Skip to main content
NCBI home page
European Journal of Medical Research logoLink to European Journal of Medical Research
. 2025 Aug 20;30:777. doi: 10.1186/s40001-025-03044-x

Ferroptosis-related hub genes and immune cell dynamics as diagnostic biomarkers in age-related macular degeneration

Jinquan Chen 1,2,#, Zhao Long 2,#, Dandan Shi 1, Qian Zhang 1, H Peng 2,✉
PMCID: PMC12366384  PMID: 40830492

Abstract

Background

Age-related Macular Degeneration (AMD) is widely acknowledged as a principal cause of vision loss in the elderly. Currently, the therapeutic interventions available in clinical practice fail to achieve satisfactory outcomes. Therefore, it is imperative that we approach the progress of AMD from novel perspectives in order to explore new therapeutic strategies.

Method

We obtained transcriptomic data from the macular and the peripheral retina from patients with AMD and a control group from the Gene Expression Omnibus (GEO) database. Through Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, we identified differentially expressed genes (DEGs) that were significantly enriched in functions associated with ferroptosis. Subsequent application of machine learning techniques enabled the identification of key hub genes, whose diagnostic potential was further validated. Additionally, the expression of these hub genes was corroborated in both animal and cellular models. Finally, we performed a functional enrichment analysis of these hub genes.

Results

In the macula of patients with AMD, 452 DEGs were identified, while in the peripheral retina, 222 DEGs were discovered. Within the macula, 19 genes were associated with ferroptosis, compared to 3 in the peripheral retina. Consequently, the macular was selected as the primary focus of the study. Subsequent screening of these 19 genes using LASSO regression, Support Vector Machine (SVM), and Random Forest algorithms identified four hub genes: FADS1, TFAP2A, AKR1C3, and TTPA. Consequently, we utilized cigarette smoke extract (CSE) to either stimulate retinal pigment epithelial (RPE) cells in vitro or administer it via intravitreal injection, thereby establishing in vitro and in vivo models of AMD. Results from RT-PCR and Western blot analyses revealed an upregulation of FADS1, AKR1C3, and TTPA, while TFAP2A exhibited decreased expression. Finally, we investigated the infiltration of immune cells within the macular and performed a functional enrichment analysis of the hub genes.

Conclusion

We identified four key ferroptosis-related genes (FRGs)—FADS1, AKR1C3, TFAP2A, and TTPA—that possess diagnostic relevance for AMD and correlate with immune cell infiltration. Moreover, significant changes in both mRNA and protein expression levels of these genes have been observed in in vitro experiments and mice models.

Keywords: Age-related macular degeneration, Ferroptosis-related genes, Machine learning, Immune cells

Introduction

In industrialized nations, AMD is a primary cause of visual impairment in individuals aged 50 and above [1]. While anti-VEGF therapy has shown partial efficacy in treating neovascular AMD (nAMD), its exorbitant cost and suboptimal therapeutic outcomes remain significant challenges [2, 3]. Therefore, exploring the pathogenesis of AMD is currently crucial for identifying novel therapeutic approaches.

The primary pathological mechanism underlying AMD is characterized by oxidative stress and chronic inflammation, which lead to the damage of RPE cells [4]. This damage results in structural disruption and functional dysregulation of the retina [5]. Furthermore, the aging process exacerbates the accumulation of iron ions within RPE cells [6]. Iron-induced cell death, a regulated cell death process that is dependent on iron and triggered by lipid peroxidation [7], is marked by the accumulation of reactive oxygen species (ROS) [8] and lipid peroxidation products (LPO), along with a concomitant decrease in glutathione (GSH) and glutathione peroxidase 4 (Gpx4) [9, 10]. This form of cell death is closely associated with the pathogenesis of neurodegenerative diseases, resistance to tumor therapies, and ischemia–reperfusion injury [11]. Therefore, investigating the role of iron-induced cell death in AMD pathogenesis is essential, as it may provide critical insights into the disease mechanisms and facilitate the development of novel therapeutic strategies.

This study utilizes a variety of bioinformatics approaches to identify FRGs in patients with AMD. These genes demonstrate significant diagnostic potential and have been corroborated through external datasets. Additionally, the study explores the enriched signaling pathways linked to these genes and their association with immune cell infiltration. The expression of key hub genes was validated using both in vivo and in vitro models. Our research offers new insights into the progression of AMD from an innovative perspective.

Materials and methods

Date sources

We sourced the GSE29801 gene set from the GEO database as our primary dataset for analysis, using GSE49107 as the validation cohort.

DEG identification and analyses of functional enrichment

We utilized the “limma” package to assess differential gene expression between macula and the peripheral retina. Subsequently, we performed functional enrichment analysis of these DEGs using the GO and KEGG databases.

Identification of hub-genes

We have identified 322 genes associated with ferroptosis and conducted an intersection analysis with genes differentially expressed in the macula and the peripheral retina to isolate candidate genes. Subsequently, these FRGs from the macular were utilized as candidates for hub gene selection. We derived hub genes by intersecting the candidate genes identified through three analytical methods: LASSO regression, SVM, and Random Forest.

Analysis of gene set enrichment

Gene Set Variation Analysis (GSVA) is an unsupervised gene set enrichment method that assesses changes in the expression levels of gene sets specific to each sample. We used GSVA to analyze hub genes, identifying biological pathways associated with the progression of AMD [12].

Immune cell infiltration

CIBERSORT is an advanced computational method used to estimate the relative abundance of cell components from complex tissue transcriptome data. This algorithm, based on linear Support Vector Regression (SVR), accurately distinguishes and quantifies the proportions of various cell types in mixed cell samples. We utilized CIBERSORT to analyze gene sets from the macular, obtaining proportions of immune cell infiltration [13].

The isolation of human primary RPE and treatment with CSE

Under sterile conditions, the eye is dissected around the periphery of the iris to remove anterior chamber structures, including the lens and iris, thereby exposing the retina and posterior ocular structures. Using fine tweezers or a needle, the RPE layer beneath the retina is gently scraped. This step is performed under a microscope to ensure precise separation of the RPE cells without damaging the retina. The isolated RPE cells are then transferred to a culture container filled with a specific medium containing essential growth factors and antibiotics to support cell growth and minimize contamination. The culture dish with RPE cells is placed in an incubator, typically set at 37 degrees Celsius with 5% CO2. The medium is refreshed regularly based on the condition of cell growth. Cell morphology and proliferation are monitored and recorded. Once the cells reach an appropriate density, they are ready for subsequent experiments [14].

We treated primary RPE cells with 0.5% CSE for 24 h. Subsequently, we extracted cellular proteins and RNA using protein extraction solutions and RNA extraction kits, respectively, to prepare for further experiments.

Construction of animal models

We sourced twenty mature C57BL/6 mice from the institution's animal facility. Following anesthesia with isoflurane, a 2 µl dose of 0.5% CSE was injected into the vitreous body of each mouse using a microsyringe. Seven days post-injection, the mice were euthanized via an overdose of pentobarbital sodium. Subsequently, their eyeballs were removed under a microscope for subsequent experiments.

Observation of ferroptosis using electron microscopy

Tissue samples are initially fixed using glutaraldehyde solution. Following fixation, the samples undergo a dehydration process involving a series of ethanol solutions with increasing concentrations, and are subsequently embedded in epoxy resin. The embedded samples are then sectioned into ultra-thin slices ranging from 50 to 70 nm in thickness using an ultramicrotome, to facilitate imaging with an electron microscope. These sections are typically stained with lead citrate to enhance image contrast under the electron microscope. The prepared samples are finally observed under an electron microscope.

Quantitative RT–PCR

Total RNA was extracted from RPE cells and mouse retinal tissues using an RNA isolation kit to ensure both purity and integrity. The RNA was then reverse transcribed into complementary DNA (cDNA) using a reverse transcriptase enzyme. This cDNA was subsequently subjected to polymerase chain reaction (PCR) amplification with specific primers, which included multiple cycles of denaturation, annealing, and extension. The PCR results were compared against standards or control samples to quantify relative gene expression. To ensure accuracy, calibration curves and normalization against housekeeping genes were employed.

Western blotting

Protein samples were first extracted from RPE cells and mouse retinal tissues. The concentration of these proteins was subsequently quantified using the BCA assay to ensure consistent sample loading onto the gel. Proteins were separated according to their molecular weight via SDS-PAGE (sodium dodecyl sulfate–polyacrylamide gel electrophoresis). The samples, mixed with a loading buffer, were heated and loaded into the wells of the gel. Electrophoresis was carried out until the proteins were sufficiently resolved. Following separation, the proteins were transferred from the gel to a PVDF (polyvinylidene fluoride) membrane using a wet transfer method. The membrane was then blocked with a bovine serum albumin (BSA) solution to prevent non-specific binding. It was incubated with a primary antibody specific to the target protein, followed by washing to remove unbound antibodies. A secondary antibody conjugated to horseradish peroxidase (HRP) was then applied. The presence of the HRP-conjugated secondary antibody was detected using chemiluminescent, colorimetric, or fluorescent substrates. The resulting signal was visualized with a digital imaging system to quantify and document the protein bands.

TF-gene interactions and gene-miRNA regulatory network construction

To investigate the regulatory interactions involving the identified hub genes, we utilized the NetworkAnalyst platform. Specifically, potential transcription factor (TF) regulating these hub genes were predicted using the ChEA3 database (based on enrichment of publicly available TF binding data). Concurrently, experimentally validated microRNAs (miRNAs) targeting the hub genes were identified using the TarBase v9.0 database. These results were integrated to explore potential upstream TF regulators and downstream miRNA interactions for the hub genes.

Statistical analysis

Statistical analyses for this study were performed using R software (version 4.2.2). A P-value of less than 0.05 was deemed statistically significant. All P-values were calculated using a two-tailed approach.

Results

DEGs identified between AMD and healthy in the macular and the peripheral retina

We performed a separate analysis of the macular and peripheral retina within the GSE29801 dataset. In our examination of AMD patients, we identified 251 genes with reduced expression and 201 genes with increased expression in the macular region compared to healthy controls (Fig. 1A). In the peripheral retina, we found 91 genes with decreased expression and 131 genes with increased expression (Fig. 1B). A heatmap was created to visually represent the top 50 DEGs exhibiting the most significant regulatory differences between the macular (Fig. 1C) and peripheral retina (Fig. 1D) in AMD patients compared to healthy. GSEA enrichment analysis revealed that DEGs in the macular region were significantly enriched in the Transmission Across Chemical Synapses and Neuronal System pathways (Fig. 1E). In contrast, DEGs in the peripheral retinal region were predominantly enriched in the Cellular Responses to Stimulation and NABA Matrisome pathways (Fig. 1F).

Fig. 1.

Fig. 1

Open in a new tab

DEGs in AMD patients compared to healthy. A Volcano plots illustrating DEGs in the macula of AMD compared to healthy. B Volcano plots illustrating DEGs in the peripheral retina of AMD compared to healthy. C The heatmap showed the DEGs in macula. D The heatmap showed the DEGs in the peripheral retina. E GSEA reveals functional enrichment of DEGs in the macula. F GSEA reveals functional enrichment of DEGs in the peripheral retina

The functional enrichment of DEGs in AMD patients and healthy individuals

Functional enrichment analyses were conducted separately for DEGs in the macula and peripheral retina. GO enrichment analysis of DEGs in the macular region revealed primary enrichment in biological processes (BP) related to system development, regulation of multicellular organismal processes, and animal organ development. In terms of cellular components (CC), significant enrichment was observed in the extracellular region, extracellular space, and cell surface. For molecular functions (MF), DEGs were predominantly enriched in signaling receptor binding, amide binding, and organic acid binding pathways (Fig. 2A).

Fig. 2.

Fig. 2

Open in a new tab

Functional enrichment analysis of DEGs. A, B The top 6 functional enrichment in BP, CC, and MF analysis in the macula and peripheral retina. C, D The KEGG analysis of DEGs in the macula and peripheral retina. E, F Integration of KEGG and GO analyses to identify modules associated with DEGs in the macula and peripheral retina

In contrast, GO enrichment analysis of DEGs in the peripheral retina highlighted significant enrichment in BP pathways associated with response to abiotic stimulus, response to unfolded proteins, and response to topologically incorrect proteins. Enrichment was also notable in CC pathways such as the extracellular region, extracellular space, and basal part of the cell. In the MF category, prominent enrichment was observed in pathways related to molecular function regulator activity, hormone activity, and protein folding chaperone (Fig. 2B).

KEGG pathway enrichment analysis showed that DEGs in the macular region were primarily enriched in pathways such as Neuroactive ligand-receptor interaction, the PI3K-Akt signaling pathway, and the MAPK signaling pathway (Fig. 2C). Conversely, DEGs in the peripheral retina were significantly enriched in pathways including Protein processing in the endoplasmic reticulum, the Pentose phosphate pathway, and Renal cell carcinoma (Fig. 2D).

Combined GO and KEGG analyses revealed that DEGs in the macular region were predominantly enriched in three GO modules: GO:0062023, GO:0005788, and GO:0030198 (Fig. 2E). In contrast, DEGs in the peripheral retina were mainly enriched in the GO modules GO:0045178, GO:0016323, and GO:0009925 (Fig. 2F).

Expression and functional analysis of genes associated with ferroptosis in AMD

To further ascertain the expression of FRGs in the retinas of patients with AMD, we conducted an intersection analysis between DEGs in the macula and the peripheral retina with those associated with ferroptosis. The Venn diagram analysis revealed that among the DEGs in the macula, 19 FRGs were identified (Fig. 3A). Similarly, in the peripheral retina, 3 FRGs were identified (Fig. 3B).

Fig. 3.

Fig. 3

Open in a new tab

Identification of FRGs in the combined microarray set of GSE29801. A The interaction of the DEGs and Ferroptosis-related genes in the macula. B The interaction of the DEGs and Ferroptosis-related genes in the the peripheral retina. C GO analysis of 19 FRGs. D KEGG analysis of 19 FRGs. E Integration of KEGG and GO analyses to identify modules associated with individual FRGs. F Integration of KEGG and GO analyses to identify modules associated with individual different expression gene

Thus, we have identified FRGs in the macula as the primary focus of our study. We conducted a functional enrichment analysis on these 19 FRGs, finding through GO analysis that these FRGs are predominantly involved in processes such as small molecule metabolic processes, fatty acid metabolic processes, and small molecule biosynthetic processes (Fig. 3C). Furthermore, KEGG enrichment analysis revealed that these 19 FRGs are significantly associated with the biosynthesis of unsaturated fatty acids, ferroptosis, and fatty acid metabolism (Fig. 3D). The subsequent analysis, conducted in collaboration with GO and KEGG, revealed that these FRGs are predominantly enriched in three GO modules: GO:0006631 (fatty acid metabolic process), GO:0001228 (DNA-binding transcription activator activity, RNA polymerase II-specific), and GO:0001676 (long-chain fatty acid metabolic process) (Fig. 3E).

Subsequently, we utilized volcano plots and heatmaps to depict the expression profiles of these genes in the macula of AMD (Fig. 4A and B). Notably, AKR1C3, BRD3, CP, FADS1, FADS2, MYCN, SLC7A11, SREBF1, TF, TTPA and SCD were found to be upregulated, while ACSF2, AR, ATF3, GALNT14, HSPB1, NRAS, TFAP2A and TGFB1 displayed downregulation (Fig. 4C).

Fig. 4.

Fig. 4

Open in a new tab

FRGs in the microarray set of GSE29801. A Volcano plot shows the 19 FRGs. B Heatmap shows the 19 FRGs. C The expression of 19 FRGs between AMD and healthy

Identification of hub genes through machine learning techniques

To pinpoint hub genes for AMD, we applied Lasso, SVM, and Random Forest analyses to 19 FRGs. LASSO analysis isolated 12 hub genes from these candidates (Fig. 5A), while subsequent SVM and Random Forest analyses each identified their own top 10 hub genes (Fig. 5B and C). Cross-analyzing these results, we identified four pivotal genes: FADS1, TFAP2A, AKR1C3, and TTPA, illustrated in a Venn diagram (Fig. 5D).

Fig. 5.

Fig. 5

Open in a new tab

Machine algorithms for identifying signature genes are illustrated as follows. A The penalty plot of the LASSO model, which includes error bars representing standard errors, and the LASSO plot demonstrating the reduction in coefficient sizes as the penalty parameter k increases. B The top 10 genes ranked by relative importance in the SVM model. C The top 10 genes ranked by relative importance in the Random Forest model. D The interaction among the LASSO, SVM, and Random Forest models

We analyzed the expression of hub genes in the macula associated with AMD. The results indicate an increased expression of AKR1C3, FADS1, and TTPA, while the expression of TFAP2A was found to be decreased (Fig. 6A). ROC curve analysis demonstrated AUC values of 0.634 for FADS1, 0.636 for TFAP2A, 0.657 for AKR1C3, and 0.626 for TTPA (Fig. 6B). We identified the genes associated with the four hub genes through GeneMANIA (Fig. 6C) and subsequently visualized their interactions using a chord diagram (Fig. 6D). In our final analysis, we determined the chromosomal locations of the hub genes within the human genome: FADS1 and TTPA are located on chromosome 11, AKR1C3 is on chromosome 10, and TFAP2A is on chromosome 6 (Fig. 6E).

Fig. 6.

Fig. 6

Open in a new tab

The performance of the hub genes in the GSE29801 dataset. A The expression levels of hub genes were compared between patients with AMD and healthy. B ROC analysis demonstrated the diagnostic efficacy of the hub genes. C The Protein–Protein Interaction (PPI) network of four FRGs, constructed using Genemania. D Interactions among four hub genes. E Chromosomal localization of hub genes

An external validation cohort was utilized to evaluate the diagnostic efficacy of the hub genes in AMD patients. We observed a significant elevation in the expression levels of these hub genes within the AMD group of the external validation cohort (Fig. 7A). ROC curve analysis demonstrated AUC values of the hub genes were as follows: AKR1C3 (AUC = 1.000), FADS1 (AUC = 0.944), TFAP2A (AUC = 0.889), and TTPA (AUC = 0.889) (Fig. 7B).

Fig. 7.

Fig. 7

Open in a new tab

The performance of the hub genes in GSE49107. A The expression of hub genes between AMD patients and healthy. B ROC showed the diagnostic performance of the hub genes. *P < 0.05

Immune cell infiltration in AMD macula

To investigate immune cell infiltration in the macular region of AMD patients, we utilized the R package"CIBERSORT"for analysis. Our results revealed increased infiltration of naive B cells, plasma cells, CD8 + T cells, regulatory T cells (Tregs), resting NK cells, M0 macrophages, M1 macrophages, M2 macrophages, and neutrophils in the macular region associated with AMD. Conversely, there was a decrease in the infiltration of resting memory CD4 + T cells, activated NK cells, monocytes, and resting mast cells (Fig. 8A).

Fig. 8.

Fig. 8

Open in a new tab

The association between immune cell infiltration and FRGs is presented as follows: A Comparative analysis of immune cell infiltration between AMD patients and healthy. B The proportion of immune cell infiltration within the samples. C–F The relationship between hub genes and significantly different levels of immune cell infiltration. Statistical significance is indicated as follows: “ns” means P ≥ 0.05, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001

Further analysis of the proportions of various immune cells in individual samples indicated that, compared to the healthy control group, the macular region of AMD patients had a higher proportion of naive B cells, Tregs, follicular helper T cells, and M0 macrophages (Fig. 8B).

To explore the relationships between hub genes and immune cells, we conducted individual gene analyses. Our findings demonstrated positive correlations between AKR1C3 and follicular helper T cells, Tregs, and naive B cells. FADS1 was positively correlated with M1 macrophages, resting mast cells, and M0 macrophages. TFAP2A exhibited positive correlations with CD8 + T cells, monocytes, and activated NK cells. TTPA was positively correlated with eosinophils, follicular helper T cells, and Tregs (Fig. 8C to F).

The expression of hub-genes in vitro experiments

After isolating primary RPE cells from the eye, the expression of Bestrophin and ZO-1 proteins in the RPE cells was observed using electron microscopy (Fig. 9A). We stimulated primary RPE cells with CSE and, using electron microscopy, observed that the extract could induce mitochondrial changes in these cells, including a reduction in volume, increased mitochondrial membrane density, and a decrease or disappearance of mitochondrial cristae. Additionally, we noted damage to the cell and organelle membrane structures (Fig. 9B). Western blot results showed an increase in the protein expression of AKR1C3, FADS1, and TTPA, while TFAP2A expression decreased (Fig. 9C). Similarly, RT-PCR results indicated that, compared to the control group, the expression levels of AKR1C3, FADS1, and TTPA were elevated in the primary cells of the experimental group, whereas TFAP2A expression was reduced (Fig. 9D).

Fig. 9.

Fig. 9

Open in a new tab

CSE-induced modulation of hub gene expression in primary RPE cells. A Fluorescence microscopy reveals that primary RPE cells express Bestrophin and ZO-1. B Use electron microscopy to observe ferroptosis in RPE cells induced by CSE. C Western blot analysis Hub genes expression of CSE treated RPE cells and control. D The expression of FADS1, TFAP2A, AKR1C3, and TTPA in CSE treated RPE cells and control were quantified by qRT–PCR. *P < 0.05, **P < 0.01, ***P < 0.001

The expression of hub-genes in mice model

Similarly, we constructed mice model of retinal inflammation by injecting CSE into the vitreous body of mice. Electron microscopy revealed damage to the cell and organelle membrane structures in retinal sections of mice from the experimental group. This pathological change was not observed in the control group mice (Fig. 10A). Western blot analysis showed elevated protein expression levels of AKR1C3, FADS1, and TTPA, alongside a reduction in TFAP2A protein expression (Fig. 10B). Consistently, RT-PCR results revealed that, compared to control mice, the experimental group mice exhibited increased expression of AKR1C3, FADS1, and TTPA mRNA, while TFAP2A mRNA expression was decreased (Fig. 10C).

Fig. 10.

Fig. 10

Open in a new tab

Expression of hub genes in mice model. A Utilize electron microscopy to investigate ferroptosis in the retina of mice induced by CSE. B Western blot analysis Hub genes expression of retina  from AMD mice and control. C The expression of FADS1, TFAP2A, AKR1C3, and TTPA in AMD mice and control were quantified by qRT–PCR. *P < 0.05, **P < 0.01, ***P < 0.001

Gene set enrichment analysis of hub genes

To determine the functions of hub genes, we conducted functional enrichment analyses for individual hub genes using the"GSVA"package. The results reveal that AKR1C3 is associated with antigen processing and presentation, immune receptor activity, and cytokine binding (Fig. 11A). Similarly, FADS1 is related to the integrin-mediated signaling pathway, antigen processing and presentation of peptide antigen, and growth factor binding (Fig. 11B). TFAP2A is linked to the photoreceptor disc membrane, respiratory chain complex, and inner mitochondrial membrane protein complex (Fig. 11C). Additionally, TTPA correlates with functions associated with the MHC protein complex on the lumenal side of the endoplasmic reticulum membrane and olfactory receptor activity (Fig. 11D).

Fig. 11.

Fig. 11

Open in a new tab

GSEA of hub genes in AMD is presented as follows. A GSEA of FADS1 in AMD utilizing KEGG pathway analysis. B GSEA of TFAP2A in AMD employing KEGG pathway analysis. C GSEA of AKR1C3 in AMD through KEGG pathway analysis. D GSEA of TTPA in AMD using KEGG pathway analysis

Construction of the TF-gene interactions and Gene-miRNA regulatory network

We utilized NetworkAnalyst to investigate TF-gene interactions and Gene-miRNA regulatory networks associated with hub genes. Subsequently, we employed Cytoscape version 3.9.3 to visualize the regulatory networks. As a result, we identified a total of 112 transcriptional regulatory relationships (Fig. 12A) and 224 miRNA relationships (Fig. 12B).

Fig. 12.

Fig. 12

Open in a new tab

TF-gene and Gene-miRNA interactions. A Diagram of TF-gene interactions, where rend circles represent genes. B Diagram of Gene-miRNA interactions, where rend circles represent genes

Discussion

AMD is a persistent and localized inflammatory response, ultimately resulting in visual impairment among patients [15]. Its primary pathological mechanism involves oxidative stress and chronic inflammatory damage to RPE cells, leading to structural disruption and functional dysregulation [16]. Furthermore, with the sustained inflammatory response, there is an accelerated accumulation of iron ions within RPE cells. This study investigated the DEGs in the macula and the peripheral retina of both AMD patients and healthy. Analysis of the macula identified 19 FRGs, with 3 FRGs found to be relocated upon comparison with the peripheral retina. Consequently, it was inferred that iron death predominantly occurs in the macular area. Subsequently, employing machine-based screening, we identified 4 FRGs, namely FADS1, AKR1C3, TFAP2A, and SREBF1. Finally, we established in vivo and in vitro models of AMD, corroborating the expression patterns of these genes.

Ferroptosis represents a form of regulated cell death characterized by iron-dependent lipid peroxidation. In response to metabolic stress, cells modulate their lipid metabolism to ensure survival, potentially altering their susceptibility to iron-mediated toxicity [17]. Previous research has indicated a significant impact of lipid metabolism on iron toxicity [18]. Fatty acid desaturase 1 (FADS1) is recognized as a crucial enzyme in the biosynthesis of biologically active metabolites, including the conversion of docosahexaenoic acid (DGLA) to arachidonic acid [19]. Epidemiological studies have identified an association between AMD and alterations in lipid metabolism [20]. Specifically, patients with AMD exhibit comparatively lower triglyceride (TG) levels than controls. Additionally, genetic variations in APOE and CETP contribute additively to variations in high-density lipoprotein cholesterol (HDLC) and apolipoprotein AI levels [21, 22]. Notably, FADS1 is highly expressed in the serum of individuals with AMD. APOE and CETP genotypes exert a cumulative effect on HDLC and apolipoprotein levels, whereas FADS1 is highly expressed in the serum of patients with AMD [23]. Researchers hypothesize that the upregulation of FADS1 expression may enhance sensitivity to iron dysregulation. Consequently, we propose that FADS1 may participate in the pathogenesis of AMD by modulating iron metabolism. AKR1C3, a member of the aldoketoreductase (AKR) family, plays various cellular roles in regulating prostaglandins, steroid hormones, and retinoic acid metabolism [24]. AKR1C3 is considered a promising biomarker and therapeutic target for several diseases, including acute myocardial infarction, hepatocellular carcinoma, and metastatic melanoma. It is involved in regulating multiple biological processes, such as cell proliferation, epithelial-mesenchymal transition, and inflammatory responses [25]. Evidence suggests that upregulation of AKR1C3 expression can reduce intracellular ROS levels, increase GSH, and thereby enhance the GSH/GSSG ratio, leading to a reduction in cellular inflammation [26, 27]. However, there is currently a lack of direct evidence regarding the role of AKR1C3 in the pathogenesis of AMD. Our in vitro and in vivo experiments have demonstrated elevated AKR1C3 expression, but its specific functional contributions require further investigation. Transcription Factor AP-2 Alpha (TFAP2A) is a pivotal DNA-binding protein with the capability to recognize specific sequences, and it can function synergistically with viral and cellular promoter to modulate the expression of a multitude of genes [28]. KEGG pathway enrichment analysis indicates that TFAP2A is intimately associated with oxidative processes. Previous research has demonstrated that the suppression of TFAP2A expression can impede the activation of oxidative stress-related genes such as heme oxygenase 1 (HO-1), nuclear factor erythroid 2-like 2 (NRF2), and NAD (P)H quinone dehydrogenase 1 (NQO1) [29, 30]. Consequently, it is hypothesized that elevated TFAP2A expression in patients with AMD could exacerbate oxidative stress within the macula. Sterol regulatory element-binding transcription factor 1 (SREBF1) plays a critical mediating role between TP63 and fatty acid metabolism, regulating the biosynthesis of fatty acids, sphingolipids (SL), and glycerophospholipids (GPL) [31]. Research has indicated that increased expression of SREBF1 can mitigate cellular iron-induced death through the SREBF1-stearoyl-CoA pathway, thereby reducing the accumulation of lipid metabolites [32]. These lipid metabolites are primary constituents of drusen and may contribute to the progression of AMD.

Numerous studies have indicated a close association between the progression of AMD and immune cell infiltration [33]. Our investigation revealed an increase in the expression of various immune cells in the macular of AMD patients. Neutrophils, a key component of the innate immune system, have been shown to be elevated in patients with nAMD [34]. Neutrophils contribute to the progression of choroidal neovascularization (CNV) by secreting matrix metalloproteinase 9 (MMP-9), which degrades and remodels the extracellular matrix, thereby compromising the integrity of the RPE barrier [35]. Furthermore, the AKT2/NF-kB/LCN-2 signaling pathway has been implicated in the inflammation associated with AMD [36]. It is well established that macrophages play a crucial role in tissue repair, regeneration, and fibrosis following injury. Macrophage M1, or pro-inflammatory macrophages [37], can inhibit tumor growth but may also contribute to tissue damage. In contrast, macrophage M2, or anti-inflammatory macrophages, are primarily involved in tissue repair and angiogenesis [38]. Research has indicated that M1 macrophages are active during the early stages of nAMD, whereas M2 macrophages become prominent in the later stages of the disease [39]. Additionally, an increase in macrophage expression has been observed in the subretinal choroid of patients with geographic atrophy (GA) type AMD [40]. Studies suggest that reducing macrophage expression in the subretinal space may lead to RPE cell death, thereby advancing AMD [41]. T lymphocytes can drive M1 macrophage polarization through the release of pro-inflammatory cytokines, linking innate and adaptive immunity in the early stages of AMD [42]. Consequently, a detailed investigation of the immune microenvironment in the retinas of AMD patients could provide a more comprehensive understanding of the disease’s pathogenesis.

This study, through integrated analysis of the ChEA3 and TarBase v9.0 databases, reveals a significant connection between the transcriptional regulatory network of hub genes and the ferroptosis pathway in AMD. The predicted 112 transcription factors  (e.g., NRF2, TP53, STAT3) and 224 miRNAs  (e.g., hsa-miR-155-5p, hsa-miR-21-5p) collectively target core ferroptosis genes (e.g., GPX4, ACSL4), forming a critical regulatory network governing oxidative stress and lipid peroxidation. Specifically, suppression of NRF2 function (regulated by miR-21-5p, among others) [43] and activation of TP53/STAT3 (promoting hsa-miR-155-5p expression) act synergistically to disrupt the redox balance in RPE cells [44]. This disruption leads to the accumulation of lipid ROS and mitochondrial damage, thereby driving two core AMD pathological phenotypes: RPE atrophy (associated with GA) and CNV [45]. Notably, the hub genes TTPA, FADS1, and AKR1C3 are directly involved in retinal antioxidant defense and the maintenance of lipid homeostasis. Dysregulation of their expression may amplify the ferroptosis cascade. Targeting this regulatory network – for instance, by reducing retinal iron accumulation via iron chelators or designing nanotherapeutics antagonizing hsa-miR-155-5p may represent a novel strategy for intervening in AMD progression [46].

In summary, our exploratory study has several limitations. First, the data used in our investigation were obtained from a small cohort of AMD patients sourced from a public database, which may limit the generalizability of our findings. Second, to further validate our insights, we anticipate acquiring a larger number of clinical trial samples for subsequent research, as the available external validation data involve a relatively small patient sample size. Third, a more detailed examination of the roles of these key genes and ferroptosis processes in the progression of AMD is needed in animal or cellular models, to corroborate our findings through molecular-level investigations.

Conclusions

In summary, our study identified Four FRGs in the macular of AMD patients: FADS1, TFAP2A, AKR1C3, and TTPA. Additionally, we examined immune cell infiltration in AMD patients and explored its association with these four FRGs, thereby offering novel insights into the onset and progression of AMD.

Acknowledgements

The authors extend their heartfelt gratitude to the GEO repository for its generous provision of unrestricted access to the data. This access was instrumental in enabling the comprehensive analysis and subsequent findings of this research.

Author contributions

The conceptualization of the study and the initial drafting of the manuscript were undertaken by CJQ and ZL. Data analysis and figure formatting were performed by SDD and ZQ. The revisions of the manuscript were supervised by PH. All authors contributed to the article and approved the final version prior to submission.

Funding

This study received funding from the National Natural Science Foundation of China (Grant No. 81670881).

Data availability

The article and supplementary materials provide a thorough summary of the key findings from this research. For additional inquiries, please contact the author at chenj187@163.com.

Declarations

Ethics approval and consent to participate

This study received ethical approval from the Ethics Committee of The First Affiliated Hospital of Chongqing Medical University (Chongqing, China), with the approval number IACUC-CQMU-2023-0355.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher's Note

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

Jinquan Chen and Zhao Long have contributed equally to this work.

References

  • 1.Kim J, Han SY, Min H. Ginseng for an eye: effects of ginseng on ocular diseases. J Ginseng Res. 2020;44(1):1–7. 10.1016/j.jgr.2018.11.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Singh A, Faber C, Falk M, Nissen MH, Hviid TV, Sørensen TL. Altered expression of CD46 and CD59 on leukocytes in neovascular age-related macular degeneration. Am J Ophthalmol. 2012. 10.1016/j.ajo.2012.01.036. [DOI] [PubMed] [Google Scholar]
  • 3.Yin X, He T, Yang S, Cui H, Jiang W. Efficacy and safety of antivascular endothelial growth factor (Anti-VEGF) in treating neovascular age-related macular degeneration (AMD): a systematic review and meta-analysis. J Immunol Res. 2022;2022:6004047. 10.1155/2022/6004047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Lee J, Kang HG, Kim HR, et al. Pigmentary abnormality without significant drusen as a risk factor for late age-related macular degeneration. Sci Rep. 2022;12(1):769. 10.1038/s41598-022-04798-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Zhang N, Peairs JJ, Yang P, et al. The importance of Bcl-xL in the survival of human RPE cells. Invest Ophthalmol Vis Sci. 2007;48(8):3846–53. 10.1167/iovs.06-1145. [DOI] [PubMed] [Google Scholar]
  • 6.Zhao T, Guo X, Sun Y, et al. Iron accumulation and lipid peroxidation in the aging retina: implication of ferroptosis in age-related macular degeneration. Aging Dis. 2021;12(2):529–551. 10.14336/AD.2020.0912. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Song D, Dunaief JL. Retinal iron homeostasis in health and disease. Front Aging Neurosci. 2013;5:24. 10.3389/fnagi.2013.00024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Chen D, Tavana O, Chu B, et al. NRF2 is a major target of ARF in p53-independent tumor suppression. Mol Cell. 2017;68(1):224-32.e4. 10.1016/j.molcel.2017.09.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Miao Y, Chen Y, Xue F, et al. Contribution of ferroptosis and GPX4’s dual functions to osteoarthritis progression. EBioMedicine. 2022;76: 103847. 10.1016/j.ebiom.2022.103847. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Luo X, Gong HB, Gao HY, et al. Oxygenated phosphatidylethanolamine navigates phagocytosis of ferroptotic cells by interacting with TLR2. Cell Death Differ. 2021;28(6):1971–89. 10.1038/s41418-020-00719-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Yan HF, Zou T, Tuo QZ, et al. Ferroptosis: mechanisms and links with diseases. Signal Transduct Target Ther. 2021;6(1):49. 10.1038/s41392-020-00428-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Li J, Liu X, Wang J, et al. Identification of immunodiagnostic blood biomarkers associated with spinal cord injury severity. Front Immunol. 2023;14:1101564. 10.3389/fimmu.2023.1101564. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Jiang F, Mao Y, Lu B, Zhou G, Wang J. A hypoxia risk signature for the tumor immune microenvironment evaluation and prognosis prediction in acute myeloid leukemia. Sci Rep. 2021;11(1):14657. 10.1038/s41598-021-94128-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Marcu IC, Eberhard N, Yerly A, et al. Isolation of human small extracellular vesicles and tracking of their uptake by retinal pigment epithelial cells in vitro. Int J Mol Sci. 2020. 10.3390/ijms21113799. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Curriero FC, Pinchoff J, van Landingham SW, Ferrucci L, Friedman DS, Ramulu PY. Alteration of travel patterns with vision loss from glaucoma and macular degeneration. JAMA Ophthalmol. 2013;131(11):1420–6. 10.1001/jamaophthalmol.2013.4471. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Rullo J, Far PM, Quinn M, et al. Local oral and nasal microbiome diversity in age-related macular degeneration. Sci Rep. 2020;10(1):3862. 10.1038/s41598-020-60674-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Ding C, Ding X, Zheng J, et al. MiR-182-5p and miR-378a-3p regulate ferroptosis in I/R-induced renal injury. Cell Death Dis. 2020;11(10):929. 10.1038/s41419-020-03135-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Xu J, Zhou F, Wang X, Mo C. Role of ferroptosis in pregnancy related diseases and its therapeutic potential. Front Cell Dev Biol. 2023;11:1083838. 10.3389/fcell.2023.1083838. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Fuchs Weizman N, Wyse BA, Antes R, et al. Towards improving embryo prioritization: parallel next generation sequencing of DNA and RNA from a single trophectoderm biopsy. Sci Rep. 2019;9(1):2853. 10.1038/s41598-019-39111-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Oczos J, Sutter I, Kloeckener-Gruissem B, et al. Lack of paraoxonase 1 alters phospholipid composition, but not morphology and function of the mouse retina. Invest Ophthalmol Vis Sci. 2014;55(8):4714–27. 10.1167/iovs.14-14332. [DOI] [PubMed] [Google Scholar]
  • 21.Wong MY, Lewis M, Doherty JJ, et al. 25-hydroxycholesterol amplifies microglial IL-1β production in an apoE isoform-dependent manner. J Neuroinflammation. 2020;17(1): 192. 10.1186/s12974-020-01869-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Wang Y, Snel M, Jonker JT, et al. Prolonged caloric restriction in obese patients with type 2 diabetes mellitus decreases plasma CETP and increases apolipoprotein AI levels without improving the cholesterol efflux properties of HDL. Diabetes Care. 2011;34(12):2576–80. 10.2337/dc11-0685. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Paun CC, Ersoy L, Schick T, et al. Genetic variants and systemic complement activation levels are associated with serum lipoprotein levels in age-related macular degeneration. Invest Ophthalmol Vis Sci. 2015;56(13):7766–73. 10.1167/iovs.15-17035. [DOI] [PubMed] [Google Scholar]
  • 24.Johnson DE, Redner RL. An ATRactive future for differentiation therapy in AML. Blood Rev. 2015;29(4):263–8. 10.1016/j.blre.2015.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Wang Y, Yang J, Chen S, Wang W, Teng L. Identification and validation of a prognostic signature for thyroid cancer based on ferroptosis-related genes. Genes. 2022. 10.3390/genes13060997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Zhou C, Wang Z, Li J, et al. Aldo-keto reductase 1C3 mediates chemotherapy resistance in esophageal adenocarcinoma via ROS detoxification. Cancers (Basel). 2021. 10.3390/cancers13102403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Liang J, Cao Y, He M, et al. AKR1C3 and its transcription factor HOXB4 are promising diagnostic biomarkers for acute myocardial infarction. Front Cardiovasc Med. 2021;8: 694238. 10.3389/fcvm.2021.694238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Gestri G, Osborne RJ, Wyatt AW, et al. Reduced TFAP2A function causes variable optic fissure closure and retinal defects and sensitizes eye development to mutations in other morphogenetic regulators. Hum Genet. 2009;126(6):791–803. 10.1007/s00439-009-0730-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Pu M, Li C, Qi X, et al. MiR-1254 suppresses HO-1 expression through seed region-dependent silencing and non-seed interaction with TFAP2A transcript to attenuate NSCLC growth. PLoS Genet. 2017;13(7): e1006896. 10.1371/journal.pgen.1006896. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Huang HX, Yang G, Yang Y, Yan J, Tang XY, Pan Q. TFAP2A is a novel regulator that modulates ferroptosis in gallbladder carcinoma cells via the Nrf2 signalling axis. Eur Rev Med Pharmacol Sci. 2020;24(9):4745–55. [DOI] [PubMed] [Google Scholar]
  • 31.Li LY, Yang Q, Jiang YY, et al. Interplay and cooperation between SREBF1 and master transcription factors regulate lipid metabolism and tumor-promoting pathways in squamous cancer. Nat Commun. 2021;12(1):4362. 10.1038/s41467-021-24656-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Budge JD, Knight TJ, Povey J, et al. Engineering of Chinese hamster ovary cell lipid metabolism results in an expanded ER and enhanced recombinant biotherapeutic protein production. Metab Eng. 2020;57:203–16. 10.1016/j.ymben.2019.11.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Terracina F, Caruana R, Bonomo FP, Montalbano F, Licciardi M. Gastro-resistant microparticles produced by spray-drying as controlled release systems for liposoluble vitamins. Pharmaceutics. 2022. 10.3390/pharmaceutics14071480. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Niazi S, Krogh Nielsen M, Sørensen TL, Subhi Y. Neutrophil-to-lymphocyte ratio in age-related macular degeneration: a systematic review and meta-analysis. Acta Ophthalmol. 2019;97(6):558–66. 10.1111/aos.14072. [DOI] [PubMed] [Google Scholar]
  • 35.Zeng J, Wang Y, Zhu M, et al. Neutrophil extracellular traps boost laser-induced mouse choroidal neovascularization through the activation of the choroidal endothelial cell TLR4/HIF-1α pathway. FEBS J. 2023;290(22):5395–410. 10.1111/febs.16928. [DOI] [PubMed] [Google Scholar]
  • 36.Ghosh S, Shang P, Yazdankhah M, et al. Activating the AKT2-nuclear factor-kappaB-lipocalin-2 axis elicits an inflammatory response in age-related macular degeneration. J Pathol. 2017;241(5):583–8. 10.1002/path.4870. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.He J, Pham TL, Kakazu A, Bazan HEP. Recovery of corneal sensitivity and increase in nerve density and wound healing in diabetic mice after PEDF plus DHA treatment. Diabetes. 2017;66(9):2511–20. 10.2337/db17-0249. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Gauthier T, Chen W. Modulation of macrophage immunometabolism: a new approach to fight infections. Front Immunol. 2022;13: 780839. 10.3389/fimmu.2022.780839. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Kauppinen A, Kaarniranta K, Salminen A. Potential role of myeloid-derived suppressor cells (MDSCs) in age-related macular degeneration (AMD). Front Immunol. 2020;11:384. 10.3389/fimmu.2020.00384. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Camelo S, Lavelette S, Guillonneau X, Raoul W, Sennlaub F. Association of choroidal interleukin-17-producing T lymphocytes and macrophages with geographic atrophy. Ophthalmologica. 2016;236(1):53–8. 10.1159/000446587. [DOI] [PubMed] [Google Scholar]
  • 41.Yi C, Liu J, Deng W, et al. Macrophage elastase (MMP12) critically contributes to the development of subretinal fibrosis. J Neuroinflammation. 2022. 10.1186/s12974-022-02433-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Tan W, Zou J, Yoshida S, Jiang B, Zhou Y. The role of inflammation in age-related macular degeneration. Int J Biol Sci. 2020;16(15):2989–3001. 10.7150/ijbs.49890. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Liu YJ, Guo ZX, Wang SN, Liu YX, Wei Y. Fucoxanthin pretreatment ameliorates visible light-induced phagocytosis disruption of RPE cells under a lipid-rich environment via the Nrf2 pathway. Mar Drugs. 2022. 10.3390/md20010015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Ma XQ, Xie YY, Gong YJ, et al. Silibinin prevents TGFβ-induced EMT of RPE in proliferative vitreoretinopathy by inhibiting Stat3 and Smad3 phosphorylation. Invest Ophthalmol Vis Sci. 2023. 10.1167/iovs.64.13.47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Nashine S. Potential therapeutic candidates for age-related macular degeneration (AMD). Cells. 2021. 10.3390/cells10092483. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.SanGiovanni JP, SanGiovanni PM, Sapieha P, De Guire V. MiRNAs, single nucleotide polymorphisms (SNPs) and age-related macular degeneration (AMD). Clin Chem Lab Med. 2017;55(5):763–75. 10.1515/cclm-2016-0898. [DOI] [PubMed] [Google Scholar]

Associated Data

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

Data Availability Statement

The article and supplementary materials provide a thorough summary of the key findings from this research. For additional inquiries, please contact the author at chenj187@163.com.

Articles from European Journal of Medical Research are provided here courtesy of BMC

RESOURCES