Abstract

Purpose

   Diabetic retinopathy (DR) is a significant complication of diabetes,
   with complex pathogenesis involving epigenetic modifications. This
   study aimed to explore the epigenetic regulatory mechanisms
   contributing to DR.

Methods

   DR-related data, including DNA methylation, mRNA, and miRNA expression
   datasets, were obtained from the Gene Expression Omnibus database.
   Differential gene expression analysis was performed to identify
   differentially methylated genes and expressed mRNAs and miRNAs.
   Cross-analysis established the methylation-expression and miRNA-mRNA
   regulatory networks. A comprehensive DR-related epigenetic regulatory
   network was constructed, identifying hub genes. The expression
   characteristics of these hub genes in various immune cells were
   examined using single-cell RNA sequencing.

Results

   We identified 10,716 differentially methylated genes, 1,181
   differentially expressed mRNAs, and 615 differentially expressed miRNAs
   in DR. The methylation-expression regulatory network was associated
   with pathways such as TGF-beta and ErbB signaling. The miRNA regulatory
   network was linked to pathways related to cellular senescence,
   adherents junctions, and endocytosis. Five hub genes were identified:
   TFRC, AP2M1, AP2A1, DAB2, and PPP1CB. Single-cell RNA sequencing
   revealed specific expression of these genes in particular immune cells,
   highlighting their potential roles in DR pathogenesis.

Conclusion

   This study constructed a comprehensive epigenetic regulatory network
   for DR and identified key regulatory genes, offering new insights into
   the molecular mechanisms underlying DR and potential therapeutic
   targets for diagnosis and treatment.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s13098-025-01723-7.

   Keywords: Diabetic retinopathy, DNA methylation, miRNA, Epigenetic
   regulation, Single-cell RNA sequencing

Introduction

   Diabetic retinopathy (DR) is one of the most common microvascular
   complications of diabetes, and its prevalence continues to rise in
   parallel with the increasing incidence of diabetes [[26]1]. The
   pathogenesis of DR is complex and involves multiple biological
   processes. Hyperglycemia induced by diabetes triggers a series of
   microvascular abnormalities [[27]2] and neurodegenerative changes
   [[28]3], along with chronic low-grade inflammation [[29]4], which
   further exacerbates retinal vascular damage. Current treatments for DR,
   such as panretinal photocoagulation (PRP) and anti-vascular endothelial
   growth factor (anti-VEGF) therapies, can only slow down disease
   progression but do not cure it completely [[30]5]. Therefore,
   elucidating the pathogenesis of DR is critical for advancing
   therapeutic development.

   In recent years, a growing body of evidence suggests that epigenetic
   modifications play an important role in the onset and progression of DR
   [[31]6]. Epigenetics, which refers to the regulation of gene function
   and expression without altering the DNA sequence, involves mechanisms
   such as DNA methylation, histone modification, and non-coding RNA
   (e.g., miRNA and lncRNA) regulation, thereby having a significant
   impact on the diseases [[32]7, [33]8]. DNA methylation, an essential
   epigenetic modification, typically occurs in CpG islands within gene
   promoters and regulates gene expression by adding a methyl group to the
   fifth carbon of cytosine residues. This modification is pivotal in gene
   imprinting, X-chromosome inactivation, genome stability, and chromatin
   structure [[34]9]. miRNAs, a class of small non-coding RNAs, regulate
   gene expression at the post-transcriptional level by binding to target
   mRNAs, inhibiting their translation or promoting their degradation.
   miRNAs are widely involved in diverse physiological processes such as
   cell cycle, cell growth, metabolism, and synaptic plasticity [[35]10].

   Epigenetic modifications, particularly DNA methylation and miRNA
   regulation, are closely linked to gene expression in DR. In a
   hyperglycemic environment, alterations in DNA methylation can affect
   the expression of retina-related genes. For instance,
   hyperglycemia-induced DNA methylation can enhance the transcription and
   expression of the matrix metalloproteinase-9 (MMP-9) gene, leading to
   mitochondrial damage and accelerated apoptosis of retinal vascular
   endothelial cells [[36]11]. Additionally, miRNAs contribute to DR
   development by inhibiting the expression of target genes. Abnormal
   expressions of miRNAs, such as miR-21 and miR-93, are closely
   associated with DR progression, modulating key physiological parameters
   such as blood glucose levels and insulin resistance [[37]12, [38]13].
   Targeting these epigenetic modifications may offer effective
   interventions to slow or even reverse DR progression.

   Despite extensive research on various epigenetic modifications in DR, a
   comprehensive characterization of the underlying epigenetic mechanisms
   remains lacking. This study addresses this knowledge gap by
   investigating the integrated epigenetic regulatory landscape of DR
   through comparative analysis of DNA methylation patterns, mRNA
   expression profiles, and miRNA signatures in clinical samples from DR
   patients. We constructed epigenetic regulatory networks to identify hub
   regulatory genes and performed single-cell RNA sequencing to further
   explore the potential functions and mechanisms of these hub genes. The
   results of this study are expected to provide new molecular targets for
   the early diagnosis and treatment of DR.

Materials and methods

Data collection

   To investigate the epigenetic regulatory mechanisms in DR, the
   following data were integrated from several publicly available
   datasets. These included [39]GSE121820, comprising DNA methylation data
   from 10 DR patients and 10 healthy controls; [40]GSE189005, containing
   gene expression data from 10 DR cases and 9 controls; [41]GSE189002,
   which encompassed miRNA expression profiles from 10 DR cases and 9
   controls; and [42]GSE204880, providing single cell sequencing data from
   DR patients and normal controls.

Data processing

   Differentially methylated genes (DMGs) were identified from the
   [43]GSE121820 dataset with a threshold of P-value < 1e-3. Differential
   expression analysis for mRNA and miRNA datasets was performed using the
   limma package in R software, with filtering criteria set at |log2 fold
   change| > 0.7 and P-value < 0.05.

Epigenetic network construction of methylation and miRNA

   Cross-analysis between DMGs and differentially expressed genes (DEGs)
   was conducted to identify methylation-associated DEGs. Potential target
   genes of differentially expressed miRNAs were predicted using the
   Starbase database [[44]14] ([45]http://starbase.sysu.edu.cn/), and
   cross-analysis was performed between the predicted target genes and
   differentially expressed mRNAs to explore miRNA regulatory mechanisms
   [[46]15].

Comprehensive epigenetic network construction and hub gene screening

   Cross-analysis of DNA methylation, mRNA, and miRNA networks was
   conducted to identify potential regulatory genes. Protein-protein
   interaction (PPI) network among these genes was mapped using STRING
   [[47]16], and hub genes were identified via the MCODE clustering
   algorithm implemented in Cytoscapes (3.9.1) [[48]17].

Functional enrichment analysis

   The R package clusterProfiler(4.8.3) was used for Gene Ontology (GO)
   and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional
   enrichment analyses of DEGs [[49]18], with significance defined as
   adjusted P-value < 0.05 corrected by the Benjamini-Hochberg method
   [[50]19].

Immunoinfiltration analysis

   Immune cell infiltration levels were assessed using single-sample gene
   set enrichment analysis (ssGSEA) implemented in the R package GSVA
   (1.48.3) [[51]20], leveraging immune-related gene sets including 782
   genes and 28 cell types [[52]21].

Single-cell analysis

   Single-cell RNA sequencing data were processed using the Seurat package
   (5.2.1) [[53]22]. Initially, quality control was applied to filter out
   low-quality cells, with removal of cells showing > 20% ribosomal gene
   expression and > 3% erythrocyte gene expression. Subsequently, the
   NormalizeData function was used to normalize the filtered data, the
   FindVariableFeatures function was employed to identify highly variable
   genes (HVGs), and the ScaleData function was utilized to normalize the
   HVGs. Principal component analysis (PCA) was conducted based on HVGs,
   and cell clustering was performed using the FindNeighbors and
   FindClusters functions. Finally, cell clusters were annotated according
   to typical marker genes provided by the reference gene set, followed by
   integration of kidney and retinal data using Harmony.

Statistical analysis

   Data analysis and visualization were conducted using R software version
   4.1.2. Differential expression analysis was performed using the limma
   package, with significance thresholds set at |logFC| > 0.7 and
   P-value < 0.05.

Results

Screening of differentially methylated genes

   Analysis of the [54]GSE121820 dataset identified 10,716 DMGs by
   comparing DNA methylation levels between DR patients and healthy
   controls with P-value less than 1e-3 (Fig. [55]1a and Supplementary
   Table 1). To further investigate the functions of these filtered genes
   and their involvement in signaling pathways, GO and KEGG functional
   enrichment analyses were performed. GO enrichment analysis revealed
   that these DMGs were significantly enriched in biological processes
   such as regulation of GTPase activity, positive regulation of GTPase
   activity, regulation of small GTPase mediated signal transduction,
   activation of GTPase activity, and peptidyl-lysine modification
   (Fig. [56]1b). KEGG analysis indicated significant enrichment in
   pathways including hippo signaling pathway, signaling pathways
   regulating pluripotency of stem cells, hepatocellular carcinoma,
   adherens junction, and cell cycle (Fig. [57]1c).

Fig. 1.

   [58]Fig. 1
   [59]Open in a new tab

   Differentially methylated genes. (a) Number of DMGs identified based on
   different p values. (b) GO functional enrichment analysis. (c) KEGG
   enrichment analysis

Screening of differentially expressed mRNA

   Based on screening criteria of |log2FC| > 0.7 and P < 0.05, 1,181
   differentially expressed mRNAs were identified from the [60]GSE189005
   dataset, comprising 885 upregulated genes and 296 downregulated genes
   (Fig. [61]2a and Supplementary Table 2). The top five significantly
   upregulated mRNAs were RNF182, SLC14A1, BBOF1, SELENBP1, and SLC4A1,
   while the top five significantly downregulated mRNAs were SNORA11,
   SCARNA5, SNORA38B, RNU12, and SNORA80E. A heatmap of these DEGs clearly
   distinguished patients with DR from healthy individuals (Fig. [62]2b).
   GO analysis indicated that the differentially expressed mRNAs were
   primarily enriched in biological processes related to myeloid cell
   development, erythrocyte differentiation, myeloid cell differentiation,
   mitotic cell cycle phase transition, and erythrocyte homeostasis
   (Fig. [63]2c). KEGG analysis showed significant differences in 34
   pathways (P < 0.05), including ribosome, ubiquitin mediated
   proteolysis, IL-17 signaling pathway, and (Fig. [64]2d).

Fig. 2.

   [65]Fig. 2
   [66]Open in a new tab

   Differentially expressed mRNA. (a) A Volcano plot visualizing
   differential mRNA in DR patients compared to healthy controls. (b)
   Heatmap of the differential mRNA distinguishing DR patients from
   healthy controls. (c) Histogram showing the top 15 GO terms ranked by
   the GO functional enrichment analysis. (d) Histogram showing the top 15
   significant KEGG pathways

Analysis of differential miRNA expression

   In the [67]GSE189002 dataset, 615 differentially expressed miRNAs were
   identified with the selection criteria (|log2FC| > 0.7 and P < 0.05),
   including 368 upregulated and 247 downregulated DEGs (Fig. [68]3a and
   Supplementary Table 3). The top five significantly upregulated miRNAs
   were hsa-miR-5189-3p, hsa-miR-1226-3p, hsa-miR-4685-3p, hsa-miR-584-3p,
   and hsa-miR-6747-3p, while hsa-miR-29a-3p, hsa-miR-143-3p,
   hsa-miR-27a-3p, hsa-miR-30e-3p, and hsa-miR-31-5p were the top five
   significantly downregulated miRNAs. A heatmap demonstrated the
   differential expression of miRNAs between DR patients and healthy
   individuals (Fig. [69]3b). GO analysis showed enrichment in biological
   processes such as regulation of protein catabolic process, cell growth,
   peptidyl-serine modification, protein localization to nucleus, and
   post-embryonic development (Fig. [70]3c). KEGG results implicated
   several disease-associated pathways, including protein processing in
   endoplasmic reticulum, adherens junction, proteoglycans in cancer,
   cellular senescence, and regulation of actin cytoskeleton
   (Fig. [71]3d).

Fig. 3.

   [72]Fig. 3
   [73]Open in a new tab

   Differential miRNA expression profile. (a) A Volcano plot visualizing
   differential miRNA expression between DR patients and healthy controls.
   (b) Heatmap of the differential miRNA expression between DR patients
   and healthy controls. (c) Histogram showing the top 15 GO terms ranked
   by the GO functional enrichment analysis. (d) Histogram showing the top
   15 significant KEGG pathways

Construction of differential mRNA methylation regulatory network

   A regulatory network for mRNA methylation was constructed to explore
   the relationship between DMGs and DEGs in DR. 537 differentially
   methylated mRNA were obtained through cross-analysis (Fig. [74]4a and
   Supplementary Table 4). GO analysis demonstrated that these genes were
   primarily involved in mitotic cell cycle phase transition,
   proteasome-mediated ubiquitin-dependent protein catabolic process,
   regulation of ubiquitin protein ligase activity, regulation of mitotic
   cell cycle phase transition, and regulation of proteolysis involved in
   protein catabolic process (Fig. [75]4b). The KEGG pathways mainly
   included ubiquitin mediated proteolysis, ribosome, TGF-beta signaling
   pathway, cell cycle, and ErbB signaling pathway (Fig. [76]4c).

Fig. 4.

   [77]Fig. 4
   [78]Open in a new tab

   The mRNA methylation regulatory network in DR. (a) Venn diagram of the
   differentially methylated mRNA. (b) GO functional enrichment analysis
   of the differentially methylated mRNA. (c) KEGG pathway enrichment
   analysis of the differentially methylated mRNA

Construction of miRNA-mRNA regulatory network

   The miRNA-mRNA regulatory network for DR was constructed by
   cross-analysis between differentially expressed miRNAs of potential
   target genes and differentially expressed mRNAs, resulting in 220
   miRNA-mRNA pairs (Fig. [79]5a and Supplementary Table 5). GO analysis
   revealed that 73 DEGs were significantly enriched in biological
   processes such as clathrin-dependent endocytosis, postsynaptic
   endocytosis, postsynaptic neurotransmitter receptor internalization,
   neurotransmitter receptor internalization, and carboxylic acid
   transmembrane transport (Fig. [80]5b). These genes were associated with
   124 signaling pathways, highlighting the involvement of oocyte meiosis,
   cellular senescence, adherens junction, endocrine and other
   factor-regulated calcium reabsorption, and endocytosis (Fig. [81]5c).

Fig. 5.

   [82]Fig. 5
   [83]Open in a new tab

   The miRNA regulatory network in DR. (a) A miRNA-mRNA regulatory
   network. (b) GO functional enrichment analysis of DEGs. (c) KEGG
   pathway enrichment analysis of DEGs

Construction of DR-related epigenetic regulatory network and identification
of hub genes

   Cross-analysis of the mRNA methylation network and miRNA-mRNA network
   identified 32 genes co-regulated by both epigenetic regulatory networks
   (Fig. [84]6a and Supplementary Table 6). The heatmap illustrated that
   these intersection genes distinctly distinguished DR patients from
   healthy controls (Fig. [85]6b). GO analysis was performed to describe
   the biological processes (BP), cellular components (CC), and molecular
   functions (MF) correlated with these genes. In terms of BP, these genes
   showed significant enrichment in coated vesicle membrane. Genes in the
   CC category were primarily related to low-density lipoprotein particle
   receptor binding, clathrin-coated endocytic vesicle membrane,
   clathrin-coated vesicle membrane, and clathrin-dependent endocytosis.
   In the category of MF, these genes were mainly enriched in clathrin
   binding, lipoprotein particle receptor binding, cargo adaptor activity,
   clathrin adaptor activity, and clathrin-coated pit (Fig. [86]6c). KEGG
   analysis demonstrated significant associations with signaling pathways
   such as clathrin-dependent endocytosis, clathrin-coated pit,
   clathrin-coated vesicle membrane, clathrin-coated endocytic vesicle
   membrane, and coated vesicle membrane (Fig. [87]6d). PPI network
   analysis showed 38 nodes and 50 edges in total (Fig. [88]6e). The hub
   gene modules were defined using the MCODE plugin in Cytoscape, with
   five hub genes identified, including TFRC, AP2M1, AP2A1, DAB2, and
   PPP1CB (Fig. [89]6f).

Fig. 6.

   [90]Fig. 6
   [91]Open in a new tab

   The epigenetic regulatory network in DR. (a) Venn diagram containing
   three lists of epigenetic regulation-related genes. (b) Heatmap of the
   intersection genes distinguishing DR patients from healthy controls.
   (c) GO functional enrichment analysis of the intersection genes. (d)
   KEGG pathway enrichment analysis of the intersection genes. (e) PPI
   network of the intersection genes. (f) The significant modules of hub
   genes

Single-cell expression of hub genes

   Single-cell RNA sequencing was employed to examine the expression
   characteristics of DR-related hub genes across various immune cell
   types, including B cells, retinal endothelial cells (EC-R), glomerular
   capillary endothelial cells (GC-EC), macrophages, neutrophils,
   podocytes, and retinal pericytes (RPC) (Fig. [92]7a). Notably, PPP1CB
   exhibited high expression in B cells, while macrophages demonstrated
   elevated AP2A1 expression. AP2M1 and PPP1CB were highly expressed in
   RPC. Hub genes such as TFRC, AP2M1, AP2A1, and DAB2 were specifically
   expressed in EC-R and GC-EC (Fig. [93]7b and c). Figure [94]7d
   illustrated the expression state of the hub genes in DR tissue compared
   to normal tissue. The Uniform Manifold Approximation and
   Projection(UMAP) of single-cell gene expression distribution revealed
   distinct clustering patterns (Fig. [95]7e).

Fig. 7.

   [96]Fig. 7
   [97]Open in a new tab

   Single-cell expression of hub genes in DR. (a) Major cell types. (b)
   Bubble plot showing the average expression levels of hub genes and
   fractions of expressed cells. (c) Heatmap of hub gene expression in the
   cell types. (d) Comparison of hub gene expression in the cell types
   between DR patients and healthy controls. (e) UMAP plots with the
   expression of hub genes in the cell types

   To further elucidate the molecular mechanisms of diabetic
   complications, the analysis of the aforementioned hub genes was
   extended to diabetic kidney disease (DKD). In DKD, EC-R and GC-EC
   showed specific expression of AP2A1 and DAB2. RPC expressed high levels
   of AP2A1 and PPP1CB, while podocytes exhibited high expression of
   AP2M1. Other cells displayed enrichment of DAB2 and TFRC
   (Figs. [98]8a-d). In order to gain further insights into the
   transcriptomic features of immune cells in diabetic complications, the
   DR and DKD data were integrated. Batch effect correction was performed
   on the merged dataset using Harmony [[99]23], thereby achieving an
   integrated single-cell transcriptome.

Fig. 8.

   [100]Fig. 8
   [101]Open in a new tab

   Single-cell expression of hub genes in DKD. (a) Bubble plot showing the
   average expression levels of hub genes and fractions of expressed
   cells. (b) Heatmap of hub gene expression in the cell types. (c)
   Comparison of hub gene expression in the cell types between T2DK
   patients and healthy controls. (d) UMAP plots with the expression of
   hub genes in the cell types

Discussion

   This study provided a comprehensive multi-omics analysis of the
   epigenetic regulatory mechanisms in DR, highlighting the critical roles
   of DNA methylation, mRNA, and miRNA in modulating disease progression.
   The findings revealed hub genes and their regulatory networks in DR,
   offering new insights into the pathophysiology of DR and potential
   molecular targets for clinical interventions.

   Our study further confirmed significant alterations in DNA methylation
   in DR, which could potentially modulate the expression and function of
   specific genes involved [[102]24]. Several DMGs identified in DR were
   found to be closely associated with signaling pathways such as hippo
   signaling pathway and hepatocellular carcinoma. Previous research has
   indicated that the hippo signaling pathway involved in angiogenesis is
   implicated in abnormal vascular responses in DR [[103]25]. Moreover,
   Azuma et al.‘s investigation has revealed that DR stands as an
   independent risk factor for the development of hepatocellular carcinoma
   (HCC) in patients with nonalcoholic fatty liver disease (NAFLD)
   [[104]26].

   At the mRNA level, our study identified a series of differentially
   expressed mRNAs in DR, primarily associated with pathways like
   ribosome, ubiquitin mediated proteolysis, and IL-17 signaling.
   Ribosomal protein networks have been observed to change significantly
   following retinal ischemia and reperfusion (I/R) injury [[105]27],
   suggesting that ribosomal damage may lead to retinal cell dysfunction
   [[106]28]. Studies have shown that changes in mRNA do not significantly
   alter basic functions of retinal pigment epithelium (RPE) cells, but
   may regulate cellular responses to diabetes and hypoxic stress through
   the ubiquitin mediated proteolysis pathway [[107]29]. The IL-17
   signaling pathway is particularly relevant in the activation of retinal
   microglial cells, which secrete pro-inflammatory cytokines and
   chemokines, exacerbating retinal inflammation and vascular damage, thus
   contributing to the progression of DR [[108]30, [109]31].

   Our study also identified a series of miRNAs specifically expressed in
   DR, which may exert key regulatory roles in DR development through
   pathways such as adherens junction and cellular senescence. Genes
   involved in the adherens junction pathway show abnormal methylation and
   differential expression, potentially altering intercellular adhesion
   and signal transmission, thereby affecting vascular stability and
   extracellular matrix organization [[110]32]. In DR, activation of the
   Notch1 signaling pathway leads to damage in the adherens junction
   pathway between vascular endothelial cells, and particularly results in
   the dissociation of VE-cadherin from β-catenin, disrupting vascular
   barrier function and increasing vascular permeability [[111]33]. Under
   hyperglycemic conditions, retinal cells undergo significant senescence
   changes, characterized by the secretion of various inflammatory factors
   and growth factors that constitute the senescence-associated secretory
   phenotype (SASP), promoting the spread of cellular senescence and
   pathological neovascularization, thereby exacerbating the pathological
   conditions of DR [[112]34–[113]36].

   The integration of DNA methylation and miRNA expression data allowed us
   to identify methylated DEGs and construct a mRNA methylation regulatory
   network in DR. In this network, pathways such as the TGF-beta signaling
   pathway and ErbB signaling pathway were found to be associated with DR.
   Elevated TGF-β expression in DR tissue suggests activation of TGF-beta
   signaling pathway, which may affect the retinal fibrotic process and
   vascular barrier function [[114]37, [115]38]. High glucose levels in
   diabetes can lead to aberrant activation of the ErbB signaling pathway,
   promoting apoptosis and inhibiting cell proliferation, which may be
   potentially linked to impaired corneal epithelial healing in diabetes
   [[116]39]. Abnormal activation of the ErbB signaling pathway is also
   associated with inflammation and neovascularization in DR, increasing
   risks of hemorrhage and vision loss [[117]40].

   A miRNA-mRNA regulatory network was also constructed and revealed
   critical pathways for specific miRNAs in DR. The network including
   important hub miRNAs such as hsa-miR-27a-3p and hsa-miR-31-5p, were
   found to be closely associated with signaling pathways such as cellular
   senescence, adherens junction, and endocytosis. Mastropasqua et al.
   have reported downregulation of hsa-miR-27a-3p in DR, which may be
   related to the role of hsa-miR-27a-3p in insulin signaling and glucose
   metabolism [[118]41]. Additionally, hsa-miR-31-5p has been identified
   to be involved in the pathological process of DR [[119]42]. It has been
   demonstrated that the activation of soluble epoxide hydrolase (sEH) in
   the retina in DR promotes the endocytosis of cell adhesion molecules
   like VE-cadherin and N-cadherin, disrupting vascular endothelial cell
   junctions, leading to vascular leakage and retinal edema [[120]43].

   Through comprehensive DR-related epigenetic regulatory network
   construction, five hub genes were identified, including TFRC, AP2M1,
   AP2A1, DAB2, and PPP1CB. Studies have shown that regulation of
   transferrin receptor (TFRC) and related iron metabolism pathways may
   provide novel strategies for the prevention and treatment of DR
   [[121]44, [122]45]. Disabled-2 (DAB2), a protein associated with the
   endocytosis of VEGFR2 (vascular endothelial growth factor receptor 2),
   may become impaired with consequent disruption of VEGFR2 signaling and
   inhibited angiogenesis, highlighting its potential as a therapeutic
   target for DR [[123]46]. Evidence suggests that Arf-directed
   GTPase-activating protein 1 (ARAP1) may regulate the phagocytic
   function of retinal pigment epithelial (RPE) cells through interactions
   with proteins like AP-2 complex subunit alpha-1 (AP2A1) [[124]47].
   Following retinal ischemia/reperfusion (I/R) injury, the AP2M1 level is
   markedly upregulated, aiding in the regulation of retinal
   neurotransmitter release and improving the survival of retinal ganglion
   cells (RGCs) [[125]48]. Changes in the activity of protein phosphatase
   1 (PP1) also have a significant impact on RGC survival and function in
   DR [[126]49], and PPP1CB, a subtype of PP1, is likely to influence
   retinal function under diabetic conditions.

   Furthermore, single-cell RNA sequencing elucidated the distinct
   expression of DR-related hub genes in different immune cells. The five
   identified hub genes were found to be associated with immune cells
   implicated in DR pathogenesis and their abnormal expression in specific
   cell types might drive DR onset and progression [[127]8]. The high
   PPP1CB expression in B cells may be linked to metabolic regulation
   [[128]50], while the elevated AP2A1 expression in macrophages suggests
   its potential role in the inflammatory response of DR [[129]51]. The
   pronounced expression of AP2M1 and PPP1CB in retinal pericytes (RPCs)
   may affect retinal vascular stability and inflammation [[130]52]. The
   high expression of TFRC, AP2M1, AP2A1, and DAB2 in EC-R and GC-EC
   further indicates their role in diabetes-induced vascular abnormalities
   and retinal damage [[131]53, [132]54]. The single-cell analysis also
   revealed key roles for these hub genes in DKD. The specific expression
   of AP2A1 and DAB2 in EC-R and GC-EC may influence microvascular
   permeability [[133]55], and the high levels of AP2A1 and PPP1CB in RPCs
   may be related to kidney filtration function and vascular damage
   [[134]56]. These findings are similar to those observed in DR and could
   suggest shared epigenetic regulatory mechanisms between DR and DKD,
   particularly in endothelial cells [[135]57]. The differential
   expression of these hub genes in immune cells reveals their potential
   role in diabetic complications through immune cell activation and
   endothelial cell function, highlighting their potential as biomarkers
   and therapeutic targets in diabetic complications.

   Although this study has revealed the epigenetic regulatory mechanisms
   of DR through multi-omics integration, several limitations remain.
   First, the analysis utilizes a small sample size of high throughput
   sequencing data, making it difficult for the analysis to filter
   further, which may limit the generality and statistical robustness of
   the findings. Second, while the study focused on DNA methylation and
   miRNAs in DR, other epigenetic mechanisms such as histone modifications
   and long non-coding RNAs (lncRNAs) require further exploration. Lastly,
   the influence of epigenetic modifications on gene expression in DR
   needs further functional validation to clarify the biological roles of
   key genes and pathways.

Conclusion

   This study systematically investigated the epigenetic regulatory
   mechanisms of DR. Through an integrative multi-omics analysis, we
   identified significant alterations in DNA methylation, and mRNA and
   miRNA expression associated with DR, and constructed a comprehensive
   epigenetic regulatory network, highlighting five hub genes. Single-cell
   RNA sequencing analysis further revealed the specific expression of
   these hub genes in various immune cell, underscoring their crucial
   roles in the onset and progression of DR. The findings not only offer
   new insights into the pathophysiology of DR but also provide potential
   molecular targets for its diagnosis and treatment.

Electronic Supplementary Material

   Below is the link to the electronic supplementary material.
   [136]Supplementary Material 1^ (350.2KB, csv)
   [137]Supplementary Material 2^ (2.5MB, csv)
   [138]Supplementary Material 3^ (553KB, csv)
   [139]Supplementary Material 4^ (4KB, csv)
   [140]Supplementary Material 5^ (4.7KB, csv)
   [141]Supplementary Material 6^ (241B, csv)

Acknowledgements