary artery disease patients. J Inflamm Res 2023;16:4031–42.ArticlePubMedPMCPDF
• 8. Bian X, He J, Zhang R, Yuan S, Dou K. The combined effect of systemic immune-inflammation index and type 2 diabetes mellitus on the prognosis of patients undergoing percutaneous coronary intervention: a large-scale cohort study. J Inflamm Res 2023;16:6415–29.ArticlePubMedPMCPDF
• 9. Laera N, Malerba P, Vacanti G, Nardin S, Pagnesi M, Nardin M. Impact of immunity on coronary artery disease: an updated pathogenic interplay and potential therapeutic strategies. Life (Basel); 2023. 13:p. 2128.ArticlePubMedPMC
• 10. Hansson GK. Inflammation, atherosclerosis, and coronary artery disease. N Engl J Med 2005;352:1685–95.ArticlePubMed
• 11. Xiao Q, Hou R, Xie L, Niu M, Pan X, Zhu X. Macrophage metabolic reprogramming and atherosclerotic plaque microenvironment: fostering each other? Clin Transl Med 2023;13:e1257.ArticlePubMedPMC
• 12. O’Rourke SA, Neto NG, Devilly E, Shanley LC, Fitzgerald HK, Monaghan MG, et al. Cholesterol crystals drive metabolic reprogramming and M1 macrophage polarisation in primary human macrophages. Atherosclerosis 2022;352:35–45.ArticlePubMed
• 13. Certo M, Rahimzadeh M, Mauro C. Immunometabolism in atherosclerosis: a new understanding of an old disease. Trends Biochem Sci 2024;49:791–803.ArticlePubMed
• 14. Xue S, Su Z, Liu D. Immunometabolism and immune response regulate macrophage function in atherosclerosis. Ageing Res Rev 2023;90:101993.ArticlePubMed
• 15. Tacconi S, Vari F, Sbarigia C, Vardanyan D, Longo S, Mura F, et al. M1-derived extracellular vesicles polarize recipient macrophages into M2-like macrophages and alter skeletal muscle homeostasis in a hyper-glucose environment. Cell Commun Signal 2024;22:193.ArticlePubMedPMCPDF
• 16. Rendra E, Riabov V, Mossel DM, Sevastyanova T, Harmsen MC, Kzhyshkowska J. Reactive oxygen species (ROS) in macrophage activation and function in diabetes. Immunobiology 2019;224:242–53.ArticlePubMed
• 17. Zhao N, Yu X, Zhu X, Song Y, Gao F, Yu B, et al. Diabetes mellitus to accelerated atherosclerosis: shared cellular and molecular mechanisms in glucose and lipid metabolism. J Cardiovasc Transl Res 2024;17:133–52.ArticlePubMedPDF
• 18. Lee YG, Lee J, Byeon SE, Yoo DS, Kim MH, Lee SY, et al. Functional role of Akt in macrophage-mediated innate immunity. Front Biosci (Landmark Ed) 2011;16:517–30.ArticlePubMed
• 19. Zhang QY, Zhang HY, Feng SG, Yao MD, Ding JJ, Li XM, et al. Macrophage metabolic reprogramming ameliorates diabetes-induced microvascular dysfunction. Redox Biol 2025;79:103449.ArticlePubMed
• 20. Russo S, Kwiatkowski M, Govorukhina N, Bischoff R, Melgert BN. Meta-inflammation and metabolic reprogramming of macrophages in diabetes and obesity: the importance of metabolites. Front Immunol 2021;12:746151.ArticlePubMedPMC
• 21. Wang T, Dong Y, Yao L, Lu F, Wen C, Wan Z, et al. Adoptive transfer of metabolically reprogrammed macrophages for atherosclerosis treatment in diabetic ApoE −/− mice. Bioact Mater 2022;16:82–94.ArticlePubMedPMC
• 22. Rong B, Jiang H, Zhu W, Yang G, Zhou X, Lyu Z, et al. Unraveling the role of macrophages in diabetes: impaired phagocytic function and therapeutic prospects. Medicine (Baltimore); 2025. 104:p. e41613.ArticlePubMedPMC
• 23. Wang Y, Xu L, Luo S, Sun X, Li J, Pang H, et al. The m6A methylation profiles of immune cells in type 1 diabetes mellitus. Front Immunol 2022;13:1030728.ArticlePubMedPMC
• 24. Vallecillo-Viejo I, Kubala S. Redox regulation of m6A methyltransferase METTL3 in β-cells controls the innate immune response in type 1 diabetes. Pediatrics 2024;154:S57–8.ArticlePubMedPDF
• 25. Ren Y, Li Z, Li J, Liang R, Wang Z, Bai Y, et al. M6 A mRNA methylation: biological features, mechanisms, and therapeutic potentials in type 2 diabetes mellitus. Obes Rev 2023;24:e13639.ArticlePubMed
• 26. Chen C, Xiang Q, Liu W, Liang S, Yang M, Tao J. Co-expression network revealed roles of RNA m6A methylation in human β-cell of type 2 diabetes mellitus. Front Cell Dev Biol 2021;9:651142.ArticlePubMedPMC
• 27. Hao WY, Lou Y, Hu GY, Qian CY, Liang WR, Zhao J, et al. RNA m6A reader YTHDF1 facilitates inflammation via enhancing NLRP3 translation. Biochem Biophys Res Commun 2022;616:76–81.ArticlePubMed
• 28. Li Z, Teng M, Jiang Y, Zhang L, Luo X, Liao Y, et al. YTHDF1 negatively regulates Treponema pallidum-induced inflammation in THP-1 macrophages by promoting SOCS3 translation in an m6A-dependent manner. Front Immunol 2022;13:857727.ArticlePubMedPMC
• 29. Wang Y, Liu L, Li J. LncRNA KCNQ1OT1 depletion inhibits the malignant development of atherosclerosis by miR-145-5p. Microvasc Res 2022;139:104236.ArticlePubMed
• 30. Ma X, Liu H, Chen F. Functioning of long noncoding RNAs expressed in macrophage in the development of atherosclerosis. Front Pharmacol 2020;11:567582.ArticlePubMedPMC
• 31. Shan L, Guo M, Dai Y, Wei L, Zhang W, Gao J. Comprehensive analysis of m6A modification lncRNAs in high glucose and TNF-α induced human umbilical vein endothelial cells. Medicine (Baltimore); 2023. 102:p. e33133.ArticlePubMedPMC
• 32. Fu S, Zhou Q, Peng X, Hu Y, Xiong J, Liu F. METTL3/ YTHDC1 mediates up-regulation of lncRNA OGRU in an m6A-dependent manner involving in oxidative stress and inflammation of HG-induced Müller cells. Immunol Lett 2025;272:106972.ArticlePubMed
• 33. Meng L, Lin H, Huang X, Weng J, Peng F, Wu S. METTL14 suppresses pyroptosis and diabetic cardiomyopathy by downregulating TINCR lncRNA. Cell Death Dis 2022;13:38.ArticlePubMedPMCPDF
• 34. Xie W, Wu D, Ren Y, Jiang Y, Zhang H, Yang S, et al. OIP5-AS1 attenuates microangiopathy in diabetic mouse by regulating miR-200b/ACE2. World Neurosurg 2020;139:e52–60.ArticlePubMed
• 35. Fu JX, Sun GQ, Wang HL, Jiang HX. LncRNA OIP5-AS1 induces epithelial-to-mesenchymal transition and renal fibrosis in diabetic nephropathy via binding to miR-30c-5p. J Biol Regul Homeost Agents 2020;34:961–8.ArticlePubMed
• 36. Wang Y, Zhuang H, Jiang XH, Zou RH, Wang HY, Fan ZN. Unveiling the key genes, environmental toxins, and drug exposures in modulating the severity of ulcerative colitis: a comprehensive analysis. Front Immunol 2023;14:1162458.ArticlePubMedPMC
• 37. Hu HJ, Qiu J, Zhang C, Tang ZH, Qu SL, Jiang ZS. Hydrogen sulfide improves ox LDL induced expression levels of Lp PLA2 in THP 1 monocytes via the p38MAPK pathway. Mol Med Rep 2021;23:358.ArticlePubMedPMC
• 38. Schwarz N, Fernando S, Chen YC, Salagaras T, Rao SR, Liyanage S, et al. Colchicine exerts anti-atherosclerotic and -plaque-stabilizing effects targeting foam cell formation. FASEB J 2023;37:e22846.ArticlePubMedPDF
• 39. McCready S. A dot blot immunoassay for UV photoproducts. Methods Mol Biol 1999;113:147–56.ArticlePubMed
• 40. Zhang JK, Li Y, Yu ZT, Jiang JW, Tang H, Tu GL, et al. OIP5-AS1 inhibits oxidative stress and inflammation in ischemic stroke through miR-155-5p/IRF2BP2 axis. Neurochem Res 2023;48:1382–94.ArticlePubMedPDF
• 41. Li X, Cao Q, Wang Y, Wang Y. Retracted article: LncRNA OIP5-AS1 contributes to ox-LDL-induced inflammation and oxidative stress through regulating the miR-128-3p/CDKN2A axis in macrophages. RSC Adv 2019;9:41709–19.ArticlePubMedPMC
• 42. Cao F, Wang S, Tan Q, Hou J, Li Y, Ma W, et al. YTHDF1-mediated m6A modification of GBP4 promotes M1 macrophage polarization in acute lung injury. Respir Res 2025;26:11.ArticlePubMedPMCPDF
• 43. Li M, Zhong X, Zhao Z, Zeng Z, Yuan Q, Xiao X, et al. The expression of m6A enzymes in the hippocampus of diabetic cognitive impairment mice and the possible improvement of YTHDF1. Brain Res 2022;1777:147766.ArticlePubMed
• 44. Li HM, Tang F, Wang LJ, Huang Q, Pan HF, Zhang TP. Association of N6-methyladenosine readers’ genes variation and expression level with pulmonary tuberculosis. Front Public Health 2022;10:925303.ArticlePubMedPMC
• 45. Zhang Y, Wang R, Tan H, Wu K, Hu Y, Diao H, et al. Fufang Zhenzhu Tiaozhi (FTZ) capsule ameliorates diabetes-accelerated atherosclerosis via suppressing YTHDF2-mediated m6A modification of SIRT3 mRNA. J Ethnopharmacol 2023;317:116766.ArticlePubMed
• 46. Zhuang A, Tan Y, Liu Y, Yang C, Kiriazis H, Grigolon K, et al. Deletion of the muscle enriched lncRNA Oip5os1 induces atrial dysfunction in male mice with diabetes. Physiol Rep 2023;11:e15869.PubMedPMC
• 47. Niu X, Zhang J, Hu S, Dang W, Wang K, Bai M. LncRNA Oip5-as1 inhibits excessive mitochondrial fission in myocardial ischemia/reperfusion injury by modulating DRP1 phosphorylation. Cell Mol Biol Lett 2024;29:72.ArticlePubMedPMCPDF
• 48. Yue Q, Liu Y, Ji J, Hu T, Lin T, Yu S, et al. Down-regulation of OIP5-AS1 inhibits obesity-induced myocardial pyroptosis and miR-22/NLRP3 inflammasome axis. Immun Inflamm Dis 2023;11:e1066.ArticlePubMedPMC
• 49. Zhang H, Gu Y, Gang Q, Huang J, Xiao Q, Ha X. N6-methyladenosine RNA modification: an emerging molecule in type 2 diabetes metabolism. Front Endocrinol (Lausanne); 2023. 14:p. 1166756.ArticlePubMedPMC
• 50. Sun YH, Zhao TJ, Li LH, Wang Z, Li HB. Emerging role of N6-methyladenosine in the homeostasis of glucose metabolism. Am J Physiol Endocrinol Metab 2024;326:E1–13.ArticlePubMed
• 51. Hu T, Liu CH, Lei M, Zeng Q, Li L, Tang H, et al. Metabolic regulation of the immune system in health and diseases: mechanisms and interventions. Signal Transduct Target Ther 2024;9:268.ArticlePubMedPMCPDF
• 52. Wang J, Wu Q, Wang X, Liu H, Chen M, Xu L, et al. Targeting macrophage phenotypes and metabolism as novel therapeutic approaches in atherosclerosis and related cardiovascular diseases. Curr Atheroscler Rep 2024;26:573–88.ArticlePubMedPMCPDF
• 53. Gindri Dos Santos B, Goedeke L. Macrophage immunometabolism in diabetes-associated atherosclerosis. Immunometabolism (Cobham) 2023;5:e00032.ArticlePubMedPMC
• 54. Matsuura Y, Shimizu-Albergine M, Barnhart S, Kramer F, Hsu CC, Kothari V, et al. Diabetes suppresses glucose uptake and glycolysis in macrophages. Circ Res 2022;130:779–81.ArticlePubMedPMC
• 55. Patterson MT, Xu Y, Hillman H, Osinski V, Schrank PR, Kennedy AE, et al. Trem2 agonist reprograms foamy macrophages to promote atherosclerotic plaque stability: brief report. Arterioscler Thromb Vasc Biol 2024;44:1646–57.ArticlePubMedPMC
• 56. Xu W, Wei Z, Dong J, Duan F, Chen K, Chen C, et al. Global metabolomics reveals the metabolic dysfunction in Ox-LDL induced macrophage-derived foam cells. Front Pharmacol 2017;8:586.ArticlePubMedPMC
• 57. Shirai T, Nazarewicz RR, Wallis BB, Yanes RE, Watanabe R, Hilhorst M, et al. The glycolytic enzyme PKM2 bridges metabolic and inflammatory dysfunction in coronary artery disease. J Exp Med 2016;213:337–54.ArticlePubMedPMCPDF
• 58. Penny HL, Sieow JL, Gun SY, Lau MC, Lee B, Tan J, et al. Targeting glycolysis in macrophages confers protection against pancreatic ductal adenocarcinoma. Int J Mol Sci 2021;22:6350.ArticlePubMedPMC
• 59. Xu B, Liu Y, Li N, Geng Q. Lactate and lactylation in macrophage metabolic reprogramming: current progress and outstanding issues. Front Immunol 2024;15:1395786.ArticlePubMedPMC
• 60. Wang J, Han Y, Huang F, Tang L, Mu J, Liang Y. Diabetic macrophage small extracellular vesicles-associated miR-503/IGF1R axis regulates endothelial cell function and affects wound healing. Front Immunol 2023;14:1104890.ArticlePubMedPMC
• 61. Malhi NK, Luo Y, Tang X, Chadha RS, Tapia A, Yuan D, et al. Mapping endothelial-macrophage interactions in diabetic vasculature: role of TREM2 in vascular inflammation and ischemic response. bioRxiv [Preprint] 2024 May 14 https://doi.org/10.1101/2024.05.14.594235.Article
• 62. Basher MA, Hallinan AR, Lee CK. Heterogeneity-preserving discriminative feature selection for disease-specific subtype discovery. Nat Commun 2025;16:3593.ArticlePubMedPMC
• 63. Chellappan D, Rajaguru H. Machine learning meets metaheuristics: bald eagle search optimization and red deer optimization for feature selection in type II diabetes diagnosis. Bioengineering (Basel); 2024. 11:p. 766.ArticlePubMedPMC
• 64. Jazieh C, Arabi TZ, Asim Z, Sabbah BN, Alsaud AW, Alkattan K, et al. Unraveling the epigenetic fabric of type 2 diabetes mellitus: pathogenic mechanisms and therapeutic implications. Front Endocrinol (Lausanne); 2024. 15:p. 1295967.ArticlePubMedPMC
• 65. Gao R, Liu M, Yang H, Shen Y, Xia N. Epigenetic regulation in coronary artery disease: from mechanisms to emerging therapies. Front Mol Biosci 2025;12:1548355.ArticlePubMedPMC
• 66. Liu J, Li K, Cai J, Zhang M, Zhang X, Xiong X, et al. Landscape and regulation of m6A and m6Am methylome across human and mouse tissues. Mol Cell 2020;77:426–40.ArticlePubMed

Figure & Data

References

Citations

Citations to this article as recorded by  

• Molecular basis of coronary artery disease–malignancy comorbidity: inflammation, immunometabolism, thrombosis, and cardio-oncology translation
  Junlin Li, Yan Zhao, Ming Bai
  Frontiers in Immunology.2026;[Epub]     CrossRef